RDP 2013-10: Stochastic Terms of Trade Volatility in Small Open Economies 2. Literature Review

Our paper complements other work studying the macroeconomic effects of uncertainty and time-varying volatility. Examples here include Bloom (2009), who explores the short-run fluctuations of output, employment and productivity growth after shocks to macroeconomic uncertainty, Justiniano and Primiceri (2008), who shed light on the sources of changes in US macroeconomic volatility, and Christiano, Motto and Rostagno (2010), who study the effects of idiosyncratic risk shocks on entrepreneurs' productivity. Also closely related to our paper are Fernández-Villaverde et al (2011), who examine shocks to the volatility of sovereign debt interest rates, and Fernández-Villaverde et al (2012), who study how changes in uncertainty about future fiscal policy affects aggregate economic activity. Our main contribution to this literature is empirical. We document time-varying volatility in a variable, the terms of trade, that has not previously been studied and explore the effects of changes in this volatility.

We also build on the literature examining the macroeconomic consequences of terms of trade shocks. Many papers in this literature have examined terms of trade shocks using calibrated business cycle models. These typically conclude that terms of trade shocks are an important driver of small open economy business cycles. For example, Mendoza (1995) concludes that terms of trade shocks account for around half of the fluctuations in GDP in developing countries and slightly less in advanced economies. In a model calibrated to match features of a standard developing economy, Kose and Riezman (2001) find that terms of trade shocks account for 45 per cent of output volatility and 86 per cent of investment volatility. And, in a model calibrated for Canada, Macklem (1993) finds that a 10 per cent temporary deterioration in the terms of trade – a large but not unprecedented shock for the economies in Figure 1 – reduces real GDP by almost 10 per cent and investment by almost 20 per cent.

Other papers in this literature have adopted a more reduced form approach and have examined the effects of terms of trade shocks in VAR models. These papers typically conclude that terms of trade shocks have smaller effects than is implied by structural business cycle models. For example, using a panel VAR covering 75 developing countries, Broda (2004) concludes that a 10 per cent permanent deterioration in the terms of trade reduces the level of GDP by around 1 per cent, and that terms of trade shocks explain between 10 to 30 per cent of the volatility of GDP growth. Similarly, Collier and Goderis (2012) find that a 10 per cent rise in commodity prices increases the level of GDP by around 1 percentage point after two years for a typical developing country. Our contribution to this literature is to highlight an additional channel – volatility – through which the terms of trade can have macroeconomic effects.

Alongside the literature examining the dynamic effect of shocks to the level of the terms of trade, another empirical literature documents a link between terms of trade volatility and long-run economic growth. Using a panel of 35 advanced and developing economies over the period 1870 to 1939, Blattman, Hwang and Williamson (2007) conclude that, for commodity producers, a one standard deviation increase in terms of trade volatility (in their sample, from 8 per cent to 13 per cent per year) causes a 0.4 percentage point reduction in annual per capita GDP growth. In related work, Williamson (2008) attributes much of the gap in economic performance in the early 19th century between economies in Western Europe and those in Eastern Europe, the Middle East and east Asia to the fact that the latter regions experienced more terms of trade volatility. Focusing on more contemporary patterns, Bleaney and Greenaway (2001) estimate a cross-country panel regression using data from 14 sub-Saharan African countries over the period 1980 to 1995 and also conclude that terms of trade volatility, measured as the residuals from a GARCH model of the terms of trade, reduces GDP growth.

Papers in this literature have also explored links between terms of trade volatility and the volatility of other macroeconomic variables. For example, using a panel of countries, Easterly, Islam and Stiglitz (2000) show that times of high terms of trade volatility tend to be correlated with times of more volatile GDP growth, while Andrews and Rees (2009) also establish a link with consumption and inflation volatility. Our paper complements this literature by illustrating the links between terms of trade volatility and macroeconomic outcomes in a fully specified macroeconomic model, and by tracing out the dynamic effects of changes in terms of trade volatility on output, external accounts and prices.