Credit markets and the propagation of Korea's 1997 financial crisis.

1. Introduction

In 1997, the East Asian economies faced a sudden financial crisis. Three countries, Indonesia, Korea, and Thailand, received emergency loans from the IMF. The rapid decline in economic activity that occurred in the wake of speculative attacks at that time has led to no shortage of explanations for the crisis among policy makers. According to the IMF, for example, fundamental weaknesses in the financial intermediaries of the crisis economies were largely at fault. Radelet and Sachs (1998), on the other hand, contend that the crisis was a self-fulfilling one with roots in the inherent instability of international financial markets. In either case, what does seem clear is that the crisis involved changes in credit market conditions that reflected more than purely monetary phenomena. In this paper, we seek to further our understanding of the crisis in South Korea by quantifying the importance of such nonmonetary factors in precipitating a credit "crunch" that acted most emphatically through small and medium-sized enterprises.

Our methodology includes structural regression models that use versions of the Lucas (1972) supply equation modified to include indicators of credit market conditions. We then extend the analysis to consider the longer-term effects of money, credit conditions, and real exchange rates on industrial production with a series of vector autoregressive (VAR) systems. The yield spread between corporate and government bonds, and the ratio of dishonored commercial bills to the total value of bills to be cleared, serve as indicators of the state of the credit market. We then assess the relative abilities of these variables to explain the length and depth of Korea's financial crisis across heavy and light industry.

We find that increases in yield spreads and the dishonored bills ratio, whether driven by increases in business risk or lowered expectations among investors about the future of the Korean economy, had effects that extended well beyond shifts in simple precautionary and speculative demands for the won. We also find that these effects were strongest for light industry, in which small and medium-sized firms accounted for more than 70% of value added in 1999, and that the dishonored bills ratio explains more of the economic decline than the yield spread in our econometric models. We interpret these findings as consistent with the operation of a mechanism much like a "credit channel" and associated "flight to quality" through which increases in risk and the cost of credit intermediation cause the brunt of a credit crunch to fall disproportionately on smaller firms. Though resembling a credit channel in its effects, however, our story differs from the standard treatment because events did not begin with routine policy intervention by the Bank of Korea that operated exclusively through bank credit.

2. Background

At the end of 1997, Korea experienced its first financial crisis, which resulted in the nation's worst economic performance in 40 years. Real GDP growth was -6.7% in 1998, which stands in sharp contrast to the 7.7% growth achieved during the "miracle" years of 1960 to 1996. (1) Moreover, in 1998 the unemployment rate reached 6.8%, up from only 2.6% the year before. The previous postwar high of 7.1% had been reached more than 30 years earlier in 1964. It is clear that the financial crisis coincided with a period of severe economic stress.

Whether the crisis was anticipated or not is crucial to investigating its causes. Krugman (1998) argues that it was indeed anticipated, and that foreign investors expected to be bailed out. If this were the case, however, foreign creditors would have disinvested upon observing negative signals about the future of the economy, and this does not appear to have been the case. Further, there is little empirical evidence that the crisis was anticipated. Table 1 summarizes economic conditions in 1997. Foreign direct investment (FDI) to Korea was on the rise. Over the first nine months of 1997, there was 50% more FDI than there had been over the corresponding period in 1996. And even though the current account balance showed a large deficit due to worsening terms of trade, net capital inflows had up to then been able to cover most of the deficit. Overall, there was a shortfall of only US$0.5 billion until October. The dishonored bills ratio was relatively low until the crisis, but then rose nearly sevenfold over the next three months.

The crisis seemed to stem from the failure of small investment banks. When the capital market opened in the 1990s, the Korean government granted new business permits to more than 20 such banks--permits that included the right to borrow abroad. By 1997, some of these new banks had already experienced liquidity problems due to maturity mismatches and bad investments, and in October the ratio of loans in default exceeded 5% for five of these banks. As they tried to obtain more U.S. dollars to repay short-term debts, changing expectations about the path of exchange rates increased the demand for U.S. dollars at the worst possible time. Kaminsky and Reinhart (1999) describe the type of downward spiral that ensued. The government intervened to protect the won, yet the currency continued to weaken as foreign reserves at the Bank of Korea came under increasing pressure. Finally, with its reserves drained, the Korean government requested an emergency loan from the IMF on October 21.

Even after the emergency loan was approved on December 3, however, the won continued to fall and did not stabilize until January 28, 1998, when foreign creditors agreed to roll over the short-term debt of many Korean financial intermediaries. Interestingly, the solution came from market interactions between debtors and creditors rather than intervention by the Korean government or the emergency loan. The difficulties involved in arriving at a monetary solution motivate us to consider nonmonetary factors, such as those driven by a severe credit crunch, in explaining the length and depth of the economic downturn.

3. Credit Market Conditions and Real Activity

Monetary theory often focuses on two main channels through which a shift in the supply of money can affect real economic activity. The first is the traditional money view, in which changes in the money supply affect the demand for household consumption and business investment through the interest rate. The second is the lending or "credit" channel, in which a decrease in the money supply causes a reduction in the supply of loans to the private sector. For the latter to operate, it is necessary that: (i) banks adjust their loan supply in response to monetary shocks, (ii) loans are not perfect substitutes on a bank's balance sheet, and (iii) there is imperfect substitutability for firms between bank loans or other commercial bills and bond issues (Bernanke and Blinder 1988). When these conditions are met, interest rates on loans increase relative to those on other securities, and the widening of the spread between risky and safer financial assets reflects the extent to which monetary policy has become more restrictive. An active credit channel can also generate a flight to quality in which banks choose to lend to larger and safer borrowers at times of monetary stringency (Bernanke, Gertler, and Gilchrist 1996). This means that some firms, particularly smaller ones, end up relying more on existing collateral and internal funds to support their operations. Our contribution extends the idea of a lending channel to analyze a financial crisis in which the real sector is not so much affected by a declining supply of money as by increases in the costs of providing financial intermediation services, and how this might generate a flight to quality in the open market for credit.

Recent studies have taken varied approaches to examining the credit crunch in Korea, with most focusing on the bank-lending channel. Using commercial banks' data, for example, Ferri and Kang (1998) found that less well-capitalized banks tended to increase their lending rate and reduce loans during the crisis. Consistent with this, Kim (1998) found a large excess demand for bank loans after the crisis. In contrast, Ghosh and Ghosh (1999) found little evidence of an excess demand for credit. Borensztein and Lee (2002) analyze the credit crunch using firm-level data and conclude that firms affiliated with Korea's largest informal business networks, the "chaebols," appear to have lost some of the preferential access to credit that they enjoyed in the precrisis period, suggesting that the credit crunch affected a wide range of firms. Ding, Domac, and Ferri (1998) link the credit crunch to sharp increases in yield spreads between risky and risk-free assets. Investigating five different countries, Indonesia, Korea, Malaysia, the Philippines, and Thailand, they observe that these effects were significant only for Korea and Malaysia.