The Evidence Base

Informing Policy in Health, Economics & Well-Being
A collaboration with
USC Dornsife Center for economic and social research

A Little Push Can Make a Big Difference: Temporary Savings Promotions Help the Poor Earn More in the Long Run

In spite of recent economic growth, over 700 million people throughout the world still live in extreme poverty. Identifying cost-effective ways of helping the poor is therefore a critical priority for global development. One increasingly popular policy option is to give unconditional grants – either in kind or in cash – to low-income beneficiaries. These grants can have big impacts: for example, in Ghana researchers gave entrepreneurs a one-time grant of $120 – over the next year, the average entrepreneur was earning $20 more per month compared to a control group that did not receive grants. To put this in perspective, investing that $120 in the U.S. stock market would yield a total return of just $8.40 after one year (assuming a long-run average return of 7 percent per year). This begs the question: if entrepreneurs have access to such high returns, do they really need a big grant to kick start business growth? Or could a lower-cost intervention encourage them to grow their businesses on their own?

In a recent study, I ask whether short-term – but very high-powered – interest rates on bank savings can help the poor catalyze business growth. To do this, I conducted a randomized controlled trial in rural Kenya with 1,558 low-income individuals. Everyone who participated was invited to open individual and joint bank accounts (individuals participated in the study with their spouses); these bank accounts were randomly assigned temporary “promotional” interest rates of 0, 4, 12, or 20 percent. These interest rates lasted for just 6 months – after that time, the new bank accounts offered no interest at all. What I find is striking: although the average interest payment was roughly $0.50, people who received 20 percent interest on an individual account earned $18 more per month 3-4 years after the experiment. This is a big deal in my study setting, where average monthly income in the 0 percent interest group was just $70.

Why did the interest rates have such persistent effects on income? To answer this question, I use administrative records of take-up and use of the experimental bank accounts, as well as detailed survey data on savings behavior, bank savings, other assets, and income sources 3 and 4 years after the experiment. First, I document what happened in the short-run (the 6 month promotional interest period). Here, I find that participants were more likely to open and use their accounts when there was a high promotional interest rate. The impacts on saving at six months were modest, though: moving from a no-interest account to a 20 percent interest account increased the 6-month closing balance of individual accounts by $1.31 and of joint accounts by $2.08.

Next, I decompose the long-run (3-4 year) effects, which turn out to be entirely driven by entrepreneurship. Study participants who received the highest interest rate on their individual account were significantly (10 percentage points, or a 26 percent increase from the mean) more likely to be entrepreneurs and had substantially more business profit and capital at follow-up. In contrast, I don’t find any significant impact on other sources of income, like wage or agricultural earnings.

I also find clear evidence that higher joint interest rates don’t have the same positive impacts on business outcomes. It’s important to note that in Kenya, businesses are typically individually owned and operated. Moreover, I find some suggestive – though less statistically robust – evidence that couples who received higher joint rates invested more in “joint” assets like livestock and the home, and also reported greater levels of spousal agreement regarding financial decisions. This suggests that the individual and joint rates may have primed individuals to invest in different types of assets over the long run.

Customers participate in a drawing for interest rates.

I then use long-run changes in income and assets to estimate a return to capital. Although the rates of return I estimate are high (19-21 percent per month), they are in line with other estimates from cash grant experiments like the Ghana study. What’s more, when people have access to such high returns, very small short-run investments can easily generate my long-run treatment effects, provided people continuously reinvest the proceeds of the investment. Without this sustained behavior change, it is very difficult to explain my results.

But why would the temporary interest rates help people permanently change savings and investment behavior? I argue that explicitly incentivizing savings was key: to make this point, I compare the effects of the interest rates to that of a randomly-assigned cash payment, which did not explicitly incentivize savings. Twenty percent of study respondents were selected for this “cash prize”, which ranged from $0.13-$3.75 with an average value of $3.09. The prizes were delivered within three months of study enrollment. Although the cash payment increased short-run bank account balances more than the interest rates, it had no detectable impact on long-run economic outcomes. This implies that that the effects of the interest rate cannot be explained by changes in the short-run capital stock alone.

Instead it seems that explicitly incentivizing study participants to save in the short run was important because it helped them establish new savings and investment practices that persisted beyond the expiration of the promotional interest rates. For example, one potential explanation is Gary Becker and Kevin Murphy’s theory of habit formation. As participants engaged in savings and investment during the promotional interest period, they may have developed a savings “habit” that made it easier to keep saving and investing in the future. Alternatively, the individual interest rate may have helped entrepreneurs set aside more money that was explicitly earmarked for business, as suggested by Richard Thaler’s theory of mental accounting. The patterns in my data suggest that both of these mechanisms may be at play.

The experiment’s results are exciting because they show that individuals with high returns to capital have the capacity to improve their economic situation without sustained external assistance or a large, expensive “big push” intervention like a cash grant – rather, these individuals simply need the “right push”.