Uncovering Subjective Models from Survey Expectations

Expectations about different macroeconomic aspects correlate with each other. I perform a structural test in the framework of the noisy information model and show that individual forms their expectations on multiple macroeconomic variables jointly rather than independently, thus causing these expectations to be correlated with each other. In particular, they have a subjective model of the economy. They believe economic conditions will be worse during episodes 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. To alleviate the concern of possible mis-specification of the linear noisy information model, I then propose an innovative generic learning model that can cover a large class of expectation formation models, including those that are standard in the literature. The effect of signals on expectational variables is estimated with Recurrent Neural Network. I found again realized inflation increases household's perceived future unemployment rate change whereas the actual unemployment rate hike will lower their expected inflation. The pessimistic effect of inflation is not because of household's belief on the interest rate and is particularly strong after late 1990s. 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 caution, as such policy may induce pessimistic responses among households