A0520
Title: Factor-augmented nonstationary panels with multiple structural changes
Authors: Qu Feng - Nanyang Technological University, Singapore (Singapore) [presenting]
Abstract: Nonstationary panels are widely used in empirical studies in economics and other related fields. Multiple structural changes are considered in nonstationary heterogeneous panels with common factors. Breaks could occur in slopes and error factor loadings. Unobserved error factors are treated as additional regressors. Thus, different breakpoints in slopes and error factor loadings are treated as multiple breaks in linear regression models. As previously, unobserved error factors can be proxied by cross-sectional averages of observable data. We show that the breakpoints in both slopes and error factor loadings can be consistently estimated by least squares in both cases of i) nonstationary factors and ii) nonstationary regressors. Monte Carlo simulations are conducted to verify the main results in finite samples.
