An evaluation of equilibrium business cycle models in the presence of statistical nonlinearities.

摘要

The dissertation consists of three essays. The first essay finds significant nonlinearities in several cyclical components macroeconomic time series across countries. Standard equilibrium models of business cycles successfully explain most first and second moments of these time series. Nevertheless, this essay shows that a model of this class cannot replicate nonlinear features of the data. Applying the Efficient Method of Moments (EMM) methodology to build an algorithm that searches over the models parameter space establishes the parameterization that best allows replication of all statistical properties of the data. The results show that this parameterization captures nonlinearities in investment but fails to account for observed properties of consumption.; The second essay uses the EMM methodology to study the impact of borrowing constraints on small open economies. The evidence of the last 20 years of recurring output busts and rapid reversals of the current account in emerging markets indicates that domestic agents may not be able to borrow in international capital markets to fully insure themselves against internal and external shocks. This essay models this phenomenon as a form of excess volatility by introducing a financial friction into a stochastic model of a small open economy. The financial friction limits the current account deficit to a fixed fraction of gross domestic product. The essay shows that conditional volatility and asymmetry are significant statistical characteristics of the GDP and current account that reflect the excess volatility and the current account reversals. The economic model can explain the conditional volatility and asymmetry of Mexican GDP and the current account.; The third essay documents the dynamic properties of national output, its components, and the current account for five OECD countries, using a seminonparametric (SNP) estimator. There is strong evidence of conditional volatility for almost all time series as well as significant nonlinearity, detected particularly in GDP, net exports, investment time series.