Prices are, at least in part, what consumers expect them to be. This is why monetary authorities are keeping a close eye on inflation expectations, which affect consumer behavior today.

However, little is known about how consumer expectations are formed. Central banks tend to focus their attention on professional analysts and financial markets rather than households, because economists tend to assume that household inflation expectations are well defined (That is to say, they do not vary in response to short-term developments).

Yet, when they surveyed 25,000 americans in 2018 what they believed the average inflation rate in the United States to be, less than 20% of respondents answered "about 2%." Nearly 40% indicated a figure above 10% (Coibion, Gorodnichenko, and Weber, 2022).

Not only are most households' expectations not well defined, but they also tend to overestimate future inflation.

We used data from the New York Fed's survey of consumer expectations and found that, between 2011 and 2018, men expected inflation to increase by about, on average, 4% over 12 months, while women expected a rate of 6% (a difference that holds regardless of the level of financial literacy).

This suggests that there may be a "gender gap" in inflation expectations. To confirm whether this is the case, we spoke to heads of households, both men and women, responsible for making purchases for their households.

On average, women expect higher inflation than men, but this is true only in "traditional households" where women do the shopping. In households where the man does the shopping, the difference disappears.

To better understand how exposure to price changes influences expectations, we conducted another survey in which we asked participants directly which sources of information they consider most important when estimating inflation (D'Acunto et al., 2021). It turns out that households ranked grocery shopping as the most relevant source of information.

To test this again, we use data from the 50,000 households participating in the NielsenIQ Homescan panel. Using information on what households buy, where they shop and how much they pay, we constructed a household-specific price index and found that, on average, households most affected by inflation expected an inflation rate 0.7 percentage points higher than other households.

However, not all price changes are equally important. If they affect product categories that are important to consumers, or foods that they consume regularly such as milk and eggs, immediate increases in headline inflation expectations are observed in times of both high and low inflation. Households also tend to pay more attention to price increases than decreases.

These factors explain why households changed their inflation expectations in the summer of 2021, when most central banks were still predicting temporary inflationary pressures: prices rose in the categories that consumers cared most about. More importantly, these findings imply that even if central banks manage to contain inflation in the short term, it will take a while for household inflation expectations to come back down.

The simpler the better

There is another factor contributing to household inflation expectations: communication. The more complex policies are, the more difficult they are to explain and, therefore, the less likely they are to influence expectations.

In D'Acunto et al. (2020), we compare the impact of announcements of future consumption tax increases with expectations targeting (a statement indicating the likely future stance of monetary policy). According to the prism of the New-Keynesian model, both policies should have the same effect on inflation expectations. However, they differ quite significantly in their complexity and in the required understanding of the economy.

This is borne out by the data. We use the German version of the European Commission's consumer survey and find that Germans changed their inflation expectations and spending plans only after then-Chancellor Angela Merkel's announcement in november 2005 that consumption taxes would increase by 3 percentage points in january 2007.

In contrast, the announcement in the summer of 2013 by then European Central Bank (ECB) President Mario Draghi that interest rates would remain at current levels or decline (the first time the ECB explicitly used expectations targeting as a policy tool) had no impact on German households' inflation expectations or their spending patterns.

Given these findings, we conducted a series of surveys on how central banks could communicate their policies more effectively. For example, we asked thousands of people in Finland (D'Acunto et al., 2020) about their expectations that their incomes would change and about their sociodemographic characteristics, and then divided the sample into three groups: a control group that did not receive new information and two groups that did receive new information. To these two groups we provided truthful information about the policy measures taken by the ECB in the spring of 2020, for which we used messages from the official Twitter account of Olli Rehn, governor of Finland's central bank. But the content varied depending on the group.

One group received a "target" communication, i.e. a message specifying a policy objective without explaining the measures the central bank would implement to achieve it. Another group received information about the "instrument," the specific policy that was implemented to achieve the objective. All respondents were asked the same questions again. Our results show that only communication specifying the target effectively improves people's income expectations.

In Coibion, Gorodnichenko and Weber (2022), we focus on the medium of message delivery. We find that using simple terms such as "current inflation," "inflation target," or "inflation forecast" is more effective in controlling people's inflation expectations. There is no doubt that the source of information is important: newspaper coverage of the Federal Reserve, while simpler to read, has less impact on expectations than official statements from the Federal Open Market Committee of the Federal Reserve. This is due to U.S. households' assessment of the credibility of various news sources.

When it comes to information about the economy, the average rating of newspapers is low, while that of social media and Twitter is high. These results suggest that central banks cannot rely on the media to convey monetary policy announcements to households.

Translated by: A.M