jQuery slideToggle Effect

Working Papers

Learning and Subjective Expectation Formation: A Recurrent Neural Network Approach (JMP)

In this paper, I propose a flexible non-parametric method using Recurrent Neural Networks(RNN) to estimate the dynamic structure of most expectation formation models in macroeconomics. This approach does not rely on restrictive assumptions of functional forms and parametric methods but nests the standard approaches of empirical studies on expectation formation. Applying this approach to data on macroeconomic expectations from the Michigan Survey of Consumers(MSC) and a rich set of signals available to U.S. households, I find qualitatively new results: (1) agents' expectations about the future economic condition have asymmetric and non-linear responses to signals; (2) agents' attentions shift from signals about the current state to signals about future: they behave as if they were adaptive learners in ordinary periods and become forward-looking as the state of economy gets worse; (3) the content of signals on economic condition, rather than the volume of these signals, plays the most important role in creating the attention-shift. My method also allows me to apply the Double Machine Learning method to assess the statistical significance of these empirical findings. Finally, I show these stylized facts can be generated by a model with rational inattention, in which information endogenously becomes more valuable when economic status gets worse.

Monetary Policy when the Phillips Curve is locally quite flat (with Paul Beaudry and Franck Portier, under revision)

This papers examines an environment where the Phillips curve is locally quite flat and in which a cost channel of monetary policy may also be at play. The aim of the paper is two fold. First, we highlight conditions under which a relatively flat Phillips Curve can invalidate standard prescriptions regarding how monetary policy should be conducted to control inflation. In particular, we discuss the desirability of a "Go Big or Stay Home" principle for monetary policy under a certain parameter configuration. This principle refers to the desirability of either responding to shocks in a very aggressive way or not at all, if one aims is to maintain inflation close to target. In the second part of the paper we explore the empirical relevance of the parameter configuration that would favor this principle. To this end, we provide structural estimates from full model estimation as well as provide evidence based on more direct estimation of the Phillips curve. The results from both exercises give substantial support to the configuration of interest. We use the framework to offer an explanation to recent puzzles associated the behavior of inflation including the possibility of falling into a low inflation trap.

Uncovering Subjective Models from Survey Expectations

Expectations about different macroeconomic aspects correlate with each other. Using Michigan Survey of Consumers (MSC), I found consumers' inflation expectation is positively correlated with expectations on unemployment status. Such a correlation is inconsistent with realized data, professionals' belief, and the standard New Keynesian Model. I then perform a structural test in the framework of noisy information model and show that consumers form their expectations on multiple macroeconomic variables jointly rather than independently, thus causing these expectations to be correlated with each other. These results imply the consumers have a subjective model about how macroeconomics variables are correlated that is different from the professionals and the reality. In particular, consumers believe economic conditions will be worse during episode with extensive inflation news, even if there's only mild inflation, causing their average expectation on inflation to co-move with that of unemployment and business condition. These patterns call for explanations on how agents form beliefs on interactions between macroeconomic variables that are different from the actual structure of data. They also suggest Central Bank should use inflation-related expectation management policy with cautious, as such policy may induce pessimistic responses among households.

Convergence Across Castes (with Viktoria Hnatkovska and Amartya Lahiri, submitted)

India has witnessed a remarkable catch-up by the historically disadvantaged scheduled castes and tribes (SC/STs) towards non-SC/ST levels in their education attainment levels, occupation choices as well as wages during the period 1983-2012. Using a heterogenous agent, multi-sector model we show that sectoral productivity growth during this period can explain 75 percent of the observed wage convergence between the castes. Inter-sectoral net flows of workers are key as they account for 3/4 of the predicted convergence. Absent these net flows, the caste wage gaps would have marginally widened. Selection effects, while present in the model, account for just a quarter of the predicted wage convergence. We also find that affirmative action policies that reduced skilling costs for SC/STs may have reduced the levels of the caste wage gaps at all times but played a limited role in accounting for the dynamics of the wage gap. Growth was key for the dynamic wage convergence.

Work in Progress

Learning Pandemics: the Informational Content of Testing (with Davide Alonzo)
Bounded Rationality in Planning (Joint with Giovanni Gallipoli, Wei Li and Jesse Perla)
Disagreement about Inflation Forecast and its Impact on Business Cycle