한국주식시장에 합리적 거품이 존재하는가-이론과 실증분석-

Testing for Rational Bubbles in the Korean Stock Market : Theory and Empirical Evidence

As Korean stock prices have recently shown wide gyrations, there has been heated debate surrounding the presence of bubbles in Korean stock prices. Unfortunately the literature on bubbles in the Korean stock market is few and far between. The purpose of this study is to establish theoretical foundations for rational bubbles and to provide empirical evidence on the presence of rational bubbles in the Korean stock market. There has been also a resurgence of interest in bubbles in stock prices primarily through the work of Hamilton and Whiteman (1985), Diba and Grossman (1988a, b, c), Evans (1991), Froot and Obstfeld (1991), and others. The bubble models elaborated by these authors represent a significant departure from the conventional paradigm in that they reinterpret rational bubbles in terms of market fundamentals. These bubbles may be termed intrinsic bubbles, as opposed to extraneous bubbles in the traditional view. It has been widely believed that bubbles do not lend easily themselves to direct testing. The attractive feature of the intrinsic bubble specification may be found in its ability to derive testable implications for bubbles by investigating the stationarity properties of stock prices and dividends or by parameterizing a specific bubble relationship as a function of market fundamentals. However, the existing approach to intrinsic bubbles still remains unsatisfactory. As Evans criticizes, Diba and Grossman's stationarity tests are unable to detect an important class of rational bubbles. The ability of Froot and Obstfeld's parametric tests to discover bubbles is also doubtful. The essence of their tests is that if the price-dividend ratio is significantly (and nonlinearly) related to current dividends, the hypothesis of no bubbles is rejected. More fundamentally, the recent tests for intrinsic bubbles may be characterized as being indirect in that explicit measures of bubbles are not directly related to market fundamentals. In this study we formulate an information error model which allows one to derive bubble measures in a straightforward manner. This study provides a new method of testing for bubbles by specifying bubble measures as the Weibull distribution. This study is the first attempt to apply the Weibull distrubution to the test of rational bubbles. There is not only a parallel between the burst of a speculative bubble and a material's burning out, but also there is a good reason to believe that measured bubbles can be appropriately modeled as the Weibull specification. A bubble is a rare event. Like other rare events, bubbles can be approached in terms of the instantaneous rate at which an event occurs after duration t since some prior event has occurred. There were three episodes when bubbles were suspected to disrupt Korean stock price movements over the past two decades or so : (l) 1982.1 - 1989.3, (2) 1992.1 - 1994.12, and (3) 1998.9 - 2000 1. Since the size of the third sample is too small, we have extended the observations of the third sample period from January 1992 to February 2000. Our empirical analysis reveals that there is no evidence for extraneous bubbles in the Korean stock market. However, there indeed exist intrinsic bubbles during the 1982.1-1989.3 period, but no evidence for the presence of intrinsic bubbles is found during the 1992.1-1994.12 and 1992.1-2000.2 periods. These findings appear to be in accordance with previous studies on rational bubbles in the Korean stock market. Thus, we can conclude that a considerable portion of stock price hikes observed in the latter part of the 1980s was attributable to the bubble component. We have also conducted two different cointegration tests: One is concerned with the Granger (1987) theorem, and the other is concerned with the Diba-Grossman (1988a, b, c) and Yuhn (1996, 1997) theorems Granger theorem states that if the stock market is efficient, the prices of two assets cannot be cointegrated (=linear cointegration). On the othe