How robust is the result that the cost of "leaning against the wind" exceeds the benefit? Response to Adrian and Liang

The main result in Svensson (2017) and its previous versions is that, given current knowledge and empirical estimates, the cost of using monetary policy to "lean against the wind" for financialstability purposes exceeds the benefit by a substantial margin. Adrian and Liang (2016a) conduct a sensitivity analysis of this result, state that "the result that costs exceed benefits rely critically on assumptions about the change in unemployment in a recession or crisis, the crisis probability, and the elasticity of crisis probability with respect to the interest rate," and provide alternative assumptions that they assert would overturn the result. This paper shows that Adrian and Liang's alternative assumptions are hardly realistic: they exceed existing empirical estimates by more than 11, 13, and 40 standard errors. Adrian and Liang furthermore do not comment on the extensive sensitivity analysis already done in previous versions of Svensson (2017), which supports the robustness of my result.