Empirical studies on the nonlinear economic models

This dissertation conducted two empirical studies on nonlinear economic models. Chapter 1 used a threshold state space model to analyze the transition from chronic inflation to hyperinflation. Chapter 2 explored whether the shocks of different sizes have an asymmetric effect on the economy. By incorporating a threshold-type non-linearity, we discussed the difference between chronic versus hyperinflation in Chapter 1 and big versus small shocks in Chapter 2.; Using Brazilian data, Chapter 1 examined how the structure of the macroeconomy changed when it evolved from chronic inflation to hyperinflation. Bruno and Fischer (1990) claim that there exist two inflation equilibria in an economy, and the stability of these equilibria depends on two parameters: the semi-elasticity of money demand with respect to the expected inflation rate and the speed of adjustment of the expected inflation rate. We specified the expected inflation rate as a state variable and estimated the threshold state space model, using Cagan's money demand function for the measurement equation and the adaptive expectation formation for the state equation. We found that the stability condition of the high inflation equilibrium held during chronic inflation, while it was violated during hyperinflation. The threshold level of inflation which separated chronic inflation from hyperinflation was estimated to be 14.8% per month and, if using this definition, hyperinflation started after December 1985 until July 1994 in Brazil.; Chapter 2 explored whether different-sized unanticipated monetary shocks have asymmetric effects on output. The standard impulse response function, based on the linear VAR, assumes that the impact of a shock is proportionate to its size. However, Blanchard and Kiyotaki (1987) argue that, due to menu costs, small monetary shocks have a disproportionately large impact on output compared to large monetary shocks. Motivated by Cover (1992), we use FIML to jointly estimate the money supply process and the output process in which big and small unanticipated monetary shocks are incorporated. We found that big shocks had a disproportionately small impact on output, and the estimated coefficients for big shocks were uniformly less significant than the coefficients for small shocks. Our result supports the neutrality of big monetary shocks.