- A new NBER working paper shows that existing US surveys of professional forecasters and households poorly proxy firms’ inflation expectations.
- Their expectations also appear unanchored, much like households’, displaying significant cross-sectional disagreement and large quarter-on-quarter revisions.
- Despite recent research showing inflation expectations matter for firms’ employment and investment decisions, very few are aware of the Fed’s inflation target or inflation over the last 12 months.
Most understand that if individuals expect higher inflation, prices will follow suit. This self-fulfilling prophecy comes from firms, households and banks all making forward-looking decisions based on where they believe prices will be. A new NBER working paper analyses the results from a new US firm-based survey of inflation expectations. It finds that:
- On average, firms’ inflation expectations are higher than professional forecasters’, and they follow similar dynamics to household expectations.
- There is widespread disagreement among firms about short-term and long-term inflation forecasts, which increases in periods of greater economic uncertainty.
- Most firms (65%) are ignorant of the Fed’s inflation target. For those that think they know, the average implied a target rate of 2.9%.
- Chief executive officers (CEOs) are also unlikely to know the current inflation rate, despite around 40% stating that aggregate inflation matters for their pricing decisions.
Inflation expectations are arguably the most important input into central bankers’ current and future inflation forecasts. However, being unobservable makes them notoriously hard to measure. There are two approaches to addressing this. The first is a model-based measure, the second a survey-based measure. Models rely on estimation and typically include financial variables and inflation series. Surveys directly elicit expectations from respondents.
While several inflation expectation measures exist, the NBER paper’s authors point out an absence of systematic survey data on the macroeconomic expectations of US firms. Those that exist tend to focus on expectations about firm-specific outcomes (e.g., firm-level uncertainty or costs). Those that extract firm-level expectations about aggregate conditions are either very small/non-representative (Livingstone Survey) or qualitative (Duke CFO Survey).
Survey of Firms’ Inflation Expectations (SoFIE)
The authors introduce a new US-based survey, the SoFIE, to measure the inflation expectations of firms. The survey started in 2018, runs quarterly, and builds upon a pre-existing privately run survey of CEOs. Each wave contains an average of 300-600 firms, most (40%) of which are small companies with 1-19 employees. Around 35% of firms participate only once, and the average participation is 3.3 waves.
In the survey, respondents answer two quantitative questions about inflation and monetary policy. One consistently measures their quantitative expectations about US inflation over the next 12 months. The second rotates across different formulations investigating long-run inflation expectations, perceived recent inflation, uncertainty about future inflation risk, and knowledge of the Fed’s inflation target. The survey is unique for these latter formulations and for measuring expectations of aggregate inflation. The authors build a case for including this survey because firms, as opposed to households and professionals, are central to aggregate inflation beliefs.
A Case for Measuring Firms’ Inflation Beliefs
The authors compare the one-year ahead inflation expectations from the SoFIE survey against the Michigan Survey of Consumers, the Survey of Professional Forecasters and the Cleveland Fed (Chart 1). This comparison shows firms’ inflation expectations deviate significantly from households’ and professional forecasters’, suggesting the latter two are no substitute for a representative survey of firms’ beliefs.
For example, during the pandemic, professional forecasters significantly reduced their inflation expectations. Households, in contrast, immediately raised their inflation expectations from 2.9% in 2020 Q1 to 4.0% in 2020 Q2, and these have remained high through 2021 Q1. Meanwhile, firms initially displayed little change in their inflation expectations but ultimately raised them to 2.8% by 2021 Q1 and 3.2% by 2021 Q2.
Interestingly, the deviation of firms’ expectations from professional forecasters’ increases when the willingness of managers to provide forecasts in the survey decreases. This finding suggests that in periods of greater uncertainty, or when firms are paying less attention to inflation dynamics, professional forecasts are less representative of firms’ expectations.
Firms’ mean forecasts are generally lower than households’, but higher than professionals’ (Chart 1). This is also true for the standard deviation (or level of disagreement) among firms. The level of disagreement is around 1.4pp on the mean during the sample – higher than for professionals (1.1pp) but lower than households (3.3pp). Also, all firms are equally likely to disagree, regardless of size and sector.
Are Inflation Expectations Anchored?
For inflation expectations to be anchored, (i) average beliefs should be close to the central bank’s inflation target; (ii) beliefs should not be too dispersed across agents; (iii) agents should be confident in their forecasts; (iv) agents should display small forecast revisions, especially at longer horizons; and (v) there should be little co-movement between revisions in long-run and short-run inflation expectations.
On the first point, firms’ inflation expectations deviated significantly from the Fed’s 2% target at the beginning and end of the limited three-year sample (Chart 1). For long-run (five-year ahead) forecasts, firms’ expectations were similar to households’, whose long-run forecasts were consistently just under 3% over this period. So, by this metric, firms seem to have poorly anchored inflation expectations.
For the second point, the cross-sectional dispersion of forecasts must be low. For example, the dispersion of long-run forecasts among FOMC members is zero: they all agree that inflation will be 2% in the long run. For professional forecasters, the dispersion is around 0.2-0.3pp, and for households it is 2.5pp. The dispersion in firms’ forecasts ranges from 1-2ppt, again suggesting expectations are unanchored. Greater dispersion is also related to lower confidence regarding beliefs around future inflation, required for the third point.
The fourth point requires revisions in individuals’ inflation forecasts to be small, since agents expect the central bank to be able to keep inflation stable over long enough horizons. Plotting the distributions of revisions of one-year-ahead (beige) and five-year-ahead (black outline) forecasts, we can see that firms display very large revisions (Chart 2). On average, firms revise long-term inflation expectations by 1.4%, compared with 0.2% for professional forecasters.
Lastly, a commonly used approach to assess whether expectations are anchored is to examine the co-movement of changes in short-run and long-run inflation expectations. The idea is that transitory economic shocks can affect short-run expectations but not long-run ones when the latter are anchored. On this front, the authors find a strong positive relationship between managers’ long-run and short-run inflation expectations, indicating the prediction is wrong.
Inattention to Monetary Policy
If expectations are unanchored, firms are either unaware of central bank inflation targeting or do not consider the central bank credible and so use the latest inflation print for forecasting instead. On this front, the authors find that long-run expectations are related to (incorrect) beliefs about the inflation target, while short-run expectations are better explained by perceptions of recent inflation.
Specifically, the average firm believed the Fed was trying to achieve an inflation rate of 2.9% in the long run. This aligns with the mean five-year ahead inflation forecast in the sample. However, 65% of firms answered that they did not know. This share has actually fallen during the pandemic as inflation concerns garner significant media attention.
When asked about beliefs regarding inflation over the previous 12 months, very few firms could provide the right answer (Chart 3). The inaccuracy was more noticeable than for the inflation target. Nevertheless, the authors find that 12-month ahead forecasts are more closely tied to what respondents believed inflation was over the last 12 months.
The results suggest that in the now-long history of low and stable inflation, firms are paying less attention to central bank communications and are likely to be unaware of the current inflation rate. This is surprising given 38% of firms responding to an Atlanta Fed survey indicated that aggregate inflation influenced their pricing decisions. Also, recent studies in Italy and New Zealand find that firms’ inflation expectations influence employment and investment decisions.
The inherently forward-looking nature of most firms’ decisions, such as price setting, employment, and capital expenditures, implies a key role for firms’ expectations in future inflation. This research reveals that managers are ill-informed about central bank inflation targets and unaware of current inflation. Perhaps most concerning is that short-term inflation expectations, which are most likely to feed into business decisions, are unanchored and positively influenced by perceptions of recent inflation. If the Fed is unable to communicate clearly that higher inflation prints are temporary, business owners could adjust their expectations upwards leading to more permanent increases in inflation.
Candia B., Et Al., (2021), The Inflation Expectations of US Firms: Evidence from a new survey, NBER, Working Paper (28836), https://www.nber.org/papers/w28836