It is well-documented that relatively few women manage investment funds. In 2019, for example, women accounted for 37.5 percent of all lawyers, 49 percent of judges, 34.5 percent of economists, 19 percent of surgeons, and 26 percent of chief executives, according to the U.S. Census Bureau. In contrast, the percentage of funds managed by women is not only low, it has barely changed: It was 10.3 percent in 2016 and 11 percent in 2020.

While there are several explanations for the employment gap between men and women in various industries, financial economists have argued that investors in mutual funds discriminate against women, investing far less in female-managed mutual funds than in male-managed funds. In response, rational fund companies might hire fewer women, given that the companies generate their profits from fees charged on assets under management.

In a new paper, “Do Investors Pay Less Attention to Women (Fund Managers),” we document a different and previously unstudied type of gender bias in investor behavior, which we term gender-based attention bias. Gender-based attention bias refers to the tendency to pay less attention to women than men.

Examining gender bias in corporate finance is challenging because it is difficult to establish how much performance depends on an individual executive’s skill and effort – especially when so many executives contribute to a firm’s value. In contrast, in mutual funds with a single manager, it is relatively straightforward to attribute performance to managerial effort and to relate investment flows (rewards) to performance (effort).

In our paper, we examine whether gender-based attention bias affects the well-documented investment flow-performance relationship in the mutual fund literature. Previous studies have found that there is a positive correlation between prior mutual fund performance and subsequent fund flow, commonly termed the flow-performance sensitivity of the fund. It has, however, been difficult to examine whether investor fund flows are affected by gender bias because of a lack of data on investor-level fund flows. Consequently, the flow-performance literature typically focuses on aggregate fund-level flows, making it difficult to isolate individual investors’ different responses to fund managers’ gender.

In contrast to the previous literature, we examine whether the flow-performance sensitivity for individual investors is affected by managerial gender, using a unique dataset provided by a large fintech platform in China. The fintech platform allows users to make and receive payments to and from other people and businesses and also provides in-app access to a variety of saving and investment products with different risk levels. The entry barrier to investing over the platform is low, with the typical minimum investment amount being 10 renminbi (equivalent to approximately $1.50). The investment products offered on the platform include open-ended mutual funds managed by fund managers affiliated with fund management companies independent of the platform. The fintech platform does not have in-house fund managers itself. It only serves as a portal to fund investments with significantly lower transaction fees than traditional brokers. The app allows the investors to rank funds on the basis of raw returns over the prior one-, three-, six,- and 12-month return horizons. Importantly, it provides clear information on the gender of the fund managers (including a photograph in a large number of cases) to the investors at the time of investment. Prior papers that have examined the effect of gender on fund flows are unable to establish that fund investors are even aware of who is managing their fund. In China, this information is almost the first piece of information investors receive when choosing to invest on the app.

Our data are based on a random sample of investors drawn from the platform and consisting of 172,672 retail investors’ monthly investment positions in 253 domestic stock funds from August 2017 to July 2019, for a total of 2.35 million user-fund-month observations. The database contains monthly data on each investor’s fund investment and redemption amounts and details of capital gains or losses for every fund owned by the investor in that month. It also contains characteristics of each investor on the platform, such as gender, age, monthly payment amounts, places of residency, and risk aversion levels (surveyed through a questionnaire upon users’ registration with the platform).

We document strong evidence of a differential flow-performance sensitivity for male- and female-managed funds, respectively. We show that flow-performance sensitivity is significantly weaker for female-managed funds. We document decreased flow sensitivity to performance for women across all the raw return horizons, from one- to 12-months, that the fund allows the investor to sort over. Simply put, when female-managed funds do well, they experience significantly lower fund inflows than do male-managed funds. However, when these funds perform poorly, they experience relatively lower fund outflows than male-managed funds.

Our results are robust to controlling for fixed effects at the investor level, allowing us to address potentially omitted variable concerns arising from differing investor backgrounds or personalities. The results are also robust to including time-fixed effects, reducing a potential omitted variable bias caused by common economic shocks. All the regressions also control for manager characteristics, such as their educational backgrounds and their length of tenure as managers of the funds, and for fund characteristics, such as fund objectives, fund size, and fees.

An alternative explanation for the muted flow-performance sensitivity for female managers is that male managers’ past performance is a better predictor of their future performance. Hence investors rationally invest in what they believe will be better-performing funds. We show, however, that past performance does not predict future performance for Chinese fund managers and, moreover, that there is no gender difference in the predictive ability for future performance.

The Chinese sample of fund managers is similar to the data studied in other countries in that only 15 percent of the sample of managers are female. Hence, we investigate whether the attention bias arises because investors face higher search costs when searching for female fund managers. For example, an investor who is actually gender-neutral towards the choice of fund manager might appear biased because of the high search costs involved in finding female managers. To address the visibility problem, we use two approaches. First, we match each female manager to a similar male manager in each month using a propensity-score matching (PSM) approach that uses a host of managerial and fund characteristics as covariates. Although PSM does not allow us to establish causality, the covariates of managerial and fund characteristics are well balanced in the matched sample. Attention bias continues to be significant in this matched fund sample, where search costs for male and female managers are likely to be approximately similar.

Second, we use a Natural Language Processing (NLP) technique to extract names from 400,000 financial news articles in Chinese and count how frequently each manager’s name appears in the article each month. We show that the level of attention bias is unaffected by the level of media coverage. Although investors do respond positively to media coverage of male managers, for similar levels of media coverage, the flow response is less sensitive for female than male managers.

We next show that there are no statistically significant cross-sectional differences in attention bias among investors with different gender, age, and risk-aversion levels and who live in cities of different population sizes. Attention bias appears significant across all our investor groups. Similarly, we show that attention bias appears to be innate to investors. In a sub-sample of all first-time investors, we show that, at the time of the first investment, investors are less likely to invest in female-managed funds for the same level of performance.

To examine if there is a causal relationship between the gender of female managers and investor fund flows, we employ a three-stage instrumental variable regression. We use two different instrumental variables, the first being the proportion of illiterate women among all women in the municipal district where the investor resides, and the second being the proportion of female new-borns among all new-borns in the municipal district where the investor resides. Both instrumental variables do not directly drive investors’ fund flow decisions but are likely to be related to investors’ gender biases, conditional on investor characteristics that we control for. Our instrumental variable regressions confirm the existence of gender-based attention biases away from female managers, which cause investors to pay less attention to female-managed funds.

With its unique dataset matching user-level flows to specific funds, our paper is the first to document the existence of a gender-based attention bias away from women in the professional finance industry. As we noted earlier, previous researchers have proposed a customer-based discrimination explanation for the lack of women in the mutual fund industry. We show that, in contrast, the attention bias works both ways. Though investors appear more sensitive to fund performance when the fund manager is male, the sensitivity is bi-directional. Investors are less sensitive to underperforming female managers. For mutual fund companies, this may lower the volatility of flows into the fund, increasing the attractiveness of women to the industry.

This post comes to us from P. Raghavendra Rau, the Sir Evelyn de Rothschild Professor of Finance at the University of Cambridge, and from Jinhua Wang, a PhD candidate at the University of Cambridge. It is based on their recent paper, “Do Investors Pay Less Attention to Women (Fund Managers)?” available here.