Breakdown of covered interest parity: mystery or myth?

Most of the post-crisis covered interest parity literature treats the persistent cross-currency basis as an anomaly requiring an explanation in terms of dollar shortage, regulation or limits to arbitrage. This paper, by two Hong Kong Monetary Authority researchers, argues that the anomaly is largely definitional: Libor is an unsecured borrowing rate carrying a counterparty risk premium, a currency swap is a secured transaction in which the exchange of principals means “the parties effectively hold each other’s loan as collateral,” and therefore a dealer who continued to price the swap off the raw Libor differential would be the one behaving irrationally. On that reading the basis is the rational adjustment for the difference between the two money markets’ counterparty risk premiums, and the phrase “breakdown of CIP” is a myth in the specific sense that it implies market malfunction. The argument is set out in three steps: a cross-currency basis swap is shown to be equivalent to a series of shorter-term FX swaps, so the CCBS basis is approximately equal to the CIP deviation; the exchange of principals is argued to strip counterparty risk but not liquidity risk, which is why a basis remains even when Libor is replaced with OIS, repo or government bond rates that carry “negligible liquidity risk premium”; and a risk-adjusted CIP condition is derived in which the forward premium depends on weighted averages of each currency’s OIS and interest rate swap rates, with two testable restrictions – a zero constant and coefficients that sum to unity within each currency. The empirical work covers seven five-year currency pairs referenced to three-month rates, four with a dollar leg (USD/EUR, USD/GBP, USD/CHF, USD/JPY) and three with a euro leg (EUR/GBP, EUR/CHF, EUR/JPY), over 22 September 2009 to 30 June 2017 (from 13 January 2010 for the Swiss franc pairs, limited by CHF OIS availability), with 1,887 to 2,029 observations per regression and observations more than five standard deviations from the mean deleted, retaining 99.0-99.7 percent of the data. A Durbin-Wu-Hausman test after GMM estimation cannot reject exogeneity of the four interest rates at the 10 percent level or higher for any pair, so the authors report OLS. The constant is close to zero and insignificant in every regression; all four coefficients lie between zero and one; and the within-currency sums are 99.2, 96.1, 100.2 and 98.0 percent for the dollar leg and 100.3, 100.1 and 98.5 percent for the euro leg. Because the swap market separates the two risk premiums, the model yields a decomposition of the Libor-OIS spread: averaged across the four dollar pairs, the counterparty risk premium is 22.3 percent of the total risk premium in USD Libor against 75.8 percent in EUR Libor (23.7 and 76.7 percent in the restricted model), with pair-by-pair USD shares of 17.5 percent against the euro, 36.8 against sterling, 18.9 against the franc and 16.1 against the yen. Adjusted R-squared runs 0.62 to 0.80 for most pairs in first differences, falling to 0.36 for EUR/CHF and 0.52 for EUR/JPY. Two further results support the “myth” reading: the bases satisfy a triangular no-arbitrage relationship, so they are “not arbitrarily determined but fairly priced”; and non-zero bases persist between pairs with no dollar leg at all, which the authors treat as a challenge to purely dollar-centred explanations. The authors are explicit about what they concede: they “acknowledge the possibility that they are determined by the limits to arbitrage caused by plausible constraints such as capital charges resulting from recent regulatory reforms,” and they note that the Libor-OIS spread is “not a perfect measure of the risks for the CCBS market,” so their estimated risk shares “are likely to be underestimated.”

Alfred Wong | Macro Paper Warehouse

Breakdown of covered interest parity: mystery or myth?