Distress in the financial sector can be transferred to the real economy. The authors develop an early warning indicator called "CATFIN," which is based on the catastrophic risk of financial firms. CATFIN can predict future macroeconomic slowdowns six months before they occur. The authors also show that the financial sector provides lending activities to other parts of the economy and that an increase in systemic risk in the banking sector may lead to an overall macroeconomic decline.
The authors develop an early warning indicator called "CATFIN" and demonstrate the measure’s ability to predict macroeconomic declines. In addition to making predictions regarding the U.S. GDP growth rate, CATFIN can accurately predict European and Asian GDP growth rates. The authors argue that because banks have a special role in society, risk taking in the banking sector is linked to real economic performance. Unfortunately, banks do not always consider the effect (i.e., cost) that their actions will have on society when making decisions.
How Is This Research Useful to Practitioners?
Policymakers at central banks consider the economic cycle when modifying their money supply and bank exposure limits. CATFIN may provide regulators with early warnings that would enable them to update their policies and reduce economic fluctuations. CATFIN also provides information on future volatility and thus can be an interesting metric for option and other derivatives traders and portfolio managers. Interestingly, the CATFIN measure can be applied at any point in time.
Previous research has indicated that financial shocks and illiquidity in the financial sector are transferred to the real economy, but researchers have focused mainly on microlevel measures. But microlevel systemic risk analysis has no predictive power, so the authors concentrate on the macro level.
They demonstrate that combining several risk measures results in a measure (CATFIN) that has significant predictive power. First, they use the generalized Pareto distribution to model extreme losses. Second, they consider the skewed generalized distribution, which takes into account the entire shape of the distribution. Lastly, they apply empirical data to determine a nonparametric value at risk. The nonparametric approach has the advantage of more accurately specifying the distribution of the risk factors.
CATFIN can explain liquidity from either a demand perspective—that is, investment decreases if volatility and uncertainty increase—or a supply perspective—that is, banks decrease borrowing when outcomes are more volatile. Furthermore, CATFIN is able to forecast financial market volatility by considering index options and credit default swap spreads. The authors show that CATFIN can predict bank lending activity a year in advance and that an increase in CATFIN is an indicator of reduced bank profits.
How Did the Authors Conduct This Research?
The authors use financial firms’ monthly returns in excess of the one-month Treasury rate. Monthly equity returns are collected from several U.S. stock indices from January 1973 to December 2009. International stock data and international GDP data are collected from the Datastream database; extreme stock returns are excluded from the sample.
By performing a regression analysis against the Chicago Fed National Activity Index, which measures GDP growth, the authors test CATFIN’s predictive power on GDP growth. Their results indicate that CATFIN is statistically significant both with and without other control variables. They conclude that CATFIN is able to predict recessions.
The authors also extend their tests to five other alternative U.S. GDP measures and show that CATFIN successfully predicts economic downturns 5–12 months in advance. In some instances, CATFIN gives a false early warning indicator. The authors suggest that these falsely predicted downturns might be the result of regulators’ intervention when the economy is headed for a downturn, in which case CATFIN’s information is still valuable.
Next, the authors test if CATFIN has predictive power when taking into account three microlevel measures of financial risk; they find that it remains a robust early warning indicator. They use Hill’s tail risk estimation (Annals of Mathematical Statistics 1975), the financial firm’s marginal expected shortfall, and the distance to default. The results indicate that CATFIN has predictive power but that the other three metrics do not.
By testing nonfinancial firms separately, the authors assess whether nonfinancial firm data can predict economic slowdowns. They combine the nonfinancial firm data to create a "fake bank" and use it to determine that simple diversification within financial institutions does not explain CATFIN’s predictive power. They also demonstrate that the special role of banks is not limited to banks that are "too big to fail."
This article is very useful, and the CATFIN metric can be used by a wide range of practitioners, from economists and regulators to option traders. The evidence the authors provide is thorough; they show CATFIN’s results in several countries and for several economic growth measures and compare it with other metrics. CATFIN can be applied widely, in such areas as GDP growth, lending activity, and bank profits. Although CATFIN works well, the reader is left wondering if using one or two of the measures CATFIN takes into account (e.g., generalized Pareto distribution or skewed distribution) would lead to similar results. Nevertheless, the evidence provided for CATFIN is deep and thus convincing.