Abstract
This thesis consists of two independent chapters on monetary policy.
Chapter 1 examines the role of house price expectations in influencing the inflation expectations of households. We use two survey datasets for the United States which report quantitative expectations for both inflation and house prices. We find that households tend to overweight house price expectations when forming their inflation expectations, relative to an accounting benchmark. Examining cross-sectional heterogeneity in the data reveals a significant effect of the cognitive abilities of households as more sophisticated households don’t overweight house price inflation as much. We model this household behaviour in a two-sector New Keynesian model, with an overweighted and a non-overweighted sector, and analytically derive a welfare loss function consistent with the micro-foundations of the model. In this setup, we show that to gauge the correct interest rate response, it is imperative for the central bank to be aware that some sectors are overweighted by consumers and that movements in expected inflation in such sectors are important for optimal monetary policy.
This work is motivated by the role of the salience of large changes in determining inflation expectations of households. The contribution of this chapter is twofold. First, we find a novel channel of salience through house price expectations; this makes a case for the central bank to monitor the housing sector beyond the usual, very important, financial stability concerns. Second, the paper breaks new ground by presenting a simple model that applies more generally to understand the monetary policy implications of over-weighting in any good. This provides a framework that can be adapted to analyse some findings from the previous literature.
Chapter 2 estimates the impact of monetary policy shocks using a model-averaging framework. In this work, we use various measures of monetary policy shocks that have been previously identified in the literature, spanning different identification methods. We use these shocks as instruments, and using a model averaging technique, embed them in the first-stage of an SVAR-IV. Model averaging is a flexible method to select instruments and the literature shows it to have a more favourable trade-off between bias and efficiency in the case of a 2SLS, relative to other methods of instrument selection. We also conduct a small Monte Carlo simulation to explore the gains from using this methodology. In this context, we then look at the impact of monetary policy shocks in the US and use all the available estimates of monetary policy shocks for our sample period.
In assessing the finite sample properties of the model averaging estimator in a simple Monte Carlo study, we find model averaging performs better than a single instrument and when using all instruments as in the standard approach. However, in the empirical application, we find only marginal gains in using model averaging in terms of the correct response of output and prices in response to a contractionary monetary policy shock. The profile of the impulse functions remains quite similar with and without model averaging, when all instruments are used. Despite the limited gains from this approach, this methodology allows us to gauge which instruments, and hence identification strategies, drive the results.
The pursuit of finding the true monetary policy shock has been ongoing in macroeconomics. The literature on shock identification is constantly evolving and is replete with varied identification strategies. The resulting mixed evidence on the impact of unexpected changes in monetary policy leaves room for further research, which this work aims to contribute to.