신용보증지원이 중소기업의 수익성과 성장성에 미치는 효과에 대한 패널분석

Small and medium-sized enterprises(henceforth ‘SMEs') play a very great role in a country's economy. They account for about 90% of firms in a nation and create a great number of jobs, which has a crucial economic meaning. Despite this importance, SMEs are encountering lots of difficulties in every phase of business activities such as, marketing, financing, personnel, production, etc. Among them, the financing problem may be the most important and severe one, which is mainly due to the high credit risk of SMEs in an environment of informational asymmetry. Moreover, SMEs generally lack the ability to provide sufficient physical assets for financial intermediaries as collateral. Thus, they are usually supported by the government with a specialized credit guarantee program when financing their projects. In Korea, KODIT, KIBO, and KOREG are institutions providing such credit guarantee. This paper empirically examines the effects of government-sponsored credit guarantee program in Korea, using the annual data of SMEs for the period of 2000 through 2007. Estimating the effectiveness of the credit guarantee program is of vast importance because the program is financed by public funds. If it turns out to be ineffective unlike the initial expectation of the program, meaning that the government is wasting public funds, the program needs to be reformed to achieve the real effectiveness or abolished to be replaced by more effective financing measures for the SMEs. Studies on the treatment effect of credit guarantee in Korea have been mainly dependent on the input-output analysis and cost-benefit analysis. Such macroeconomic approaches, however, cannot provide accurate estimates for the effects of the policy. Recently, a few researchers have begun to pay attention to the microeconomic approach such as, cross-sectional regression analysis, propensity score matching(PSM), etc. Analyzing the treatment effect with cross-sectional data, however, cannot deal with the heterogeneity residing in each group and is subject to the endogeneity problem resulted from it. In order to avoid those problems and to more accurately estimate the effect, we employ a fixed-effects panel regression model in our analysis. In our specification, particularly, is the guarantee variable included with a one-year lagged form, which is believed to relieve above mentioned problems to a great extent. We construct a huge data set with the help of the Korea Credit Guarantee Fund(KODIT) and Korea Enterprise Data(KED). KODIT data comprise of data on 10,411 firms having credit guarantee for 8 consecutive years, data on 9,430 firms with credit guarantee for 1 to 7 years, and data on 230 firms without credit guarantee for the whole sample period. KED data, in contrast, only comprise of data on 6,367 firms without credit guarantee for the whole sample period. In actual analyses, we winsorized variables with apparently incorrect or extraordinarily large or small values. Thus, the number of observations differs between analyses depending on which variables are included in the regression. After all, total observations on average after winsorization amount to more than 150,000, which is far greater than the data in existing studies. We believe that using a panel approach with an enormous data set increased the accuracy of the estimates greatly. We concentrate on the profitability and the growth potential among various performance indexes for a business firm. Particularly, we use ROA, ROS(return on sales) as variables representing the profitability, and GRS(growth rate of sales), GRV(growth rate of value added) as variables representing the growth potential of enterprises. For the main explanatory variable, we use GUR, a dummy variable for the credit guarantee, assigning 1 to firms with amounts of credit guaranteed and 0 to firms without credit guaranteed as of the end of the year. In addition, variables indicating firm-specific characteristics like firm history, firm size, etc. and variables measuring diverse financial structure of a firm are included to control for their impact on the firm performance. Finally, a few of macroeconomic variables such as, the growth rate of real GDP, won-dollar exchange rate, and the growth rate of stock price are included to reflect the overall effects of economic environments. In contrast to the initial object of the credit guarantee program, we find no evidence that it positively affects the profitability and the growth potential of SMEs. In the baseline analyses conducted with whole sample, ROA and ROS turn out to have a significant negative relationship with GUR at the 1% level in all models, indicating the provision of credit guarantee does not improve the profitability of a firm. In sub-period analyses, however, the performance indices turn out to have somewhat different results. The performance indices show higher negative coefficients with GUR in a period of large and increasing amount of guarantee than in a period of small and decreasing amount of guarantee. This may support the existing idea that one important shortcoming of the program is to allow marginal firms to survive in the industry, leading to the weak overall performance of guaranteed firms. Even though the results reveal that the credit guarantee program in Korea is not currently working well, it does not imply that the program should be abolished soon. It is because there are no appropriate alternatives to help the SMEs finance their projects. This requires that the current credit guarantee program be reformed to meet the requirement of government authorities by raising its effectiveness in upgrading the performance of business firms. In order to draw some insights into this issue, we try to analyze the effects of size and duration of credit guarantee in case it is provided. In a sub-sample analysis where only guaranteed SMEs are included, the size of guarantee turn out to negatively affect the performance of SMEs. On the other hand, it is found that the duration of guarantee increases the firms' performance except ROA. Meanwhile, it is possible that the same amount of guarantee may produce different outcomes when it is delivered to different type of firms. If it is correct, differentiating the size of credit guarantee according to the type of firms or focusing on a specific type of a firm should be encouraged. When we analyze the effect of guarantee program by industry type, it turns out that the effect was most prominent in the prospective service firm. Above results lead us to a conclusion that the current system of credit guarantee does not work well, suggesting that more efforts are needed to make the program indeed effective. Regarding this, finding an optimal size of guarantee, selective provision of credit guarantee by industry type should be a good example of those efforts. This study, as far as we know, is the first one to use the panel regression approach in analyzing the treatment effect of credit guarantee program in Korea. Accordingly, the results may have many points to be improved, and we eagerly want as many works as possible to be followed in the future.