The Microstructure of Covered Interest Arbitrage

In this study, we use a proprietary dataset of New Taiwanese (NTD) and U.S. Dollar (USD) swaps transactions to examine different factors that may impact the implied covered interest returns. This is the first study using actual transaction level data from swaps market to test covered interest arbitrage. The data covers trade-to-trade swaps transactions from August 2001 to July 2003. It includes each transaction’s spot rate, swap rate, buyer’s identity, seller’s identity, spot date and swap date. Using interest rate parity, we calculate the implied covered interest returns as follows.

Annualized Implied Covered Interest Return=|(1+r_T)-(F/S)*(1+r_U)|*(360/Days)

where r_T is the Taiwan interest rate, r_U is U.S. interest rate, F is forward rate, and S is spot rate. Since contract term of a swaps transaction can range from 1 week to 1 year, to make all returns comparable, we annualized all the returns. The annualized implied returns are economically large with a median value of 15.4%. To examine the factors that impact the returns, three groups of factors are included in the regression, namely the market order flow, the initiator’s identity, and market timing. Results suggest that traders will engage in larger transactions when the implied returns are high, however, greater daily volume in the market is associated with smaller implied returns.  Foreign banks and newer Taiwanese banks are more likely to achieve higher returns.  Additionally, higher implied returns occur for transactions with more irregular durations or occurring in the last week of the quarter, first week of the quarter and immediately after 9/11.