Crowds, Crashes, and the Carry Trade
Currency carry trades have been shown to suffer sudden, extreme losses. A popular explanation is that these losses are to some extent driven by leveraged carry trade speculators amplifying negative shocks through forced unwinding of their positions. This implies that the likelihood of large carry trade losses (crashes) increases with the level of carry trade activity (crowdedness). To test this, I develop a measure of crowdedness based on daily, abnormal, currency return correlation among the target currencies. Using four decades of daily carry trade returns, I show that between 40% and 50% of the largest carry trade losses occur in periods of high crowdedness. I further demonstrate that high levels of crowdedness double the probability of realizing an extreme carry trade loss after controlling for FX volatility, FX liquidity, equity volatility and funding liquidity. Finally, I show that carry trade crowdedness negatively forecasts monthly carry trade returns.
PRESENTED AT: Aalto University, Amsterdam Business School, BI Norwegian Business School, HEC Montréal, Lund University, Norwegian School of Economics, Nova School of Business and Economics, Stockholm Business School, Stockholm School of Economics, Universidad Carlos III de Madrid, University of Georgia, University of Gothenburg, University of Texas at Dallas, Vrije Universiteit Amsterdam, Warwick Business School.
(with Patrick Augustin, Marti G. Subrahmanyam and Davide Tomio)
Abstract: We provide empirical evidence of a significant complementarity between the size of a country's debt and the net amount of insurance purchased against default by its government, based on a novel data set of net notional amounts outstanding for single-name sovereign credit default swaps (CDS) from October 2008 to September 2015. Domestic and international debt, the underlying reference obligation for many CDS contracts, reflect different information sets and, together with the size of the economy, explain up to 75% of the cross-country variation in net insured positions. Unlike for CDS spreads, for which a single principal component accounts for 54 percent of the cross-sectional variation, common global factors explain only up to 7 percent of the variation in sovereign CDS net notional amounts outstanding, consistent with findings that net sovereign insurance is driven primarily by country-specific risk. We further pinpoint two economic channels that explain the net trading in sovereign CDS: (a) country-specific credit risk shocks that change banks’ capital requirements based on regulatory rating thresholds, and (b) the issuance, but not the announcement, of domestic and international debt. All our findings suggest a strong hedging motive for the use of sovereign CDS.
PRESENTED AT: AFA 2018, Bank of Italy, China International Conference in Finance, IFSID Conference on Structured Products and Derivatives, Kiel Workshop on Empirical Asset Pricing, 2016 HEC-McGill Winter Finance Workshop, New York University Stern School of Business, Stockholm School of Economics.
(with Egle Karmaziene)
Abstract: This paper examines the effects of the 2008 short-sale ban on exchange traded funds (ETFs). Short sales of banned stocks decreased significantly during the ban period. However, we demonstrate that a portion of that decrease was reabsorbed by financial-sector ETFs and the biggest and most liquid ETF - the S&P 500 Spider. We argue that short selling equity ETFs was a viable method of circumnavigating the ban. Additionally, we offer evidence that the supply of ETF shares available for lending was able to be increased rapidly to meet the demand through ETFs' unique creation mechanism (“create-to-lend”).
PRESENTED AT: Bank of Lithuania, Columbia Business School, International Monetary Fund, McGill University, Nordic Finance Conference, Stockholm School of Economics, University of Cape Town.
Industry Coverage: "The short on shorting ETFs: The art of create to lend." Eurex Group Institutional Insight, May 2014.
Work in Progress
The Benchmark Currency Stochastic Discount Factor
(with Erik Fredriksen)
The question of what the appropriate pricing kernel is for pricing the positive expected excess returns for investments in high interest rate currencies and the negative expected excess returns for investments in low interest currencies remains open in the literature. A number of competing candidate currency pricing kernels currently exist. We use a relative entropy minimization approach to extract the most likely pricing kernel to price the interest rate sorted currency portfolios out-of-sample. This pricing kernel delivers superior cross-sectional fit and smaller pricing errors than the currently available currency return pricing models. Moreover, our pricing kernel offers an intuitive benchmark to which the existing candidate currency pricing kernels can be compared and presents a tractable framework within which we survey the existing literature on currency risk premiums.