Structural change in cointegrated systems: Theory and applications

My dissertation studies theoretical and empirical aspects of structural change in cointegrated systems. Chapters 1 and 2 provide theoretical analyses of the relevant issues while Chapter 3 uses the tools developed to study the Feldstein-Horioka puzzle. Chapter 4 studies data dependent rules in the context of dynamic least-squares cointegrating regressions.; In Chapter 1, I consider testing for multiple structural changes in cointegrated systems and derive the limiting distribution of the Sup-Wald test under mild conditions on the errors and regressors. It is shown that as long as the intercept is allowed to change across regimes, inference is possible even if stationary variables are allowed. I also propose a sequential testing procedure that permits consistent estimation of the number of breaks. The procedures are based on a modified Wald test that has good size and monotonic power.; In Chapter 2, I derive the consistency, rate of convergence and the limit distribution of the estimated break fractions. I show that if the coefficients of the integrated regressors are allowed to change, the estimates are asymptotically dependent so that confidence intervals need to be constructed jointly. If, however, only the intercept and/or the coefficients of the stationary regressors are allowed to change, the estimates are asymptotically independent.; Chapter 3 argues that the observed strong positive saving-investment correlation arises due to the failure to account for instabilities in the long run relationship between them. I use the instability tests of Chapter 1 and the cointegration test of Arai and Kurozumi (2005) extended to allow for multiple breaks. I find that for most countries the cointegrating relationship has changed, mostly towards a lower saving-investment correlation regime. While the saving-investment link varies closely with country openness, there is hardly any association with country size.; Chapter 4 studies the use of data dependent rules in the context of the selection of the number of leads and lags in the dynamic least-squares cointegrating regression. The assumptions imposed by Saikonnen (1991) preclude the use of information criteria to select the number of leads and lags. I show that his results remain valid under weaker conditions which permit the use of such rules.