Corporate investment and European monetary policy

2 downloads 0 Views 173KB Size Report
small firms' investments are particularly dependent on the share of long-term debt in their financing and on monetary policy conditions, and that compared to ...
DRAFT FOR COMMENTS

Corporate investment and European monetary policy

Alex Cobham1

Abstract. This paper assesses the different structures of investment finance in the main EMU member economies, using the BACH database of balance sheet data, aggregated by firm size. Two main sets of results are obtained. First, each country/size class group is characterised using the net sources and uses of finance methodology. The importance of long-term debt as a source of investment finance, and its reduction in the later part of the 1990s, is emphasised. The second contribution of the paper is to conduct panel regressions examining the determinants of the level and returns to investment, paying particular attention to the differential impact of economic conditions and monetary policy. The main findings to emerge are that the level of small firms’ investments are particularly dependent on the share of long-term debt in their financing and on monetary policy conditions, and that compared to larger firms their growth is more strongly associated with investment. The broad conclusion is that while stability should remain a central macroeconomic goal, financial market policy aimed at encouraging the expansion of a stable supply of long-term credit is also critical for the employment and growth prospects of the eurozone.

1

Supernumerary Fellow in Economics, St Anne’s College, University of Oxford. Contact address: Finance and Trade Policy Research Centre, 21 St Giles, Oxford OX1 3LA; [email protected]. I am grateful to Valpy FitzGerald, Richard Mash and participants at the EIFC workshop at Warwick Business School (February 2004) for comments.

Corporate investment and European monetary policy Patterns of employment and economic growth depend on the underlying investment behaviour of firms (and the technological progress embodied in, and resulting from those investments). The impact of European economic and monetary union (EMU) on corporate investment will take place within the eurozone through two main channels – firstly, through the resulting changes in financial markets, and secondly through the effects of the single monetary policy.

This paper aims to further understanding of these channels by using a database of aggregated firms’ balance sheet data to examine the investment behaviour and the financing of investment, for firms of different sizes in the largest EMU member economies: France, Germany, Italy and Spain. To this end it considers four main questions: ♦

How do the (gross and net) sources of finance for firms’ investment vary?



How do levels of investment vary?



How does the cost of capital vary?



How do returns to investment differ?

For each of these questions, the role of firm size-class and industry sector are considered, and the impacts of economic and in particular monetary policy conditions assessed. A panel of data for the period 1988-2001 is used to evaluate each question in turn, in section 4. Section I first introduces the data and highlights some key facts of the growth and employment characteristics of firms of different sizes. Section II sets out the methodology to be used in assessing sources of investment finance for the corporate sector.

Section III then presents the results of the sources of finance

analysis. Section IV presents regression analysis to identify the determinants of the external sources of finance and the level of firms’ investment, along with results showing the impact of firm size, patterns of external financing and monetary policy conditions on the quality of firms’ investments. A final section concludes.

2

I. Data outline and main characteristics

I.i: Basic features of the data set

The data forming the basis for empirical work comes primarily from the BACH data set of balance sheet data. Aggregate data are reported for three size classes of firm and for all firms together. The size classes are defined by turnover of firms, with the smallest size of firms having turnover up to €7m, the largest in excess of €40m, and an intermediate size between the two.

The data, which are annual observations stretching at the longest from 1982 to 2000, are split by country, firm size class and into 18 industrial sectors (see table 1 for the full breakdown with NACE codes).

For each size-class, country and sector, 95

balance sheet variables are given, including: (i)

general variables (e.g. the number of enterprises covered, total assets, turnover, employment, value added);

(ii)

balance sheet data on assets (e.g. fixed and current assets) and liabilities (e.g. subscribed capital, and debt due in less than or more than one year);

(iii)

a profit and loss account including data on the firms’ incomings and outgoings (e.g. materials and staff costs, financial income and depreciation), and finally

(iv)

some further information (including e.g. acquisitions of assets and distribution of profits).

Data is used from a sliding sample which is maintained over two year periods. For each sample year, there exists data for that year as the reference year (year t, which allows comparison with the previous year) and also data for the same year as the base year (i.e. year t-1, to observe the change to the following year). The required financial ratios can be observed as they change within a constant sample from one year to the next, but without the survivor bias of using a maintained sample for the whole period nor the sample composition bias of using reference year data alone. As long as a firm-size approach is adopted, the effects of the unrepresentativeness of the national samples are minimised and the data can provide a relatively balanced picture of changes in the population. 3

Sliding sample data is available for the countries under consideration for the following years: France: 1983/4-2000/1, Germany: 1987/8-1998/9 and Italy and Spain: 1982/3-2000/1. The quality and comparability of data for each of the sample countries before the period 1987-1988 is dubious however, and hence the range of data used is from 1987/8 to 2000/1 for France, Italy and Spain, and from 1987/8 to 1998/9 for Germany. Other data on economic and monetary policy conditions are taken from the IMF’s International Financial Statistics CD-ROM.

I.ii: Descriptive analysis

Figure 1 shows the difference in size distributions between the firms sampled for the BACH database for each of the chosen countries, and the actual distribution in those countries. Medium and large firms are over-represented throughout. In each case, it can be clearly seen that smaller firms are under-represented, and that this problem is most acute in the Italian case. This is unsurprising given the relative dominance of small firms in the Italian case – it has the most extreme distribution of our sample countries – and the relative difficulties of gathering data from amongst this class. Figure 1 illustrates the precise extent of over- or under-representation of each size class in each country.

Unless good data on medium- sized and large firms is excluded simply to maintain the relative size distribution, it is almost inevitable that the practical scale problems of collecting data from the relatively large populations of small firms will lead to such a pattern.

The BACH data are as broad as exist, however, and they provide a

reasonable cross-section by which to study economy-wide phenomena.

Figure 2 illustrates the pattern of sizes across countries in 1999 (the last year for which data is available for all countries). Panel (a) shows average levels of turnover per firm. On average, Italian small firms are much larger than those of any other country. Italian large firms are the smallest of those in the four countries considered. Spanish small and medium-sized firms are the smallest. German medium-sized and large firms are the largest, while French firms are centrally positioned in each size 4

class. Panel (b) shows similar patterns in employment levels; the main difference is that of medium-sized firms, the Spanish have the lowest levels of turnover per firm but the highest employment.

Figure 3 shows the breakdown across industry sectors.

Panels (a) and (b) show

average firm size in terms of turnover and employee numbers, respectively. Some industries score above average on both, and these are simply characterised by larger firms – e.g. sector 14, manufacture of transport equipment. As panel (c) makes clear, those industries which score highly only on turnover (employee numbers) – e.g. sector 11, energy and water (sector 16, hotels and restaurants) – are characterised by particularly high (low) capital/labour ratios.

Figure 4 shows that industry factors appear to dominate country factors in terms of these simple ratios. Although German (Italian) firms tend to exhibit slightly higher (lower) levels of average turnover and employment per firm in any given sector, country scores are closely clustered for each sector while levels vary strongly across sectors. The main implication of this is that firms in a given industry are likely to be characterised by largely similar structures of production, regardless of nationality.

II. Methodology, data structure and variable generation

II.i: The ‘net sources’ approach

Analysis of the corporate sector’s sources and uses of funds for investment has been used to provide answers to questions about different countries’ financial market structures, perhaps most famously by Corbett & Jenkinson (1996 and 1997). Following an approach initiated by Mayer (1988), 2 the authors used national flow of funds data in order to contrast the sources of investment finance in Germany, Japan the UK and the US. A key feature of their approach was to focus on net sources of finance – the contribution of bank debt, for example, was considered net of the

2

Although the idea of using flow of funds accounts to examine net financing of economic activity had long been the object of discussions on their construction – see, e.g., Atlee (1962). 5

corporate sector’s holdings of cash and deposits, and that of equity issues net of equity purchases.

Their results – as shown in table 2 – reinforced the view of Mayer that the marketbased versus bank-based dichotomy of financial structures was a false or at least seriously exaggerated one: ‘in purely quantitative terms, the contribution of German banks over the last 25 years has been no greater than that of their counterparts in the “market-based” financial systems’ of the UK and US (p.74). The latter are based on high internal finance rather than high equity financing as is often assumed, while the system of Japan reflects a wider range of positive net sources which Corbett and Jenkinson characterise as ‘balanced external finance’.

There are two main benefits to using the net sources approach.

Firstly, market

characteristics such as the practice of holding generally high bank balances in Japan can be treated without generating distortion – that is, the real contribution of the bank sector to corporate investment can be seen. Secondly, since it is the net flow of new investment which is generally considered critical to future growth, only examination of the net sources of new finance for investment will shed light on the financial market determinants of economic performance.

In the example of Japanese bank balances, higher gross loan levels are not necessarily evidence of a higher contribution by the banking sector to corporate investment. Instead, the contribution of the banking sector is assessed in terms of its net contribution in each period to the actual (net) investments made. The levels (stocks) of balances held and loans outstanding are undoubtedly an important characteristic of a financial market, and continue to be a central component of work on the contribution of financial markets to economic development (e.g. see the World Bank’s Database on Financial Market Structure).

However, to the extent that flows of investment drive future growth, it is new flows of bank debt to the corporate sector net of increases in firms’ bank balances that capture the contribution of the banking system to economic performance. It is unsurprising then that research at the World Bank such as Beck & Levine, 2000, fails to find any 6

connection between having a market-based or bank-based system and resulting performance in (financially) externally dependent industries. 3

The net sources

approach pulls aside the “veil” of gross financial market characteristics, and hence allows examination of the origin of financial flows for corporate investment and growth.

The other significant result in the net sources literature is especially relevant here. Cobham, Cosci & Mattesini (1999) use company balance sheets instead of flow of funds data (an approach this paper also follows), in order to apply the approach to Italy. This allows them to discriminate by size of firm, rather than treating the corporate sector as a homogenous unit. The results show that smaller firms make the greatest use of new equity, and less of internal finance. In the aggregate, the authors argue that the Italian combination of relatively high internal finance with relatively high debt should not be view as occupying an intermediate point between the AngloSaxon systems and that of Japan. Rather, their results are suggestive of an alternative system – one which ‘has much in common with France, and the two countries could be viewed together as having a ‘Mediterranean’ type of financial system’ (p.342). This view is considered further here.

II.ii: Criticisms of the approach

The net sources approach has not been without its critics. Hackethal & Schmidt (2000) seek to reclaim the ‘bank-based’ tag for Germany, and raise a number of objections to the Mayer and Corbett & Jenkinson methodology and results (MCJ hereafter). They show that MCJ are ‘double-netting’ their debt flows because as well as netting off sources against uses, they use data in which repayments are already netted off outstanding loan va lues. Using a simple model in which firms finance 70% of all physical investment with loans, they show how the MCJ approach can make (net) internal finance appear to be responsible for 100% of investment. 4

3

It is perhaps surprising though that the question continues to be asked, when rejection of the dichotomy seems well-accepted among researchers examining mainly industrialised economies. 4 Proceeding on the basis that loan maturity in years is equal to the number of firms or sectors, and that the latter invest sequentially and only one per year. 7

Hackethal & Schmidt carry out an alternative analysis, using assumptions about loan maturity to calculate the contribution to investment financing of gross (of repayments) loan contributions; these they find to be much higher than are those of MCJ. Green, Murinde & Suppakitjarak (2002) accept the validity of Hackethal & Schmidt’s central criticism, but argue that: ‘[t]he problem with this approach is that it is entirely predictable that gross flows make an arithmetically larger contribution to gross investment than do net flows, and the assumptions needed to derive gross figures using aggregate data necessarily border on the heroic’ (p.8).

Two further points can be made in response to Hackethal & Schmidt. Firstly, the results of the simple model are noticeably altered by each of the following: (i)

introduction of a growth trend – the model assumes zero economic growth; and

(ii)

allowing for debt maturity in excess of half the length of the economic cycle.

The other major criticism of Hackethal and Schmidt (2000) is a more general one: that despite the focus on the reconstruction of ‘true’ gross loan figures – which is, as noted, problematic – the central point is an insistence on examining gross rather than net flows.

Hackethal & Schmidt seem not to address the arguments (above) for using net rather than gross data, and at least if the question under consideration concerns corporate investment then these cannot be overlooked. Green et al. (2002) do allow that for other questions the methodology advanced by Hackethal & Schmidt could be appropriate. This paper focuses on the patterns of investment financing in the largest EMU members, and hence the analysis of net rather than gross sources (reconstructed or otherwise) is essential.

One other criticism made of MCJ is that it ignores the likelihood (or necessity?) that firms will require greater quantities of non- investment funds in order to operate efficiently at a larger scale. Consider a small firm which has invested in order to grow, using external equity finance, for example. If it reaches twice its initial size, it may reasonably need twice the amount of working capital – in the form of a bank overdraft – as it did previously. If it also doubles its cash buffer stock, and if for simplicity this 8

holding was previously equal to the working capital overdraft, the resulting net flow of bank finance will be zero.

On the one hand, it can reasonably be held that this is eminently appropriate as the contribution of bank finance to the investment was indeed zero.

However, an

alternative view would be that if the (new, larger) working capital is absolutely necessary to the functioning of the firm post-expansion, then the role of bank finance is being greatly undervalued by the net sources approach. 5 As will be seen in the analysis that follows, it is not the case that firms of different sizes maintain the same financial ratios. However, it remains possible that important features of corporate financing are overlooked by focusing only on the net sources of funds for investment, and for that reason gross sources of finance are also reported.

II.iii: Flow-of- funds vs balance-sheet data and approaches

Perhaps the most valid point made by Hackethal & Schmidt (2000) is that the MCJ approach reveals more about the financing of the corporate sector as a whole tha n about that of individual firms. It will not shed light on the financing of individual firms – quite apart from the issue of netting repayments, the number and identities of firms in the corporate sector are changing over time. The unchanging cylindered samples of the BACH database mean that the issue of numbers and identities of firms is largely dealt with, while it is precisely the aim of this work to consider the financing of the sector as a whole.

Green et al. (2002) use this to support their preference for disaggregated (balance sheet) over flow-of- funds data. They argue that reservations about the international comparability of the former have been overstated, and point to the use of WorldScope and Global Vantage databases (respectively) for international comparison by McClure, Clayton & Hofler (1999) and Rajan & Zingales (1995). Green et al. also argue that the availability of published company accounts is now such that samples can be constructed that need to be smaller than those used in the construction of national accounts. However, in looking at France, Germany and Italy, these two

5

I am grateful to Valpy FitzGerald for this point. 9

papers used data on, respectively, only 99, 78 and 33 firms (McClure et al.) and 225, 191 and 118 firms (Rajan & Zingales).

Table 3 shows, respectively, three different ‘sources and uses’ methodologies for dealing with flow of funds data, and their approximate equivalent as assets and liabilities methodologies for dealing with balance sheet data. Balance sheet data provide only stock data on firms’ financial positio ns, but by taking differences in stocks from year to year flow variables can be straightforwardly generated. Note that the (sources- uses) approach of Singh & Hamid (1992) differs mainly in netting depreciation off both investment and internal funds.

This allows examination of

changes in net assets, rather than gross investment, which may capture better the component of firm behaviour which results in growth. It will however less accurately embody firms’ economic activity.

Green et al. (2002) raise one final question about the use of balance sheet data: that flow data from balance sheets may capture not only the financing of new investment but also a part of the stock adjustment process to some optimal capital structure. However, if new investment is financed in the preferred way according to some pecking order (following Myers, 1984), then the net flows financing that investment will capture this preference – even if differs from that previously captured in the company’s financial structure. The authors very clearly favour balance sheet data over flow-of- funds, but they do also point out that, for different research objectives, different data may be superior.

In this case, our focus is on the aggregate financing of different size classes of firm. The aim is to understand how these size classes – which play very different roles in the provision of employment and in their contributions to growth – finance their operations and hence generate macroeconomic outcomes. Cobham (2000, 2001) argued that smaller firms generally find the availability of financing less stable, and that the resultant uncertainty causes more volatile investment outcomes in terms of and growth and mortality rates. For this reason, returns to investment should be higher for larger size classes of firm, even if individual small firms grow faster than their larger counterparts.

In seeking to assess such hypotheses about the 10

macroeconomic outcomes of microeconomic behaviours then, it is necessary to consider data aggregated by size class – exactly what the BACH database provides.

II.iv: BACH sources and uses

Following the assets and liabilities approaches set out in table 3, we can use the BACH balance sheet format in order to construct gross and net sources breakdowns of the data. We differ from the approaches set out in a number of important ways.

A main distinguishing feature of the approach followed here concerns the definition of investment. Investment has been variously defined to include some or all of the following: the changes in tangible, intangible and financial fixed assets, depreciation, current investments and changes in inventories. The treatment of each requires care.

The distinction between tangible and intangible assets – where only changes in the former are releva nt for investment – seems unsustainable. It may have made sense (see the net tangible assets method in table 3) to net off increases in intangibles against shareholders’ funds, when the former category referred mainly to the ‘goodwill’ of a business which was acquired – effectively the amount by which the shareholders had ‘overpaid’ for the acquisition. Given the increasing importance of other intangibles – ranging from highly complex R&D investments, to simple acquisition of new computer software – this treatment of intangibles seems less and less likely to be sustainable.

As such, investment is here defined to include net

changes in both tangible and intangible fixed assets.

Financial fixed assets and current investments are excluded. The former include ‘five types…: •

shares in subsidiaries



shares in associated companies



shares in other investments



loans to subsidiaries, associated companies and other investments



other loans.’

6

6

Regulations given in Stolowy & Lebas (2002, p.483). 11

On balance, I do not consider that changes in these types of asset represent new, otherwise uncounted, investment.

If the resulting allocations of finance lead to

investments in the particular country in question, then ideally these will appear elsewhere in the BACH database; including this as new investment would then involve a form of double-counting. 7

“Current investments” are defined under international accounting regulations as ‘investments that are by their nature readily realizable and are intended to be held for not more than one year’ (IAS 25, §4, IASC 1994), and can be considered in four separate categories: ‘financial asset held for trading’, ‘held-to-maturity investments’, loans and receivables originated by the enterprise’ and ‘available for sale financial assets’ (IAS 39, §10, IASC 1998). 8 There is no strong basis for counting any of these as part of firms’ investment in productive fixed assets.

Unlike Singh and Hamid, we do not net off depreciation, so that it is gross investment which is used as the denominator. On theory grounds this can be justified as more directly relating to the creation of new assets, regardless of the extent of depreciation which is chosen to be written down in accounts. On practical grounds, it ensures a positive denominator (allowing meaningful gross and net sources of funds analysis) in rather more cases. 9 Finally, we exclude inventory investment on the grounds that it should be viewed as representing ‘current’ rather than fixed assets.

A second difference in our approach is that the structure and detail of the BACH balance sheet data allows us to follow an approach closer to that of the sources-uses methodology. In particular, rather than collapse equity issues and internal funds into “shareholders’ funds” (as in table 3), we are able to separate out three categories: -

Retained profits refer to (changes in) the profit or loss for the given year, and profit or loss brought forward, and include depreciation;

-

Reserves and provisions refer to (changes in) revaluation and other reserves and in pension-related and other provisions;

7

Note however that this leads inevitably to lower values of the main net sources, since these holdings of long-term financial assets must now be netted off the gross sources. 8 Both regulations quoted in Stolowy & Lebas, 2002, p.362.

12

-

Equity refers to (changes in) shareholders’ subscribed capital and the share premium account (which captures the effect of changes in the value of shares).

Panel (a) of table 4 shows the gross sources- uses (assets- liabilities) approach as it is applied here to the BACH data. Panel (b) shows the net sources-uses approach, of which the only controversial feature concerns the treatment of financial fixed assets. As noted, we do not treat these as investment. The BACH database (see table 5) gives the breakdown of financial fixed assets (C.3) into ‘Shares in affiliated undertakings and participating interests’ (C.3.1/3) and ‘Other financial fixed assets’ (C.3.8). The first of these is likely to refer to the first two of the five types of asset listed above, the second to the remaining three. We consider ‘Shares in affiliated undertakings and participating interests’ as holdings of equity, and therefore new flows should be netted off against new flows of own equity.

The allocation of ‘Other financial fixed assets’ is less straightforward. The category combines loans to related companies and to other investments with shares in other investments. As such, it is not possible to state in what proportion the values here should be netted against inflows of equity and long-term debt. Given that most of the equity out-stocks will be captured by the category ‘Shares in affiliated undertakings and participating interests’, we expect a significantly higher loan proportion and therefore allocate ‘Other financial fixed assets’ to long-term debt for the net sources analysis. As such, caution should be applied – in particular, the possibility should be borne in mind that flows of net equity may be over-valued at the expense of those of net long-term debt.

9

40 observations, or just over 1% of the total, were deleted from the sample where the value of the investment variable was negative; these do not appear to have been systematically distributed by size, country, sector or year. 13

III. Sources of finance for corporate investment

III.i: General results

Table 6 presents the results of applying to the BACH database sample the sources and uses methodology outlined above.

Panel (a), showing the breakdown of gross

sources, indicates that firms in general saw large expansions of external finance in the form of equity issues. This is consistent with the view of (the latter half at least of) the 1990s as a period of significant stock market development across Europe. Less obvious are the expansions of short-term bank credit and trade credit which also occurred.

Long-term credit saw smaller increases on average.

Spain saw

significantly larger expansion in their (gross) funding position than any other country, which is consistent with some catch- up in the country’s financial markets and also the controversial financial behaviour of her larger firms towards the end of the period (see e.g. Ontiveros, 2004).

Panel (b) shows the net sources pattern.

Netting off the equivalent uses of funds

leads to a notable reversal in the relative importance of sources. The increases in equity and short (bank and trade credit) are more than offset by increases in (respectively) long-term equity holdings, cash and short-term financial assets and trade debt owed. The result is that the most significant external source of net new funds for the corporate sector in each country is seen to be long-term debt. Note that the difference between total net sources and 100% reflects changes in inventories – i.e. industry in France and Germany managed to reduce their ratio of stocks to total assets by 2% and 4.5% of investment per annum respectively over the period, while in Italy and Spain the ratio increased by 7.2% and 1.4% of investment respectively.

Reserves and provisions form a part of internal sources, though they may not necessarily be destined for investment – while it is common practice to seek to place a share of cyclically high profits into provisions to allow profit smoothing, there are also uses such as pension provisioning which are more ‘real’ in the sense that they represent certain future liabilities. Retained profits are seen to be somewhat higher than the equivalent values obtained in previous research (see table 2), which reflects 14

one weakness of the BACH database: coverage of the variable ‘distribution of profit’ (dividends) is incomplete, meaning that to ensure the constructed sources are fully comparable this variable has been omitted. The resulting variable ‘retained profits’ therefore (unavoidably) overestimates the true contribution of this source to new investment.

In one sense, the results in panel (b) support the criticisms which have been made of the approach – that looking over the long-term will play down the contribution of debt through repayment netting, and that netting uses off sources will undervalue the contribution made by e.g. equity when both issues and holdings rise. The results also show the value of the approach however. Over the period, firms in each country have significantly expanded their equity issues and short-term debt.

Because equity

holdings and short-term credit owing have risen by yet more, both the gross and net sources results are necessary in order to see that firms are involved in a relatively higher volume of transactions under these headings, but have not been able to tap either source for greater values of funds which are freely available for investment. The contribution of external sources of funding is, on average, only positive for longterm debt.

III.ii: Firm size results

Table 7 shows the breakdown of results by size class of firm, and the clear patterns which emerge. While the gross contributions of short-term credit appear ambiguous, it is clear when short-term uses are netted off that in each country except France this is a positive source of investment funds for small firms but not large. The same pattern is observed with respect to trade credit and payments received (now aggregated) – see panels (a) and (b) of figure 5. By contrast, panel (c) shows that smaller firms are more reliant on long-term credit for their investment finance. The exception here is Italy, where the pattern indicates that smaller firms increased their long-term debt holdings (credit owed to the firms) more than their liabilities, while large firms increased their debt but reduced their long-term creditors.

Equity issues (gross values) were systematically higher for larger firms. The net 15

position (panel d of figure 5) shows that these same firms also had much higher equity holdings in other businesses, with the effect that equity became a significant net use of funds. Small firms on the whole faced equity as a small net use, although in France they had high levels of both gross issue but net use. Medium firms emerge as most likely – on average – to have benefited from equity as a net source of investment finance.

III.iii: Results across periods

To analyse the results further, it is informative to consider the sources of funds over different time periods.

To guide the cho ice of period, output gap measures are

constructed for each country by taking the residuals from linear regressions of each country’s GDP (log) on lagged values of itself, a time trend and the time trend squared. Quarterly data from 1980Q1 to 2003Q2 were used, to minimise end-ofsample bias. German data was adjusted for reunification. As figure 6 shows, despite some differences, there are broadly similar patterns across the four sample countries. This, and the 1999 end of the German sample, suggests the following periods: 198893 (generally above trend GDP growth), 1994-99 (generally flat and below trend growth), and (for France, Italy and Spain) 2000-01 (return to above trend growth).

Table 8 shows the breakdown of net sources by country and firm size for this division of time periods. Of particular interest is the difference in external funding patterns between the first and second periods. Each six- year span captures different basic conditions – the first, above trend growth, and the second, below trend growth. Figure 7 shows how each firm size class in each country saw their use of long-term credit and of equity change (by subtracting the values for 1988-93 from those for 1994-99, so that positive outcomes indicate an increase in that source’s net contribution to investment finance). The difference between the panels is striking. Small firms everywhere saw the contribution of long-term debt fall sharply, while medium and especially large firms were relatively protected. The net contribution of equity followed almost exactly the opposite pattern: the changes tended to be positive, but in each case bar Italy were most positive (or least negative) for small firms. In the Italian case, large firms saw long-term debt fall but equity increase; this is the main 16

exception to the clear pattern that emerges.

The changes in net sources capture the outcome of three main processes: changes in preferred capital structure and changes in the availability of types of funds due to changes in financial market structure or in economic conditions.

It is not then

possible to immediately ascribe the pattern in figure 7 to the latter, especially when the process of European financial market integration was underway – however partially – during the second period at least.

Figure 8 shows the correlations between sources of finance, which imply differing pecking orders of finance. If firms are able to access external sources of finance to maintain investment levels during economic downturns, when profits are lower, the we would expect to see negative correlations between external and internal sources. If however firms are subject to rationing, then they will have more access to external finance during the upswing and hence we would expect to see complementarity between external and internal sources. Gertler & Gilchrist (1993) find evidence of such an effect in US credit markets, showing that during downturns it is smaller firms that see access reduced most strongly.

The correlations of new net external flows with internal finance is shown in panel (a). Long-term debt is complementary to internal finance for small firms, but a substitute for medium firms and more so for large. The implication is that firm size does predict rationing, in that long-term debt moves with internal finance for small firms but against for larger. All firms substitute equity and short-term debt for internal finance, though these effects are strongest for small firms. This implies that these sources of finance are not rationed. It is worth noting that smaller firms’ equity access is more likely to be in the form of insider or third-party equity, while larger firms are able to access public markets. That (more expensive and less stable) short-term debt is most strongly negatively correlated with internal finance for smaller firms suggests that a cost of the rationing of long-term debt is to force smaller firms to less suitable sources.

The net sources analysis suggests that smaller firms are indeed rationed in their use of 17

(long-term) debt, as expected. It also shows that the contribution of long-term debt to investment fell when the period 1988-93 is compared to 1994-99. To understand the implications for the levels and results of investment – to see whether we should be concerned – it is necessary to extract results about the role of external financing in investment decisions and outcomes. For this we turn to regression analysis.

IV. Results

IV.i: The cost of capital

As well as considering the role of net flows of finance, we wish in what follows to use a measure of the cost of capital. As the exact make- up of non- interest financing costs is unclear in the BACH data, the expectation must be that it may include a significant component which bears no direct relation to the costs of obtaining finance. Certainly, it is higher for larger firms which does not coincide with any rationing-related view and seems more likely to indicate firms’ choices of financing than costs they are forced to accept. For this reason then, the variable created captures only the interest costs of financing, and expresses these as a percentage of total debt.

The question of interest is the extent to which firm size and economic conditions determine this interest cost of capital. The equation tested is as follows:

CostKt = α 1Outgap t−1 + α 2 ∆Outgap t−1, t + α 3 Interest t −1 + α 4 ∆Interest t−1, t + α 5 ∆CPI t−1, t + α 6 LtDebt t + α 7 StDebt t + α 8 LtDebtStock t + α 9 StDebtStock t + α 10Turnovert−1 + α 11const … (1) where: CostK is the interest cost of capital to firms Outgap is the level of the output gap, ∆ Outgap the change Interest is the level of the real interest rate, ∆ Interest the change ∆ CPI is the change in the consumer price index, or inflation

LtDebt is the (net) share of new flows of long-term debt to the funding of new investment, n4i 18

StDebt is the (net) share of new flows of long-term debt to the funding of new investment, n1i+n2i+n3i LtDebtStock is the gross stock of long-term debt outstanding, as a % of total assets StDebtStock is the gross stock of short-term debt outstanding, as a % of total assets

The cost of capital in a given period is a function of economic conditions, firm size and firms’ financial characteristics. Higher levels of the output gap will be associated with previous above trend expansion, and therefore is expected to be associated with a higher cost of capital if money is tight at the end of a boom and easy after a downturn. Larger changes in the output gap will be associated with accelerating economic expansion and therefore a lower cost of capital. Both the lagged level and the change in (real) interest rates would be expected to feed directly into the cost of capital. Higher inflation will be associated with expectations of tighter monetary policy, and possibly with higher levels of economic uncertainty, which in turn would lead to a higher cost of capital.

The greater use of debt, both in general (gross stocks) but also in a given period (net flows) is expected to contribute to higher interest costs. With higher stocks this is evident, though with flows it may be less so – e.g. firms may use less debt when the cost is higher, so reverse causality is possible. For this reason, we consider the lagged values of both stocks and flows, as they contribute to the cost of capital at the end of that period (t-1).

Firm size is expected to be significant, finally, if the above

characteristics do not fully capture the differences in the cost of capital. Specifically, size will be negatively related to the cost of capital if credit rationing occurs more among smaller firms and if the cost of capital proxies for access to finance.

Model (1) in Table 9 shows the results of estimation of equation 1. The hypotheses above are all supported, though only weakly (at the 10% significance level) for the effect of net flows. These are dropped in model (2) without loss of explanatory power. In model (3), the output gap measures are replace by their interaction terms with dummies for whether the group involves firms of size 1 (small), 2 (medium) or 3 (large). If economic conditions impact firms of different sizes differently, we would expect to see statistically significant differences in the coefficients on these terms. 19

This is indeed the case.

Further testing of model (3)’s predictions reveals the

differences between the output gap level coefficients to be significant between the variables for sizes 1 and 2 (at the 5% level), but only weakly (at the 10% level) for size 1 and 3, and not at all for size 2 and 3. This, and the coefficients themselves, provide weak support for the view that previous above trend expansion increases the cost of capital more for medium and large firms. This is contrary to the findings of Gertler & Gilchrist (1993) that small firms suffer more from tighter money periods.

Looking at the change in the output gap yields strong support for their view however. Smaller firms are seen to benefit more from accelerating growth (and equivalently suffer most from slowing growth), and hypothesis testing reveals that the break is between small and medium firms on the one hand, and large firms on the other. The difference in coefficients between small and large firms is significant at the 1% level, as is that between medium and large firms.

This suggests that there is a much

stronger effect of constraints being lifted from smaller firms during periods of accelerating growth (relative to trend) than from their large counterparts. Firm size is negatively and significantly associated with the cost of capital, implying that there are reasons beyond the financial structure of firms and broader economic conditions which are responsible for smaller firms facing higher costs (and possibly less access) – the credit rationing view is supported.

IV.ii: The level of investment

We test to what extent the level of investment (as a percentage of total assets at the start of the period, which we label I t ) is determined by factors reflecting general economic conditions, firm size, performance and financial structure, and sectoral dependence on external finance: I t = α 1Outgapt −1 + α 2 ∆Outgapt −1,t + α 3 Interestt −1 + α 4 ∆Interestt −1,t + α 5 ∆CPI t −1,t + α 6 CostK 6 + α 7 ∆CostKt −1,t + α 8 LtDebtStock t + α 9 StDebtStock t + α 10 EqStockt + α 11 LtDebtt + α 12 StDebtt + α 13 Equityt + α 14 Intextt + α 15Turnovert −1 + α 16 const

…(2)

20

where: Equity is the (net) share of new flows of long-term debt to the funding of new investment, n5i EqStock is the gross stock of equity issued, as a % of total assets Intext is the ratio of internal finance to external finance in the funding of new investment, (s6+s7)/(n1+n2+n3+n4+n5). 10

Logical investment decisionmaking for a firm of any size would favour taking advantage of higher growth periods to invest more, but cutting back when (or before?) growth has peaked. We may predict, in line with Gertler & Gilchrist (1993), that smaller firms will see a bigger effect of economic and monetary policy conditions on their external financing and hence on their investment levels. The model of Cobham (2001) however suggests that while the return to smaller firms’ investments will be reduced by the uncertainty of obtaining financing during these periods, the actual effect on the level is in fact ambiguous. If small firms invest when they can – that is, when sufficient finance from whatever source is available – then their level of investment may be less sensitive to the economic cycle, if changing access to capital is captured by the interest cost. one interpretation however.

The cost of capital variables are open to more than Negative association with investment levels could

equally imply a role as a proxy for access, or alternatively a firm having some power over the timing of its investment and choosing to invest when the cost of capital is lower.

Comparison of the role of net flows and gross stocks of different sources of finance will shed light on the marginal investment decisions of firms.

If net flows of

individual sources are more strongly associated with investment, the implication is that those sources are used especially where investment is being increased. If gross stocks are more strongly associated with investment, it suggests either that these (if the coefficient is negative) represent obstacles to accessing further finance for investment, or alternatively, with a positive coefficient, that the existence of the stock (e.g. of issued equity) is associated with higher investment behaviour by managers. If firm size is positively related with investment, after allowing for financial structure, 10

NB The financing identity generates problematically high correlations between internal finance and the main external sources, so the information is conveyed by use of the ratio Intext instead. 21

this would imply that larger firms are able to invest at a higher rate than small. The reverse would suggest that the need for growth of smaller firms dominates that of large, consistent with the former being younger and further from their optimal size (or minimum efficient scale).

Table 10 presents the results. As we are interested here in how every variable varies across firm size classes, we do not carry out regressions on all sizes jointly with interaction terms as for the cost of capital, but instead separate the panel into three sub-samples: those of small, medium and large firms. For the latter, the first regression (model 5) test equation (2). However, the correlation between the net flows of long-term debt and equity exceed 0.6 and hypothesis testing indicates that the null of equal coefficients cannot be rejected. In models (6) and (7) these are therefore replaced by variables representing the same information: the sum and ratio of the two flows.

The output gap variables do not show a clear pattern. While both the level and change are positively associated with higher investment for small firms, for medium the level is negatively associated but the change positively. The latter pattern is consistent with firms investing more during higher growth periods, but less when the economy has peaked. Large firms follow the same pattern, but the results are insignificant. Small firms alone appear to invest more through the peak. This might be interpreted as giving weak support to the view of Cobham (2001) that small firms invest when they can. The level of real interest rates is also positively associated with investment for small firms, which seems similarly counterintuitive. However, these results disappear when the cost of capital is removed, which suggests that they do indeed imply that – allowing for changes in the cost of capital as a proxy for access – small firms do invest more in ‘worse’ economic conditions.

Small firms’ investment levels are most strongly reduced by their cost of capital, which supports the view of it as a proxy for access.

Medium firms see higher

investment when the interest cost of capital rises more quickly, which is hard to explain. Large firms’ investment reacts as strongly to the cost of capital as does that of small firms, which raises the problem above of making the correct inference. 22

Theory may suggest that in the case of the smaller firm, the role as proxy for access drives the result; but that the same result for large firms stems from their ability to be more flexible about the timing of their investments. The regressions do not allow us to distinguish with certainty.

Of the gross stocks of finance, only short-term debt for large firms is significant. The higher the stock, the lower the investment level, consistent with repayment taking priority. Among net sources, a pattern emerges by firm size. Large firms – for whom we cannot distinguish between the roles of long-term debt and equity – see a stronger association with investment from these sources than from short-term debt, though all are significant and the null of equal coefficients cannot be rejected. For medium firms, short-term debt and, more strongly, equity, are positively associated with investment – surpisingly, the share of long-term debt in net sources is not. For small firms, all three are significantly positively associated with investment, as would be expected. It is however long-term debt that plays the most important role, followed by equity and then the (only weakly significant) short-term debt. The implication is that long-term debt is the most important source of investment finance for small firms, but equity dominates for medium while the distinction is unimportant for large firms.

Finally, firm size itself (captured by the log of turnover per firm, lagged so as not to capture period t growth effects) is negatively and significantly associated with investment in the case of small and medium-sized firms. That is, smaller firms in these size classes – but not that of large firms – invest relatively more. This is consistent with a view of smaller firms in these size classes seeking growth to reach some minimum efficient scale, a position already achieved by a higher proportion of large firms.

IV.iii: Investment and growth

The final stage of this analysis is to examine how investment affects the growth of firms. We regress growth on lagged investment, the output gap, real interest rate and (log of) lagged turnover per firm. The output gap and interest rate are intended to capture basic economic conditions as simply as possible, as the focus is on the role of 23

firm size and in particular, investment. We expect tighter monetary policy and higher levels of economic expansion above trend to be associated with lower growth. Logged turnover per firm will capture the simple effects of firm size. The coefficient on the investment level, if significant, should indicate whether the growth response to further investment is positive or negative.

Models (1) and (3) in table 11 show the results for this basic model for (proportional) growth of sales and, as an alternative measure, value added. The output gap and real interest rate act as expected, with higher levels of each associated with lower firm growth. The firm size variable is negative and strongly significant, indicating that – all other things being equal – firms’ growth is negatively related to size: smaller firms grow more slowly on average. Investment is significant and negatively associated with sales growth, but insignificant in determining growth of value added.

The

former result implies that higher investment restricts sales growth in the subsequent period, suggesting investment beyond the optimal level.

Models (2) and (4) show the impact on the results of substituting for the investment level three interaction terms of investment level with size (simply small, medium or large). The results indicate the differential responses of growth to investment, by firm size class.

Looking at sales growth shows that while the interaction term with

investment is insignificant for small and medium firms, it is negative and significant a the 1% level for large firms. The negative result for investment overall derives then from large firms, so if there is overinvestment it is by this size class. The same result emerges even more clearly from the value added regression: the insignificant investment result hid a pattern of positive and significant small firm investment with negative and significant large firm investment. The simple implication is that large firms are over- investing in relation to the optimum while small firms under invest.

24

V. Conclusions

Two main sets of results have been presented in this paper. The net sources of funds analysis showed the importance of short-term and trade credit for small (as against larger) firms, especially in Germany. The contribution of long-term credit as a net source shows the clearest pattern of being inversely related to firm size, while equity is net use of funds for both small and large – only for medium- sized firms does this differ. Comparing 1994-99 with 1988-93, small firms in each country, but above all in France, saw a notable reduction in their ability to tap long-term credit for new investment financing; medium firms were less affected, and large firms least of all. The pattern in equity was somewhat reversed – in France, Germany and Spain (though the latter saw falls in each size class), the pattern was of greater expansion (or smaller reduction) among smaller firms. Italy was the exception, as small firms saw across the board reductions in their external financing.

Although there is evidence of a turnaround in these patterns with the continuing upturn in 2000-01, the concern remains that long-term debt is falling for smaller firms in particular. The positive correlation of long-term debt with internal finance for small firms only implies that they are able to expand their use of this source only when profits are already high; that is, rationing varies over the economic cycle and is strongest during downturns when profitability is also low.

This may be of particular concern as the patterns in table 8 show that greater reliance on debt financing characterises the small firm in each country (over the sample period) bar Italy. High levels of acquisition of financial assets has seen low net use of external funds across the sample, leaving Italian and Spanish firms’ use of long-term credit as the main (external) net source. The latter’s financial strategies (which have contributed to high non-production related profits) are responsible for the large role of equity as a net use of funds (see Ontiveros, 2004) and also the high internal finance flows.

Regression analysis yielded rather more precise results. The (interest) cost of capital was shown to be most sensitive to changes in the output gap for smaller firms, 25

implying that instability has higher costs for them in terms of access to credit. This is over and above the direct inverse effect of firm size on the cost of capital. Regressions on the level of investment showed further that the cost of capital is most strongly associated with reduced investment among smaller firms. After allowing for this effect, the level of investment was seen to be least sensitive to changes in economic conditions among small firms. This provides support for the view that the effect of the threat of financial constraints hanging over smaller firms is to reduce their ability to wait, as argued by Cobham (2001).

The contribution of different net sources of external finance to investment levels was seen to differ clearly across firm size class. Large firms drew no greater return in investment level from any individual source, as would be expected if they face least financial market imperfections.

Medium firms convert the net equity issues into

investment most readily, while for small firms long-term debt plays the most important role.

Finally, a simple analysis of the contribution of investment to

subsequent growth found that large firms appear to over- investment while small firms in contrast underinvest.

While a great deal of confirmatory work is required, the findings of this paper should raise concerns about the potential disintermediation associated with European financial market integration.

Small firms appear most exposed to economic

instability, which given their role in providing employment especially may make the eurozone’s restricted counter-cyclical policy a more urgent area for discussion. That small firms would appear (a) to have most growth to offer from additional investment, (b) gain most benefit in additional investment from greater access to long-term debt and (c) saw a significant withdrawal of long-term debt over the second half of the 1990s, is surely cause for grave concern and should provide impetus for further research into the restructuring of the European banking system.

Small firms appear to require stable sources of long-term finance if investment volumes and returns are to be improved. Given their importance in employment provision, and current unemployment levels, such a goal seems eminently sensible for EMU member governments. While growth may currently stem primarily from larger 26

firms, this appears to be largely independent of their external financing (and even in spite of their investment levels).

Intervention then will be most effective where

smaller firms’ access to external finance is its intermediate target. The dangers of disintermediation should be re-emphasised in light of the analysis presented here.

27

References Atlee, J., 1962, ‘Comment’, in The flow-of-funds approach to social accounting – Appraisal, analysis and applications: Studies in Income and Wealth 26: NBER Report by the Conference on Research in Income and Wealth, Princeton: Princeton University Press. Banque de France, 1999, Information notice 114 (May). Bartelsman, E., S. Scarpetta & F. Schivardi, 2003, ‘Comparative Analysis of Firm Demographics and Survival: Micro-Level Evidence for the OECD Countries’, OECD Economics Department Working Paper 348. Beck, T. & R. Levine, 2000, ‘New firm formation and industry growth: Does having a market- or bank-based system matter?’ mimeo, Washington, D.C.: World Bank. Cobham, A., 2000, ‘Making Bad Decisions: Firm Size and Investment under Uncertainty,’ Queen Elizabeth House Working Paper Series 39. Cobham, A., 2001, ‘EMU, monetary policy and the role of financial constraints’, EIFC (INTECH-UNU) Working Paper 01-6. Cobham, D., S. Cosci & F. Mattesini, 1999, 'The Italian financial system: Neither bank based nor market based', The Manchester School 67:3, pp.325-345. Corbett, J. & T. Jenkinson, 1996, 'The financing of industry, 1970-1989: An international comparison', Journal of the Japanese and International Economies 10, pp.71-96. Corbett, J. & T. Jenkinson, 1997, 'How is investment financed? A study of Germany, Japan, the United Kingdom and the United States', The Manchester School Supplement, pp.69-93. Gertler, M. & S. Gilchrist, 1993, ‘Monetary policy, business cycles and the behaviour of small manufacturing firms’, Finance and Economics Discussion Series 93-4, Washington, D.C.: Board of Governors of the Federal Reserve System . Green, C.J, V. Murinde & J. Suppakitjarak, 2002, ‘Corporate Financial Structures in India’, Economic Research Paper 02/4, University of Loughborough. Hackethal, A. & R. Schmidt, 2000 (revised 2003), 'Financing patterns: Measurement concepts and empirical results', Finance and Accounting Working Paper 33, Frankfurt: Johann Wolfgang Goethe-Universitat. Johannsson, D., 1999, ‘The number and the size distribution of firms in Sweden and other European countries,’ Stockholm School of Economics, Industrial Institute for Economic and Social Research Working Paper. Mayer, C., 1988, ‘New issues in corporate finance’, European Economic Review 32, pp.1167-1189. McClure, K., R. Clayton & R. Hofler, 1999, ‘International capital structure differences among the G7 nations: a current empirical view’, European Journal of Finance 5, 141-164. Myers, S., 1984, ‘The capital structure puzzle’, Journal of Finance 34, pp. 575-592. Ontiveros, E. (ed.), 2004, 1987-2003: Integración Económica y Financiera de España, Madrid: Escuela de Finanzas Aplicadas. Rajan, R. & L. Zingales, 1995, ‘What do we know about capital structure? Some evidence from international data’, Journal of Finance 50, 1421-1460. Singh, A. & J. Hamid, 1992, ‘Corporate financial structures in developing countries’, IFC Technical Paper 1, Washington, D.C.: International Finance Corporation. Stolowy, H. & M. Lebas, 2002, Corporate Financial Reporting: A Global Perspective, London: Thomson. 28

Table 1: Industrial sector coverage of the BACH database #

Sector name

NACE codes

11

Energy and water, including refining activities

10+11+12+23+40+41

12 1 2 3 13 14 4 5 6 7 8

Manufacturing industry Intermediate products Extraction of metalliferous ores and preliminary processing of metal Extraction of non-metalliferous ores and manufacture of non-metallic mineral products Chemicals and man-made fibres Investment goods and consumer durables Manufacture of metal articles, mechanical and instrument engineering Electrical and electronic equipment including office and computing equipment Manufacture of transport equipment Non-durable consumption goods Food, drink and tobacco Textiles, leather and clothing Timber and paper manufacture, printing Other manufacturing industries not elsewhere specified Building and civil engineering

13+27.1+27.2+27.3+27.4 14+26 24 27.5+28+29.1-29.6+33 30+31+32+29.7 34+35 15+16 17+18+19 20+21+22 25+36 45

Trade 9 15 10 16 17 18

Wholesale trade, recovery services Sale of motor vehicles, wholesale and retail trade Retail trade Hotels, restaurants Transports and communications Other services not elsewhere specified

51 50.1+50.3+50.4 52.1-52.6+50.5 55 60+61+62+63+64 50.2+52.7+67+70+71+72 +73+74+75+80+85+90 +91+92+93+95

Source: Banque de France, 1999.

Table 2: Net sources of finance, 1970-94 (% of physical investment) Internal Bank finance Bonds New equity Trade credit Capital transfers Other Statistical adjustment

Germany 78.9 11.9 -1.0 0.1 -1.2 8.7 1.4 1.2

Italy 1 83.7 20.9 -6.4 4.7 -6.5 -3.6 --

Japan 69.9 26.7 4.0 3.5 -5.0 -1.0 0.0

Sources: Corbett & Jenkinson, 1997 except Italy (Cobham et al., 1999). Notes. 1: Data for Italy are for the period 1983-93 only.

29

UK 93.3 14.6 4.2 -4.6 -0.9 1.7 0.0 -8.4

US 96.1 11.1 15.4 -7.6 -2.4 --4.4 -8.3

Table 3: Sources and uses methodologies Panel (a) Sources and uses methods Sources (s) Gross sources-uses method

Uses (u)

1 EAITBD1 2 3 decrease in inventories 4 equity issues 5 non-bank debt 6 bank debt 7 other current sources2 8 Total sources (=ΣsI) Net asset growth method (Singh & Hamid, 1992)

dividends paid gross fixed investment increase in inventories equity purchases non-bank debt purchases cash and deposits other current uses2 Total uses (=Σui)

1 EAITBD1 – depreciation - dividends paid 2 3 4 equity issues 5 non-bank debt 6 long-term bank debt 7 8 Total sources (=Σsi) Net sources-uses method (Corbett-Jenkinson)

gross fixed investment - depreciation - asset sales increase in inventories - decrease in inventories equity purchases non-bank debt purchases cash and deposits - short-term bank debt other current uses2 - other current sources2 Total uses (=Σui)

1 2 3 4 5 6 7 8

EAITBD1 - dividends paid Gross fixed investment Increase in inventories - decrease in inventories Equity issues – equity purchases non-bank debt - non-bank debt purchases bank debt - cash and deposits Other current sources1 - other current uses1 Total sources (=ΣsI)

Total uses (=Σui)

Panel (b) Balance sheet methods Liabilities (d) Total assets method 1 2 3 4 Shareholders' funds 5 non-bank debt 6 Bank debt 7 Other current liabilities2 8 Total liabilities (=Σ di) Net assets method 1 2 3 4 Shareholders' funds 5 non-bank debt 6 Bank debt 7 8 Total liabilities (=Σ di) Net tangible assets method 1 2 3 4 Shareholders' funds – intangibles 5 Non-bank debt - security holdings 6 Bank debt - cash and deposits 7 Other current liabilities2 - other current assets2 8 Total liabilities (=Σ di)

Assets (a) net fixed assets intangibles inventories equity investments security holdings cash and deposits other current assets2 Total assets (=Σa i) net fixed assets Intangibles Inventories equity investments security holdings cash and deposits other current assets2 - other current liabilities2 Total assets (=Σa i) Net fixed assets Inventories Equity investments

Total assets (=Σa i)

Source: Green et al. (2002). Notes: 1. Earnings after interest and tax, before depreciation. 2. Including trade credit. 30

Table 4: Balance sheet structure in the BACH database BACH code A. C. C.1 C.1.1 C.1.5 C.2 C.2.1 C.2.2 C.2.3 C.2.4 C.3 C.3.1/3 C.3.8 D. D.1 D.1.1 D.1.4 D.1.5 D.2 D.2.1 D.2.7 D.3 D.4 E. AE.

Balance sheet - assets (% of total assets) Subscribed capital unpaid Fixed assets (C = C.1 + C.2 + C.3) Intangible fixed assets (C.1 = C.1.1 + C.1.5) Formation expenses Other intangible fixed assets Tangible fixed assets (C.2 = C.2.1 + C.2.2 + C.2.3 + C.2.4) Land and buildings Plant and machinery Other fixtures Payments on account and assets in construction Financial fixed assets (C.3 = C.3.1/3 + C.3.8) Shares in affiliated undertakings and participating interests Other financial fixed assets Current assets (D = D.1 + D.2 + D.3 + D.4) Stocks (D.1 = D.1.1 + D.1.4 + D.1.5) Raw material and consumables Payments on account Other stocks Debtors (D.2 = D.2.1 + D.2.7) Trade debtors Other debtors Current investments Cash at bank and in hand Prepayments and accrued income Total assets (A + C + D + E) Balance sheet - liabilities (% of total assets)

F. F.2 F.3 F.4 F.10 F.101 F.102 I. I.1 I.2 I.4 I.10 I.101 I.102 J. J.1 J.4 K. L. L.1 L.2 L.3 L.4 L.5 L.6 FL.

Creditors: amounts payable within one year (F = F.2 + F.3 + F.4 + F.10) Amounts owed to credit institutions Payments received on accounts of orders Trade creditors Other creditors (F.10 = F.101 + F.102) Other financial creditors Other non financial creditors Creditors: amounts payable after more than one y ear (I =I.1 + I.2 + I.4 + I.10) Debenture loans Amounts owed to credit institutions Trade creditors Other creditors (I.10 = I.101 + I.102) Other financial creditors Other non financial creditors Provisions for liabilities and charges (J = J.1 + J.4) Provisions for pensions and similar obligations Other provisions Accruals and deferred income Capital and reserves (L = L.1 + L.2 + L.3 + L.4 + L.5 + L.6) Subscribed capital Share premium account Revaluation reserve Reserves Profit or loss brought forward Profit or loss for the financial year Total liabilities (F + I + J + K + L)

Identity

A + C + D + E = F + I + J + K + L = AE = FL = 100%

31

Table 5: Methodology used to calculate gross and net sources and uses of finance Panel (a): Gross sources method GROSS SOURCES (LIABILITIES)

GROSS USES (ASSETS)

s1. Short-term bank/OFI debt

u1. Short-term (non-trade) debtors plus cash plus current investments u2. Trade debtors

s2. Trade credit s3. Payments received (payments received on account plus accruals and deferred income)

u3. Payments made (payments made on account plus prepayments and accrued income) u4. Inventories

s4. Long-term bank/OFI debt s5. Equity issued

u5. Subscribed capital unpaid plus shares in participating and affiliated interests

s6. Reserves and provisions s7. Retained profits after tax and financial charges, before depreciation u6. Financial fixed assets (excluding equity positions in related enterprises) u7. Investment: net fixed assets (tangible and intangible) before depreciation Identity: s1+ s2 + s3 + s4 + s5 + s6 + s7 = u1 + u2 + u3 + u4 + u5 + u6 + u7

Panel (b): Net sources method NET SOURCES (LIABILITIES) n1=s1-u1. Short-term bank/OFI debt net of short-term (nontrade) debtors, cash and current investments

‘NET’ USES (ASSETS)

n2=s2-u2. Trade credit net of trade debtors n3=s3-u3. Payments received (payments received on account plus accruals and deferred income) net of payments made (payments made on account plus prepayments and accrued income) 4. Inventories n4=s4-u6. Long-term bank/OFI debt net of financial fixed assets (excluding equity positions in related enterprises) n5=s5-u5. Equity issued net of subscribed capital unpaid and shares in participating and affiliated interests s6. Reserves and provisions s7. Retained profits after tax and financial charges, before depreciation 7. Investment: net fixed assets (tangible and intangible) before depreciation Identity: (s1 - u1) + (s2 - u2) + (s3 - u3) + (s4 - u6) + (s5 - u5) + s6 + s7 = u4 + u7

NB. Tables show stock variables from BACH balance sheet structure, which are used in the analysis to generate flow variables from two-year cylindered data, adjusting for inflation. The suffix ‘i' is used to denote that variables are given as a fraction of investment u7, e.g. n5i = n5/u4 and so on.

32

Table 6: Means, 1988-2001 (1988-99 for Germany) Panel (a): Gross sources (liabilities) France

Germany

Italy

Spain

Total

Short-term (non-trade) credit

s1I

13.7

22.3

14.9

30.8

18.5

Trade credit

s2I

9.5

2.1

15.3

17.0

11.2

Payments received

s3i

-2.2

-4.0

0.2

-0.5

-1.5

Long-term (non-trade) credit

s4i

8.2

2.8

4.0

11.7

6.1

Equity issued

s5i

15.0

14.3

8.5

34.6

15.5

Reserves and provisions

s6i

27.4

14.5

18.3

20.5

20.4

Retained profits

s7i

96.4

104.9

102.5

128.4

105.2

168.0

156.8

163.7

242.6

175.3

Total sources

Panel (b): Net sources (liabilities) France

Germany

Italy

Spain

Total

Net short-term credit*

n1i

-8.1

-3.5

-7.8

-18.4

-8.6

Net trade credit

n2i

2.6

0.1

-8.8

0.6

-2.3

Net payments received

n3i

-1.2

-5.0

2.5

-1.1

-0.7

Net long-term credit

n4i

3.2

0.3

9.5

6.4

5.3

Net equity issued

n5i

-22.2

-15.7

-8.9

-35.0

-18.0

Reserves and provisions

n6i

27.4

14.5

18.3

20.5

20.4

Retained profits

n7i

96.4

104.9

102.5

128.4

105.2

98.0

95.5

107.2

101.4

101.3

Total sources

Results are mean values for sector and country (weighted by total assets) * Short-term (non-trade) credit net of short-term (non-trade) debtors, cash and current investments.

33

Table 7: Means by firm size, 1988-2001 (1988-99 for Germany) Panel (a): Gross sources (liabilities) Gross short-term credit

Gross trade credit and payments received

Gross long-term credit

Gross equity issued

Reserves and provisions

Retained profits

Total gross sources

Small Medium Large Small Medium Large Small Medium Large Small Medium Large Small Medium Large Small Medium Large Small Medium Large

France 7.9 16.3 13.7 6.5 14.2 6.6 26.8 9.3 7.0 11.4 19.6 15.7 23.7 23.8 30.0 91.9 90.1 97.9 168.2 173.4 171.0

Germany 23.2 24.3 21.9 30.3 8.9 -7.4 15.3 6.8 1.6 1.6 8.5 14.9 6.8 5.1 17.6 87.6 94.4 106.3 164.8 148.0 155.0

Italy -7.0 37.1 3.2 -47.4 22.0 17.7 7.0 9.4 3.4 -0.9 10.0 9.1 -3.7 20.3 20.5 85.3 82.1 125.4 33.2 180.9 179.3

Spain 23.4 12.5 27.0 25.9 24.8 15.7 19.0 12.9 -9.0 2.2 15.1 15.0 36.2 31.4 18.0 83.1 72.3 159.5 189.9 169.1 226.2

Small Medium Large Small Medium Large Small Medium Large Small Medium Large Small Medium Large Small Medium Large Small Medium Large

France -10.1 -6.6 -9.1 1.6 -3.5 1.5 25.5 7.6 0.6 -24.6 -0.3 -27.7 23.7 23.8 30.0 91.9 90.1 97.9 108.2 111.2 93.2

Germany 13.3 8.1 -5.5 23.7 0.1 -8.9 11.8 3.2 -1.0 -1.2 0.3 -19.9 6.8 5.1 17.6 87.6 94.4 106.3 141.9 111.2 88.6

Italy 1.2 21.5 -31.9 1.0 -17.1 -5.8 4.5 6.7 12.5 -3.7 0.8 -14.6 -3.7 20.3 20.5 85.3 82.1 125.4 84.6 114.3 106.0

Spain 7.8 -0.6 -26.1 2.4 0.0 -3.0 17.3 14.0 -2.0 -7.4 -1.8 -49.8 36.2 31.4 18.0 83.1 72.3 159.5 139.4 115.3 96.5

Panel (b): Net sources (liabilities) Net short-term credit

Net trade credit and payments received

Net long-term credit

Net equity issued

Reserves and provisions

Retained profits

Total net sources

Results are mean values for sector and country (weighted by total assets) * Short-term (non-trade) credit net of short-term (non-trade) debtors, cash and current investments.

34

Table 8: Net sources of funds France

Germany

Italy

Spain

1988-93 1994-99 2000-01 1988-93 1994-99 1988-93 1994-99 2000-01 1988-93 1994-99 2000-01 Short-term credit

Trade credit

Long-term credit

Equity

Small

-22.5

-1.0

-12.4

20.8

Medium

-14.9

Large

-12.2

-1.2

0.8

0.2

-36.2

Small Medium

-1.0

2.6

-7.0

-1.8

Large

1.9

Small Medium

38.8

12.2

7.8

41.3

-1.5

-7.2

38.3

6.3

22.4

22.3

-0.7

1.7

-0.6

-0.1

-23.8

-8.3

27.1

-2.9

-15.0

-12.3

72.2

-4.7

13.4

13.2

9.3

12.7

-34.9

-19.3

22.7

-8.7

18.6

9.5

22.8

12.0

-14.8

7.5

-35.6

-98.6

9.5

-67.8

11.2

11.9

-10.5

1.8

5.8

-6.5

-10.9

-19.3

3.9

-1.3

-7.1

3.0

-20.5

-14.0

11.7

-14.4

-3.5

-12.6

26.3

11.8

6.2

13.9

-0.7

23.4

5.6

1.2

7.4

8.7

0.4

12.6

8.3

34.5

Large

1.8

1.2

9.0

-1.6

-1.0

-6.4

-9.4

75.1

-75.4

-47.3

126.4

Small

-76.8

6.2

7.8

-4.2

3.8

-7.2

-5.1

3.5

-0.4

-9.1

-21.2

Medium

-12.2

3.0

20.9

-3.2

4.1

-4.6

-2.0

12.9

7.9

-4.1

-20.7

Large

-35.0

-24.4

-18.1

-9.9

-27.4

-20.2

-1.9

-30.2

29.9

-104.2

-43.5

39.7

14.5

10.7

17.1

-5.1

7.8

-21.7

10.5

38.6

34.6

35.6

Reserves and Small provisions Medium Retained profits

3.6

33.2

19.4

9.9

13.3

-1.3

22.5

13.5

29.9

31.7

28.0

38.4

Large

34.2

22.9

56.1

16.5

19.1

27.4

7.5

38.5

-1.4

49.0

-23.7

Small

93.0

93.3

93.6

78.9

99.7

59.9

121.8

68.8

52.6

101.5

112.4

Medium

87.4

99.0

72.2

85.8

103.4

68.1

96.1

73.4

39.0

102.1

74.1

Large

96.0

104.7

72.6

92.1

118.9

81.9

142.4

142.3

141.4

249.9

58.5

35

Table 9. Regression results for interest cost of capital

Independent variables Outgap, lagged Outgap, change Interest rate, lagged Interest rate, change Inflation Net flow of long-term debt, lagged Net flow of short-term debt, lagged Gross stock of long-term debt, lagged Gross stock of short-term debt, lagged Log of turnover per firm, lagged

(1) Interest costs 9.030 (9.55)*** -18.986 (15.89)*** 0.405 (39.92)*** 0.281 (26.16)*** 0.343 (19.10)*** 0.006 (1.82)* 0.008 (1.66)* 0.038 (11.22)*** 0.035 (7.84)*** -0.545 (6.14)***

(2) Interest costs 9.045 (9.56)*** -18.889 (15.82)*** 0.405 (39.92)*** 0.280 (26.11)*** 0.343 (19.09)***

5.547 (6.16)*** 2535 201 0.78

5.639 (6.27)*** 2535 201 0.78

0.038 (11.36)*** 0.035 (7.80)*** -0.555 (6.25)***

Size 1 dummy * Outgap, lagged Size 2 dummy * Outgap, lagged Size 3 dummy * Outgap, lagged Size 1 dummy * Outgap, change Size 2 dummy * Outgap, change Size 3 dummy * Outgap, change Constant Observations Number of sicosec R-squared

(3) Interest costs

0.404 (39.93)*** 0.279 (26.11)*** 0.339 (18.83)***

0.038 (11.27)*** 0.036 (7.96)*** -0.586 (6.54)*** 6.900 (5.17)*** 10.803 (7.92)*** 9.620 (7.10)*** -23.074 (12.66)*** -20.017 (10.97)*** -13.554 (7.41)*** 5.951 (6.55)*** 2535 201 0.78

Notes: *** indicates significant at 1% confidence level, ** at 5% and * at 10%. Regressions are fixed effects (R-squared is within-group value). Hausman test rejects random effects specification at 1% significance level and Breusch-Pagan LM test rejects pooling of groups, also at 1% level.

36

Table 10. Regression results for level of investment, u7 Independent variables Outgap Outgap, change Interest rate Interest rate, change Inflation Interest cost of capital Interest cost of capital, change Gross stock, longterm debt Gross stock, shortterm debt Gross stock of equity issued Net flow of long-term debt Net flow of short-term debt Net flow of equity Ratio, internal: external Log turnover per firm, lagged Net flow, long-term debt + equity Ratio, net long-term debt: equity Constant Observations Number of sicosec R-squared¹

Small u7

Small (2) u7

Small u7

11.002 (1.69)* 17.751 (2.17)** 0.234 (2.54)** -0.171 (2.19)** -0.408 (2.88)*** -0.595 (3.71)*** -0.008 (0.06) 0.002 (0.11) -0.010 (0.33) 0.009 (0.38) 0.323 (4.33)*** 0.098 (1.86)* 0.206 (3.86)*** -0.000 (0.22) -2.829 (5.33)***

10.701 (1.68)* 17.862 (2.25)** 0.228 (2.65)*** -0.168 (2.28)** -0.412 (2.97)*** -0.593 (4.36)***

6.931 (1.09) 35.853 (5.24)*** -0.034 (0.53) -0.086 (1.20) -0.587 (4.35)***

30.311 (6.70)*** 845 67 0.13

0.322 (4.35)*** 0.098 (1.88)* 0.199 (3.98)*** -2.865 (5.55)***

30.597 (6.88)*** 845 67 0.13

0.307 (4.21)*** 0.230 (4.71)*** -2.447 (4.81)***

26.270 (6.06)*** 845 67 0.10

Med’m (4) u7

Med’m (5) u7

-21.435 (3.08)*** 36.351 (3.77)*** -0.131 (1.32) -0.080 (0.97) -0.202 (1.32) 0.172 (0.98) 0.331 (2.29)** 0.034 (1.28) -0.039 (1.09) 0.048 (1.62) -0.041 (0.32) 0.210 (2.24)** 0.313 (3.31)*** 0.000 (0.50) -2.151 (1.90)*

-27.791 (4.91)*** 31.217 (4.28)*** -0.120 (1.93)*

25.720 (2.32)** 845 67 0.09

0.417 (3.27)***

0.200 (2.21)** 0.315 (3.41)*** -1.701 (2.35)**

22.080 (3.19)*** 845 67 0.08

Large (6) u7 -5.632 (0.80) 9.194 (1.09) 0.017 (0.20) 0.122 (1.38) 0.301 (2.89)*** -0.318 (2.58)*** 0.053 (0.38) -0.017 (0.86) -0.085 (4.39)*** 0.012 (0.50) 0.194 (2.39)** 0.121 (2.07)** 0.120 (1.75)* 0.001 (0.66) 0.024 (0.09)

6.337 (1.81)* 845 67 0.09

Large (7) u7 -5.873 (0.84) 10.411 (1.24) 0.022 (0.25) 0.131 (1.49) 0.305 (2.95)*** -0.324 (2.64)*** 0.050 (0.36) -0.018 (0.90) -0.085 (4.45)*** 0.016 (0.64) 0.094 (1.99)** 0.001 (0.66) 0.043 (0.16) 0.149 (2.57)** 0.005 (1.61) 6.069 (1.75)* 845 67 0.09

Large (8) u7

0.133 (1.95)* 0.297 (3.71)*** -0.376 (4.68)***

-0.085 (4.64)***

0.091 (1.97)**

0.145 (2.59)*** 6.966 (14.77)*** 845 67 0.08

Notes: *** indicates significant at 1% confidence level, ** at 5% and * at 10%. Regressions (1)-(5) are fixed effects: Hausman test rejects random effects specification at 1% significance level and BreuschPagan LM test rejects pooling of groups, also at 1% level. Regressions (6)-(8) are random effects: Hausman test cannot reject specification, Breusch-Pagan rejects pooling. ¹ The reported R-squared are within -group values for regressions (1)-(5), overall values for (6)-(8).

37

Table 11. Regression results for firm growth (1) Growth of sales Investment level, lagged Output gap Interest rate Log turnover, lagged

-0.001 (1.98)** -1.402 (21.86)*** -0.012 (15.74)*** -0.100 (17.25)***

Size 1 * Investment level, lagged Size 2 * Investment level, lagged Size 3 * Investment level, lagged Constant Observations Number of sicosec R-squared

1.049 (18.30)*** 2535 201 0.25

(2) Growth of sales

-1.402 (21.93)*** -0.011 (15.48)*** -0.098 (17.07)*** 0.001 (0.90) 0.000 (0.49) -0.003 (4.23)*** 1.034 (18.05)*** 2535 201 0.26

(3) Growth of value added 0.000 (0.16) -1.173 (16.08)*** -0.009 (11.24)*** -0.076 (11.60)***

0.799 (12.25)*** 2535 201 0.15

(4) Growth of value added

-1.173 (16.12)*** -0.009 (10.99)*** -0.075 (11.41)*** 0.002 (2.07)** 0.001 (1.30) -0.002 (2.56)** 0.784 (12.02)*** 2535 201 0.15

Notes: *** indicates significant at 1% confidence level, ** at 5% and * at 10%. All regressions are fixed effects: Hausman test rejects random effects specification at 1% significance level and Breusch-Pagan LM test rejects pooling of groups, als o at 1% level.

38

Figure 1. Relative firm size distributions in BACH sample as % of actual 10000

1000

100

10

1 France

Germany

Italy

Spain

Figure 2. Average firm size, 1999 (a) Turnover per firm, €m 1000

100

10

1

France

Germany

Italy

Small

3.0

3.3

5.6

2.1

Medium

15.3

17.2

14.3

12.7

319.1

567.1

207.6

388.4

Large

Spain

(b) Employees per firm 10000

1000

100

10

1

France

Germany

Italy

Spain

Small

26

26

32

27

Medium

89

90

65

96

1158

1916

893

1714

Large

39

Figure 3. Average firm size by industry sector, 1999 (a) Turnover per firm, €m

(b) Employees per firm

18

18

17

17

16

16

15

15

14

14

13

13

12

12

11

11

10

10

9

9

8

8

7

7

6

6

5

5

4

4

3

3

2

2

1

1 0

500

1000

0

1000

2000

3000

4000

(c) Total assets per employee, thousands of € 900 800 700 600 500 400 300 200 100 0

1

2

3

4

5

6

7

8

9

10

15

16

17

K/L ratio 207 295 149 246 149 178 126 213 284 117 838 238 193 200 329

74

307 183 235

40

11

12

13

14

18

All

Figure 4. Average firm size by sector and country, 1999 (a) Turnover per firm, €m 10000

1000

100

10

1 1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

6

7

8

9

10

11

12

13

14

15

16

17

18

(b) Employees per firm 10000 France Germany Italy Spain 1000

100

10

1 1

2

3

4

5

Source (figures 1-4) : BACH database, own calculations. Log scale used. Fig.1: BACH database averages for 1988-2001 (Germany 1988-99), ‘actual’ data: Eurostat for 1983-91 (Johansson, 1999).

41

Figure 5: Means by firm size, 1988-2001 (1988-99 for Germany) Panel (a) Short-term (non-trade) credit Gross

40

Net

30 Small

35

Medium

30

Large

20 10

25 20

0

15 -10

10 5

-20

0 -30 -5 -10

-40 France

Germany

Italy

Spain

France

Panel (b) Trade credit Gross

Germany

Italy

Spain

Italy

Spain

Net

40

25

30

20

20

15

10

10

0

5

-10

0

-20

-5 Small

-30 -40

-10

Medium Large

-15

-50

-20 France

Germany

Italy

Spain

France

Germany

Figure 5 continued Panel (c) Long-term credit Gross

Net

30

30

Small Medium

25

25

Large 20

20

15 15 10 10 5 5

0

0

-5 -10

-5 France

Germany

Italy

Spain

France

Panel (d) Equity issued Gross

Germany

Italy

Spain

Italy

Spain

Net

20

10 Small Medium Large

15

0

-10 10 -20 5 -30 0

-40

-50

-5 France

Germany

Italy

Spain

France

Germany

Figure 6: Output gap, 1988-2001 0.06 France Germany Italy Spain

0.05 0.04 0.03 0.02 0.01

-0.02 -0.03 -0.04

Figure 7: Change in net sources, 1988-93 to 1994-99 Panel (a): Long-term credit

Panel (b): Equity

40

100

Small Medium 20

Large

50

0 0

-20 -50

-40 -100

-60

-150

-80 France

Germany

Italy

Spain

France

Germany

Italy

Spain

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

1989

1988

0 -0.01

Figure 8: Correlations of net sources of funds with internal finance, by firm size

0.2 0

-0.2 -0.4 -0.6

Small

-0.8

Medium Large

-1 Short-term debt Long-term debt

Equity