Bank Restructuring in Post-Crisis Asia Don ... - Harvard University

19 February 1998

Test Report Title

Bank Restructuring in Post-Crisis Asia

Don Hanna & Yiping Huang

Paper prepared for the Asian Economic Panel Center for International Development Harvard University Cambridge, MA April 2000

The views expressed in the paper are our own and do not represent those of Salomon Smith Barney or Citibank. We would like to thank Wilbur Maxino and Adrienne Lui for able research assistance.  2001 by Don Hanna & Yiping Huang. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including  notice, is given to the source.

1

19 February 1998

Test Report Title

Bank Restructuring in Post-Crisis Asia Don Hanna & Yiping Huang April , 2001

Abstract We assess the quantitative effects in Asia of the 1997-1998 banking crisis at both the macroeconomic and bank levels, comparing our findings to broader studies of banking crises. Asia’s crisis was both more severe and more costly than others, but with little evidence of lasting bank runs. We than assess the nature of reforms within a structure that organizes measures between those designed to address externalities or market structure, those treating information asymmetries and those addressing legal infrastructure. We compare these measures to results from Barth, Caprio and Levine on the effects of banking structure and regulation on financial development, finding that reforms to date have under emphasized private monitoring and have concentrated too many assets in state-owned institutions. Despite extensive efforts at broadening legal infrastructure, progress has been poor, especially in Indonesia and Thailand.

Don Hanna Salomon Smith Barney/Citibank 20 fl, 3 Exchange Square 8 Connaught Rd Central, Hong Kong SAR, China [email protected]

2

Yiping Huang Salomon Smith Barney/Citibank 20 fl, 3 Exchange Square 8 Connaught Rd Central, Hong Kong SAR, China [email protected]

19 February 1998

Test Report Title

Introduction Almost four years have passed since much of Asia erupted into financial turmoil sparked by Thailand’s July 2 devaluation. The turmoil and contagion of that event has in turn sparked a host of papers seeking to understand the nature of the crisis. Authors have put forward a variety of alternative explanations. Some have seen the crisis as the result of excessive short-term capital inflows in the face of liberalization that drove real exchange rates out of line and increased fragility so that panic set in after Thailand’s devaluation 1 . Others have seen the crisis as the result of moral hazard created by an implicit or explicit backing of financial institution deposits in the face of poor lending decisions prompted by a poorly regulated system plagued by adverse selection. 2 There is one common thread in almost any explanation put forward: the financial system. This is obvious in stories that focus on excessive short-term foreign debt or in implicit financial sector guarantees. The Thai devaluation itself is closely linked to the inconsistency between the governments efforts at supporting a fixed exchange rate and its unwillingness to mount a sustained interest rate defense that would cripple its financial system. The depth and breadth of the crisis links to the depth and breadth of deterioration in the banks of four most heavily hit economies (see Figure 1) 3 . The ham-handed efforts at closing financial institutions in Indonesia were an important element of the bank runs that sparked massive liquidity support by Bank Indonesia in late 1997 and early 1998 4 . Further support for the importance of financial institutions can be gleaned from the respective stabilization programs, either IMF-supported or not. Each contained as a central pillar banking or financial reform (see Appendix A). Given the centrality of Asia’s financial system to its crisis, it is worthwhile to examine two sets of related issues. The first is to look at the actual performance of Asia’s banking sector over the crisis, both at the macro and at the bank level (see Section 2). The second is to analyze what policy steps have been taken in remaking Asia’s banking in the face of the macroeconomic and firm performance and to assess what the remaining agenda is for financial sector reform. 1

See Radelet and Sachs (1998) for this view There are various studies in this vane: Dooley (1998); Roubini et al.(1998); Meltzer. UCLA (2000) lay out a theoretical model of banking/currency crises that has at its heart the interaction of moral hazard and asymetric information. 3 Of course Figure 1 doesn’t tell us much about the causality between GDP growth and problem loans. Indeed, the poor growth performance of Hong Kong and Singapore in 1998 relative to their modest deterioration in the banking system hint that the whole story is not about Asia’s financial system. India and China’s woeful daunting bad loans, but resilient growth do the same, but from the opposite end of the spectrum. We return to this issue below. 4 On this see the post mortem by the IMF in Goldstein et al.(2000). The point is also made in Radelet and Sachs (1998). 2

3

19 February 1998

Test Report Title

Figure 1. Deviation from trend GDP growth and non-performing loans ratio to total loans Indonesia % 4 2 0 -2 -4 -6 -8 -10 -12 Mar-98

Malaysia % % 4 2 0 -2 -4 -6 -8 -10 Mar-98 Sep-98 Mar-99 Sep-99 Mar-00 Sep-00 Real GDP YoY growth: deviation from trend (left scale) NPL ratio (right scale)

% 60 50 40 30 20 10 0 Sep-98 Mar-99 Sep-99 Mar-00 Sep-00 Real GDP: deviation from trend (left scale) NPL ratio (right scale)

Korea % % 6 4 2 0 -2 -4 -6 -8 -10 Mar-98 Sep-98 Mar-99 Sep-99 Mar-00 Sep-00 Real GDP YoY growth: deviation from trend (left scale) NPL ratio (right scale)

20 15 10 5 0

Thailand 10 8 6 4 2 0

% 4 2 0 -2 -4 -6 -8 -10 Mar-98

% 50 40 30 20 10 0 Sep-98 Mar-99 Sep-99 Mar-00 Sep-00 Real GDP YoY growth: deviation from trend NPL ratio

Source: Bank of Indonesia CEIC, Salomon Smith Barney/Citibank estimates Note: NPLs are those of the banking system, except Korea, which is of the domestic commercial banks. Malaysia measure includes Islamic banks. Thailand measure excludes branches of foreign banks.

How have Asian banks fared? Methodology and data In order to explore economy-wide and bank-level responses during and after the Asian financial crisis. We construct two simple data sets, one economy-wide and another bank level. In the economy data set, we have data for nine Asian economies including China, Hong Kong, Indonesia, Korea, Malaysia, Philippines, Singapore, Taiwan and Thailand. All economies are included in the econometric exercise but are divided into two groups: crisis-affected and other Asian economies. This enables us not only to improve the statistical efficiency of the estimation, but also to capture the possible impacts of the crisis on the other Asian economies. The bank data set contains information for 29 banks from only three economies — Malaysia, Thailand and Hong Kong. Again, Hong Kong bank data were included for comparison. This data set is smaller than we would like, but does include two of the four hardest hit banking systems (Malaysia and Thailand). The data are for listed banks in each of the countries. This creates some bias as key state-owned banks or privately held banks are not listed5 . All the variables are expressed in percentage forms. For more details about the data and definitions of the variables, see the appendix.

5

4

Need table with sample bank assets/total bank assets.

19 February 1998

Test Report Title

To identify the impacts and responses of the economies and banks, we apply the following simple framework6 :

Yit = φC + α0 T0 + α1 T1 + α2 T2 + α3 T3 + β0 TN 0 + β1 TN1 + β2 TN 2 + β3TN 3

Where Yit is the dependent variable, such as GDP growth rate or the ratio of overhead expenses to total assets, being investigated for i-th economy/bank in t-th year. C is the constant. T0 , T1 , T2 , T3 are yearly dummies for 1997, 1998, 1999 and 2000, respectively, for crisis-affected economies or banks. Similarly, TN 0 , TN 1 , TN 2 , TN 3 are yearly dummies for the other Asian economies or banks. The regression also includes some country dummies to capture the country-specific effects. But the results of these are not reported for simplicity of discussion. The key of the econometric results is to look at the significance, sign and values of the coefficient

ˆ k ’s and βˆ l ’s. If the estimates are significant and negative, then it implies that the estimates α dependent variable dropped significantly in that particular year, in comparison with the pre-crisis levels. In order to resolve the heteroskedasticity problem of the data, we apply the White heteroskedasticity-consistent estimation approach.

Economy-wide experiences Much of what we find with regard to the effects on Asia of the crisis is by now well-known. The crises had significant negative impacts on Asian growth, almost by definition. GDP growth in the crisis-affected economies fell, on average, by 4.5 percentage points in 1997 (see Figure 2). But given that Thailand didn’t devalue until the middle of 1997, the damage was greatest in 1998: a fall of 17.9 percentage points. GDP growth was still 3 percentage points lower in 1999 than in the precrisis period. In 1997, growth of other Asian economies was not affected. But these economies lost 5.8 percentage points and 2.3 percentage points in 1998 and 1999, respectively. In 2000, GDP growth of both crisis-affected economies and other Asian economies recovered to their pre-crisis levels.

6

This methodology has been applied by Demirgüç-Kunt et al.(2000) to their broader (but dated) study of banking and currency crises. 5

19 February 1998

Test Report Title

Figure 2. . Responses of output, investment and bank deposit/credit (country data) GDP

Credit

Investment/

Deposit

Deposit/

Growth

growth

GDP ratio

growth

GDP ratio

-4.495***

-1.956

-2.366*

2.785

-0.125

(3.8)

(0.3)

(1.9)

(0.5)

(0.2)

1998

-17.880***

-27.481***

-16.726***

-6.825

-2.125***

(15.0)

(4.8)

(13.5)

(1.3)

(3.0)

1999

-2.950**

-15.478***

-17.673***

-6.874

-1.125

(2.5)

(2.7)

(14.3)

(1.3)

(1.6)

2000

-1.783

-14.980**

-13.971***

-1.162

0.125

(1.5)

(2.3)

(11.3)

(0.2)

(0.2)

-0.385

-1.89

0.867

-0.53

-1.083

(0.4)

(0.3)

(0.8)

(0.1)

(1.6)

1998

-5.819***

-14.792**

-2.351**

1.112

-1.333**

(5.5)

(2.6)

(2.1)

(0.2)

(2.0)

1999

-2.267**

-14.807**

-3.887***

-2.318

0.667

(2.1)

(2.6)

(3.5)

(0.5)

(1.0)

2000

0.455

-11.277*

-3.769***

-3.773

1.408**

(0.4)

(1.8)

(3.4)

(0.7)

(2.0)

7.185***

14.480***

25.474***

7.854**

39.492***

(8.2)

(3.5)

(27.9)

(2.1)

(77.9)

-125.06

-154.22

-127.51

-185.83

172.83

Crisis economies 1997

Other economies 1997

Country dummies not reported Constant Log likelihood

Note: Numbers in parenthesis are asymptotic t-statistic, and *, ** and *** indicate significance level of 10, 5 and 1 percent, respectively. The sample contains nine countries, including crisis-affected economies Indonesia, Korea, Malaysia, and Thailand and other Asian economies China, Hong Kong, Philippines, Singapore and Taiwan, and covers seven years, 1994-2000. The total number of observation is 63 Source: Salomon Smith Barney/Citibank estimates

The results above support the view that, although output declines were deep, the effects of the Asian financial crises on GDP growth were short-lived. In an earlier study on experiences of 35 countries, Demirgüç-Kunt et al. (2000) also found the impacts of the crisis on output growth to be short-lived. However, the magnitudes of the impact from their study (less than 4 percentage points in the first two years) were much smaller than what we discover here. The high estimates might be related to the high growth rates during the base years — in 1994, 1995 and 1996 most of the crisisaffected economies suffered more or less the overheating problem. Furthermore, Asia’s banking systems were very large in relation to GDP (see Figure 3). This could well have contributed to a greater disruption in firm financing due to the banking crisis.

6

19 February 1998

Test Report Title

Figure 3. Total Domestic Credits in the Economy (as Percent of GDP) 1994

1995

1996

1997

1998

1999

2000

Crisis Asia Indonesia

50.2

51.3

53.8

64.9

54.9

58.5

63.2

Korea

140.8

142.6

152.5

167.3

177.6

172.1

159.8

Malaysia

105.5

119.2

132.4

150.8

146.7

131.7

122.7

Thailand

101.0

108.2

111.6

141.6

140.2

135.0

117.6

Other Asian Economies China

42.1

39.9

39.2

43.4

50.3

54.6

n.a.

Hong Kong

124.5

129.8

137.4

153.9

155.7

147.8

n.a.

Philippines

48.5

56.9

69.4

79.2

69.8

64.2

n.a.

Singapore

58.8

62.3

67.1

73.5

88.1

84.8

79.4

176.8

179.0

177.2

175.1

178.1

180.4

177.8

Brazil

57.7

36.6

38.0

40.0

49.1

47.7

n.a.

Mexico

34.5

28.1

16.9

28.7

26.2

22.2

n.a.

Russia

31.7

23.6

25.1

26.6

41.1

32.7

n.a.

Turkey

25.6

28.0

34.2

36.9

40.0

45.3

n.a.

Taiwan Other Emerging Markets

Source: IMF, National Sources, Salomon Smith Barney/Citibank calculations

The impacts of the crisis on credit growth and investment/GDP ratios, however, last longer than the GDP effects. For the crisis-affected economies, credit growth stayed unchanged in the first year, 1997, but was 27.5, 15.5 and 15 percentage points lower than the pre-crisis levels in 1998, 1999 and 2000, respectively. The investment/GDP ratio fell even in the first year and declined more significantly in the following three years. In 2000, the ratio was still 14 percentage points lower than the pre-crisis level. The much greater effects on investment ratio compared to the findings from the study by Demirgüç-Kunt et al. (2000) were probably also related to the overheating and excessive investment in the pre-crisis years. Again Asia’s I/GDP ratios are far higher than those of the average country. The pattern of the impacts on the other Asian economies was basically the same, although the effects were much smaller in magnitude. The magnitude of the continued drop in investment to GDP, especially given the recovery in GDP back to pre-crisis levels is an issue to which we return below. Changes in bank deposits show a different story than credit, GDP or investment. The crises did not have any significant impact on deposit growth in either crisis-affected or other Asian economies. The deposit/GDP ratios dropped by 1-2 percentage points in the second year after the crisis began, 1998, in the two groups of economies but recovered in the following year. In other Asian economies, deposit/GDP ratios were actually 1.5 percentage points higher than the pre-crisis levels. The overall finding is that bank deposits were not seriously affected by the crisis. This suggests that systemic bank runs did not accompany panic in the financial sector. This may be attributable to the deposit insurance system or the implicit government guarantee of the deposits or the presence of banks perceived as safer (state-owned or foreign-owned). In Indonesia, for instance, the government introduced a partial guarantee policy for deposits in 1997 and then switched to a

7

19 February 1998

Test Report Title

blanket guarantee scheme in early 1998. Thailand had implicit guarantee policy for depositors and Korea had a deposit insurance system. There is also an argument that the annual data we applied here may not capture the short-lived bank runs. The finding is at odds with the typical pattern of deposit/GDP ratios in banking crises, which show slight increases, generally because GDP growth slumps more than deposit growth. Once again, this finding could be the result of the unusually large GDP effects of the Asian banking crisis. The impacts on real interest rate variables were also limited (see Figure 4). In crisis-affected economies, real interest rates (which is a policy variable) only increased significantly, by 7 percentage points, in the first year, 1997. They recovered quickly in the following year. In other Asian economies, the effects on real interest rates stay much longer, which were equivalent to 5.3, 6.2, 5.8 and 5.6 percentage points higher in 1997, 1998, 1999 and 2000, respectively. For some economies of this group, such as China and Hong Kong, the rises of real interest rates were probably because of price deflation. Inflation rates were substantially lower during the crisis years for this group of economies (this is shown later in analysis of inflation rate below). Figure 4. Responses of interest rates (country data) Real

Real

Real

Interest

Interest rate

deposit rate

lending rate

spread

6.834***

1.39

1.761

-2.205*

(2.8)

(0.2)

(1.6)

(1.9)

1.327

9

5.539***

0.98

(0.5)

(1.4)

(5.0)

(0.8)

1.542

-5.231

0.884

0.543

(0.6)

(0.8)

(0.8)

(0.5)

2.567

3.967

-1.966*

-1.675

(1.1)

(0.6)

(1.8)

(1.4)

5.278**

-3.449

0.193

0.462

(2.4)

(0.7)

(0.2)

(0.5)

1998

6.208***

-4.056

0.393

-0.194

(2.9)

(0.8)

(0.4)

(0.2)

1999

5.814***

-3.588

-1.509

0.104

(2.7)

(0.7)

(1.6)

(0.1)

2000

5.554***

-4.28

-1.667*

0.07

(2.6)

(0.9)

(1.7)

(0.1)

0.448

5.867

8.746***

2.912***

(0.3)

(1.4)

(11.1)

(3.7)

-169.64

-153.6

-114.23

-100.57

Crisis economies 1997 1998 1999 2000 Other economies 1997

Country dummies not reported Constant Log likelihood

Note: Numbers in parenthesis are asymptotic t-statistic, and *, ** and *** indicate significance level of 10, 5 and 1 percent, respectively. The sample contains nine countries, including crisis-affected economies Indonesia, Korea, Malaysia, and Thailand and other Asian economies China, Hong Kong, Philippines, Singapore and Taiwan, and covers seven years, 1994-2000. The total number of observation is 63. Source: Salomon Smith Barney/Citibank estimates

Real deposit rates, however, did not change significantly in either the crisis-affected or the other Asian economies. This suggest that households’ real income from their bank deposits stayed largely

8

19 February 1998

Test Report Title

stable during the years of the Asian financial crisis. This would also help explain the lack of much change in deposit to GDP ratios. The real lending rates rose significantly in the second year in the crisis-affected economies. They recovered, however, in the following year and became significantly lower, by 2 percentage points, than the pre-crisis levels in 2000. The other Asian economies’ real lending rates were also 1.7 percentage points lower than the pre-crisis levels in 2000, but they did not experience rise in previous years. In crisis-affected economies, the ex-ante interest spread also narrowed by 2.2 percentage points in the first year of the crisis but recovered in the following year. The other Asian economies witnessed no change in interest spread during the period of financial crisis. Finally, the inflation rates of the crisis-affected economies did not change in the first year of the crisis but experienced a significant increase, by 14 percentage points on average, in the second year (see Figure 5). They recovered to the pre-crisis levels thereafter. The inflation rates of the other Asian economies did not change during the first two years of the crisis but experienced sharp declines in the third and fourth years, by 7.3 and 7 percentage points respectively. These changes in the later years, again, reflected the deflation experienced by some of the economies. Figure 5. Responses of inflation rate, exchange rate and fiscal balance (country data) Inflation

Fiscal surplus/

Cent Bank funds/

rate

Depreciation

GDP

bank assets

-1.023

19.769

-0.449

4.976***

(0.3)

(1.3)

(0.5)

(3.9)

1998

13.947***

87.694***

-2.769***

2.873**

(3.7)

(5.9)

(2.9)

(2.2)

1999

0.399

-12.368

-4.352***

-1.512

(0.1)

(-0.8)

(4.6)

(1.2)

2000

-3.483

1.399

-3.449***

-0.138

(0.9)

(0.1)

(3.7)

(0.1)

x

x

x

x

1997

-4.481

2.174

0.947

-1.629

(1.3)

(0.2)

(1.1)

(1.1)

1998

-5.129

11.352

-0.293

-2.803*

(1.5)

(0.8)

(0.3)

(1.8)

1999

-7.349**

-3.538

-2.149***

-2.676*

(2.2)

(0.3)

(2.6)

(1.7)

2000

-7.013**

0.16

-2.632***

-1.296

(2.1)

0.0

(3.1)

(0.7)

x

x

x

x

5.546*

1.176

-4.112***

3.605***

Crisis economies 1997

Other economies

Country dummies not reported Constant Log likelihood

(2.0)

(0.1)

(5.9)

(3.4)

-197.96

-284.53

-110.16

-88.35

Note: Numbers in parenthesis are asymptotic t-statistic, and *, ** and *** indicate significance level of 10, 5 and 1 percent, respectively. The sample contains nine countries, including crisis-affected economies Indonesia, Korea, Malaysia, and Thailand and other Asian economies China, Hong Kong, Philippines, Singapore and Taiwan, and covers seven years, 1994-2000. The total number

9

19 February 1998

Test Report Title

of observation is 63. Source: Salomon Smith Barney/Citibank estimates

The crisis-affected economies also witnessed very sharp depreciation of their currencies, by an average of 87.7 percentage points compared to the pre-crisis levels, in the second year. The crisis, however, did not have statistically significant impact on currency values of the other Asian economies. This result was probably related to the virtual fixed exchange rates in China and Hong Kong, as we did observe exchange rate movements in Philippines, Singapore and Taiwan. China’s one-time depreciation at the beginning of 1994 may have also affected the results (see Figure 6). Figure 6. Asian Currencies Depreciation Depreciation (%) 50 40 30 20 10 0 -10 -20 1994

1995 China

1996 Hong Kong

1997 Philippines

1998 Singapore

1999

2000

Taiwan

Source: CEIC

The trajectories of both inflation rates and exchange rates of the crisis-affected economies were very different from the earlier experiences of the other economies. Demirgüç-Kunt et al. (2000), for instance, found persistent and significant impacts on inflation and exchange rate even in the fourth year of the crisis. For the Asian crisis economies, inflation rates surged and currencies depreciated only in the second year and then recovered to the pre-crisis levels. This finding hints at two possible explanations: first, the relative lack of real exchange rate misalignment in Asia prior to the crisis (see Hanna (2000)) and to initially strong fiscal positions that allowed the governments to absorb large losses in salvaging their financial systems without sparking inflationary expectations. The crisis had large direct impacts on the fiscal system. These were probably related to the expenditure on banking restructuring and other programs designed to cushion the blow from falling GDP (such as proactive fiscal policy in many economies), as well as from the falling GDP itself (see Figure 7). This was partly confirmed by the greater reduction in fiscal balances in these economies compared to the findings from the other sample economies (Demirgüç-Kunt et al, 2000). The crisis-affected economies’ fiscal balance as shares of GDP were reduced, on average, by 2.8, 4.4 and 3.4 percentage points, compared to their pre-crisis levels, in the second, third and fourth 10

19 February 1998

Test Report Title

year of the crisis period, respectively. Fiscal deficits of the other Asian economies also increased in the third and fourth year.

Figure 7. Indonesia -- Estimated fiscal cost borne from recapitalization of the banking system % of GDP 60

Rp671 tln

50

40

Rp453 tln

30

Rp218 tln

20

10

0 Total Bonds issued (estimated final)

Bank Indonesia

Banks 1/

Source: IMF Staff Country Report No. 00132 (Nov 2000) available at http://www.imf.org Note: 1/ Includes a Rp21 trillion contingency, partly to cover the cost of raising state banks' capital from 4% to 8% CAR by end-2001

Changes in the ratios of central banks funds for deposit money banks to banks’ total assets were quite different in the two groups of economies. The ratios increased significantly during the first and second year in crisis-affected economies and then recovered afterwards. The ratios were, however, reduced in the second and third year in the other Asian economies. The findings reflect the significant liquidity support given to the crisis country banks by their central banks.

Bank-level experiences In the next step we examine changes during at the bank level the crisis years. Such exercise is at least helpful in two respects. On the one hand, we can look at changes in some of the variables, such as interest rates and deposit/credit growth, to see if the bank-level data agree with the economy-wide data. If not, then there is probably an important structural question, in addition to the obvious sample bias issue. Second, the bank level data enable us to explore changes in profitability and portfolio shifts. Banks’ performance in crisis-affected economies (Malaysia and Thailand), measured by gross and net profitability, was not affected in the first year of the crisis (see Figure 8). But from the second year, both total and net returns on assets declined significantly, by 3.2 and 1.2 percentage points, respectively. This negative impact continued in 1999 and 2000 (1999 was the worst). Bank

11

19 February 1998

Test Report Title

performance in Hong Kong was affected in the same pattern, but the magnitudes of the effects were somewhat smaller compared to those in Malaysia and Thailand. The crisis also had important effects on interest margin (net interest income) and overhead costs from the second year of the crisis in Malaysia and Thailand. The increases in overhead costs in the second and the fourth year was somewhat puzzling as usually when banks run into crisis they are forced to improve their efficiency (and cut costs). One explanation is the rapid decline of real assets of the banks, which is evident in analysis below, outpacing the reduction in overhead expenses (see Figure 9). For Hong Kong banks, the negative impact on net interest income was evident. But there was no significant effect on overhead costs throughout the period. Figure 8. Crisis aftermath – evidence from bank level data (I) Return on average asset

Profitability

Interest margin

Overhead

-0.636 (1.3)

0.168

0.18

0.013

(0.5)

(1.0)

1998

-3.191***

0.0

-1.212***

-0.745***

0.489*

(6.6)

(3.4)

(4.0)

(1.6)

1999

-4.355***

-1.540***

-1.246***

0.264

(9.1)

(4.3)

(6.8)

(0.9)

2000

-1.659***

-1.152***

-0.379**

0.933***

(3.5)

(3.3)

(2.1)

(3.1)

-0.517

-0.529

-0.474**

0.054

(0.9)

(1.3)

(2.2)

(0.1)

1998

-1.716***

-0.870**

-0.626***

0.111

(3.1)

(2.1)

(2.9)

(0.3)

1999

-1.586***

-0.911**

-0.730***

-0.038

(2.9)

(2.2)

(3.4)

(0.1)

2000

-0.913*

-0.858**

-0.664***

-0.027

(1.7)

(2.1)

(3.1)

(0.1)

2.279***

2.718***

3.023***

1.277***

(9.3)

(15.1)

(32.3)

(8.4)

-336.86

-283.74

-170.2

-252.69

Banks of crisis economies (Malaysia and Thailand) 1997

Banks of other economies (Hong Kong) 1997

Country dummies not reported Constant Log likelihood

Note: Numbers in parenthesis are asymptotic t-statistic, and *, ** and *** indicate significance level of 10, 5 and 1 percent, respectively. The sample includes 29 banks from Hong Kong (12), Malaysia (9) and Thailand (8). Every bank has annual observation from 1995 to 2000. The total number of observation is 174. Source: Salomon Smith Barney/Citibank estimates

One phenomenon after the crisis began was the rapid rise of nonperforming loans. This resulted in sharp increases in loan loss provisions (from the first year) and loan loss reserves (from the second

12

19 February 1998

Test Report Title

year) (see Figure 9)7 . Loan loss provision also increased in the second and third year among Hong Kong banks, but their loan loss reserves did not change. Growth of banks’ deposits fell sharply in the second year, by 14.2 percentage points, and declined even further in the fourth year, by 18.7 percentage points. But because the growth rates were very high in the pre-crisis period (around 23%), deposits were still growing for most banks after the crisis. This decline deserves further exploration, as in our country-level analysis, the crisis did not have a statistically significant impact on growth of deposits (though the real deposit growth data do show a decline in 1998-2000). Figure 9. Crisis aftermath – evidence from bank level data (II) Loan loss provision

Loan loss reserves

Growth of real assets

Growth of real loans

Growth of real deposits

0.720*

0.425

4.822

-2.854

-5.116

(1.9)

(0.5)

(1.0)

(-0.7)

(0.8)

1998

2.100***

3.690***

-17.148***

-19.154***

-14.193**

(5.8)

(4.5)

(3.5)

(4.6)

(2.3)

1999

3.428***

6.220***

-4.207

-11.685***

4.091

(9.5)

(7.6)

(0.9)

(2.8)

(0.7)

2000

0.751**

3.145***

-18.929***

-27.763***

-18.725***

(2.1)

(3.8)

(3.9)

(6.7)

(3.1)

0.321

-0.649

-1.764

1.565

-7.201

(0.8)

(0.7)

(0.3)

(0.3)

(1.0)

1998

1.102***

0.472

-22.577***

-30.418***

-22.403***

(2.6)

(0.5)

(4.0)

(6.4)

(3.2)

1999

1.190***

1.427

-17.01***

-29.985***

-18.797***

(2.9)

(1.5)

(3.0)

(6.3)

(2.7)

2000

0.513

1.136

-11.083*

-18.409***

-14.526**

(1.2)

(1.2)

(2.0)

(3.8)

(2.1)

-0.022

2.161***

22.771***

28.053***

25.258***

(0.1)

(5.2)

(9.1)

(13.4)

(8.2)

-285.08

-427.4

-741.41

-711.64

-778.12

Banks of crisis economies (Malaysia and Thailand) 1997

Banks of other economies (Hong Kong) 1997

Country dummies not reported Constant Log likelihood

Note: Numbers in parenthesis are asymptotic t-statistic, and *, ** and *** indicate significance level of 10, 5 and 1 percent, respectively. The sample includes 29 banks from Hong Kong (12), Malaysia (9) and Thailand (8). Every bank has annual observation from 1995 to 2000. The total number of observation is 174. Source: Salomon Smith Barney/Citibank estimates

One possible explanation is that deposits were moving among financial institutions. A quick look at the names of the banks included in this study suggests that they are mostly private banks. One change during the crisis was shift of deposits toward the state-owned and foreign banks. In Thailand, though, the rise in foreign-owned bank deposits was only about 3.5 pp of the total deposit base to 5.5% in late 2000. In Malaysia, foreign banks gained about 2 pp during the peak of the

7

In our data, loan loss provisions are the amounts, both general and specific, shown as an expense in the banks’ income statements. The loan loss reserves are the amounts, again both general and specific, shown on the banks’ balance sheets. 13

19 February 1998

Test Report Title

crisis, with another 1.5 pp gain in since mid-2000. These changes don’t look significant enough to explain the discrepancy between the macro and micro data. Similar to deposits, growth of real loans also fell significantly in the second year. This continued in the next two years. The initial decline in loan growth was consistent with the sudden increase in real lending rate that we discovered in the country-data analysis and to the drop in GDP growth and investment. But negative impacts on loan growth appeared to be more persistent in Malaysia and Thailand — the effects were still quite significant in the fourth year, 2000. This is somewhat different from past experiences of other crisis economies in which loan growth slowdown was often short-lived (Demirgüç-Kunt et al. 2000). The changes, though, match the overall profile apparent in the economy-wide data. Changes to growth of real assets show the same pattern — no significant impact in the first year but sharp decline from the second year and thereafter. It is interesting to note that decline of real asset growth of Hong Kong banks was greater than that for banks in Malaysia and Thailand. One difference is that Hong Kong banks’ lending concentrated heavily in the non-tradable sector, especially properties. With the gradual realignment of the real exchange rate, the prospects for the property sector remained glum, limiting banks’ interest in mortgage lending. After the crisis began, Hong Kong banks chose to hold more cash and dues from banks (see results below). Another interesting finding from the econometric exercise is the lack of portfolio shift at the beginning of the crisis in Malaysian and Thailand banks, except for the decline of cash and dues from the banks (see Figure 10). Usually a quick shift away from bank loans is observed from the early stage of the crisis. But for the group of the banks we look at, the loan/asset shares only fell significantly in the fourth year, 2000. This may be related to the timing of write-down and sale of bad assets to the AMC in the crisis-affected economies, or the effects of fx denominated loans 8 . In Thailand, for example, Krung Thai bank, which represents some 15% of commercial bank assets, didn’t sell its nonperforming loans to a national AMC until the September 2000. Figure 10. Crisis aftermath – evidence from bank level data (III) Cash/ assets

Deposits/ assets

Equity/ assets

Loan/ Other earn assets/ assets assets

-5.865**

-2.693

-0.97

0.103

(2.3)

(0.9)

(1.3)

0.0

(0.2)

1998

-7.266***

3.334

-0.833

0.659

8.013***

(3.0)

(1.1)

(1.2)

(0.3)

(3.3)

1999

-3.800*

4.593*

-1.891***

-0.102

12.667***

(1.6)

(1.6)

(2.7)

0.0

(5.2)

2000

-3.048

5.795*

-1.935***

-5.544**

9.790***

Banks of crisis economies (Malaysia and Thailand) 1997

8

14

0.576

Need to check with banking team on Korea/Singapore data and on fx/lc asset split.

19 February 1998

Test Report Title

(1.2)

(2.0)

(2.7)

(2.5)

(4.1)

8.638***

6.671*

1.716**

-1.103

4.780*

(3.1)

(2.0)

(2.1)

(0.4)

(1.7)

1998

9.638***

8.681**

1.907**

-2.646

14.079***

(3.4)

(2.6)

(2.3)

(1.0)

(5.0)

1999

13.124***

9.251***

1.798**

-7.045***

10.532**

(4.7)

(2.7)

(2.2)

(2.8)

(3.8)

2000

10.823***

8.513**

1.523*

-7.940***

6.762**

(3.8)

(2.5)

(1.9)

(3.1)

(2.4)

19.521***

71.119***

9.027***

61.741***

80.156***

(15.8)

(47.6)

(25.0)

(55.5)

(65.4)

-616.02

-648.64

-402.76

-597.45

-614.17

Banks of other economies (Hong Kong) 1997

Country dummies not reported Constant Log likelihood

Note: Numbers in parenthesis are asymptotic t-statistic, and *, ** and *** indicate significance level of 10, 5 and 1 percent, respectively. The sample includes 29 banks from Hong Kong (12), Malaysia (9) and Thailand (8). Every bank has annual observation from 1995 to 2000. The total number of observation is 174. Source: Salomon Smith Barney/Citibank estimates

The proportions of other earning assets, including deposit with banks, government securities, other investments and equity investment, however, increased significantly from the second year and stayed at high levels in the following two years. The shares of cash and dues from banks dropped from the first year until the fourth year. This is consistent with a decline in risk preference by banks as they moved to hold government paper, often issued to cover the costs of financial system restructuring. It is also consistent with a rise in other investments or equity as a result of loan workouts and restructurings. Surprisingly, portfolio shifts in Hong Kong banks were not only more responsive but also in different directions in some cases. Cash and due from banks rose (rather than fell) from the first year and remained high in the following three years. The shares of equity also increased. The rise in equity is consistent with an effort to bolster depositor confidence in the aftermath of the regional crisis. Note that unlike most of the rest of Asia, Hong Kong banks do not carry deposit guarantees.

A framework for organizing bank restructuring Our macro data overview shows large, persistent effects from the Asian crisis on credit and investment, but little lasting effect on real interest, real exchange rates or real growth. It also shows banks with little shrinkage in deposits (at least in the economy-wide data), with shrinking returns on assets and falling interest margins. The macro effects on credit and investment are in general larger and more persistent than has typically been the case, with the effects on interest rates and exchange rates are less so. Is there something in the nature of the reforms in the Asia’s crisis banking systems that might explain these findings? Rather than launch into the myriad array of financial reform measures, we

15

19 February 1998

Test Report Title

take a moment to lay out a set of organizing themes. This doesn’t amount to a theory, so much as an effort at laying out various key shortcomings in Asian banks or, indeed, banks more generally 9 . Externalities, Moral Hazard and “Too Big to Fail” Problems One of the first topics that arises in discussions banks’ weaknesses is the problem of moral hazard created most governments’ unwillingness to allow the systemic collapse of its banking system. This reluctance stems from the broader economic costs foisted on society by the inability to use money as a means of payment in the event of a systemic collapse. Moral hazard springs from the ability of an individual bank to put it losses to the government (or whatever institution backs up the bank’s liabilities) in the event of a systemic crisis 10 . Couple this with a financial system with a few dominant institutions and, voila, the problem of banks that are too big to fail arises. Knowing that the institution’s losses are shared, but its gains accrue to its owners, banks take larger risks. Larger risks than raise the chances of an actual systemic crisis.11 An alternative is the New Zealand solution: invite in global banks so large that they are unlikely to see their solvency put at risk by events in any one country. For this course to make sense, though, requires that global institutions value their reputation highly enough not to allow the failure of a local branch. The political economy of this solution is also difficult for many countries to stomach as it implies less sensitivity of the banks to non-regulatory coercion by government. We return to this topic below. There are a number of ways to address the issue of moral hazard and “too big to fail” issues. One is too limit the size of any individual bank relative to the system. This solution, though, is not much in vogue today given the view that large institutions are more efficient, either because of lower costs or greater geographic diversity or because of greater product diversity. Another solution is to separate money as a means of payment from the risky assets of the banks, backing money with low risk government bonds.12 Separating money from the banks risky loans then allows for the payments system to function even if individual banks fail. The difficulty with this system, is the proliferation of near monies in comparison with government debt (see Figure 11). 9

In the interest of brevity, we will use banks to mean both banks and other financial intermediaries unless the context requires us to make a distinction. 10 It also implies that cost of this option is not recovered from the financial system 11 Plant et al. (2000) sets out a useful taxonomy for banking distress/crisis that relates the dynamics of bank behavior to a bank’s liquidity and its capital base. As bank capital and liquidity shift, so too does a bank’s preference for risk and the costs of moral hazard. 12 This was the structure used in the US in the 19th century during the period of free banking (see White 198?) 16

Test Report Title

19 February 1998

If government paper is issued to back broader money aggregates it is entirely possible that the issuance of that debt will undermine the government’s credit quality and the raison d’etre of the system. Note that to some extent, the ex post outcome in Asian crisis countries resembles 19th century solution in that banks’ holdings of government paper have risen sharply in Indonesia, Thailand, Korea and Malaysia (see Figure12). Figure 11. Narrow Money (M1) as Percent of Government Debt 1994

1995

1996

1997

1998

32

38

50

17

19

107

123

117

74

52

Malaysia

50

57

68

70

52

Philippines

12

14

17

16

15

158

201

241

143

66

Indonesia Korea

Thailand Source: CEIC , IMF

Figure 12. Total claims on government (as percent of total loans and advances) %

128%

165%

20 18 16 14 12 10 8 6 4 2 0 1994

1995 INDONESIA

1996

1997 KOREA

1998

1999

2000

THAILAND

Note: Of commercial banks Source: CEIC

The most common solution is to impose regulation/supervision by the insurer of the banks’ liabilities in an effort to minimize that chances that excessively risky loans are put on the balance sheet. Another is to design a deposit guarantee system that protects some but not all deposits so that depositors have an interest in avoiding risky banks. Deposit guarantee schemes usually protect small depositors so as to minimize the number of people effected by a bank collapse and because small depositors may not have the ability to assess the risks of the banks in which they place their money. Asymmetric information This, though, can lead to a second issue: asymmetric information. Bankers will naturally know less about their borrowers prospects than the borrowers themselves. Furthermore, borrowers with 17

19 February 1998

Test Report Title

riskier projects will be willing to pay higher interest rates (on the assumption that riskier projects come with higher returns). Adverse selection can then result, whereby the quality of the pool of borrowers declines as banks raise interest rates13 . Since banks profits depend not on the interest rate charged, but the on the interest collected, profits can fall if credit quality falls fast enough as interest rates rise14 . Banks would be subject to what might be called the Groucho Marx effect. Groucho, a Jewish comedian during a time when most posh American clubs restricted membership to gentiles, once remarked that he wouldn’t want to join any club that would have him. With adverse selection and high enough interest rates, bankers wouldn’t want to make a loan to any borrower who would takeit. Solutions to asymmetric information include forcing borrowers to make co-payments or to put up collateral to either align their interest with those of the bank (with co-payments) or to divorce the bankers return from the outcome of the project. A further solution is for bankers to develop longterm relationships with their borrowers in an effort to increase the chances that the borrower will value the prospects of future loans in assessing his/her decision to borrow for any particular project. The question of long-term relationships as a means of overcoming asymmetric information problems has a checkered history. The literature on corporate governance from the late 80’s and early 90’s was generally positive about long-term relationships as a means of allowing banks to smooth lending through more intimate knowledge of the their borrowers and through leverage gained as the chief source of finance (see Mayer 1988). The poor performance of Japanese banks in the 1990’s has pointed to the downside: (see Boot & Thakor (2000). The work by Barth et al. (2001) cited below is perhaps less pessimistic. They find that restrictions on bank equity holdings are not associated with sounder financial development. The findings, though, also support the need for strong private sector monitoring that may force banks to hone their credit skills rather to become captives of their borrower. Another element of dealing with asymmetric information is to simply put more information into public hands. But this then raises the problem of free riding. If information on a borrower is entirely public, but costly to generate and the loan made to the borrower is tradeable, then bankers will underinvest in information. That results from the ability of other bankers to simply lend to the same firms receiving credit from those institutions that invest in information collection. 13

This problem was first laid out in banking context by Stiglitz and Weiss (1983?). This issue has been a hugely contentious one in assessing the Asian financial crisis. It at one point pitted the IMF against the World Bank’s chief economist. The former argued that high interest rates were needed to stabilize exchange rates, while the latter argued that such 14

18

19 February 1998

Test Report Title

One solution to the problem is to force the monitoring costs onto the firms, with verification through a public regulator. For bank loans, the problem of underinvestment in monitoring is limited by the lack of secondary markets in most loans. This effectively privatizes the gains from monitoring. Legal Infrastructure Banking by its very nature is an activity in which resources are given today conditional on outcomes in the future. In the event that the future doesn’t lead to the success of a borrower’s project, the issues surrounding bankruptcy law, liens and dispute resolution come squarely into view. Collateral may be an elegant solution to the problem of adverse selection, but only if the banker can effectively take control of the collateral in the event of default. Antiquated or inadequate laws and capricious adjudication have undermined the value of collateral as protection against bad loans (See La Porta et al. (1997)). Indeed, the issue of dispute resolution can be seen as a subset of a larger issue, the strength of outside investors’ protection. La Porta et al. (2000) argue that measures of financial market development, be it bond or equity markets, are closely linked to measures of the strength of legal protections for investors. Issues of Market Structure and Prudential Regulation The issues of moral hazard and economic/allocative efficiency rise a thicket of issues around the optimal number, size and regulatory structure to put in place for a stable and effective financial system15 . Barth, Caprio & Levine (2001) shed some light on these issues using a database on banking supervision and regulation in over 100 countries. They use the database to analyze how different regulatory structures, shares of government ownership, etc., affect measures of financial development and stability. They find that, in general: 1.

government ownership of banks is associated with sub-par financial development as measured by the extent of private credit to GDP, overhead, interest margins or the chance of a banking crisis

2.

Restrictions on banking activities do not improve the stability of financial systems or their efficiency

rates would so undermine bank profits and growth that further capital flight and depreciation would ensue. See Goldstein et al.(2000) for references to this debate. 15 Barth et al. (2001) divide their database into ten categories: 1) bank activity regulations variables, 2) Mixing banking/commerce regulations , 3) Competition regulations, 4) capital

19

19 February 1998

3.

Test Report Title

Foreign bank presence, while enhancing volatility, fosters growth through reduced fragility and greater cost and allocative efficiency

4.

Regulatory structures that rely on private market forces lead to more stable financial systems

What reforms have been taken? With this general schema in mind—issues of externalities, asymmetric information, legal and supervisory structures—we now turn to the question of what changes have actually taken place and what issues remain to be tackled.

Market Structure & Externalities: Dealing with unviable banks One of the key issues each set of programs dealt with was the treatment of unviable institutions. Steps taken in closing down, merging or rescuing banks are crucial to instilling confidence, to the costs of financial reform and to the incentives for future risk-taking by banks. The changed structure of Asian banking systems as a result of the treatment of banks is shown in Figure 13. Figure 13. Market Structure changes in Asia's Financial Sector Number of commercial banks taken over/sold to foreigners/nationalized

Number of private domestic banks (market share)

Number of state Number of nonbank banks (market financial share) institutions (market share)

Market share of three largest

Indonesia

4/0/12

122 (21%)

43 (78%)

245 (1%)

34% (35%)

Republic of Korea

5/2/04

18 (37%)

10 (58%)

11 (5%)

24% (16%)

Malaysia

1/0/1

21 (56%)

3 (18%)

37 (26%)

34% (28%)

Thailand

0/2/4

13 (48%)

6 (45%)

22 (7%)

48% (47%)

banking institutions, December 2000 (end 1997)

Source: World Bank: East Asia Recovery and Beyond and SalomonSmithBarney (check figures)

Indonesia: The very first step Indonesia adopted in bank restructuring was the closure of sixteen non-viable commercial banks in November 1997. This was perhaps the most criticized element of the early stabilization program as the banks closed did not encompass all of Indonesia’s problem banks. When deposit guarantees were not extended to the rest of the system, a series of bank runs that cost Bank Indonesia almost Rp.220 trillion (US$51 billion at the then current exchange rate or 22% of GDP) ensued. By January 1998 Indonesia had declared a blanket government guarantee.

regulations, 5) supervisory actions, 6)supervisory resources, 7) private monitoring, 8) deposit insurance, 9) market structure and 10) outcomes. 20

19 February 1998

Test Report Title

Bank closure continued to serve as a means of bank restructuring in the following years — seven banks were closed in April 1998, three were closed in August 1998, 38 were closed in March 1999 and two joint venture banks were closed in April 1999. From November 1997 to the end of 1999, there were a total of 66 banks closed. Together, these banks accounted for about 13.5 percent of the banking sector’s total liabilities. While depositors, large and small, were protected in every closing, bank owners were not. Their equity was wiped out. Indonesia’s state banks had long had serious asset quality problems and poor performance.16 They accounted for about 40 percent of total liabilities of the banking sector assets in 1997 and have accounted for almost 2/3rds of the recap costs for the banking system. During the crisis, significant efforts of restructuring were devoted to the state banks. In June 1998, the government dusted off a plan for merger of state banks. Bank Bumi Daya and Bapindo were identified as the first two to be merged. But in September 1998, the government announced the formal merger of four state-owned banks, Bank Dagang Negara, Bank Ekspor Impor Indonesia, Bank Bumi Daya and Bank Pembangunan Indonesia, and the corporate business of a fifth state bank, Bank Rakyat Indonesia, into a new institution, Bank Mandiri. The NPLs of all the five banks were transferred to the Indonesian Bank Restructuring Agency (IBRA) set up in early 1998 to handle the assets of failed institutions. Bank Mandiri became the largest bank in the country, accounting for 25 percent of the country’s total banking sector liabilities. As can be seen in Figure 13, the three largest banks coupled now control 34% of banking assets, roughly double the pre-crisis share. This is hardly good news for the “too-big-to-fail” problem. Similar steps were undertaken with the private banks. First a series of portfolio reviews of 128 private banks, completed in February 1999 were used to classify banks capital adequacy. In March 1999 the government moved to deal with the weak banks according to the level of measured capital adequacy ratios (CARs). Thirty-eight banks were closed, seven banks were taken over by the IBRA, and nine banks were identified for possible government assisted recapitalization 17 . Of the nine banks eligible for recapitalization, one could not come up with the necessary capital and has been taken over by IBRA. Eventually seven of the eight banks received injections of public funds.

16

State bank reform was a key part of Indonesia’s original efforts at deregulating its banking system in 1983. 17 Private banks were eligible for recapitalization if the owners were willing to contribute 20% of the recap needs with the government providing the additional 80%. Owners were also given the right to buyback the government’s equity at the initial price plus the cost of debt issued in support of the capital infusion. This was an implicit subsidy to the old owners as the government was taking equity, but demanding only a risk-free bond return. The subsidy was justified as the cost of maintaining some portion of the banking system in local private hands. 21

19 February 1998

Test Report Title

Besides the treatment of pre-existing state banks, 13 formerly private banks came under IBRA’s control. These include almost all of the largest pre-crisis private banks. The difference in treatment does seem related to size, hinting at the too-big-to-fail problem. However, management at the banks taken over (BTO) have been changed, dampening the incentive for risk-taking by these institutions. The quality of the 1999 decisions on bank closures/take overs/recaps has been tarnished by the recent problems of Bank Internasional Indonesia (BII). It was once one of the largest private banks, and a key component of the Wijaya group companies. It was judged mildly undercapitalized in 1999 and given public money 4:1 with equity from the Wijayas. Massive related lending, though, drove a huge hole in the balance sheet in late 2000 when Asia Pulp and Paper, the flagship of the group, became illiquid. The extent of the related loans flew in the face of supervisory norms and hints at favoritism. Other banks which had massive related loans were taken over by IBRA in 1999, while BII was not. Indonesia’s restructuring efforts have had minimal involvement from foreign banks. To date only two management contracts involving foreign banks are in place. An effort by Standard Chartered Bank to buy an Indonesia bank ran afoul of bank employees and became ensnared in a major political scandal that still plagues the central bank. Korea: In Korea, the government first reviewed merchant banks’ portfolio positions and their rehabilitation programs in mid-1997. These institutions had come under pressure from heavy investments in Indonesia, Thailand and other emerging market bonds that suffered after the Thai devaluation. Then in December 1997, it decided to close fourteen merchant banks and require the other sixteen to follow a timetable to achieve capital adequacy ratios of at least 6 percent by June 1998 and 8 percent by June 1999. All deposits, selected liabilities and ‘good’ financial assets of the suspended merchant banks were transferred to a newly established merchant bank, Hanareum Banking Corporation (HMBC). The remaining assets and liabilities were transferred into bankruptcy and eventually sold to KAMCO, the Korean Asset Management Company. Of the sixteen ‘better’ banks, the government suspended one and merged another. The remaining fourteen operated under the government’s close monitoring. Just this month another two private banks announced a merger which will create the countries largest bank. In June 1998, the government decided that another five commercial banks, Dongwa, Dong Nam, Dae Dong, Chung Chong and Kyungki, were non-viable. Five stronger banks acquired these banks. Again, in December 1998, two commercial banks in distress, Korea First and Seoul, were intervened and were effectively nationalized in early 1999. The plan was that these nationalized 22

19 February 1998

Test Report Title

banks would be privatized later. After ten months of negotiation, the government signed a binding agreement in September 1999 to sell a controlling stake of 51 percent in Korea First Bank to Newbridge Capital. According to the agreement, the government would take on all NPLs, add to the bank’s reserves should its existing reserves fail to meet accounting standards and take on any existing loans that go bad in the ensuing two years. Like Indonesia, Korea’s efforts at dealing with unviable institutions were piecemeal. While the mistake of partial guarantees was avoided—a blanket guarantee was given in December 1997— issues at non-bank financial institutions were left unresolved until 2000—insurance companies, mutual saving and finance companies and investment trust companies. A less cheerful comparison is the repeated rounds of recapitalizations using public funds—most recently last fall. While early recaps were done after portfolio reviews, these reviews have proved too optimistic. This is particularly disturbing given the magnificent economic performance from late 1999 through the third quarter of 2000. Recaps and takeovers have, like Indonesia, resulted in a larger proportion of assets are now under government-owned banks, now nearly 60% of all assets. Concentration of bank assets has also risen, boding poorly for moral hazard linked to “too-big-to-fail problems”. Selective closures can bring up issues of favoritism that undermine their otherwise disciplining effect on other banks’ future behavior. Unlike Indonesia, Korea has seen a larger role for foreign banks as has Thailand (see Figure 17). Thailand: The Thai government’s efforts, after fitful starts in early 1997, were some of the most aggressive in treating failed institutions after the July devaluation. The government established the Financial Restructuring Authority (FRA) and the Asset Management Corporation (AMC) in October 1997 the dispose of the assets fifty-eight finance companies whose operations were suspended. Of these suspended finance companies, fifty–six were closed down and their assets taken over to be auctioned by the FRA. The assets of the suspended finance companies were over 20% of the overall assets of the financial banking system. The Bank of Thailand intervened in two additional banks, Union Bank and Laem Thong Bank, and five additional finance companies. Merger and restructuring of government-owned entities (Krung Thai Bank and Krung Thai Tanakit) with intervened financial institutions and their restructuring also formed part of the strategy. One bank (BBC) was liquidated, two banks (FBCB and LTB) were absorbed by two state banks (KTB and Tadhanasin Bank), one bank (UBB) and twelve finance companies that were intervened would be eventually integrated with a state-owned finance company (KTT), and three banks (RAB, BMB and SCIB) would eventually be privatized. The

23

19 February 1998

Test Report Title

government also initiated the sale of four intervened banks by providing public resources. Nakornthon Bank (NTB) was the first bank acquired by a foreign strategic investor, Standard Chartered Bank, which agreed in September, 1999 to purchase a 75 percent stake for TB12.4 billion. In another transaction, United Overseas Bank of Singapore (UOB) agreed in November 1999 to purchase Radanasin Bank (RSB). Malaysia: The strategy that the Malaysian government adopted toward non-viable is different from either Indonesia, Korea or Thailand. It have opted for industry consolidation through merger rather than closures. This less urgent response reflects two facts: better capitalized banks and far less fx exposure which resulted in smaller NPL ratios (see Figure 14). The absence of insolvent institutions allowed Malaysia to focus on transferring NPLs out of going concerns and only more recently addressing questions of market structure. Also note the prior to the crisis Malaysia had a significant amount of its assets controlled by foreign banks (23% versus less than 5% in Thailand or Korea). This is consistent with the work of Barth et al that finds a positive correlation between foreign bank participation and financial soundness. In July 2000 the government announced a plan to bring about a consolidation of the financial industry which covers twenty-one banks. The purpose was to improve the competitiveness, and so to ensure survival of the domestic banking industry as foreign competition intensified. The process of mergers, though was a politically charged one, with accusations of manipulation leveled at the Finance Minister. The initial plans for consolidation to 6 groups was subsequently revised to 10. The merger plans for most of the ten were finalized last December. While existing asset concentrations aren’t much changed from pre-crisis levels, the new merged structure will leave each bank potentially “too-big-to-fail” putting stress on either official supervision or private monitoring (see section on prudential supervision below). Figure 14. Non-Performing Loans (Percent to Total Loans)

24

Test Report Title

19 February 1998

% 16

60

14

50

12 40

10 8

30

6

20

4 10

2

0 0 Mar-98 Jun-98 Sep-98 Dec-98 Mar-99 Jun-99 Sep-99 Dec-99 Mar-00 Jun-00 Sep-00 Dec-00 Korea

Malaysia

Indonesia

Thailand

Note: NPLs are those of the banking system, except Korea, which is of the domestic commercial banks. Malaysia measure includes Islamic banks. Thailand measure excludes branches of foreign banks. Source: CEIC

Governance Changes Besides the numbers on mergers/takeovers/privatizations, the future behavior of banks will also depend on changes in management and internal oversight. 18 Here one finds a similar ranking to that involving the extent of restructuring or fragility. Indonesia shows both the most closed board system, the least amount of managerial changes and the least reliance on outside expertise. Korea shows the most aggressive changes, will Malaysia shows fewer changes but , at least in terms of outside directors, a more open structure to begin with. Note that the comments on foreign experts is misleading. The high percentage of foreign controlled banking assets creates and opportunity for Malaysians to develop expertise in foreign banks and then shift to local banks.

Figure 15.Changes in Corporate Governance and Management of Banks Corporate governance Country

Independent outside directors Changes in top management

Management

In majority-owned domestic banks

Performance-based pay

Hiring of foreign experts in domestic banks

Indonesia

Rare

State bank management changedRare

Limited

Republic of Korea

Two-thirds of board seats

In 6 of the 11 major banks

Being introduced

Frequent

Limited

Rare

Occupied by outside directors Malaysia 18

In place

In 1 of 33 banks

XXX (2001) throws some cold water on the importance of changes in boards of directors for corporate performance. They find that in a sample of US firms changes in the composition of the board to install outside directors or the establishment of larger boards were not correlated with improved firm performance. 25

19 February 1998

Thailand

Test Report Title

In place

In 3 of 11 banks

Limited

Frequent

Source: S. Claessens, et. al., Financial Restructuring in East Asia: Halfway There?, Financial Sector Discussion Paper No. 3, The World Bank, September 1999

Resolution of NPLs One of the key issues in restoring the health of banks is the question of how best to deal with nonperforming loans (NPLs). This involves at least three sets of issues: 1.

regulatory issues (loan loss provisioning standards, tax treatment, timing of provisioning)

2.

use of bad loans centralized asset management companies (AMCs)

3.

legal support for loan restructuring or foreclosure and liquidation.

We deal with the regulatory and legal issues below and concentrate for the moment on the questions surrounding AMCs. There are a number of choices surrounding the structure and role of an AMC: 1.

should it be centralized or left as a subsidiary of each financial institution; if centralized,

2.

should assets transfer at book, at market to the AMC and should transfers be mandatory or voluntary and how should AMC purchases be funded

Centralization or Not: All the crisis countries adopted a least one centralized, state-owned AMC (KAMCO, Danaharta in Malaysia, IBRA in Indonesia and the FRA and the TAMC in Thailand,) 19 . The arguments in support of this were (and are): gains from being able to consolidate assets; the simplification of restructuring through the reduction in the number of creditors (and hence free rider problems); more efficient use of scarce restructuring talent and the ability to use government’s extra-legal power in pursuit of quicker resolution. Korea established the Korea Asset Management Corporation (KAMCO) in August 1997 and a special fund was set up within KAMCO to which the banks could be allowed to sell their NPLs. In November that year, the government removed KAMCO from the Ministry of Finance and Economy (MOFE) and re-established it under the control and supervision of the newly established Financial Supervisory Commission (FSC) as a public agency to manage non-performing assets transferred from distressed financial institutions.

19

China has set up individual AMCs under each of the major state-owned commercial banks, but without setting up a central AMC. 26

19 February 1998

Test Report Title

Malaysia established its asset management company, Danaharta, in May 1998. Danaharta was designed to purchase NPLs from financial institutions whose NPL-to-total loan ratio exceeded 10 percent. It was also given the power to appoint special administrators to manage the affairs of distressed companies to facilitate and accelerate their restructuring. The Financial Restructuring Agency (FRA), set up in 1997, handled the assets of intervened finance companies. The early auctions that involved non-core assets were considered successful, with 53 percent of the value of the assets recovered, while the later auctions involving core assets were less successful, recovering only 25.4 percent of book value.. The FRA had largely finished liquidating the assets of closed finance companies by the end of 1999. Thailand initially rejected a centralized AMC for the resolution of NPLs in operating banks out of concern that credit discipline would weaken once banks sold off their loans. As a result, after a relatively quick process of finance company asset sales, commercial bank NPLs were left either on the books of the banks or transferred to wholly-owned subsidiary AMCs. Thailand’s concern over credit discipline looks more like a response to a political reality than to an economic one. Taking point three above, if one transfers the NPL at fair market value and forces the selling bank to realize the loss, it is not clear that bank managers have any incentive make bad lending decisions going forward. Thailand’s decision, then, looks more like a realization that determining fair market value would not be easy given the accounting standards in place, the magnitude of the distressed loans relative to market transactions and the political clout of the borrowers. Hence the decision to leave the loans with the banks who made them. Transfer prices and funding : Transfer prices in Korea and Malaysia, while initially contentious, have proved so far to be reasonable. In Danaharta’s case, prices were determined after portfolio reviews. In Korea, more uniform discounts were applied based on the type of asset. Each entity funded the purchases through issuance of its own, government-backed bonds. As with the RTC in the US, the use of agency bonds, rather than sovereign government paper may have pushed up the cost of funding the entities (check on magnitude of this cost). Both KAMCO and Malaysia (see Figure 16) have been able to recover on average an amount greater than their book value for assets sold. By the end February 2001, KAMCO had spent KW35.6 trillion to purchase KW89.4 trillion (face value) of NPLs which was about 11 percent of total bank outstanding loans. It had recovered KW 21.5 trillion from the sale/restructuring of bad debts with an original face value of KW46.5 trillion and a cost of KAMCO of KW19.3 trillion.

27

19 February 1998

Test Report Title

Danaharta, meanwhile, by the end of 2000 had acquired RM20.5 billion at an average discount of 55%. It held another RM27.0 billion under management. At the same time it had restructured/approved for restructuring RM35.8 billion with an expected recovery rate of 66%. Of course, one shouldn’t confuse good ex post outcomes with the absence of ex-ante subsidies. Still, the recoveries have meant that both AMCs have been able to reduce the net cost to the government of financial restructuring. Figure 16. Malaysian Recovery Proceeds Asset Group

Expected Recovery*

Recovery Pending Receipt

Recovery Received

Stock ** (valued as at 31-Dec-00)

Cash/ Realized in cash

(a)

(b)

(c=a-b)

(d)

(e=c-d)

Cash

8725

4502

4223

--

4223

Performing Loans

9689

2964

6725

5450

1275

Securities

4089

3465

624

620

4

Properties

1298

838

460

334

126

23801

11769

12032

6404

5628

Total

Source: DANAHARTA Operations Report (as at December 31, 2000)

IBRA followed a different tack, receiving assets at a value of 1 rupiah. The government then injected the necessary capital into the selling banks covering the hole created by the sale of the NPLs to IBRA. This method, while lowering IBRA hurdle, created a large gross financing cost for the government. Within the IBRA, Asset Management Investment (AMI) and Asset Management Unit (AMU) were created to lead the restructuring efforts in the most illiquid and insolvent banks and to manage the assets acquired in the bank resolution process. AMI is responsible for managing equity holdings and AMU is in charge of acquiring NPLs from frozen or merged banks and managing them. And all outstanding liquidity support from Bank of Indonesia extended in the in early stage of bank closure were all transferred to IBRA. As of March 2001, the AMU holds Rp307 trillion in loans of which about 40% have finalized restructuring plans. The AMI holds Rp.94 trillion in corporate assets. In Indonesia, while any recovery on loans is a plus, the amount of recovery as a proportion of the portfolio has been small. The AMU has sold Rp.4.6 trillion (out of Rp.94 trillion), while cumulative loan recoveries have reached Rp.19.8 trillion (on a face value of Rp.307 trillion). The low recovery on corporate assets held by the AMI is in part of the result of in value due to both overly optimistic initial valuation assumptions and to the length of time it has taken IBRA to re-sell the assets. The deterioration reflects the costs of delay and of political interference. IBRA in its brief existence has had four heads and frequent turn-over of senior staff. The turn-over is due to the political pressures that IBRA faces (its assets total some 40% of GDP) and to the difficulties with restructuring in a capricious and ill-defined legal environment (more on this below). Figure 17. Financing East Asia's Restructuring (figures need updating)

28

19 February 1998

Action

Test Report Title

Indonesia

Korea, Rep. Of

Malaysia

Thailand

$21.7 billion

$23.3 billion

$9.2 billion

$24.1 billion

(18 percent of GDP)

(5 percent of GDP)

(13 percent of GDP)

(20 percent of G

31.4 (9/99)

17.9 (9/99)

24.0 (12/99)

38.4 (5/00)

10.1

17.5

35.6 (5/00)

Initial government response Liquidity support Non-Performing Loans (NPLs) NPLs/Total Loans (%)

NPLs/Total loans after transfer to AMCs 12.4 Peak NPL ratio

46.8

Financial distress resolutions Bank closures

66 of 237

None

None

1 of 15

Closures of other financial institutions

None

More than 200

None

59 of 91

Mergers

9 nationalized or state banks are being merged 8 of 26 absorbed by other banks 58 to be merged into 10 by Dec-01 3 banks and 13

Banks temporarily nationalized

12

4

1

4

Bank recapitalization strategies Public funds for recapitalization

Bonds equivalent to $40 bn issued in 1999; $20 Government bn injected $50 bn into 9Danamodal injected $1.7 billion

Government inje

to be issued in 2000

private banks an

commercial banks; 5 of 6 major

into 10 institutions

banks now 90% controlled

into public banks

by the state Private funds for recapitalization Majority foreign ownership of banks

$9 bn 0

2

Not allowed. Foreign bank share

4 completed, 1 p

is significant, however Weak financial institutions still in system Many weak commercial banks

Some weak nonbank financial Institutions

Source: World Bank: East Asia Recovery and Beyond

Recapitalization Besides, closures/mergers and NPL disposal, the other key ingredient in market structure has been efforts to restore bank capital. Figure 17 lays out some of the key parameters for each of the main crisis countries (and China). We provide details below, but a few points stand out: 1) government capital injections have been extensive in Indonesia, Korea 2) almost no use has been made of foreign strategic partners in recapitalizing Asian banks 3) lingering threats to capital and to government costs persist in NBFIs, state-owned banks and weaker private banks The most costly recapitalization for the government is easily Indonesia. The total estimated cost of bank restructuring has risen to Rp636 trillion, or over 52 percent of GDP (Kawai 2000). About onethird of the total cost is connected with the cost incurred by Bank Indonesia, which incurred losses from past liquidity support to closed banks, banks taken over by the IBRA and a former state bank. The Bank Indonesia was also required to finance payments of claims under the guarantee on the closed banks. About half of the total cost is for recapitalizing state banks. Nearly one-fifth is related to recapitalization of twelve banks taken over by IBRA. Private banks only account for a small fraction of total cost. But this cost may rise further if restructuring is delayed and banks’

29

Some weak pub

banks and financ

19 February 1998

Test Report Title

performance continue to deteriorate. The costs of recapitalization are financed through the issuance of government bonds, which began in May 1999. Figure 18. Assessing Bank's Ability to Handle Their NPLs (numbers need updating) Indonesia

Malaysia

Thailand

Total Loans (US$bn)

41

104

132

Current Capital

-25

14

12

New Capital Required to Lift Capital Adequacy to 8% (a)

30

-

-

5

-

8

35

-

8

Additional Capital Required to Meet Provision Requirement (b) Total Capital Shortfall (US$bn) (a) + (b) Source: Salomon Smith Barney estimates

The Korean government originally set aside KW64 trillion, about 14 percent of GDP, for resolving the crisis in the banking and financial sectors. That figure, by the end of 2000, had risen to KW150 trillion (check figure; FSS website Oct 28 publication). In October 1998, the Korea Deposit Insurance Corporation (KDIC) injected W5.78 trillion into the five commercial banks, which acquired one ailing bank each. KDIC later injected an additional W2.26 trillion into the same five banks. The seven under-capitalized banks private banks, Korea Exchange, Hanil, Cho Hung, Commercial, Kangwon, Chungbuk, and Peace, received conditional support. They were required either to merge into healthy banks or to arrange mergers among under-capitalized banks with government entity assistance and resulting government ownership. Hanil and Commercial, for instance, merged and established a new bank called Hanvit. Through its fiscal support, the government became its major shareholder. The initial efforts at stemming the panic quickly moved to dealing with the stock of bad debts and reducing the likelihood of new crisis. These included efforts to separate good from bad institutionswith the closure or merger of a number of financial institutions. In Malaysia, Danamodal was established in July 1998 as a special purpose vehicle with the purpose of recapitalizing financial institutions whose capital adequacy ratios fell below 9 percent. By June 1999, Danamodal had inject RM6.4 billion in capital into ten financial institutions in the form of Exchangeable Subordinated Capital Loan (ESCL). Originally it was anticipated that up to twentythree financial institutions would become undercapitalized and would require up to RM16 billion in capital injections through Danamodal. But later the government announced that Danamodal would need significantly less. Some of the institutions that received capital injections have begun repay Danamodal. Danamodal’s funding is comprised of paid-up capital of RM3 billion provided by Bank Negara Malaysia and RM7.7 billion obtained by issuing zero coupon bonds to fifty-seven banking institutions.

30

19 February 1998

Test Report Title

The Bank of Thailand adopted a market-led approach of recapitalization of the financial institutions. The strategy was based on a progressive strengthening of provisioning requirements designed to prompt financial institutions to obtain fresh capital, supervisors taking over those institutions that would not be able to recapitalize. In August 1998, the government announced a program for the injection of public funds (Bt300 billion) into troubled financial institutions subject to safeguards. The program consisted of two parts. For the tier one recapitalization, the government would recapitalize the institution up to 2.5 percent of capital. Beyond this level, the government would inject Tier one capital by matching private investors’ fresh capital. For Tier two recapitalization, the government would inject capital as the institution wrote off NPLs due to corporate debt restructuring or increased net lending to the private sector. In the end relatively little of the public funds were used in private banks (roughly Bt50 billion) reflecting concerns on the banks part about the loss of control possible with government involvement. The poor reception of the August program was one of the catalysts for the new Thaksin administrations pledge to establish a centralized AMC, the Thai AMC to speed commercial bank restructuring.

Financial supervision and legal framework A large part of the efforts at restructuring Asia’s banks have focused on the need for supervision that meets international standards in an effort to forestall future crises. It has also focused on legal infrastructure, both the help salvage value of the wreckage of this crisis and to improve the credit culture going forward. The stakes on getting the prudential framework right have gone up with the concentration of assets in a reduced number of institutions.20 We look at four key issues: 1) deposit insurance, 2) loan classification and provisioning standards, 3) capital adequacy standards, and 4) the institutional structure for supervision.

Deposit Insurance. One of the key elements of the new prudential regulatory structure is deposit insurance. Such insurance was critical to the avoidance of persistent bank runs—as is evidenced by the cash of Indonesia. At the beginning, the Indonesian government made no efforts to either compensate for shareholders’ losses or to guarantee repayments of liabilities, except for small depositors (for up to Rp20 mn per depositor per bank). But faced with spiraling runs, the authorities in January 1998, announced a blanket guarantee for depositors and creditors (excluding subordinated debt) of locally incorporated banks. Thailand extended deposit insurance in the late summer of 1997 as it moved to close NBFIs. Korea did the same in December 1997. Malaysia,

31

19 February 1998

Test Report Title

interestingly, choose not to put in place an explicit deposit insurance program, but rather emphasized an MOF commitment to protect depositors. The problem much of Asia now faces is how limit the long-term pernicious incentive effects of blanket deposit guarantees. Barth et al (2000) and Demirguc-Kunt and Detraigache (2000) present important evidence that over time that broader deposit insurance increases the chances of a crisis. Coupling that with the positive correlation between private monitoring and bank development hints that relying heavily on tighter supervision is a more difficult means for improving outcomes. The first step is to put in place in place more restrictive guarantees so that private parties have a stronger incentive to police institutions. Thailand, Indonesia, and Korea all have plans for new laws/regulations to limit deposit guarantees. Getting the new laws passed, though, is the hurdle. Korea has already postponed implementing lower guarantees out of concern for the lingering weakness of banks. Weaker guarantees needs stronger banks, especially stronger capital bases. Figures show average CARs well above 8% minimum for most banks; but official delays on deposit insurance, and recurrent recaps and unrecognized loses all point to the caution that needs to be applied to the capital adequacy figures.

Loan Classification, Provisioning : The discussion of deposit guarantees leads directly into the issue of loan loss provisioning and capital adequacy. Provisioning has been an area of ferment across Asia as regulators have struggled get a handle on the extent of NPLs and of losses (see Table 10). Standards are now uniformly tighter and more in line with best practices. Figure 19. Changes in Prudential Standards in East Asia Standard

Indonesia

Korea

Malaysia

Thailand

Loan classification (days elapsed

No change - 180 days

Lowered from 180 to 90 days

No change –

Lowered from 360

180 days

to 90 days.

before considered past due) Loan los provisioning: substandard/

From 0/50/100 to 10-15/50/100

doubtful/loss (percent) Interest accrual

From 20/75/100 (backward-looking No change – to 20/50/100 (forward-looking)

0/50/100

From 0/50/100 to 20/75/100

Lowered from up to 6 months to up Lowered from up to 6 months to up No change - up to 6 to 3 months; no clawback. Source: World Bank: East Asia Recovery and Beyond

to 3 months, with clawback.

Lowered from up to 6 months to up months, w/ clawback. to 3 months; no clawback.

Unfortunately best practice loan classification and provisioning requires more information and judgement than system’s based on whether the “check is in the mail”. Assessing the adequacy of cash flows, the valuations of collateral are skills that build over time, both in banks and in supervisors. But beyond the problems of a new, more information intensive process, the crisis countries still suffer from a regulatory bias to push the costs of the crisis out over time. The

20

Looking at the concentration of banking assets is somewhat misleading as to the fragility of the financial system including capital markets because of the rise in the latter. 32

19 February 1998

Test Report Title

Hyundai provisioning example is a case in point. With this continued conflict, confidence in banks’ reporting on asset quality and hence provisioning is likely to stay weak. Capital Adequacy Standards Inadequate provisioning means overstated capital. Is that a problem? Luckily in the very short term it is less of a problem than some would think. One of the virtues of capital is to act as buffer for depositors from loan losses. In the presence of a deposit guarantee, capital is the co-payment that protects the guarantor through aligning the capital owner’s incentives with the guarantor. But one hardly needs a co-payment if the owner of the capital and the guarantor are one and the same. In state-owned back backed by state guarantees they are just that. Hence, excessive concern about capital standards, at least in state banks, at least for now, looks out of place.

The domination of the state in banks creates opportunities and challenges. The chief opportunity created by state ownership isn’t the slack it can provide in capital standards, but rather it is the ability to directly control the allocation of credit. This has been, for example, a power that the Korea government has used (some would say abused) in forcing corporate restructuring. Wider state ownership of assets through institutions like Danaharta or IBRA has also allowed for the establishment of special foreclosure authority in an effort to speed debt restructuring. Unfortunately the chief challenge of state ownership is direct control of credit. The highest loan losses are uniformly found in state dominated institutions, be it Krung Thai in Thailand, Bank Bumiputera in Malaysia, Philippine National Bank in the Philippines or the five main state banks in Indonesia. Government has a lousy track record in allocating credit as La Porta et al. point out.

Sell-down of government stakes is key Improving the flow of credit so that NPLs don’t resurface is key to limiting the costs of state intervention and ensuring fiscal soundness, especially in Indonesia, Thailand and China where public debt to GDP is gathering steam. Every country has a program for the sale of governmentowned stakes. Some are more aggressive, and more successful, than others. In Indonesia, the government has so far only been able to sell a 20% stake in BCA, formerly the nation’s largest bank. IBRA plans to sell its stake in Bank Danamon, the largest bank in the country, in the next 12 months, but a deteriorating political environment will make that sale a tough one. The Republic, meanwhile, is planning the sale of a stake in Bank Mandiri, the second largest

33

19 February 1998

Test Report Title

bank in the country. These sales, though, will still leave the state with significant holdings. Remember, state-owned banks held some 45% of the banking system’s assets before the crisis. Korea has been more aggressive in selling financial institutions. However, repeated bouts of regulatory forbearance on over-indebted borrowers have meant expanding infusions in NBFIs. Last May the government injected public money into two investment trust companies, with more capital going into state-owned banks last November. The government now plans to privatize banks in 2002. Starting earlier would be a boost to the long-term resilience of Korea’s financial system. There are two hugely positive opportunities created by the process of selling off state assets. 1. First, the proceeds of asset sales will be absolutely critical to reducing the size of government debt and, hence, the ultimate cost of the financial crisis. 2. Second, moving assets into private hands, in an environment of enhanced competition and private market oversight, is likely to speed the process of corporate reform that is a must for avoiding the pitfalls of future restructuring.

Consolidation/restructuring of supervisory authorities Besides moves to tighten classification and loan supervision another theme across Asia has been the move to consolidate regulatory authority in one entity. Indonesia’s Central Bank Law states that banking supervision will be taken out of Bank Indonesia by end-2002, so that banking supervision will be consolidated with that of non-bank financial institutions (NBFIs) and the securities markets in a new supervisory institution. Korea first created an independent agency, the Financial Supervisory Commission (FSC), in July 1997 with a mandate to supervise and restructure all banks and non-bank financial institutions. The organization was expanded into the Financial supervisory Service (FSS) in January 1999 by merging four financial supervisory agencies for banks, non-banks, securities and insurance. From late 1997, Korea’s financial restructuring programs aimed at achieving five key objectives: restoring the confidence of depositors, investors and creditors; establishing effective frameworks for resolution of non-viable banks and for recapitalizaing weak but viable banks; introducing an effective NPL resolution mechanism including the establishment of a centralized asset management company; the introduction of international standards for financial regulation and consolidated supervision; and capital markets development. The authorities have issued regulations on connected

34

19 February 1998

Test Report Title

lending, loan classification and provisioning, exposure to foreign exchange risks, coverage of deposit insurance scheme, and accounting standards. Malaysia hasn’t been as aggressive in consolidating authority, though it has moved bolster supervision. Relative to other Asian crisis economies, Malaysia had a better supervisory framework before the crisis. It was further enhanced by the introduction of Danaharta and Danamodal. Bank Negara Malaysia traditionally applies CAMEL rating for banks. During the first half of 1998, it introduced the proper definition of NPL, though it subsequently applied regulatory forebearance in an effort to cushion the system’s immediate capital needs. Thailand’s legal and regulatory framework for supervision of financial institutions is still being revised. The government is attempting to strengthen the role of the BOT in the supervision and regulation of financial institutions, including prompt corrective action authority, and put the supervisory regime in line with international standards. BOT supervisory capacity is also being strengthened. As part of financial institutional reform, the Thai government is also preparing a draft deposit insurance law. A focus on consolidated or strengthened supervisors misses an important opportunity: greater incentives for private monitoring. There has been progress here: subordinated debt issuance in Thailand and Korea; new, more frequent and extensive disclosure standards for banks, especially as regards loan quality and distribution; requirements for outside directors in Korea and Malaysia and higher standards (and liability) in Indonesia. But the use of market discipline looks more like an adjunct rather than the leader in Asia’s prudential supervisory role.

Legal Infrastructure: a Lynchpin A quick perusal of Appendix A shows a fair amount of attention to legal changes in Asia—a new banking law in Indonesia; hundreds of laws passed on Christmas Eve in Korea including a new Central Bank Law; bankruptcy code revisions in across Asia, changes to laws on secured interests, on collateral, on mergers and acquisitions. But this welter of legislation tough as it has been to pass—key laws are still pending in Thailand—is the easy part. Improved adjudication is absolutely critical to the protection of outside investors that in turn sparks financial development and growth. Here progress across Asia is decidedly mixed.

35

19 February 1998

Test Report Title

Part of the problem is simply a question of experience. But the issue runs deeper. In Korea, once a decision is made, foreclosure or reorganization can and has occurred relatively rapidly, though the court system is getting increasingly clogged. In SE Asia, though, the well publicized cases of Manulife in Indonesia or TPI in Thailand point to tremendous power in the hands of well-connected borrowers as against creditors. In Malaysia, the recent Malaysian Airlines deal or the Timedotcom deal point to a similar story of aggrieved investors (some of whom are now suing). The ability of creditors to recover has deteriorated to such an extent in Indonesia that even some distressed debt investor are pulling out. They would prefer to hold the equity of the borrowers, but can’t get a majority of the shares and certainly don’t want to be minority shareholders!

Weak legal infrastructure is slowing restructuring outright or pushing creditors into more cosmetic changes in loan agreements in the hopes that either their bargaining power will improve with reforms in the future, or that a surge in growth will bail them out. The global macro environment of the moment is dimming the latter’s chances. In the meantime the reaction of lenders is caution. As we noted at the start of the paper, credit growth remains well below earlier norms. With subdued investment and moderate capacity utilization and generally better cash flows, weaker credit flows are not a binding constraint now. But capacity use will rise with growth unless investment surges too. But outside finance is unlikely to come in the aftermath of such blatant evidence of the risks to creditors if the environment sours. How can Asia speed the implementation of a legal infrastructure that is more supportive of outside investors, of private sector monitoring or the enforcement of prudential regulations? Greater discretion by outside investors is a first step. Investments in firms which despite the legal environment have honored their obligations. Other processes are more long-term. Krozner (1997) argues that a push for technical change can be a catalyst for reform as such change creates new centers of wealth and authority. The dot com boom is a recent case in point, where despite the bust, significant shifts are occurring on the back of the technology. Unfortunately, how technical in the IT world feeds through into the legal infrastructure for finance is not entirely clear. One channel, though, that could be encouraging for Asia and financial restructuring is the increased weight of trade in world activity. Trade must increasingly mean dealing with

36

19 February 1998

Test Report Title

unknown counterparties where legal recourse becomes a more important element of business. As exporters come to rely on legal protections abroad, they could create a more powerful voice for similar protections at home. The real depreciations engendered by the crisis have helped to bolster the prospects of exporters (though the current global slowdown is not). As poorly functioning banking system become more of a constraint for exporters, Asia may find a group willing to support legal reform and through it bank restructuring.

37

19 February 1998

Test Report Title

Data appendix The country data set contains nine countries — Indonesia, Korea, Malaysia and Thailand of the crisis-affected economies group and China, Hong Kong, Philippines, Singapore and Taiwan of the other Asian economies group. To capture the differences caused by the crisis, the data set covers seven years from 1994 to 2000: three years before and three years after the crisis year, 1997. Data are mostly drawn from the database maintained at the Asia Pacific Economic and Market Analysis of the Salomon Smith Barney and Citibank in Hong Kong. GDP growth

Rate of growth of real GDP

Investment/GDP ratio

Ratio of gross investment to GDP

Real credit growth

Rate of growth of credit by deposit money banks, deflated by GDP deflator

Deposit growth

Rate of growth of deposits in deposit money banks, deflated by GDP deflator

Deposit/GDP ratio

Ratio of total deposits in deposit money banks to GDP

Real interest rate

Nominal interest rate (nominal rate on short-term government securities or, if that is not available, the rate charged by the central bank to domestic banks such as the discount rate) minus inflation rate

Real lending rate

Bank average lending interest rate minus rate of change of GDP deflator

Real deposit rate

Bank average deposit interest rate minus rate of change of GDP deflator

Interest spread

ending rate minus deposit rate

Inflation rate

Rate of change of the GDP deflator

Depreciation

Rate of change of the nominal exchange rate (period average, percentage)

Fiscal balance/GDP ratio

Government budget balance divided by GDP

Central bank funds/bank assets Loans from the monetary authorities to deposit money banks divided by total assets of those banks The bank-level data set was constructed by the Bank Analysts in the Asia Pacific Equity Research of the Salomon Smith Barney. The sample contains a total 29 banks, of which 8 are from Thailand, 9 are from Malaysia and 12 are from Hong Kong. Hong Kong bank data are also included in econometric estimation to both improve the statistical efficiency of the estimates and to provide a comparison of bank behavior in other economies. The data covers six years from 1995 to 2000 (as data for 1994 were generally not available). Net profit/assets

Ratio of after tax profits to total assets

Gross profit/assets

Ratio of gross profit to total assets

Interest margin

Ratio of net interest income (interest income minus interest expenditure) to total assets

Overhead/assets

Ratio of overhead expenses (personnel expenses and other non-interest expenses) to total assets

38

19 February 1998

Test Report Title

Loan loss provision

Ratio of loan loss provisions to total assets

Loan loss reserves

Ratio of loan loss reserves to total assets

Cash/assets

Ratio of cash and dues from banks to total assets

Deposits/assets

Ratio of total deposits (demand, saving, time, interbank and other deposits) to total assets

Equity/assets

Ratio of equity to assets

Loan/assets

Ratio of loans (commercial loans, public sector loans, consumer loans, secured loans and other loans, net of LLR) to total assets

OEA/assets

Ratio of other earning assets (deposit with banks, government securities, other investments and equity investment) to total assets

Asset growth

Growth rate of real total assets (applying CPI as deflator)

Credit growth

Growth rate of total real credit (applying CPI as deflator)

Deposit growth

Growth rate of total real deposits (applying CPI as deflator)

39