A PANIC ANALYSIS OF STOCK PRICES: NEW EVIDENCE FOR INDUSTRIALIZED COUNTRIES

This paper presents an investigation of the time series properties of stock price indexes for a panel of 18 countries spanning from December 1969 to May 2007. Unlike previous studies in the field, we employ the PANIC procedures of Bai and Ng (2004a, b), which explicitly allow for strong forms of cross-sectional dependence and enable us to determine the source of the non-stationarity in the observed series of stock prices, i.e. whether it stems from the stochastic behavior of the common factor and/or the idiosyncratic components. Overall, there is strong evidence of non-stationarity in stock prices, which appears to be driven by a common stochastic trend. The computation of half-lives of shocks to the idiosyncratic components through impulse-response functions corroborates the findings obtained with PANIC. First, mean-reverting country-specific components suggest the existence of cross-sectional predictability of stock prices. Second, there is no evidence of time-series predictability of stock prices, since the global shock appears non-stationary in levels and exhibits an infinite half-life.