The author uses a rich universe of data to identify the key drivers of the financial crisis for each country and then summarizes the common causes across countries. Unlike earlier studies, the author determines a country-specific dating of the crisis period that allows for deeper insights into the drivers specific to each country.

The author finds that common drivers of the crisis include high levels of credit growth, elevated exposure to external funding, high levels of real activity, and low levels of international reserves. An interesting observation is that countries with food products constituting a significant percentage of their exports tend to fare better as global trade constricts. Unusual results for Belarus and Ukraine are also noted.

Although many of the findings are not necessarily new, the author’s approach of using four different measures of crisis severity, country-specific dating, and various statistical methods to test approximately 100 explanatory variables and their myriad of interaction effects is different.

This research is particularly relevant to fund managers, asset consultants, and strategists with a focus on global and emerging market strategies because it may be a useful input into strategic asset decisions at the country level. It may also be useful for risk management purposes to identify and manage exposure to these risk factors.

The author collects macroeconomic and financial data dating back to 1980 from numerous sources for 63 advanced and emerging economies. Because the crisis affected different countries at different times, he determines a separate crisis commencement date for each country. Four estimates of the severity of the crisis are calculated for each country: Two measures are based on the short-term impact of the crisis and two are based on longer-term measures. The short-term measures include a measure of the cumulative loss in real output and the depth of the crisis relative to real output in the year prior to the commencement of the crisis. The two longer-term measures consider the loss in real output relative to longer-term growth trends; the first measure assumes that the fall in growth during the crisis was temporary and the second measure assumes that trend growth has been permanently lowered.

The author then runs a series of regressions for each crisis measure. Given that approximately 100 variables are available for inclusion in the regression, model averaging methods are used to reduce the impact of model uncertainty. The benefit of this approach is that it tends to lead to results that are less affected by the decision to include/exclude variables.

He uses Bayesian techniques to determine the weighting given to each model, and the importance of each variable in explaining the crisis measure under investigation is assessed by comparing it with the posterior inclusion probability.

Sophisticated econometric techniques are used in this research, which may make it difficult to replicate the results. But the conclusions are broadly in line with earlier research.

There are many views on the underlying causes of the global financial crisis, with high credit growth generally considered to be a major driver. This view has been embraced by banking regulators with an increased focus on the dangers of elevated credit growth and the need for higher levels of capital in the banking system as credit growth increases. Although the adage that “history never repeats, but it rhymes” seems applicable, there is the risk that the next financial crisis may be caused by a different combination of drivers that may not have been as closely monitored as those that caused the last crisis.