Abstract
This Selected Issues paper characterizes the rapid expansion of Hong Kong Special Administrative Region’s (Hong Kong SAR) economic ties with the Mainland over the last two decades. It examines the possible impact on Hong Kong SAR of policy developments in the Mainland. The paper concludes that as integration has progressed, developments in various sectors of the Hong Kong SAR economy have become increasingly tied to developments on the Mainland. This paper also analyzes the initial episode of strong-side pressures on the Hong Kong dollar and, in particular, the Hong Kong Monetary Authority response.
III. Procyclicality of Credit in the Banking System in Hong Kong SAR1
A. Introduction
1. Credit growth is procyclical in most countries and can often amplify the economic cycle. Credit naturally tends to expand during upturns and contract during downturns but can fuel excessive investment during booms and deepen cyclical downturns when the financial system fails to adequately discipline lending. Asset prices and property prices, in particular, can play an important role in this process. Rising asset prices encourage more rapid credit growth by raising the value of the collateral against which credit is offered. Similarly, during a downturn, the decline in asset prices can exacerbate the decline in credit.
2. Some degree of procyclicality in credit is inherent in any economy, and can occur even in the absence of financial market imperfections. The returns on investment are often sensitive to macroeconomic conditions and a firm may choose to reduce investment during a recession even though it has access to a willing lender. Households may also choose to postpone purchases that they would finance with credit during downturns.
3. The banking systems can amplify the natural tendency towards a procyclical credit through a number of channels. For example:
Capital requirements can cause banks to tighten credit following losses during a downturn in order to limit the decline in their capital ratios, which can deepen the recession.
Agency problems can increase the role of collateral and borrower’s prospective cash flows in lending decisions (Bernanke and Gertler, 1989), making credit more sensitive to changes in asset prices and borrowers’ balance sheets. During cyclical downturns, for example, the decline in cash flow and collateral value can reduce the creditworthiness of some borrowers, leading to a contraction in credit. 2
Possible coordination problems can exacerbate procyclicality when the decision to reduce lending during a downturn is the prudent response for each individual bank but leads to a collectively inefficient outcome as the aggregate impact on credit can deepen the downturn.
4. Since some degree of procyclicality is normal, the policy challenge facing country authorities is to prevent the financial system from exacerbating it. This would limit the costs of high procyclicality, including the risk to financial stability from deeper cyclical downturns, the lower quality of investment when credit is growing rapidly and the failure to finance profitable investment projects when credit is contracting. There may be little supervisors can do about information and agency problems. However, they can try to limit the extent to which the banking system amplifies procyclicality due to weak credit assessment and an excessive reliance on collateral. They could improve incentives to better measure and control credit risk over the cycle. Efforts to expand securitization and insurance of property loans could reduce the banks’ risk exposure to the sector. More extreme measures include adjusting capital adequacy ratios over the cycle and provisioning guidelines to encourage portfolio diversification and dampen credit expansions and contractions.
5. The correlation of credit growth with respect to growth in GDP and in property prices tends to be higher on average in Hong Kong SAR than in comparable Asian economies. Craig et al. (2004) study procyclicality in East Asian financial systems. They provide a cross-country comparison of the correlations between growth in credit to the private sector and growth in GDP and in property prices. Table III.1 shows some of their findings. While stronger positive correlations suggest higher procyclicality, caution is needed when interpreting these correlations or comparing them across countries.3
Correlation Coefficients with Year-on-Year Growth in Real Credit 1/
Figures based on quarterly data from 1984 to the present available through CEIC.
Correlation Coefficients with Year-on-Year Growth in Real Credit 1/
| Year-on-Year Growth in Real GDP | Year-on-Year Growth in Real House Prices | |
|---|---|---|
| Hong Kong SAR | 0.44 | 0.66 |
| Australia | 0.50 | 0.34 |
| Mainland China | -0.15 | n.a. |
| Japan | 0.65 | 0.85 |
| Korea | 0.35 | 0.35 |
| Malaysia | 0.49 | 0.71 |
| New Zealand | 0.03 | 0.10 |
| Philippines | 0.43 | n.a. |
| Singapore | 0.35 | 0.42 |
| Thailand | 0.35 | -0.23 |
| Average | 0.33 | 0.39 |
Figures based on quarterly data from 1984 to the present available through CEIC.
Correlation Coefficients with Year-on-Year Growth in Real Credit 1/
| Year-on-Year Growth in Real GDP | Year-on-Year Growth in Real House Prices | |
|---|---|---|
| Hong Kong SAR | 0.44 | 0.66 |
| Australia | 0.50 | 0.34 |
| Mainland China | -0.15 | n.a. |
| Japan | 0.65 | 0.85 |
| Korea | 0.35 | 0.35 |
| Malaysia | 0.49 | 0.71 |
| New Zealand | 0.03 | 0.10 |
| Philippines | 0.43 | n.a. |
| Singapore | 0.35 | 0.42 |
| Thailand | 0.35 | -0.23 |
| Average | 0.33 | 0.39 |
Figures based on quarterly data from 1984 to the present available through CEIC.
6. In Hong Kong SAR, the concentration of property lending in bank loan portfolios could contribute to procyclicality. The link between collateral values and credit is stronger in property-related lending because property is the primary form of collateral and property prices tend to be highly sensitive to economy-wide shocks.4 Thus, a more concentrated exposure to property raises the vulnerability of the banking system to aggregate shocks that affect property prices, which can contribute to procyclicality in lending. These effects have been dampened, however, by the high level of liquidity in the banking system.
7. The procyclicality of credit has not posed a threat to Hong Kong SAR due to its well supervised and sound banking system. A major potential cost of high procyclicality is the risk to financial stability from deeper cyclical downturns. The very well capitalized and highly liquid banking system should be able to effectively weather such downturns, suggesting that this risk has been, and is likely to remain, very low. The resilience of Hong Kong SAR’s banking system in the wake of the Asian crisis is particularly noteworthy, and facilitated the recovery from the crisis.
B. Features of Hong Kong SAR’s Financial System Relevant to Procyclicality
8. Hong Kong SAR has a large and well developed equity market, which should attenuate the effects of procyclical credit. The stock market in Hong Kong SAR is the largest in Asia after Japan. Its market capitalization currently stands at about HK$5.5 trillion, which is roughly three times the size of outstanding bank loans. Equity financing can reduce procyclicality since it makes firms less vulnerable to cyclical changes in the supply of credit.
9. The bond market remains relatively small, with bank lending accounting for a large share of debt financing. The outstanding stock of Hong Kong dollar debt in 2003 (excluding that in the Exchange Fund) was HK$438 billion, of which 32 percent is associated with Authorized Institutions (AIs). Local corporates account for only HK$33 billion, which is less than 2 percent of total loans for use in Hong Kong SAR. An expansion of the bond market could potentially dampen procyclicality.
10. The share of property loans in bank lending has steadily increased, going from 39 percent in 1991 to 58 percent in 2003, even as banks became more liquid. Figure III.1 plots the evolution of loans measured in constant prices broken down by three different categories including two types of property loans—private residential loans and loans for property building, construction, development and investment—and other types of loans for use in Hong Kong SAR.5 Loans for use in Hong Kong SAR declined following the Asian financial crisis, but by 2000 they had resumed their growth path, albeit at a slower rate than prior to the crisis. However, property loans continued to expand throughout the period, increasing their share in total loans.
11. The increasing share of property loans in bank lending is largely attributable to weak credit demand by other sectors following the Asian crisis. The weak demand from other sectors, combined with the good performance of property loans despite the bursting of the property bubble, led to a continuing expansion of the latter. In fact, property-related loans appear to have been safer than other types of loans during this period (Gerlach et al., 2004). Specific features of the Hong Kong SAR property market help attenuate credit risk in this sector. For example, the practice of pre-selling units reduces the leverage of developers, the prudential limit on the loan to value ratio is relatively low, and the stigma of defaulting has prevented it even in cases where mortgages had negative equity. Finally, many home buyers have substantial net worth and so may be better able to avoid default when there are adverse shocks.
C. Extent of Procyclicality of Bank Lending in Hong Kong SAR
12. The growth rate of bank loans is positively correlated with the growth rate of GDP and with property prices. Quarterly data on bank loans are available broken down by sector beginning in 1981Q4. 6 All series are converted to constant 2000 prices using the GDP deflator. All growth rates are based on year-on-year changes. Table III.2 presents correlation coefficients between growth in loans and growth in GDP and in property prices, as well as the elasticities implied by univariate regressions. The results indicate that:
Growth in loans for use in Hong Kong SAR is strongly positively correlated with growth in GDP, and the estimated elasticity suggests a strong comovement (Figure III.2) that has strengthened over the last 10 years.
The comovement of loans with equity prices is relatively weak.
Growth in loans for use in Hong Kong SAR is correlated with changes in property prices and the correlation and the associated elasticity have increased substantially in the last ten years (Figure III.3). This increase in comovement may be due to the simultaneous large declines in property prices and loans following the Asian crisis
Correlation Coefficients and Elasticity of Growth in Loans for Use in Hong Kong SAR with Respect to: 1/
Elasticities based on a univariate regression of year-on-year real growth in the corresponding variable on the real growth in loans. The symbol * indicates a statistically significant coefficient at the 5 percent level.
Correlation Coefficients and Elasticity of Growth in Loans for Use in Hong Kong SAR with Respect to: 1/
| 1981Q4-2004Q1 | 1994Q2-2004Q1 | |||
|---|---|---|---|---|
| Real Growth in: | Correlation | Elasticity | Correlation | Elasticity |
| GDP | 0.22 | 0.37 * | 0.35 | 0.52 * |
| Hang Seng index | -0.23 | -0.06 * | -0.03 | -0.01 |
| Residential property prices | 0.22 | 0.10 * | 0.61 | 0.22 * |
| Office property prices | 0.32 | 0.10 * | 0.31 | 0.10 |
Elasticities based on a univariate regression of year-on-year real growth in the corresponding variable on the real growth in loans. The symbol * indicates a statistically significant coefficient at the 5 percent level.
Correlation Coefficients and Elasticity of Growth in Loans for Use in Hong Kong SAR with Respect to: 1/
| 1981Q4-2004Q1 | 1994Q2-2004Q1 | |||
|---|---|---|---|---|
| Real Growth in: | Correlation | Elasticity | Correlation | Elasticity |
| GDP | 0.22 | 0.37 * | 0.35 | 0.52 * |
| Hang Seng index | -0.23 | -0.06 * | -0.03 | -0.01 |
| Residential property prices | 0.22 | 0.10 * | 0.61 | 0.22 * |
| Office property prices | 0.32 | 0.10 * | 0.31 | 0.10 |
Elasticities based on a univariate regression of year-on-year real growth in the corresponding variable on the real growth in loans. The symbol * indicates a statistically significant coefficient at the 5 percent level.
Total Loans by Type of Use (2000 Constant Prices)
(In billions of Hong Kong Dollars)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.Total Loans by Type of Use (2000 Constant Prices)
(In billions of Hong Kong Dollars)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.Total Loans by Type of Use (2000 Constant Prices)
(In billions of Hong Kong Dollars)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.
Real Growth of GDP and of Loans for Use in Hong Kong SAR
(Year-on-year percentage change)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.Real Growth of GDP and of Loans for Use in Hong Kong SAR
(Year-on-year percentage change)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.Real Growth of GDP and of Loans for Use in Hong Kong SAR
(Year-on-year percentage change)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.
Real Growth of Loans for Use in Hong Kong SAR and of Property Prices
(Year-on-year percentage change)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.Real Growth of Loans for Use in Hong Kong SAR and of Property Prices
(Year-on-year percentage change)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.Real Growth of Loans for Use in Hong Kong SAR and of Property Prices
(Year-on-year percentage change)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.13. Craig et al. (2004) show that provisions expand when loans contract, contributing to procyclicality, and that credit comoves with the economic cycle in Hong Kong SAR. These authors use a panel of bank-level data from BankScope to estimate the determinants of loan growth, asset margin and actual provisioning rates. The estimated elasticity of loan growth with respect to GDP growth is about 1, while the estimated elasticity of the actual provisioning rate with respect to loan growth is about -2.
14. The larger role of the property sector in the lending portfolio of banks appears to contribute to the procyclicality of credit. Property-related loans move with the property cycle, with the strength of this comovement being stronger for residential property prices than for commercial property prices (Table III.3). The role of property collateral values in property-related lending may help explain this comovement. Demand-side effects may also contribute (for example, higher property prices increase the size of the loan needed to finance a given property purchase). As bank lending becomes more concentrated in the property sector, the procyclicality of credit could increase because property prices are relatively volatile and procyclical, which leads to comparable movements in collateral values. These series are plotted in Figure III.4.
Correlation Coefficients and Elasticity of Growth in Loans with Respect to Growth in Property Prices in the Last Ten Years (1994Q2-2004Q1) 1/
Elasticities based on a univariate regression of year-on-year real growth in the corresponding property variable on loan variable. The symbol * indicates a statistically significant coefficient at the 5 percent level.
Correlation Coefficients and Elasticity of Growth in Loans with Respect to Growth in Property Prices in the Last Ten Years (1994Q2-2004Q1) 1/
| Real Growth in Residential Property Prices | Real Growth in Office Property Prices | |||
|---|---|---|---|---|
| Real Growth in | Correlation | Elasticity | Correlation | Elasticity |
| Private residential loans | 0.31 | 0.10 | -0.18 | -0.05 |
| Loans for property building, const., dev. and inv. | 0.50 | 0.31 * | 0.30 | 0.16 |
| Nonproperty loans for use in Hong Kong SAR | 0.62 | 0.26 * | 0.45 | 0.16 * |
Elasticities based on a univariate regression of year-on-year real growth in the corresponding property variable on loan variable. The symbol * indicates a statistically significant coefficient at the 5 percent level.
Correlation Coefficients and Elasticity of Growth in Loans with Respect to Growth in Property Prices in the Last Ten Years (1994Q2-2004Q1) 1/
| Real Growth in Residential Property Prices | Real Growth in Office Property Prices | |||
|---|---|---|---|---|
| Real Growth in | Correlation | Elasticity | Correlation | Elasticity |
| Private residential loans | 0.31 | 0.10 | -0.18 | -0.05 |
| Loans for property building, const., dev. and inv. | 0.50 | 0.31 * | 0.30 | 0.16 |
| Nonproperty loans for use in Hong Kong SAR | 0.62 | 0.26 * | 0.45 | 0.16 * |
Elasticities based on a univariate regression of year-on-year real growth in the corresponding property variable on loan variable. The symbol * indicates a statistically significant coefficient at the 5 percent level.
Real Growth of Property and Nonproperty Loans for Use in Hong Kong SAR and of Property Prices
(Year-on-year percent change)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.Real Growth of Property and Nonproperty Loans for Use in Hong Kong SAR and of Property Prices
(Year-on-year percent change)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.Real Growth of Property and Nonproperty Loans for Use in Hong Kong SAR and of Property Prices
(Year-on-year percent change)
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.15. Procyclicality in the property sector can also spill over into other sectors. Movements in property prices could contribute to the procyclicality of credit to other sectors where property is also used as collateral. In fact, the comovement of nonproperty-related loans with the property cycle is roughly the same as some types of property loans (Figure III.3). This suggests that such spillovers do occur and that property collateral may indeed play a significant role in nonproperty-related lending in Hong Kong SAR.
D. Options for Policy Makers
16. Banks do not internalize the effects of their actions on the rest of the economy so their actions may deviate from what is optimal from the perspective of policy makers. For example, reducing credit during a downturn may be an individually optimal prudential response for each bank but could lead to a collectively inefficient outcome by contributing to a credit crunch that exacerbates the downturn. As a result, policy makers may be more willing to accept lower credit growth during upturns in order to reduce the need for an adjustment during downturns relative to what would result from banks’ lending decisions. One step in this direction would be for policy makers to ensure that property lending is indeed based on capacity to repay as well as collateral. This would reduce the potential for property price changes to exacerbate the cycle and reduce the risk of default in downturns. Bank supervisors in Hong Kong SAR have already taken steps to ensure that banks adequately take capacity to repay into account.
17. There are some measures available to the Hong Kong SAR authorities to dampen procyclicality. One option would be to expand the capacity to securitize and insure property-related risks. Supervisors could further encourage the securitization of property loans as a way to increase diversification and to transfer the risk to other institutions better able to bear it. This would also help to develop bond markets in Hong Kong SAR. At present, banks may prefer to maintain their exposure to the property sector given the weak demand by other sectors. Banks are holding a large amount of liquidity. The loans to deposits ratio (denominated in Hong Kong Dollars) is currently about 85 percent, which is a historical low and significantly below its pre-crisis level (Figure III.5.). This liquidity can dampen procyclicality. However, if credit demand by the nonproperty sector picks up and banks reduce their liquidity, there will be more scope for the financial system to exacerbate the procyclicality of credit.
Ratio of Hong Kong Dollar Loans to Deposits
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.Ratio of Hong Kong Dollar Loans to Deposits
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.Ratio of Hong Kong Dollar Loans to Deposits
Citation: IMF Staff Country Reports 2005, 062; 10.5089/9781451816914.002.A003
Source: CEIC.18. Policy makers could also encourage banks to adjust their capital ratio and provisioning over the cycle to help reduce procyclicality. This would lead banks to hold more capital against risks during cyclical upswings, which would tend to slow loan growth. It would also provide more room for bank capital to decline in downturns, which would dampen procyclicality. Banks may make provisions based on short-term measures of risk for longer-term credits. This could cause them to raise provisions during downturns and lower them during expansions.7 Options for provisioning guidelines include:
Bank provisioning based on long-term average loss rates, effectively extending ex-ante general provisioning to take into account risks arising from procyclicality. Such a system, similar to the “statistical provisioning” introduced in Spain in 2000, would tend to slow lending growth during booms since provisions will reflect long-term average risks, not the lower short-term risks. It would also attenuate the decline in lending during downturns, since smaller increases in provisions would be needed when risks materialize. However, such an approach faces a number of technical challenges, such as ensuring conformity with International Accounting Standard (IAS) 39.
Requiring higher provisions for loans in sectors where the banking system as a whole concentrates lending (i.e., the property sector). This would encourage diversification away from those sectors.
19. At present, the procyclicality of credit does not seem to be a significant problem for Hong Kong SAR’s economy. It is possible that much of the observed procyclicality is driven by demand and structural issues. However, if the future of the procyclicality of credit becomes a concern, the authorities would have several options available to respond.
References
Bernanke, Ben, and Mark Gertler, 1999, “Agency Costs, Net Worth, and Business Fluctuations,” American Economic Review, Vol. 79, pp. 14–31.
Borio, Claudio, Craig Furfine, and Phillip Lowe, 2001, “Procyclicality of the Financial System and Financial Stability: Issues and Policy Options,” BIS Papers No. 1, BIS background paper.
Craig, Sean, E. Phillip Davis, and Antonio Garcia, 2004, “Sources of Pro-Cyclicality in East Asian Financial Systems,” Draft Working Paper, IMF.
Gerlach, Stefan, Wensheng Peng, and Chang Shu, 2004, “Macroeconomic Conditions and Banking Performance in Hong Kong: A Panel Data Study,” Hong Kong Monetary Authority Research Memorandum.
Prepared by Marcos Chamon, ext. 35867 (APD) and Sean Craig, ext. 38714 (MFD).
This effect is stronger if banks tend to rely largely on collateral when allocating credit because their credit risk assessment capacity is weak.
Correlations indicate comovement but do not provide information on the amplitude of the cycle. In addition, correlations may change over time and they tend be higher during periods of high volatility. Also, structural factors such as financial liberalization may lead to a one-time increase in the level of credit that will affect its correlation with other variables, even though that one-time change is not due to cyclical factors. Nevertheless, correlation coefficients remain a useful first approximation to the extent of procyclicality.
The correlation between quarterly year-on-year real changes in property prices and real GDP growth over the last ten years is over 0.6.
The GDP deflator is used to convert current prices to constant prices. That index is more appropriate than the CPI for our purposes because of its broader coverage.
We focus on loans for use in Hong Kong SAR for two reasons. First, their cyclical properties are more relevant to Hong Kong SAR’s economy than the ones of external loans. Second, much of the changes in external loans are driven by offshore booking practices of foreign banks, which do not affect credit supply or credit demand in Hong Kong SAR.
Gerlach et al. (2004) show that provisions track non-performing loans very closely in Hong Kong SAR. Craig et al. (2004) show that provisions expand when credit contracts and vice-versa.
