Does it matter that households disagree about inflation?

Central banks are increasingly concerned about rising inflation expectations among consumers, but not all consumers are alike: there is a wide spread of different expectations across households. In a new paper (Macaulay, 2022), I show that heterogeneity in the processes used to form these expectations influences how shocks affect the aggregate economy. Applying this to inflation in the UK, I find that features of expectation formation imply high inflation can become entrenched in household expectations. However, this is only a risk among a certain subset of households, with particular beliefs about the economy.

Forming an expectation involves taking some information, and passing that through a model to map from the information to the variable being forecasted. A simple process for inflation expectations, for example, would take observations of prices today (information) and project them to the future using a rule of thumb (model). When we see different expectations across households, that could therefore be because they have different information, or they could be using different models, or it could be both [1].

Disagreement will affect the transmission of macroeconomic shocks if these components of expectations are correlated with each other. Intuitively, a shock will have a larger effect if the people who get the most information about it are also the people whose models say that shock is important for the variables that matter to them - because in that case, the information is concentrated among those who will react most strongly to it.

Expectations Heterogeneity in the UK

I examine this effect in the UK using unique features of the Bank of England Inflation Attitudes Survey to measure (i) household information about inflation, and (ii) their subjective models of how inflation affects the wider economy. Two key facts emerge from that data.

First, households who believe inflation makes little difference to the economy are systematically less likely to make use of information about inflation than other households (Figure 1). Second, more households believe inflation weakens the economy when current inflation is high (Figure 2). That fact is also present within a period: those who believe inflation weakens the economy perceive that current inflation is 59 b.p. higher on average than other households in the same quarter.

The Effects of an Inflation Shock

To explore the consequences of this, I simulate the effects of a one-off spike in inflation in a model built to match this data. The model features households choosing how much information to acquire about inflation based on their subjective models (to match Figure 1). If a household believes inflation is currently high, they update their subjective models to be more negative about the consequences of inflation (to match Figure 2).

Figure 3 shows average perceived inflation after the simulated shock, for two groups: those who start out believing inflation makes the economy weaker (black) and stronger (blue). When inflation rises, both groups observe some information about it, so their perceived inflation rises and their subjective models get more negative. The ‘weaker’ group now believe inflation is even more damaging to the economy, so it is even more important to them to process information about inflation. When the shock dies away, these households therefore observe that very precisely, so their perceptions of inflation fall quickly back to target.

The ’stronger’ group, however, have updated from believing inflation makes the economy stronger, to on balance believing that inflation makes little difference. They now believe there is less need to pay attention to inflation, and so when inflation falls they don’t observe it. They therefore keep on perceiving – and expecting – high inflation, long after the shock has disappeared. Consistent with this, in the survey data we see that among those who believe inflation makes the economy weaker, higher perceived inflation is associated with more processing of inflation information. That correlation is reversed among the other households.

Policy Implications

Policymakers are therefore right to worry about recent high levels of inflation becoming entrenched in expectations – among the minority of households who believe inflation makes the economy stronger.

That insight matters for two reasons. First, it tells us whose expectations to track most closely. But second, it implies that if high inflation does become entrenched in their expectations, that will affect the way the economy reacts to shocks going forward. High expected inflation leads a group of previously well-informed people who believed inflation makes the economy stronger to become uninformed. Information therefore becomes even more concentrated among those with negative models of the effects of inflation, meaning that aggregate consumption will react even more negatively to inflation shocks in the future.

At the time of writing, however, that cat is not yet out of the bag. In the Inflation Attitudes Survey, average one-year inflation expectations rose by 1.9 percentage points between August 2021 and February 2022. Among the people who say inflation makes the economy stronger, however, expectations rose by less than 1 percentage point. It is not too late, therefore, to keep inflation expectations anchored and avoid the consequences analysed above.

Alistair Macaulay

alistair.macaulay@economics.ox.ac.uk

https://sites.google.com/site/alistairmacaulayecon

+44 7515 471376

Notes

[1] Recent evidence suggests that there is indeed substantial heterogeneity across households in both of these components of expectations (Beutel and Weber, 2021; Macaulay and Moberly, 2022).

References

Beutel, J. and Weber, M. (2021) “Beliefs and Portfolios: Causal Evidence“, Working Paper

Macaulay, A. (2022) “Heterogeneous information, subjective model beliefs, and the time-varying transmission of shocks”, CESifo Working Paper, no. 9733

Macaulay, A. and Moberly, J. (2022) “Heterogeneity in imperfect inflation expectations: theory and evidence from a novel survey”, Department of Economics Discussion Paper Series, University of Oxford