RDP 2001-05: Understanding OECD Output Correlations 4. Conclusions

In this paper we have identified a number of variables that can explain bilateral growth correlations among a sample of OECD countries. Consistent with earlier work by Frankel and Rose we generally find a significant role for bilateral trade in goods, with higher bilateral trading intensities associated with higher output growth correlations. The volatility of bilateral exchange rates is also found to be significant with lower bilateral exchange rate volatility associated with higher correlations.

In addition to these two transmission mechanism variables we find significant effects on output growth correlations from variables that measure certain characteristics or institutional features of economies. In particular, good accounting standards, similar legal systems, common language and openness to new technology are important factors. Unfortunately we are not able to determine the exact role played by these characteristic variables. It seems likely that either they make it easier for shocks to be transmitted between countries or they cause a pair of countries to be more susceptible to a similar type of shock.

Also worthy of note are variables that are not found to be significant factors in explaining bilateral growth correlations. These include: the similarity of monetary policy in the two countries (as measured by the volatility of the spread between two countries' short-term real interest rates); the extent of integration of long-term bond markets; common industry structure and differences in structural reform. While some of these variables are individually significant they typically do not survive in a multivariate regression framework. The lack of significance of industry structure in our general model is a particularly interesting finding in light of the strong effects reported by Imbs (2000).

Our empirical results can be used to help understand the nature of the strong correlation between Australian and US output growth over the last twenty years. Using our preferred model we can explain about 80 per cent of the actual Australian–US correlation of 0.66. In our view this is an impressive result particularly given that our model was designed to fit bilateral correlations across 17 OECD countries. According to our results, the high correlation between Australia and the US is probably not due to the direct transmission of shocks from the US to Australia, via trade or financial markets or through coordinated monetary policies. Overwhelmingly the results in this paper suggest that it is the presence of common characteristics and institutions in Australia and the US that primarily explains the high correlation. Why exactly these characteristics and institutions help to explain output growth correlations between countries remains an open question. It may be that countries with similar characteristics respond similarly to common shocks. Alternatively, there may more subtle transmission channels at work – perhaps the transmission of ideas, including the effects on output of a similar approach to the take-up of new technologies.