Nonlinearity tests of the Mexican stock market returns: Lyapunov coefficients

Abstract

We examine the non-linearity of the Mexican Stock Market daily returns. We find empirical evidence to reject lineal specifications in the behavior of the stock returns. As a consequence, most of the findings based on lineal methods regarding the stock market in Mexico may be questioned. We also test a random walk specification versus an alternative hypothesis of chaos in the Mexican stock market index, IPC. To achieve this, we design a statistic based on Lyapunov dominant exponent by using local polynomial regression methods. The empirical distribution of the statistic is obtained through the surrogate data method. Finally, the test concludes that the hypothesis of random walk cannot be rejected.