What are the effects of Large Scale Asset Purchases on asset prices in the US?

Abstract

In this thesis we research the effects of the Federal Reserve’s Large Scale Asset Purchase programs on asset prices in the United States, mainly stock prices and long term interest rates. First, an event study is employed to capture the effects of the announcements regarding LSAP, which gives an indication on what effects the LSAP programs had on asset prices. In addition, we employ a Vector Autoregression and a Vector ErrorCorrection model to analyse the actual effects of the asset purchases done through the LSAP. From the event study we find that LSAP announcements have a statistically significant negative impact on five, ten and thirty year treasury yields. The results indicate a downward shift of around 0.4% for these yields. These findings are in line with what other studies have found, as well as what is expected from theory. The effects of LSAP announcements on the S&P500 index are not significant, which could be due to the fact that LSAP programs are not directly affecting the S&P500 index, and thus it is not captured in the short estimation window in the event study. From the Vector Autoregression we find that an onestandard deviation shock to the Securities held outright by the Federal Reserve (measure of LSAP) has a positive effect on the S&P500 index of about 1.5% in the first few months, before the effect declines back towards equilibrium after about a year. For ten year treasury yields we find that an onestandard deviation shock to Securities held outright by the Federal Reserve results in a positive response of about 0.10%, before it moves back to equilibrium after around a year. The results from the Vector ErrorCorrection model is similar to the results from the Vector Autoregression model, with the most noteworthy difference being that the effects appears to be more permanent. The positive reaction on the ten year treasury yields due to a shock in Securities is not in line with what we would expect, and what other studies have found. The reason for this could be that we have somehow misspecified our models, or that a more complex model that opens up for structural restrictions is better suited for this analysis. i