Essays on expectations and learning in macroeconomics

Tang, Li ORCID: 0000-0001-7119-3186 (2022). Essays on expectations and learning in macroeconomics. University of Birmingham. Ph.D.

Tang2022PhD_Redacted.pdf
Text - Redacted Version
Restricted to Repository staff only until 31 December 2030.
Available under License All rights reserved.
Download (11MB) | Request a copy

Abstract

This thesis consists of three major chapters (papers), Survey Forecasts, Sentiment and Stock Market Volatility, Forecast Disagreement about Macroeconomic Relationships and Output Gap Estimation and Monetary Policy with Imperfect Knowledge. The core aspect is around expectations and adaptive learning in macroeconomics. I looked at the expectation formations in the framework of three major macroeconomic models: the Lucas-type consumption-based asset pricing model, the stochastic growth/real business cycle model and the New Keynesian model. Three different types of expectation formation of adaptive learnings are examined.

The first chapter (paper) documents new evidence that survey forecasts of stock prices are not anchored by forecasts of economic fundamentals in US stock markets. This evidence is at odds with a wide range of asset pricing models with various information assumptions. The paper develops and estimates a Lucas-type consumption-based stock pricing model with adaptive learning and sentiment shocks, which replicates this evidence, together with a set of equity pricing facts. The model suggests that about two- thirds of the fluctuations of stock price-dividend ratios are driven by shifting investors’ expectations as a consequence of the dynamic interaction between the sentiment shocks and investors’ learning behavior.

Using survey forecasts data, the second chapter (paper) studies if professional fore- casters utilize long-run co-integration relationships among macroeconomic variables to forecast future as postulated in workhorse stochastic growth models. There exists a significant heterogeneity among forecasters, the majority of whom do not use these long- run relationships and generally make more accurate forecasts (comparing with those who use). We identify potential causes for the survey evidence and discuss its implications. Simple parsimonious recursive forecasting models are fitted to the data as one way to approximate the expectation formation process of the forecasters.

The third chapter (paper) bridges the econometrics literature on macroeconomic de- trending (or output gap estimation) and analysis of monetary policy and welfare. It analyzes the welfare properties of the combined choices of a detrending method from eleven methods and interest rates. Detrending interacts with policy decisions and macroeconomic outcomes, which has important welfare implications. The paper offers novel and significantly different perspectives on the desirability of the detrending methods from the literature. For each detrending method, we characterize optimized simple interest rules and compute the associated welfare losses. Moreover, the reliability of various detrending methods is evaluated with a crucial difference from a common practice in existing literature. These issues are analyzed under alternative assumptions on the underlying trend and alternative policies, such as a higher inflation target, dual man- date, average inflation targeting and negative interest rates.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):