Why is voluntary financial education so unpopular? Experimental evidence from Mexico

Size: px

Start display at page:

Download "Why is voluntary financial education so unpopular? Experimental evidence from Mexico"

Bethanie Johns
5 years ago
Views:

1 Why is voluntary financial education so unpopular? Experimental evidence from Mexico Miriam Bruhn, World Bank Gabriel Lara Ibarra, World Bank David McKenzie, World Bank Understanding Banks in Emerging Markets EBRD, September 5, 2013 All opinions expressed in this work are those of the authors alone and do not necessarily represent those of the World Bank or the partnering institutions.

2 Why care about financial education? Levels of financial literacy tend to be low in developed and developing countries (Xu and Zia, 2012) Lower financial literacy is associated with Lower savings Lusardi and Mitchell (2009), Klapper and Panos (2011) Less responsible borrowing behavior Lusardi and Tufano (2008) Higher probability of mortgage delinquency and default during the subprime mortgage crisis Gerardi, Goette, and Meier (2010) Lower probability of strategic default post-crisis Burke and Mihaly (2012)

3 Policy Responses Governments aim to promote financial education, particularly in developing countries where access to financial products has been expanding Financial education in schools Regulators encourage financial institutions to educate consumers Adult financial education courses Online training Financial institutions provide training to delinquent or at-risk clients

4 Impact of Financial Education School financial education seems to increase savings (to some extent) Bernheim et al. (2001), Bruhn et al. (2013) General purpose adult training courses tend to have only small effects on financial behavior Effects are larger for subgroups (low ex-ante literacy) or when training is provided at a teachable moment Duflo and Saez (2011), Cole, Sampson, and Zia (2009), Gibson, McKenzie, and Zia (2012), Doi, McKenzie, and Zia (2012), and Seshan and Yang (2012)

5 Training Take-Up is Often Low Voluntary financial education is widely available today, yet seldom used Willis (2011, p. 430) Brown and Gartner (2007) Only 2 out of 3,200 people completed an online course offered by a U.S. credit card provider (0.06%) Only 384 out of 42,000 cardholders completed online training provided by a U.S. bank (0.9%) Still only 6.5% of cardholders completed a course when offered a 60 minute phone card Impact evaluations in developing countries often have take up rates below or around 40 percent (despite intensive screening and handholding)

6 Teacher, leave them kids alone The Economist, Feb 16th 2013 Financial education has had disappointing results in the past.

7 Our Paper Collaborate with a financial institution that provides a widely accessible training course in Mexico City Conduct randomized experiments to investigate the following questions Are there economic or behavioral constraints which prevent more individuals from participating in financial education programs? Are there any benefits to these marginal individuals from doing so, or are they rationally choosing not to participate in such training?

8 Preview of Results Training take-up seems to be low due to low benefits Find no evidence that time-inconsistency, high discount rates, high transportation costs or uncertainty about benefits influence take-up The training increases financial knowledge, but has only a temporary effect on savings and no effect on borrowing behavior for individuals who need to be incentivized to attend the training

9 Outline Financial literacy training course Experimental design and study sample Take-up and reasons for low attendance Follow-up data collection Impact of training on financial knowledge, behavior, and outcomes

10 Financial Literacy Training Course Large-scale program offered free of charge Several locations in Mexico City Classrooms with capacity for 20 participants Two sessions per day (Monday Saturday) Sessions often not filled to capacity About 0.6 percent of Mexico City s adult population takes the course each year

11 Course Content Half-day course, covering Savings Savings instruments, how to save more, budgeting Retirement and pension funds Credit cards Fees, how to decipher a credit card statement Responsible use of credit Credit score and history, good practices Computer-based training, with group exercises, take-home material

12 Outline Financial literacy training course Experimental design and study sample Take-up and reasons for low attendance Follow-up data collection Impact of training on financial knowledge, behavior, and outcomes

13 Experimental Design Randomized encouragement design First, screen study sample for interest in participating in a financial education course Randomly assign interested individuals to either Treatment group: Receives phone-call, inviting them to sign up for the specific financial education course, phone reminder for attendance (+ extra incentives) Control group: Receives no invitation to the course

14 Screener Survey I: Mailing Campaign Collaborated with a financial institution to send screener survey to 40,000 clients in Mexico City Letter + two-page survey with pre-paid return envelope (+ phone and online option) Expected response: 800 1,200, based on typical 2-3 percent mailing response rate Actual response: 42 = 0.1 percent

15 Screener Survey II: Facebook Created Facebook page for financial literacy with link to an online survey Facebook ad pointing to this page was displayed 16 million times 1,240 fans of the Facebook page 119 survey responses

16 Screener Survey III: Street + Branch Face-to-face survey in public locations (one month) and outside branches of our partner financial institution (2 weeks) Completed questionnaires 6,945 from street survey 2,294 from branch survey Only about 50 percent had valid contact information (as verified by follow-up phone-calls)

17 Treatment Randomization Final sample includes 3,503 people 1,752 treatment group 1,751 control group Stratified randomization by type of screener survey, gender, education, financial institution client or not, financial behavior reported on screener survey

18 Outline Financial literacy training course Experimental design and study sample Take-up and reasons for low attendance Follow-up data collection Impact of training on financial knowledge, behavior, and outcomes

19 Initial Take-Up in Treatment Group About 60 percent of the treatment group signed up for a course session Only 1/3 of these people actually attended, despite phone reminders on the day before Overall take-up rate was 17.8 percent Investigate reasons for low take-up

20 Potential Reason for Non-Attendance Course has no (or only small) benefits Assuming it has sizeable benefits, people may not attend if Benefits accrue in the future (high discount rates) Costs are too high (e.g. transportation) They are uncertain about the benefits Randomly assign incentive treatments to address and test these reasons

21 Experimental Interventions to Overcome Barriers to Attendance 1. US$72 gift card for completing the course 2. US$36 gift card for completing the course 3. US$36 gift card received one month after completing the course 4. Free taxi ride to and from the course location 5. Video CD with positive testimonials from people who had attended the course

22 Take-Up Rates by Incentive Group % No extra incentive $72 now $36 now $36 later Free transportation Testimonials Full treatment group Only individuals who could be reached Low take-up seems to be due to low benefits, not high discount rates, high transportation costs, or uncertainty about benefits

23 Which individuals attend training? Take-up rate was 28 percent among individuals who are clients of a financial institution and 18 percent among those who are not Measure impact of the course only for financial institution clients Other characteristic that are significantly associated with greater take-up Having a bachelor s degree or higher Age (older people more likely to take-up)

24 Outline Financial literacy training course Experimental design and study sample Take-up and reasons for low attendance Follow-up data collection Impact of training on financial knowledge, behavior, and outcomes

25 Follow-Up Data Collection 15 minute survey on financial knowledge, behavior, and outcomes Conducted over the phone and in person if phone interview was not possible Attrition: 29 percent in treatment group, 25 percent in control group (sample size about 1,500) Baseline characteristics are balanced across treatment and control groups for people who replied

26 Administrative Data Our partner financial institution provided administrative data on their clients 470 clients found in their database (matched by name and address) Due to confidentiality reasons, did not receive individual data, but averages and medians for the treatment and control group Savings account balance, credit card balance, percentage of credit card debt paid off

27 Project Timeline J F M A M J J A S O N D J F M A M J J Mailing campaign Facebook campaign Street/Branch s urvey Invitation to course Follow-up survey Administrative data Follow-up survey conducted 6-8 months after the course Admin data for 17 months throughout the project

28 Outline Financial literacy training course Experimental design and study sample Take-up and reasons for low attendance Follow-up data collection Impact of training on financial knowledge, behavior, and outcomes

29 Estimating Treatment Impact OLS intention-to-treat (ITT) estimates y i,s,m = α + βcourseinvite i,s,m + γ s d s + δ m d m + ε i,s,m Control for randomization strata dummies (d s ) and month of follow-up interview dummies (d m ) OLS local average treatment effect (LATE) estimates: course attendance is instrumented with course invite Effect of training for people who are induced to attend it as a result of our interventions, but would not attend it without the incentives we provided

30 Impact on Financial Knowledge Control ITT Treatment LATE Treatment Mean Difference Difference Knowledge index (average of 8 components below) *** *** (0.0094) (0.0261) (1) Knows what UDI is "Unidad de Inversion" (0.0150) (0.0426) (2) Knows deposit insurance exists up to 400,000 UDIs *** *** (0.0188) (0.0533) (3) Knows what a credit report is ** ** (0.0245) (0.0684) (4) Knows credit card cycle is 30 days (0.0251) (0.0701) (5) Knows they have 20 days to pay credit card w/o interest (0.0168) (0.0472) (6) Knows that what CAT is "Costo Anual Total" * * (0.0216) (0.0602) (7) Knows what an AFORE (pension fund) is ** ** (0.0219) (0.0613) (8) Knows retirement age is (0.0234) (0.0659) Notes: Robust standard errors in parentheses. Statistical significance: * 10 percent, ** 5 percent, *** 1 percent.

31 Impact on Savings Behavior Control ITT Treatment LATE Treatment Mean Difference Difference Savings behavior index (avg.of 5 components) (0.0124) (0.0348) (1) Checks bank transactions regularly (0.0226) (0.0644) (2) Keeps track of expenses (0.0206) (0.0582) (3) Makes a budget (0.0211) (0.0596) (4) Has a savings goal (0.0250) (0.0705) (5) Cut expenses in past 3 months * * (0.0247) (0.0698) Notes: Robust standard errors in parentheses. Statistical significance: * 10 percent, ** 5 percent, *** 1 percent.

32 Impact on Savings Outcomes Control ITT Treatment LATE Treatment Mean Difference Difference Savings outcomes index (avg. of 3 components) ** ** (0.0147) (0.0414) (1) Has any type of savings (0.0200) (0.0566) (2) Saved more than zero during past 6 months (0.0192) (0.0524) (3) Saves more each month than a year ago * Notes: Robust standard errors in parentheses. Statistical significance: * 10 percent, ** 5 percent, *** 1 percent. 1 Includes bank savings account, caja de ahorro, tanda and other non-retirement savings Results are similar when controlling for whether the individual received a monetary incentive payment for participation in the financial literacy course

33 Admin Data on Savings Median Savings Account Balance Course participation US$ Dec-10 Jan-11 Feb-11 Mar-11 Apr-11 May-11 Jun-11 Jul-11 Aug-11 Sep-11 Oct-11 Nov-11 Dec-11 Jan-12 Feb-12 Mar-12 Apr-12 May-12 Control group Treatment group

34 Impact on Credit Card Behavior Control ITT Treatment LATE Treatment Mean Difference Difference Currently has at least one credit card (0.0210) (0.0596) Panel A: Credit card behavior index and components Credit card behavior index (avg.of z-scores of 6 components) (0.0229) (0.0652) (1) Knows credit limit (0.0209) (0.0594) (2) Knows interest rate (0.0193) (0.0552) (3) Checks statement every month (0.0209) (0.0594) (4) Fraction of past 6 months where paid balance in full (0.0170) (0.0479) (5) Fraction of past 6 months where made only the min. payment (0.0138) (0.0389) (6) Fraction of past 6 mths where got cash through the credit card Notes: Robust standard errors in parentheses. Statistical significance: * 10 percent, ** 5 percent, *** 1 percent. 1 Included in credit card behavior index with negative sign.

35 Impact on Credit Card Outcomes Control ITT Treatment LATE Treatment Mean Difference Difference Panel B: Credit card outcomes index and components Credit card outcomes index (avg. of z-scores of 3 components) (0.0416) (0.1184) (1) Issuer blocked credit card during past 6 months (0.0096) (0.0270) (2) Fraction of past 6 mths where was charged late paymt fees (0.0064) (0.0183) (3) Fraction of past 6 mths where was charged overdraft fees (0.0034) (0.0098) Notes: Robust standard errors in parentheses. Statistical significance: * 10 percent, ** 5 percent, *** 1 percent.

36 Admin Data on Credit Cards (I) Median Credit Card Balance US$ Course participation - Dec-10 Jan-11 Feb-11 Mar-11 Apr-11 May-11 Jun-11 Jul-11 Aug-11 Sep-11 Oct-11 Nov-11 Dec-11 Jan-12 Feb-12 Mar-12 Apr-12 May-12 Control group Treatment group

37 Admin Data on Credit Cards (II) Average Percentage of Credit Card Debt Paid Off Each Month Course participation % Dec-10 Jan-11 Feb-11 Mar-11 Apr-11 May-11 Jun-11 Jul-11 Aug-11 Sep-11 Oct-11 Nov-11 Dec-11 Jan-12 Feb-12 Mar-12 Apr-12 May-12 Control group Treatment group

38 Impact on Loan Behavior and Outcomes Control ITT Treatment LATE Treatment Mean Difference Difference Panel A: Loan behavior index and components Loan behavior index (avg.of 3 components) (0.0118) (0.0334) (1) Applied for a loan from any source during past 6 months (0.0210) (0.0594) (2) Went to a pawn shop to get credit during past 6 months (0.0152) (0.0429) (3) Stopped servicing outstanding debt during past 6 months (0.0180) (0.0523) Panel B: Loan outcomes index and components Loan outcomes index (avg. of z-scores of 2 components) (0.0427) (0.1206) (1) Currently has a loan (from any source) (0.0234) (0.0660) (2) Total outstanding debt as percentage of annual income (1.1899) (3.2220) Notes: Robust standard errors in parentheses. Statistical significance: * 10 percent, ** 5 percent, *** 1 percent.

39 Summary and Conclusion The financial literacy course increased financial knowledge Positive effect on savings is only temporary No effect on borrowing behavior Individuals who do not attend the course voluntarily seem to have little benefit from doing so

40 Alternative Ways of Providing Financial Education Course/classroom setting for the provision of financial education may not be appealing to the general public Higher take-up for people with a bachelor s degree Could provide financial education through alternative channels Video games Soap operas (Berg and Zia, 2013)

41 Baseline Data Stratification Variables Control Treatment Mean Difference Baseline survey conducted in branch Client of partner institution (vs. other institution) Made savings deposit during past month Has credit card Paid more than credit card minimum in all past 6 months Has bachelor's degree or higher Female Sample Size Notes: *, **, and *** indicate statistically different from control mean at the 10, 5 and 1% levels respectively Sample excludes individuals who are not clients of a financial institution 1 Conditional on having a credit card

42 Other Baseline Variables Control Treatment Mean Difference Age Occupation is employee Paid credit card late in past 6 months Monthly household income is above US$ Monthly household expenditure is above US$ Sample Size Notes: *, **, and *** indicate statistically different from control mean at the 10, 5 and 1% levels respectively Sample excludes individuals who are not clients of a financial institution 1 Conditional on having a credit card

The minimal impact of a large-scale financial education program in Mexico City

Accepted Manuscript The minimal impact of a large-scale financial education program in Mexico City Miriam Bruhn, Gabriel Lara Ibarra, David McKenzie PII: S0304-3878(14)00029-7 DOI: doi: 10.1016/j.jdeveco.2014.02.009