Yield curve arbitrage in the USD interest rate swap market: returns and risks in 2012-2021

In my master’s thesis, I study the performance of yield curve arbitrage in the USD interest rate swap market in 2012-2021. To my knowledge, yield curve arbitrage has been extensively studied only in the article that I replicate in my thesis. The article finds strong performance for yield curve arbitrage in the USD interest rate swap market in 1988-2004. Yield curve arbitrage aims at finding and taking advantage of relative mispricings on a given rate curve. I apply monthly data and I model the USD swap rate curve with the two-factor Vasicek short rate model and with the two-factor Cox-Ingersoll-Ross short rate model. The models assume that the development of a rate curve is explained by two stochastic factors. Yield curve arbitrage aims at being hedged against changes in the factors. My modelling indicates mispricings on the swap rate curve and I open monthly arbitrage trades for the largest mispricings. I calculate the market values for the open arbitrage trades and I calculate the returns by following the market values. I close the arbitrage trades when the mispricings disappear or when the 12-month time limit is exceeded I try to study whether yield curve arbitrage has provided better returns in 2012-2021 than in the study period of 1988-2004 of my reference article. I also try to observe whether there are differences in the performance of the short rate models and if my data and methodology affect my results. The means of returns of yield curve arbitrage are mostly positive but the confidence intervals of the means are wide. The distributional measures of the returns, skewness and kurtosis, indicate good performance in some arbitrage implementations and bad performance in others. I can not conclude that my returns would be different from the returns of my reference article or that one of the short rate models would provide better performance than the other. The data and methodology appear to have effect on the arbitrage performance. Yield curve arbitrage aims at having limited exposure to the market risks. My multifactor regressions indicate that the arbitrage does not provide market risk corrected positive returns and that the arbitrage is exposed to market risks. On the other hand, the regression results suggest that the arbitrage has exposure to exploiting the mispricings on the swap rate curve. I compare my returns to the returns of a comparable hedge fund index. The performance of the hedge fund index has been strong in 2012-2021 and the correlation between my returns and the returns of the index is low. The hedge fund index is exposed to some of the same market risks as my yield curve arbitrage returns. My analysis does not indicate strong support for yield curve arbitrage having been profitable in the USD interest rate swap market in 2012-2021. However, my results are inconsistent and inconclusive. My results are likely due to the problems in my methodology and due to my difficult study period in the 2010s when USD rate markets were extraordinary