Limits to Arbitrage During the Crisis: Funding Liquidity Constraints & Covered Interest Parity

Arbitrage ensures that covered interest parity holds. The condition is central to price foreign exchange forwards and interbank lending rates, and reflects the efficient functioning of markets. Normally, deviations from arbitrage, if any, last seconds and reach a few basis points. After the Lehman bankruptcy, instead, arbitrage profits were large, persisted for months and involved borrowing in dollars.
The first goal of this paper is to measure precisely such deviations from arbitrage. By using high frequency prices from novel datasets , we are able to exactly replicate two major arbitrage strategies: "secured" and "unsecured" arbitrage. Specifically, arbitrage can be undertaken by borrowing and lending funds on secured terms, as would a hedge fund, or on unsecured terms, as would a bank's proprietary trading desk (prop desk). The distinction draws on that made in Brunnermeier and Pedersen (2009). Both arbitrage strategies yield very similar and consistent results.
The second goal of the paper is to investigate why arbitrage broke down. Empirical analysis suggests that insufficient funding liquidity in dollars kept traders from arbitraging away excess profits. Risk factors, instead, are mostly insignificant. Liquidity factors involve intermediaries cutting back loans to arbitrageurs in order to shrink their balance sheet, and hoarding liquidity to cover their own funding strains. In addition, it seems that arbitrageurs themselves had insufficient pledge-able capital to fund their arbitrage trades.
This paper contributes to the literature on slow moving capital. It shows how capital constraints turn into enduring arbitrage opportunities. For policymakers, it shows that certain unconventional monetary policy measures (in particular central bank swap lines) were effective at providing the necessary funding liquidity in dollars and re-establishing arbitrage across money markets.