|About the Book|
The first two chapters of this dissertation are devoted to the estimation of the number of factors in linear factor models. Based on the generalized method of moments estimation, two different methodologies to estimate the number of factors areMoreThe first two chapters of this dissertation are devoted to the estimation of the number of factors in linear factor models. Based on the generalized method of moments estimation, two different methodologies to estimate the number of factors are proposed. The methods are appropriate for panels with a large (small) number of cross-section observations and a small (large) number of time-series observations. They are robust to heteroskedasticity and time series autocorrelation of the idiosyncratic components. All necessary procedures are similar to the instrumental variables estimation, so they are computationally easy to use. In addition, the methods can be used to determine which observable variables are correlated with the latent factors without estimating them. Monte Carlo experiments show that the proposed estimators have good finite-sample properties. As an illustrative application, the methodology is used to analyze the international stock markets. Results imply that international stock returns are explained by one strong global factor. This factor seems to be highly correlated with the US stock market factors. This result can be interpreted as evidence for market integration. Also there is evidence of two weak factors related mostly with the European and the Americas markets. The third and final chapter of the dissertation is a careful factor analysis of the term structure of credit spreads. The estimation shows that credit spread innovations are subject to three common factors: two strong factors and one weak factor. A novelty is that the factors are extracted using canonical relations between credit spreads and a set of observable or estimated variables. This approach appears to estimate the factors in credit spreads better than the conventional principal component approach. The first strong factor is related to the contemporaneous state of the economy. The second strong factor represents investors expectations about future economic conditions, and is shown to have predictive power for the state of the economy over a two-quarter horizon. The weak factor is mainly related to the error-correction processes in short-term spreads. Although the weak factor is not a major determinant of credit spreads, it is needed to obtain a factor model representation of the data.