Panel Conditioning in the Survey of Consumer Expectations

Abstract

Panel conditioning occurs when participation in previous survey rounds affects how respondents answer questions in later rounds. I document panel conditioning effects in reported inflation expectations and other responses in the Federal Reserve Bank of New York Survey of Consumer Expectations. Even controlling for attrition effects, inflation point forecasts and uncertainty decline gradually over a respondent's first few survey rounds. Point forecasts are more than 3 percentage points higher in a respondent's first round and 2 percentage points higher in her second round than in her final round, and revisions are more common between earlier rounds. I provide several examples of how panel conditioning can influence results when using the survey data to monitor expectations, study expectations formation, or study the link between expectations and behavior.

Keywords: panel conditioning, consumer expectations, forecasts, inflation expectations, survey data

JEL Classification: D80, D83, D84, E31