Identification of fiscal SVARs in small open economies using trading partner forecast errors as instruments
We identify structural vector autoregressions with external instruments (SVAR-IV) to study the dynamic effects of fiscal policy. Our main contribution is a novel instrument for aggregate output shocks of a small open economy. Unexpected shocks in domestic output are proxied by forecast errors of professional forecasters in trading partner economies. Our instrument relies on two key assumptions. Firstly, unexpected changes in trading partners are correlated with unexpected shocks of an open economy (relevance). Secondly, unexpected fiscal shocks of a small economy are unrelated with the forecast errors of its trading partners (exogeneity). Test results show that this instrument is relevant. We find suggestive evidence that our instrument is more credibly exogenous than the prevailing TFP instrument. We apply our instrument to estimating fiscal SVAR models of two countries, Canada and Finland, and find that estimates of the spending multiplier are sensitive to conventional identification assumptions.
- JEL: E62, C26, C32
- Publication in PDF-format