Abstract
This paper empirically tests, using the daily closing values of the Dow Jones Industrial Average (DJIA) data from February 5, 1971 through July 1, 2011 whether stock prices in the United States are both nonlinear and nonstationary by applying the unrestricted two-regime Threshold Autoregressive unit root tests (TAR) proposed by Caner and Hansen (2001). Unlike the traditional linear unit root tests, the TAR model simultaneously tests for the presence of nonlinearity and nonstationarity in the data generating processes. The implications of the outcomes of the unit root tests for structural analysis and econometric forecasting can hardly be exaggerated. The findings of the paper indicate that DJIA stock price time-series is nonlinear and nonstationary in two regimes. Stock prices exhibiting randomness suggest that the data generating process is nonstationary and thus makes it extremely difficult to accurately forecast stock price changes. The evidence of the presence of a unit root in stock prices also implies that the weak-form efficient market hypothesis is valid for the U.S. stock market.