New Indicator of Household Attentiveness and Its Implications

Mention that the countercyclical attentiveness is good for monetary policy transmission – people pay more attention.

We construct individual-level measures of attentiveness to economic news and attitude toward economic conditions based on microdata from the Michigan Survey of Consumers (MSC, henceforth). For this, we exploit the following question from the MSC:

“During the last few months, have you heard of any favorable or unfavorable changes in business conditions? What did you hear?”

Figure 1 reports the indices of attentiveness and attitude. The two indices show quite different dynamics: The attentiveness index is countercyclical but the attitude index is procyclical. Households pay more attention to economic news when macroeconomic conditions deteriorate. The attentive index indicates that their memory of recent macroeconomic events is consistent with the realized cyclical effects of these events on the aggregate economy. This observation suggests that increased exposure to unfavorable economic news is the key driver of countercyclical attentiveness. Consistent with findings in coibion2015information, households endogenously become more attentive to economic news as uncertainty increases during recessions.⁴⁴4ahn2023portfolio documents that, in particular, stock market participants become more attentive during periods of heightened uncertainty.

Next, we explore factors determining households’ attentiveness with the following empirical model:

where the regressor $X_{it}$ includes educational attainment, income, homeownership, stock-market participation, and strong political affiliation. The MSC started to occasionally ask respondents if they would think of themselves as “a Republican, a Democrat, or an Independent” starting in the mid-1980s. Respondents who identify themselves as Republican (Democrat) are followed up with a question on whether they would call themselves as strong Republican (Democrat). On average, about 28 percent of respondents self-identify as Republican while 31 percent as Democrat. Among Republicans (Democrats), about 59 percent (63 percent) would call themselves strong Republican (Democrat). We use time-fixed effects $\alpha_{t}$ to control for business-cycle fluctuations.

The estimation result is reported in Column (1) of Table 1. Individuals who are college graduates, have higher incomes, own homes, and participate in the stock market pay more attention to economic news than others. Last, households with strong political affiliations also show higher attentiveness.

Does high attentiveness represent a better understanding of the current economic situation and lead to more accurate forecasts? In this section, we examine the effects of individual attributes on the accuracy of inflation forecasts and see whether individuals showing high attentiveness produce better projections.⁵⁵5ahn2023portfolio show that stockholders produce more accurate inflation forecasts than non-holders, and ahn2022effects find that homeowners update their expectations on inflation and real activity in a way that is consistent with the intended effects of monetary policy. We use 1-year ahead inflation forecast error which is measured as the absolute difference between 1-year ahead inflation forecast and the realized CPI inflation rate. We focus on inflation forecasts because the numeric responses are recorded so that the accuracy of expectation is quantitatively evaluated, unlike other expectations that are expressed in a qualitative manner (e.g., improve vs. deteriorate). We employ the same specification as Equation (1) but replace the dependent variable with inflation forecast errors.

Column (2) in Table 1 reports the estimation result. On average, those who pay more attention to economic news produce more accurate inflation projections. Specifically, being homeowners, stock-market participants, college graduates and individuals with higher incomes reduces errors in forecasting inflation.

One exception is having strong political affiliations, which deteriorates the accuracy of inflation forecasts despite increased attentiveness. To examine why strong political affiliation leads to biased forecasts, we dive into the inflation forecasts of a sub-sample of attentive households when different parties are in office. Consider the following empirical model:

where $\text{Democrat}_{it}$ and $\text{Republican}_{it}$ are indicators of whether the respondent is self-identified as a strong Democrat or Republican. The baseline is households that do not have strong political affiliations. We control for time-fixed effects and other demographic-fixed effects such as homeownership, stock market participation, educational attainment, and the rank of income.

Table 2 reports the empirical results. The estimated coefficients of $\text{Democrat}_{it}$ and $\text{Republican}_{it}$ represent forecast errors of those affiliated households relative to the baseline group when Republican (in Column (1)) or Democrat (in Column (2)) is incumbent. Households with strong party affiliations make significantly larger forecast errors relative to those who do not when the rival party is incumbent. ⁶⁶6It is interesting to note that Democrats seem to be less over-optimistic when their supporting party is in power.

What explains this bias in inflation projections? One important source is excess pessimism about the economic outlook when the supporting party loses its presidential position. Figure 2 shows the polarization of attitudes between the two groups of different political affiliations. Specifically, Republicans’ attitude toward the economic outlook spiked at the end of 2016 when the Republican candidate (Donald Trump) won the presidential election, while the attitude of Democrats plunged after the election. Similarly, Democrats’ attitudes spiked at the end of 2020 when the Democratic candidate (Joe Biden) won the presidential election, while Republicans’ attitudes deteriorated abruptly. In contrast to the two groups, independents do not show sudden changes in the attitude associated with the presidential election.⁷⁷7This is consistent with the findings of kamdar2022polarized.

This observed excess pessimism has important implications for anchored inflation expectations which is a primary goal of monetary policy. Table 3 provides evidence that attentive households, on average, report lower and more accurate inflation forecasts. However, pessimism – captured as the negative attitude – is associated with higher inflation projections (Column (1)) and larger forecast errors (Column (2)). Specifically, after controlling individual characteristics and business cycle fluctuations, households who recall only negative news make higher inflation projections and larger forecast errors. Excess pessimism as a result of political polarization likely leads to excess volatility in inflation expectations and increased disagreement about the inflation outlook, which likely reflects the weakening of anchored inflation expectations.

To summarize, high attentiveness, on average, improves the accuracy of households’ inflation projections. However, strong political affiliations increase attentiveness but are biased toward polarized economic news, which eventually deteriorates the accuracy of inflation forecasts.

will rewrite this after the other parts are done This research has implications for monetary policy. Our finding shows that inflation expectations become polarized across individuals with different political affiliations in times of political uncertainty such as the presidential election. As a result, households’ inflation expectations may show excess volatility and individuals may disagree more about the inflation outlook, which can eventually lead to unanchored inflation expectations. As a result, political uncertainty can eventually weaken the effectiveness of monetary policy. In this context, the central bank’s independence and clear communication about policy path are particularly important in times of high political uncertainty, as they help prevent the spillover of political uncertainty to the macroeconomy.