World prices and business cycles of a small open input-output economy

In a recent paper, Schmitt-Grohe and Uribe (2018) question the major importance assigned to terms-of-trade shocks in the explanation of business cycles of developing countries, by conventional wisdom. They estimate a country-specific structural vector autoregression (SVAR) model on 38 countries, and show that the share of variance of macroeconomic indicators explained by terms of trade shocks, lies on average around 10%, rather than 30% obtained with calibrated business cycle models in seminal papers (Mendoza, 1995; Kose, 2002). They also perform country by country comparisons of the empirical results and theoretical predictions of a calibrated three-sector version of the small open economy (SOE) model introduced by Mendoza (1995). Following a rigorous methodology, they find that, once variables are measured in same units as in the data, the theoretical model confirms on average the empirical share of variance of 10%. However, at the country level, results appear to reject any relationship between theoretical and empirical predictions, although models share the same exogenous process of terms of trade shocks and are estimated on same data. They conclude, therefore, that the resolution of this disconnect is likely to imply improvements in both models.

The goal of this paper is to focus on the theoretical side of their study, and to address the problem of this disconnect through an alternative SOE model where all sectors are inter-connected as described by input-output tables from the National Accounts, and where all goods can be either consumed or jointly employed as fixed and/or intermediate capital. We first develop this model in a dynamic general equilibrium framework and then investigate its implications in the analysis performed by Schmitt-Grohe and Uribe (2018). Through a slightly different calibration strategy and a similar estimation of structural parameters country by country, we find that the average effect of terms of trade shocks on variables of around 10% is confirmed, and that the disconnect between the importance of terms of trade shocks assigned by the theoretical and SVAR model is reduced at the country level. It appears indeed that beyond different debt premium elasticities, capital adjustment costs and processes of terms of trade shocks, the different input-output economic structure and sector-specific labor elasticities across countries, contribute to the explanation of the various impacts of terms of trade shocks. The data sources include the Organization for Economic Cooperation and Development (OECD), World Input-Output (WIO) Database, and the World Bank World Development Indicators (WDI) Database.