Are household inflation expectations in Europe consistent with the Phillips curve?

JEL codes:
D84
E31

Understanding the factors driving variation in inflation expectations is crucial for enhancing macroeconomic models. Household inflation expectations can lead to overinvestment in real assets (Sims, 2009), speculative financial behavior (Nimark, 2012), and increased economic sensitivity to shocks (Badarinza & Buchmann, 2011). We will use the ordinary least squares approach to study heterogeneity in household inflation expectations. Additionally, household expectations significantly influence economic actions and overall growth (Coibion et al., 2018). The Phillips curve is a fundamental concept in macroeconomics and monetary policy, indicating a negative correlation between economic slack and inflation. We will examine if European households' inflation expectations align with the Phillips curve predictions. Using fixed effects models, we will analyze whether household expectations match theoretical predictions, considering static, adaptive, and rational expectations.

Furthermore, we use modified Phillips curve expectations models that account for both observed and anticipated mistakes in inflation expectations. Results from these models show that households exhibit adaptive expectations rather than rational expectations. Finally, a hybrid Phillips curve model including lagged energy and food inflation is employed, as swings in commodity prices may impact expectations. This study contributes to the literature by analyzing survey data to determine if agents' expectations align with economic theory. Previous research on professional forecasters suggests their forecasts align with economic theory models, including the Phillips curve (Fendel et al., 2011; RŌlke, 2012; Casey, 2020; Clements, 2024), as well as the heterogeneity of inflation expectations among European households.

The Consumer Expectations Survey of the European Central Bank (ECB) provides microdata from eleven European countries: Austria, Belgium, France, Germany, Greece, Finland, Italy, Ireland, Netherlands, Portugal, and Spain. The survey began in 2020 and offers monthly data for 49 months, from April 2020 to April 2024. We examine household inflation predictions for the next 12 and 36 months, perceptions of past inflation, and forecasts for joblessness and economic growth over the next 12 months. Each variable is winsorized at one percent in each tail. Demographic variables considered include gender, household size, income bracket, and education level. We also provide ECB statistics on the monthly unemployment rate, food and energy components, and the annual rate change of the Harmonized Index of Consumer Prices (HICP).

Our findings align with previous research. Specifically, middle-aged individuals (50-70 years old), women, those with lower incomes, and less education tend to express higher inflation expectations. These results are consistent with studies by Bryan and Venkatu (2001) and Bruine de Bruin et al. (2010). Additionally, in line with Souleles (2004) and Anderson (2008), women, individuals with lower incomes, and those with less education show larger prediction errors in inflation expectations. In every Phillips curve model, the relationship between inflation and unemployment expectations was positive and did not align with theoretical predictions. This result mirrors findings by Kamdar (2019), Candia (2020), Weber (2023), and Kipson (2024). Examining the hybrid model, households place greater weight on the forward-looking component rather than the lagged inflation component, similar to Clements (2024).