We employ the pairs trading algorithm popularized by Gatev et al. (1999, 2006) on U.S. stocks data from 1963 until 2019. As Do and Fa
(2011) we find that the pairs trader does not earn significant returns after transaction costs. We hypothesize that the pairs trader is still making a profit as compensation for providing liquidity during shocks to market and funding liquidity proxied by the VIX and TED-spread, respectively. We find evidence that the pairs trader on average is net long illiquid stocks, measured as the Amihud (2002) illiquidity measure. During sub-periods with shocks to market and funding liquidity, the pairs trader increases his/her loading on even more illiquid securities, but equally so on the long and the short leg. The pairs trading strategy has become more unprofitable in recent years, but if one implements during sub-periods with scarce market and funding liquidity we do find an indication of slightly higher returns after transaction costs, but not a statistically significant compensation. A liquidity-adjusted CAPM model does not provide an indication that the pairs trader is compensated for taking liquidity risk. We therefore come to the conclusion that the pairs trader is not compensated, to a significant degree, for taking liquidity risk during sub-periods characterized by liquidity shocks. We speculate that the pairs trader in a Gatev et al.(1999, 2006) framework is sub-optimally selecting its pairs to ideally reflect a liquidity risk premium exploiting trade.
|Educations||MSc in Finance and Investments, (Graduate Programme) Final Thesis|
|Number of pages||129|