Date of Award

Winter 1986

Project Type

Dissertation

Program or Major

Economics

Degree Name

Doctor of Philosophy

Abstract

Vector Autoregression (VAR) is an alternative to structural econometric model building, specifically in forecasting. The challenge of the VAR technique has been, however, limited to models of national and regional economies.

This dissertation extends the scope of the VAR technique with the construction of a multi-country Link Vector Autoregressive (LIN-VAR) model based on eighteen OECD economies. It offers a unique way of combining the linkage specifications of global structural econometric models with the VAR technique. Each national economy in the LINK-VAR model is specified by four macroeconomic variables: output, nominal money supply, prices, and a short-term interest rate. Linkages among countries are formulated through link variables created from IMF's multilateral trade weighting scheme. The dissertation empirically investigates the extent to which an international VAR model can improve forecasting over VAR models which are limited to domestic influences.

Two alternative linkage mechanisms were developed. The first mechanism links the four domestic variables of the typical country model to the other countries with a different link variable: an output link, a money supply link, a price link, and an effective exchange rate link. The second linkage mechanism includes both a foreign demand variable using the output link and an economic/financial price variable using the effective exchange rate link in each equation of the domestic model. Using monthly data for the period July 1972-June 1984, the two versions of the LINK-VAR model and individual country closed economy VAR models are estimated to generate twelve period ahead ex-post forecasts for statistical evaluation.

The results from both linkage experiments emphatically suggest that forecasts of domestic variables improve substantially when a linkage mechanism is added to domestic VAR models. This argument is more forceful when the forecast evaluation focuses on country-by-country comparisons.

Share

COinS