The Gender Optimism Gap
Men are significantly and persistently more optimistic than women about the economic outlook
Summary
We document a significant and persistent gap in optimism about the economy between men and women in the United States from 1978 to the present. On average, men are roughly 10 index points more optimistic than women, compared to an average index value of 84 over the period. In only 13 out of the 570 months since 1978 have women expressed more optimistic views than men. The gap does not exhibit noticeable time trends or cycle variation. However, it is larger when there is a Republican in the White House and smaller when there is a Democrat in control. Only a small share of the gap is explained by available control variables.
The Gender Optimism Gap in Consumer Sentiment
The University of Michigan’s Index of Consumer Sentiment (ICS) is based on a nationally representative survey of approximately 600 Americans. Each month, the survey asks respondents questions about their current views and future expectations for the economy. The ICS is a diffusion index based on these responses, with scores above 100 indicating that more respondents have positive than negative views of the economy and scores below 100 indicating the opposite.
Michigan releases respondent-level data, allowing researchers to observe how different groups of consumers view the economy. We use this data to construct consumer sentiment series for men and women.1
As the chart above shows, men have persistently held more positive views on the economy than women. Over the 1978-2025 period, the average gap in consumer sentiment between men and women is 9.9 index points compared to an average index value of 84.4. In only 13 out of the 570 months since 1978 have women expressed more optimistic views than men. The gender gaps in the indexes of current economic conditions and expected future conditions are similar, at 8.4 and 10.9 index points, respectively.2
The figure below shows the gap in levels (men-women) and as a proportion of overall sentiment (men-women/total). The gap in overall sentiment, as well as in current conditions and expectations, does not exhibit noticeable time trends or cyclical variation. However, since at least 2000, the gap has been sharply larger during times when a Republican was in the White House and lower when a Democrat was in the Oval Office. This reflects the combined effects of a gender gap in partisanship and the strong effects of partisanship on consumer sentiment: Men are more likely to be Republicans, and Republicans are more optimistic when there is a Republican president, expanding the gap when there is a Republican in charge, and shrinking it when a Democrat is in control.
Available Controls Explain Only A Small Part of the Gap
To investigate the extent to which controls explain the gap, we run regressions of overall consumer sentiment on a female indicator and available control variables. The table below shows these results, using different time windows based on when the control variables were available.
The available controls explain little of the gender optimism gap. For instance, the gap only drops from 9.8 to 8.4 when we control for demographics, income, and political affiliation in the 2017-2025 sample, where we have the richest set of controls (columns 4-7). The results are similar when we use data from 1990-2025, where we have all but the political control variables (columns 2-3). The table also shows the well-known partisanship effect, which we have investigated in more depth in other work.3
What Are the Potential Explanations?
In email correspondence, Alice Evans of King's College London, and the author of the upcoming book The Great Gender Divergence, raised three possible explanations for the gap. First, women are over-represented in lower-paid sectors, which may make them more concerned about economic volatility. But this wouldn't explain why the gap persists despite women's economic rise. Second, women tend to consume different media, which may reinforce negativity. If this were true, smartphones would have increased polarisation, yet this does not seem true for consumer sentiments. Finally, Evans noted that women tend to score higher on measures of anxiety and empathy, whereas men tend to be much more confident. Differences in personality (whether via nature or nurture) may shape how we perceive economic flux. While these theories seem plausible to us, the data used in this analysis does not allow us to explore them in further detail. Understanding what causes the gap and its implications for economic behavior are intriguing questions for future research.
The University of Michigan releases sentiment indices by gender, but these are on a three-month moving average basis. For consistency, we utilize the monthly gap in sentiment constructed from the microdata.
These numbers are consistent with the findings of Jacobsen et al (2014) and Majumder and Gallagher Robbins (2024), who both use the ICS data to examine the gender optimism gap. Additionally, Morning Consult, a global survey research firm, also produces a weekly consumer sentiment index and shows a gender gap of a similar magnitude over their time sample (2018-present). International evidence further suggests that the gender optimism gaps exist across countries: Dawson (2023) finds that women are more risk-averse and less optimistic than men in U.K. survey data; using data from Sweden, Bjuggren and Elert (2019) show men are more optimistic than women but are also more likely to be wrong about the future state of the economy.
In our other work, using data through November 2023, we found a total partisanship effect between Democrats and Republicans of 20 index points. The effect here is 35 points, which may reflect growing partisanship over the last 2 years and the additional covariates in the regression.







Nice analysis, mates!
Why wouldn't the explanation be that women have long faced gender wage/inc/occ gaps? (Even w occupational upgrading, there are still within occ gender wage gaps.) Hard to control for this in monthly data but you could put in such a control (using Census ASEC data) for yearly data. Fewer obs, of course, so you'd have to cut down on the IVs, but maybe worth going there.