How to forecast an economic meltdown

by Hansi Mehrotra

3rd December 2015

Economic downturns are notoriously difficult to forecast, as evidenced by the 2007–10 financial crisis that surprised most of the world’s top economists. It has been difficult to identify indicators that consistently flag brewing recessions in time for policy-makers to take preventive action. If we hope to stave off crises with improved monetary policy, we could use a better warning system.

By combining indicators used to measure financial-sector risk, Chicago Booth’s Stefano Giglio and Bryan T. Kelly, with Arizona State’s Seth Pruitt, created a more sensitive alarm: a systemic-risk index that they argue is capable of forecasting widespread economic distress. Their statistical model appears to have predictive powers beyond any of its individual components.

The researchers evaluated 19 systemic-risk indicators typically used for measuring financial-system strength in the United States, as well as 10 measures used for both the United Kingdom and Europe. These indicators include standard regulatory measurements, such as levels of debt, risky assets and equity, liquidity, and default and credit spreads. The researchers also measured equity volatility, or the price swings in financial-sector stocks. The data focus on the 20 largest financial institutions in each region.

While much previous research focused on the recent financial crisis, Giglio, Kelly, and Pruitt analyzed the signals from indicators over multiple decades, through several business cycles. The data cover 1946–2011 for the US, 1978–2011 for the UK, and 1994–2011 for Europe.

The expanded time periods allowed the researchers to see that many individual indicators accurately predicted economic shocks only when certain general economic conditions are in play. For example, high debt and low liquidity accurately flagged subsequent lower national production activity at times, but not consistently.
Of the financial measurements, high volatility in the financial sector was the strongest and most consistent predictor of economic downturns, the study finds. By contrast, large swings in the prices of other sectors of equities had little, if any, forecasting power. Also, financial stocks performed worse than the rest of the market when there was broader economic trouble ahead. The findings suggest that financial-sector issues can arise ahead of, rather than as a result of, declining economic conditions.

The researchers amplify the predictive abilities of the volatility indicators by combining them into an index with other predictive, but less consistent, indicators. Their systemic-risk index predicted downturns in industrial production three months into the future, in every region and every time period. For example, the index predicted a one-in-five chance that annualized quarterly industrial-production growth would drop by at least 3 percentage points ahead of the recent worldwide recession. In normal times, there is a one-in-five chance of annualized quarterly industrial production growth declining nearly one-and-a-half percentage points, according to the authors’ estimates.

The researchers note that neither the index nor its individual components are entirely useful for forecasting dramatic economic upswings. Absence of volatility in financial stocks, for example, is not necessarily a sign of boom times ahead. These findings are in line with other research suggesting that financial-system stress can amplify broad economic trouble, turning a modest downturn into a recession. But stability in the financial sector doesn’t trigger extremely high production growth.

Giglio, Kelly, and Pruitt also find that the Federal Reserve cuts the federal funds rate when risks in the financial system rise. Although the action is intended to encourage economic activity, the researchers find that the preemptive action doesn’t fully stave off a downturn.

Work cited

Stefano Giglio, Bryan T. Kelly, and Seth Pruitt, “Systemic Risk and the Macroeconomy: An Empirical Evaluation,” Journal of Financial Economics, forthcoming.

——–

This article originally appeared in Capital Ideas, a publication of the University of Chicago (http://www.chicagobooth.edu/capideas)