This Selected Issues paper reviews the aggregate balance-sheet analysis that describes the improvement in Thailand's overall balance sheet since the crisis, and also highlights the potential vulnerabilities. It assesses the public debt and contingent liabilities, and developments in the banking sector. It discusses the operations of Specialized Financial Institutions and related regulatory issues, the financial and corporate sector restructuring, and also presents an overview of developments in nonperforming loans and assets. It reviews the growth without credit feature of Thailand, which explains how firms have financed their operations after the crisis.


This Selected Issues paper reviews the aggregate balance-sheet analysis that describes the improvement in Thailand's overall balance sheet since the crisis, and also highlights the potential vulnerabilities. It assesses the public debt and contingent liabilities, and developments in the banking sector. It discusses the operations of Specialized Financial Institutions and related regulatory issues, the financial and corporate sector restructuring, and also presents an overview of developments in nonperforming loans and assets. It reviews the growth without credit feature of Thailand, which explains how firms have financed their operations after the crisis.

V. Financial and Corporate Sector Restructuring1

A. Introduction

1. Thailand has made substantial progress in restructuring the financial sector and resolving non-performing loans (NPLs).2 Some progress has also been made in strengthening the balance sheets of the corporate sector, and recent improvements in profitability and balance sheet ratios are encouraging signs.3 However, despite the earlier progress private banks still have substantial amounts of NPLs on their books that limit their operations. The corporate sector is further burdened by the large amounts of NPLs transferred to asset management companies (AMCs), now part of the pool of non-performing assets (NPAs)4, The progress in reducing nonperforming assets is hampered by a weak legal framework and the sheer number of cases that imposes large administrative burdens on both the financial and the legal system. This chapter presents an overview of the flows and stocks of NPLs and a discussion of how resolution of NPLs and NPAs may affect growth prospects. It then details some of the progress made by the Thai Asset Management Corporation (TAMC) before concluding with a section on recent and potential future initiatives to deal with financial and corporate sector restructuring.

B. A History of Non-performing Loans and Assets

2. Although non-performing loans have declined substantially since the peak in 1999, NPL levels remain high, especially for private banks (Chart). Initially, the NPL problem was much more severe at state-owned banks but, as they have benefited from larger transfers of NPLs to AMCs, the situation is now reversed. At the beginning of 2003, NPLs at private banks were over 20 percent of the total loan portfolio, while state-owned banks had NPLs on average of around 8 percent of total loans, down from the peak of around 70 percent.5


NPLs (% of total loans)

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005

3. Although outflows from the pool of NPLs have exceeded inflows and reduced the stock of NPLs substantially, new and re-entry NPL flows have been significant. Since the last quarter of 1999, outflows from the pool of NPLs have amounted to a total of B 3.1 trillion (or two thirds of the initial loan portfolio). Restructured NPLs account for close to 40 percent of the decline, transfers to AMCs another 30 percent, loan reclassification for close to 10 percent, and write-offs and other for the remaining 20 percent. However, there have also been substantial inflows to the pool of NPLs (on the order of 27 percent of the initial loan portfolio), with new and re-entry NPLs contributing almost equal amounts (Charts). The large share of reentry NPLs has raised questions about the quality of restructuring deals. Cumulative over the period, re-entry NPLs are over 40 percent of restructured and reclassified loans, or in other words, 4 out of 10 “resolved” NPLs have re-entered the pool of NPLs. More recently, the rate of new NPLs has declined, while re-entry NPLs remain relatively high and volatile, which raises concerns about the quality of past restructurings. In 2002, the inflows to the pool of NPLs were around B 300 billion (or almost 8 percent of the total loan portfolio in 2002) despite the pick-up in economic activity.


Cumulative NPL lnflows

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005


Cumulative NPL Outflows

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005

4. With easy access to credit prior to the crisis, many sectors of the economy contributed to the build-up of NPLs at banks. Manufacturing accounted for around a quarter of peak level NPLs, while real estate and wholesale trade each accounted for around a sixth.6 Somewhat more surprising is perhaps that personal credits accounted for a non-trivial amount of NPLs (10 percent of peak levels). Their share has increased over time to 15 percent as such credits have heen resolved at a slower rate than NPLs in other sectors. The recent increase in credit to households (albeit from low levels) is a development that should he closely monitored to avoid a further build up of household’ NPLs. This sectoral picture is somewhat misleading in terms of actual resolution of distressed assets, since manufacturing and real estate loans now constitute a relatively large share of distressed assets with the TAMC. However, the prominence of these sectors on bank’ balance sheets has diminished.


Total NPLs by Sector 50% of GDP at the Peak in 1999

(Sector NPLs in percent of GDP)

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005


Total NPLs by Sector 14% of GDP at the end of 2002

(Sector NPLs in percent of GDP)

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005

5. Because much of the reduction in bank’ NPLs has been achieved by transfers to AMCs, system-wide non-performing assets (NPAs) remain high. Distressed assets in financial institutions and AMCs are around B 2 trillion, or almost 40 percent of GDP.7 The TAMC has become the largest holder of NPAs, with the book value of impaired assets close to 15 percent of GDP. Almost two thirds of system-wide NPAs is accounted for by AMCs, and of this, only a small fraction is in private AMCs. A successful restructuring of NPAs in the AMCs is therefore key lo ensure that taxpayers are not unduly burdened by the financial crisis. NPLs in private banks are also substantial (at around 11 percent of GDP), and resolution of these would significantly improve the financial position of banks and most likely improve the prospects for new credits.


Total Digressed Assets at 50%of GDP at their peak in 1999

(Distressed assets in percent of GDP)

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005


Total Distressed Assets at of GDP in 2002

(Distressed assets in percent of GDP)

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005

6. Thailand is lagging regional peers in terms of outstanding amounts of distressed assets. Korea enjoys the lowest levels of NPAs at 7 percent of total loans, while currently, only Indonesia has distressed assets to total loans at a comparable level with Thailand’s 37 percent. To a large extent, Korea’s favorable position today reflects better initial conditions, but in addition, restructuring has reduced the NPA ratio by two thirds, compared to Thailand where the progress has been significantly slower. Contributing to the relatively speedy asset resolution in Korea is the work of KAMCO, which purchased a large share of problem loans early on and used market mechanisms to dispose of them. Thailand relied on a more private sector-based resolution strategy initially and the TAMC was not set up until 2001.8


Distressed Assets in 1999 and 2002

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005

C. Potential Gains from Restructuring

7. The benefits from resolving NPLs and NPAs can be substantial over the medium term as both the financial and corporate sectors will be strengthened. Policy recommendations on speeding up financial and corporate sector restructuring are based on arguments that the level of distressed assets bears heavily on the financial sector and the Thai economy, Those arguments include: (i) Corporate’ investment decisions and financing options as well as consumer demand are hampered by uncertainties about the conditions for their debt that still needs to be restructured, (ii) NPLs contribute to risk aversion that reduce bank’ willingness to lend, which may restrict the scope for real sector growth, (iii) The high level of NPLs is costly for banks and reduces profits in banks, (iv) The stock of distressed assets locks up resources from a productive use in the economy.

8. As an indication of the link between NPLs and the future earnings prospects of corporates, sectors with relatively low NPL ratios have seen more favorable developments in share prices (Chart). If banks are less willing to lend to companies in sectors that account for large amounts of NPLs, this may lead to reduced earning potentials for the sector. Using the Bank of Thailand’s (BOT) data on NPLs by sector since 1999 to end-2002 and SET’s sector indices in panel regressions reveal a statistical significant relationship between NPL ratios and share prices. Although the estimates do not reveal what the impact on macroeconomic growth is from resolving NPLs, it clearly shows that sectors with low NPL ratios enjoy higher valuations than those with high NPLs. To the extent improvements in stock prices are correlated with improved earnings potentials that generate economic growth, this is another indication of the benefits of NPL resolution.


Sector panel 1999–2002

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005

9. The low-NPL sectors have indeed benefited from greater access to bank credit, which may be a partial explanation of the correlation between sectoral NPLs and share prices. Using the same data, but looking at sectoral credit growth, there is a statistically significant relationship between low NPL ratios and access to credit (Chart). These results are obtained in simple panel regressions, and arc robust to specification changes that vary between a common intercept and fixed effects and to the inclusion of time dummies. Firms may be able to finance investments and operations from alternative sources, such as retained earnings or by issuing debt, during some periods. However, bank credit is an important part of a well-functioning financial system and could facilitate growth in the corporate sector. The link between economic growth and access to credit is discussed in more detail in Chapter VI.


Sector panel 1999–2002

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005

10. Due to uncertainties about recovery rates, NPLs on banks balance sheet may affect their willingness to extend credit even if provisioning is adequate in a regulatory sense. If banks provision according to prudential regulations and provisioning requirements are expected to meet future losses, banks should in principle not worry about NPLs, but would base their lending decisions on their capital ratios and lending opportunities. If banks arc assumed to be fully provisioned against future losses, the existence of a relationship between NPLs and credit growth suggest that uncertainty surrounding the resolution of NPLs is sufficient to discourage banks from extending credit. Using a panel data set covering 1999–2002 of individual bank’ NPLs, credit growth and capital ratios, the estimates suggest that there is a statistically significant relationship between NPLs and credit growth, even after controlling for the capital position of banks. The relationship seems to be stronger for private banks than for state-owned banks, which could reflect different lending policies and attitudes towards risk, or perhaps less likely, different perceptions of the credit-worthiness of borrowers (Chart).


Panel of Private Banks 1999–2002

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005

11. It is argued that the lack of good borrowers rather than banks’ willingness to lend is the factor constraining currently lending. The argument is based on the observation that there is excess liquidity in the system, and that banks thus would extend loans if they could find good borrowers. However, if risk aversion makes bank charge borrowers a higher interest rate than they otherwise would, lowering risk aversion and thus lending rates would in most circumstances increase the profitability of firms and improve the borrower’ credit worthiness. The volume of good credits is therefore a function of lending rates and not a static concept, and restructuring bank’ balance sheets may help increase the pool of credit-worthy borrowers. Finally, a more practical reason for a link between large NPL portfolios and slow credit growth may be that bank managers are so busy dealing with NPL-customers that there is not enough time to spend to evaluate new projects. These additional arguments suggest that resolving NPLs would increase banks ability and willingness to extend credit.

12. Current and future costs associated with the stock of NPLs in Thai banks are mostly caused by uncertain recovery values and to a lesser extent by funding costs. Once NPLs are provisioned 100 percent, i.e. the losses from NPLs have been fully recognized, the carrying cost is zero even if the assets remain on bank’ balance sheets. However, a large share of NPLs are only partially provisioned since some recovery from these loans is expected. These NPLs affect bank’ profitability in two ways. First, there is a funding cost associated with the non-provisioned part of the loan. In the current low interest rate environment, with deposit rates around 1 percent, this cost is limited (see Chapter IV for more details on bank’ balance sheets). The second cost is related to potentially insufficient provisioning of current NPLs, which would lead to additional provisionings in the future that reduce profits. The large share of NPLs and the great dispersion in provisioning rates suggests that additional provisioning may be a substantially greater burden on the banking system than the direct funding costs. Estimates of these costs naturally depend on assumptions on the adequacy of current provisioning and future economic developments, but in adverse scenarios, these costs could be more than ten times the current carrying cost of NPLs. In other words, the level and variance of bank’ profits will be impacted more by changes in loan valuations than by carrying costs, again suggesting that resolving NPLs would reduce uncertainties and may increase banks’ risk appetite.

13. Returning assets that are currently tied up in the restructuring process to more productive uses can have a significant impact on growth. With NPAs on the order of 40 percent of GDP, this effect in itself could be of the order of half of a percent of GDP per year if it is assumed that the market value of the NPAs is on average 50 percent of book value and the assets return around 2.5 percent additionally per year once they are out of the pool of NPAs. The impact on GDP growth is clearly dependent on the assumptions, and a more detailed analysis of the underlying assets would be needed to produce a reliable number. Quantifying the macrocconomjc impact of a healthier banking system on growth is even more difficult, but overall, it is clear that the potential benefits are substantial over the medium term, and suggests that NPA resolution should be a priority area for enhancing future growth prospects.

14. A cross-country comparison between GDP growth rates and NPA ratios provide some macroeconomic evidence that further supports the case for rapid NPA resolution. For the Asian countries compared above, there has been a remarkably strong correlation between average NPAs and growth rates between 1999 and 2002. If the simple regression line is taken as a. measure of how reduced NPAs affect growth rates, this would suggest that growth picks up by almost one percentage point for every ten percentage point reduction of NPAs. This chart is based on five data points and obviously does not address issues of simultaneity or control for other important variables that affect growth, so the quantification of the effect is only suggestive. The strong relationship is nonetheless noteworthy and consistent with the microeconomic stories discussed above.


Distressed Assets vs GDP Growth

(Averages 1999–2002)

Citation: IMF Staff Country Reports 2004, 001; 10.5089/9781451836806.002.A005

D. Restructuring by the TAMC9

15. Taken at face value, the restructuring progress by the TAMC has been impressive. Of the B 761 billion transferred to the TAMC, resolution of B 535 billion worth of assets (or 70 percent of total) have been approved by the Executive Committee. Around half of the progress comes from debt and business restructuring, and rehabilitation in the central bankruptcy court. Foreclosure accounts for most of the remaining 50 percent of assets approved for resolution.

16. The initial picture of substantial progress is less clear once more detailed information about the resolution process is analyzed. Of the 795 debt restructuring cases, only 294 cases have been signed by the debtors, and almost two thirds of the cases under this heading still have some way to go before a final agreement can be reached. A similar qualification applies to the cases that are in the process of being foreclosed, since foreclosure is currently used as a way of bringing debtors to the negotiating table, if this strategy works, more cases will be restructured rather than foreclosed. It is also possible that some cases currently classified under debt restructuring will eventually move to the foreclosure procedure, given that many restructuring cases have to be finalized and signed by the debtor.

article image

17. In order to evaluate the progress of the TAMC, more information will have to be disclosed on a frequent basis. A key measure to the success of the TAMC is cash recoveries from all types of resolution procedures. So far, cash recoveries amount to around B 8 billion, or 1.5 percent of the B535 billion book value of all the “resolved” cases. The expected 45 percent recovery rate thus require substantial cash flows, going forward To monitor the progress to achieve the target, cash flow projections should be published and compared with actual cash recoveries. Since a substantial share of cash recoveries will only be due many years from now, additional information on the progress should be made available to evaluate the progress in the near-term. For example, more details can be made public on where in the restructuring or foreclosure process cases are (e.g, deals signed or assets transferred to the TAMC). This type of information could then be published regularly as the IT system and back-office routines are developed.

E. Looking Ahead

18. To speed up the restructuring process, the BOT is actively pursuing strategies to resolve NPLs in banks and the government is searching for new ways to reduce outstanding NPAs and tackle strategically important cases. The BOT has recently presented new initiatives with ambitious timelines to restructure NPLs. The government is also pushing to make progress in restructuring impaired assets transferred to the TAMC, and is pursuing alternative strategies to reduce distressed assets.

19. The progress of restructuring debt under the aegis of the Corporate Debt Restructuring Advisory Committee (CDRAC) seems to have slowed significantly in the last couple of years. Although substantial progress was made initially with the help of CDRAC (around B 2.5 trillion had been restructured this way by the end of 2001), the amounts restructured seem to have plateaued, and reached B 2.8 trillion at the end of 2002.10 One factor is that the more complicated cases are the ones remaining. Some cases have also become re-entry NPLs due to overly optimistic cash flow projections or creditor’ unwillingness to take up-front hair-cuts. The BOT has recently introduced more ambitious timelines for cases in the restructuring process. However, so far the impact on the pace of restructuring has been limited.

20. The CDRAC process will face an additional challenge at the end of the year, when the tax incentives granted to debtors that restructure debt under the BOT guidelines expire. The CDRAC has pushed for an extension of the tax incentives, but so far this has not been approved. The reason for making the tax incentive expire at some date is to reinforce the urgency of reaching a restructuring agreement. However, once the tax incentives have expired, it is not clear what the benefits arc for the debtor to take part in the BOT-sponsored negotiations. If the incentives indeed are removed and fewer debtors choose to restructure debt this way, this may be another argument for considering reopening the TAMC for additional transfers.

21. The government has recently decided to set up mutual funds that could facilitate corporate sector restructuring. The so-called Vayupak funds will be 10 year closed-end funds, include the government’s holding of financial sector shares, and invest in shares and distressed loans in important financial institutions and corporates in need of capital. The private sector will also be invited to invest in the fund. The government views this as an additional tool to facilitate the needed restructuring of me financial and corporate sectors. However, some concerns have been voiced that if guarantees are attached to the fund, this would add to the contingent liabilities of the government, and perhaps more importantly, signals the intention of the government to intervene in a way that may distort incentives for private sector debt work-outs. Since the funds are not yet operational, it is too early to evaluate the merits and drawbacks of this approach to restructuring.

22. There are a number of factors that will influence how willing and able both debtors and creditors are to restructure NPAs without the help of the government. On the debtor side, a sustained economic recovery and a low interest rate environment will increase their ability to pay. However, their willingness to pay will depend on pressures from the legal system and other incentives to participate in the restructuring process (e.g., tax incentives or access to new loans). On the creditor side, the restructuring process depends, inter-alia, on the willingness and capacity of financial institutions to realize losses (by offering “hair-cuts”) that may further affect their profits and balance sheets. If the restructuring process on the creditor side is held up by limited creditor rights in me legal system rather than to insufficient provisioning, this suggests that additional legal reforms would be needed to promote NPL resolution in the private sector.

23. The absence of reforms that create an enabling legal framework for debt workouts could strengthen the case for a new round of transfers of impaired loans to the TAMC. A first-best solution of banks NPL problems is to create a legal environment that balances creditor and debtor rights and facilitates restructuring of loans within the private sector. However, changing the legal framework and behaviors may take too long and motivate a more centralized resolution of NPLs that utilities the special legal powers of the TAMC. If banks have not fully provisioned against expected losses (compared to what is expected to be recovered by the TAMC rather than by the bank), pricing of NPLs transferred to the TAMC would be more complicated since the government will have to decide to what extent bank stakeholders or tax payers should bear the additional cost of restructuring.

24. If risk aversion and uncertainty about the value of NPLs arc key determinants of bank lending, this has implications for future initiatives to clean up banks’ balance sheets. In the case of further transfer of NPLs from private banks to the TAMC to help clean up banks balance sheets, the issue of gain/loss sharing arrangements will affect bank’ risk perception of this type of restructuring. In turn, this may affect bank’ willingness to make transfers to the state AMC, and perhaps more importantly, influence future lending behavior. Assuming that banks currently have provisioned for the expected value of future losses and that transfers occur at a price consistent with such provisioning, neither banks, nor the state AMC would expect to make further losses when the NPLs are resolved.

25. It is conceivable that banks would be willing to pay the government an “insurance premium” for transferring loans without future gain/loss sharing arrangements by selling loans below current valuations. This “insurance premium” would lead to an immediate loss for banks in exchange for reduced uncertainty. However, if the premium is too high, this could reduce CARs below regulatory or desired levels. The question of pricing is further complicated if estimated collateral values differ between banks and the government. If a bank is deemed to have overvalued collateral, it may need to raise capital even if NPLs are transferred at expected values, and without any discount to avoid future gain/loss sharing arrangements. The uncertainty associated with collateral valuations, potentially undcrprovisioned banks, and aversion to gain/loss sharing arrangements could make price determination rather complicated and slow down the resolution of NPLs on private banks’ balance sheets.

26. The pricing of NPLs and how to deal with gain/loss sharing is ultimately a political decision. It has to weigh the benefits from cleaned up balance sheets of banks against the fiscal cost and possible redistribution of income between bank shareholders and tax payers. The benefit from relieving the banks from a gain/loss sharing arrangement is that banks may become less risk averse and extend more credit. However, transfers prices could also be viewed by debtors as setting benchmarks for “hair-cuts”, and may affect borrowers willingness to service their debt. It may also have more long-term implications for the credit culture that has to be considered in resolving current NPLs. Finally, the government carries the risk of future losses if recoveries are realized below expectations, which would affect the fiscal position. Obviously, the government could also stand to benefit if economic and other conditions improve debtors ability and willingness to pay their debts.


Prepared by Torbjön Becker.


For earlier treatments of financial and corporate sector restructuring, see Chapters II and III in Thailand: Selected Issues, IMF Staff Country Report No. 00/21 and Chapter II in Thailand: Selected Issues, IMF Country Report No. 01/147.


For details on corporate sector performance, see Haksar and Kongsamut, 2003, “Dynamics of Corporate Performance in Thailand,” forthcoming IMF Working Paper.


The definition of NPAs that is used includes NPLs at banks as well as the impaired assets transferred to AMCs. In other words, an NPL that is no longer with a bank is not called an NPL, but is still part of the pool of distressed assets until it is resolved.


The Bank of Thailand’s definition of NPLs changed in December 2002. The earlier definition included loans that were more than 3-month overdue and excluded doubtful of loss loans with mil provisioning, whereas the new definition includes loans classified as substandard, doubtful, doubtful of loss and loss.


The sectoral breakdown of NPLs include loans at foreign branches and finance companies, and the total level of NPLs will therefore differ from the previous chart that focuses on Thai commercial banks. Also, the “other” sector includes in order of importance, import businesses, banking and financial services, export businesses, agriculture, public utilities, mining and hire-purchase businesses.


Views may differ on what to include in the measure of distressed assets. Here we use a rather conservative measure in that restructuring deals that have been reached at the TAMC arc not subtracted from the pool of NPAs. Once progress on restructuring is more verifiable, for example by cash recoveries, part of the impaired assets at the TAMC and other AMCs could be taken out of the measure of NPAs. On the other hand, loans restructured and reclassified by banks immediately leave the pool of NPLs at banks. This may be overly optimistic in this measure of the overall level of distressed assets given the high rate of re-entry NPLs.


For a comparison of centralized AMCs, see Chapter III in Thailand: Selected Issues, IMF Country Report No. 01/147.


For an introduction to the TAMC, see Chapter III in Thailand: Selected Issues, IMF Country Report No. 01/147.


For more details on the operation of CDRAC, see Chapter II in Thailand: Selected Issues, IMF Staff Country Report No. 00/21.

Thailand: Selected Issues
Author: International Monetary Fund