Lessons for the foreign exchange market from the global financial crisis

Michael Melvin, Mark Taylor 06 November 2009



The timing of the subprime crisis, which became the global crisis, is well known; see, for example, New York Fed (2009). The impact on the foreign exchange markets has been much less discussed.

The crisis in the foreign exchange (FX) market came relatively late. In the early summer of 2007, it was apparent that fixed income markets were under considerable stress. Then, in July 2007 the dominoes began to fall as supposedly market-neutral equity portfolios suffered huge losses and it was common to hear talk of a five (or larger) standard deviation event, reflecting the failure of traditional financial risk analysis to capture systemic effects. Foreign exchange market participants watched these other markets with growing trepidation. Their fears were realised on 16 August 2007 when a major unwind of the carry trade occurred and many currency market investors suffered huge losses. We thus date the beginning of the crisis in the foreign exchange market as August 2007.

Four stages in the FX crisis

An efficient way to visualise the various stages of the crisis in the currency market is provided by the returns to the so-called “carry trade,” a popular foreign exchange investment strategy of taking long positions in high-interest rate currencies financed by short positions in low-interest rate currencies. Figure 1 displays a standard carry trade index over the crisis period.

Figure 1. Deutsche Bank carry index

Beginning in August 2007, the first wave of the subprime crisis hit the foreign exchange market as losses in equity and fixed income portfolios spilled over as portfolio managers deleveraged and liquidated winning positions, including their currency positions.

November 2007 saw credit problems increase as firms found it increasingly difficult to issue commercial paper, and there was a flight to quality in which yields on US Treasury securities fell substantially. Investors shed risk from their portfolios, including foreign exchange investments.

Risk appetite fell again in March 2008 as rumours of Bear Stearns’ imminent demise began to circulate. The orderly rescue and sale of Bear Stearns to JP Morgan Chase restored some stability to the market. The market came to believe that there is a “too big to fail” doctrine, so that counterparty risks were perceived as manageable going forward. Then, in September 2008, this confidence-building exercise was undone as Lehman Brothers was allowed to fail, as policymakers began to worry about the moral hazard implications of bank rescues.

Lehman Brothers’ failure imposed huge losses on many of their counterparties who were unable to collect on obligations owed them. Post-Lehman, there was a dramatic fear across the market as to where losses hid and who might be next to go under. Institutions were monitoring counterparty exposures more carefully than ever and some institutions, considered more at risk than others, found their client base shrinking.

Post-Lehman: Risk, volatility, and transaction costs

The failure of Lehman added a new dimension to perceptions of risk. Post-Lehman exchange rates experienced unprecedented levels of volatility and FX transaction costs rose dramatically.

When market makers provide liquidity to the market, they assume inventory positions in currencies as a result of their trades. The greater volatility, the greater risk they face from holding positions. As a result, the bid-ask spread rises to compensate them for this risk.

In the fall of 2008, FX spreads widened dramatically (Table 1). In the pre-crisis period, the spread on a two-way price quoted by a market-maker to a hedge fund might be in the range of 4 “pips,” e.g. an offer to buy dollars at 1.1525 Swiss francs each or to sell dollars at 1.1529 Swiss francs. During the post-Lehman crisis period, the spread widened to at least 16 pips.

Table 1. FX bid-ask spreads for $50 million, in pips

Values are for risk-transfer trades in the range of $50-100 million, in which a counterparty requests a two-way price from a market maker in a bilateral transaction. These should be considered representative of the periods under consideration. Larger trades would generally have wider spreads. Values are in “pips,” for instance if the spread on EUR-USD is 1, then the spread would be something like 1.3530-1.3531. Both spot and 3-month forward swap spreads are given.

In the worst of times, spreads on particular currencies were at least 400% wider than normal times. In addition, there were times in the fall of 2008 when it was difficult to trade at all in normal sizes.

Even more dramatic than the spot spreads was the widening that occurred in spreads for forward delivery. Table 1 also contains data on indicative spreads for 3-month swap (i.e. forward minus spot) quotations. In the pre-crisis period, swap spreads on the dollar-Swiss franc would be around 0.4 pips; post-Lehman, they were about 15 pips.

Regulatory and legislative reaction and the foreign exchange market

The cost imposed by the financial crisis has resulted in a legislative and regulatory reaction to rein in risk-taking and speculative behaviour. One effort has been attempting to reduce compensation at banks that have accepted government assistance. In one instance, a UK bank paid no bonuses for 2008. This government reaction to the crisis is not surprising, but it is doubtful that those setting the rules fully understand the implications of the changes they are forcing on the financial industry.

The losses experienced by financial institutions did not come from foreign exchange trades, imposed compensation restrictions treat the foreign exchange function the same as other areas of the bank. We expect bank employees to respond in a predictable manner to a changed incentive structure. Since compensation is severely limited compared to the past, the risk-return tradeoff has changed in a manner that is probably consistent with public policy; less incentive to take risk results in less risk taking.

For example, in the foreign exchange market, market-making dealers are expected to provide liquidity to their counterparties and then manage the risk of their positions while earning a profit for their banks. Competition across banks resulted in tight spreads and a willingness to provide good two-way prices for large trade size. This willingness to bear risk on the “sell-side” was beneficial to the “buy-side” bank clients. In fact, given the large spreads reported in Table 1 and the large volume of trading that occurred in 2008, bank profits from foreign exchange were very large.

In the aftermath of the crisis, dealers are charging wider spreads and dealing in smaller amounts than in the past. This will lower the bank’s risk exposure but impose greater costs on the banks’ clientele – non-bank financial institutions, corporate customers, governments, central banks, international travellers, and others who benefit from liquidity and risk-management services provided by banks.

A predictable implication of the public policy response to the financial crisis is to lower liquidity and raise the risks and costs associated with non-bank currency trades. The “buy-side” faces greater costs associated with currency trading along with greater volatility of exchange rates. It should be more difficult for non-banks to transfer their currency risks to a bank than in the past, while the non-bank entities face greater risk in the foreign exchange market than they used to. It is not clear that there is a net gain to society from these changes.

A practical risk-control strategy for FX managers

One practical issue that arises concerning the crisis in the FX market is whether FX portfolio managers could have protected their portfolio by an appropriate risk control strategy.

In order to illustrate such a strategy without recourse to proprietary methods, we constructed a global financial stress index that is similar in many respects to the index recently proposed by the IMF (2008). The index is a composite variable built using market-based, publicly available indicators in order to capture four essential characteristics of a financial crisis: large shifts in asset prices, an abrupt increase in risk and uncertainty, abrupt shifts in liquidity, and a measurable decline in banking system health indicators.1

In order to ascertain whether an extreme value of the financial stress index has been breached, we subtract off a moving-average mean and divide through by a moving-average standard deviation. The result then gives a measure of the how many standard deviations the index is away from its time-varying mean. As can be seen from Figure 2, the global financial stress index crosses the threshold of one standard deviation above the mean for most of the major crises of the past twenty years or so, including the 1987 stock market crash, the Nikkei/junk bond collapse of the late 1980s, the 1990 Scandinavian banking crisis, and the 1998 Russian default/LTCM crisis.

Figure 2. Global financial stress index

The 1992 European exchange rate mechanism crisis is less evident at the global level. Most interestingly, however, the global financial stress index shows a very marked effect during the recent crisis. Mirroring the carry unwind in August 2007, there is a brief lull in the index as it drops below one standard deviation from its mean before leaping up again in November 2007 to nearly 1.5 standard deviations from the mean.

The global financial stress index then breaches the two-standard deviation threshold in January 2008 and again in March 2008 (coinciding with the near collapse of Bear Stearns). With the single exception of a brief lull in May 2008, when index falls to about 0.7 standard deviations above the mean, it remains more than one standard deviation above the mean for the rest of the sample, spiking up in October to more than four standard deviations from the mean following the Lehman Brothers debacle in September.

We simulated the returns an investor could have earned from investing naively in the Deutsche Bank Carry Return Index and the returns from investing in the index in normal periods and closing out the position in stressful periods when the FSI exceeds a value of 1. Over the entire 2000-2008 period studied, the naïve carry strategy yields an information ratio of -0.3 (the annualised return divided by the annualised standard deviation of returns) while the risk-controlled carry strategy yields a ratio of 0.69. Over the more recent 2005-2008 period, the naïve information ratio is -0.66 while the risk-controlled information ratio is 0.31.

In this regard, we see that the financial stress index and similar market stress indicators may have practical potential to add value to foreign exchange investments.2


1 We examine the same group of seventeen developed countries as in the IMF study but, in contrast to the IMF analysis, we built a ‘global’ financial stress index based on an average of the individual country indices. See Melvin and Taylor (2009) for further details.

2 This is likely to be true even after allowing for transaction costs and implementation lags – see Melvin and Taylor (2009).


International Monetary Fund (2008), “Financial Stress and Economic Downturns,” World Economic Outlook: Chapter 4, 129-158.

Melvin, Michael and Mark P. Taylor (2009), “The Crisis in the Foreign Exchange Market,” CEPR Discussion Paper 7472, September (forthcoming in the Journal of International Money and Finance special issue on the Global Financial Crisis, December 2009).

New York Fed (2009).”Timelines of Policy Responses to the Global Financial Crisis.”



Topics:  Exchange rates International finance

Tags:  global crisis, foreign exchange, FX markets

Head of Currency and Fixed Income Research, Global Market Strategies Group, Barclays Global Investors

Dean and Professor of Finance, Olin Business School, Washington University in St Louis, USA; Research Fellow CEPR; Senior Fellow, Brookings Institution


CEPR Policy Research