Banks andBank Systems

Banks and Bank Systems

issued quarterly

2007

2

Ukrainian Academy of Banking of the National Bank of Ukraine Publishing Company "Business Perspectives" Banks and Bank Systems International Research Journal Volume 2, Issue 2, 2007 Issued from 2006 Published quarterly ISSN 1816-7403 ISSN online 1991-7074 Heads of the Editorial Board: Prof. Anatoliy Yepifanov Dr. Fedir Shpyh Journal addresses the national as well as international issues of the day, particularly those concerning the most burning aspects of economy organization, performance and governance. Address: Publishing Company “Business Perspectives” Dzerzhynsky lane 10, Sumy 40022 Ukraine E-mail: [email protected] URL: http://www.businessperspectives.org The authors are responsible for reliability of information which materials published contain. Reprinting and reproduction of published materials are possible in case of referring to author and edition. © Publishing Company “Business Perspectives” All rights reserved

Banks and Bank Systems / Volume 2, Issue 2, 2007

3

CONTENTS

J.A. Consiglio Financial Services Privatisation in the CEECs – An Overview

4

Ivana Valová New Capital Rules According to Basel II

14

Matthias Menke, Dirk Schiereck Private Equity Investments in the Banking Industry – The Case of Lone Star and Korea Exchange Bank

22

Christophe J. Godlewski An Empirical Investigation of Bank Risk-Taking in Emerging Markets Within a Prospect Theory Framework. A Note

35

Mete Feridun Financial Liberalization and Currency Crises: The Case Of Turkey

44

Alper Ozun, Atilla Cifter Industrial Production as a Credit Driver in Banking Sector: An Empirical Study With Wavelets

69

AUTHORS OF THE ISSUE

81


4

FINANCIAL SERVICES PRIVATISATION IN THE CEECs – AN OVERVIEW J.A. Consiglio* Abstract Financial services privatization in the Central and Eastern European countries (CEECs) is an account of significant differences in strategy processes and implementation, and of strongly asymmetric evolution paces of bank and economic restructuring. The inherited background disorganization factor, different motivating factors, extent and pace of the process, and the lingering argumentative positions on its real need, make this one of the most fascinating contemporary banking history themes. Key words: privatisation, financial services, CEECs. JEL classification: D53, E44. Popular Western media presentation of the post-communist block of countries often suggests their being homogenously placed in the economic transition ladder. The spread of characteristics of the CEECs’ banking sectors shows quite sufficiently how differently they must be considered from each other: for example in aspects like the pace of change, privatization infrastructures, legal and social elements. Just as much as the move from the economics of communism to the economics of democracy required a particularly sequenced methodology, similarly the shift from privatization to market structuring, inclusive i.e. of the element of M&As in the financial services industry (FSI), could not be expected to flourish before privatization too had evolved to status and levels of both mass acceptability and efficient methodology. By the end of the 1980s transition from communism to democracy still had much to suggest a perception there of a big black box. The regimes were “familiar” with both systems in terms of their respective economic implications, but the paths from one economic system to the other very often were not as familiar to them. Even if some exceptions must be made, a substantial part of the literature on this much-dealt-with ground of privatization in the CEECs has predominantly been characterized by holistic approaches that consider the process as a feature of economic transition that has generally common, and totally undifferentiated, characteristics across the whole spectrum of these countries’ economies. This paper will focus on some of the more particular characteristics of a specific sector’s, financial institutions’ privatisation, in that region. It must be considered as part of a lengthier study that this author has made which includes, inter alia, various country case studies, synthesisation of deduced specific country characteristics of the process into a general model of the process in the whole region, and adoption of the analytical evolution model to the experience of a small Mediterranean island’s analogous privatization experience. Economists and policymakers in the CEECs possessed a lot of information on the transition from democracy to communism, and on its economic parallel i.e. from capitalism to central planning and collective property. But no instance or guide was available for the reverse process. But in this context it is useful to note that Rossini (1998) draws attention to the fact that the process towards a socialist economy had also partially been undertaken by several Western democracies who, in this context, one could say were wrongly described as “market economies”1.

* University of Malta, Malta. Rossini G. (1998). Review of The Economics of Post-Communist Transition by O. Blanchard – Economic Notes, Banca Monte dei Paschi di Siena, No. 1/1998. 1

© J.A. Consiglio, 2007.


5

Despite the Cold War, during the 1950s and 1960s there seemed to be an inclination towards a form of convergence of some Western economies towards a system in which the original spirit of capitalism could coexist with a strong presence of the state in the economy. One version of this was touted as “the social market economy”. Such convergence attempts, it can be argued, came to a gradual halt in the 1980s when in most Western European countries the state could control almost a half of the flow of wealth produced, with the other half being allocated through market institutions. A classical and influential example of such a situation was Italy with its IRI (Istituto per la Ricostruzione Industriale). By contrast, for several historical reasons such convergence did not proceed on the other side of the Iron Curtain. One could argue that one of the main obstacles against moving towards a systemic organization more akin to the capitalist one was not the large diffusion of state property and state presence in all sectors of the economies there (including in the FSI), but rather the concomitant centralization of all relevant, and less relevant, economic decisions.

The “disorganization” factor Blanchard (1998), when exposing the characteristics of the economics of post-Communist transition, is prompt to emphasise the effect of the disappearance of central planning in former Communist countries1. There was a ‘disequilibrium’ factor which explains the poor performance of some of those economies in both a total and sectorial sense. Differently said, the restructuring of Eastern Europe was muddled by a lack of that coordination which in capitalist economies is provided partly by markets and partly by the state, even when reacting to unexpected crises. Markets in post-communist transition countries did not exist, not at least in the freely functional and operative sense that the Western world knew. The state was neither ready to use macroeconomic instruments, nor to undertake any industrial policy which had to operate without compulsory planning. And yet there were differences amongst some of those former Communist countries, the most noteworthy being that of the timing and speed with which some of them started their economic revolution, regardless in some cases of heavy social distress that was involved. Poland, for example, was able to adopt certain forms of market legislation and practices well ahead of others. A good example here was that of antitrust law, which it adopted early on (indeed even before certain Western European countries). Hungary too was fairly early off the mark with a variety of legislative changes. Other countries were loath to decisively undertake the same route, and it became ever clearer that different adjustment speeds, and a different politico-economic psychosis, prevailed within this group of vastly different countries. Perhaps the best illustration of this was that, in various ways, they were deeply, awkwardly, and differently dependent on the notion of “nothing succeeds like what at least popularly appears to be a success”, a scenario that various astute politicians of the old guard for long still continued to play with wisely for their own personal interests. The study of privatization in the financial services sector in the economics of post-Communist transition must therefore also concern itself with some of the reasons why these Eastern European countries experienced different adjustment speeds in this sector, and why restructuring was so awkward to implement. The response of output to transition policies was, in this sector as in others, often U-shaped, with some countries still trapped in the bottom of the U. Expectations often went wrong, and particularly disappointed were those who though that recovery was just a matter of freeing latent animal spirits of economic agents. When we touch upon the long-term nature of the whole process, issues relating to various factors, including the historical FSI evolution models applicable, the restructuring context and methodologies applied, some very important human resources factors, and others, are all important and ex-

1

Blanchard O. (1998) – The Economics of Post-Communist Transition – (Clarendon Press, Oxford).


6

plain why some CEECs stayed long in this position and why, for example, none of the CEECs has – up to our times – so far evolved as a reputable, dynamic, competitive financial centre able to hold its own with the London, New York, Paris, Frankfurt, and Tokyos of the Western world. Gardner and Molyneux (1990, p. 143 et seq.) hold as the traditional explanation for the development of financial centres in the nineteenth and early twentieth centuries “the way in which [these] tended to dominate international trade financing and capital export, and the important role they have played in the world economy”1. For the CEECs this was certainly not the case. In their economic environment the privatization process was nothing like what many in the West would imagine it should be. If Russia is taken as only one brief example of this, at a period when official policy was professedly very much in favour of the process, the economy was showing a sharp decline in output which was much larger than the decrease in employment. The simple explanation was that in most cases redundant workers were retained on the books of firms, and adjustment took place in the form of reduced wages rather than dramatic lay-offs. And the FSI was often no exception to this approach. This sort of softer adjustment carried out in many firms has, as one possible explanation, the fact that many were in fact cases of insider privatizations. Insider control, alongside state control, is viewed by Bonin J.P. et al. (1998, p. 1 et seq.) as having been continued, and often entrenched, from the fact that privatization, as practiced in many of the economies in transition (EITs), often failed to insure independent governance2. But insider privatizations, one must admit, are not the only explanation for reallocation and restructuring not to have produced immediate effects. What we refer to as “disorganization” was often the more conspicuous culprit which, in cumulative terms over national levels, also caused derailment risks of the totality of the economies. Establishing markets where there are only centrally planned links is anathema to the processes of restructuring and reallocation. In the absence of properly functioning markets, privatization in the CEECs could not be undertaken according to the full methodologies and contexts known in Western Europe. Blanchard (op cit) reaches the conclusion that insiders privatization was not only one of the most common ways (excepting possibly in the Czech Republic) of pushing takeovers, but it may also have been the most efficient, given the actual conditions of financial markets. It avoided dramatic underpinning and made the adjustment more gradual, even if painful over a longer period. This however is not a view generally shared by all Western economists, many of whom would not agree with him on the assumed ground that inefficiency may undermine insiders privatization.

Objectives The raising of public efficiency was of course one of the stated objectives of privatization programmes in many countries, including in the CEECs group. But it is not the only one. Others include: ♦ A concomitant improvement of the quality and range of services to citizens. ♦ The redeployment of national resources in a more efficient manner. ♦ Allowing governments to concentrate on what are – according to a particular political stance – considered as being their sole or core activities in their economic role, and in the process encouraging an independent “enterprise culture”, and ♦ The turnaround or consolidation of public finances. From a more restrictedly FSI viewpoint, privatizations are often viewed as the way forward for: ♦ ♦ 1

The introduction, enhancement, and development of local capital markets. Encouraging competition and modernization in the sector.

Gardener E.P.M., Molyneux P. (1990). Financial Centres in Changes in Western European Banking – (Routledge, London). 2 Bonin J.P. et al. (1998). Banking in Transition Economies – (Edward Elgar, Cheltenham).


♦ ♦

7

Attracting foreign and domestic private investment in infrastructure which enhances technology through strategic partnerships. Labour market restructuring, i.e. away from state-owned enterprises (SOEs).

Achieving all or several of these objectives is easily traceable as a leitmotiv of many prominent CEEC politicians’ (sincerely, or otherwise, motivated!) speeches and writings during the first decade after 1989. But that analysis quickly brings up the timing factor as often raising the issue of whether the process should in fact have been done when it was. Or, indeed, when it was not. The fact however remains that with the breakup of the Soviet Union, and the start of market-oriented reforms in many former socialist of Central and Eastern Europe, that was a period where the prospect of privatizing inefficient state-owned companies started to figure not only, on the one hand in academic writings, but also in the new popular mens, the new general mindset, which started to create and form mass and popular perceptions. As one of the headline events symbolizing change from central planning to capitalism, privatization seemed to promise an end to the inefficiencies of the former, and in mass perception became the key to freezing the resources and talents of people everywhere, but more so in the CEECs, where the ideal of lifting living standards to those of the industrial countries of Western Europe became a holy grail.

When did it start? Havrylshyn and McGettigan (1999), who were prominent in the International Monetary Fund’s (IMF) Europe II, and Policy Development & Review, departments, summed up some analysts’ views holding that when the CEECs embarked on the privatization route in the way of severally apparently nationally embraced formal programmes, there existed no formally theoretical base to guide the practical process of economic transition. For many analysts there only existed generic – or even sometimes insistently specific – theories on capitalism and socialism. This is perhaps a somewhat surprising position if one considers the fact that the IMF argues in its own favour a claim for historical antecedent in this area – [its 1977 strict pressurizing of the British Labour government that it effects a sale of its then held shares in British Petroleum (BP)] – prior, that is, to the introduction by the Thatcher government in the UK in 1979 of a formal lengthy programme of divestitures. That was one year after Prime Minister Harold Wilson – who had made much of his intention to modernize Britain with a much touted “white hot technological revolution” – had resigned. In the UK several Tory politicians have claimed credit for that country’s privatization process, but later former Chancellor Sir Nigel Lawson held that the paucity of references to privatization in the Conservative Party’s 1979 general election manifesto was in actual fact, according to him, a reflection of “Lady Thatcher’s lack of enthusiasm” for privatization1. But Megginson, Nash, and Van Randeburgh (1994), and again Jones, Megginson, Nash and Netter (1999)2, also argue that “the first large-scale, ideologically-motivated denationalization programme 1

“The Financial Times Guide to Privatisation” – Jan 8th, 1996. Vide also FT, Oct 15th, 1999 where Robert Peston significantly reports thus, under “Thatcher’s flagship finally runs aground”: “The death sentence has been posted. The 20-year campaign to privatize public services is over. No less a figure than the Prime Minister (Tony Blair) read the last rites over the archetypical Thatcherite policy this week. “Today the issue is not ruling back government, hammering trade unions, or more and more and more privatization of public services”, Tony Blair wrote in a little-noticed article in Monday’s issue of London’s “Times” newspaper. “It is investment in public services, allied to their reform and modernisation…………….”. 2 Megginson W.L., Nash R.C., & Van Randeburgh M. (1994). “The Financial and Operating Performance of Newly Privatised Firms: International Empirical Analysis”, The Journal of Finance Vol. XLIX, No. 2, June. Jones S.L., Megginson W.L., Nash R.C., & Netter J.M. (1999). “Share issue privatizations as financial means to political and economic ends”, Journal of Financial Economics, No. 53, pp. 4-5 and 217-253. Other discussions of the purely historical aspects of the development of privatization are to be found in, inter alia, Jenkinson & Meyer (1988), Van Der Walle (1989), Shirley & Nells (1991), Brada (1996), Bennell (1997), Yergin & Stanislaw (1998), and The World Bank (1995).

8


of the post-war era was launched by the government of Konrad Adenaur (one of the so-called “founding fathers” of the European Union) in the FDR. In 1961 (i.e. four years after the EU’s birth) the German government sold a majority stake in Volkswagen1 in a public share offering heavily weighted in favour of small investors. Four years later the German government launched an even larger offering of shares in VEBA, the mining and heavy industry giant. Both offerings were initially very favourably received, increasing the number of private shareholders in Germany from around 500,000 to some 3 millions, but the appeal of share ownership did not survive the first cyclical downturn in stock prices, and the government was actually forced to bail out many small shareholders. The Adenauer government’s 1960’s privatization programme had officially announced objectives which read identical to those of the two decades’ later Thatcherite plans. When one then pursues the historical iter of privatizations even in other European countries (Denmark, Italy, France, and from the mid-1990s in the EU generally), in Asia (Malaysia, Singapore, Japan), and South America (Chile2), one sees that, regardless of ideological basis, the objectives are nearly always similar, viz: ♦ Raising revenue for the state ♦ Promoting increase efficiency ♦ Reducing government interference in the economy ♦ Promoting wiser share ownership ♦ Providing opportunity to introduce competition ♦ Exposing SOEs to market conditions ♦ Developing national capital markets. After Germany and the UK, the next major Western European nation to pursue privatization as a core element of its politico-economic policy was France. Jacques Chirac’s conservative government which came to power in March 1986 was declaredly committed to selling off not only the financial and industrial groups which had been nationalized during 1981 and 1982, but also the large banks nationalized even further back by General Charles De Gaulle in 1945. With the benefit of hindsight it can be said, for the period between 1965 and 1979, that the number of non-European governments who implemented deeply well enough thought out scientific policies of privatization was fairly small. Later, after 1987, privatisation programmes spread rapidly round the globe, and the 1989/1990 events in the former USSR and Eastern Europe thereafter shifted the “movement” to that part of the world. Shirley and Nells (1991) motivate the imperative for privatization in these countries as being that “to create a market economy as quickly as possible, using all available methods, and almost regardless of the social cost entailed”3. However, irrespective of when, and to where, an exact chronological start of this widely discussed economic policy innovation is to be accredited, it can be argued that from the decade between 1989 and 1999 – which was the first whole and continuous period of the transition economies’ experience with privatization – some elements for the cobbling together of a workable “model” of transition or transformation which, it can be imaginatively said, should be ‘grateful’ or ‘owe a lot to’ privatization, can be individualized.

1

There is a touch of irony in the fact that the German term “Volkswagen” actually means “People’s Car”. Possibly the one which attracted most media interest was the case of Chile, where the Pinochet government, which gained power after the ouster of Salvador Allende in 1973, attempted to privatize companies that the Allende regime had nationalized during its short but eventful reign. However the process was poorly executed, and required very little equity investment from purchasers of assets being divested. Thus, many of these same firms were renationalized once Chile entered its debt and payments crisis in the early 1980s. Chile’s second privatization programme, which was launched in the mid1980s, and relied more on public share offerings, than direct asset sales (where the government acted as both creditor and seller) was much more successful. These Chilean vicissitudes are assessed in more detail in Yotopoulos (1989). 3 Shirley M. & Nells J. (1991). “Public Enterprise Experience” – World Bank, Washington DC, Publication No. 9800; (1991) – “Public Enterprise Reform – The Lessons of Experience” – World Bank, Washington DC. 2


9

In considering some of these elements it is easy to see that all are – at least theoretically – such as can be hypothesized in a general manner for the financial services sector. Consider, inter alia, ♦ ♦ ♦ ♦

The forcing through of moves from a sellers’ to a buyers’ market – through price liberalisation this is also possible in the market for financial services. The enforcing of a hard budget restraint – through privatization itself and the elimination of various government support mechanisms the FSI is, again, a possible contributor1. The reallocating of resources from old to new activities – closures and bankruptcies, combined with the establishing of new enterprises, within the FSI is again theoretically relevant. The restructuring within surviving firms – even in the FSI labour rationalization, product line changes, and new investment, are relevant factors to be considered.

Extent and pace Any hypothesized model of the evolution of the CEECs’ FSI privatization perhaps carries too much to resist analogy with Schumpeterian creative destruction. And yet, again naturally a posteriori, it can now be said that this became the generally accepted road down which developing countries chose to travel. Between 1988 and 1992 the pace of privatization in these countries decreased dramatically. From 6% of total world privatization sales income in 1988, the developing countries’ share rose to 42% in 1992. In 1995 an ILO report stated that proceeds from the sale of public enterprises in developing countries rose from just over US$2 bn in 1988 to almost US$ 20 bn in 19922. And indication of to what extent privatization proceeds became for these countries the more important source element for their financing needs, when compared to their previous heavy dependence on external financing, is suggested by the following World Bank figure covering just the beginning of these countries’ experiences with privatization.

Fig. 1. Provatisation proceeds in developing countries ($bn) By the first half of 1995 worldwide privatization was progressing at a much slower pace compared to the same period in 1994. During that period total proceeds from the sale of SOEs stood at US$ 18.4 bn, down from the peak reached in the comparable previous year period of almost US$ 37

1 2

There is of course a ‘flipside’ soft budget constraint position that can also be made, with concomitant issues. “Privatisation surge in developing countries” – Financial Times, April 26th, 1995.

10


bn. Privatisation International (1995) – published in the US – however maintained that a large number of deals were already in the pipeline for the second half of 1995, or for early in 1996, and so, “barring an adverse turn in markets, the value of privatization worldwide could still reach US$ 45 bn or more for 1995”1. World Bank calculations show that privatization raised US$ 270 bn worldwide in the period between 1988 and 1993. As the above figure suggests, developing countries received a large share of those revenues. State sell-offs, in Latin America, in Asia, and in other developing countries elsewhere raised a total of US$ 96 bn during the period in that group of countries, with more than a third of it coming from foreign investors. That size of activity was large in itself, but both in its factually completed component as well as in what was being programmed down the road, was in fact to an extent that at one stage loomed dangerous in the eyes of the OECD. In 1995 the OECD warned that the current privatisation programmes in its member countries were so large that their implementation would have a powerful impact on the countries’ financial systems. It estimated that these programmes could result in equity offerings totaling US$ 200 bn during the successive five years, creating the crucial issue of whether the financial markets would be able to absorb this. The OECD’s report included warnings referring, inter alia, to “fatigue among retail investors”, and to the possibility that “future privatizations would have either to reduce the tranches specifically directed to retail investors or to enhance the attractiveness of the offering to these investors”2. The digression here towards OECD experience, and away from CEECs specifically, is explained by the intention of underlining the substantial differences in both the economic and political environments, and timings, within which privatizations occurred in these different scenarios of the world. Whilst the developed world presented issues like ♦ equity prices of offerings needing to remain strong to support investor appetite for new issues, ♦ downturns in equity markets leading to shifts away from equity and thus potentially undermining privatization plans, ♦ some large-scale privatizations having negative effects on share prices (e.g. in 1994), and having to compete in a rising interest rates market, ♦ meanwhile when some governments carried out privatization programmes in the transition economies a key question constantly bothering investors and analysts was “Is the country, or that sector in that country, ready?” Even when the whole concept has long been accepted as an ideal or methodology along the road of economic transition, the banking sector in Central Europe was factually not yet totally ready for privatization, and forcing the issue was likely to hurt rather than help banking development.

A non-proven case? For a long time many observers’ first look at the progress made by the CEECs towards transition was summed up in the question about what is the extent of the private sector’s share of an economy. But whilst this was a legitimate way of measuring a country’s progress away from central planning, it did not, on its own, necessarily follow that bank privatization was always essential or, when effected if needed, that it was well timed. The simple, key, standard question would need to be: what was privatization trying to achieve in that particular country, and how would it increase banking efficiency? Answers from different CEECs are less simple, and specific case-studies3 1

“First-half slowdown in worldwide privatisations” – Financial Times, July 18th, 1995. “Privatisation and Capital Markets in OECD Countries” in “Financial Market Trends (OECD), Feb 1995. See also “OECD warns on scale of privatization programmes”, FT, March 6th 1995. Specific case studies of countries where such scenarios effectively came very close to reality feature in the Ph.D. thesis of which this paper is a component. 3 The financial services privatization experiences of Hungary, the Czech Republic, the Slovak Republic, Romania, Estonia, and Russia, are extensively analysed in my 2006 Ph.D. thesis on the subject, as referred to on Page 1. 2


11

highlight issues like the extent to which Central Europe’s economic and industrial vicissitudes in post-privatisation years showed much progress away from state planning, whether and at what stage had the state really arrived in total readiness to give up control of the banking sector, and again indeed whether arguably it should have at all. Michael Kapoor, who in 1996 was making a strong case against “forced quasi-privatisation” distracting the CEEC banks from acting more commercially, assessing risks better, and generally offering better service, maintained that at its simplest privatisation is a bottom-line question: the commercial orientation of private companies means that they should operate more profitably than state dinosaurs1. For corporates needing fundamental refocusing to survive, the argument is compelling, and in many cases privatization was indeed an essential step to commercialization. But the bottom-line argument was at times of limited relevance to Central European banks, some of which – according to certain yardsticks – were actually amongst the most profitable in the world. The excessive caution they were charged with stemmed as much as anything from a determination to keep profits up. Indeed, problems did as often as not come from naïve commercialization (e.g. chasing new business when distinctly unqualified to gauge risks) as from complacency. Moreover, many would indeed argue that the banking system in any of these countries was run on commercial, rather than political, lines. Along the way, whilst state influence certainly remained there, direct control had long gone. Nor could privatization be seen as the guaranteed key to bank restructuring, for there was actually no evidence that privatization per se would speed up the process. At times, even the IMF seemed to be maintaining an arm’s length approach to the issues of privatisation’s worth as an essential component of bank restructuring methodology within the much wider sphere of appropriate macroeconomic policy in transition economies2. Indeed the presence of competition seems to be a far more effective catalyst to development than mere transfer of ownership. Most observers for example agreed that Hungary’s OTP Bank restructured more effectively than the Czech Republic’s Sporitelna, even though the Czech savings bank was privatized significantly earlier. Up to early in 1996 the case for bank privatization in the CEECs was, at best, unproven, and at that point in time, when the difficulties of selling what were effectively large institutions in some of these economies were factored in, the conclusion was that it was better to concentrate on increasing their efficiency – if necessary even under state ownership – than on privatizing them. The situations in Hungary, the Czech Republic, and Slovakia are specific examples of that particular conjuncture. Most of the literature originating from non-CEEC sources suggests that the perception around the beginning of 1996 was that Central European banks were not yet ready for privatization, and that forcing the issue was likely to hurt rather than help banking development. The general popular approach was that of looking at the progress made towards transition, and posing as a first quote the private sector’s share of the economy. But whilst this was a legitimate way of measuring these countries’ progress away from central planning, it did not necessarily follow that bank privatization was essential. The simple question which had to be posed was: what was privatization trying to achieve, and how would it increase banking efficiency? The answer was always less simple, and highlighted the extent to which Central Europe’s economic and industrial future at that stage still remained unplanned. The state was not yet ready to give up control of the banking sector, nor, arguably, should it have done so.

1

Kapoor M. (1996). “Are banks ready?”, Business Central Europe, February, pp. 7-11. It is significant, for example, that privatization is totally absent in all the papers published in the IMF’s 1997 publication “Systemic Bank Restructuring and Macroeconomic Policy” – (Alexander W.E., Davis J.M., Ebrill L.P., Lindgren C.J. (eds) – which was the joint product of the IMF’s Fiscal Affairs and Monetary & Exchange Affairs Departments arising “from the need to advise countries on how to deal with banking system problems, and on how to consistently incorporate the macroeconomic aspects of systemic bank restructuring into IMF policy advice and IMF-supported adjustment programmes”.

2

12


At its simplest, privatization is a bottom-line question: its defenders will consistently hold that the commercial orientation of private companies means that they should operate more profitably than state-owned dinosaurs. For companies needing fundamental refocusing to survive, the argument is compelling and in many cases is an essential step to commercialization. But the bottom-line argument was of limited relevance to Central European banks, which were arguably amongst the most “profitable” in the world. The excessive caution they were rightly charged with stemmed as much as anything from a determination to keep profits up. Indeed, problems did as often as not come from naïve commercialism (e.g. chasing new business when unqualified to gauge risk) as from complacency. By early 1996 few would deny that the banking system in any of these countries was run on commercial, rather than political, lines – while state influence was certainly there, direct control had long been gone. Nor could privatization be seen as the key to bank restructuring, for there was no evidence that privatization per se was speeding up the process. The presence of competition was a far more effective catalyst to development than mere transfer of ownership. For example, Hungary’s OTP Bank restructured more effectively than the Czech Republic’s Sporitelna, even though the Czech savings bank had been privatized significantly earlier. At that stage it therefore appeared that the case for bank privatization in the CEECs remained at best unproven, and when difficulties of selling such large institutions were factored in the obvious conclusion was that it would be better to concentrate on increasing their efficiency, if necessary still under state ownership, than on privatizing them. The most obvious example of this debate was Poland, where international creditor pressure was forcing the authorities to take drastic steps to privatize the long-stagnant banking industry. But, because it was impossible to sell the banks on what were effectively shallow domestic capital markets, privatization was likely to be a sham by any meaningful criteria. If banks could be sold to private investors working through aggressively managed funds, then the debate would have been different, because there would be an emphatic distancing from the state, and corporate governance could be imposed by the market. But with 40% of Polish stock market capitalization already dominated by the banks, significant further issues were impossible, meaning that the shares would, one way or other, be given away (unless the banks were to be sold to foreigners, which was then politically unlikely), and market control, as opposed to legal ownership, would still be lacking. 1996 – WHAT THE POLISH GOVERNMENT WANTED The structure of Polish banking was too dispersed, and the Government intended to rationalize the industry around existing regional and specialized banks. In early 1996 the structure was: ♦ Five state-owned commercial banks ♦ Three state-owned socialized banks ♦ Agricultural and cooperative banks (BGZ SA Holding) ♦ Private banks ♦ Foreign banks. Post-reform it wanted it to be: ♦ Two or three banking groups with some foreign capital and remaining state minority share (formed around Handlowy Bank and Pekao Bank) ♦ PKO BO state-owned savings bank ♦ BGZ SA (food economy bank) and cooperative banks ♦ Two or three private banking groups formed out of the regional banks, probably with a remaining state shareholding ♦ Socialised banks (e.g. mortgage). Source: Business Central Europe, Feb. 1996, p. 7.


13

Ironically, this very lack of domestic capital appeared to be as the only thing which, in an inverse sort of manner, motivated the banks themselves towards privatization. Up to around 1994 Poland’s largest bank, Bank Handlowy, showed absolutely no eagerness to privatize: ownership, to it, was irrelevant. Two years down the road it became desperate to do so, not out of conversion to Thatcherite idealism, but because it had become seriously scared by its inability to compete with larger banks. It needed growth capital; there was none available in Poland; that meant going to international markets; and that in turn required the transparency of a listed – and therefore private – company. Looking prophetically ahead the then deputy director of the bank put it very wisely: “If we don’t have access to international markets, then in ten years’ time we’ll be the 20th biggest bank [in Poland], not the biggest”.

Conclusions It is easy to see how the debate on financial services privatization in the CEECs can become entangled in conflicting positions between issues related to systemic banking sector problems on the one hand, and those associated with individual banks in the region on the other. A well-functioning banking infrastructure should be considered as a public good everywhere. The different CEECs’ perception of the various elements discussed here is one important explanation of why the different governments there addressed banking sector problems with heterogeneous (sometimes outrightly inconsistent and often questionable) levels of urgency. And this must be seen alongside the fact that systemic banking problems often turn into full scale banking crises with related negative effects for macroeconomic growth. Those CEECs which did consider the banking sector positively (e.g. Estonia) did so because there was realization of the fact that the size and impact of a systemic banking crisis on economic output and growth are dependent on the stage of development of the financial sector, and of course its linkages with the real sector. Actually however there are two views to be considered here. On one hand is the argument that in a big country, say, the US, if a significant portion of such a country’s banking system would collapse, the relative adverse consequences would be larger than if a similar event occurred in a smaller country. Matousek (1997) on the other hand does not think that this is quite so where big developed economies are concerned1. His view is that it is ‘in many transition countries [that] bank credits still represent the only financial channel for the real sector, and that, moreover, a small number of commercial banks tend to have dominant positions as key lenders to the biggest companies”. Therefore a collapse of such key institutions in the CEECs could be very harmful indeed for the real sector as well as for the emerging financial sector, something which, he holds, is not the case elsewhere. This paper has not looked at the actual transformation and restructuring process of the banking sector in Central and Eastern Europe, a process which can generally be said to have commenced around 1989-1990, and which can justifiably be held as yet another consequence of known political events. Systemic national bank system restructuring during the 1990s developed into a discipline of its own, had a dynamic of its own, and comprised comprehensive programmes to rehabilitate significant parts of countries’ banking systems, with the ultimate objective being that of providing vital bank services on a sustainable basis. Close study of policy assessments and strategic steps relevant to the different situations in each CEEC invariably reveals many issues that make this possibly one of the most fascinating areas of study in contemporary banking history.

1 Matousek R. (1997). Comments to paper by Hans Bloomestein in “The New Banking Landscape in Central and Eastern Europe” – (OECD, Paris).


14

NEW CAPITAL RULES ACCORDING TO BASEL II Ivana Valová* Abstract The Basel Committee on Banking Supervision (known as “the Basel Committee”) was established by the central-bank Governors of the group of ten countries at the end of 1974. In 1988 the Basel Committee on Banking Supervision decided to introduce a capital measurement system for a credit risk commonly referred to as the Basel Capital Accord (known as “Basel I”). Amendment to the Basel Capital Accord to incorporate market risks was issued by the Basel Committee on Banking Supervision, in 1996. The final version of the New Basel Capital Accord (known as “Basel II”), covered operational risk, was released in June 2004. The article, “New capital rules according to Basel II”, is devoted to the problem risks by credit financial institutions. The paper dedicated to the importance of risks, capital adequacy, risk measurement and risk management, and advantages and disadvantages of the new capital rules are described. The Czech National Bank, a central bank of the Czech Republic, defines the prudential framework for banking business and cooperates with banks to implement the New Basel Capital Accord too. The paper talks about trends and actual situation in accordance with the New Basel Capital Accord and some interesting things being related to the Czech Republic too. Key words: Capital adequacy, Czech national bank, Basel Committee on Banking Supervision, Basel Capital Accord, New Basel Capital Accord, risk management. JEL classification: G18.

Introduction A creditable and stable banking sector is one the basic preconditions for a functioning economy. But such stability is not guaranteed by market mechanisms alone. The activities of banks are governed by a number of injunctive regulations. We have to be aware that banking sector is somewhat different from other sectors. It is a specific area with specific banking products and services, with specific risks and management. Because of these the banking sector has to be regulated and supervised. Regulators want to ensure that banks and other financial institutions have sufficient capital to keep them out of difficulty. Regulators try to protect depositors and also the wider economy. The reason is that the failure of a big bank has extensive knock-on effects. The risk of knock-on effects that have repercussions at the level of the entire financial sector is called systemic risk.

Objective and Methodology The aim of the paper is to give brief information on a theory of the Basel Capital Accord, and namely of the New Basel Capital Accord, issuing by the Basel Committee on Banking Supervision. The article points the moral that it is necessary and very important for each bank to measure, manage and monitor the banking risks. The attention is given to capital adequacy and basic changes in the risk management in accordance with Basel II and advantages and disadvantages that the new rules bring to banks. The basic method of submitted article is the deduction. It is gone from common pieces of knowledge and theory to particulars.

*

© Ivana Valová, 2007.

Masaryk University, Czech Republic.


15

Results Banking Regulations have several goals: improving the safety of the banking sector, levelling the competitive playing field of banks through setting common benchmarks for all players, promoting sound business and supervisory practices. Regulations have a decisive impact on risk management. The regulatory framework sets up the constraints and guidelines that inspire risk management process of banks. Regulations promote better definition of risks, and create incentives for developing better methodologies for measuring risks. In 1988, the Basel Committee on Banking Supervision1 issued the Basel Capital Accord. This accord established minimum levels of capital in order to strengthen the soundness and stability of the banking system as a whole and create a more “level playing field” in competitive terms among internationally active banks. Since 1988, the framework has been introduced in virtually all countries with internationally active banks. In 1999, the Basel Committee decided to replace the Basel Capital Accord with a more risk-sensitive agreement. The new framework is based on current risk management techniques. Banking risks Banking risk is uncertainties resulting in adverse variations of profitability or in losses. There are a large number of risks in the banking sector. Most of them are well known. In connection with Basel II, the credit risk, market risk and of recent years operational risk too are very often discussed. The first of all risks in terms of importance is credit risk. Credit risk is the risk of loss due to a deterioration of the credit standing of a borrower. We may not forget the view of the risk differs for the banking portfolio and the trading portfolio. Traditional measures of the credit quality of debts are ratings. We know the internal rating and external ratings. Internal ratings use bank, and external ratings are made by rating agencies Moody's, Standard & Poor's and so on. There are various types of ratings. Rating is ordinal measures of credit risk, but it is not sufficient to value credit risk. Because of that the portfolio models are used. Market risk is the risk of adverse deviations of the mark-to-market value of the trading portfolio, due to market movements, during the period required to liquidate the transaction. The period of liquidation is critical to assess such adverse deviations. If it gets longer, so do the deviations from the current market value. The last risk, we will talk about is operational risk. The New Basel Capital Accord defines operational risk as “The risk of direct of indirect loss resulting from inadequate or failed internal processes, people and systems or from external events”. Operational risk covers people's risk (it means human errors), processes risk (for example errors in the recording process of transactions), technical risk (model errors, the absence of adequate tools for measuring risks) and information technology risk (system failure). Capital adequacy Capital adequacy exists for a long time and it is the main pillar of the regulations. The two most important capital adequacy requirements are those specified by the Basel Committee on Banking. The first implemented accord, known as Basel I, was focused on credit risk and set up the minimum required capital as a fixed percentage of assets weighted according to their nature in 1988. The range of regulations extended gradually later. A major step was the extension to market risk, with the 1996 Amendment. Basel II enhances the old credit risk regulations.

1

The Basel Committee was established in 1974 by supervisors of the Group of Ten (G10) countries.


16

Capital adequacy accordiny to Basel I was defined as a single number that was the ration of a banks capital to its assets. There were two types of capital – tier 1 and tier 2. The requirement was that tier 1 was at least 8% of assets. Each class of asset has a weight of between zero and 100%. The weighted value is multiplied by the weight for that type of asset. The Capital Accord is to be replaced by New Capital Accord. The new capital framework is based on three pillars: minimum capital requirements, a supervisory review process, and effective use of market discipline. With regard to minimum capital requirements, the Basel Committee set that a modified version of the actual Basel Capital Accord should remain the standardised approach. But there are possibilities for some banks to use internal credit ratings and portfolio models. And so a capital requirement of a bank can be in relation to its particular risk profile. It is also possible that the Accord's scope of application be extended, so that it fully captures the risks in a banking group. Basic characteristics of BASEL II Basel II is a regulatory capital adequacy framework, which will be implemented by all banks located in the EU countries and by all internationally active banks in non-EU G10 countries. The main objective of the framework is to improve security and soundness of the financial system. The new package is the set of consultative documents that describes recommended rule for enhancing credit risk measures, extending the scope of capital requirements to operational risk, providing various enhancements to the existing accord and detailing the supervision on market discipline pillars. The New Basel Accord provides a menu of options, extended coverage and more elaborate measures, in addition to descriptions of work in progress, with yet unsettled issues to be streamlined in the final package. The New capital accord contains three basic pillars: 1. Pillar – Minimum capital requirements The pillar contains minimum capital requirements for credit risk, market risk, and now also covers operational risk. The pillar offers a wider range of risk measurement approaches for determining capital requirements, including banks' own internal models. Different options for credit risk are: 1.

Standardised Approach (SA) – It is the simplest method. The risk weights are derived from ratings set by external credit assessment institutions or export credit agencies.

2.

One of two internal ratings-bases (IRB) ♦ Foundation IRB Approach (FIRB) – By the method bank uses own estimates of the probability of default of its client and banking supervisory authorities determine the other characteristics. ♦ Advanced IRB approach (AIRB) – All the components are determined by banks.

Measurement methods for the market risk are unchanged, but there is change in the definition of the trading book and assessment of the capital requirement in the case of a small trading book. Banks can use one of three basic methods for measurement of operational risk: 1. Basic indicator approach (BIA) – Calculation of the capital charge as a fixed percentage of the bank's net income. 2. Standardised approach (STA) – Calculation of the capital charge separately for each business line, as a fixed percentage. 3. Alternative standardised approach (ASA) – banking supervisory authority may allow bank to use another alternative indicator for commercial or retail banking business lines. 4. Advanced measurement approaches (AMA) – Banks are allowed to use their own various internal methods and models. The methods and models have to be approved by the supervisory authority. Table 1 shows an overview of the methods for risk management in accordance with Basel II.


17

Table 1 The methods for risk management within Pillar 1 Type of risk

Methods

CREDIT RISK

Standardized Approach

Foundation Internal Rating Based Approach

MARKET RISK


Internal Models Approach

OPERATIONAL RISK

Basic Indicator Approach


Advanced Internal Rating Based Approach

Advanced Measurement Approach

2. Pillar – Supervisory review process The second pillar is intended primarily on the process of assessment of each financial institution's capital adequacy by the banking supervision. Very important point is soundness and quality of the bank's management and control mechanisms. It is necessary for bank to have internal processes in place to assess so-called Capital Adequacy Assessment Process (known as CAAP). The absolute minimum of capital adequacy is still 8% of the value of risk-weighted assets. 3. Pillar – Market discipline The third pillar is focused on the issue of transparency and information disclosure. Each bank has to disclose more detailed information about its activities (for example publication made about the methods used to calculate capital adequacy or own approaches to measure and control risks). All banks have to do the core disclosure requirements. The Czech National Bank´s approach Following part of the article is devoted to the Czech National Bank's approach and is worked out namely in terms of information recovering from the Czech National Bank and websites. In the Czech Republic banking regulation and supervision are regulated by the Czech National Bank (known as “CNB”). The CNB defines the prudential framework for banking business1. Banks have to adhere to that framework. Regulations contain the terms and conditions of entry into the banking sector and setting prudential rules for specific areas of banking business. Czech National Bank Banking Supervision checks whether banks are adhering to those rules2. The Czech Banking Association is a voluntary association of legal persons, which do business in banking and in nearly connected areas. One of the objects of its activity is to present and promote the common interest of its members in the Government and the Czech National Bank. The core part of the Czech Banking Association activities consists in an active involvement in the preparation of laws and lower legal provisions, securities, regulating banking supervision, capital market and so on3. The CNB and the Czech Banking Association cooperate with banks to implement the New Basel Capital Accord. New rules are also reflected in the re-casting of Directive 2000/12/EC4 and Directive 93/6/EEC5 (hereinafter referred to as the EC Directive). The Czech Republic – as European Union member

1

Under Act No. 21/1992 Coll. on Banks, the Czech national bank is authorised to issue regulations. For more information see www.cnb.cz. 3 For more information see www.czech-ba.cz. 4 Directive 2006/48/EC of the European Parliament and of the Council of 14 June 2006 relating to the taking up and pursuit of the business of credit institutions (recast version of Directive 2000/12/EC). 5 Directive 2006/49/EC of the European Parliament and of the Council of 14 June 2006 on the capital adequacy of investment firms and credit institutions (recast version of Directive 93/6/EC). 2

18


state – will have to implement. The CNB agrees with the proposed revisions to these Directives and supports and regards, namely, follows the elements1: ♦ Promoting safety and soundness in financial systems and enhancing competitive equality and international comparability among credit institutions and investment firms as a result of introducing the new rules. ♦ Implementation of the new rules on the date proposed in the Directive (January 1, 2007, whereas some advanced approaches will not be applied until January 1, 2008). ♦ Implementation on a solo and consolidated basis for all credit institutions, i.e. banks, electronic money institutions and credit unions, and for investment firms in the Czech Republic. ♦ An individual, more risk-sensitive approach tailored to the institution's risk profile. ♦ Flexibility as regards choice of method and the use of more sophisticated and accurate risk management methods for determining capital requirements. ♦ Timely preparation of the banking sector for implementation of the new rules, on which the CNB and the financial sector will need to work together (see below). ♦ Active co-operation at international level. ♦ An internationally uniform interpretation of the new. CNB Banking Supervision has to prepare for a fundamental change in banking sector regulation in connection with Basel II. This will involve elaborating approaches that are as objective as possible, incorporating them into the Czech laws and regulations, and subsequently applying them in practice. It will also be necessary to establish a uniform interpretation of the EC Directive rules and requirements and to ensure sufficient transparency of procedures, especially where the supervisor has the option of taking an individual approach to banks (such as in risk profile assessment). At the same time, banks must have the opportunity to adapt to the new methods sufficiently in advance. It is therefore vital for the CNB to work with the Czech Banking Association, with individual banks and with the Czech Chamber of Auditors in both the preparatory phase and the implementation phase. These objectives can only be achieved through active co-operation at international level as well. The Czech Republic was involved in the preparation and implementation of Basel II – via its membership of the Core Principles Liaison Group (known as “CPLG”) and the CPLG Working Group on Capital. The CNB is also involved in the work of the relevant EU committees and working groups. In late 2001, the CNB prepared a questionnaire surveying banks' preparedness for introducing the new approaches for credit risk. The CNB has also addressed the issue of training and has organised several seminars on Basel II with internal and external instructors). Now that the final version of Basel II is known and the texts of the EC Directives are mostly final, the Czech National Bank's most important tasks at present are as follows2: ♦ to implement the EC Directives into the Czech legislation and regulations, ♦ to enhance the awareness of both specialists and the general public (through presentations, seminars and publications in the press), ♦ to co-operate actively within the Joint Project of the CNB, the CBA and the CA CR, ♦ to enhance the skills of the Joint Project participants and prepare for the implementation of the new rules, ♦ to participate actively in the QIS5 quantitative impact study organised by the Basel Committee on Banking Supervision in cooperation with the CEBS (the study focuses on impact assessment and recalibration of Basel II and the EC Directive, as the case may be),

1 2

For more information see www.cnb.cz. For more information see www.cnb.cz.


♦

19

to develop cooperation with foreign supervisory authorities and, where possible, to conclude agreements on cooperation in respect of supervision of individual bank groups (especially with regard to "home/host issues" such as data validation and in the area of national discretion).

The time schedule for the implementation of the new rules also places considerable demands on supervisory authorities. Because the new regulations implementing the EC Directive have not been issued yet, banks are not to be permitted to calculate tier capital requirements using the current and future rules in parallel.

Discussion In articles, new or other public origins we can read very often that the New Capital Accord take up and amends the Basel Capital Accord. I take it that it could be relatively misleading formulization for unfamiliar people. The Basel Capital Accord made the first imaginary step towards banking supervisions harmonization. For the first time we could hear of minimal capital adequacy for a credit risk and later for a market risk too. There was only one method of risk measurement methodologies. The New Basel Capital Accord is focused on providing more sensitive and accurate risk measurement. It induces credit financial institutions to enhance their risk management abilities. Unlike Basel II, it gives more comprehensive approach and the flexible options for measuring risks and includes operational risk. For operational risk the first pillar offers a wider range of risk measurement methodologies, including banks´ own internal models, qualifying criteria etc. I think it is necessary to cover up operational risk. But it is very hard to find the way of measurement and control of the risk. The New Basel Capital Accord accordingly includes three pillars. The first pillar contains minimum capital requirements for a credit, market and operational risk. The second pillar covers supervisory review process, and the third one contains market discipline. For the second and third pillars we could not talk of taking up and amending the Basel Capital Accord. Table 2 takes down important changes and basic differences between the Capital Accord and the New Capital Accord. Table 2 Basic differences between the Capital Accord and the New Capital Accord Basel I

Basel II

1. Banking Supervision

Centred on capital adequacy

Three pillars of Banking Supervision

2. Capital requirements

For credit and markets risk

For credit, markets and operational risk

3. Setting capital requirements

One method only

More than one method

4. Risk weights (and height of capital requirements for a credit risk)

Dependence on client type and independence from endured risk

Dependence on riskiness of client (it is inferred from external rating by standardised method, and from internal rating by IRB methods)

5. Risk measurement methodologies – banks´ own internal models and qualifying criteria

By market risk only

By market risk, credit and operational risk too

6. Administrative costs

Low costs

High costs

7. Motivation to risk management quality

Banks are not motivated

Banks are motivated to risk management quality, they can reach for lower capital requirements

20


The table shows there are significant differences between Basel I and Basel II. The New Basel Capital Accord is more emphasis on banks´ own internal methodologies, supervisory review and market discipline. It is possible to tell advantages and disadvantages the New Basel Capital Accord. The CNB and the Czech Banking Association have cooperated to implement Basel II. On May 21 2007, a new framework was published. The new framework comes into force on July 1 2007 and implements the New Capital Accord taking into account the EC Directives. In principle the CNB respects opinion of the Basel Committee on Banking Supervision, or more precisely European Commission. Banks in the Czech Republic knew of the new rules preparation. They had enough information on the basic principles of Basel II and on-coming changes. Banking management has been trained by the CNB. Banks prepared their systems in advance. Because of the facts I think the new framework is not surprise to banks and will not be the cause of their difficulties. In this respect the new framework shall have no radical impact on banks. But on all accounts I have taken the view the Czech National Banking system will be more transparent and safer for depositors, stakeholders and banks too.

Conclusion Basel Capital Accord established a risk measurement framework with a minimum capital standard of 8% and 5 risk classes. This framework proved competent. Because of further development of financial markets, introduction of new financial products and emergence of sophisticated risk management techniques, Basel I leaved off performing its purpose and it was necessary to find out more sufficient framework. Contribution of Basel II is more refined scale of risk classes, multiple options for calculation minimum capital requirements and introduction of a capital charge for operational risk. Table 3 Minimum capital ratio calculation Total capital ≥8% Risk-weighted assets for credit, operational and market risk

Table 3 shows a basic minimum capital ratio calculation. An item “credit risk” is changed by Basel II, “operational risk” item is new, and items “total capital” and “8%” stand similar. The other difference between Basel II and Basel I is three pillars approach for establishment of required capital: Pillar 1 – minimum capital requirements (Calculation of risk weighed assets for credit risk, operational risk and market risk subject to strict minimum requirements); Pillar 2 – supervisory review (A bank-wide, integrated risk governance model has to be introduced, with a more a more comprehensive supervisory review to assess alignment of the bank's capital with its risk profile); Pillar 3 – Market discipline (More extensive disclosure requirements to provide more transparency to stakeholders with respect to a bank's risk profile). The objective of the capital requirements is to have in place a comprehensive and risk-sensitive framework and to foster enhanced risk management amongst financial institutions. This will


21

maximise the effectiveness of the capital rules in ensuring continuing financial stability, maintaining confidence in financial institutions and protecting consumers. The implementation of Basel II will accelerate convergence of supervisory practices. Basel II intended to facilitate the establishment of effective systems of management, especially in the area of credit and operational risk. The CNB agrees with the proposed revisions to these Directives. The new rules should enhance risk management by credit institutions and investment firms and improve their capital coverage. It is possible to expect that the new framework will result in greater market transparency and bolster overall stability in financial markets.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

DVOŘÁK, P. Bankovnictví pro bankéře a klienty. 3. vyd. Prague: Linde, 2005. 681 s. ISBN 80-7201-515-X. FABOZZI, F.J., The handbook of asset/liability management: state-of-the-art investment strategies, risk controls, and regulatory requirements. Rev. ed. Boston: Irwin, c1996. vi, 506 s. ISBN 1557388008. JÍLEK, J. Finanční rizika. 1. vyd. Prague: GRADA Publishing, 2000. 635 s. ISBN 80-7169-579-3. PETRJÁNOŠOVÁ, B. Bankovní management. 1. vyd. Brno: Masaryk University, 2004. 132 s. ISBN 80-210-3481-5. UYEMURA, D.G., Financial risk management in banking: the theory & application of asset & liability management. New York: McGraw-Hill, c1993. xviii, 361 s. ISBN 1557383537. Act No. 6/1993 Coll. of 17 December 1992, on the Czech National bank, as amended by later Acts. Act No. 21/1992 Coll. of 17 December 1991, on Banks, as amended by later Acts. Decree of the Czech National Bank No. 522 of 15 September 2004, amending Decree of the Czech National Bank No. 333/2002 Coll., stipulating the prudential rules of parent undertakings on a consolidated basis. Decree of the Czech National Bank No. 333 Coll. of 3 July 2002, stipulating the prudential rules of parent undertakings on a consolidated basis. Provision of the Czech National Bank No. 2 of 3 July 2002, on capital adequacy of banks and other prudential rules on a solo basis. Directive 2006/48/EC of the European Parliament and of the Council of 14 June 2006 relating to the taking up and pursuit of the business of credit institutions (recast version of Directive 2000/12/EC). Directive 2006/49/EC of the European Parliament and of the Council of 14 June 2006 on the capital adequacy of investment firms and credit institutions (recast version of Directive 93/6/EC). http://www.bis.org (official websites of the Basel Committee on Banking Supervision). http://www.cnb.cz (official websites of the Czech national bank). http://www.czech-ba.cz (official websites of the Czech banking association). http://www.europa.eu.int (portal of the European union).

22


PRIVATE EQUITY INVESTMENTS IN THE BANKING INDUSTRY – THE CASE OF LONE STAR AND KOREA EXCHANGE BANK Matthias Menke*, Dirk Schiereck** Abstract This paper examines success factors for the value creation of a private equity fund investing in a bank – based on Lone Star’s acquisition of Korea Exchange Bank in 2003. Despite the value destruction mergers and acquisitions in the banking industry have shown, Lone Star turned the bank successfully around. Considering regulatory restrictions and limitations to lever a transaction, the private equity fund also capitalized on the recovery of financial markets after the financial crisis in Asia during the late nineties. With regard to these circumstances success factors of the performed value creation are evaluated by the case study approach. Key words: Private Equity, Bank, Acquisition, Value Creation, South Korea. JEL Classification: G14, G21, G24, G34.

Introduction Are private equity funds able to create shareholder value in the banking industry? Considering the research on mergers and acquisitions (M&A) with regard to banking and private equity, the field of (financial) industry-specific private equity investments remains largely unexplored. Existing studies about M&A transactions in the banking industry have shown certain value destruction. Negative value impacts for banks as acquirers question the value-creation potential for other investors, in particular private equity funds. Compared to transactions in other industries, the acquisition of a financial institution has certain particularities. Besides regulatory restrictions and limitations to lever a transaction due to capital adequacy, the generation of cost synergies is challenging. In the case of an acquisition of a bank, public opinion plays also a role in the success of the transaction – especially if the bank is listed on a stock exchange. Furthermore, given the narrow investment period of a fund, the financial sponsor will already need to have factored in measures to prepare a successful exit to reach a required internal rate of return. Taking this background into consideration, we analyze the value creation of a financial buyer through the acquisition of a bank. The analysis is performed by the case study approach. As an immediate reaction to the financial crisis in Asia during the late nineties, which also hit the Republic of Korea (Korea), the Asian country received a support package of USD 57 billion from the International Monetary Fund (IMF) in December 1997. The IMF provided these funds under the condition that structural reforms would have to be initiated in the Korean economy1. As a result of the initiated measures banking business in Korea became attractive for national and international players2. Besides the geographic distinction of acquiring companies, strategic (e.g., Citigroup and Standard Chartered) and financial players (e.g., Lone Star, Newbridge Capital, and The Carlyle Group) entered the Korean market through the acquisitions of banks to participate in the positive economic developments.

* ** 1 2

European Business School, International University, Germany. European Business School, International University, Germany.

See Ahn, Choong Yong (2001), p. 1. “[…] Korea has achieved an annual growth rate of 5.5 per cent during the past five years […].” OECD (2005), p. 11.

© Matthias Menke, Dirk Schiereck, 2007.


23

Performing one of the largest M&A transactions in the Korean market in the aftermath of the 1997 crisis, Lone Star acquired a majority stake in Korea Exchange Bank (KEB) in 2003. The private equity fund intended to sell its stake in 2006. Due to the ongoing allegations by Korean authorities concerning irregularities and potential fraud in connection with Lone Star’s acquisition, the intended sale of Lone Star’s KEB stake to Kookmin Bank had been cancelled by the private equity fund. Despite these allegations, the following case study investigates the value creation by Lone Star1. Also, key factors for a successful acquisition of a bank by a private equity fund – including turnaround measures – are exemplarily discussed. Besides the described developments in Korea and the resulting transactions in the financial industry, there are similar transactions (e.g., in Japan and Germany) where financial investors acquired banks, initiated and implemented turnaround measures, or already exited their investment2. The rest of the paper is organized as follows. A brief overview of the theoretical background and current research will be presented. Taking the regulatory status of the selected transaction into consideration, banks as investment objects as well as economic and banking developments in Korea will be discussed. Key factors of the transaction and the following turnaround measures will be elaborated upon to provide a basis for the empirical analysis. Therefore, the strategic rationale, the execution of the transaction, and post-acquisition measures will be analyzed. The latter factors and the results of the empirical analysis serve as a foundation for the concluding discussion of key factors for a successful acquisition.

Theoretical Background and Current Research The private equity market represents a financial source for enterprises. This organized market can be described as follows: “[…] professionally managed equity investments in [registered and] unregistered securities of private and public companies. An equity investment is any form of security that has an equity participation feature. […]”3. Within its investments, financial investors focus on the acquisition of majority shareholdings which are in late financing rounds. The research about private equity has been intensified in the last decade. Studies about tasks, functions, fundraising, organization, value creation, and performance analysis, or publications concerning financing stages and exits with geographical or industry-specific focus have been performed4. Also statistics about the private equity investments in the financial industry will be shown separately (e.g., European Venture Capital Association). Besides the performed research and statistics, no further breakdown of sub-industries (such as banking industry) will be provided. Worldwide, banks are subject to regulatory conditions which refer to the contribution of services as well as to the associated refinancing of their business. Apart from national laws, international

1 On November 23, 2006, Lone Star announced the cancellation of the purchase and sales agreement between the private equity fund and Kookmin. Shortly after that announcement, the Korean prosecutors revitalized its claim on December 7, 2006, and prepared a lawsuit against Lone Star. Due to the ongoing allegations, no final judgment can be made at this time. Therefore, the selected procedure will be pursued without consideration of the described allegations. See Irvine, Steven (2006), pp. 1-5, and Lone Star (2006), pp. 3-8. 2 Examples for transactions in Japan are engagements of Lone Star in Tokyo Star Bank and First Credit Corp., Cerberus’s investment in Aozora Bank, and Ripplewood’s acquisition of Shinsei Bank. See Securities Data Corporation (SDC) Database, Thomson Financial. 3 Fenn, George. W./Liang, Nellie/Prowse S. (1995), p. 2. This definition of private equity and the market needs to be extended, due to the fact that the targeted bank was listed at the Korean Stock Exchange. 4 Considering research with an industry-specific focus, studies explored different industries: high-technology, biotechnology, software, services, telecommunication and networking, medical equipment, and computer hardware. Exemplarily, the following studies can be named; Armstrong et al. (2005) analyze the relationship between “Venture-Backed Private Equity Valuation and Financial Statement” and explore six of the listed industries. Wu (2001) examines a dataset within the hightechnology industry between 1986 and 1997, focusing on the “Choice between Public and Private Equity Offerings”. Within his study about “The Value Relevance of Financial Statements in Private Equity Markets,” Hand (2004) analyzed US biotechnology firms. Loos (2006) analyses the value creation of financial investors through their investments in Europe and the United States based on a dataset of more than 3,000 leveraged buyout transaction from various industries.

24


sets of rules have been extended in recent years (such as Basel II). For example, the specified 8% minimum capital requirement of Basel II forces banks to determine the amount of regulatory capital for unexpected losses of their uncovered loans in more detail. In addition to the changing regulatory environment, banks face economic concerns and the need for restructuring in light of national or international developments. According to Beitel et al. (2003), technological change, increasing demand of shareholder value by shareholders, currency alignments, as well as globalization and increasing competition can be exemplarily named for these developments. Following these remarks, it is essential for private equity funds, by investing in banks, that the new regulations will be adopted, that the bank has sufficient regulatory capital, and a loan portfolio with a high portion of a qualitative customer base; there should also be the opportunity to settle the bad loan portfolio.

Post-crisis Economic and Regulatory Developments in Korea’s Financial Sector The Korean government outlined the following strategic cornerstones to implement the requested IMF restructuring of the financial sector1: 1) Provision of liquidity support, time-bound deposit guarantees, and intervention in important nonviable institutions to quickly restore the stability of the financial system; 2) Intervention in nonviable institutions, acquisition of non-performing loan portfolios, and usage of government funds for recapitalization as restructuring measures to revitalize the financial system; 3) Adoption of international regulatory and supervisory best practices to strengthen the existing legal framework; 4) Implementation of measures to reduce the dependency of corporate distress and financial institutions exposed to the credits. In order to build up trust and to ensure the operability of the financial systems, the government guaranteed all deposits of financial institutions until end of 2000 and negotiated prolongations for currency debts with foreign banks. In addition to that, foreign investors were also allowed to own commercial bank shares. To adopt best international regulatory practices, the Korean supervision consolidated the independent authority Financial Supervisory Commission (FSC) and the executive agency Financial Supervisory Service (FSS). Both are responsible for granting and revoking banking licenses and are the regulatory authorities for banking and non-banking financial institutions. Since 1998, foreigners have been able to serve as executives for Korean banks. As an additional measure to strengthen the corporate governance of banks, non-executive committees of outside directors have been introduced in banks. In 2000 the chaebol2 companies faced financial difficulties again. Therefore, the regulatory authorities had to revitalize their restructuring activities3. To accelerate changes and economic restructurings, the government asked foreign investors to participate. Due to Korea’s reform efforts – especially in the financial sector where “[…] the restoration of healthy bank balance sheets has strengthened the transmission of monetary easing to the economy”4 – the positive economic development could already be seen in the following years. At the beginning of 2003, the economic outlook for 2003 and 2004 was positive – projecting an output growth of 5.5 to 6%. In addition to these growth expectations, Korea’s Central Bank retained its medium term inflation target of 2.5 to 3.5 %. While inflation was in that expected range in 2002, the policy interest rate could be increased by 0.25% from its record low of 4%. This was the economic environment Lone Star was facing in early 2003 while considering the acquisition of a majority stake in KEB.

1

See Chopra Ajai et al. (2001), p. 36, and Dymski, Gary (2004), pp. 17-19. “A chaebol is a Korean conglomerate where various firms are loosely linked through their shareholders. There is generally no holding structure, at least for the group as a whole.” Delhaise, Phillipe F. (1998), p. 46. 3 By the end of 2000, the Korean government had spent approximately USD 106,482.2 million to support the restructuring of the financial sector. See Ahn, Choong Yong (2001), p. 30. 4 OECD (2003), p. 9. 2


25

Selection of the Transaction The acquisition of 51% of KEB by Lone Star in August 2003 corresponds to a value of USD 1,171.7 million and was one of the largest M&A transactions in the Korean market in the aftermath of the 1997 crisis. The investment in KEB, which had been publicly listed since 1994 and was the fifth largest Korean bank representing 6.1% of Korea’s total bank assets, was also the biggest overseas investment in Korea’s financial industry. The public listing of KEB enables an analysis based on publicly available data. In addition to the deal size, it is one of the landmark transactions in the private equity industry focusing on the acquisition of banks. Annual figures from 2003 to 2005 are available for the analysis of balance sheets as well as profit and loss data. Also, the capital market data will be used for the analysis of the value creation. In accordance with other case studies prepared in the field of banking M&A (e.g., Calomiris and Karceski, 1998), the transaction partners and their motives will be described in an initial step. Following that description, the investigation and turnaround measures will be discussed in detail to serve as a basis for the empirical analysis.

Description of Korea Exchange Bank The following description of KEB covers the economic base, reorganization plans, and organizational issues as well as an overview about the shareholder structure. With total assets of USD 68,604.2 million by the end of 2002, KEB was the fifth largest Korean nationwide bank. At that time, KEB had a common equity of USD 1,180.0 million and a market capitalization of USD 713.7 million. The Asian financial crisis in the late nineties also hit the KEB; a number of bankruptcies of corporate clients resulted in an increase of loan loss provision and subsequently to the disposal of nonperforming assets. As an immediate action, a corporate restructuring program was initiated and led to a reduction of the workforce and a subsidiary restructuring. The management additionally initiated a “five actions plan”1 to prepare KEB for the future: 1) To improve the credit process by the adoption of loan screening techniques – and therefore to reduce the bad debt amount – KEB set up an internal “Bad Bank” division; 2) Measures for the enhancement of transparency and responsibility were introduced. Therefore, the bank reorganized its Board of Directors, enabling nonexecutive directors and working committee members to influence the new corporate governance structure; 3) KEB switched from a function- to a customer-oriented organization through the implementation of a business unit system; 4) The bank set up a new risk management system; 5) The management initiated a project to reorganize the information and technology infrastructure. From 1998 to 2000, several measures to recapitalize the bank had been implemented. As a foreign bank and strategic investor, Commerzbank AG (Commerzbank) participated in these capital measures. The German bank injected capital USD 290.7 million (in 1998), USD 228.1 million (in 1999), and USD 166.0 million (in 2000) and therefore increased its shareholding to 32.55%. KEB’s merger with its subsidiary, Korea International Merchant Bank, in January 1999 led to an increase of KEB’s paid in capital of USD 1,280 million. Following the five-step plan initiated in 1998, KEB presented in 2000 an updated turnaround plan to the FSA and received the approval to continue with its business operations independently. Optimizing the capital base, reducing nonperforming loans, and raising the profitability were the core elements of the restructuring plan approved by the FSA. By the end of 2002, KEB had the vision to become the “First Choice Bank for Customers, Shareholders, and Employees”2. In light of the poor performance in 2002 – net income decreased significantly in 2002 to USD 56.4 million from USD 241.5 million in 2001 – the management focused on the realignment of its business strategies and strengthening internal capabilities of KEB. Therefore, services in retail banking, corporate banking, global banking, foreign ex1 2

See KEB (1999), pp. 3-5. KEB (2003), p. 22.

26


change, trade financing, and risk management, as well as products for its broad national and international private and corporate customer base, emerged from the selected organizational structure. To operate the business, the Executive Committee and its subcommittees led 9 Business Units, 20 Banking Groups, 34 Divisions, and 4 Temporary Divisions (Marketing, Loan, Personnel, Risk Management Steering, and Capital Markets). On December 31, 2002, KEB’s shareholder structure was as follows: Commerzbank (32.55%), Bank of Korea (10.67%), Export-Import Bank of Korea (32.50%), and other shareholders (24.28%). Bank of Korea and Export-Import Bank were government-owned institutions. Commerzbank as well as the Korean government, were dependent on each other. Neither the government nor Commerzbank had a controlling stake that could be divested or alternatively used without the other’s permission or future decision support. Due to the deteriorating capital position of KEB in early 2003 and the unwillingness of KEB’s major shareholders to inject additional capital, they endorsed the active search for an external investor.

Description of Lone Star Until May 2006 Lone Star had invested in almost 50 separate investments in Korea worth USD 5 billion. The private equity fund performed its first engagement in the Asian country in 1998. Before the acquisition of KEB, Lone Star was an active player in the acquisition of nonperforming loan portfolios, real estate investments, and distressed banks (only in Japan) in the Japanese and Korean markets. As a specialized private equity investor with entrepreneurial focus that invests in distressed real estate, distressed debt, distressed companies and distressed banks, Lone Star acquired bad debt portfolios and real estate in Korea (e.g., a loan portfolio of KEB Credit Services Co. Ltd. [KEBCS] after an extensive due diligence). Besides, Korea Lone Star is mainly active in the United States, Canada, Japan and Germany. Until the end of 2005, Lone Star had raised USD 13.25 billion for its six funds since its founding and realized annual returns between 9 and 28% for its first five funds1.

Strategic Rationale and Description for the Transaction As briefly described above, KEB was facing financial challenges in 2003 and the majority of existing shareholders were not willing to invest additional money. At that time Lone Star had analyzed KEB and its subsidiaries for months; it was their intention to turn the bank around, increase organizational efficiency, and pursue the best available exit option following the economic recovery of the bank and the Korean economy. In addition to financial engineering measures were opportunities to increase the efficiency of the existing and invested capital. The governance structure in terms of the board and the shareholding structure, also provided potential for further optimization. “Lone Star was the only realistic potential buyer at that time willing to provide the necessary capital injection of about USD 750 million.”2 On August 27, 2003, Lone Star Fund (LSF) IV signed a Memorandum of Understanding with KEB regarding a Share Subscription Agreement to acquire 51%; a new issue of shares served as a capital injection to recapitalize KEB. Also, as part of the transaction, Commerzbank and Export-Import Bank of Korea sold shares to Lone Star. The right for Lone Star to acquire further shares from the two major shareholders until October 31, 2006, was granted by a call option3. The Korean regulatory authorities approved the transaction as of September 2003, but it was requested by the FSC and the FSS to pre-notify the regulatory authori-

1

See Effinger, Anthony/Yu, Hui-yong (2005), pp. 40-42. Appendix B of Lone Star (2006). 3 On June 2, 2006 Lone Star Fund exercised its option to raise its interest to 64.63% from 50.53%. See SDC Database, Thomson Financial. 2


27

ties in advance of the execution of the options1. Additional key terms of the acquisition were the agreement of a lock-up period for two years after closing, no put back-option of non-performing loans, as well as the right for Lone Star to nominate seven of the ten members of the Board of Directors. After the execution of the transaction, the shareholder structure was as follows: LSF-KEB Holdings (51.00%), Commerzbank (14.75%), Bank of Korea (14.00%), Export-Import Bank of Korea (6.18%), and other shareholders (14.07%). Due to the financial situation of KEB, Lone Star could not expect any dividend payments at short notice, but had to focus on the immediate implementation of turnaround measures.

Post-acquisition Measures Management and Corporate Governance After the acquisition by Lone Star, a new management team, including a new Board of Directors, was implemented within KEB. This team was experienced and operations-focused while defining a core-competency-focused strategy. Immediately after its implementation, the management team initiated corporate restructuring measures. Within a month after the closing (October 30, 2003) of the Lone Star’s KEB acquisition, the Board of Directors had decided to merge KEB with its subsidiary KEBCS. The latter offered a wide range of credit card services. KEBCS was established by a spin-off. KEB’s credit card operation was separated in 1988. The encountered growth path after the public offering in 2001 slowed down sharply in 2002 and early 2003. A shortfall of receivables and sales growth, in combination with credit losses and high delinquency ratios, forced the management to initiate turnaround measures for KEB’s subsidiary2. The merger of KEBCS and KEB became effective on February 28, 2004, and diluted Lone Star’s shareholding to 50.53%. Besides the reduction of complexity and alignment of the governance structure, reduced funding costs – through an improved rating for KEBCS – could be exemplarily named as reasons behind the merger. By taking full control, additionally, the credit card exposure could be stabilized and controlled. The corporate restructuring of KEB also took place in Europe and the United States. Strategic and Operational Developments The new management team adopted immediately global management standards and a new management philosophy to enhance the competitiveness of KEB. Changing the organizational structure as of December 2002, KEB reshuffled its organizational frontline structure in profit centers to the divisions Global Corporate Banking, Retail Banking, and Credit Card. In line with this reshaping was the strategic focus on small and medium enterprises, wealthy clients, as well as customers with deposit accounts and no KEB credit cards in its Credit Card Division. To support these divisions, three operational business groups were set up in May 2004 as cost centers: Credit Management-, Service Delivery-, and Information Technology Group. Since then, Risk-, Financial-, Human Resources Management, and Corporate Communications serve as support functions for the banking business divisions and operational groups. In addition, a new structure performance measurement system was introduced. In order to reduce bad assets, KEB improved its loan quality through credit control processes and improved the customer classification in the Global Corporate Bank division by industry type. In early 2004, KEB initiated measures to improve the cost efficiency. The organizational restructuring of the headquarters was completed in June 2004. Caused by the peaked Information Technology (IT) depreciation charges in 2003, a focus within the initial restructuring phase was on IT spending. Also at that time, a staff realignment and branch remodeling program had been started. In contrast to the stated initiatives, short-term effects could not be expected on the profit-and-loss 1 2

See FSS (2003), p. 4, and KEB (2006), p. 97. See KEB (2006), p. 144, and KEBCS (2003), pp. 26-41.


28

account. In opposition to that, a focus on trading and settlement operations, as well as reengineering banking operations and a reallocation of staff, led to early positive results. Based on this initial success in the first full accounting year of Lone Star’s ownership, the management focus for 2005 was as follows1: 1) Build marketing excellence for targeted customer segments by trade-oriented corporate banking, affluent retail banking, and focus on fixed income and international settlement as well as credit card business; 2) Enhancement of KEB image; 3) Refinement of credit reviews and deal-structuring capabilities; 4) Maximization of capital efficiency and building up of the capital base; 5) Continuous focus on process and cost efficiency (by being not the biggest but the best by implementing smarter, faster, and more efficient procedures); 6) Expansion on global standard Human Resource improvement systems. Balance Sheet Restructuring and Recapitalization Due to dependencies on the Korean chaebol and the economic turmoil as a result of the Asian crisis in the late nineties, KEB faced a large amount of corporate bad debt on its balance sheet. After the initial recovery of the corporate sector starting in 2002, the private household sector lacked liquidity and, therefore, the bad debts in this sector increased as well2. Already initiated measures had been further enforced after Lone Star’s engagement. The continuation of prudent credit management (e.g., Hynix Semiconductor Inc.) and the sale of non-performing loan-portfolios as well as new loans led to the lowest amount of loan loss reserves since the end of the Asian crisis. By the end of 2003, KEB had also restructured its debt by swapping the maturity from short-term to long-term, taking advantage of the low interest level in Korea at that time. Lone Star’s capital injection of approximately USD 900 million covered the losses in 2003 and increased shareholders’ equity. The completion of the merger with KEBCS caused an increase “in debentures amounting to” USD 1,932.1 million “converted from the merger”3. Also, USD 784.4 million of the bank’s capital was used to complete the merger. The measures to restructure the balance sheet revoked an improved Bank for International Settlement (BIS) capital adequacy ratio (BIS Ratio). The highest net income in KEB’s history of USD 2,056.4 million was the primarily driver for the improvement of the BIS ratio by the end of 2005 and also represented the best performance of a nationwide bank in Korea (13.68%). The rating agencies – Moody’s Investor Service (Moody’s), Standard and Poors (S&P), and Fitch Ratings (Fitch) – acknowledged the refinancing efforts by KEB and lifted their ratings in 2005/2006 (Moody’s: Baa2/P-2; S&P: BBB/A-2; Fitch: BBB+/F2). Through the execution of the described measures, KEB became a profitable bank. In order to examine the impact of Lone Star on the shareholder value creation, share-price developments, its balance sheet figures, as well as a comparison with its competitors, an empirical analysis will be performed.

Empirical Analysis Description of the Research Method and Definitions Initially, the short-term and long-term value creation will be analyzed in an event study. Within the following benchmark analysis, it will also be explored if KEB shows a higher value creation than other banks. Also, as part of the empirical analysis the long-term performance of fundamental data will be analyzed. Following the definition of the event, the event window will be defined. The temporal deferment is necessary, in order to determine the expected net yield for an estimation period and for the event window, which includes periods before and after the event. 1

KEB (2005), p. 29. See Dymski, Gary (2004), p. 19. 3 KEB (2005a), p. 34. 2


29

August 27, 2003, is defined as event day (t = 0). Based on t the event window will be expanded to 40 trading days before and 40 trading days after the event day (T = [-40,+40], while t ε T). Through the selection of the trading days before and after t, alternative events distorting the results should be avoided. A further detailed view takes place via the analysis of intervals within the originally selected window from T. To analyze the impact of the announcement that Lone Star acquires a KEB stake, daily stock returns are used (Total Return Index – TRI). Within these returns, dividend payments are considered. They are also adjusted by corporate actions. Market-adjusted models assume a linear relationship between the return of any security and the return of the market portfolio. As this is a commonly used model, the following formula will be used to calculate the expected return on security i in period t (Rit*)1. Rit* = αi + βiRmt + εit

(1)

Thereby, αi and βi are estimated over the defined estimation period by an ordinary least squares (OLS) regression of the respective stock versus the national banking index in Korea (Rm). The underlying assumption for the calculation of formula (1) is a linear relationship between the return of security i and the overall market Rm. Based on the expected return (Rit*) and the return on security i in period t (Rit), the abnormal return (ARit) is calculated as the difference between Rit and Rit* and therefore the value creation. The calculated daily abnormal returns will be summarized to quote the cumulated return for the respective intervals. Short-term Analysis In the following short-term analysis, the described methodology is applied on the defined event day. Table 1 provides an overview of the abnormal returns within the event window for KEB and its national competitors Kookmin Bank and Shinhan Financial Group. Table 1 Short-term abnormal returns of Korean banks Abnormal return in %

Abnormal return in %

Intervals

Korea Exchange Bank

Kookmin Bank

Shinhan Financial Group

[-40;0] [-35;0] [-30;0] [-25;0] [-20;0] [-15;0] [-10;0] [-5;0] [-2;0] [-1;0] [0]

2.33 4.45 10.83 11.53 9.71 10.97 10.19 13.31 -1.66 1.68 3.63

-4.28 -5.44 -6.24 -5.68 -6.37 -4.63 -2.45 -1.73 -1.67 -1.86 -1.17

7.23 6.65 -0.28 -1.86 -5.54 -4.24 -7.92 -7.28 -3.00 -1.62 -0.56

Intervals

Korea Exchange Bank

Kookmin Bank


[-40;+40] [-35;+35] [-30;+30] [-25;+25] [-20;+20] [-15;+15] [-10;+10] [-5;+5] [-2;+2] [-1;+1]

13.00 17.68 23.62 16.67 28.99 15.42 24.42 23.27 14.31 15.36

-9.46 -10.16 -9.85 -7.58 -9.72 -5.42 -3.94 -0.44 -2.78 -3.93

11.99 7.11 5.45 5.53 2.16 5.49 2.16 -6.78 -3.64 -2.55

On the event day, KEB’s stock price increased by more than 3.6%. Also, in all other intervals within the event window – except for two – KEB had the highest value creation. This result corre-

1

Peterson, P. (1989), pp. 39-55.


30

sponds with M&A banking research, where a positive value impact on the acquisition target can be recognized. In addition, the impact of Lone Star as an experienced investor with financial strength also had a positive impact on the value creation. Due to KEB’s described situation, its value creation could be interpreted as a reaction to Lone Star’s investment and the expectation of a successful turnaround. Even though a positive impact of Lone Star’s engagement can be seen, the presented analysis provides only a short-term view, and the long-term value creation needs to be analyzed. Long-term Analysis of Capital Market and Fundamental Data The long-term value creation of KEB’s share-price development is analyzed on daily stock returns in accordance with the short-term analysis. In addition to the presented procedure, extensions to consider the extended time period had to be made: 1) Extension of the time frame to 39 months in advance and 36 months after the months of the announcement of Lone Star’s KEB acquisition; 2) Setup of a benchmark portfolio. Based on these extensions, the long-term abnormal returns are calculated. Thereby value creation in the subsequent three years of t is calculated in the intervals ([0-12 months], [0-24 months], [0-36 months]). For the abnormal returns, which are also the difference between daily and expected returns, αi and βi were determined by an OLS regression. For this regression 36 months, starting three months preceding August 2003, serve as an estimation period. With these results, the expected return is calculated. Cumulated daily abnormal returns are presented in Table 2. It is recognizable that KEB outperformed its local competitors within the first 36 months after the transaction. Table 2 Long-term abnormal returns of Korean banks Abnormal return in % Intervals

Korea Exchange Bank

Kookmin Bank


[0-12 months] [0-24 months] [0-36 months]

34.82 51.00 23.71

-45.97 -74.11 -97.54

8.85 9.60 -2.58

Despite KEB’s success, the presented long-term event study approach does not cover all particularities. Exemplarily deviations in the calculation of long-term abnormal return can be stated. Therefore, the benchmark approach is used to apply an acknowledged research approach. Out of Thomson Financials Datastream Banking Index Asia, five benchmark companies were selected by comparables (market capitalization as measurement for the size and the expected return; market-to-book ratio to evaluate the return by calculating the market value in relation to the bookvalue as measurement for the expected growth). Adopting these criteria, Towa Bank, Fukui Bank, Awa Bank, Metropolitan Bank and Trust Company, and RHB Capital were selected. Due date for the selection criteria was the last year-end balance sheet date (December 31, 2002) before Lone Star’s acquisition. The return for KEB and the five selected banks was calculated for the same intervals like in the long-term event study. By subtraction of the respective average of the five banks from KEB’s performance, the abnormal return of KEB is determined. The results are presented in Table 3.


31

Table 3 Long-term abnormal return calculated with benchmark approach Intervals

Return Korea Exchange Bank in %

Average benchmark in %

Abnormal return Korea Exchange Bank in %

[0-12 months] [0-24 months] [0-36 months]

26.40 103.18 118.16

-7.06 4.57 14.03

33.47 98.61 104.13

In all three intervals KEB outperformed the average benchmark in the respective periods. Due to the adjustments of the cons of the previously discussed approaches, as well as the clear results of the value creation initiated by Lone Star in KEB, the latter approach can be qualified as the best in the context of this paper. To also evaluate the described qualitative measures and the calculated value creation based on share-price developments, the fundamental data is analyzed. Since Lone Star’s engagement, the implementation of financial and operating measures can be acknowledged. Considering an increase of total assets in 2005 after its initial “balance sheet shakeup” with significant reserves for loan losses in 2003 and a decline in 2004, the net loans grew slightly in 2005; in comparison with the increase of net-loans in the years before Lone Star’s engagement, this considerable generation of new credit business can be accounted to the new credit and risk processes. KEB’s balance sheet restructuring as well as an optimized “interest management” caused an increased net interest margin. As one important driver for KEB’s performance improvement, the interest income also increased the retained earnings. This can be acknowledged in Tables 4 and 5. Table 4 Reserves for loan losses and net interest margin of Korean banks Reserves for loan losses as % of Total loans

Net interest margin in %

Banks

2002

2003

2004

2005

2002

2003

2004

2005

Korea Exchange Bank Kookmin Bank Shinhan Financial Group

2.56 2.39 1.71

5.18 2.66 2.88

2.00 2.25 1.95

1.38 1.78 1.67

2.19 3.73 2.36

2.76 3.89 2.18

2.70 3.33 2.52

3.41 3.99 3.22

The growth of the latter earnings increased the return on average equity (RoAE) of KEB significantly. The resulting BIS Ratio of 13.68% by the end of 2005 is the highest a Korean bank has reached within the last years (see Table 5). Table 5 Return on average equity and BIS Ratio of Korean banks RoAE in %

BIS Ratio in %

Banks

2002

2003

2004

2005

2002

2003

2004

2005

Korea Exchange Bank Kookmin Bank Shinhan Financial Group

3.00 13.33 17.83

-42.73 -7.99 6.11

20.41 6.25 14.11

44.84 20.78 18.17

9.31 10.41 10.92

9.32 9.81 10.49

9.47 11.14 11.94

13.68 12.84 12.27

This brief retrospective review of the financial performance also represents clear indications for the success of the turnaround initiated by Lone Star, but focuses on past performance. Therefore,

32


the latter approach can only be an indicative approach to support the calculated value creation by the benchmark approach. A summary will be provided in the following to conclude the key drivers for success.

Determination of Success Factors and Evaluation of the Presented Analysis The presented case study approach provides an in-depth view of the measures and implications initiated and performed by the financial investor. Even without the exit of the US fund, various key factors for the successful turnaround can be named. Lone Star’s profound knowledge of the Korean market, as well as its long-term experience in the financial industry, served as a catalyst for the success of KEB’s turnaround. In addition, the following criteria can be identified as critical success factors for the performed value creation: 1) Macroeconomic environment: The selection of a bank which was located and operating in a country with growth potential after an economic downturn was a key driver for Lone Star’s success. Additionally the low interest rates in Korea and the regulatory reforms within the financial sector enabled Lone Star to implement financing measures and to act in a stabilized economic environment; 2) Target selection, deal preparation, and execution: The extensive knowledge of the target market served as an essential reason for the turnaround. Based on this and the negotiated key deal terms – especially the acquisition of the majority shareholding and the ability to nominate key decision makers – the private equity fund was able to initiate measures for enabling the turnaround; 3) Governance and corporate measures: The adoption of international management practices and the execution of corporate measures, in addition to a stringent credit and risk management, were essential to assign performance goals and to generate synergies of scope; 4) Financial reengineering program: Restructuring the balance sheet on the asset and liability side supported by the interest level was a key in executing the new business strategy; 5) Business strategy: The clearly identified customer orientation was the basis for the extension of business with the existing customer base and the acquisition of new clients. Additionally the focus on KEB’s core competencies, in combination with achieved quality improvement, was an important factor for the success; 6) Human Resource measures: The reshuffling of the employee structure, in combination with and implementation of an incentive system, was the basis for the cultural development. Even though each of the listed factors has significant impact on KEB’s performance, the selected and implemented mix of the measures is the essence of the success of Lone Star’s value creation. The presented research results are based only on a single transaction. To verify these results, a broader data sample should be analyzed. Even if there might be only a limited number of transactions where a private equity fund invested in a regulated bank, the evidence of the results should be supplied. In addition, the conditions of the acquisition can be questioned. On the one hand, this transaction was partially a government privatization and, therefore, represents an ownership change that rarely happens in comparison to banking M&As and private equity investments in other industries. On the other hand, therefore, the selection of competitors and selected benchmarks can be questioned. Finally, a final judgment of the value creation performed by Lone Star cannot be ultimately made due to the exit, which was not performed until the end of November 2006. Additionally, it needs to be considered that Lone Star had negotiated favorable terms in its initial deal to restructure KEB (e.g., board seats), but parts of the initiated and implemented measures required the approval and know-how of the other majority shareholders as well and not only the selected Lone Star representatives.

Conclusion and Outlook The objective of this case study was to analyze the value creation of a private equity fund in the banking industry. The performed analysis clearly states the positive impact which Lone Star’s engagement had for the performance of KEB. Being trapped in a regulated environment and, addi-


33

tionally, continuously monitored by quasi-government shareholders, a significant portion of KEB’s performance can be attributed to Lone Star. The selected approach provides a comprehensive overview of the measures performed by Lone Star and the associated limitations the fund had to deal with. Therefore, it fills the gap of the performed banking M&A and private-equity field research. In order to extend the current research on banking and private equity, the presented case study serves as a valuable addendum in named fields.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

Ahn, Choong Yong. 2001. “Financial and Corporate Sector Reform in South Korea: Towards a New Development Paradigm”. Working Paper Draft prepared for presentation at the MDT Workshop at Tokyo University, Seoul. Armstrong, Chris/Davila, Antonio/Foster, George. 2005. Venture-Backed Private Equity Valuation and Financial Statement Information, Stanford Graduate School of Business, Stanford. Bank for International Settlement. 2006. International Convergence of Capital Measurement and Capital Standards – A Revised Framework, Comprehensive Version, Basel. Bank of Korea. 2006. Annual Report 2005, Seoul. Beitel, Patrick/Schiereck, Dirk/Wahrenburg, Mark. 2003. Explaining the M&A-success in European Bank Mergers and Acquisitions, Oestrich Winkel. Berger, Allen N./Udell, Gregory F. 1998. The Economics of Small Business Finance: The Roles of Private Equity and Debt Markets in the Financial Growth Cycle, in: Journal of Banking & Finance, o. Jg (1998), Vol. 22, pp. 1-52. Brav, Alon/Gompers, Paul A. 1997. Myth or Reality? The Long-Run Underperformance of Initial Public Offerings: Evidence from Venture and Nonventure Capital-Backed Companies, in Journal of Finance, Vol. 52, No. 4. Brown, S./Warner, J. 1980. Measuring security price performance, Journal of Financial Economics, 8, pp. 205-258. Brown, S./Warner, J. 1985. Using Daily Stock Returns - The Case of Event Studies, in: Journal of Financial Economics, Vol. 14, pp. 3-31. Calomiris, C.W. /Karceski, J. 1998. Is the bank merger wave of the 1990s efficient? Lessons learned from nine case studies, The AEI Press, Washington, D.C. Chopra, Ajai/ Kang, Keneth/Karasulu, Meral/Liang, Hong/Ma, Henry/Richards, Anthony. 2001. From Crisis to Recovery in Korea: Strategy, Achievements, and Lessons; IMF Working Paper WP/01/154, Washington, D.C. Cuny, C.J./Talmor, E. 2006. A theory of Private Equity Turnarounds, Working Paper, Washington University in St. Louis, London Business School. Delhaise, Philippe. 1998. Asia crisis: the implosion of the banking and finance systems, Singapore. Dymski, Gary. 2004. Korea’s Post-Crisis Banking Crisis and the Global Bank Merger Wave, Sacramento. Effinger, Anthony/Yu, Hui-yong. 2005. Bad-Loan Power Play; in: Bloomberg Markets, March 2005; New York. European Venture Capital Association (EVCA). 2005. EVCA Yearbook 2005, Annual Survey of Pan-European Private Equity & Venture Capital Activity, Brussels. Fama, Eugene. 1991. Efficient Capital Markets: II, in: The Journal of Finance, Vol. 46, No. 5, pp. 1575-1617. Fenn, George W./Liang, Nellie/Prowse S. 1995. The Economics of the Private Equity Market, Federal Reserve Board, Washington D.C. Fraser-Samson, Guy. 2007. Private equity as an asset class, West Sussex. FSS. 2003. Weekly Newsletter, Vol. IV, No. 34, Seoul. FSS. 2006. BIS Capital Adequacy Ratio of Domestic Banks: End-2005, Press Release March 8, 2006, Seoul.

34


22. Gompers, Paul A. 1995. Optimal Investment Monitoring and the Staging of Venture Capital, in: The Journal of Finance, Vol. L, No. 5, pp. 1461-1489. 23. Hahm, Joonho. 2005. The Korean Model of Corporate Governance: Issues and Lessons in Reform of Bank Governance (Woori Bank); Hills Governance Center at Yonsei University, Working Paper Series No. 05-01, Seoul. 24. Hand, John R.M. 2004. The Value Relevance of Financial Statements in Private Equity Markets, University of North Carolina at Chapel Hill. 25. Irvine, Steven. 2006. Korea’s real winners: local banks not foreign equity. http://www.financeasia.com/article.aspx?CIID=69792 (accessed: December 21, 2006). 26. Kang, Chungwon. 2003. From the Front Lines at Seoul Bank: Restructuring and Reprivatization, IMF Working Paper, Washington, D.C. 27. KEB. 1999. Annual Report 1998, Seoul. 28. KEB. 2001. Annual Report 2000, Seoul. 29. KEB. 2003. Presentation 2002 Year-end Report, Seoul. 30. KEB. 2005. Presentation 2004 Year-end Report, Seoul. 31. KEB. 2005a. Annual Report 2004, Seoul. 32. KEB. 2006. Annual Report 2005, Seoul. 33. KEBCS. 2003. Roadshow Presentation May 2003, Seoul. 34. Kester, Carl/Luehrmann, Timothy. 1995. Leveraged Buyouts – besser als ihr Ruf, in: Harvard Business Manager, 17. Jg. (1995), Heft 4, S. 75-85. 35. Knedlik, Tobias/Stroebel, Johannes. 2006. The role of banking portfolios in the transmission from currency crises to banking crises – potential effects of Basel II, Institut für Wirtschaftsforschung Halle – Discussion Papers, No. 21, Halle. 36. Korea Development Bank. 2006. Korea’s M&A Market and Role of Financial Investors; in: 2nd Quarterly Review, Seoul. 37. Lerner, Joshua/Hardymon, Felda/Leamon, Ann. 2005. Venture Capital and Private Equity. A Casebook, Boston. 38. Lone Star. 2006. John Grayken – Press Conference, New York. 39. Loos, Nicolaus. 2006. Value Creation in Leveraged Buyouts, Analysis of Factors Driving Private Equity Investment Performance, Dissertation Universität St. Gallen. 40. OECD. 2003. OECD Economic Surveys Korea, Volume 2003/5, Paris. 41. OECD. 2005. Economic Surveys Korea, Paris. 42. Peterson, P. 1989. Event Studies: A Review of Issues and Methodology, in: Quarterly Journal of Business and Economics, Issue 28, pp. 36-66. 43. Pilloff, Steven J./Santomero, Anthony M. 1996. The Value of Bank Mergers and Acquisitions; Wharton Working Paper 97-07, Philadelphia. 44. Thomson Financial. Datastream, SDC and Worldscope Database, New York. 45. Van der Geest/van Frederikslust. 2001. Initial Returns and Long-Run Performance of Private Equity – Backed Initial Public Offerings on the Amsterdam Stock Exchange, Working Paper for EFMA 2001 Lugano Meetings, Rotterdam. 46. Wu, Ylin. 2000. The Choice between Public and Private Offerings, Washington University, St. Louis.


35

AN EMPIRICAL INVESTIGATION OF BANK RISK-TAKING IN EMERGING MARKETS WITHIN A PROSPECT THEORY FRAMEWORK. A NOTE1 Christophe J. Godlewski* Abstract The purpose of this note is to investigate the validity of some behavioral conjectures as alternative explanations of bank risk-taking behavior. We especially focus on the different valuation of gains and losses relative to a reference point and the changing attitude toward risk conditional on the domain (gains vs. losses) features (Tversky and Kahneman, 1992). We follow a methodology based on Fiegenbaum and Thomas (1988) and the Fishburn (1977) measure of risk, applied to a sample of banks from emerging market economies. Results show that the Tversky and Kahneman (1992) framework could provide an alternative explanation of risk-taking behavior in the banking industry. Key words: Cumulative Prospect Theory, bank risk taking, emerging market economies. JEL Classification: C12, C31, D81, F39, G21.

1. Introduction In order to investigate the deviations of agents from traditional finance models, relying on perfect information and coherent beliefs, behavioral finance models based on cognitive psychology propose specific features of agents' behavior, relaxing the individual rationality hypothesis (Shleifer, 2000; Barberis and Thaler, 2002). Another crucial feature of a model that aims at explaining trading behavior for example is the hypothesis made on investors’ preferences and the way they evaluate risky choices. Prospect Theory is one of such theories, due to Kahneman and Tversky (1979) and Tversky and Kahneman (1992). It is the most successful one because of its capacity to capture and fit results obtained in laboratory experiments. Its starting point is a critique of the expected utility theory as a descriptive model of decision making under risk. Following experimental results, agents usually under-weight the probable results compared to certain one (certainty effect), which implies risk aversion when gains are certain and risk loving when losses are certain. Agents also exhibit a lack of coherence in their preferences when the same choice is differently presented (isolation effect and framing). The Prospect Theory's formulation provides several important features: 1) utility is defined on the gains and losses, and not on the final wealth value, 2) the evaluation function form, particularly its concavity in the gains domain – agents are risk-averse on the gains and risk loving on the losses – with a kink at the origin showing a greater sensibility to losses (loss aversion), 3) non-linear transformation of probabilities: small one are overestimated and agents are more sensible to differences of probabilities at higher probability's levels. The principals of judgment and perception are possible thanks to the use of the value function. The value is treated as a function of two elements: the asset's value as a reference point and the amplitude of changing from this starting point.

* 1

Louis Pasteur University, LaRGE, Strasbourg, France.

I would like to thank Marie-Hélène Broihanne for valuable discussions and suggestions. I also thank the participants of the AFFI International Conference 2005, Paris-La Défense, France, the Global Finance Conference 2005, Trinity College, Dublin, Ireland and the Campus for Finance 2006 Conference, Koblenz, Germany. The usual disclaimer applies. © Christophe J. Godlewski, 2007.

36


Prospect Theory features can be applied to study investors behavior (like insufficient or naive diversification or excessive trading) (see Odean, 1998; Odean, 1999; Barber and Odean, 2000, Barber and Thaler, 2001). It has also applications in corporate finance. For instance, Wiseman and Gomez-Mejia (1998) build a behavioral managerial risk taking agency model, through the linkage of corporate governance mechanisms and prospect theory features (especially framing). Their main contribution concerns an alternative risk formulation compared to the agency theory, based on loss aversion and not risk aversion. In a behavioral framework, preferences will be unstable due to the framing feature, contrary to the agency theory that assumes constant preferences. The same choice can be presented in the potential gains or losses domain, altering traditional agency theory results. In this framework, changing the performance benchmark for the manager affects its reference point (translating the gains and losses domains), and therefore may adversely alter its risk taking behavior. In addition, the use of compensation mix in order to establish proper incentives for the manager, aligned with the principal interests, may also adversely affect agent's risk taking behavior in such a framework. The application of behavioral finance features to investigate risk-taking in the banking industry is of central interest in this paper. As far as we know, this area has received scarce attention from the behavioral finance perspective1, although risk-taking remains the core activity of banks. It has been proven that excessive risk taking2 is the principal bank default factor (see for example Pantalone and Platt, 1987; and O.C.C., 1988). The last 20 years have witnessed several bank failures throughout the world, particularly in emerging market economies (EME) (Bell and Pain, 2000). The interest for bank failures comes mainly from its costs: financial losses for the stakeholders (shareholders, clients, and deposits insurance fund), loss of competitiveness, and a potential destabilization of the financial system, through the contagion mechanisms, when several individual failures lead to a banking crisis. The resolution of these failures is a waste of resources, particularly scarce in EME (Honohan, 1997)3. Several explanations of the excessive risk taking sources can be found in the literature4: inefficient corporate governance mechanisms (Gorton and Rosen, 1995; Knopf and Teall, 1996; Simpson and Gleason, 1999; Anderson and Fraser, 2000), inadequate bank capital regulation (see Jeitschko and Jeung, 2005), intense market competition (Keeley, 1990; Cordella and Yeyati, 2002; Repullo, 2004; Boyd and Nicolo, 2005), and an adverse regulatory, institutional and legal environment (Barth et al., 1999, 2000, 2001, 2002; La Porta et al., 1997, 1998, 2000). Alternative explanations of excessive risk taking in banks seem neglected. As risk-taking decisions are made upon human subjective judgment and especially perception of risk, it seems quite natural to engage in the behavioral perspective to better investigate and understand this process. Effectively, the final decisions concerning credit approvals and loan terms are based on many different attributes, from which experience and the judgment of the credit staff continue to play a significant role (Crouhy et al., 2001). Bowman (1980, 1982) findings are of particular interest in this perspective because they provide the basis of the so-called risk-return paradox. The prospect theory's feature stipulates that risk attitude is determined by the outcome's relation to a reference point and not the outcome's level. Therefore, some testable hypotheses are provided by Kahneman and Tversky (1979) framework: when performance is below a given target level, decision makers should be risk seeking, and when performance is above the target level, they should be risk-averse.

1

Shen and Chih (2005) empirically investigate earnings management in banks within a prospect theory framework. This can be defined as a level of risk-taking that amplifies the bank’s probability of default above an acceptable level by the different partners of the bank, especially the shareholders and the regulator. 3 For example, the banking crises in Indonesia (1997) and Thailand (1997-1998) cost about 50-55% and 42.3% of the GDP (fiscal contribution) respectively in terms of restructurization. 4 See Godlewski (2006) for a survey. 2


37

Fiegenbaum and Thomas (1988) tested these predictions using accounting data, defining benchmark returns as median returns, and dividing the firms of their sample in two groups – above and below target. Their results strongly corroborated the presented prospect theory predictions. Jegers (1991) replicates Fiegenbaum and Thomas (1988)'s methodology using Belgian accounting data, testing some new return and risk variables, like ROA (return on assets) in addition to ROE (return on equity), which should take into account managerial performance view, and cash flow on equity, and a coefficient of variation, defined as the standard deviation of returns divided by the average return, in addition to the variance of returns. Jegers (1991) calculates each firm's time average return, ranks firms according to these values, and divides the firms into 2 equally sized groups: those with above and respectively below target returns, the target being defined as the median return. Then, for each group, Spearman rank correlations between return and risk and the negative association ratio are calculated. The results corroborate those of Fiegenbaum and Thomas (1988). Finally, Johnson (1994) also places his analysis of risk-taking in banks in a behavioral finance framework, following Fiegenbaum and Thomas (1988), and using Fishburn (1977) measure of risk, defined as dispersion about the mean outcome. Johnson (1994) tests several measures of return and risk for a sample of US commercial banks for the 1970-1989 period. He uses standard measures of return like ROA and ROE, as well as primary capital ratio. Risk is measured as standard deviation of outcome. The study aims at examining historical data to determine whether there is any evidence consistent with prospect theory, by measuring the relationship between outcome variability and distance from target. Targets are defined as the median values of return variables. Banks are classified in two separate groups according to this target, and correlation between distance to target and standard deviations are computed. The statistical tests are based on Kendall τ correlation coefficient. The obtained results also corroborate Fiegenbaum and Thomas (1988) conclusions. Following this literature, we aim to empirically investigate risk taking in the banking industry in emerging markets in a Cumulative Prospect Theory framework. We focus on the banking industry in a specific framework – emerging market economies – where risk-taking behavior can become adverse, generating excessive risks and therefore amplifying bank's default probability, thus affecting negatively the whole economy. The specificities of these economies, mainly historical heritage (political, economic, social, moral, …), restructuring process in progress, rapidly evolving economic reality, inadequate regulatory, institutional and legal environment, may foster excessive risk taking, affecting the perception of risk by the bankers. For example, an evolving economic environment forces the banker to constantly adapt his appreciation of risk. An inadequate institutional or legal environment may bias banker's risk perception. Therefore, an investigation of this risk perception in a behavioral finance framework is important. The rest of the note is organized as follows. Section 2 describes the methodology and the data used in this study. Section 3 presents the results and their discussion. Finally, section 4 concludes and proposes further research perspectives.

2. Methodology and data In the present study, we follow Johnson (1994)'s methodology for the formalization of the tested hypothesis in order to provide empirical evidence dealing with bank risk taking based on prospect theory features. We use a pooled sample of 894 commercial banks for the 1996-2001 period from two main areas of emerging market economies – South-East Asia and South and Latin America (see Table 1). The accounting data come from Bankscope.


38

Table 1 Banks' in sample frequency by country Country

Banks

Frequency

Argentina

151

16.89

Bolivia

23

2.57

Colombia

104

11.63

Ecuador

63

7.05

Indonesia

68

7.61

Korea (South)

33

3.69

Mexico

95

10.63

Malaysia

82

9.17

Peru

100

11.19

Thailand

54

6.04

Venezuela

121

13.54

894

100

Source: Bankscope.

We calculate several return and risk measures, following the existing literature, but also trying to propose some alternative measures. The definition of the variables used in this study and their descriptive statistics are provided in Table 2. Table 2 Variables definition and descriptive statistics Variable

Calculation

M.

Med.

S.D.

Min.

Max.

ROE

Net Income/Equity

-2,97

7,86

193,16

-4864,15

2057,90

ROA

Net Income/Total Assets

0,26

0,73

5,84

-112,21

23,66

EQTA

Equity/Total Assets

10,80

9,86

8,19

-120,92

53,45

SPREAD1

Interest Income/Total Loans

26,31

21,71

16,53

4,57

162,68

SPREAD2

Interest Income/Total Operating Income

11,62

8,50

10,07

1,53

111,01

NPLGL

Non Performing Loans/Gross Loans

11,09

7,40

11,50

0,00

89,59

LLRNPL

Loan Losses Reserves/Non Performing Loans

98,81

70,24

104,45

3,18

846,15

LLRGL

Loan Losses Reserves/Gross Loans

6,71

5,16

6,17

0,00

60,24

NLTA

Net Loans/Total Assets

57,04

57,41

13,69

25,38

92,35

M.: mean, Med.: median, S.D.: standard deviation, Min.: minimum, Max.: maximum.

Concerning the return measures, we use “traditional ones”, like the ROE (reflecting rather the shareholder point of view), the ROA (reflecting rather the management point of view) and the EQTA (reflecting the shareholder, the management and the regulator points of view). We also use SPREAD1 and SPREAD2 measures that focus more precisely on the bank's credit activity and should give a more adequate perspective on return in commercial banks. Concerning the risk measures, apart from the standard deviations of the return variables discussed above, we also investigate the usefulness of standard deviations of the “loss measures” mainly NPLGL (reflecting a potential loss for the bank), LLRNPL, LLRGL (both reflecting management's


39

perception of risk and its coverage with reserves which alter the profitability of the bank) and NLTA (which reflects both potential future returns but also potential problems in term of reserves and/or losses). We also investigate the framing issue, testing the correlations between risk and return measures in different domains – gains versus losses. Therefore, we test the significance of the correlation coefficient between measures of return and risk crossing the domains (gains and losses). The Kendall τ correlation coefficient measures the strength of the relationship between two variables, and like Spearman's rank correlation, is carried out on the ranks of the data. It ranges from +1 to -1, with a positive correlation indicating that the ranks of both variables increase together, whilst a negative correlation indicates that the rank of one variable increases the other one decreases. Its main advantage is the possibility for direct interpretation of the statistic in terms of probabilities of observing concordant or discordant pairs. Our tests rely on time average and their standard deviations measures, as well as median of these variables. The medians of the employed measures represent the target levels – the reference points for the bank. We work with 9 zones which are: Zone 1 – ROE, Zone 2 – ROA, Zone 3 – EQTA, Zone 4 – SPREAD1, Zone 5 – SPREAD2, Zone 6 – NPLGL, Zone 7 – LLRNPL, Zone 8 – LLRGL, Zone 9 – NLTA.

3. Results and discussion The Fishburn's measures of risk are the distance of the variable from the target level. For each zone, we split the sample in 2 areas: ABOVE and BELOW, corresponding respectively to banks above and below the target level – the median of the variable corresponding to the zone. In Tables 3 and 4, we compute Kendall τ correlation coefficients between the standard deviation of the variable and the distance to the target level corresponding to the zone and by area. Table 3 Correlations results between standard deviation and distance to benchmark measures (gain domains) Area

Zone 1

Zone 2

Zone 3

Zone 4

Zone 5

(ROE) ABOVE

-0.0851**

(ROA)

(EQTA)

(SPREAD1)

(SPREAD2)

-0.0962***

-0.0418

-0.1706***

BELOW

0.1675

0.1772*

-0.1498**

0.0115

0.0464

-0.0357

Kendall τ correlation coefficients between the standard deviation and the distance to median are shown for each zone, by area. ***, ** and *: statistically significant at 1%, 5% and 10% levels respectively.

Concerning the correlation results in the gain domains for the Zones 1-5, we observe significant and negative Kendall τ coefficients for each zone (except Zone 3 corresponding to the EQTA variable) in the ABOVE area. We can interpret these results in the following way: for banks located above the target level in the gains domain, bankers exhibit a risk averse behavior, as the standard deviation and the distance to median are negatively correlated. It may correspond to a “defensive attitude”, as being above the target in terms of outcome implies preserving the privileged position, and therefore exhibiting risk aversion. For banks located below the target level, the relationship between these 2 measures is not significant.


40

Table 4 Correlations results between standard deviation and distance to benchmark measures (loss domains) Area

Zone 6

Zone 7

Zone 8

Zone 9

(NPLGL)

(LLRNPL)

(LLRGL)

(NLTA)

ABOVE

-0.1182*

-0.0635*

-0.0824

-0.0996***

BELOW

0.0045

-0.1734

-0.028

0.0513

Kendall τ correlation coefficients between the standard deviation and the distance to median are shown for each zone, by area. *** and *: statistically significant at 1% and 10% levels respectively.

Concerning the correlation results in the loss domains for the Zones 6-9, we observe mixed evidence. In the ABOVE area, except for the Zone 9, corresponding to the NLTA variable, other Kendall coefficients are weakly significant and negative, the coefficient being not significant for Zone 8 (LLRGL). For banks above the target levels in terms of potential losses (NPLGL) or their coverage (LLRNPL), bankers exhibit a risk aversion behavior. Having, for example, a level of NPLGL above the target level implies a more risk averse attitude, as these potential losses may drive the bank into default. The Kendall correlation coefficients for the BELOW area are all not significant. In Tables 5 and 6 we propose to cross the domains (gains vs. losses) in order to investigate the framing issue which is one of the crucial feature of Prospect Theory. The same choice may be presented in alternative ways (as a gain versus as a loss), affecting the editing phase of an agent, and therefore affecting its preferences. We do this in the following manner: in Table 5 we compute Kendall correlation coefficients between standard deviations of gain measures (ROE, ROA, SPREAD1, SPREAD2) and distance to median losses measures (NPLGL, LLRNPL, LLRGL, NLTA, corresponding to the Zones 6-9). In Table 6, we invert the measures, showing Kendall coefficients between standard deviations of loss measures and distance to median gains measures (Zones 1-5). Table 5 Correlations results between standard deviation and distance to benchmark measures (cross gain vs. loss domains)

SDROE

Zone 6

Zone 7

Zone 8

Zone 9

(NPLGL)

(LLRNPL)

(LLRGL)

(NLTA)

ABOVE

BELOW

ABOVE

BELOW

ABOVE

BELOW

ABOVE

BELOW

0.0859

0.0901***

-0.0294

-0.0758

0.0748

0.089**

-0.0811**

-0.0478 -0.079

SDROA

0.0977

0.0919***

-0.0287

-0.0963

0.085

0.0958**

-0.0835**

SDSPREAD1

0.0979

0.0955***

-0.0181

-0.0881

0.0843

0.1037***

-0.0779**

-0.16

SDSPREAD2

0.0784

0.0788***

-0.0209

-0.1167

0.0573

0.0956**

-0.075**

-0.0257

Kendall τ correlation coefficients between the standard deviation and the distance to median are shown for each zone, by area. ***, ** and *: statistically significant at 1%, 5% and 10% levels respectively.

Concerning the results shown in Table 5, we observe significant Kendall τ correlation coefficients only for the BELOW areas for Zone 6 and Zone 8, and for the ABOVE area for Zone 9. The results for the BELOW areas seem to indicate that banks located below target levels in terms of potential losses (NPLGL) and their (costly) coverage (LLRGL) exhibit risk loving behavior, as the relationships between the distance to median and standard deviations of return measures is significantly positive. Being under such target “leaves room” for aggressive risk taking within the bank.


41

As to the ABOVE results, we observe significantly negative Kendall coefficients between the distance to target in terms of NLTA and the standard deviation of the return measures. This may be interpreted as a feature of risk aversion on the side of the banker, as being above a target level of loans volume compared to total assets restrain the risk taking attitude materialized in terms of standard deviations of return variables. This volume of loans represents potential revenues but may also transforms into NPL, enhancing the bank's risk of default, contrary to NPLGL or LLRGL variables, which are proxies of ex post excessive risk taking, already materialized. Table 6 Correlations results between standard deviation and distance to benchmark measures (cross loss vs. gain domains) Zone 1

Zone 2

Zone 3

Zone 4

Zone 5

(ROE)

(ROA)

(EQTA)

(SPREAD1)

(SPREAD2)

A.

B.

A.

B.

A.

B.

A.

B.

A.

B.

SDNPLGL

-0.0511

0.1455

-0.069**

0.0214

-0.1321**

0.0371

-0.2177***

0.0101

-0.1931***

-0.0183

SDLLRNPL

-0.05

0.1195

-0.067**

-0.0013

-0.1118*

0.035

-0.2075***

0.0007

-0.1823***

-0.0272

SDLLRGL

-0.0545

0.1221

-0.0738**

0.0013

-0.1264**

0.0338

-0.2108***

0.0103

-0.1877***

-0.0187

SDNLTA

-0.0585*

0.1169

-0.0771**

0.0013

0.1165*

0.0359

-0.2135***

0.0004

-0.1789***

-0.0145

Kendall τ correlation coefficients between the standard deviation and the distance to median are shown for each zone, by area. ***, ** and *: statistically significant at 1%, 5% and 10% levels respectively. A.: ABOVE, B.: BELOW.

Turning to the interpretation of the results in Table 6, we observe significant negative Kendall τ correlation coefficients only for the ABOVE areas for Zones 2-5. Concerning the Zone 2, corresponding to the ROA, we can interpret these results as indicating risk averse behavior rather on the management side, as the relationship between the distance to the ROA target and the standard deviation of losses measures is negative1. Concerning the Zone 3, corresponding to the EQTA variable, we also observe significantly negative Kendall τ correlation coefficients between the distance to EQTA target and standard deviations of losses measures (except for the standard deviation of NLTA). We can interpret this result in a similar manner as for the Zone 2, except that it may reflect in this case the shareholders point of view, as they are the main contributor to the bank's equity. Banks located above the EQTA target exhibit risk avert behavior, as the distance to this target is a proxy measure of the equity cushion or franchise value, which expected loss seems to discipline the risk taking behavior. Finally, concerning the results for the Zones 4 and 5, we also observe significantly negative Kendall correlation coefficients between the distances to SPREAD1 and SPREAD2 target levels and the standard deviations of loss measures. This may also be interpreted as risk adverse behavior feature, as being located above such level implies a more prudent and conservative risk taking behavior. Overall, we observe that in a loss framework, being below a target level seems to affect bank risk taking in a risk loving fashion. On the contrary, in a gain framework, being above a target level has a significant impact on risk taking, in a rather risk averse fashion.

4. Conclusion Cumulative Prospect Theory provides an alternative framework for risk taking analysis, especially excessive risk taking in banks, which remains the major determinant of their failure. Although, the literature dealing with these issues remains scarce.

1

This relationship is not significant for the Zone 1 (ROE – shareholder point of view).


42

This note provides an empirical insight into the investigation of the usefulness of the behavioral framework for risk-taking analysis in banks from emerging market economies. The results tend to support the usefulness and pertinence of the Cumulative Prospect Theory features as alternative explanation for risk taking behavior within banks. Banks located above target level (measured in several different ways) tend to exhibit risk adverse behavior. However, further investigation is needed in order to better understand the behavioral finance contribution to risk taking analysis in banks. First, other benchmark variables should be tested (for example mean or maximum values, as well as benchmark values calculated for best agency rated banks). Second, it would be interesting to apply tournament and ranks theories (Brown et al., 1996; Busse, 2001; Taylor, 2003) especially to investigate rating grades effect on risk taking behavior, and the quantification of rating's default probability, in order to test the probabilities' deformation with an adequate methodology – a crucial prospect theory feature.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

Anderson, R.C., and D.R. Fraser, 2000, Corporate control, bank risk taking and the health of the banking industry, Journal of Banking and Finance 24, 1383-1398. Barber, B., and T. Odean, 2000, Trading is hazardous to your health: The common stock performance of individual investors, Journal of Finance 55, 773-806. Barber, B., and Odean T., 2001, Boys will be boys: Gender, overconfidence, and common stock investment, Quarterly Journal of Economics 141, 261-292. Barberis, N., and R.H. Thaler, 2002, A survey of behavioral finance, Working Paper Series 9222 NBER. Barth, J.R., G. Caprio, and R. Levine, 1999, Financial regulation and performance: Crosscountry evidence, Working Paper 2037 World Bank. Barth, J.R., G. Caprio, and R. Levine, 2000, Banking systems around the globe: Do regulation and ownership affect performance and stability?, Working Paper 2325 World Bank. Barth, J.R., G. Caprio, and R. Levine, 2001, The regulation and supervision of banks around the world: A new database, Working Paper 2588 World Bank. Barth, J.R., D.E. Nolle, T. Phumiwasana, and G. Yago, 2002, A cross-country analysis of the bank supervisory framework and bank performance, Economic and Policy Analysis Working Paper 2 Office of the Comptroller of the Currency. Bell, J., and D. Pain, 2000, Leading indicator models of banking crises – a critical review, Financial Stability Review 12 Bank of England. Bowman, E.H., 1980, A risk/return paradox for strategic management, Sloan Management Review 21, 17-31. Bowman, E.H., 1982, Risk seeking by troubled firms, Sloan Management Review 23, 33-42. Boyd, J., and G.D. Nicolo, 2005, The Theory of Bank Risk Taking and Competition Revisited, Journal of Finance 60, 1329-1343. Brown, K.C., W.V. Harlow, and L.T. Stark, 1996, Of tournaments and temptations: An analysis of managerial incentives in the mutual fund industry, Journal of Finance 51, 85-110. Busse, J.A., 2001, Another look at mutual fund tournaments, Journal of Financial and Quantitative Analysis 36, 53-73. Cordella, T., and E. Levy Yeyati, 2002, Financial opening, deposit insurance, and risk in a model of banking competition, European Economic Review 46, 471-485. Crouhy, M., D. Galai, and R. Mark, 2001, Prototype Risk Rating System, Journal of Banking and Finance 25, 47-95. Fiegenbaum, A., and H. Thomas, 1988, Attitudes toward risk and the risk-return paradox prospect theory explanations, Academy of Management Journal 73, 337-363. Fishburn, P.C., 1977, Mean-risk analysis with risk associated with below-target returns, American Economic Review 67, 126-166.


43

19. Godlewski, C.J., 2006, Regulatory and Institutional Determinants of Credit Risk Taking and Bank’s Default in Emerging Market Economies: A Two Step Approach, Journal of Emerging Market Finance 5 (2), 183-206. 20. Gorton, G., and R. Rosen, 1995, Corporate control, portfolio choice and the decline of banking, Journal of Finance 50, 1377-1420. 21. Honohan, P., 1997, Banking system failures in developing and transition countries : Diagnosis and predictions, Working Paper 39 Bank for International Settlements. 22. Jegers, M., 1991, Prospect theory and the risk-return relation: Some Belgian evidence, Academy of Management Journal 34, 215-225. 23. Jeitschko, T., and S. Jeung, 2005, Incentives for Risk-Taking in Banking: A Unified Approach, Journal of Banking and Finance 29, 759-777. 24. Johnson, H.J., 1994, Prospect theory in the commercial banking industry, Journal of Financial and Strategic Decisions 7, 73-89. 25. Kahneman, D., and A. Tversky, 1979, Prospect theory: An analysis of decision under risk, Econometrica 47, 263-291. 26. Keeley, M.C., 1990, Deposit insurance, risk, and market power in banking, American Economic Review 80, 1183-1200. 27. Knopf, J.D., and J.L. Teall, 1996, Risk taking behavior in the us thrift industry: Ownership structure and regulatory changes, Journal of Banking and Finance 20, 1329-1350. 28. La Porta, R., F. Lopez de Silanes, and A. Shleifer, 1998, Law and finance, Journal of Political Economy 106, 1113-1155. 29. La Porta, R., F. Lopez de Silanes, and A. Shleifer, and R.W. Vishny, 1997, Legal determinants of external finance, Journal of Finance 52, 1130-1150. 30. La Porta, R., F. Lopez de Silanes, and A. Shleifer, 2000, Investor protection and corporate governance, Journal of Financial Economics 58, 3-27. 31. O.C.C., 1988, Bank failure an evaluation of the factors contributing to the failure of national banks, Discussion paper Office of the Comptroller of the Currency. 32. Odean, T., 1998, Are investors reluctant to realize their losses ?, Journal of Finance 53, 17751798. 33. Odean, T., 1999, Do investors trade too much ?, American Economic Review 89, 1279-1298. 34. Pantalone, C.C., and M.B. Platt, 1987, Predicting commercial bank failures since deregulation, New England Economic Review, pp. 37-47. 35. Repullo, R., 2004, Capital Requirements, Market Power, and Risk Taking in Banking, Journal of Financial Intermediation 13, 156-182. 36. Shen, C.-H. and H.-L. Chih, 2005, Investor Protection, Prospect Theory, and Earnings Management: An International Comparison of the Banking Industry, Journal of Banking and Finance 29, 2675-2697. 37. Shleifer, A., 2000, Inefficient Markets: An Introduction to Behavioral Finance (Oxford University Press). 38. Simpson,W., and A. Gleason, 1999, Board Structure, Ownership, and Financial Distress in Banking Firms, International Review of Economics and Finance 8, 281-292. 39. Taylor, J.D., 2003, Risk taking behavior in mutual fund tournaments, Journal of Economic Behavior and Organization 50, 373-383. 40. Tversky, A., and D. Kahneman, 1992, Advances in prospect theory: Cumulative representation of uncertainty, Journal of Risk and Uncertainty 5, 297-323. 41. Wiseman, R.M., and L.R. Gomez-Mejia, 1998, A behavioral agency model of managerial risk taking, Academy of Management Review 23, 133-153.


44

FINANCIAL LIBERALIZATION AND CURRENCY CRISES: THE CASE OF TURKEY Mete Feridun* Abstract This article aims at identifying the determinants of currency crises in Turkey in the period of 1980:01-2006:06. A broad set of explanatory variables was tested through the signals approach and bivariate and multivariate logit regressions. The same procedure is then repeated for the postcapital account liberalization period (1989.09-2006:06). The results obtained are novel and deviate widely from the existing literature. The findings suggest that conventional crisis indicators fail to provide a satisfactory explanation for the crises experienced in Turkey. For the period spanning 1980:01-2006:06, banking sector fragility index, short-term debt/international reserves, bank reserves/bank assets, US GDP, M1, and US 3-month T-Bill rate have been identified as significant leading indicators by both the signals approach and logit regressions. Analyzing the post-capital account liberalization period spanning 1989:09-2006:06 in isolation, strong evidence is obtained confirming the importance of US federal funds rate, banking sector fragility index, US GDP, and US 3-month T-Bill rate by both approaches. Overall, the results confirm the significance of global economic conditions, and suggest that financial liberalization has rendered the Turkish economy vulnerable to currency crises. Key words: Currency crises, logit regression, signals approach, Turkey. JEL classification: F31, F37.

1. Introduction Turkey is one of the primary examples of the emerging economies that bought into the promises of the IMF-prompted financial liberalization policies in 1980s. The first phase of the liberalization process in the country was initiated in the early 1980s with the deregulation of the interest rates on bank deposits. The second phase of the financial liberalization process was completed in the late 1980s when all the restrictions on capital movements were lifted leading to a period characterized by financial openness and subsequent speculative attacks on the Turkish lira. It has been widely argued that financial liberalization was to blame for the currency crises that the country experienced during this period (see, for example, Yeldan, 1998; Alper, 2001; Erugrul and Selcuk, 2001; Ekinci, 2002; and Seyidoglu, 2003). Nevertheless, there exists no rigorous empirical attempt to support this assertion in the literature on Turkish currency crises, which contains only a handful of empirical studies. These studies extend only over the post-capital account liberalization period and focus on the crises of 1994 and 2000-2001, paying less attention to the periods of unsuccessful speculative attacks on the Turkish lira and they ignore the imperative distinction between pre- and post-capital account liberalization periods. In the light of this motivation, the present study investigates the root causes of currency crises in Turkey in two separate sample periods representing both the entire liberalization era (1980:012006:06) and the post-capital account liberalization episode (1989:09-2006:06). Another novelty of the present research is that it contemplates the episode of currency crunch that the country recently experienced in May 2006, which has not received any empirical attention to date. The literature on currency crises is full of numerous and futile empirical efforts aiming at devising a successful Early Warning System (EWS) to predict future crises through monitoring the behavior of certain variables. It is now widely accepted that it is not possible to predict crises reliably because, particularly after the liberalization of capital flows, currency crises have been increasingly arising

*

© Mete Feridun, 2007.

Loughborough University, United Kingdom.


45

from self-fulfilling panic or pure contagion effects, i.e. inherently unpredictable market sentiments. Even if a successful EWS model can finally be devised, the signals identified by this model would presumably affect the behavior of both the policymakers and the financial market participants, which would quickly render the model obsolete. Therefore, it is more feasible to direct efforts towards the identification of weaknesses that typically render economies vulnerable to crises. For this reason, the present article departs from the existing literature in that it does not concern itself with the prediction of crises. Identification of the factors indicating the vulnerability of an economy to currency crises is essential for the design of effective strategies to avoid future currency crises and to strengthen the macroeconomic structure. The rest of the article is structured as follows: The next section will introduce the data and explain the two methodologies that we will use. Section III and IV will present the empirical results obtained from the analysis of two sample periods. Section V will point out the conclusions that emerge from the study.

2. Data and Methodology 2.1. Data A considerable number of variables can be considered as indicators of vulnerability to currency crises. Essentially, the choice of which variables to select depends on the perceived causes of crises as well as on the variables suggested by the earlier studies in the theoretical and empirical literature on currency crises. Accordingly, we selected our variables to proxy the conditions of current account, capital account, financial sector, real sector, fiscal sector, the global economy, and the domestic political setting. The data is monthly and spans the period between 1980:01 and 2006:06 unless mentioned otherwise. As Goldstein et al. (2000) explain, particularly in the context of signals approach, monthly data allows us to learn much more about the timing of the leading indicators, including differences among indicators in the first arrival and persistence of signals. Nonetheless, some variables were available only in annual or quarterly frequency. Following the existing empirical literature on crises (see, for example, Kumar et al 2003), these series have been interpolated using cubic spline technique from annual and quarterly data1. In order to enhance the possibility of identifying the crisis factors, the present study employs forty-two variables from various sources such as International Financial Statistics Database of the IMF, the Central Bank of the Republic of Turkey’s Electronic Data Delivery System, World Bank’s World Development Indicators Database, European Central Bank’s Statistical Warehouse, US Federal Reserve Board Database and the Global Development Finance Statistics Database. Appendix I presents the list of potential pre-crisis indicators considered, provides justification to their selection, and indicates the sources of the data. A disadvantage of using high frequency data is the possible presence of seasonal effects. This problem is circumvented by using the data in 12-month percentage changes based on the suggestion of, inter alia, Eliasson and Kreuter (2001) and Jacobs et al. (2005)2. This practice eliminates seasonal effects, avoiding the possible non-stationarity problem of the variables in levels, and renders the indicators more comparable across time (Goldstein et al., 2000). Crisis Definition A growing body of studies uses a weighted index consisting of exchange rates, interest rates and reserves which was first introduced by Eichengreen et al. (1995). These studies either adopt ex1

Kumar et al. (2003) argue that such interpolation is appropriate as a monthly observation from an interpolated annual series is based in part on the realization for the year (quarter) in which the crisis occurs and, at any given moment, we possess interim estimates of the annual (quarterly) data over the coming year. The current analysis does not lag any series so interpolation appears a reasonable approach. Still, we checked whether this has affected our results by repeating our estimations and calculations lagging the interpolated annual data by 12 months. The parameter values and t-statistics only slightly changed. 2 This filter has not been used for the real effective exchange rate overvaluation, excess of real M1 balances, and the interest rates.

46


actly the same index or use modified versions of it (See, for example, Herrera and Garcia, 1999; Kibritcioglu et al., 1998; Krkoska, 2000; Gelos and Sahay, 2001; Cepni and Kose, 2006; and Walter, 2006). In the present study, we will adopt a modified version of this index to take both successful and unsuccessful attacks on the Turkish lira into consideration. As Kaminsky et al. (1998) explains, such an index can be used to analyze speculative attacks under both fixed and flexible exchange rate regimes. Therefore, application of this crisis index in the present context is well justified as Turkey followed both fixed and a floating exchange rate regimes in the sample period. During the period under study, Turkish Central Bank followed various exchange rate policies: Prior to 1984, a fixed exchange rate was in effect. Between 1984 and 1993 the exchange rate changed daily in the context of a crawling peg exchange rate regime. After 1993, a managed-floating exchange rate regime was used until January 2000 when the country signed a stand-by agreement with the IMF. With this agreement a pre-announced crawling-peg against a dollar-German mark basket was adopted. However, this peg did not last long and collapsed with the currency crisis of February 2001. Since then, the Central Bank has been using a flexible exchange rate system. Therefore, we use an Exchange Market Pressure (EMP) index which would be applicable in the context of all exchange rate regimes. Accordingly, a currency crisis is assumed to occur when a speculative attack on the Turkish lira results in an official devaluation, or sharp depreciation of the currency, or forces the authorities to defend the currency by expending large volumes of international reserves or by sharply raising interest rates. Eichengreen et al.’s (1995) Exchange Market Pressure (EMP) index is chosen particularly because it is a model-independent, weighted index and it takes into consideration a reference country. Hence, it is more informative than the other variants of the EMP index in the literature. The weights attached to the three components of the index, which are the inverse of the standard deviation for each component, equalize the volatilities of the three components and prevents the component with the highest volatility dominating the index. The choice of which exchange rate to use in the index is somewhat arbitrary. We depart from the existing crisis literature in that we use a Deutsche mark 1 and US dollars basket, which are the two prevalent foreign currencies in Turkey2. The EMP index is calculated as follows: EMPt = αΔet + βΔ(it - i*t) - γ(Δτ,t-Δr*t),

(1)

where α, β and γ are weights that equalize the conditional volatilities of each component. More specifically, α=(1/σe), β=(1/σi), and γ=(1/σr) where σe is the standard deviation of et, σi is the standard deviation of (it - i*t) and σr is the standard deviation of (Δτ,t-Δr*t). Δeit is the monthly change in the Deutsche mark-US dollar exchange rate basket, i denotes the domestic interest rate (3-month deposit rate), i*t corresponds to the same variable but for the country of reference (US prime loan rate). Following Girton and Roper (1977), τ,t denotes the ratio of foreign reserves (net of gold) to domestic money (M1) for the domestic country, and r*t denotes the same concept for the country of reference, i.e. United States3,4. The higher the standard deviation, the lower weight would be imposed on the corresponding variable. A positive value of the index measures the depreciation pressure of the currency that can be signaled by a nominal depreciation, a widening of the interest rate spread, or a loss of foreign reserves, whereas a negative value of the index measures the appreciation pressure of the currency.

1 In January 1999, the euro was introduced and completely replaced Deutsche mark in December 2001. For the sake of consistency we consider Deutsche mark for the whole period under study and use the official fixed parity (1 euro = 1.95583 Deutsche mark) to recalculate exchange rates for the period 1999:01-2006:06. 2 We weighted both series by 0.5 following Kipici and Kesriyeli (1997) who weighted these two currencies equally in an effort to calculate an index of real effective exchange rates for Turkey. 3 In a time of capital inflow reversal, the central bank must be prepared to cover all its liquid liabilities with reserves unless the fixed exchange rate policy is abandoned. The appropriate yardstick with which to evaluate the abundance of reserves is a measure of liquidity compared with the stock of foreign exchange reserves, since, at a time of currency crisis, the larger the stock of privately held domestic liquid assets, the larger the contingent demand for foreign assets (Calvo, 1998; Karfakis and Moschos, 2004). Thus the ratio of foreign reserves to M1 is used as a liquidity measure. 4 This definition of crisis is a major departure of the present study from the existing literature on Turkish currency crises.


47

A currency crisis is considered to occur when the EMP index exceeds a certain threshold value. We identify months in which the index of speculative pressure is at least 1.5 standard deviations above the sample mean as instances of speculative attacks, i.e. currency crises. The value of 1.5 is used following Eichengreen et al. (1996) and Herrera and Garcia (1999) as it gives the best estimation of crises1. The threshold value is determined as: Threshold value = μEMP + 1.5(σEMP) = 1.0763,

(2)

where μEMP is the mean of the index, and σEMP is the standard deviation of the index. A currency crisis is observed when the value of the EMP index exceeds this threshold value. Accordingly, a dummy variable is introduced to take the value of 1 if a crisis occurs and 0, if otherwise. Nevertheless, to avoid counting the same crisis more than once, we set our “exclusion window”2 as 12 months. In other words, 12 successive months immediately after the crisis take the value of 0 regardless of whether the value is above the threshold or not3. In light of these considerations, the crisis months that have been identified by the index are 1980:02, 1981:05, 1983:07, 1985:01, 1991:02, 1994:02, 1995:12, 2001:02, and 2006:06. Figure 1 shows the graphical representation of the estimated EMP index and the threshold. 5 4 3 2 1 0 -1 -2 80 82 84 86 88 90 92 94 96 98 00 02 04 06 EMP Index

Threshold

Fig. 1. EMP index

3. Methodology Based on the identified crisis episodes, we will investigate the determinants of currency crises in Turkey using signals approach and logit regressions. 3.1. Signals Approach The signal approach is a non-parametric methodology introduced by Kaminsky et al. (1998). It involves monitoring the evolution of a number of economic indicators that show a behavior which is different in tranquil period and prior to a crisis. When an indicator exceeds or falls below its 1 We tested different thresholds and found that a higher threshold misses the currency crunch of May 2006 whereas a lower threshold leads to too many crisis episodes. 2 Depending on the frequency of data used, exclusion windows of various lengths have been used in the literature. For instance, Moreno (1995) used a 5-month, Eichengreen et al., (1994) used a 6-month, Glick and Moreno (1999) used a 12month, and Frankel and Rose (1996) used a 3-year exclusion windows. Based on the inspection of the frequency of the crisis months identified by our threshold (particularly during early 1980s), we opt for a 12-month exclusion window. 3 Accordingly, we did not count these observations as currency crises: 1981:08, 1981:11, 1982:05, 1984:01, 1991:03, 1994:03, 1994:04, 2001:03, 2001:04.

48


own pre-determined threshold within a given period, this is interpreted as a crisis signal. This period is defined as the signaling horizon, or crisis window. In the literature, the crisis window spans from 6 months to 24 months1. In the present analysis, we defined it as 12 months in light of the number of observations and the frequency of identified crisis episodes. If an individual threshold is set too loose, the indicator will catch all the crises, but will also issue too many false signals, i.e. noise. If the threshold is set too tight, the indicator will not issue any false signals, but it may as well miss the crises. The outcome for each indicator can be considered in terms of a two by two matrix as shown in Table 1. Table 1 Crises-Signals Matrix Crisis within 12 months

No crisis within 12 months

Signal issued

A

B

No signal issues

C

D

Source: Kaminsky et al. (1998).

For each variable, there are four possible categories A, B, C, and D. A is the number of months a good signal was sent (a crisis is correctly signaled), B is the number of months a false alarm signal was sent, C is the number of months in which no signal was sent but a crises followed, and D is the number of months in which no signal was sent and no crises followed. Any fluctuations of an indicator beyond its pre-determined threshold are considered abnormal and are taken as a signal that a crisis could occur in the next 12 months. An optimal signal is the one that is followed by a crisis within this signaling horizon. The threshold level for each variable is chosen to minimize the noise-to-signal ratio (NSR) which is the ratio of false signals to good signals and is calculated as: NSR = [B/(B+D)]/[A/(A+C)].

(3)

A signaling device that issues signals at random times would obtain an NSR equal to unity. Hence, those indicators which produce more false alarms than good signals, i.e. those having an NSR of above unity, are not helpful in predicting crises (El-Shazly, 2002). For each of the indicators, a two-step procedure is used to obtain the optimal set of thresholds: First, thresholds are defined in relation to percentiles of the distribution of observations of the indicator. Second, a grid of a reference percentile is considered and the optimal set of thresholds is defined as the one that minimizes the NSR ratio. In order to determine the variable-specific optimal threshold values, one of the two grids of reference percentiles between 70% and 95%, or 5% and 30% of the distribution are employed depending on the expected impact of the variable. For some variables a decline in the indicator increases the probability of a crisis, hence the threshold is below the mean of the indicator. For other variables the opposite is the case. The information about the expected impact of each variable on crisis likelihood is given in Table 2. Following the establishment of the relevance of the chosen set of variables as the leading indicators of currency crises by the signals approach, we will test the validity of the functional relationship between the dichotomous variable of currency crises using an econometric analysis since the signals approach ignores the interaction among variables, which may obscure the actual reasons for crises. Having established the crisis index as a binary variable, we will use a limited dependent model. Compared to probit models, logit models typically perform better when the dependent variable is

1 For instance, El-Shazly (2002) uses a 6-month, Brüggemann and Linne (2002) use a 18-month, and Kaminsky et al. (1998) use a 24-month signaling horizon.


49

not evenly distributed between the two outcomes (Manesse et al., 2003). As in the data only 30% of all outcomes are crisis entries, we opt for a logit model1. 3.2. Logit Model Logit models resolve some of the disadvantages associated with the signals approach. For instance, indicators are not transformed into dummies. So, information on the relative importance of each indicator is retained. Besides, regression results are easily interpreted as the probability of a currency crisis. Traditional econometric modeling suggests that we estimate models with numerous explanatory variables and successively eliminate variables with relatively low t-statistics (Kumar et al., 2003). Nevertheless, owing to the large number of explanatory variables2, we will successively eliminate the candidate variables by applying a general-to-specific model selection methodology suggested by Manasse et al. (2003), Linne and Bruggemann (2002), and Krznar (2004). Before moving to a multivariate framework, individual logit models with two variables will be estimated to test the possibility of any functional form between the crisis index and the contemporaneous values of the explanatory variables. Variables that are significant at 10% with correct signs are selected into the final model. Normally, a crisis model should consist of the variables in lagged form. However, it is difficult to decide on the appropriate lag length of monthly variables and is cumbersome to test all possible lags with the large number of variables considered. This issue is circumvented in the literature by, inter alia, Berg et al. (1999), Busssiere and Fratzcher (2002), Komilainen and Lukkarilla (2003)3, and Krznar (2004) by employing a certain crisis-window in which all values of the crisis index take the value of 1. Following these authors, the present study adopts a 12-month crisis-window spanning the year before each particular crisis episode since potential explanatory variables are expected to worsen prior to crises. This allows the use of data without any lags and increases the number of ones in the sample from a statistical standpoint (Krznar, 2004)4.

4. Empirical Results for the Period of 1980:01-2006:06 4.1. Results of Signals Approach Table 2 presents the estimated thresholds and the results ranking the potential early warning indicators of currency crises according to their NSRs. Since NSR is a measure of the relative proportion of false signals to correct signals, closeness of values to zero indicates a high quality leading indicator of crises. We also calculate the unconditional crisis probability P(C) and the conditional crisis probability P(C|S)5. The conditional probability should display higher scores than the simple probability of crisis if the indicator has useful information (Kaminsky et al., 1998). From the estimates reported in Table 2, it is clear that the set of indicators for which the conditional probability

1 In order to compare the sensitivity of the results to the assumption of a logit versus a probit model, we also estimated the regressions using the latter. The estimated parameters from a logit require scaling before they are comparable to those obtained from a probit estimation. Maddala (1983) suggests multiplying the logit parameters by √3/π while Amemiya (1981) suggests that one multiply by 0.625. We found that the latter scaling factor produced closer results. In general, the scaled estimated parameters were broadly similar, especially when the parameters in question had large t-statistics. The fact that the parameters were not closer reflects the fact that the crisis events we are modeling are in the tail of the distribution i.e. there are muck more non-crisis periods than crisis periods, so the fat-tailed nature of the logistic distribution affected the results. The results of the probit models are available from the author upon request. 2 A logit model cannot accommodate all 42 variables simultaneously as a large number of independent variables in the model would increase the probability of linear dependence between individual independent variables, i.e. multicollinearity (Krznar, 2004). 3 Crisis index of Komulainen and Lukkarilla (2003) takes on the value of 1 in the month of crisis itself and the preceding 11 months. However, including the crisis month values of the explanatory variables may bias the results. Hence the crisis index in the present analysis does not take value of 1 in the crisis month and in the immediate aftermath. 4 The values of the currency crisis index are equal to 1 not in the month of the crisis but in the preceding 12 months because including the crisis months’ values would bias the results. This way, explanatory variables used in estimations will have leading indicator characteristics. 5 P(C) is calculated as (A+C)/(A+B+C+D). P(C|S) is calculated as A/(A+B). A, B, C, and D represent the cells in the matrix in Table 2.


50

of a crisis is lower than the unconditional probability is the same as the set for which the NSR is higher than unity. Table 2

Average Lead time of the first signal (in months)

Persistence of signals

P(Crisis/ Signal)P(crisis) (A+C)/(A+B+C+D)

88 210

0.22

0.67

0.31

4.55

9.0

4

55

153

0.31

0.56

0.28

3.23

7.0

Short Term Debt/International Reserves

+

95

7

6

89 209

0.38

0.54

0.31

2.63

6.2

Bank reserves/bank assets

-

5

14

14

46 146

0.38

0.50

0.27

2.63

3.2

Current Account Balance/GDP

-

95

4

4

92 214

0.44

0.50

0.31

2.27

4.0

FDI/GDP

-

5

2

2

94 218

0.44

0.50

0.30

2.27

8.0

US GDP

-

90

7

7

89 208

0.45

0.50

0.31

2.22

6.4

Spread between US 3-month T-Bill and Federal Funds Rate

+

85

12

12

84 198

0.46

0.50

0.31

2.17

5.6

M1

+

85

11

12

85 199

0.50

0.48

0.31

2.00

11.0

US Real T-Bill Rate

+

95

8

9

88 205

0.50

0.47

0.31

2.00

4.0

Federal Funds Rate

+

85

6

7

90 209

0.52

0.46

0.31

1.92

8.7

Foreign Liabilities/Foreign Assets of Banks

+

95

17

20

79 185

0.55

0.46

0.32

M2/International Reserves

+

80

13

16

83 193

0.57

0.45

0.31

1.82

9.7

NSR

4

5

D

8

90

C

5

+

B

-

USD LIBOR Rate

A

Banking Sector Fragility Index

Variable

Threshold (in percentile)

P(Crisis/ Signal) A/(A+B)

Expected Impact on Crisis likelihood

Signals Approach (1980:01-2006:06)

International Reserves/GDP

-

25

33

40

63 149

0.62

0.45

0.34

1.75

6.8

Capital Inflows/GDP

+

85

15

20

81 187

0.62

0.43

0.32

1.61

9.4

Public Debt/GDP

+

90

15

21

81 186

0.65

0.42

0.32

1.61

10.0

Fiscal Balance/GDP

-

10

11

16

85 195

0.66

0.41

0.31

1.54

3.3

Imports

+

95

27

38

69 157

0.69

0.42

0.33

1.52

9.7

Domestic Credit/GDP

+

70

17

26

79 179

0.72

0.40

0.32

1.45

8.3

Industrial Production Index

-

15

12

19

84 191

0.72

0.39

0.31

1.39

10.0

Commercial Bank Loans to Private Sector

+

95

16

26

80 180

0.76

0.38

0.32

1.39

4.5

Commercial Bank Loans to Public Sector

+

80

24

38

72 160

0.77

0.39

0.33

1.32

9.3

Exports

-

20

26

41

70 155

0.77

0.39

0.33

1.30

9.8

M2 Multiplier

+

85

20

33

76 169

0.78

0.38

0.32

1.30

10.8

Reserve Money/GDP

-

20

30

51

66 141

0.85

0.37

0.33

1.28

11.7

Portfolio investments/GDP

-

5

5

12

55 157

0.85

0.29

0.26

1.18

8.1

Government consumption/GDP

+

80

3

6

93 213

0.88

0.33

0.30

1.18

9.3

Excess real M1 balances

+

90

24

44

72 154

0.89

0.35

0.33

1.14

8.3

GDP per capita

-

5

16

33

80 173

0.96

0.33

0.32

1.12

9.0

Commercial Bank Deposits

-

25

44

78

52 100

0.96

0.36

0.35

1.04

6.4

Deposit money banks net past due loans/total loans

+

95

6

18

54 162

1.00

0.25

0.25

1.04

10.6

CPI Growth

+

85

18

44

78 160

1.15

0.29

0.32

1.00

6.3

Short-Term Debt/Long-Term Debt

+

75

3

10

57 161

1.17

0.23

0.26

0.87

10.5


51




90 199

1.26

0.26

0.31

0.85

36

84

174

1.37

0.25

0.31

0.79

8.0

Contagion Dummy

+

N/A

3

10

93 209

1.46

0.23

0.30

0.73

9.3

Real Effective Exchange Rate Overvaluation

+

85

8

30

88 184

1.68

0.21

0.31

0.68

8.9

Domestic Real Interest Rates

+

95

6

23

90 193

1.70

0.21

0.31

0.60

11.3

Stock Market Index

-

12

2

16

22

96

1.71

0.11

0.18

0.59

5.7

Real Interest Rate Differential

+

95

4

20

92 198

2.20

0.17

0.31

0.58

4.0

Central Bank Credit to Public Sector/GDP

+

93

1

8

59 165

2.77

0.11

0.26

0.45

3.3

Trade Balance/GDP

-

5

2

19

94 201

4.15

0.10

0.30

0.36

2.0

NSR

17

12

D

6

90

C

N/A

+

B

+

Oil prices

A

Government Changes

Variable




Table 2 (continued)

6.0

Another desirable feature in the potential leading indicators is that signals be more persistent prior to crises during the 12-month window than at other times. Table 2 presents a summary measure of the persistence of the signals measured as the average number of signals per period during the precrisis period compared to tranquil times. Indicators are ranked according to their performance. The indicator issuing the most persistent signals is the same as the indicator that has the lowest NSR. The opposite is the case for the indicator issuing the least persistent signals. A drawback of the signal’s approach is that, in focusing on the 12-month window prior to the onset of the crisis, the criteria for ranking the indicators do not distinguish between an indicator that sends signals well before the crisis occurs and one that signals only when the crisis is imminent. In order to evaluate the performances of indicators, one should also consider the average number of months prior to crisis the first good signal occurs because a variable with lower NSR can be a useful leading indicator of currency crises only if it sends warning signals sufficiently early to enable policymakers to take preemptive measures to prevent approaching crises. Therefore, in addition to the ranking of the indicators according to their ability to predict crises, lead time of the signal is also estimated. Table 2 also presents the average number of months in advance of the crisis when the first signal occurs. On average, first ten indicators send the first signal 6.4 months before the crisis erupts, with M1 having the longest lead time and bank reserves/bank assets having the shortest. Overall, it can be concluded that the identified leading indicators are indeed leading as signaling, on average, occurs sufficiently early to allow for preemptive policy actions. Overall, the results suggest that only a handful of variables may be considered to consistently provide information about vulnerability to a currency crisis in the sense that they correctly signal crises with negligible noise, and also provide signals early enough enabling policy-makers to take preventive measures. The variables which can provide some useful information about the risks of a possible crisis are banking sector fragility index, USD LIBOR rate, short-term debt/international reserves, bank reserves/bank assets, current account balance/GDP, FDI/GDP, US GDP, the spread between 3-month US T-Bill and federal funds rate, M1, and US real T-Bill rate. Overall, results of the signals approach for the period of 1980:01-2006:06 suggest that indicators related to real sector and fiscal sector variables are not useful as leading indicators for crises. The same conclusion applies to current account variables with the exception of current account balance/GDP. On the other hand, capital account variables and the variables reflecting the global economic conditions are generally found to be functional as leading indicators. Above all, financial


52

sector variables, especially those indicating the fragility of the banking sector, are found to be the foremost indicators of currency crises for the period that we studied. 4.2. Results Logit Regressions Results of bivariate logit models which investigate the possibility of functional forms between the dichotomous crisis index and the contemporaneous values of the individual explanatory variables are presented in Table 3. Positive values of each coefficient imply that increasing the variable will increase the probability of the crises while negative values imply the opposite. The size of each estimated coefficient reflects the relative effect of the variable on the predicted probability for crises. Nonetheless, interpretation of the coefficient values is complicated by the fact that estimated coefficients from a binary dependent model cannot be interpreted as the marginal effect on the probability of crises. Hence, marginal effects of the significant explanatory variables, which we compile into a general logit model, are estimated by taking the derivatives of the parameter estimates. Results of the variable-by-variable logit regressions show that 3-month US real T-Bill rate, US GDP, foreign liabilities/foreign assets of banks, bank reserves/bank assets, M2/international reserves, banking sector fragility index, short-term debt/international reserves, M1, and federal funds rate are significant at 10% level. The signs of the estimated coefficients coincide with what we expect from economic theory. Based on the results of the variable-by-variable analysis, we combine those variables that appear to help predict crises into a general logit model1. Table 3 Coefficient Estimates of the Logit Models with Two Variables (1980:01-2006:06) Variable


Logit Coefficient

Standard Errors

Z-Statistic

P>|z|


+

-0.666895

0.999015

-0.667553

US Real T-Bill Rate

+

0.008402

0.002782

3.019994***

0.5044 0.0025

Fiscal Balance/GDP

-

0.020389

0.672398

0.030323

0.9758

GDP per capita

-

-2.070894

1.601406

-1.293172

0.1960

US GDP

-

-0.322746

0.057608

-5.602442***

0.0000


+

2.514881

3.425531

0.734158

0.4629


+

0.146872

1.969395

0.074577

0.9406


-

-1.151832

1.206750

-0.954491

0.3398

M2 Multiplier

+

1.225859

2.789007

0.439532

0.6603


+

0.513308

0.763213

3.672562***

0.0050


-

-0.239688

2.190382

-3.109428***

0.0029

Imports

+

0.927825

0.793768

1.168887

0.2424


-

-3.859221

3.347362

-1.152914

0.2489

Exports

-

0.503813

1.003442

0.502085

0.6156


+

0.010684

0.002837

3.766075***

0.0002


+

8.922337

5.767289

3.547059***

0.0018


+

1.078855

1.159989

0.930056

0.3523

Capital Inflows/GDP

+

-0.000425

0.031489

-0.013500

0.9892

Reserve Money/GDP

-

-3.821631

2.349475

-1.626589

0.1038

Domestic Credit/GDP

+

-1.236772

3.307736

-0.373903

0.7085

CPI Growth

+

-0.417301

1.140233

-0.365979

0.7144

1 Before moving to multivariate analysis we checked the selected series for multicollinerarity. We did not find evidence of strong correlation between any series.


53

Table 3 (countinued) Expected Impact on Crisis likelihood

Variable

Logit Coefficient

Standard Errors

Z-Statistic

P>|z|


+

-3.80E-05

0.000149

-0.254824

0.7989


+

0.606612

0.411827

3.472979***

0.0078


-

0.010293

0.021844

0.471207

0.6375


+

-0.006289

0.768977

-0.008178

0.9935


+

-0.010884

0.003120

-3.489023

0.0005


-

-0.000126

0.003399

-0.037191

0.9703 0.6071


+

-0.022013

0.042810

-0.514195


+

-0.103147

0.137315

-0.751170

0.4526


-

3.470851

5.229210

0.663743

0.5069

Trade Balance/GDP

-

3.19E-05

0.000346

0.092174

0.9266

Stock Market Index

-

-0.685296

1.319430

-0.519388

0.6035

Public Debt/GDP

+

-0.064192

1.052700

-0.060979

0.9514

Real Interest Rates

+

-0.022013

0.042810

-0.514195

0.6071

Government Changes

+

-0.602633

0.657268

-0.916876

0.3592

Oil prices

+

-0.296546

1.485923

-0.199570

0.8418

M1

+

0.009452

0.005096

1.854778*

0.0636

FDI/GDP

-

0.005685

0.033729

0.168546

0.8662

Federal Funds Rate

+

0.008960

0.002764

3.241148***

0.0012

USD LIBOR Rate

+

-0.004510

0.003646

-1.236853

0.2161

Spread between 3-month US T-Bill and Federal Funds Rate

+

0.000171

0.000428

0.399943

0.6892

Contagion Dummy

+

-1.297566

1.067719

-1.215269

0.2243

* Significant at the 10% level, ** Significant at the 5% level, *** Significant at the 1% level. Table 4 presents the results of the final logit model. The results indicate that the movements in the explanatory variables are correlated with the incidence of currency crises in the expected manner. The statistical characteristics of the model are favorable. All the variables are significant at 10% level. The LR measure confirms the general statistical significance of the model. Hypothesis of no significance of all the coefficients in the model was rejected with significance at 1% level. In addition, McFadden R-squared indicates fairly good goodness-of-fit for the model. Table 4 Coefficient Estimates of the Logit Models with Multiple Variables (1980:01-2006:06) Variable


Logit Coefficient

Standard Errors

Z-Statistic

P>|z|

Marginal Effecta

3-month US Real T-Bill Rate

+

0.008613

0.003646

2.362595**

0.0181

0.365876

US GDP

-

-0.326416

0.059787

-5.459626***

0.0000

-0.067065


+

0.010634

0.003151

3.374403***

0.0007

0.001954


-

-13.16992

7.280977

-1.808812*

0.0705

-2.775423


+

0.023775

0.005654

4.204567***

0.0000

0.003878


54

Table 4 (continued) Variable


Logit Coefficient

Standard Errors

Z-Statistic

P>|z|

Marginal Effecta


+

32.68563

11.12350

2.938432***

0.0033

5.556543

Short Term Debt/International Reserves1

+

3.133247

0.887642

3.529852***

0.0004

0.554764

M1

+

0.040817

0.009460

4.314826***

0.0000

0.007324

Federal Funds Rate

+

0.038007

0.018335

2.072873**

0.0382

0.008001

-1.555654

1.079058

-1.441679

0.1494

Constant McFadden R-squared2: 0.765001 LR statistic (5 df)3: 140.3485***

Significant at the 10% level. *** Significant at the 1% level. a Marginal effects are calculated at sample means.

A comparison of the results obtained from the logit analysis with those obtained from the signals approach indicates that the variables identified by both approaches as leading indicators of crises do coincide. In a sense, this also serves as a confirmation of the robustness of the results obtained by each approach. Next, we will repeat our analysis for the post-capital account liberalization period to see if the results would change.

5. Empirical Results for the Post-Capital Account Liberalization Period (1989:09-2006:06) Liberalization of capital flows has exposed economies to speculative short-term capital movements and rendered them vulnerable to currency crises (Grabel, 1995). Hence, particularly in the postcapital account liberalization period, global liquidity conditions and financial flows are very likely to influence the vulnerability of the economy to currency crises (Kumar et al, 2003). Turkish economy is by no means an exception. Although Turkish financial liberalization efforts root back to early 1980s, the full capital liberalization was declared in August 1989 with the liberalization of the capital account. Since our sample period includes both pre- and post-capital account liberalization periods, following Komulainen and Lukkarilla (2003), we will analyze the post-capital account liberalization period (1989:09-2006:06) separately in order to investigate whether the capital account liberalization has changed the causes of currency crises in Turkey4. 5.1. Results of Signals Approach As evident from Table 5, the signals analysis indicates that variables related to the global liquidity conditions and the US monetary policy are indeed useful leading indicators of currency crises for the post-capital account liberalization period. These variables are US federal funds rate, US GDP, 1 To check that the short-term debt to reserves ratio is not significant simply because the denominator diminishes before a crisis, we tested the level of short-term debt measured as a percentage of GDP. The result is the same: this variable is significant and contributes to the goodness-of-fit of the model. 2 McFadden R2 is a measure of the goodness-of-fit of a model that is obtained when the ratio of the log of the function maximum with a restriction on parameters (all parameters equal zero) and the log of the probability function maximum without the restriction regarding the parameters are deducted from one; it corresponds to R2 as a measure of goodness-offit of models estimated by OLS (Krznar, 2004). 3 LR measure is equal to the multiple of (-1) and the difference between the logarithm of the maximum of the probability function with a restriction on parameters (in this case the restriction requires all the parameters to be equal to zero) and an “average” logarithm of the function probability maximum without a restriction. Therefore a larger LR measure relates to a higher statistical significance of the model. LR measure is analogue to the F measure in the models estimated by OLS (Krznar, 2004). 4 The pre-liberalization sample includes 6 crisis periods and the post-liberalization sample includes 4 crisis periods based on our crisis definition and the exclusion window of 12 months.


55

and the spread between 3-month US T-Bill and federal funds rate. Overall, we note that a number of variables are useful as leading indicators in both the entire sample period and the post-capital account liberalization period. These variables are banking sector fragility index, short-term debt/international reserves, current account balance/GDP, US GDP, the spread between 3-month US T-Bill and federal funds rate, M1, 3-month real US T-Bill rate, foreign liabilities/foreign assets, and international reserves/GDP. Nonetheless, capital inflows/GDP and portfolio investments/GDP are not identified as helpful leading indicators in the post-capital account liberalization period as expected. An interesting finding is that government consumption/GDP was found to be the most useful indicator of crises for the period under study. Another interesting finding is that although FDI/GDP and USD LIBOR rate are strong leading indicators for the period of 1980:012006:06, they are not among the useful indicators for the post-capital account liberalization period. Nonetheless, the results of the signals approach are, at most, suggestive and it would be erroneous to reach a definitive conclusion based on these results alone. Hence, we will seek to find evidence to support these findings using logit regressions for the same period. Table 5





NSR

1

0

59

140

0.00

1.00

0.30

#DIV/0!

1.00

85

4

1

56

136

0.11

0.80

0.30

9.09

7.50


-

5

6

2

54

133

0.15

0.75

0.31

6.67

8.33


-

5

12

7

48

122

0.27

0.63

0.32

3.70

3.80

US GDP

-

90

5

4

55

132

0.35

0.56

0.31

2.86

5.25


+

95

6

5

54

130

0.37

0.55

0.31

2.70

9.00


+

85

11

9

49

121

0.38

0.55

0.32

2.63

5.25

Imports

+

95

13

11

47

117

0.40

0.54

0.32

2.50

9.25

M1

+

85

11

10

49

120

0.42

0.52

0.32

2.38

11.00

US Real T-Bill Rate

+

95

3

3

57

135

0.43

0.50

0.30

2.33

2.00

FDI/GDP

-

5

2

2

58

137

0.43

0.50

0.30

2.33

10.00


-

95

4

4

56

133

0.44

0.50

0.30

2.27

4.00


+

95

7

7

53

127

0.45

0.50

0.31

2.22

8.40

C

80

+

B

+

Federal Funds Rate

A


Variable


D


Signals Approach (1989:09-2006:06)


-

25

15

15

45

111

0.48

0.50

0.32

2.08

7.80

Public Debt/GDP

+

90

8

9

52

124

0.51

0.47

0.31

1.96

1.75

USD LIBOR Rate

+

90

6

7

54

128

0.52

0.46

0.31

1.92

7.25

Fiscal Balance/GDP

-

10

5

6

55

130

0.53

0.45

0.31

1.89

9.25

Exports

-

20

16

21

44

104

0.63

0.43

0.32

1.59

11.20


-

5

9

13

51

119

0.66

0.41

0.31

1.52

9.33


-

25

33

39

27

69

0.66

0.46

0.36

1.52

11.20

GDP per capita

-

5

11

16

49

114

0.67

0.41

0.32

1.49

4.40


+

80

19

27

41

95

0.70

0.41

0.33

1.43

9.40

Capital Inflows/GDP

+

85

9

14

51

118

0.71

0.39

0.31

1.41

8.00


56





0.30

5

57

133

0.72

+

85

15

23

45

103

0.73

0.39

0.32

1.37

9.25

+

70

14

23

46

104

0.78

0.38

0.32

1.28

9.50

Reserve Money/GDP

-

20

20

33

40

88

0.82

0.38

0.33

1.22

11.00

M2 Multiplier

+

85

17

29

43

95

0.83

0.37

0.33

1.20

10.00


+

90

14

25

46

102

0.84

0.36

0.32

1.19

9.40


+

95

8

15

52

118

0.85

0.35

0.31

1.18

10.50


+

80

5

10

55

126

0.88

0.33

0.31

1.14

4.00


+

95

8

16

52

117

0.90

0.33

0.31

1.11

6.67


-

15

5

11

55

125

0.97

0.31

0.31

1.03

4.67

Stock Market Index

-

12

4

9

56

128

0.99

0.31

0.30

1.01

9.50

+

CPI Growth Domestic Credit/GDP

C

3


B

93

Variable

A

NSR

0.38

D



Table 5 (continued)

1.39

8.00


+

95

7

18

53

116

1.15

0.28

0.31

0.87

4.67

Oil prices

+

90

8

23

52

110

1.30

0.26

0.31

0.77

5.80


+

85

6

18

54

117

1.33

0.25

0.31

0.75

9.33


+

75

3

11

57

127

1.59

0.21

0.30

0.63

7.67

Trade Balance/GDP

-

5

3

15

57

123

2.17

0.17

0.30

0.46

7.67

Government Changes

+

N/A

3

15

57

123

2.17

0.17

0.30

0.46

6.00


+

95

3

16

57

122

2.32

0.16

0.30

0.43

10.00

Contagion Dummy

+

N/A

1

11

59

129

4.71

0.08

0.30

0.21

12.00

5.2. Results of Logit Regressions Results of the bivariate logit models covering the post-capital account liberalization period are summarized in Table 6. Strong evidence emerged that US federal funds rate, M2/international reserves, banking sector fragility index, foreign liabilities/foreign assets of banks, US GDP, and 3month US real T-Bill rate are significant in explaining the occurrence of crises. These results are generally in line with the logit estimates obtained for the period of 1980:01-2006:06. Nonetheless, bank reserves/bank assets, short-term debt/international reserves, and M1 did not turn out to be significant in the post-capital account liberalization period. These findings confirm the power of some of the leading indicators obtained from signals approach for the post-capital account liberalization period albeit with the exception of a number of variables such as government consumption/GDP, short-term debt/international reserves, M1, imports, bank reserves/bank assets, and the spread between 3-month US T-Bill rate and federal funds rate. We combined the significant series into a general logit model as shown in Table 71. The results are encouraging. Overall, the significance of federal finds rate, banking sector fragility index, 3-month US T-Bill rate, US GDP, and short-term debt/GDP has been verified by both the signals approach and the logit analysis.

1 Again, prior to building the general model we checked the selected series for multicollinerarity and did not find evidence of correlation among any pairs.


57

Table 6 Coefficient Estimates of the Logit Models with Two Variables (1989:09-2006:06) Variable

a


Logit Coefficient

Standard Errors

Z-Statistic

P>|z|


+

0.822053

2.659485

0.309102

US Real T-Bill Rate

+

0.010207

0.003131

3.260192***

0.7572 0.0011

Fiscal Balance/GDP

-

4.116872

2.862182

1.438368

0.1503

GDP per capita

-

-7.160001

0.323427

-1.656094

0.9677

US GDP

-

-0.522162

0.157648

-3.312200***

0.0009

Commercial Bank Loans to Public Sector Excess real M1 balances

+ +

3.052109 0.094991

4.395581 2.396808

0.694359 0.039632

0.4875 0.9684


-

-1.097750

2.026194

-0.541779

0.5880

M2 Multiplier

+

0.056193

3.453267

0.016272

0.9870


+

1.386015

1.644995

1.842569*

0.0995

Bank reserves/bank assets Imports

+

0.143599 0.743850

2.163535 1.278177

0.066372 0.581962

0.9471 0.5606


-

-10.46990

1.523010

-0.314808

0.0206

Exports

-

0.116801

1.417013

0.082428

0.9343


+

0.016376

0.003708

4.416972***

0.0000


+

21.67176

7.839483

2.764437***

0.0057

Commercial Bank Loans to Private Sector Capital Inflows/GDP

+ +

-2.629292 -0.007097

2.030772 0.034298

-1.294726 -0.206918

0.1954 0.8361

Reserve Money/GDP

-

-4.709064

2.927666

-1.608470

0.1077

Domestic Credit/GDP

+

-4.477376

4.110504

-1.089252

0.2760

CPI Growth

+

1.258228

1.460749

0.861358

0.3890

Short-Term Debt/Long-Term Debt Short Term Debt/International Reserves

+ +

-0.000166 -0.917198

0.000181 0.563459

-0.919483 -1.627799

0.3578 0.1036


-

0.013345

0.020372

0.655063

0.5124


+

-0.176459

0.772831

-0.228328

0.8194


+

-0.009030

0.003342

-0.701625

0.4369


-

0.000481

0.003372

0.142787

0.8865

Real Interest Rate Differential Real Effective Exchange Rate Overvaluation

+ +

0.052814 -0.100980

0.068849 0.194724

0.767099 -0.518578

0.4430 0.6041


-

10.10039

7.842765

1.287861

0.1978

Trade Balance/GDP

-

-1.86E-05

0.000401

-0.046339

0.9630

Stock Market Index

-

-0.685296

1.319430

-0.519388

0.6035

Public Debt/GDP

+

-1.345362

1.766604

-0.761552

0.4463

Real Interest Rates Government Changes

+ +

0.054108 -1.395247

0.069086 1.057219

0.783207 -1.319734

0.4335 0.1869

Oil prices

+

0.150665

1.851018

0.081396

0.9351

M1

+

0.008254

0.006245

1.321796

0.1862

FDI/GDP

-

0.016220

0.033704

0.481232

0.6304

Federal Funds Rate USD LIBOR Rate

+ +

0.011817 -0.010521

0.003204 0.004621

3.688661*** -0.476999

0.0002 0.6228


+

8.21E-05

0.000477

0.172196

0.8633

Contagion Dummy

+

41.11223

NA

NA

NA

Marginal effects are calculated at sample means. * Significant at the 10% level, ** Significant at the 5% level, *** Significant at the 1% level.


58

Table 7 Coefficient Estimates of the Logit Models with Multiple Variables (1989:09-2006:06) Variable


Logit Coefficient

Standard Errors

Z-Statistic

P>|z|

Marginal Effect

Federal Funds Rate

+

0.043643

0.018852

2.315019

0.0206

0.004763


+

0.013797

0.004018

3.433677

0.0006

0.000534


+

0.015805

0.003960

3.991537

0.0001

0.002653


+

0.404196

0.186959

2.161955

0.0306

0.005223

US GDP

-

-0.495452

0.215346

-2.300729

0.0214

-0.008432

US Real T-Bill Rate

+

0.040472

0.018699

2.164449

0.0304

0.003112

-3.234149

1.295861

-2.495754

0.0126

Constant McFadden R-squared1: 0.683660 LR statistic (5 df)2: 141.71014***

* Significant at the 10% level. *** Significant at the 1% level.

5.3. Sensitivity Tests We carried out two sensitivity tests suggested by Manasse et al. (2003) to see how robust the estimated logit model is. First, we dropped observations with extreme values for the variables included in the logit. The direction of influence of the variables for which the extreme values were removed remains unchanged, and the coefficient estimates did not exhibit large falls in the z value. Second, we re-entered several random variables that dropped out of the specification process into the model to ensure that our specification process was not adversely affected by an omitted variable bias. In none of these cases did we see the model's goodness-of-fit improved. Hence, we concluded that the results of the model are robust3.

6. Conclusions In this article, we have used signals approach and logit regressions to explore the causes of currency crises in Turkey for the period of 1980:01-2006:06. Overall, our findings suggest that conventional crisis indicators fail to provide a satisfactory explanation for crises despite the economic intuition: We did not find strong evidence indicating an obvious linkage between the macroeconomic fundamentals and currency crises. For the entire period spanning 1980:01-2006:06, only banking sector fragility index, short-term debt/international reserves, bank reserves/bank assets, US GDP, M1, and US 3-month T-Bill rate have been identified as significant leading indicators by both the signals approach and logit regressions. Still, the fact that banking sector fragility index turned out a significant leading indicator is not surprising as it has been widely documented in the literature that banking sector problems and currency crises are interrelated. This is particularly in line with the literature on Turkish currency crises where the fragility of the banking sector has fre-

1 McFadden R2 is a measure of the goodness-of-fit of a model that is obtained when the ratio of the log of the function maximum with a restriction on parameters (all parameters equal zero) and the log of the probability function maximum without the restriction regarding the parameters are deducted from one; it corresponds to R2 as a measure of goodness-offit of models estimated by OLS (Krznar, 2004). 2 LR measure is equal to the multiple of (-1) and the difference between the logarithm of the maximum of the probability function with a restriction on parameters (in this case the restriction requires all the parameters to be equal to zero) and an “average” logarithm of the function probability maximum without a restriction. Therefore a larger LR measure relates to a higher statistical significance of the model. LR measure is analogue to the F measure in the models estimated by OLS (Krznar, 2004). 3 Results are available from the author upon request.


59

quently been noted as one of the leading causes of currency crises in Turkey (See, for example, Celasun, 1998; and Ozatay and Sak, 2002). Analyzing the post-capital account liberalization period between 1989:09 and 2006:06 in isolation, we find evidence that the importance of US federal funds rate, banking sector fragility index, US GDP, and US 3-month T-Bill rate has been confirmed by both approaches, suggesting that these results are not driven by the specific method of estimation. Additionally, strong evidence emerged that foreign liabilities/foreign assets of banks significantly increase the probability of currency crises in the post-capital account liberalization period. In both samples, indicators pertaining to global economic conditions substantially increase the likelihood of currency crises. Signals analysis, in particular, revealed that these indicators become more important in explaining crises during the post-capital account liberalization period, while conventional variables such as current account deficit/GDP diminish in significance. On the other hand, logit estimates indicate that bank reserves/bank assets, short-term debt/international reserves, and M1 are not significant indicators of crises after the liberalization of capital flows. On the whole, there exists a general consensus on the significance of banking sector fragility index, US GDP, US real T-Bill rate by both approaches for both sample periods. The importance of indicators of global economy is an interesting and novel result, and indicates an increased vulnerability to downturns in global capital markets. The fact that banking sector variables are indeed leading indicators of currency crises for this period is along the lines of the literature where it has been widely argued that financial liberalization increases the possibility of a crisis if the banking sector is fragile. This, coupled with our results confirming the significance of global economic conditions, reveals that financial liberalization has indeed rendered the Turkish economy vulnerable to crises. A possible explanation to this is that the greater degree of openness in financial sector provided greater scope for speculative attacks due to global liquidity conditions. The explanation for the contradiction between the evidence that emerged in the present article and those provided by the literature lies in the selection of sample period, data frequency, and the methodology. In fact, results of any empirical work on currency crises, including those of the present analysis, must be treated with caution due to several technical limitations of available methodologies. In particular, certain issues such as the definition of a currency crisis, selection of the time horizon of the pre-crisis period, dependence of the results on the choice of an arbitrary threshold value, frequency of data, and use of certain series in interpolated form, and the small number of crisis episodes with different characteristics, may affect the statistical reliability of our results. In particular, the previous studies on Turkish currency crises considered narrower sample periods focusing on particular crisis incidences rather than analyzing multiple crises from a broader perspective. On the whole, our results suggest that currency crises are not all alike, even in the context of a single country, and that it is a difficult endeavor to spot any common patterns across various crises episodes. The results on several variables such as international reserves/GDP, interest rates, real effective exchange rate overvaluation, and current account balance/GDP were not as anticipated by the review of the literature. For the sample periods under study, we failed to detect strong empirical evidence to agree that these variables were among the underlying causes of currency crises in Turkey. In a sense, this means that, in the case of Turkey, even if various indicators that are commonly known to form the background to currency crises are followed systematically, incipient problems that may eventually lead to currency crises might not be detected. Obviously, relying solely on the result of the present analysis might not be sufficient to detect future financial crises either. It is practically impossible to recognize and correctly interpret warning signals for all currency crises because each has its own characteristics. Nonetheless, close monitoring of the identified leading indicators would vastly assist the policy makers in forestalling potential crises. In conclusion, the major contribution of the present article is the identification of certain variables whose variation in a certain trend may help policy-makers to foresee future crises. The results of the present study emphasize the need for a careful monitoring of various indicators of financial


60

sector and global economic conditions by the central banks. Given the high degree of international capital mobility, the results obtained in this analysis are also relevant for other emerging markets and for countries intending to liberalize. Although identification of these variables can not replace the sound judgment of policy-makers in guiding policy, it still plays an important role in emphasizing the areas that require special attention.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

Alper, C.E. (2001). “The Turkish Liquidity Crisis of 2000”, Russian and East European Finance and Trade, 37 (6), November-December, p. 58-80. Amemiya T, (1981). “Qualitative response models: A survey”, Journal of Economic Literature, 19: 4, 1483-1536. Arora, V. and Cerisola, M. (2001). “How Does U.S. Monetary Policy Influence Sovereign Spreads in Emerging Markets?”, IMF Staff Papers, Vol. 48, No. 3. Barro, R.J. (1997). “Optimal Debt Management of Indexed and Nominal Debt”, NBER Working Paper 6197. Berg, A., E. Borensztein, G.M. Milesi-Ferretti and C. Pattillo (1999). “Anticipating Balance of Payments Crises – The Role of Early Warning Systems”, IMF Occasional Papers, No. 186. Berg, A. and Patillo, C. (1999), “Are Currency Crises Predictable? A Test”, IMF Staff Papers, Vol 46 No 2, pp. 107-138. Borensztein, E., M.Chamon, O. Jeanne, P. Mauro and J. Zettelmeyer (2005). “Sovereign Debt Structure for Crisis Prevention”, IMF Occasional Paper No. 237. Brüggemann, A.; Linne, T. (2002). “Are the Central and Eastern European Transition Countries still vulnerable to a Financial Crisis? Results from the Signals Approach”, IWHDiscussion Papers No. 157. Calvo, G. (1998). “Varieties of Capital-Market Crises”, in G. Calvo and M. King, (eds) The Debt Burden and its Consequences for Monetary Policy, McMillan. Celasun, O. (1998). “1994 Currency Crisis in Turkey”, World Bank Policy Research Working Paper No. 1913. Cepni, E. and Kose, N. (2006). “Assessing the Currency Crises in Turkey”, Central Bank Review, 1 (2006), pp. 37-64. Dermirguc-Kunt, A. and E. Detragiache (1997). “The determinants of banking crises in developing and developed countries”, IMF Working Paper 106, International Monetary Fund, Washington, D.C. Demirguc-Kunt, A. and E. Detragiache (2000). “Monitoring banking sector fragility: a multivariate logit approach”, World Bank Economic Review, 14 (2), 287-307. Edison, H., (2003). “Do Indicators of Financial Crises Work? An Evaluation of an Early Warning System”, International Journal of Finance and Economics 8, 11-53. Eichengreen, B., Rose, A.K. and Wyplosz, C. (1994), “Speculative Attacks on Pegged Exchange Rates: An Empirical Exploration with Special Reference to the European Monetary System”, NBER Working Paper, No: 4898 (October). Eichengeen, B., A.K. Rose and C. Wyposz (1995). “Speculative Attacks on Pegged Exchange Rates: Exchange Market Mayhem: The Antecedents and Aftermath of Speculative Attacks”, Economic Policy, November, pp. 251-296. Eichengreen, B., Rose, A.K. and Wyplosz, C. (1996). “Contagious Currency Crises”, NBER Working Paper, No. 5681 (July). Ekinci, N.K. (2002) “Anatomy of the Recent Crisis in Turkey”, Journal of Economic Cooperation, Vol. 23, No. 1. El-Shazly, A. (2002). “Financial Distress and Early Warning Signals: A Non-Parametric Approach with Application to Egypt”, Paper presented at the 9th Annual Conference of the Economic Research Forum, Sharjah, UAE, October 2002. Eliasson, A. and Kreuter, C. (2001). “On currency crisis models: A continuous crisis definition”, Presentation at 10th International Conference on Banking and Finance. Esquivel, G. and F. Larrain (1998). “Explaining Currency Crises”, HIID, June.


61

22. Frankel, J. and Rose, A. (1996). “Currency Crashes in Emerging Markets: An Empirical Treatment”. International Finance Discussion Papers, Board of Governors of the Federal Reserve System, No. 534 (January). 23. Fratzcher, M. (2002). “On Currency Crises and Contagion”, ECB Working Paper 139. Frankfurt: European Central Bank. 24. Gelos, R.G. and Sahay, R. (2001). “Financial market spillovers in transition economies”, The Economics of Transition, The European Bank for Reconstruction and Development, vol. 9 (1), pp. 53-86, March. 25. Girton, L. and Roper, D. (1977). A monetary model of exchange market pressure applied to the postwar Canadian experience, The American Economic Review, 67 (4), September, pp. 537-548. 26. Glick, R. and R. Moreno (1999). “Money and Credit Competitiveness and Currency Crises in Asia and Latin America”, Center for Pacific Basin Money and Economic Studies Working Paper PB99-01, Federal Reserve Bank of San Fransisco. 27. Goldstein, M. Kaminsky G. L. and C.M. Reinhart (2000). “Assessing Financial Vulnerability: An Early Warning System for Emerging Markets, Washington, DC: Institute for International Economics. 28. Grabel, I. (1995). “Speculation-led Economic Development: A Post-Keynesian Interpretation of Financial Liberalization Programmes in the Third World”, International Review of Applied Economics, 9, 2, pp. 127-149. 29. Hardy, D. and C. Pazarbasioglu (1999). “Determinants and Leading Indicators of Banking Crises: Further Evidence”, IMF Staff Paper Vol. 46, No. 3. 30. Herrera, S. and C. Garcia, (1999). “User’s Guide to an Early Warning System for Macroeconomic Vulnerability in Latin American Countries”, World Bank Working Paper, November. 31. Jacobs J.P.A.M., Kuper, G.H. and A. Lestano (2005). “Currency crises in Asia: A multivariate logit approach”, CCSO Working Paper No: 2005/6. 32. Jotzo, F., (1999). “The East Asian Currency Crisis: Lessons for an Early Warning System”, APSEM Working Paper # 99/2, Australia National University, Canberra. 33. Kamin, S.B., J.W. Schindler, and S.L. Samuel (2001). “The contribution of domestic and external sector factors to emerging market devaluations crises: an early warning systems approach”, International Finance Discussions Papers 711, Board of Governors of the Federal Reserve System, Washington, D.C. 34. Kaminsky, G., S. Lizondo, and C. Reinhart, (1998). “Leading Indicators of Currency Crises”, International Monetary Fund Staff Papers 45, 1-48. 35. Kaminsky, G.L. and C.M. Reinhart (1999). “The twin crises: the causes of banking and balance-of-payments problems”, American Economic Review, 89 (3), 473-500. 36. Kaminsky, G,L.; Reinhart, C.L. and Végh, C.A. (2003). “The Unholy Trinity of Financial Contagion”, Journal of Economic Perspectives, 17(4), pp. 51-74. 37. Kibritcioglu, B; Kose, B and Ugur, G. (1998). “A Leading Indicators Approach to the Predictability of Curreny Crises: The Case of Turkey” Hazine Dergisi, No. 12, July, pp. 1-27. 38. Kibritçioğlu, A (2003). “Monitoring Banking Sector Fragility”, Arab Bank Review, 5/2: 5166 39. Kipici, A.N. and Kesriyeli, M. (1997). “Reel Doviz Kuru Tanimlari ve Hesaplama Yontemleri”, TCMB Arastirma Genel Mudurlugu, Yayin No 97/1, Ankara. 40. Komulainen, T. and Lukkarila, J. (2003). “What Drives Financial Crises in Emerging Markets ?”, What drives financial crises in. emerging markets?. Emerging Markets Review, Vol. 4, pp. 248-272. 41. Krkoska, L., (2000). “Assessing Macroeconomic Vulnerability in Central Europe”, European Bank for Reconstruction and Development, Working Paper No. 52. 42. Krznar, I. (2004). “Currency Crisis: Theory and Practice with Application to Croatia”, Croatian National Bank Working Paper, Vol. 12, August, pp. 1-46. 43. Kumar, M., U. Moorthy and W. Perraudin (2003). “Predicting Emerging Market Currency Crashes”, Journal of Empirical Finance 10, pp. 427-454.

62


44. Lanoie, P. and S. Lemarbre (1996). “Three Approaches to Predict the Timing and Quantity of LDC Debt Rescheduling”, Applied Economics 28 (2), 241-246. 45. Maddala, S. (1983). Linear Dependent and Quantitative Variables in Econometrics, Cambridge University Press, New York, NY. 46. Manasse, P.; Roubini, N. and Schimmelpfennig (2003), “Predicting Sovereign Debt Crises”, IMF Working Paper No WP/03/221. 47. Marchesi, S. (2003), “Adoption of an IMF programme and debt rescheduling”, Journal of Development Economics, 70 (2), 403-423. 48. Moreno, R. (1995). “Macroeconomic Behavior during Periods of Speculative Pressure or Realignment: Evidence from Pacific-Basin Economies” Federal Reserve Bank of San Francisco Economic Review, pp. 3-16. 49. Obstfeld, M. (1986). “Rational and Self-Fulfilling Balance of Payments”, American Economic Review, 76, pp. 72-81. 50. Ozatay, F. and Sak, G. (2002). “Banking Sector Fragility and Turkey's 2000-01 Financial Crisis”, Brookings Trade Forum, 2002, pp. 121-160. 51. Rahman, S., Tan, L.H., Hew, O.L.; Tan, Y.S. (2004). “Identifying Financial Distress Indicators of Selected Banks in Asia”, Asian Economic Journal 18 (1), 45-57. 52. Seyidoglu, H. (2003). “Uluslararasi Mali Krizler, IMF Politiklari, Az Gelismis Ulkeler, Turkiye ve Donusum Ekonomileri, Dogus Üniversitesi Dergisi, 4 (2) 2003, 141-156. 53. Walter, S. (2006). “Policy Responses to Speculative Attacks Before and After Elections: Theory and Evidence”, CIS Working Papers No. 19, Center for Comparative and International Studies, Zurich, Switzerland. 54. Yeldan, Erinç (1998). “On Structural Sources of the 1994 Turkish Crisis: A CGE Modeling Analysis” The International Review of Applied Economics, 12 (3): 397-414. 55. Zhuang, J., and M. Dowling (2002). “Causes of the 1997 Asian financial crisis: what can an early warning system model tell us?”, ERD Policy Brief, 7. Manila: Asian Development Bank.

Weak exports or excessive import growth could lead to deteriorations in the current account which can lead to currency crises (Fratzcher, 2002; Edison, 2003).

High deficits make the country vulnerable to expectation shifts and less capable to generate external revenue to finance a balance of payments problem whereas surplus is expected to indicate a diminished probability to devalue and thus to lower the probability of a crisis (Fratzcher, 2002; Kamin et al., 2001; Lanoie and Lemarbre, 1996; Marchesi, 2003).

Trade Balance/GDP


Used as a proxy of reserve adequacy. It captures to what extent the liabilities of the banking system are backed by international reserves. It assesses the short-term liquidity and convertibility of a country’s currency. Fearing devaluation, economic agents may substitute local currency for foreign currency. This ratio shows the extent to which the Central Bank can withstand this pressure. (Kamin et al., 2001; Calvo, 1998; Fratzcher, 2002).

Excessive reliance on short-term debt is an indicator of financial vulnerability as the shorter and more concentrated the debt maturity the more likely debt crises are to occur. In addition, short-term debt may increase a country’s exposure to sharp increases in interest rates, which may have additional negative consequences, as governments may need to increase taxes in order to service the debt (Barro, 1997; Borensztein et al., 2005).

Used as an indicator of reserve adequacy. A high short-term debt/international reserves ratio is a vulnerability indicator, signifying exposure to crises (Rodrik and Velasco, 1999). As a rule of thumb known as the Greenspan-Guidotti rule, international reserves should cover at least the level of short-term debt (Fratzcher, 2002).

Higher amount of FDI implies a lesser share of the current account being financed by volatile portfolio inflows and should lower the probability of crisis. Also, higher FDI ratios may be indicative of more attractive economic policies and prospects. FDIs are long-term capital inflows which increase the productive capacity of the country and produce the revenues necessary to cover future capital outflows (Evans et al., 2000).

Reflects the changes in expectations of foreign investors. When portfolio investments take a sudden drop or reverse, this can be taken as a sign of loss of confidence, and may be the immediate cause of a currency crisis (Jotzo, 1999).

Used as a proxy for financial account and vulnerability to a sudden stop of capital inflows (Komulainen and Lukkorilla, 2003).

Used as a proxy of seignorage, i.e. monetization of the government deficit. Currency crises may take place if a deficit is financed through seignorage, as this may cause agents to expect a crisis and push the economy to a bad equilibrium (Obstfeld, 1986).



Short-Term Debt/ International Reserves

FDI/GDP

Portfolio Investments/GDP

Capital Inflows/GDP

Reserve Money (High-Powered Money)/GDP

Used as a proxy for fragility of banking sector. The index consists of a weighted average of bank credits to the domestic private sector, bank deposit and the foreign liabilities of banks (See Kibritcioglu, 2003). Calculated as: [((%ΔCPS-μCPS)/σCPS ) + ((%ΔFL–μFL)/σFL) + ((%ΔDEP–μDEP)/σDEP)]/3 where %Δ indicates the difference in 12-month changes in data that has been used. CPS, FL and DEP stand for credits to the domestic private sector, the foreign liabilities of banks and bank deposit, respectively.

Used as a proxy of liquidity. High growth of M1 may indicate excess liquidity which can lead to speculative attacks on the currency thus leading to a currency crisis (Eichengreen et al., 1995).

A higher M2 multiplier indicates higher growth in money supply which may lead to higher inflationary expectations and expectations of a future devaluation of the currency. The resulting real appreciation of the exchange rate may put a peg under pressure (Bruggemann and Linne, 2002).

Banking Sector Fragility Index (BSFI)

M1

M2 Multiplier

Financial Sector

Declining level of international reserves may trigger a speculative attack against the currency and shows that a currency is under devaluation pressure. It may also be used as indicators of a country’s financial difficulty, dealing with debt repayment (Kaminsky et al., 1998; Berg and Pattillo, 1999; Marchesi, 2003).


Capital Account

CBRT EDDS

Used as a proxy of external vulnerability. Currency overvaluation could lead to deteriorations in the current account and is often perceived by the market as an indication that the country will have to devalue. It may also cause a loss of competitiveness and a recession. Ultimately it adversely affects a country’s ability to service its debt (Kaminsky et al. 1998, 1999; Dermirguc-Kunt and Detragiache, 2000).

Real Effective Exchange Rate a Overvaluation

M2 (IFS line 35 ZLF)/base money (IFS line 14ZF)

IFS line 34ZF

Foreign Liabilities of banks (IFS line 26C), credits to the domestic private sector (IFS line 32DZF), bank deposit (IFS line 24+ 25)

Reserve money (IFS line BL)/GDP(CBRT EDDS)

Capital Inflows (IFS line 78BJD)/GDP

IFS line 78BFD/GDP (Data available after 1985)

(IFS line 78 BED)/GDP

(Foreign debt with maturity of less than 1 year/international reserves) WB GDFS*

WDI*

M2 (IFS line 35ZLF)/International Reserves (IFS line 1LD)

International Reserves (IFS line 1LD)/GDP(CBRT EDDS)

CBRT EDDS

Exports-imports (IFS line 70DZF71DZF)/GDP (CBRT EDDS)

IFS line 71DZF

Excessive import growth may show that the exchange rate is overvalued which could lead to a loss in competitiveness and worsening in the current account (Kaminsky et al., 1998; Berg and Pattillo, 1999; Edison, 2003).

Imports

IFS line 70DZF

Data Source

Declining export growth implies that the government may devalue in order to empower the exports. Besides it shows a loss in competitiveness and possible problems of domestic enterprises. It also inhibits the country’s ability to earn foreign exchange to finance an existing current account deficit (Kaminsky et al., 1998; Berg and Pattillo, 1999; Bruggemann and Linne 2002).

Rationale

Exports

Current Account

Variable

APPENDIX I. Potential Crisis Indicators and Sources of Data

Banks and Bank Systems / Volume 2, Issue 2, 2007 63

1

Excessive growth of domestic credit may serve as an indicator of the fragility of the banking system as it increases the chances of bank failures due to balance sheet problems and in terms of non-performing loans and currency mismatches (Bruggemann and Linne, 2002).

Used as a proxy for the quality of the asset portfolio and the credit risk. A high ratio is an indicator of inefficiency of the financial institutions (Rahman et al., 2004).

Used as a proxy for excessive monetary expansion (Zhuang and Dowling, 2002).

Rapid growth in credit fueled by excessive monetary expansion makes the economy more vulnerable to crises (Corsetti et al.. 1998; Sachs et al., 1996).

Used as proxy of financial liberalization. High real interest rates signal a liquidity crunch and can also increase the probability of loan defaults (Kaminsky et al., 1998; Edison, 2003; Dermirguc-Kunt and Detragiache, 2000).

Used as a proxy of liquidity risk and banking sector fragility. Contractions in commercial bank deposits often reflect distress and problems in the banking sector and increase the chances of a bank run. Also, a weak banking system increases the probability of speculative attack since the investors know that the government will be reluctant to increase interest rates (Chang and Velasco, 2000; Berg and Pattillo, 1999; Edison, 2003).

Adverse macroeconomic shocks are less likely to lead to crises in countries where the banking system is liquid. High ratio indicates banks’ soundness (Dermirguc-Kunt and Detragiache, 1997).

Used as a proxy of banking sector fragility measuring exchange rate exposure and the imbalance between foreign currency denominated liabilities and foreign currency denominated assets (Corsetti et al., 1998; Kibritcioglu, 2003).

Currency and banking crises have been linked to rapid growth in credit fueled by excessive monetary expansion in many countries (Zhuang and Dowling, 2002).

Used as a proxy for lending boom, which may increase the ratio of bad loans to total assets, thereby weakening the banking system. The rapid increase of the credit to the private sector may also indicate that a large amount of credit is going to dubious projects (Kibritcioglu, 2003).

Domestic Credit/GDP

Deposit Money Banks Net Past Due Loans/Total Loans

Excess Real M1 Balances




Bank Reserves/Bank Assets



Commercial Bank Credit to Private Sector

A decline in the asset prices may lead to loan defaults. It also signals a loss of investor confidence. Besides, it indirectly measures contagion (Kaminsky et al., 1998; Berg and Pattillo, 1999; Edison, 2003; Komilainen and Lukkarilla, 2003).

Negative per capita growth is assumed to increase the policymaker’s incentives to switch to a more expansionist policy, which can be achieved through a nominal devaluation of the currency (Esquivel and Larrain, 1998).

Inflation rate is likely to be associated with high nominal interest rates and may proxy macroeconomic mismanagement which adversely affects the economy and the banking system and may lead to currency instability (Dermirguc-Kunt and Detragiache, 1997; Lanoie and Lemarbre, 1996).

Stock Market Index

GDP Per Capita

Consumer Price Index (CPI)

Higher expenditure net of revenues would have positive effect on the likelihood of crisis (Saqib, 2002). Large fiscal deficits could lead to a worsening in the current account position, which could in turn put pressure on the exchange rate (Zhuang and Dowling, 2002).

Government Consumption/GDP

Until 1999 from IFS line 80, after 1999 annual from treasury web site (www.hazine.gov.tr).

3-month US Real T-Bill Rate

High US interest rates may induce capital outflows (Edison, 2003; Kamin et al., 2001; Milesi-Ferretti and Razin, 1998). The yield on the three-month U.S. treasury bill can be considered a key short-term risk-free rate that usually serves as a benchmark for pricing other high-yield assets in world capital markets, and that would most likely reflect changes in global liquidity and economic conditions (Arora and Cerisola, 2001).

Higher indebtedness is expected to raise vulnerability to a reversal in capital inflows, and hence to raise the probability of a crisis (Lanoie and Lemarbre, 1996).

Public Debt/GDP

Global Economy

High deficits increase the vulnerability to shocks. They could lead to a worsening in the current account position, which could put pressure on the exchange rate (Dermirguc-Kunt and Detragiache, 2000; Zhuang and Dowling, 2002).

Fiscal Balance/GDP

Fiscal Sector

Economies are more vulnerable to crises when economic growth slows down. Lower output growth indicates a deceleration of the economy (Berg and Pattillo, 1999; Hardy and Pazarbsioglu, 1999).


Real Sector

Rationale

Variable

US T-bill rate (IFS 60p)-CPI (IFS line 64x)

Government Consumption (IFS line 91F and WDI)/GDP*

WDI *

Budget balance (IFS line 80 and the treasury web site)/GDP1*

IFS line 64XZF

EDDS*

ISE National -100 Index (CBRT EDDS, data available after 1996)

CBRT EDDS**

IFS line 32DZF

IFS line 12C (data available after 1986)

Foreign Liabilities (IFS line 26C)/Foreign Assets of Banks (IFS line 21ZF)

(IFS line 20ZF)/(IFS lines 21+22a+22g (data available after 1986)

IFS line 24+ IFS line 25

3-month deposit rate (IFS line 60L)

IFS line 12C (data available after 1986)

(IFS line 34/64)/(trend derived using Hodrick-Prescott (HP) filter).

CBRT EDDS (data available after 1986)

Domestic Credit(IFS line 32ZF)/GDP(CBRT EDDS)

Data Source

64 Banks and Bank Systems / Volume 2, Issue 2, 2007

Defined as foreign interest rates (3-month US deposit rate) less domestic interest rate (3-month Turkey deposit rate). The higher the differential, the larger is the probability of an outflow of reserves and may signal devaluation expectations (Komulainen and Lukkarilla, 2003).

In addition to the direct impact of changes in U.S. interest rates on rates in developing countries, interest rate spreads (the differences between yields on sovereign bonds of developing countries and U.S. treasury securities of comparable maturities), which are a proxy for country risk, have tended to move in the same direction as the changes in U.S. interest rates.

Currency crises may pass contagiously from one country to another (Eichengreen et al., 1996).The contagion dummy takes the value of one if there has been a major financial crisis during the month and zero if otherwise. Crisis dates were obtained from Kaminsky et al. (2003).

Used as a proxy for global liquidity conditions (Ades et al., 2000) changes in U.S. interest rates, or likewise in global liquidity conditions, would be expected to influence positively country risk and sovereign spreads in developing countries (Arora and Cerisola, 2001).

Captures heightened uncertainty about the expected stance of U.S. monetary policy (Arora and Cerisola, 2001).

Used as a direct measure of U.S. monetary policy. Changes in US monetary policy have been felt by developing countries through effects on the cost and availability of funds (Arora and Cerisola, 2001).

World Oil Prices


Interest Rate Spread (3-month bonds)

Contagion Dummy

USD LIBOR Rate

Spread between the yield on 3month US t-bill and the US federal funds rate

US Federal Funds Rate

The National Assembly of Turkey web site (www.tbmm.gov.tr)

FRB Online Database

US T-bill rate (IFS 60p) – FRB Online Database

ECB Database

Kaminsky et al. (2003)

FRB Online Database

IFS line 60L for Turkey – IFS line 60L for USA

IFS line 176 (Crude oil prices)1

WDI*

Data Source

1

US dollar per barrel (Spot prices).

NOTES: GDP used in calculations is obtained from CBRT EDDS is linearly interpolated from annual (until 91) and monthly (after 91). * linerarly interpolated from annual data; ** liıneraly interpolatd from quarterly data; a (CPI-based real effective exchange rate index calculated using the IMF weights for 19 countries). An increase in the index shows appreciation of the Turlish lira. Calculated as: (REER-REERt-24)/REERt-24 (Fratzscher, 2002). Data Sources Key: IFS: International Financial Statistics Database of the IMF, CBRT EDDS: the Central Bank of the Republic of Turkey’s Electronic Data Delivery System, WDI: World Bank’s World Development Indicators Database, ECB: European Central Bank’s Statistical Warehouse, FRB: US Federal Reserve Board Database, WB GDFS: Global Development Finance Statistics Database.

Government Change Dummy

Used as a proxy for political instability. Takes the value of 1 if there is election and 0, if otherwise.

High oil prices pose a danger to the current account position, and also could lead to domestic recessions (Edison, 2003).

US GDP

Political

Rationale

Higher foreign output growth should strengthen exports and thus reduce the probability of a crisis (Edison, 2003; Kamin et al., 2001).

Variable

Banks and Bank Systems / Volume 2, Issue 2, 2007 65


66

APPENDIX II. GRAPHICAL REPRESENTATION OF EXPLANATORY VARIABLES 400

.25

.20

.20

.15

300

.15

.10

.10

200

.05 100 0

.05

.00

.00

-.05

-.05 -.10

-.10 -100 80 82 84 86 88 90 92 94 96 98 00 02 04 06

-.15

-.15

80 82 84 86 88 90 92 94 96 98 00 02 04 06

80 82 84 86 88 90 92 94 96 98 00 02 04 06


Domestic Credit/GDP

M2 Multiplier

.4

.6

20

.5

.3

10

.4

.2

.3

0

.1

.2 .1

.0

-10

.0

-.1

-.1

-20

-.2

-.2 -.3 80 82 84 86 88 90 92 94 96 98 00 02 04 06

-30

-.3 80 82 84 86 88 90 92 94 96 98 00 02 04 06

80 82 84 86 88 90 92 94 96 98 00 02 04 06

.25

.5

.20

.4

.12 .08

.3

.15

.2

.10

Oil Prices

Real Interest Rate

Excess Real M1 Balances

.04

.1

.05

.00

.0

.00

-.1

-.05

-.2

-.04

-.3

-.10 80 82 84 86 88 90 92 94 96 98 00 02 04 06

-.08 80 82 84 86 88 90 92 94 96 98 00 02 04 06

-.4 80 82 84 86 88 90 92 94 96 98 00 02 04 06




1.2

3

2.8 2.4

0.8

2

2.0 0.4

1.6 1

1.2

0.0

0.8 -0.4

0

0.4 0.0

-0.8 80 82 84 86 88 90 92 94 96 98 00 02 04 06 Imports

-1

-0.4 80 82 84 86 88 90 92 94 96 98 00 02 04 06 Fiscal Balance/GDP

80 82 84 86 88 90 92 94 96 98 00 02 04 06 Government Consumption /GDP

Banks and Bank Systems / Volume 2, Issue 2, 2007 .5

40

600

.4

30

500

.3

20

.2

400

10

.1

67

300

0

.0

200

-10

-.1

100

-.2

-20

-.3

-30

0

-.4

-40

-100

80 82 84 86 88 90 92 94 96 98 00 02 04 06

80 82 84 86 88 90 92 94 96 98 00 02 04 06

CPI

1.0


.20

20

0.8

80 82 84 86 88 90 92 94 96 98 00 02 04 06

Capital Inflows/GDP

.15

10

0.6

.10

0

0.4

.05

0.2

-10

.00

-20

-.05

0.0 -0.2

-.10

-0.4 80 82 84 86 88 90 92 94 96 98 00 02 04 06

80 82 84 86 88 90 92 94 96 98 00 02 04 06

-30 80 82 84 86 88 90 92 94 96 98 00 02 04 06

Public Debt/GDP

.3



160

0.8

140 .2

0.4

120 100

.1

0.0

80 .0

-0.4

60 40

-.1

-0.8

20 -.2 80 82 84 86 88 90 92 94 96 98 00 02 04 06

5000

Short-Term Debt/International Reserves

Narrow Money (M1)

1600

.6

1200

.4

4000

.2

800

3000

80 82 84 86 88 90 92 94 96 98 00 02 04 06

80 82 84 86 88 90 92 94 96 98 00 02 04 06

Reserve Money/GDP

6000

-1.2

0

.0

2000

400

-.2

1000 0

-.4

0 -400

-1000

87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05

80 82 84 86 88 90 92 94 96 98 00 02 04 06 Trade Balance/GDP

60

2.0

80 82 84 86 88 90 92 94 96 98 00 02 04 06 GDP

.8

50

1.5

-.6


.6

40

1.0

.4

30 20

0.5

.2

10

0.0

.0

0

-0.5

-10

-.2

-20

-1.0 80 82 84 86 88 90 92 94 96 98 00 02 04 06 Foreign Liabilities/Foreign Assets of Banks

87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05

-.4 80 82 84 86 88 90 92 94 96 98 00 02 04 06

FDI/GDP

Commercial Bank Credit to Private Sector


68 200

8

.10

160

6

.08

120

4

.06 80

2

.04

40 0 -40 -80 80 82 84 86 88 90 92 94 96 98 00 02 04 06

0

.02

-2

.00

-4 -6

-.02

80 82 84 86 88 90 92 94 96 98 00 02 04 06

80 82 84 86 88 90 92 94 96 98 00 02 04 06

Real US 3-Month T-Bill Rate



16

6000

14

4000

12

2000

10

1.0 0.8 0.6

0

8

-2000

6

0.4

-4000 0.2

4

-6000 2 80 82 84 86 88 90 92 94 96 98 00 02 04 06

0.0

-8000

80 82 84 86 88 90 92 94 96 98 00 02 04 06

87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05

US GDP

Government Changes


1.2

60

.4

1.0

.3

40

0.8

.2

0.6

20

.1

0.4

.0

0

0.2

-.1

0.0

-20

-0.2

-.2 -.3

-0.4 96

97

98

99

00

01

02

03

04

05

-40 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05

Stock Market Index

-.4 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05

Portfolio Investments/GDP

Bank Reserves/Bank Assets

1.6

2000

250

1.2

1000

200 150

0

0.8

100

-1000

0.4

50

-2000

0.0

0

-3000

-50

-0.4

-100

-4000

80 82 84 86 88 90 92 94 96 98 00 02 04 06

80 82 84 86 88 90 92 94 96 98 00 02 04 06

-0.8 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05

US Federal Funds Rate

Spread between US T-Bill and Federal Funds Rate

Net Past Due Loans/Total Loans of Banks

250

1.0

.4 .3

200

0.8

.2

150

.1

0.6

100

.0

50

0.4

-.1

0

-.2

0.2

-50

-.3

-100 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 USD LIBOR Rate

0.0 80 82 84 86 88 90 92 94 96 98 00 02 04 06 Contagion Dummy

-.4 80 82 84 86 88 90 92 94 96 98 00 02 04 06 Exports

Note: The shaded areas in the graphs mark the 12-month window before crises. The greater the incidence of the flashing indicators within these windows, the more vulnerable the economy is to a crisis.


69

INDUSTRIAL PRODUCTION AS A CREDIT DRIVER IN BANKING SECTOR: AN EMPIRICAL STUDY WITH WAVELETS Alper Ozun*, Atilla Cifter** Abstract This paper examines the timescale effects of industrial production on credits volume at banks. By using industrial production in Turkey and credit volumes of Turkish banks from 3/1992-12/2006, this study employs wavelet filters to estimate multi-scale causality for scaled time series. The original data is transformed by the wavelet filter up to 5 time scales. The first wavelet coefficient captures oscillations with a period length 3 to 6 months. Equivalently, the consequent wavelets capture oscillations with a period of 7-12, 13-24, 25-48 and 49-96 months, respectively. The results of multi-scale granger causality test show that the industrial production is effective on credits volume upto 24 months, while the credits volume starts to affect industrial production after 2 years. This paper has originality in presenting multi-scale effects of industrial production as a credit driver by using wavelet analysis with Turkish data. Key words: Bank credits, industrial production, wavelets, multi-scale causality, granger causality. JEL classification: C45, E51, E23, C14.

1. Introduction The relationship between banks as the intermediates and industrial sector as money demander is the milestone of the economic life. Banks as the credit providers have crucial role in the production facilities in the industrial sector. According to demand-following hypothesis, economic growth leads to financial developments, while the reverse relationship is suggested by supplyleading hypothesis. Robinson (1952) as one of the initial supporters of the demand-following hypothesis argues that financial sector has minor effect on growth. Economic development creates demand for financial intermediates leading to growth in lending facilities of the credit institutions. On the other hand, Schumpeter (1911) already stresses the importance of financial intermediaries for economic development. Gurley and Shaw (1955) and Davis (1965), as the initial supporters of the supply-leading hypothesis underline the effects of financial system on macroeconomic growth. Patrick (1966) argues that financial sector contributes significantly to industrial growth in emerging markets, while the industrial growth increases demand for financial sector services in advanced economies. Though that argument might be accepted as valid in Latin America in 1990’s, the financial crises in Mexico, Argentina and Brazil are extreme features of this relationship. After that initial discussion in the theory of financial economics, numerous empirical researches that will be covered in the literature review part of this article have been conducted to show the relationship between growth and credit facilities of the banks. Research results mostly argue that there is a positive correlation between the financial development and economic growth in real terms. On the other hand, there is disagreement on the underlying causality. Studies often display varying results. The originality of this paper is that it uses a new methodology to display the timescale of the relationship between the production and bank credits. By employing wavelets as a filtered model to determine the timescale effects among the variables, this paper measures both the strength of the relationship between production and credits volume and also the duration of the relationship. The methodology also enables us to see the dual way causality between the variables.

**

© Alper Ozun, Atilla Cifter, 2007.

* Is Bank of Turkey, Turkey. Marmara University, Deniz Investment-Dexia Group, Turkey.

70


For empirical analysis, the paper uses industrial production and volume of bank credits in Turkey from 3/1992 to 12/2006. The wavelet algorithm enables us to scale the causal effects between the variables. In that respect, as much as the authors know, it is the first empirical work to employ the wavelets analysis in measuring the causal interrelationship between credits volume and industrial production. Working with data from Turkish markets has also an importance for the research because of the volatile business cycle in Turkish economy. The wavelets are expected to filter the high volatility to display a robust causality between the variables under examination. The time period under examination includes three financial catastrophes in 1994, 1998 and 2001 in which interest rates on credits increased above 1000% and even in 2001 the credits facilities of the banks were frozen for a while. In that respect, the methodology is proper to capture the high volatility and shocks in the economy by filtering the time series data. The paper is constructed as follows. In the next part, a theoretical framework and current literature are presented. In the third part, wavelets methodology is examined in detail. The fourth part includes the presentation and discussion of the empirical results both in terms of pragmatic and methodological perspectives. The final part is the conclusion where the research findings are summarized and suggestions for the future researches are given.

2. Theoretical Framework and Literature Review Financial systems channel household savings into the industry and allocate economic resources among firms. They are the sources connecting financial development to economic growth. Patrick (1966) argues two alternative causal relationships between financial development and economic growth. The first one, namely demand-following hypothesis, states that the demand for financial intermediation depends on the economic growth measured by real output. The alternative perspective is supply-leading hypothesis. By transferring resources from the traditional sectors to the highgrowth sectors, the financial system supports economic growth. Schumpeter (1911) argues that the financial services are essential for technological progress and and economic growth. If the financial sector is crucial for the economy, then there should be a relation between financial markets and economic growth. Gurley and Shaw (1955) are the first to examine the relationship between financial markets and industrial activity. Their study shows that financial markets extend financial power of borrowers and increase the efficiency of trade. Since this is an empirical research paper, we do not discuss the theory in deep. Instead, the theory is expressed in reviewing literature. As parallel to two different theoretical perspectives, empirical results on the direction of the causality between two variables are also contradictory. The empirical results vary on the economy and time period under examination. Methodology also matters in facing different empirical results. For example Jung (1986) finds causality for both directions between financial development and economic growth by time series analysis. On the other hand, Xu (2000), by extending the research of Jung (1986) with VAR analysis shows that the financial sector does not affect growth. However, Christopoulos and Tsionas (2004) display that causality exists from finance to growth in the long-run by using panel unit tests and cointegration analysis. King and Levine (1993) use liquid liabilities of banks and nonbank financial intermediaries over GDP, bank credit over the sum of bank credit and central bank domestic assets and credit to private enterprises over GDP to measure the effects of financial services on economic growth. They find that banking sector development can spur economic growth in the long-run. According to Allen and Oura (2004), traditional neoclassical literature on growth suggests that financing is not important. In this perspective there are two main sources of economic growth. The first source is growth within the technological frontier as a result of factor accumulation. The second one is innovation that causes the technological frontier to move outwards. They state that innovation is crucial for an economy to experience sustained growth for long-run. On the other hand, factor accumulation can still be an important part of growth for emerging economies that are a long way from the technological frontier.


71

Empirical researches on the issue for the emerging markets have also different contradictory findings. We present recent empirical findings from different economies in this part of the paper before examining previous works with Turkish data. Dritsaki and Dritsaki-Bargiota (2006) examine the causal relationship among financial development, credit market and economic growth by using a trivariate autoregressive VAR model in Greece from 1988 to 2002. They show that there is a bilateral causal relationship between banking sector development and economic growth. Bulir (1998) shows that industrial production is cointegrated with various measures of bank credits between 1976 and 1990. Although the impact of credit supply shocks on production changes, growth follows credit loosening. Asian economies have contradictory behaviours on the issue, as well. Tang (2005) examines the direction of causality relationship between bank lending and economic growth for the five ASEAN economies, namely, Malaysia, Singapore, Indonesia, Thailand and the Philippines. He uses Granger causality test to examine the demand-following hypothesis (economic growth causes bank lending), and supply-leading hypothesis (bank lending causes economic growth). The empirical results display that the supply-leading hypothesis is valid for Thailand while the demand-following hypothesis is approved by time series data of Singapore. In Malaysia, Indonesia and the Philippines, on the other hand, the variables are statistically independent. Shan et al. (2006) estimate a vector autoregression (VAR) model to examine the relationship between financial development and economic growth for nine OECD countries and China. Test results have little support for the supply-leading hypothesis. Empirical works on the relationship between bank credits and production or growth are restricted. Darrat (1999) examines the role of financial deepening in economic growth in Saudi Arabia, Turkey and the United Arab Emirates by multivariate Granger-causality tests within an errorcorrection model. Empirical results support the argument that financial deepening is a necessary causal factor of economic growth, but the strength of the evidence changes across countries. Kar and Pentecost (2006) investigate the causal relationship between financial development and economic growth in Turkey with five alternative proxies for financial development. By using Granger causality tests with cointegration and vector error correction methodology, they show that the direction of causality is sensitive to the proxy used for financial development. If financial development is measured by the money to income ratio the direction runs from financial development to economic growth. On the other hand, if the bank deposits, private credit and domestic credit ratios are used as proxy, growth leads financial development. Aslan and Kucukaksoy (2006) examine financial development and economic growth relationship for Turkey over the period of 1970-2004 by Granger causality test. The test results support the supply-leading hypothesis for Turkey. As the results of empirical researches show, there exits contradictory evidence on the causality between growth and financial sector development even in the same economy. Our research presents a new perspective on the relationship by scaling the time to show the direction and strength of the interrelated effects between the variables.

3. Methodology and Data 3.a. Methodology The wavelets methodology derives its theoretical roots from Fourier analysis. Fourier analysis states that any function can be represented with the sum of sine and cosine functions. Fourier series are expressed in Equation (1). ∞

f ( x) = b0 + ∑ (bk cos 2πkx + a k sin 2πkx) k =1

(1)


72 2π

1 b0 = 2π

ak =

1

∫ f (x) dx ,

bk =

0

1

π

2π

∫ f (x ) Cos(kx) dx

,

0

2π

∫ f (x ) Sin(kx ) dx

π

0

a0,, ak and bk can be solved with OLS. Fourier to wavelet transition is in Equation (2). 2 j −1

∞

f ( x ) = c0 + ∑

∑c

j =0

k =0

ψ (2 j χ − k )

jk

ψ (x) is the mother wavelet, mother to all dilations and translations of ψ

(2) in Equation (2). Tkacz

(2001) gives a simple example for mother wavelet in Equation (3).

1 ⎧ ⎪1 : 0 ≤ x < 2 ⎪ . 1 ⎪ Ψ ( x ) = ⎨− 1 : ≤ x < 1 2 ⎪ ⎪0 : other ⎪ ⎩

(3)

In finance, the maximal overlap discrete wavelet transform (MODWT) is used instead of discrete wavelet transform (DWT) since MODWT can work with any sample size N and wavelet variance estimator of MODWT is asymptotically more efficient than the estimator based on the DWT. The MODWT is formulated with matrices (Percival and Walden, 2000) and yields J vectors of ~ wavelet filter coefficients W j,t , for j=1,…,J and t=1,….,N/2j, and one vector of wavelet filter coef~

ficients V

~ W ~ V

j ,t

j ,t

=

j,t

through Equations (4) and (5) (Gallegati, 2005).

N

∑ w~

t=Lj

=

N

∑ v~

t=L j

Y j ,t

X j ,t

f (t − 1) ,

f (t − 1) ,

X

(4)

(5)

Y

where w j ,t and v j ,t are the scaled wavelet and scaling filter coefficients. In and Kim (2006) define wavelet covariance between two series Xt and Yt as in Equation (6). N

1 Cov (λ j ) = ~ N

t=Lj

In the equation,

λj

~ X ~Y j ,t V j ,t .

∑W

(6)

represents scale. In and Kim (2006) also define MODWT estimator of the

wavelet correlation as in Equation (7).

Cov (λ j ) , v X (λ j )v~Y (λ j )

ρ~ (λ j ) = ~

(7)


73

~ (λ ) and v~ (λ ) are wavelet variances estimated by the MODWT coefficients for where v X j Y j scale

λj

described in Equation (8) and Equation (9).

[ ]

1 N ~ 2 v~X (λ j ) = ~ ∑ WjX,t , N t =Lj

(8)

[ ].

1 N ~Y ~ v (λ j )Y = ~ ∑ Vj,t N t =Lj

2

(9)

We employ Johansen unrestricted cointegration test without trend and with constant term to examine the cointegration between the variables (Johansen, 1988; and Johansen and Joselius 1990) as expressed in Equation (10).

H1* (r ) : ∏ yt −1 + Bxt = α ( β ' yt −1 ) + ρ 0 .

(10)

Cointegration in stationary time series by Johansen procedure is set with trace and maximum eigenvalue statistics as shown in Equations (11) and (12). k

_

λtrace(r ) = −T ∑In(1 − λ i ), r = 0,1,2,3,....,n −1,

(11)

i =r +1

_

λ max( r ,r +1) = −TIn(1 − λ r +1 ) .

(12)

Granger causality test is used to see whether at least one directional causality exists between variables (Granger, 1969). Granger causality test is summarized in Equations (13) and (14). M

K

n =1

n =1

M

K

n =1

n =1

Ct = B0 + ∑ Bn Ct −n +∑α n IPt −n + ε t , IPt = B0 + ∑ Bn IPt −n +∑α n Ct −n + ε t ,

(13)

(14)

where C and IP represent change in bank credits and industrial production respectively. We apply Granger causality test with maximum 9 lags as our data are limited to 55 observations. 3.b. Data By using industrial production in Turkey and credit volumes of Turkish banks from 3/199212/2006, the paper employs wavelet filters to estimate dynamic correlation for scaled time series. Level and log-differenced series are shown in Figure 1. Quartely data as industrial production index and credits volume that is used in this paper are from Turkish Central Bank database, www.tcmb.gov.tr. The original data is transformed by the wavelet filter up to 5 time scales. The first wavelet coefficient captures oscillations with a period length 3 to 6 months. Equivalently, the consequent wavelets capture oscillations with a period of 7-12, 13-24, 25-48 and 49-96 months, respectively.

Ma 1 - r- 9 3 De 1 - c -9 3 Se p 1 - -9 4 Ju 1 - n -9 5 Ma 1 - r- 9 6 De 1 - c -9 6 Se p 1 - -9 7 Ju n 1 - -9 8 Ma 1 - r- 9 9 De 1 - c -9 9 Se p 1 - -0 0 Ju 1 - n -0 1 Ma 1 - r- 0 2 De 1 - c -0 2 Se p 1 - -0 3 Ju 1 - n -0 4 Ma 1 - r- 0 5 De 1 - c -0 5 Se p -0 6

1-

c r e d i ts v o l u m e

500,000,000.0 credits volume

1

400,000,000.0

0.6

100.000

300,000,000.0 80.000

200,000,000.0 60.000

40.000

100,000,000.0

0.0

LLcredits volume

0.3

0.2

-0.1

Bright line represents change in credits and dark line represents change in industrial production.

0

-0.05

0.1

i n d u s tr i a l p r o d u c ti o n

600,000,000.0

i n d u s tr i a l p r o d u c ti o n

1De 1 - c -9 Se 2 1 - p -9 Ju 3 1 - n -9 Ma 4 1 - r-9 D 5 1 - e c -9 Se 5 1 - p -9 Ju 6 1 - n -9 Ma 7 1 - r-9 D 8 1 - e c -9 Se 8 1 - p -9 Ju 9 1 - n -0 Ma 0 1 - r-0 D 1 1 - e c -0 Se 1 1 - p -0 2 Ju 1 - n -0 Ma 3 1 - r-0 De 4 1 - c -0 Se 4 1 - p -0 Ju 5 n06

c r e d i ts v o l u m e

74 Banks and Bank Systems / Volume 2, Issue 2, 2007 160.000

industrial production

140.000

120.000

20.000

0.000

0.1

LLindustrial production

0.5 0.05

0.4

0 -0.1

-0.15

Fig. 1. Credits volume and industrial production (level and log-differenced series)1


75

4. Empirical Results Table 1 reports Phillips-Peron (Phillips and Peron, 1988) and Augmented Dickey Fuller tests (Dickey and Fuller, 1981) of industrial production (IP) and credits volume (C) based on level, logdifferenced and time-scaled decompositon up to 5 scale. Lag lengths are determined with Schwartz Information Criteria. Series are not stationary at I(0) where stationary at I~(1) based on both Phillips-Peron (Phillips and Peron, 1988) and Augmented Dickey Fuller test (Dickey and Fuller, 1981) unit root tests at the 1% significance level. WJ1, WJ2, WJ3, WJ4 and WJ5 represent time-scale decomposition of C and IP log-differenced series. Table 1 Unit Root test results

C

Phillips- Peron test I(1)

Augmented D-F test I(1)

4.19608

0.769523

C (Log-differenced)

-5.14417***

-3.13032**

WJ1 for C

-24.6553***

-8.13458***

CWJ2 for C

-6.68074***

-7.14287***

WJ3 for C

-2.78277*

-6.49884***

WJ4 for C

-2.02804

-1.51253

WJ5 for C

-1.31928

-2.05974

-0.00553034

0.168694

IP IP (Log-differenced)

-7.26752***

-7.19877***

WJ1 for IP

-30.7972***

-7.41872***

WJ2 for IP

-4.66449***

-7.68116***

WJ3 for IP

-3.46919***

-6.3224***

WJ4 for IP

-1.7842

-4.57092***

WJ5 for IP

-1.20227

-2.44645

Notes. The table reports results of the Phillips-Perron and augmented Dickey-Fuller tests for all the time series. The number of lags has been selected using the Schwarz information criterion with a maximum of twelve lags. *, **, *** Indicate the rejection of the unit root null at the 10%, 5% and 1% significance level respectively.

Johansen cointegration test (Johansen, 1988; and Johansen and Joselius, 1990 ) results in Table 2 show that original data (C and IP) are cointegrated where time-scaled data are cointegrated up to 3rd scale at 5% significance level and cointegrated up to 4th scale at 10% significance level. This indicates that credits volume and industrial production are not only cointegrated at log-differenced level but also cointegrated based on time-scale decomposition or multi-scale cointegration. Since we employ multi-scale granger causality we will also add 5th time-scale in multi-scale causality analysis although 5th scale is not cointegrated1.

1

Granger causality test can be applied both cointegrated and noncointegrated variables in multi-scale analysis.

76


Table 2 Cointegration test results Unrestricted Cointegration Rank Test Original data (C&IP)

r=0 r≤1 r=0

WJ1 (6 months)

r≤1 r=0

WJ2 (1 year)

r≤1 r=0

WJ3 (2 years)

r≤1 r=0

WJ4 (4 years)

r≤1 r=0

WJ5 (8 years)

r≤1

Trace Stat.

Max-Eigen Stat.

20.6245*** 3.75923

20.6245*** 24.3837**

89.6598*** 31.2678***

58.392*** 31.2678***

41.7976*** 15.872***

25.9256*** 15.872***

24.8944*** 9.68213**

15.2122* 9.68213**

23.3752** 2.09998

21.2752*** 2.09998

15.6297 0.541378

15.0883* 0.541378

Notes. *, **, *** indicates significance of cointegration at the 10%, 5%, and 1% level respectively. The number of lags is selected as 4 using the Schwarz information criterion with a maximum of nine lags.

Figure 2 shows original data and scaled data with wavelet analysis. LA (8) MODWT multi-scale decomposition is applied in wavelet analysis (five different wavelet details, WJ1 to WJ5). There is a high correlation between C and IP at 5th scale or 8 years decomposition. Figure 2 shows dynamic correlation (between C and IP) for scaled time series as WJ1 to WJ5. Rolling window size is 18 data points or 4,5 years. 2001 is crises year for Turkish economy and in 2001 all scaled time series correlation became negative but not WJ2 or one year decomposition. Since the negative correlation between C and IP is not significant in the theory of credit-growth explanation this indicates that C and IP cointegrate in one year lag in crises period. After 2006 or presently WJ3 or 2 years correlation increases where other time-scale decomposition becomes nonsignificance. This evidence shows that C and IP cointegrate around 2 years lag presently. Figure 4 shows wavelet correlations between C and IP. Correlation is maximum at 5th time scale as 53% and wavelet correlation increases with time scales from WJ2 to WJ5. This indicates that C and IP are not fundamentally different starting from 6 months until 8 years or in the long-run (Lee, 1999; and In and Kim, 2006).

01 .1 0 1 2 .1 9 .0 9 0 1 6 .1 9 2 .1 9 0 1 2 .1 3 . 06 9 9 0 1 .1 9 3 . 12 94 0 1 .1 9 .0 9 0 1 6 .1 9 4 .1 9 0 1 2 .1 5 . 06 9 9 0 1 .1 9 5 .1 9 0 1 2 .1 9 6 .0 9 0 1 6 .1 9 6 .1 9 0 1 2 .1 7 .0 99 0 1 6 .1 9 7 .1 9 0 1 2 .1 9 8 .0 9 0 1 6 .1 9 8 .1 9 0 1 2 .1 9 .0 99 0 1 6 .2 0 9 .1 0 0 1 2 .2 0 0 . 06 00 0 1 .2 0 . 12 01 0 1 .2 .0 00 0 1 6 .2 0 1 .1 0 0 1 2 .2 0 2 . 06 02 0 1 .2 0 .1 0 0 1 2 .2 3 .0 00 0 1 6 .2 0 3 . 12 04 0 1 .2 0 .0 0 0 1 6 .2 0 4 .1 0 0 1 2 .2 5 . 06 0 0 .2 0 5 06

1 0,2

0,1 -0,05

-0,1

-0,1

0,08

0,06 WJ3_Credits

WJ3_IP 0,02

0,04

0

-0,02

-0,04 -0,02

-0,06 -0,03

-0,08 -0,04

0,05

0,04 WJ5_Credits

WJ5_IP

WJ4 0,008

0,03 0,004

0,02 0,002

0,01 0

0 -0,002

-0,01

-0,004

-0,02 -0,006

-0,03 -0,008

-0,04 -0,01

-0,05 -0,012

01 .1 0 1 2 .1 9 .0 9 0 1 6 .1 9 2 .1 9 0 1 2 .1 9 3 .0 9 0 1 6 .1 9 3 .1 9 0 1 2 .1 9 4 .0 9 0 1 6 .1 9 4 .1 9 0 1 2 .1 9 5 .0 9 0 1 6 .1 9 5 . 12 9 6 0 1 .1 9 . 06 9 6 0 1 .1 9 .1 9 0 1 2 .1 9 7 .0 9 0 1 6 .1 9 7 .1 9 0 1 2 .1 9 8 .0 9 0 1 6 .1 9 8 .1 9 0 1 2 .1 9 9 .0 9 0 1 6 .2 0 9 .1 0 0 1 2 .2 0 0 .0 0 0 1 6 .2 0 0 .1 0 0 1 2 .2 0 1 .0 0 0 1 6 .2 0 1 .1 0 0 1 2 .2 0 2 .0 0 0 1 6 .2 0 2 .1 0 0 1 2 .2 0 3 .0 0 0 1 6 .2 0 3 .1 0 0 1 2 .2 0 4 .0 0 0 1 6 .2 0 4 .1 0 0 1 2 .2 0 5 . 06 0 5 .2 0 06

0,3

-0,02

.1 0 1 2 .1 9 .0 9 0 1 6 .1 9 2 . 12 93 0 1 .1 9 .0 9 0 1 6 .1 9 3 .1 9 0 1 2 .1 9 4 .0 9 0 1 6 .1 9 4 .1 9 0 1 2 .1 9 5 .0 9 0 1 6 .1 9 5 . 12 96 0 1 .1 9 .0 9 0 1 6 .1 9 6 .1 9 0 1 2 .1 9 7 .0 9 0 1 6 .1 9 7 .1 9 0 1 2 .1 9 8 .0 9 0 1 6 .1 9 8 .1 9 0 1 2 .1 9 9 . 06 99 0 1 .2 0 . 12 00 0 1 .2 0 .0 0 0 1 6 .2 0 0 .1 0 0 1 2 .2 0 1 .0 0 0 1 6 .2 0 1 .1 0 0 1 2 .2 0 2 .0 0 0 1 6 .2 0 2 .1 0 0 1 2 .2 0 3 . 06 03 0 1 .2 0 . 12 04 0 1 .2 0 .0 0 0 1 6 .2 0 4 .1 0 0 1 2 .2 0 5 . 06 05 .2 0 06

Credit

01

01 .1 0 1 2 .1 9 .0 9 0 1 6 .1 9 2 .1 9 0 1 2 .1 9 3 .0 9 0 1 6 .1 9 3 .1 9 0 1 2 .1 9 4 .0 9 0 1 6 .1 9 4 .1 9 0 1 2 .1 9 5 .0 9 0 1 6 .1 9 5 .1 9 0 1 2 .1 9 6 .0 9 0 1 6 .1 9 6 .1 9 0 1 2 .1 9 7 .0 9 0 1 6 .1 9 7 .1 9 0 1 2 .1 9 8 .0 9 0 1 6 .1 9 8 .1 9 0 1 2 .1 9 9 .0 9 0 1 6 .2 0 9 . 12 00 0 1 .2 0 . 06 00 0 1 .2 0 .1 0 0 1 2 .2 0 1 .0 0 0 1 6 .2 0 1 .1 0 0 1 2 .2 0 2 .0 0 0 1 6 .2 0 2 . 12 03 0 1 .2 0 . 06 03 0 1 .2 0 .1 0 0 1 2 .2 0 4 .0 0 0 1 6 .2 0 4 .1 0 0 1 2 .2 0 5 . 06 05 .2 0 06

0,6

01 .1 0 1 2 .1 9 .0 9 0 1 6 .1 9 2 .1 9 0 1 2 .1 3 . 06 9 9 0 1 .1 9 3 . 12 9 4 0 1 .1 9 .0 9 0 1 6 .1 9 4 .1 9 0 1 2 .1 5 . 06 9 9 0 1 .1 9 5 .1 9 0 1 2 .1 9 6 .0 9 0 1 6 .1 9 6 . 12 9 7 0 1 .1 .0 99 0 1 6 .1 9 7 .1 9 0 1 2 .1 9 8 .0 9 0 1 6 .1 9 8 . 12 9 9 0 1 .1 .0 99 0 1 6 .2 0 9 .1 0 0 1 2 .2 0 0 . 06 0 0 0 1 .2 0 .1 0 0 1 2 .2 1 .0 00 0 1 6 .2 0 1 .1 0 0 1 2 .2 0 2 . 06 0 2 0 1 .2 0 .1 0 0 1 2 .2 3 .0 00 0 1 6 .2 0 3 . 12 0 4 0 1 .2 0 .0 0 0 1 6 .2 0 4 .1 0 0 1 2 .2 5 . 06 0 0 .2 0 5 06

01 .1 0 1 2 .1 9 .0 9 0 1 6 .1 9 2 .1 9 0 1 2 .1 9 3 .0 9 0 1 6 .1 9 3 . 12 94 0 1 .1 9 . 06 94 0 1 .1 9 .1 9 0 1 2 .1 9 5 .0 9 0 1 6 .1 9 5 .1 9 0 1 2 .1 9 6 .0 9 0 1 6 .1 9 6 .1 9 0 1 2 .1 9 7 .0 9 0 1 6 .1 9 7 .1 9 0 1 2 .1 9 8 .0 9 0 1 6 .1 9 8 . 12 99 0 1 .1 9 . 06 99 0 1 .2 0 .1 0 0 1 2 .2 0 0 .0 0 0 1 6 .2 0 0 .1 0 0 1 2 .2 0 1 .0 0 0 1 6 .2 0 1 .1 0 0 1 2 .2 0 2 .0 0 0 1 6 .2 0 2 .1 0 0 1 2 .2 0 3 .0 0 0 1 6 .2 0 3 . 12 04 0 1 .2 0 . 06 04 0 1 .2 0 .1 0 0 1 2 .2 0 5 . 06 05 .2 0 06


IP

0,1 0,25

0,3

0,05

0,4

0

-0,05

-0,15

Original data 0,03 0,08

0,1

-0,01

WJ2

0,006 0,08

WJ5

Fig. 2. Original data and scaled data with wavelet analysis1

Bright line represents change in credits and dark line represents change in industrial production.

77

WJ1_Credits

WJ1_IP

0,5 0,2

0,15

0 -0,1

WJ2_Credits

WJ2_IP

0,06

0,01

0,02 0,04

0

-0,04

WJ4_Credits

0

0,08

0,06

0,04

0,1

0,02

0,05 0

0 -0,02

-0,04

-0,06

-0,15 -0,08

-0,2 -0,1

WJ1 0,05

0,04

0,03

0,02

0,02 0,01

0 0

-0,01

-0,02

-0,06 -0,03

-0,04

-0,08 -0,05

WJ3 WJ4_IP

0,02

0,06 0,015

0,04 0,01

0,02 0,005

0

-0,02

-0,005

-0,04 -0,01

-0,06 -0,015


78

1.2 1

Original data WJ1 WJ2 WJ3 WJ4 WJ5

0.8 0.6 0.4

1-Jun-06

1-Jun-05

1-Dec-05

1-Dec-04

1-Jun-04

1-Jun-03

1-Dec-03

1-Dec-02

1-Jun-02

1-Jun-01

1-Dec-01

1-Jun-00

1-Dec-00

1-Jun-99

1-Dec-99

1-Jun-98

1-Dec-98

1-Jun-97

1-Dec-97

1-Jun-96

1-Dec-96

1-Jun-95

1-Dec-95

1-Jun-94

1-Dec-94

1-Dec-93

-0.2

1-Jun-93

0

1-Dec-92

0.2

-0.4 -0.6 -0.8

Fig. 3. Estimated dynamic correlation for scaled time series 60% 53%

50% 45%

40% 30% 20% 10%

12% 5%

0% -10% -20%

WJ1

WJ2

WJ3

WJ4

WJ5

-20%

-30%

Fig. 4. Estimated wavelet correlations between Credits and Industrial Production

Multi-scale granger causality test results for original and time-scaled data are shown in Table 3. C and IP are not caused each other where causality exists for time-scaled data or wavelet based decomposition analysis. IP causes C at WJ1,WJ2, and WJ3 while C causes IP at WJ4 and WJ5. In other words, IP causes C in 6 months to 2 years while C causes IP in 4 years to 8 years in the longrun. As a result IP affects C in the short-run and C affects IP in the long-run.


79

Table 3 Granger causality test Granger causality test for original data

CÆIP

IPÆC

Granger causality test for wavelet analysis

C&IP

WJ1

WJ2

WJ3

WJ4

WJ5

0.45800 (0.76603)

0.35349 (0.8401)

1.41311 (0.2462)

1.76452 (0.1540)

2.58648 (0.0505)*

2.42800 (0.0626)*

1.44553 (0.36199)

5.03170 (0.0021)*

2.40185 (0.0649)*

3.45960 (0.0156)*

0.99249 (0.4222)

1.68293 (0.1719)

Notes. The original data has been transformed by the wavelet filter (LA(8)) up to time scale 5. The significance levels are in parentheses. * indicates significance at 5% level. The first detail (wavelet coefficient) WJ1 captures oscillations with a period length 3 to 6 months. Equivalently, WJ2, WJ3, WJ4, and WJ5 capture oscillations with a period of 7-12, 13-24, 25-48 and 49-96 months, respectively.

5. Concluding Remarks In many researches conducted with data from different economies and time periods, relationship between industrial production and credit volume has been empirically figure out. However, the evidence on the direction of that relationship varies on the methodology used, economics and time periods examined. In this research paper, we use a new methodology, namely wavelets analysis, to empirically examine the time-scale relationship between industrial production and credit volume. By using data from Turkish economics for the time-period between 3/1992 and 12/2006, we try to figure out the industrial production is a credit driver for the Turkish banks. We transform the original data upto 5 time scales with wavelet filter, which captures oscillations with a period length 3 to 6 months. The consequent wavelets capture oscillations with a period of 7-12, 13-24, 25-48 and 49-96 months, respectively. The results of multi-scale granger causality test display the fact that the industrial production has significant effects on credits volume upto 24 months. However, after that period, the credits volume starts to cause industrial production to increase. We think that the paper represents interesting empirical findings with a new methodology from an emerging economy. The researches in the future might focus on alternative new methodologies that are able to show dynamic relationship between the two variables. We think that a combination of wavelets and neural networks, namely wavelet networks, might be used to see the relationship. On the other hand, it should be noted that empirical results might be biased in methodology employed. Therefore, future researches with new methodologies might present comparative analysis for different methodologies.

References 1. 2. 3. 4. 5.

Allen F., H. Oura. Sustained Economic Growth and the Financial System // Monetary and Economic Studies, Bank of Japan, 2004. – №22/S-1. – 95-119. Aslan, O., I. Kucukaksoy. – // Financial Development and Economic Growth Relationship: An Econometric Application on Turkish Economy // Istanbul University Econometrics and Statistics e-Journal. 2006. – №4/1. – 26-38. Aleš B. Income Inequality-Does Inflation Matter? // IMF Working Papers, №98/7, International Monetary Fund, 1998. Christopoulos, D., E. Tsionas. Financial Development and Economic Growth: Evidence from Panel Unit Root and Cointegration Tests // Journal of Development Economics, 2004. – №73. – 55-74. Darrat, A.F. Are Financial Deepending And Economic Growth Causally Related? Another Look At The Evidence // International Economic Journal, 1999. – №13/3. – 19-35, October.

80


6.

Davis, L.E. The Investment Market, 1870-1914: The Evolution of a National Market // Journal of Economic History, 1965. – №25/3, 151-197. Dickey, D.A., W.A., Fuller. Likelihood Ratio Statistics for an Autoregressive Time Series with a Unit Root, Econometrica, 1981. – №1057-72. Dritsaki, C., M. Dritsaki-Bargiota. The Causal Relationship between Stock, Credit Market and Economic Development: An Empirical Evidence for Greece // Economic Change and Restructuring, 2005. – №38/1. Gallegati, M. A Wavelet Analysis of MENA Stock Markets, Mimeo // Universita Politecnica Delle Marche Working Papers, 2005. Ancona, Italy. Granger, C.W.J. Investigating Causal Relations by Econometric Models and Cross-Spectral Methods // Econometrica, 1969. – №37. – 424-38. Gurley, J.G., E.S. Shaw. Financial Aspects of Economic Development // American Economic Review, 1955. – №45/4. – 515-538. Johansen, S. Statistical analysis of cointegration vectors // Journal of Economic Dynamic and Control, 1988. – №12. – 231-254. Johansen, S. and Juselius, K. Maximum Likelihood Estimation and Inference on Cointegration with Application to the Demand for Money // Oxford Bulletin of Economics and Statistics, 1990. – №52. – 169-210. In, F. and S. Kim. The Hedge Ratio and the Empirical Relationship Between the Stock and Futures Markets: A New Approach Using Wavelets, The Journal of Business, 2006. – №79. – 799-820. Jung, W.S. Financial Development and Economic Growth: Inetrnational Evidence // Economic Development and Cultural Change, 1986. – №34/2. – 333-346. Kar, M., E. Pentecost. Financial Development and Economic Growth in Turkey: further evidence on the causality issue // Economic Research Paper, 2000- №00/27, Loughborough University, Department of Economics. King, R.G., R. Levine. Finance and Growth: Schumpeter Might Be Right // Quarterly Journal of Economics 1993. – №108. – 717-737. Lee, G.G.J. Contemporary And Long-Run Correlations: A Covariance Component Model And Studies On The S&P 500 Cash And Futures Markets // Journal of Futures Markets, 1999. – №19. – 877-94. Patrick, H.T. Financial Development and Economic Growth in Underdeveloped Countries // Economic Development and Cultural Change. 1966. – №14/2. – 174-189. Percival, D.B., A.T. Walden. Wavelet Methods for Time Series Analysis // Cambridge University Press, 2000. Philips, P.C., B. P. Perron. Testing for a Unit Root in Time Series Regression // Biometrica, 1988 – №75. – 335-446. Robinson, J. The Generalization of the General Theory in The Rate of Interest and Other Essays, London: Macmillan, 1952, 69-142. Shan, J.Z., F. Sun, A. Morris. Financial Development And Economic Growth // Review of International Economics, 2001. – №9. – 443-54. Schumpeter, J. The Theory of Economic Development, Harvard University Press, 1911, Cambridge. Tang, T.C. An Examination of the Causal Relationship between Bank Lending and Economic Growth: Evidence from ASEAN // Savings and Development Savings, 2005. – №3. – 22-31. Tkacz, G. Estimating the Fractional Order of Integration of Interest Rates Using a Wavelet OLS Estimator // Studies in Nonlinear Dynamics&Econometrics. 2001. – №5/1. – 19-32. Xu, Z. Financial Development, Investment, and Economic Growth // Economic Inquiry, 2000. – №38/2. – 331-344.

7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.


81

AUTHORS OF THE ISSUE John A. Consiglio

− Ph.D., M.Phil (Eur. Studs)., MBA(Wales)., Dip Law & Adm., DipFS., FCIB., CSA. lectures in the Department of Banking & Finance of the University of Malta, Malta

Ivana Valová

− Ing., Masaryk University, Faculty of Economics and Administration, Czech Republic

Matthias Menke

− Department of Finance, Accounting & Real Estate, European Business School, Germany

Dirk Schiereck

− Professor, Department of Finance, Accounting & Real Estate, European Business School, Germany

Christophe J. Godlewski

− Pôle Européen de Gestion et d’Economie, Université Louis Pasteur – FSEG – LaRGE, Strasbourg, France

Mete Feridun

− Department of Economics, Loughborough University, United Kingdom

Alper Ozun

− Market Risk Group, Is Bank of Turkey, Turkey

Atilla Cifter

− Department of Econometrics, Marmara University, and Financial Reporting Department, Deniz Investment-Dexia Group, Turkey

82


SUBMISSION GUIDELINES FOR AUTHORS The cover page of manuscript should contain the title and name(s) of the author(s). At the bottom of this page, provide the name, author’s degree, position and the place of work as well as contact details (phone number, job or/and personal e-mail). 1. Abstract preparation guidelines 1.1. The abstract (150-200 words) should reflect the conceptual content of the article. 1.2. Include Journal of Economic Literature (JEL) classifications 2. The paper main body preparation guidelines 2.1. The paper should present the result of independent original research, undertaken by the author; it should contain the data never published before. 2.2. The paper should contain a clear description of research objective and its subject. 2.3. The methodology of research should be described in detail. 2.4. The author personal scientific contribution must be grounded in the paper. 2.5. The paper should contain basic suggestions how to solve the problem under study. 3. References in the text 3.1. References in the text are made as follows: (Myers, 2000), the former being name of the author, the latter edition year. 3.2. Example of references: Alchian A., S. Woodward. Reflections on the Theory of the Firm // Journal of Institutional and Theoretical Economics, 1987. - № 143. - pp. 110 - 136. Berle A.A., G.C. Means. The Modern Corporation and Private Property. - New York: Macmillan, 1932. - 418 рр. 4. Manuscript length 4.1. The paper should not be in less than 2000 words and should not exceed 6000 words. 5. Submission guidelines Please send your paper using one of the formats listed bellow: 5.1. An electronic copy as a MS Word .doc file (all versions accepted) or as .rtf file to Journal "Banks and Bank System": Executive Editor: [email protected] 5.2. A hard copy supplied with a disc containing electronic copy of the paper to: Publishing company “Business Perspectives” Dzerzhynsky lane, 10 Sumy 40022 Ukraine 6. Reviewing process All papers are refereed. We apply the best practice in the paper reviewing, “a double-blind reviewing”. 7. Acceptance fee We offer very democratic fee policy to our contributors. We ask to pay only those contributors whose papers have been already reviewed and accepted for publication in the journal.