Measuring Systemic Financial Risk

On a recent visit to UCSD, NYU Professor and Nobel Laureate Rob Engle called my attention to the NYU Stern Volatility Laboratory, a great resource that anyone can use to get some very interesting real-time analysis. Here I’d like to describe some of the features available for assessing the systemic risk posed by financial institutions.

The first step that Engle and colleagues propose is to calculate what they call the Marginal Expected Shortfall (MES) associated with a given financial institution. This is an estimate, based on recent dynamic variances and correlations of observed stock prices, of how much the stock valuation of a given institution would be expected to fall today if the overall market were to decline by more than 2%. This is essentially a time-varying tail-event beta, details of whose estimation can be found here.

They next used a dynamic simulation to extrapolate from the MES an estimate of how much the stock would fall in the event of a full financial crisis, defined as a 40% decline in a broad market stock index over a space of 6 months. They estimate this number to be around 18 times the daily MES.

The last step is to compare the magnitude of the decline with the firm’s current equity and liabilities, and calculate how much more capital the firm would need to raise in order to remain solvent in the event of another financial crisis. This measure, which they describe as the “systemic risk” associated with the firm, can either be reported in terms of how big the shortfall of that firm would be (in billions of dollars), or in terms of the percentage of the shortfall across all financial firms contributed by that single institution.

For example, here’s what their calculation gives you using Friday’s stock market data. If we were to experience a new financial crisis, three institutions– Bank of America, Citigroup, and JP Morgan Chase– would each need to raise about $100 billion, and between them would account for about half of the financial sector’s capital shortfall in the event of a major downturn.

Financial institutions with highest estimated systemic risk as of May 20, 2011. Source: NYU Stern Vlab.

By clicking on the heading for the Vlab page from which the above screenshot was taken, you can alternatively sort firms by any measure you like.

Another neat feature is you can go back in time to any date of interest. For example, here’s where things stood on August 29, 2008, right before the financial crisis, with firms sorted by MES. At that time, three firms: Fannie, Freddie, and Lehman– each had MES above 12, though several other institutions (remember AIG?) are assessed to have posed a bigger systemic risk.

Financial institutions with highest estimated dynamic MES as of August 29, 2008. Source: NYU Stern Vlab.

And here’s a plot of the path followed by the estimated MES for Lehman leading up to the crisis. Correlations among stock returns signaled Lehman’s growing vulnerability to a downturn.

Historical graph of estimated MES for Lehman Brothers as assessed on August 29, 2008. Source: NYU Stern Vlab.

I view these measures as a supplement to, rather than replacement for, other analyses based on direct linkages and derivative exposure. CDS could offer another useful market indicator. But the advantage of the NYU Stern approach is it can immediately draw out some of the implications of the latest stock market valuations and comovements for real-time use by regulators, investors, and business planners.

This post originally appeared at Econbrowser and is reproduced here with permission.