The Power of Practice: How We Doubled People’s Savings

If you’re a policy maker launching a new program, then you might have to consider how to get your target audience to take up and engage with the program. That is, you’re pondering how to encourage a new behavior. Oftentimes, the challenge is that even when the target audience says that a program is desirable, people’s actions often don’t reflect their intentions.

When new programs struggle to catch on, it might appear that it is because they failed to convince people of the merits of program, or because they failed to provide enough information. However, we question whether in some cases, the problem is actually a lack of fluency rather than a lack of information or desire.

To answer this question, the CAH World team conducted a randomized controlled trial in Kenya. Participants included 220 new and expecting mothers who were being enrolled into a subsidized health insurance program. In this program, each mother was given one year of free health insurance for her and her family, but could extend the health cover for a second year if she saved a total of 3,000 Kenyan Shillings (KSh) (about $30) within one year. Savings were made in M-TIBA* which is a digital wallet similar to Venmo or Paypal, but exclusively for healthcare transactions.

Our intervention

The intervention that we tested involved helping participants to practice depositing money into M-TIBA during on-boarding. Mothers randomly assigned to the treatment group practiced making four deposits. After each deposit, an onboarding agent instructed the participant to notice a confirmation SMS sent to their phone and provided positive reinforcement, e.g., “nice job, you’re remembering how to save on your own!” Participants were randomly assigned to receive the practice intervention or not, but all participants were given the same explanation and information about how to make a deposit. We also manipulated the size of the total amount deposited and whether the money the participant used to deposit was the participant’s own money, or was “gifted” money.


Overall, the practice intervention nearly doubled savings behavior. Within 3 months of the intervention, the proportion of individuals who made 1+ deposits increased from 24% to 46% and the total average amount saved increased from KSh 64 ($0.64) to KSh 138 ($1.38).

Interestingly, the practice intervention was found to be even more effective when participants deposited their own money versus gifted money. Mothers who practiced by depositing their own money saved an average of KSh 174, while those who practiced by dy depositing gifted money saved KSh 113 (excluding any money saved on the day of on-boarding). There was something particularly motivating about choosing to invest one’s own resources in the practice.

What could explain this?

We think effects such as the self-herding bias or the sunk cost fallacy may have played a role here. Still, it’s worth pointing out how the result runs counter to what we’d expect if humans were perfectly rational. Those who used their own money to make the practice deposits are immediately “poorer” compared to those who used gifted money. That is, the group that used their own money voluntarily “spent” some of their available money. The group that used gifted money didn’t spend any of their available money. Given this, you would expect the “poorer” group to save less than the group who hadn’t spent any money yet. Yet, here we see the opposite: spending money on savings leads to even more savings.

Why was the practice intervention effective?

It seems that practicing increased fluency in the steps of depositing as well as confidence in being able to save. At the end of the onboarding interaction, we asked participants if they could recall the account number required to make a deposit in M-TIBA. Those who did not practice making deposits recalled this number accurately 55% of the time compared to 93% of those who practiced. We found a similar difference between the two groups’ confidence that they could meet their savings goal.

This experiment has important implications for policy makers who are considering how to encourage users or citizens to adopt a new behavior. While ensuring users have adequate information and a desire to perform the behavior is important, we argue that these aren’t enough. Our experiment illuminates one way in which a program can help users become fluent and confident in a desired behavior. Importantly, our evidence shows that this intervention can have significant impact.


Are you interested in this partnership? Contact Matt Bodien, Joep Lange Institute Partnership Lead, at

Is your organization interested in CAH Global? Contact Jan Lindemans, Principal of CAH Global, at


* M-TIBA is a digital savings platform developed and operated by the PharmAccess Foundation and CarePay International, and supported by the Joep Lange Institute. All organizations are partners to the Center for Advanced Hindsight.