14.74 Lecture 22: Savings Constraints

Transcription

1 14.74 Lecture 22: Savings Constraints Prof. Esther Duflo May 2, 2011 In previous lectures we discussed what a household would do to smooth risk with borrowing and savings. We saw that if they can borrow and save, they will try to maintain a smooth consumption profile, especially if the shocks are temporary. Another reason to save, for a household that wants to invest in a business, is to accumulate some collateral and thus increase how much they can borrow. However, we did not discuss what would happen if households had difficulty to save. Two sources for those possible difficulties: External: Internal: 1 Savings Constraints If you want to save, where can you keep the money? 1

2 Why may a bank be reluctant to keep small savings account? Is this because they are evil? is there an easy solution to this problem? Poor households use all sorts of ways to save: Self Help Groups Rotating Savings and Credit Associations (roscas) Productive assets (e.g.: Bullocks) Possible sources of inefficiency: Roscas: you don t get the money when you need it. Deposit collectors (who will charge negative interest rates to take the money) Productive assets : what will you do with ONE bullocks! (Rosenzweig and Wolpin). An experiment which try to test the idea that households may be savings constrained (Dupas and Robinson, 2009). Open free savings account for small business owners at a local bank (waive the opening fee, which is normally $7 (for business owners who make on average about $2 a day). The accounts have no interest and a withdrawal fee of 50 cents for transfers below $8, 80 cents for withdrawals between $8 and $15 and $1.5 above. At baseline, 2% of people had an account. Researchers did a baseline with 300 people, and randomly selected half of them, to whom they offered to pay the opening fees for a savings account. After 6 months, they had people fill daily log-books on business activities and expenditures, for about 3 months. Usage: See conditional distribution function of savings. What is the main finding? Results Increase in in investment in the business. Some increase in consumption Decrease in sensitivity to shocks. 2

3 2 Savings and Self Control We just reviewed external savings constraints. In fact there may also be internal constraints. In fact, there must be such constraints, because otherwise, there would be ways to save for individuals: They could use the savings account more systematially (Dupas-Robinson experiment) They could avoid borrowing at very high interest rate (fruit vendors in Chennai) They could use fertilizer on their field, which divisible, high returns (fertilizer in Kenya). They could invest more in their businesses (in hyderabad, we calculate that spending just a little less on health would allow them to double their capital stock). We have seen that people may have preference that exhibit hyperbolic discounting : with 3 periods, the individual maximizes: Maxu(c1) + β[u(c2) + δu(c3)] People will tend to save relatively little in this model: they will always wait till tomorrow to save (since today is important). They will tend to procrastinate. Example: saving for fertilizer in Kenya. Two seasons a year: at the end of harvest, you have lots of maize. Three period: Harvest: get x. Consume, save or buy fertilizer (cost 1, plus cost of shopping f) Planting: get nothing. One more chance to buy fertilizer Harvest: Get y H if you used fertilizer, Y L otherwise you are hyperbolic discounter. viewed from Harvest utility is x f 1 + β i y H if bought fertilizer in period 1 x 1 β f + β i y H 3

4 if saved for fertilizer in period 1 and plan to buy it in period 2 x + β i y L if did not save for fertilzer. In every period, farmers can be either very impatient (β L ) or quite patient (β H ). For the patient people, x f 1 + β H y H > x + β H y L For the impatient people x f 1 + β L y H < x + β L y L Some farmers are always patient, some are always impatient, and some are patient or impatient in every period with probability p. However, they also tend to overestimate the probability that they will be patient in the future (naivete): they think it is p What will always impatient farmers do at harvest What will always patient do at harvest? and in the next period? What will farmers who happen to be patient do at harvest: They want to use fertilizer Will they save or buy it right away? How will they decide? Compare expected utility of buying fertilizer now, and expected utility of saving and buying later: x 1 p[β f + β H y H ] + (1 p)β H Y L What determines whether they buy today or save? What will happen at planting time? In this way, most people make plans to use fertilizer early in the season but do not end up actually doing it. In kenya it is the case in the data: about 98% of farmer plan to use fertilizer early on, but 20%-30% end up doing it! 4

5 What kind of interventions could help period 1 farmer to use fertilizer. The SAFI program: offer farmers to buy fertilizer right after harvest, and get free delivery. Why should it work? Why would it work better than free delivery at the time of planting? Suppose I come to people s home before harvest, and offer them the choice about when I will come back to offer program, when should taht come. Results in table for different version of the program: SAFI free delivery at harvest time SAFI with choice of timing See results in table. 5