Can Japan avert any future banking crisis?

Abstract

This paper suggests that the weaknesses of corporate governance are enough to explain the Japanese banking crisis in the 1990s. Bank size and lack of operating and management efficiency contributes to low return, thereby lead to the failure of the banks. Emergence and burst of the bubble in the late 1980s just accelerated the situation to an early crisis in the 1990s.

Acknowledgements

I would like to thank Professors Y. Okita, Kenichi Ohno and K. P. Kalirajan for their cooperation and helpful comments on the paper.

Notes

Of the failed ordinary banks, one city bank, one regional bank, two long-term credit banks and the rest are regional tier II banks.

Post-war Japanese financial system was highly regulated and banks were heavily dependent on Bank of Japan's (BOJ) subsidies (window guidance) and borrowings of ‘enterprise groups'. The system is well known as ‘main bank' system. Group affiliation interlocks stock shares among industrial enterprises, banks and other financial institutions. The shares of group member firms owned by banks form an important link in the interlocking structure of enterprise groups. This structure of Japanese banks might be compared to the so-called ‘Industrial bank' (also available in Germany as House bank) rather than modern commercial bank. It is evident from many research works that this ‘main bank' system in Japan contributed greatly to the post-war economic growth although the varieties of functions played by the main bank were not usually associated with the concept of commercial banking which has been termed as ‘convoy system' by some economists. The financial deregulations started in the late 1970s to make the system competitive.

The PL method, a non-parametric approach gives maximum likelihood estimates the survival functions by considering the fact that the probability of surviving any particular number of years from the beginning of the study is the product of the same estimate up to the preceding year. The median survival time can be obtained by linear interpolation.

The PH model has a partial likelihood function in which only the parameters are the coefficients associated with covariates. Therefore it is mostly free from omitted-variable bias. Statistical inference based on the partial likelihood has asymptotic properties.

The underlying assumptions of the hazards model are to assign the characteristics of the subjects (e.g. banks) measured at start-point or end-point or any other specified time-point to the failure (or survival) time of the subjects in a prospective/retrospective study. Nonetheless, there is no clear-cut rule about the time-specification of the covariates, but the interpretations must depend on the specification (Kiefer, 1988). Kiefer also mentioned that assigning economic meanings to coefficients is a matter of modelling and judicious use of prior information.

Considering start point (i.e. 1990), only interest income over interest expense has been found significant to the future failure of the Japanese banks. The other variables have not been found significant because of the fact that only one year's data is not enough to capture the long term time series trend.

Log of total asset has been used as a proxy of bank size.