Determinants of World Business Cycles: A Dynamic Factor Model Approach

Description

Casual observations suggest that the extent of globalization and economic interdependence among nations increase over time. Recent researches also suggest that changes in the world economic conditions would affect even the price movements of housing in some Asian countries. Thus, for academic research as well as policy analysis, it is important to understand the dynamics of the world economic condition even if the research orientation is completely local. Yet the "world economic condition" is not directly observed. Too many economic variables, both locally and globally, are potentially important and it is not practical to include all of them in the statistical model. Some of the existing literatures often need to rely on strong assumptions to construct proxies.Recent years witness a fast-growing literature which develops the theoretical foundation, as well as employs dynamic factor models (henceforth DFM) in the empirical research of economics. A merit of DFM is the capacity to summarize statistically useful information from a large number of time series data in a few factors, and hence facilitate predictions and policy analysis. Among others, James Stock (Harvard) and Mark Watson (Princeton) have demonstrated in one of their 2002 papers that 6 unobserved factors can account for much variation of 215 time series of the United States. It means that many macroeconomic variables are inter-related and the statistical method identifies only 6 significant and independent sources of shocks in the United States. Ayhan Kose (IMF), Christopher Otrok (Missouri) and Charles Whiteman (Iowa) apply DFM in the study of the world business cycle co-movements. In particular, they decompose the country macroeconomic variables into a world common component, a regional-specific component, a country-specific component, and an idiosyncratic error term.This research builds on their work and applies DFM to study the business cycle comovements among a panel of countries. In particular, we will consider the interactions among different groups of countries, including the G7, selected Asian countries and emerging markets. We will apply a dynamic factor model with a multi-level factor structure, which allows different factors to be correlated over time. For instance, the world factor may be influenced by previous period country-specific factors, and the country-specific factor affected by previous world common factor. Moreover, the previous period country-specific factors may impact other country-specific factors. Our framework will be able to identify all these dynamic interactions and would contribute to a deeper understanding of the world business cycles.