The treasury auction risk premium

Using a time series asset pricing model, I empirically show that underpricing of U.S. Treasury securities is explained by risk premia that compensate dealers for bearing price risk. This finding suggests that the Treasury could reduce underpricing by reducing the post-auction price risk (volatility) to auction participants, which can be achieved mathematically by reducing the time from auction to settlement. I calculate that underpricing cost the Treasury $46.3 billion from January 2000 through June 2016. I estimate that standardizing the settlement period to 1-day could have saved the Treasury $15.6 billion over the same period. In addition, I use the estimated model to forecast expected risk-adjusted returns (that result from underpricing) for each auction, and find that these forecasts predict Treasury auction demand. This finding suggests that auction demand depends on underpricing, albeit on an expected risk-adjusted basis. Further, this expected underpricing may actually help the Treasury to sell debt and avoid auction failures.