January 02, 2024

Question design and the gender gap in financial literacy

Anna Tranfaglia, Alicia Lloro, and Ellen Merry

Many surveys have measured people's financial literacy with a standard set of questions covering interest, inflation, and investment diversification. Results from these surveys have consistently shown that women are less likely than men to answer the financial literacy questions correctly – the so-called financial literacy gender gap.

These standard financial literacy questions are typically administered with an explicit "don't know" answer choice. Women are much more likely than men to select "don't know," and this fact largely drives the financial literacy gender gap; men and women generally select incorrect responses at similar rates. If women tend to select "don't know" for reasons unrelated to financial literacy, such as confidence, then the observed financial literacy gender gap may reflect both differences in knowledge as well as other factors. One reason this gender gap matters is that financial literacy is predictive of many financial behaviors like saving for retirement, having three months of liquid savings, and holding stocks and bonds (Bhutta, Blair, and Dettling, 2021; Hogarth and Hilbert, 2002; Lusardi and Mitchell, 2011; Board of Governors of the Federal Reserve System, 2022).

In this note, we leverage a survey experiment in the 2021 Survey of Household Economics and Decisionmaking (SHED) to explore the extent to which women's greater propensity to respond "don't know" contributes to the gender gap in financial literacy. We find that the gender gap in financial literacy shrinks when the "don't know" option is removed. Constructing a counterfactual of what we might expect if people who answered "don't know" guessed the answer to the question instead, we find that for both men and women, removing the "don't know" option increases the share of correct responses beyond what we predict would occur if people were simply guessing. Additionally, removing the "don't know" option increases the share correct by more for women than for men and shrinks the observed financial literacy gap.

SHED Experiment

Since 2013, the SHED has been conducted annually each fall to measure the economic well-being of U.S. households. The 2021 SHED had over 11,000 respondents and included the following three standard financial literacy questions from Lusardi and Mitchell (2008) that have been included on numerous other surveys to measure financial literacy (correct answer in bold):

• (Interest) Suppose you had $100 in a savings account and the interest rate was 2% per year. After 5 years, how much do you think you would have in the account if you left the money to grow?
  Answers: More than $102, Exactly $102, Less than $102, [Don't know]
• (Inflation) Imagine that the interest rate on your savings account was 1% per year and inflation was 2% per year. After 1 year, how much would you be able to buy with the money in this account?
  Answers: More than today, Exactly the same, Less than today, [Don't know]
• (Diversification) Do you think the following statement is true or false? "Buying a single company's stock usually provides a safer return than a stock market mutual fund."
  Answers: True, False, [Don't know]

Typically, the questions are administered with an explicit "don't know" response included in the set of possible answers. In the 2021 SHED, however, whether a respondent received a "don't know" answer choice for the three financial literacy questions was randomly determined. As a result, approximately half of the respondents in the 2021 SHED (n = 5,925) received the three financial literacy questions with an explicit "don't know" response option, and the remaining respondents (n = 5,949) received the questions without the "don't know" option.

This experiment builds on previous work from Bucher-Koenen et al. (2021) who examined a similar experiment conducted on the De Nederlandsche Bank Household Survey (DHS) – a panel survey of the Dutch central bank that is representative of the Dutch speaking population in the Netherlands. However, these experiments differ in that our experiment gives the two versions of the questions (with and without "don't know") to two different groups of respondents on the same survey, while their experiment gives the two versions of the questions to the same group of respondents twice, 6 weeks apart.