Does Policy Committee Diversity Affect Public Trust and Expectations?

2020-08-10

2023-06-30

Main survey: expectations of future inflation and unemployment; trust in the Fed (to adequately manage inflation/unemployment; and to ensure the economic well-being of all Americans); plans for personal spending and labor market behavior.
Follow-up survey: expectations of future inflation and unemployment; trust in the Fed (to adequately manage inflation/unemployment; and to ensure the economic well-being of all Americans);; whether read any newspaper articles about Fed over previous month; whether (within survey) choose to read article featuring Fed policy maker (randomized to be either male or female) vs. policy maker from a different organization; how they evaluate article and featured policy maker.

The experimental intervention consists of three surveys.

The first survey begins by eliciting respondents’ knowledge and beliefs about the economy. Then, each respondent is randomly allocated to one of 7 groups (with equal probability). The first group is a control group that is not shown any information about Fed forecasts (just some generic Fed-related information). The remaining groups are shown forecasts about either inflation (groups 2-4) or unemployment (groups 5-7) based on official FOMC releases, along with interventions that make more or less salient the extent of minority representation in the FOMC. All respondents are then asked about their trust in the Fed and their economic expectations are elicited again. We also elicit various demographic characteristics.
The main hypotheses we’ll test is whether the same forecast information has differential effects on respondents’ beliefs and trust in the Fed depending on whether minority representation on the FOMC is salient, and in particular whether these effects differ depending on respondents’ own demographic characteristics.

In the second survey, we again elicit respondents’ economic expectations as well as trust in the Fed, to test whether any potential effects found in the main survey persist over time. We will further examine whether the treatment in the main survey affected whether respondents followed Fed news over the intervening month, and whether they appear more informed about the Fed. Finally, we randomize respondents into three groups (orthogonal to treatment groups from the main survey). In group 1, respondents choose whether they want to read a short article based on statements by officials of the Congressional Budget Office or the Federal Reserve. In group 2 and 3, we add the names and titles of the policy makers based on whose statements the news articles are written, only one of which belongs to an underrepresented demographic group. The choices in the three groups allow us to test the extent to which the treatment in the main survey relates to subjects’ willingness to learn about the Fed’s statement relative to other institutions, and the extent to which this choice is mediated by the salience of minority representation in the policy committee.

In the third survey, we will replicate the outcomes and demographic characteristics, motivations, and beliefs indicated in the first survey, but we will assess the effects of the salience of minority representation on the FOMC at the intensive margin (number and composition of multiple FOMC members) rather than the extensive margin (whether the only FOMC member shown to subjects belongs to an underrepresented group. To this aim, subjects will be exposed to one of 8 experimental conditions. The first condition only reports generic information about the Fed, alongside the Fed's logo. The second condition reports the Fed log alongside the March 2023 economic forecasts (inflation and unemployment rate for 2023 and 2024) approved at the March 2023 FOMC meeting. All other experimental conditions report the same forecasts alongside a set of 5 FOMC members' profile pictures and names among those who took part in the March 2023 meeting. Specifically, in condition 3, the group includes 5 white men. In condition 4, three white men and two white women. In condition 5, three white men and two Black men. In condition 6, three white individuals (two men and a woman) and two Black individuals (one man and one woman). In condition 7, four Black individuals (two men and two women) and one white man. In condition 8, the same composition as in 7 but placing the white men, Fed's Chairman Jerome Powell, at the center of the group and emphasizing that he is the Chair.

The main survey begins by eliciting respondents’ knowledge about the Federal Reserve, and by eliciting their prior expectations about future inflation and unemployment. Then, each respondent is randomly allocated to one of 7 groups (with equal probability). The first group is a control group that is not shown any information about Fed forecasts (just some generic Fed-related information). The remaining groups are shown either median forecasts from the most recent FOMC Survey of Economic Projections about inflation (groups 2-4) or unemployment (groups 5-7), along with a picture of an FOMC member that is a white male (groups 2 and 5), a white female (groups 3 and 6), or an African-American male (groups 4 and 7). All respondents are then asked about their trust in the Fed (both in terms of whether they trust the Fed to adequately manage inflation and unemployment, and to ensure the economic well-being of all Americans), and their expectations about future inflation and unemployment are re-elicited. We also elicit various demographic characteristics.
The main hypotheses we’ll test is whether the same forecast information has differential effects on respondents’ beliefs and trust in the Fed depending on whether minority representation on the FOMC (either female or black) is made salient, and in particular whether these effects differ depending on respondents’ own group membership (gender and race).
In the follow-up survey, we again elicit respondent expectations about inflation and unemployment, as well as trust in the Fed, to test whether effects found in the main survey (if any) persist over time. We will further examine whether the treatment in the main survey affected whether respondents followed Fed news over the intervening month, and whether they appear more informed about the Fed. Finally, we randomize respondents into four groups (orthogonal to treatment groups from the main survey). In group 1 and 2, respondents are asked whether they would prefer to read a short article that features a statement about the future of the U.S. economy from a high-ranked official from (i) the Congressional Budget Office, or (ii) the Federal Reserve. In groups 3 and 4, respondents are asked the same question, but with policy maker names added, where in group 3 the Fed policy maker is female, and in group 4 the Fed policy maker is male. The choice in groups 1 and 2 is used to test whether treatment assignment from the main survey affects the likelihood that a respondent selects to read about the Fed. Choices in groups 3 and 4 are instead used to directly test whether female respondents become more likely to read about the Fed when the Fed policy maker is female rather than male. Among those in groups 1 and 2 who choose the Fed article, those in group 1 (2) are shown the article citing the female (male) Fed representative, to test whether this affects their evaluation of the article (differentially depending on respondent characteristics).

Randomization is performed by a computer that assigns incoming subjects to different experimental arms

Individual

No

NA

8750 in the first survey, of which 3,500 will be contacted for the second survey one month later. 10,000 in the third survey.

In the first survey, respondents will be randomly exposed to 1 of 7 treatment arms, i.e. the sample size is (approximately) 1,250 per treatment.
In the second survey, respondents will be randomly exposed to 1 of 4 treatment arms, i.e. sample size (approximately) 875 per treatment
In the third survey, respondents will be randomly exposed to 1 of 8 treatment arms, i.e. the sample size is (approximately) 1,250 per treatment.