DIME Impact Evaluation Workshop Innovations for Agriculture 16-20 June 2014, Kigali, Rwanda
Facilitating Savings for Agriculture: Field Experimental Evidence from Rural Malawi Lasse Brune University of Michigan Xavier Giné World Bank Jessica Goldberg University of Maryland Dean Yang University of Michigan 2
Motivation Agriculture employs 2/3 of the labor force and is responsible for 1/3 of GDP growth in sub- Saharan Africa (WDR 2008) Fertilizer is one of the highest-return and most under-exploited investment opportunities for smallholder farmers Many farmers are unable to use optimal amounts of fertilizer because of lack of access to finance 3
Motivation Government response has been large-scale fertilizer subsidies for smallholders (Malawi, Zambia, Tanzania, etc.) In Malawi, 11% of government budget in 2010/11 Unsustainable without continued donor support 4
Raising farm output with rural finance For today: results from the 3 rd of a series of experiments in rural Malawi aimed at raising farm output via financial service provision Insure farmers against adverse events Provide insurance against poor rainfall Facilitate credit for agricultural inputs Improve repayment via biometric identification Encourage farmers to save for their own input purchases Facilitate direct deposit into to ordinary and commitment savings accounts 5
The agricultural cycle in Malawi Rainy season May June July August September October November December January February March April Harvest Planting Hungry season 6
The agricultural cycle in Malawi Rainy season May June July August September October November December January February March April Harvest Planting Savings need to span this period Hungry season 7
Can bank accounts increase the use of farm inputs? We use an RCT to test two questions: What is the effect of direct deposit of farming proceeds into individual bank accounts? Do commitment accounts have different effects than ordinary accounts? 8
Commitment savings Commonly proposed mechanism for increasing savings Allow individuals to voluntarily restrict own access to funds Allow customers to put funds into a special account where their access is restricted for defined period Customers choose release date of funds 9
Sample 3,150 tobacco farmers currently borrowing from a Malawian MFI ~300 borrowing clubs with group liability loans Sell crop to central auction, crop proceeds (net of loan repayment) deposited into club accounts - Our intervention takes advantage of this direct deposit system 10
Randomized control trial: set-up Collaborating institution: Opportunity International Bank of Malawi (OIBM) Treatments involve: Help with account opening procedures Direct deposit of cash crop sales into individual farmer accounts Study participants randomly allocated to control group or one of the savings treatments: Ordinary savings account Ordinary + commitment accounts All clubs given financial education session on savings and budgeting for future needs Randomization at club level 11
Treatment conditions Control group No savings intervention Ordinary savings intervention Direct deposit into individual ordinary accounts Commitment savings intervention Direct deposit into ordinary and commitment accounts 42 clubs 130 clubs 127 clubs 12
Project timing (1) Baseline + offer of savings accounts (2) Endline survey April May June July August September October November December January February March April May June July August September 2009 harvest Planting Hungry season 2010 harvest 13
Summary of findings Facilitating formal savings has positive effects on important outcomes: Savings prior to next planting season Key outcomes in and after next season: agricultural inputs applied, crop output, household expenditures Actual tying of hands appears unnecessary Amounts actually saved in commitment accounts very small; most savings are in ordinary accounts 14
Total deposits into savings accounts (MK), March-October 2009 25000 20000 15000 10000 5000 *** *** 0 Control group Ordinary savings Ordinary + commitment savings 15
Savings balances immediately prior to planting (MK) 3000 2500 *** 2000 1500 *** 1000 500 0 Control group Ordinary savings Ordinary + commitment savings 16
Agricultural outcomes, 2009-2010 season 140000 120000 100000 80000 60000 40000 20000 0 ** Total value of inputs (MK) *** ** Revenue from crop sales (MK) Farm profits (MK) Control group Ordinary savings Ordinary + commitment savings 17
Channel for effects How exactly does offer of commitment accounts lead to higher inputs? Initial hypothesis: funds are locked up and can t be accessed until planting time, at which point they are used for inputs Help to solve both self- and other-control problems Distribution of commitment release dates is consistent with this 18
Channel for effects How exactly does offer of commitment accounts lead to higher inputs? Initial hypothesis: funds are locked up and can t be accessed until planting time, at which point they are used for inputs Help to solve both self- and other-control problems Distribution of commitment release dates is consistent with this But on closer inspection, it does not appear that commitment accounts help via literal tying of hands Vast majority of funds ended up in ordinary, not commitment accounts However, no conclusive evidence for alternative channels 19
In sum Facilitating direct deposit into individual savings accounts leads to higher input use, crop output, and household expenditures later on Commitment treatment effects not due to self-control Good news for welfare impact of intervention, since funds available to cope with shocks Cannot rule out mental accounting or other psychological channels Maybe people with two accounts thought differently about saving and investing than those with only one account 20
DIME Impact Evaluation Workshop Innovations for Agriculture 16-20 June 2014, Kigali, Rwanda
Commitment, Mental Accounting & Agricultural Investment Evidence from Rwanda Maria Jones, Florence Kondylis & Daniel Stein
Big idea
problem: seasonality SEASON A June 2012 September 2012 February 2013 June 2013
problem: seasonality SEASON A SEASON B June 2012 September 2012 February 2013 June 2013
problem: seasonality GAP SEASON B SEASON A SEASON B June 2012 September 2012 February 2013 June 2013
problem: seasonality GAP No time to sell harvest! SEASON B SEASON A SEASON B June 2012 September 2012 February 2013 June 2013
good intentions aren t enough! $60 Planned input expenditures $50 $40 $30 $20 $10 $0 Season A Season B
good intentions aren t enough! $60 Actual Input Expenditures $50 $40 $30 $20 $10 $0 Season A Season B
good intentions aren t enough! $60 Planned vs. Actual Input Expenditures $50 $40 $30 $20 $10 $0 Season A Planned Input Expenditures Season B Actual Input Expenditures
solutions? Kenya Offering free delivery of fertilizer right after harvest increases fertilizer use 47 70% Bigger effects than 50% subsidy on fertilizer
solutions? Kenya Offering free delivery of fertilizer right after harvest increases fertilizer use 47 70% Bigger effects than 50% subsidy on fertilizer Malawi Offering commitment savings accounts to tobacco farmers increases agricultural input use (48%), crop sales (27%), and household expenditures (27%)
Training staff at Credit Cooperatives Mobilizing Farmers. Account registration.. Saving in the new accounts.
Savings innovations in Rwanda TARGETED ACCOUNT COMMITMENT ACCOUNT
Innovations interest farmers 80% 70% 67% 60% 57% 50% 40% 30% 20% 10% 0% 27% Targeted Savings Account Attended Training 32% Commitment Savings Account Opened Account
Gender affects take-up 35% 33% 30% 28% 27% 29% 25% 20% 15% 10% 5% 0% Targeted Savings Account Men Women Commitment Savings Account
Innovations increase savings Before Season A Planting Male offered target account Male offered commitment account Female offered target account Female offered commitment account Control Group N Number of deposits per month Average amount deposited per month Maximum Balance 0.05 1.65 4.82* [0.06] [1.54] [2.63] 0.12** 0.89 1.61 [0.06] [1.04] [1.62] -0.07-0.8-0.48 [0.10] [1.2] [1.66] -0.03-0.09 1.18 [0.09] [0.97] [1.85] 0.29 4.26 5.57 1200 1200 1200
Innovations increase savings Before Season A Planting Male offered target account Male offered commitment account Female offered target account Female offered commitment account Control Group N Number of deposits per month Average amount deposited per month Maximum Balance 0.05 1.65 4.82* [0.06] [1.54] [2.63] 0.12** 0.89 1.61 [0.06] [1.04] [1.62] -0.07-0.8-0.48 [0.10] [1.2] [1.66] -0.03-0.09 1.18 [0.09] [0.97] [1.85] 0.29 4.26 5.57 0.05 1.65 4.82*
Innovations increase savings Before Season A Planting Male offered target account Male offered commitment account Female offered target account Number of deposits per month Average amount deposited per month Maximum Balance 0.05 1.65 87% [0.06] [1.54] increase 41% increase 0.89 1.61 [1.04] [1.62] -0.07-0.8-0.48 [0.10] [1.2] [1.66] Female offered commitment account -0.03-0.09 1.18 Control Group N [0.09] [0.97] [1.85] 0.29 4.26 5.57 0.05 1.65 4.82*
Innovations increase savings Before Season A Planting Male offered target account Male offered commitment account Female offered target account Number of deposits per month Average amount deposited per month 0.05 1.65 [0.06] [1.54] Deposit more often Maximum Balance Save more 0.89 1.61 [1.04] [1.62] Female offered commitment account No significant changes Control Group N 0.29 4.26 5.57 0.05 1.65 4.82*
Innovations increase savings Before Season B Planting Number of deposits per month Average amount deposited per month Maximum Balance Male offered target account 0.05 2.09 2.01 [0.07] [1.55] [2.5] Male offered commitment account 0.11* 0.47-0.71 [0.06] [1.2] [2.16] Female offered target account 0.06 0.84 2.11 [0.09] [1.38] [2.4] Female offered commitment account Control Group N 0-0.05 1.71 [0.07] [0.9] [2] 0.29 4.56 7.79 1200 1200 1200
Innovations increase savings Before Season B Planting Number of deposits per month Average amount deposited per month Maximum Balance Male offered target account 0.05 2.09 2.01 [0.07] [1.55] [2.5] Male offered commitment account 38% increase 0.47-0.71 [1.2] [2.16] Female offered target account 0.06 0.84 2.11 [0.09] [1.38] [2.4] Female offered commitment account Control Group N 0-0.05 1.71 [0.07] [0.9] [2] 0.29 4.56 7.79 1200 1200 1200
Innovations increase savings Before Season B Planting Number of deposits per month Average amount deposited per month Maximum Balance Male offered target account 0.05 2.09 2.01 [0.07] [1.55] [2.5] Male offered commitment account Female offered target account Deposit more often 0.47-0.71 [1.2] [2.16] Female offered commitment account No significant changes Control Group N 0.29 4.56 7.79 1200 1200 1200
Innovations increase investment $ fertilizers $ labor $ seeds Total input costs Male offered target account Male offered commitment account Female offered target account Female offered commitment account Control Group 4.87* 2.11 16.48 23.46* [2.93] [4.75] [11] [13.36] 3.12-0.02 21.89** 24.98** [2.91] [3.9] [8.65] [11.54] 3.4 1.25 3.32 7.97 [3.92] [3.46] [9.67] [12] 1.72-1.19 5.94 6.48 [3.05] [2.71] [7.9] [10.22] 17.31 16.91 8.07 42.29 N 1200 1200 1200 1200
Innovations increase investment $ fertilizers $ labor $ seeds Total input costs Male offered target account Male offered commitment account Female offered target account Female offered commitment account Control Group 4.87* 2.11 16.48 23.46* [2.93] [4.75] [11] [13.36] 3.12-0.02 21.89** 24.98** [2.91] [3.9] [8.65] [11.54] 3.4 1.25 3.32 7.97 [3.92] [3.46] [9.67] [12] 1.72-1.19 5.94 6.48 [3.05] [2.71] [7.9] [10.22] 17.31 16.91 8.07 42.29 N 1200 1200 1200 1200
Innovations increase investment $ fertilizers $ labor $ seeds Total input costs Male offered target account Male offered commitment account 4.87* 2.11 16.48 [2.93] [4.75] [11] 3.12-0.02 21.89** 55 59% increase Female offered target account Female offered commitment account Control Group [2.91] [3.9] [8.65] 3.4 1.25 3.32 7.97 [3.92] [3.46] [9.67] [12] 1.72-1.19 5.94 6.48 [3.05] [2.71] [7.9] [10.22] 17.31 16.91 8.07 42.29 N 1200 1200 1200 1200
Innovations increase investment $ fertilizers $ labor $ seeds Total input costs Male offered target account Male offered commitment account Female offered target account Female offered commitment account Control Group 28% increase 2.11 16.48 [4.75] [11] 3.12-0.02 270% [2.91] [3.9] increase 55 59% increase 3.4 1.25 3.32 7.97 [3.92] [3.46] [9.67] [12] 1.72-1.19 5.94 6.48 [3.05] [2.71] [7.9] [10.22] 17.31 16.91 8.07 42.29 N 1200 1200 1200 1200
Innovations increase production Total Value Harvested Total Value Sold Male offered target account Male offered commitment account Female offered target account Female offered commitment account 31.32*** 23.64*** [9.35] [6.36] 23.58*** 18.95*** [8.76] [5.73] 6.84 8.26 [16.03] [12.05] 11.4 6.12 [10.3] [6.06] Control Group Mean 43.35 25.96 N 1200 1200
Innovations increase production Total Value Harvested Total Value Sold Male offered target account Male offered commitment account Female offered target account Female offered commitment account 31.32*** 23.64*** [9.35] [6.36] 23.58*** 18.95*** [8.76] [5.73] 6.84 8.26 [16.03] [12.05] 11.4 6.12 [10.3] [6.06] Control Group Mean 43.35 25.96 N 1200 1200
Innovations increase production Total Value Harvested Total Value Sold Male offered target account Male offered commitment account 54 72% increase 23.64*** [6.36] 18.95*** [5.73] Female offered target account Female offered commitment account 6.84 8.26 [16.03] [12.05] 11.4 6.12 [10.3] [6.06] Control Group Mean 43.35 25.96 N 1200 1200
Innovations increase production Total Value Harvested Total Value Sold Male offered target account Male offered commitment account 54 72% increase 73 91% increase Female offered target account Female offered commitment account 6.84 8.26 [16.03] [12.05] 11.4 6.12 [10.3] [6.06] Control Group Mean 43.35 25.96 N 1200 1200
Take away New savings products have big impacts on agricultural investment and production
Take away New savings products have big impacts on agricultural investment and production Similar effects across two accounts Mental accounting matters most Complexity / sophistication
Take away New savings products have big impacts on agricultural investment and production Similar effects across two accounts Mental accounting matters most Complexity / sophistication Keep learning Positive outcomes observed for men only Take-up could be higher
Insuring commitments agricultural shocks health shocks
Insurance increases appeal 45% 40% 42% 35% 30% 32% 25% 20% 15% 10% 5% 0% Insured Commitment Standard Commitment
Gender affects take-up 60% 50% 43% 44% 48% 40% 30% 28% 20% 10% 0% Agricultural Shock Insurance Health Shock Insurance Men Women
Big Idea Time of financial growth in Rwanda Many opportunities to scale up and mainstream innovations -- need to know what works best! Gender dimensions
Big idea Time of financial growth in Rwanda Many opportunities to scale up and mainstream innovations -- need to know what works best! Gender dimensions Small innovations added to a large investment can make a big difference in farmers lives Simple tweak almost doubled income from agricultural sales
DIME Impact Evaluation Workshop Innovations for Agriculture 16-20 June 2014, Kigali, Rwanda
Selling low and buying high? Financial constraints to profitability Marshall Burke and Lauren Falcao UC Berkeley DIME Workshop, Kigali, June 2014
Seasonal price fluctuations At our rural study site: Typical seasonal increase is 100%!
So: buy low, sell high?
Sell low, buy high! Traders? Only half store any maize for more than 2 months. Those who do store: only store 25% of amt purchased
What s going on? Q: Why do you sell at harvest instead of later, when prices are higher? A: I need the cash.
With OAF, we offered a storage loan Offer cash (~$100) at harvest (T1) or three months after harvest (T2) to randomly selected OAF farmers Stored maize as collateral 10% interest, repay flexibly Re-visited treated and control farmers many times throughout the year to track inventories, sales, purchases, etc. Take-up was very high: >70%
But what if the loan works? Farmers might sell less at harvest, sell more later on: This could reduce the seasonal price increase! This could reduce the profitability of the loan! To understand these price effects, we also randomized the number of loan offers across markets (50).
Results at the farmer level
Results at the farmer level
Results at the farmer level Even with low price rise: 20% ROI after repayment for farmers who got the loan at harvest
What happened to prices? Difference in prices, high versus low treatment density Post-harvest prices ~5% higher in areas with lots of loans
What did these price effects mean for farmers? It reduced the effectiveness of the loan in high density areas Arbitrage opportunity was smaller It helped revenues of control farmers! (although not quite significantly so) But probably also had positive spillovers outside our experiment
Conclusions from farmer study 1. Access to well-timed credit can help improve the profitability of small farms 2. Absence of credit markets spills over into other markets that matter for the poor Missing credit market exacerbates seasonal price swings
Where are the traders? We are following 200 traders over time, collecting detailed cost & marketing data Storage does seem profitable: ROI 15-50% over 6 months But: little storage. About half store any maize, and those don t store much Why?
Why no trader storage? Difficulty in accessing credit? 75% say this is a problem But 80% also say they could get a loan if they need it (?) Price uncertainty? Rotation more profitable? 6 mo ROI is 100%! Compare: 15-50% for storage Ongoing
DIME Impact Evaluation Workshop Innovations for Agriculture 16-20 June 2014, Kigali, Rwanda
Savings and Fertilizer Initiative (SAFI) Jonathan Robinson, Michael Kremer, Esther Duflo. Center for Effective Global Action University of California
A small nudge at the appropriate time was as powerful as a heavy subsidy in promoting fertilizer use
Why don t you use fertilizer?
Can small, time-limited offers for advanced fertilizer purchase increase fertilizer use?
INTERVENTION 1 District in W. Kenya 2 seasons (long & short rains) 4 treatment arms
Basic SAFI
SAFI with Choice of Timing
SAFI during Growing Season
SAFI & 50% Subsidy
A small nudge to purchase fertilizer at the appropriate time was as powerful in promoting fertilizer use as a 50% subsidy.
SAFI SUBSIDY
DIME Impact Evaluation Workshop Innovations for Agriculture 16-20 June 2014, Kigali, Rwanda