In my last article, I summarized what the recently-released annual report from the Office of Financial Research (OFR) had to say on the stability of our financial system. I described the development of a Financial Stability Monitor that tracks five categories of financial system distress using a mix of economic indicators, market indexes and calculated measurements. The last category, contagion risk, is defined as "the vulnerability of the financial system to sudden shocks that may spread through seemingly unrelated parts of the financial system."
In the report, the OFR spends some time addressing contagion in financial networks. It is the interconnectedness of the financial system, coupled with the effects of leverage, that allow negative shocks (such as sudden drops in asset price) to be amplified in the system and result in contagion. They describe a network model which contains a set of institutions as the network "nodes" and their payment obligations as the "links" or interconnections. The example that they provide for sake of illustration includes a mortgage company, a broker-dealer, a money-center bank and a regional bank, along with non-financial players such as depositors and intermediaries.
The simple example included in the report demonstrates that depending on leverage, the degree of connectivity of the institutions, and the size of the shock (drop in asset value) the outcome can be either a relatively small systemwide loss in value or one in which "a modest write-down in the assets of one node can trigger a cascade of defaults throughout the network" with a "total loss in value ... substantially greater than the amount of the initial default."
The OFR acknowledges that it is not possible to develop a network model of our financial system that adequately reflects the interactions among all of the players. However, they are developing several methods to characterize contagion and loss amplification without such detailed knowledge. The first, "node depth," gauges the extent to which initial losses at a given node are amplified as they course through the system. (As you might imagine, in the simple model the mortgage company has the greatest node depth because it is highly leveraged and deeply interconnected.) The second metric is financial connectivity, the degree to which a financial institution is obligated to other institutions as compared to their outside obligations. An estimate of systemic financial connectivity is provided by looking at liabilities which are likely held by other financial institutions, both on and off balance sheet.
Financial connectivity of the 50 largest U.S. banks
At the same time that they put forward the results of their research, the OFR is quick to call attention to its limitations as well. The network models are being extended to cover factors that may contribute to contagion such as default costs, "fire sales" and mark-to-market losses. More critically, there is a "data gap in financial stability monitoring" with the statement that "Financial regulators do not have, even at an aggregate level, comprehensive measures of what fraction of liabilities of financial institutions are obligations to other financial institutions." If you accept (as I do) that interconnectedness is one of the keys to understanding the vulnerability of our complex financial system, this may give you pause.
In the next article about the OFR's report, I will conclude this three-part series with the OFR's progress and outlook on promoting data standards.