L8: Savings

A, B; The footnote to Table 3 notes that "in addition to the variables listed, the regressions included dummy variables for 20 regions and two years."

In order to run the regression that Paxson runs in Table 3, Column 1 (i.e. the income equation), what dummies will you need to create for variables that are not included in the dataset? (Check all that apply.) a. Region b. Year c. School type d. Gender

A; Individuals in the commitment savings group were asked to either set a time goal for how long the money saved in the account could not be withdrawn, or an amount goal whereby the money could not be withdrawn until a certain amount had been saved. Thus, restrictions on withdrawal (either based on time or the amount) served as the commitment device.

In the study Tying Odysseus to the Mast by Ashraf et al., what form did the commitment device take in the commitment savings product? a. Withdrawal restrictions b. Minimum deposit requirements c. High interest offered on the savings d. Deposit collectors

D; ROSCAs are ubiquitous around the world, while self-help groups are popular in many parts of India. The poor may also save in productive assets such as bullocks, as Professor Banerjee points out. Thus, all of these options are used by the poor as means of saving, either in lieu of, or in addition to more formal bank accounts.

The poor often use which of the following means of saving? a. Self-help Groups b. Rotating Savings and Credit Associations (ROSCAs) c. Productive Assets d. All of the above

A; One of the puzzles is why people would borrow at the high interest rate of 5% per day and not saving. Framed another way, all borrowers have access to a high return "investment" if they were to just borrow less and save the money, since they would no longer be paying the 5% per day rate on their daily loans.

The study of fruit and vegetable sellers in India found that they faced an interest rate of _________. a. 5% per day b. 5% per week c. 5% per month d. 5% per year

True; Although bricks represent a relatively illiquid form of saving, the market for bullocks in general is quite active and bullocks thus represent a comparatively liquid form of saving. However, the supply and demand for bullocks tends to be correlated within villages - since this is often determined by seasonal fluctuations - and it may thus be difficult to sell bullocks off at certain times.

True or False: Professor Banerjee explains that bullocks are generally a more liquid form of saving than bricks. True False

True; In both self-help groups and ROSCAs, members put money into a group pot at every, regularly scheduled, meeting, with one member then getting the pot each time. However, where the recipient of the amount is determined by a rotation system with a ROSCA, in the case of self-help groups, recipients are determined based on the strength of the case that they present for a loan. They then pay interest to the group on the loan that they take out from the group.

True or False: Self-help groups are a more market-oriented version of ROSCAs, with recipients of the pot of savings determined each week by a price mechanism. a. True b. False

False; ROSCAs generally follow a fairly rigid structure: regular meetings at which members put in fixed amounts, and a fixed schedule along which cash outs from the group accrue to specific individuals. Thus, situations may arise when the ROSCA does not meet the cash flow needs of members at the right time owing to the lack of flexibility in the structure.

True or False: Unlike self-help groups, ROSCAs represent a flexible arrangement for saving. True False

False; As Professor Banerjee points out, even in the Dupas and Robinson study in Kenya, although a few people benefited from having access to a savings account, a number of people did not use the account at all. Some people put no money at all into the account and infrequent usage by a large number of people is a persistent pattern observed in many cases where savings accounts are offered.

True or False: When offered low cost savings accounts, studies have found that most people use the accounts quite frequently. a. True b. False

D; Under hyperbolic discounting, individuals are generally indifferent between consumption at various points in the future i.e. consumption tomorrow is not worth too much more than consumption the day after that. However, consumption today tends to look a lot more attractive than consumption in either of those two periods, thus, individuals will tend to save very little today. Once tomorrow arrives though, the individual will follow a similar calculus and privilege consumption in the new "today" over consumption in the new "tomorrow", thereby again saving very little. Thus, while today the individual may be relatively indifferent between saving or consumption tomorrow, this calculus changes once tomorrow arrives and thus (s)he will save very little.

Under hyperbolic discounting, individuals: a. Save very little today, but save a lot in the future b. Save a lot today, but save very little in the future c. Tend to save a lot d. Tend to save very little

B; The surveys ask individuals whether they would prefer P200 today versus P300 tomorrow in one question. Individuals are then asked whether they would prefer P200 in 6 months or P300 guaranteed in 7 months. The researchers posit that individuals who exhibit a preference for P200 in the first case, but P300 in the second case (i.e. those individuals who "reverse" their preferences) are exhibiting hyperbolic preferences.

What do survey questions in the Philippines commitment savings study try to elicit as a proxy for hyperbolic preferences? a. Preference consistency b. Preference reversal c. Goal setting d. Consumption smoothing

B, C, D; Professor Banerjee does not mention any impacts on women's empowerment from the Dupas and Robinson savings access study. However, he does highlight that the study finds an increase in investment in businesses and a broad decrease in sensitivity to shocks, which allows savers to not have to reduce consumption.

What impacts from the Dupas and Robinson study does Professor Banerjee highlight? (Check all that apply) a. Increase in women's empowerment b. Increase in investment in business c. Some increase in consumption d. Decrease in sensitivity to shocks

A, D; There are two, cross-cutting treatments in the study: a cash grant that is enough for individuals to buy out their debt, and a half day financial literacy training to pick up the effects of financial illiteracy. The grant is intended to put the vendors on a path to save, or to at least help them pay down their debts and stay where they are.

What is the randomly assigned treatment in the study with vendors in Chennai? (Check all that apply) a. Cash grant for individuals to buy out their debt b. Provision of a business asset c. Access to a low cost savings account d. Financial literacy training

Transitory (.8) as opposed to .44 (permanent)

What's higher, marginal propensity to save each additional dollar of transitory or permanent income?

All of these conceptual explanations have been proposed for why the poor may not save. As Professor Banerjee details though, not every explanation is supported by the data.

Which of the following conceptual explanations are provided for why the poor, such as the fruit vendors in Chennai, may not save? (Check all that apply) a. Inability to cut back on consumption b. Vendors discount the future a lot c. Vendors may not understand compounding and the benefits of saving d. Vendors don't have access to savings

C, D; If people do not exhibit hyperbolic preferences, they would have no need for a commitment device since they are fully capable of acting in a time-consistent manner without one and saving the right amount. Moreover, even if people did exhibit hyperbolic preferences, they may not demand a commitment device if they are unaware of these inherent behavioral biases since they would not see the need for a commitment device in that case. It is only when people exhibit hyperbolic preferences and are aware of these preferences that we see demand for commitment devices.

Which of the following conditions need to be in place for there to be demand for commitment devices? (Check all that apply) a. People should have non-hyperbolic preferences b. People should not be aware of their time-inconsistent behavior c. People should exhibit hyperbolic preferences d. People should be aware of their time-inconsistent behavior

A, C, D; If past savings were high, say because of past economic conditions, then households may choose to forego some current income opportunities while still maintaining a high level of consumption, making the propensity to save seem lower than the truth i.e. we would get a biased estimate since income and savings are jointly determined. Moreover, a more patient person may both prefer higher savings and be able to maintain a higher-earning job, in which case an omitted variable (time preferences) affects both income and savings. Measurement error in income would attenuate the OLS estimate of savings, further affecting our causal interpretation.

Which of the following might be reasons why the regression of savings on total income cannot be interpreted causally? (Check all that apply.) a. Income and savings are jointly determined b. Income and savings are not correlated c. There may be omitted variables that affect both income and savings d. There may be measurement error in income

B; Say an individual received $1 in period 1 and $1 in period 2. If the individual is indifferent between these two propositions, the marginal utility of consumption in the two periods will be the same i.e. 1*u'(c1) = 1*βδu'(c2), or u'(c1)/u'(c2) = βδ. Thus, the relative valuation of $1 in period 1 versus period 2 is given by βδ. Similar exercise for period 2 and 3 shows that the relative valuation is only δ. This discrepancy is at the crux of the time inconsistent preference of the individual -- today is extra valuable relative to the future by a factor of β

An individual's utility maximization problem is given by the following (where the subscripts represent the time period): Maxu(c1) +βδu(c2) +βδ2u(c3) What represents the individual's relative valuation of utility in period 2 versus utility in period 1? a. u(c1)/u(c2) b. βδ c. βδ2 d. δ

A; Although the poor exhibit high take-up of savings products, particularly low cost ones, usage of these products is oftentimes very low. This indicates that people likely recognize the need to save, but lack the ability to overcome their own time-inconsistent preferences and actually save in the account that they open.

Demand for savings products is ________________ and usage of these products is oftentimes quite ________________. This provides some proof that people recognize their own inconsistencies, but then fall prey to them. a. High; low b. Low; high c. High; high d. Low; low

A; The Philippines study finds that preference reversal i.e. hyperbolic discounting behavior predicts take-up for females only.

Does preference reversal predict take-up of the commitment savings product? a. Yes, but for women only b. Yes, but for men only c. Yes, for both men and women d. No

C; The bank accounts in the Dupas and Robinson study do not offer any interest. Furthermore, the bank partner in the study does not offer any loans linked to the savings account. Dupas and Robinson argue that by allowing the poor to build up a buffer stock of capital, the savings accounts help in dealing with shocks so that the poor do not have to sell off their assets, or reduce consumption.

Dupas and Robinson find that people offered savings accounts do not reduce consumption by as much when faced with shocks. Their hypothesis for why this may be is that: a. People are able to borrow from the banks at which they save b. Interest earned from the bank account allows the people saving to grow their wealth c. Better buffering i.e. the bank account allows savers to maintain a buffer against shocks d. All of the above

B; Individuals in the study played a series of games whereby they were offered money in three time periods (today, tomorrow, and the day after) and asked how much money they would give up in future periods in order to get money in the present period and tomorrow. Using this data, the researchers tried to estimate what the discount rates were for the sample and who the hyperbolic discounters were.

How did the researchers identify which clients exhibited time-inconsistency in the Tying Odysseus to the Mast study sample? a. By inferring the hyperbolic discounters from those that took up the commitment savings product b. By playing a series of games in which people made choices between rewards in different time periods c. The researchers were unable to identify who exhibited time-inconsistent preferences d. By directly asking people whether they thought they exhibited time inconsistent preferences

False; A regression of savings on total income mixes up both transitory and permanent components of income since only total income is considered on the right hand side of the equation, not the disaggregated components of income. In the best case, such a regression will give a weighted average of several different propensities that depends on whether income components happened to be permanent or transitory.

Paxson (1992) attempts to estimate the marginal propensity to save out of transitory income. She states, "finding that these marginal propensities are high would indicate that farmers do use savings to smooth consumption." (p. 15) 1. True or False: Although a regression of savings on total income cannot be used to estimate the marginal propensity to save out of total income, it is still possible to use such a regression to estimate the transitory and permanent components of income. True False

B, C; On page 18 of her paper, Paxson says that in this study, transitory income "consists of deviations from average values of regional rainfall in each of four seasons (plus deviations from averages squared.)" Note that Paxson also includes year effects in calculating the predicted value for transitory income.

Paxson obtains a predicted value for transitory income in a similar way to the predicted value for permanent income. She multiplies the transitory characteristics by their respective coefficients from the regression that you estimated earlier and then adds them up to form inctrans (see equation 3 on page 18 of the paper.) Which of the following variables are considered to be transitory characteristics? (Check all that apply.) a. The land ownership variables b. The variables for deviation of rainfall from the mean c. The variables for deviation of rainfall from the mean squared d. The three "save" variables

B, D; While the point estimate on the change in savings for the marketing treatment group is high, it is not significant. Savings in the control group are found to be quite small, meaning that the increase in savings observed in the commitment treatment group is quite large in proportional terms. Although less than half as many people choose the amount goal as the time goal, those in the amount goal group save much more.

Which of the following results are seen in the Philippines commitment savings study? (Check all that apply) a. Balances after 6 months are significantly higher in the marketing treatment group b. Balances after 6 months are significantly higher in the commitment treatment group c. Savings go up in the control group as well as the two treatment groups d. Fewer people choose the amount goal than the time goal