Forecasting economic and financial variables with global VARs

Abstract

This paper considers the problem of forecasting real and financial macroeconomic variables across a large number of countries in the global economy. To this end a global vector autoregressive (GVAR) model previously estimated over the 1979Q1-2003Q4 period by Dees, de Mauro, Pesaran, and Smith (2007), is used to generate out-of-sample one quarter and four quarters ahead forecasts of real output, inflation, real equity prices, exchange rates and interest rates over the period 2004Q1-2005Q4. Forecasts are obtained for 134 variables from 26 regions made up of 33 countries covering about 90% of world output. The forecasts are compared to typical benchmarks: univariate autoregressive and random walk models. Building on the forecast combination literature, the effects of model and estimation uncertainty on forecast outcomes are examined by pooling forecasts obtained from different GVAR models estimated over alternative sample periods. Given the size of the modeling problem, and the heterogeneity of economies considered -industrialised, emerging, and less developed countries -as well as the very real likelihood of possibly multiple structural breaks, averaging forecasts across both models and windows makes a significant difference. Indeed the double-averaged GVAR forecasts performed better than the benchmark competitors, especially for output, inflation and real equity prices.

Do we need a global VAR model to forecast inflation and output in South Africa?

“…average forecasts across different GVAR specifications and different estimation windows, generally perform better than benchmark forecasts from univariate autoregressive and random walk models as well as forecasts from individual GVAR models. Smith (2013) summarises the conclusions of Pesaran et al (2009a) and reinforces the findings by adding out-of-sample data for another four quarters before re-evaluating the models.…”

“…In the first GVAR forecasting application, Pesaran, Schuermann and Smith (2009a;2009b) forecast macroeconomic and financial variables for all 26 regions in the standard 33-country GVAR of DDPS (2007) and DHPS (2007). The paper considers short-term (one quarter in advance) and medium-term (four quarters in advance) out-of-sample forecasts.…”

“…Recursive out-of sample forecasts are then determined from 2005Q1 to 2009Q4, each time for a forecast horizon ( h ) up to eight quarters. These simple models are often used in the literature as benchmark models as their forecasts are "surprisingly hard to beat" (Pesaran et al, 2009a;Smith, 2013). The use of an AR(1) model specifically for South Africa has also been stressed recently by Gupta, Kanda, Modise and Paccagnini (forthcoming).…”

Global prediction of recessions

“…7 We use Diebold-Mariano tests to check the significance of differences in forecast performance (Diebold and Mariano, 1995). To test for performance differences with respect to groups of countries, we apply the panel version of the test proposed in Pesaran et al (2009). Regarding individual countries, the GVAR model outperforms both benchmarks for a little more than half of the major economies-especially at short forecast horizons.…”

“…Though GVAR models are linear, they offer a fair degree of flexibility in modeling business-cycle dynamics of the world economy. Recent studies show that they have merits of providing good forecasts for a range of macroeconomic variables (Pesaran et al, 2009;Crespo Cuaresma et al, 2014;Dovern et al, 2015). Yet, if they also provide accuracy gains (relative to country-specific models) in terms of forecasting the business-cycle regime remains an open question.…”