Credit rationing caused by capital market imperfections is widely seen as an important phenomenon in the loan market, especially for small and medium enterprises (SMEs). Among various ways of alleviating the problem, credit guarantee schemes are one of the most important policy tools in many countries. An economic rationale for such public intervention is that it can enhance efficiency by providing additional funds for SMEs that are in fact healthy but unable to secure enough loans because of the informational gap between lenders and borrowers.
Despite many empirical studies (e.g. Riding and Haines Jr 2001, Riding et al. 2007, Cowling 2010, Uesugi et al. 2010) that evaluate the benefit of credit guarantee schemes, empirical analyses on the cost side of the policy are scarce. Two important costs of providing credit guarantees are adverse selection and moral hazard. Since credit guarantees insure banks against incurring losses from default, they are enticed to ask seemingly risky borrowers to apply for credit guarantees. Also, because credit guarantee corporations cannot distinguish low-risk borrowers from risky ones, credit guarantee schemes attract a sizable portion of risky borrowers, which results in inefficient resource allocation. This potential problem could be especially grim in Japan, where the proportion of 100% credit guarantees accounts for more than half of the total loans with credit guarantees.
Problems in Japanese credit guarantee schemes
The credit guarantee schemes in Japan have some seemingly problematic features in the light of asymmetric information. First, public credit guarantee corporations (CGCs) mainly provide full credit guarantees. Until October 2007, the coverage rate in Japan was 100%, and 80% thereafter. However, for emergency guarantee programmes for the period October 2008 to March 2011, it returned to 100%. As a result, a large portion of guaranteed loans covers 100% of defaulted loans.
Second, the Japan Finance Corporation (JFC), which is a public corporation wholly owned by the Japanese government, offers credit insurance that covers losses from subrogation. Coverage rates of credit insurance range from 70% to 80%, and hence the CGCs themselves suffer little loss from subrogation. The JFC accepts all credit insurance for CGCs, implying that CGCs have only a weak incentive to monitor banks and small businesses. Notably, the JFC has suffered substantial losses in credit insurance accounts every year – for example, 568 billion yen in fiscal year 2009 and 436 billion yen in fiscal year 2010.
Third, the rejection rate of credit guarantees is low, at approximately 10%, implying that most credit guarantee applicants are accepted. As we have mentioned, this might be due to the fact that CGCs have weak incentives to screen small business applicants because of the high coverage of credit insurance.
Finally, CGCs cannot collect sufficient soft information from applicants. Many studies (e.g. Berger et al. 2005) claim that soft information plays an important role in assessing the credit risk of small businesses. Unlike banks, which can acquire soft information through relationship lending from continuous transactions, CGCs cannot, or do not, gather enough soft information, and therefore have to rely only on hard information on small businesses. Moreover, banks that are unable to assess the risks of small businesses have strong incentives to offer them guaranteed loans.
Testing for adverse selection and moral hazard
Using data on city, regional, and shinkin banks, in Saito and Tsuruta (2014) we apply the basic positive correlation test proposed by Chiappori and Salanie (2000) to investigate whether: (i) banks that transact with risky small businesses are more likely to offer loans with guarantees (adverse selection); and (ii) small businesses with guaranteed loans are more likely to default (moral hazard).
In both cases, a positive correlation is observed between the rate of loans with guarantees and ex-post default risk. As this correlation should be assessed within the group of observationally equivalent firms, we control for the effects of observable variables by using a seemingly unrelated regression (SUR).
Our findings are consistent with the adverse selection and/or moral hazard hypotheses. The estimation results of the SUR model suggest that the null hypothesis of no correlation between the rate of loans with guarantees and ex-post default risk is rejected at the 1% or 5% level for all banks, city and regional banks, and shinkin banks. These suggest that our data are consistent with the adverse selection and/or moral hazard hypotheses. Further analyses provide us some additional knowledge about the role of self-payment in the public credit guarantee schemes. For the 100% guarantee, we observe a positive and statistically significant correlation between the rate of loans with guarantees and ex-post default risk regardless of the bank type. However, for the 80% guarantee, the results are slightly different. Correlation between the rate of loans with guarantees and ex-post default risk are positive, but not statistically significant.
To investigate the non-linear relationships between the rate of loans with guarantees and ex-post default risk, we also estimate a partial linear model. Figures 1 to 3 show the results from the partial linear model. The plotted data are dispersed widely, but Figures 1 and 2 indicate a positive correlation between the rate of loans with guarantees and ex-post default risk for all samples and 100% guarantees. Figure 3 indicates that the slope for 80% guarantees is flatter than in the case of 100% guarantees, indicating that the information problem is less severe because of the 20% self-payment.
Figure 1. Partial linear model using all samples
Note: This figure provides the estimates of a partial linear regression model with the amount of loans with guarantees/amount of loans for small businesses (y) and the amount of subrogation/amount of guarantees (z).
Figure 2. Partial linear model using 100% guarantees
Note: This figure provides the estimates of a partial linear regression model with the amount of loans with 100% guarantees/amount of loans for small businesses (y) and the amount of subrogation/amount of guarantees (z).
Figure 3. Partial linear model using 80% guarantees
Note: This figure provides the estimates of a partial linear regression model with the amount of loans with 80% guarantees/amount of loans for small businesses (y) and the amount of subrogation/amount of guarantees as the dependent variables (z).
We find statistically significant positive correlations between credit risk (subrogation rate) and the amount of guaranteed loans, indicating that a public credit guarantee programme is influenced by asymmetric information. Further investigation suggests that the association between the subrogation rate and the ratio of guaranteed loans to total loans is stronger for 100% credit guarantees than for 80% credit guarantees, implying that the ‘20% self-payment’ criteria is working as an effective mechanism for alleviating the problem, but is not enough for eliminating it.
Some economists argue that, without a rigorous empirical study, it is obvious that the Japanese credit guarantee scheme is severely affected by adverse selection and moral hazard. At the same time, however, bank officers claim that banks do not offer loans without sufficient screening and monitoring even if the loans are credit-guaranteed. Given these differing opinions, we believe that empirical analyses are essential to assess whether adverse selection and/or moral hazard are detected in the Japanese credit guarantee scheme. We also believe that our study contributes to the recent policy debate on whether CGCs should lower the rate of self-payment to under 80%.1
Editor’s note: The main research upon which this column is based (Saito and Tsuruta 2014) first appeared as a Discussion Paper of the Research Institute of Economy, Trade and Industry (RIETI) of Japan.
Berger, A N, N H Miller, M A Petersen, R G Rajan, and J C Stein (2005), “Does function follow organizational form? Evidence from the lending practices of large and small banks”, Journal of Financial Economics 76(2): 237–269.
Chiappori, P-A and B Salanie (2000), “Testing for Asymmetric Information in Insurance Markets”, Journal of Political Economy 108(1): 56–78.
Cowling, M (2010), “The role of loan guarantee schemes in alleviating credit rationing in the UK”, Journal of Financial Stability 6(1): 36–44.
Riding, A L and Haines Jr, G (2001), “Loan Guarantees: Costs of Default and Benefits to Small Firms”, Journal of Business Venturing 16(6): 595–612.
Riding, A, J Madill, and G Haines (2007), “Incrementality of SME Loan Guarantees”, Small Business Economics 29(1): 47–61.
Saito, K and D Tsuruta (2014), “Information Asymmetry in SME Credit Guarantee Schemes: Evidence from Japan”, RIETI Discussion Paper 14-E-042.
Uesugi, I, K Sakai, and G Yamashiro (2010), “The Effectiveness of Public Credit Guarantees in the Japanese Loan Market”, Journal of the Japanese and International Economies 24(4): 457–480.
1 See Nihon Keizai Shimbun (Nikkei), p.5 in the morning issue of 2 June 2014.