Nobel laureate Robert C. Merton challenged traditional models used by investors to measure sovereign and financial system credit risk and proposed an alternative framework during a keynote session at the CFA Institute European Investment Conference.
Merton began by arguing that any definition of sovereign debt should take into account another class of liabilities, top-up guarantees, which aren’t on the balance sheet but are just as real as those that are. For the United States, these guarantees amounted to an astonishing $17 trillion in 2010. “The values of these guarantees are enormous, particularly in times of stress,” Merton said. According to Merton, these explicit and implicit guarantees cause risk to propagate in substantial ways across the various sectors of the economy — household, corporate, financial, banking, and government sectors — as well as across borders. Merton is interested both in understanding what is going on with these guarantees and in measuring and monitoring risks.
Merton’s approach is based on contingent claims analysis (CCA), which models an issuer’s debt as a combination of risk-free debt and a short put option on the issuer’s assets. If the issuer defaults, the issuer has to give up the remaining value of the firm’s assets to the bondholder. The holder of the guarantee receives the promised value of the debt minus the value of assets recovered from the defaulting entity. So, the value of the guarantee, such as a credit default swap (CDS), is analogous to a put option on the assets of a borrower.
The way that macro risks build up in a non-linear fashion, which can be understood in terms of derivatives, is significant to Merton. “The insidious part of credit is the non-linearity,” said Merton. A put option is sensitive to the value of an underlying asset in a non-linear way. It exhibits “convexity,” in the jargon of derivatives. “As the asset goes down, the sensitivity goes up,” Merton explained. Similarly for a bank loan portfolio, if you have a second, same-sized shock, following from a first shock to the assets, then things move in a non-linear fashion. For Merton, if you don’t adjust risk parameters when circumstances worsen, then using old sensitivities is inappropriate: Deltas can be five times what they were, so a 2-sigma event looks like a 10-sigma event because you have measured the sensitivities incorrectly.
Worse still, during a crisis, volatility increases, so the value of guarantees rises across the board, not just within the banking sector. When governments bail out banks, they issue guarantees, or short put options, which have all the same unsavory non-linear sensitivities and characteristics as bank loan guarantees. “They are writing a guarantee on the bank assets, which are short put options. So, they are writing a put on a put,” Merton added. This means that when things go bad for governments, they do so at an accelerating rate.
Feedback loops of these risks can arise when each party is effectively guaranteeing the other: Governments get hit by weakening banks, which they are guaranteeing, and banks (holding government bonds) suffer further when sovereign debt weakens. “The weakness of each just cycles back to the other,” Merton said. Banks have interactions with other banks and with sovereign debts of their own and of different countries. Then there are interactions between sovereigns. So, an adverse feedback loop ties sovereign stresses to banking sector problems.
How can we measure such connections and their influence on credit ratings between sovereigns and financial institutions? Using data on sovereigns, banks, and insurers from January 2001 to March 2012, Merton and his co-authors propose an expected loss ratio (based on CCA) and network measures, which scale and price the value of insurance and its change over time. One market measure of this guarantee cost is seen in the form of CDS spreads. But Merton and his co-authors seek to measure the fair value of CDSs, including the government guarantee, which isn’t reflected just in the CDS. Merton’s measure uses the options market to come up with an estimate of what the CDS spread could have been if there were no government guarantees. This facilitates a study to measure credit connectedness and influence amongst institutions with the help of Granger causality tests.
Using his measures, Merton demonstrates that the banks, insurance companies, and sovereigns in the sample are dynamically connected, with one country spreading risk to another and vice versa. Network measures allow for early warnings and assessment of the system complexity. “Right now, sovereigns are having a big impact on banks, whereas previously, banks were having a big impact on sovereigns,” he said. By using these network measures, investors should be able to monitor risks and see how they change dynamically.
In view of his research, Merton proposes a unified macrofinancial framework, instead of the traditional flow and accounting frameworks used by most countries. “If we want to think about policy — monetary, fiscal, and financial stability — it’s a good idea that we should be thinking about integrating those,” Merton said.
Finally, Merton broadened the radar to include pension funds and pointed to the dire impact of monetary policy on U.S. pension funds. “What they [policymakers] have not taken account of is the unintended consequence of holding long rates lower than they otherwise would be. What does that do to pension funds? It kills them,” he said, pointing to the adverse effect of low long rates on pension liabilities. And Merton added that there is at least one unpleasant guarantee, a hidden feedback, between the state guarantors and those same pension funds: If the pension funds do fail, it’s the government that picks up the bill.