Foreign treasury purchases and the yield curve: Evidence from a sign-identified vector autoregression

I employ a sign-identified vector autoregression (VAR) in foreign Treasury purchases and the first three factors of the yield curve to estimate the dynamic impacts of foreign Treasury purchases on Treasury yields.  Although a growing literature studies this question, it does not adequately address the endogeneity of foreign Treasury purchases to yields; not only can purchases affect yields, but yields can affect purchases.  I achieve identification using Bayesian sign identification (Inoue and Kilian (2013)) and estimate effects which are larger than in most of the existing literature, but which roughly match results from event studies.  I find that an exogenous foreign purchase equal to 1% of the outstanding stock of Treasury securities in a single month significantly lowers Treasury yields across the term structure for around two years.  As a rough guide, the largest impacts come nearly a year after the shock, when the 1- and 5-year yields are each lower by perhaps 28 basis points and the 10-year yield is lower by 21 basis points.  I argue that I identify larger effects because I successfully overcome the endogeneity of purchases to yields.  That endogeneity should bias the results of incompletely identified studies toward zero.  My results are robust to reasonable changes in the identifying restrictions, and they have three significant policy implications.  First, foreign demand for Treasury securities can explain the period of low long-term interest rates in the years leading up to the financial crisis.  Part of that episode has of course been labelled the “conundrum” after Alan Greenspan (2005).  Second, my results suggest that the large scale asset purchases (LSAPs) recently undertaken by central banks around the world should be effective in lowering yields.  My impacts are larger than those found in the literature directly studying the LSAPs, though there are other channels by which LSAPs might affect yields which are not important in my context.  Finally, I decompose the behavior of long rates into the responses of term premia and expected future short rates.  Following foreign Treasury purchases, term premia persistently and substantially decline while short rates rise with a sizable lag.  This suggests that the Federal Reserve tightens policy in response to foreign purchases, but still allows a loosening of credit conditions.