Experimenting with mobile money William Jack Georgetown University Based on research with Alev Gurbuz, James Habyarimana and Tavneet Suri Africa Growth Forum, 2015 UNCC, Addis Ababa, Ethiopia

The problem: The solution:

Customer and Agent growth 2011 2009 2010 Customers Agents Customers 2008 Agents 2007

Visualizing financial inclusion (Gurbuz) Excluded Informal Formal

Financial inclusion in Kenya: 2009

Financial inclusion in Kenya: 2013

M-PESA as a risk spreading tool (Suri) Four-year panel survey of 3,000 households, 2007-2010 Negative shocks reduce consumption by about 7% - 10% But M-PESA users see no fall in consumption That is, M-PESA users are fully insured

Long run effects of M-PESA (Suri) Can M-PESA change the level of consumption, not just it’s variance? Possible impacts on: Occupational choice: farming  business? Migration: seasonal or permanent? Use a long-run follow up survey of our sample in 2014 to assess these impacts

Mobile financial services Savings Credit Subsidies

Savings: High Hopes (Habyarimana) How can savings be boosted? Increase salience of saving Provide information about future costs Reduce transaction costs Mobile Money savings account Reduce temptation Commitment Savings Account Increase awareness Regular text reminders

Mobile savings solutions Lock Savings Account Interest bearing mobile money bank account Mobile money transfer service Additional 1% interest 1-6 month maturity Goal set Early withdraws: Lose bonus interest 48 hour waiting period 1-5% interest Loan facility

Context of our study Three counties in Kenya Kisumu, Nyeri, Kilifi Geographic, ethnic, and linguistic diversity Target parents of final year primary school students Six month window ahead of transition to high school

Design: school-based randomization Weekly SMS No SMS Total Control 60 120 Mobile bank account Mobile commitment account 180 360 All groups received information on the importance of saving for high school

Sampling and randomization Kisumu Nyeri Kilifi 3 Counties 3 x 120 school meetings 13 without Std 8 2 refused 7 with no parents 338 x ~14 parents

School meeting Introduction to High Hopes and consent Parental engagement Importance of high school Thinking about the costs of high school Encouragement to save 15 minute baseline survey Treatment administration Members of T1 and T2 assisted in opening accounts if desired T2 invited to convert maturity to January 5, 2015

{ { Timeline Re-contact rate: 88% Primary school exams end High school starts 2014 2015 July Aug Sept Oct Nov Dec Jan Feb Mar Apr { { Endline school-based interviews n = 3,215 Endline invitation letters sent Lock boxes mature Jan 5, 2015 Baseline and recruitment n = 4,713 Endline invitation calls and SMSs Endline phone interviews n = 934 Re-contact rate: 88%

Baseline summary T1 T2 Control Total Number parents at meetings 1,480 1,631 1,713 4,824 Number who took survey 1,437 1,596 1,680 4,713 97% 98% Average # respondents per school 13.0 13.6 15.2 13.9 Access granted to M-PESA and M-Shwari data 1,254 1,388 1,485 4,127 87% 88% Number with M-Shwari at baseline 319 333 389 1,041 22% 21% 23% Number with LSA at baseline 6 9 11 26 0% 1%

M-Shwari take-up Post- Post- Post- Pre- Pre- Pre-

M-Shwari opening dates Control   Treatment 2 Treatment 1

LSA take-up Post-

LSA Maturity Dates January 5, 2015 Control   Treatment 2 Treatment 1

Results Stay tuned

Credit (Suri) Sustainable credit requires screening of bad risks But group-based lending and high deposits can choke off demand Asset collateralization is rare – especially for small, easy to hide, assets

Mobile credit: making credible the threat of repossession dis Small solar power units purchased on credit Cost: $75 Down-payment: $5 Repayment via mobile money Remote penalties and shut-off both feasible

Design Sample: 1,850 small retailers Ex ante assignment Control I. PAYG II. Weekly repayments Ex post assignment II(a). Full enforcement II(b). No penalty III(c). No enforcement

Take-up

Results Penalties increase repayment somewhat Even with complete lack of enforcement, many people still repay Reputations important? Kerosene use falls – health benefits

Subsidies delivered over the mobile network Maternal health care vouchers and transfers (both conditional and unconditional) Implemented in rural western Kenya Meaningful impacts on assisted deliveries CCTs just as effective as maternal vouchers

www.gui2de.org