Geographies in Depth

What’s the link between saving and food security?

Markus Goldstein
Lead Economist, Africa Region and Research Group, World Bank
Our Impact
What's the World Economic Forum doing to accelerate action on Geographies in Depth?
The Big Picture
Explore and monitor how Africa is affecting economies, industries and global issues
A hand holding a looking glass by a lake
Crowdsource Innovation
Get involved with our crowdsourced digital platform to deliver impact at scale
Stay up to date:


People in developing countries, much like people everywhere, save. And in Sub-Saharan Africa, beyond banks, folks save through a bunch of techniques – ranging from the less sophisticated under the mattress savings to the more complex community-based rotating savings and credit associations (ROSCAs). Given this plethora of savings options, one might wonder if an NGO program that set up savings groups but injected no capital or lockboxes or any other capital intensive intervention might make a difference at all.

It turns out it can.

This week I discuss a new paper by Lori Beaman, Dean Karlan and Bram Thuysbaert which joins the burgeoning literature looking at savings programs as a tool for improving consumption and income outcomes (note that the literature isn’t the only thing burgeoning – estimates indicate that savings program have grown from 2.3 million beneficiaries in 2010 to 7.5 in 2013). Beaman and co. work with Oxfam America and Freedom from Hunger to look at the impact of the Saving for Change (SfC) program in Mali. This program works with groups of about 20 women who meet weekly. At every meeting, each woman contributes an amount that was previously agreed with the group (in the case here, it’s about 48 cents) to a communal pool. Members can borrow from the pool, but they have to pay back any loans with interest (a parameter that is also agreed ex ante by the group). One interesting operational tweak is that these groups, given the low literacy rates among adult women in Mali, use an oral accounting system to keep track of things. At a previously agreed upon time, the accumulated pool is paid out to all of the members (the share-out). And then, if members want to, the cycle can repeat.

To look at the effects of the program, Beaman and co. work with the program as it expands to new areas. They randomise across 500 villages (unbalanced across treatment and control to look for spillovers).   And in a nice, operationally relevant, additional tweak they work with a higher NGO effort and lower NGO effort variant of the implementation. Key to the formation of these groups is the agent (think facilitator) who helps form these groups. What this program does is to have a hired agent who trains replicator agents who will later go out and help expand the program. In the high-NGO effort version, the replicating agent got a structured, three day training, a guide book and a certificate. In the low-NGO effort variant, these replicating agents didn’t get any formal training. But in both cases, the hired agent helped out the replicating agents.

Beaman and co. do a neat job of collecting more than one kind of data. They collect the usual panel survey of beneficiaries and control groups (6000 households), but they also collect a set of high frequency data that will allow them to look at things like consumption across time – including through the lean season. One practical note is that while the larger panel survey collects data across the entire experiment area, the high frequency data is restricted to a much smaller number of people in a random sample of villages around centrally located larger towns so that enumerators could actually make it to this sample on a regular basis.

So what do they find? Before getting to the results there are a couple of summary stats worth taking a look at. These folks do not live in a financial void – 22 percent of the women at baseline were members of ROSCAS and 35 percent had received a loan in the last 12 months. About half of them had a business.

So the first indicator Beaman and co. look at is uptake. This turns out to be not so easy – remember these groups aren’t a strictly branded intervention (although, in my experience, the branding might not help).  It turns out that at the end of the meetings for these groups, folks clap.  Hence, you can identify them as “applause groups.” Using this, they find take up of around 30 percent in treatment villages and 6 percent in control. The folks adopting seem to be coming from somewhat wealthier households (as measured by food consumption). So while this intervention does seem to be working with poor people (you’ll see why in a bit) it isn’t the poorest of the poor (given that its savings, this might be obvious). One other interesting characteristic of the adopters is that they are women who score higher on an index of intrahousehold decision making power.

Folks in treatment villages save more, about $3.65 more on aggregate. This is driven overwhelming by increased group savings.  Indeed, savings in formal institutions (which is used by a really small percent of the sample) actually declines. And there also seems to be some significant movement away from other savings groups and ROSCAs towards the SfC groups. Women in treatment villages were also significantly more likely to have received a loan in the last 12 months (about 3.3 percentage points).  Again, there are some compositional shifts at play – there is a big increase in loans from a savings group (+12 percentage points) and a decline in loans from family and friends. Given that they are paying interest on the savings group loans and not to family and friends, you get a sense of the social costs of borrowing from your family and friends. I can relate.

So what happens to economic activities? On agriculture, the results are a bit of a puzzle. There is no increase in output for the household, but there is a significant increase in output on female controlled fields. But Beaman and co. can’t find a significant increase in input costs or in land and in-kind inputs, so they don’t make much of this. There is no significant movement in business profits. Where there is a big bump is in livestock holdings – a 13% percent increase (off of a mean of $896). Given that livestock is both a consumption buffer and productive asset in these households, this is meaningful.

Looking at end-line consumption, Beaman and co. find a small (and significant only at 10 percent) increase. They also look at a chronic food security index, which shows a significant improvement in food security. Part of the explanation for this latter result comes when Beaman and co. turn to the high frequency data collection. When they look at food consumption in the lean season, they find that treatment households manage to offset the decline that affects the control group and thus, the participants in the program are doing a better job of smoothing consumption. (Indeed, the share-out timing chosen by most of the groups seems to line up with this). Turning to a range of other outcomes, there are no significant impacts on health, education, and female decision making power. But there does appear to be a significant improvement in housing quality.

Turning to the variants of program implementation, Beaman and co. find some results that suggest that higher-NGO effort (or the more structured approach as they put it) yields somewhat better results. There might be more savings in the villages with the better trained workers – the coefficient difference is big, but it doesn’t come in make it to 10 percent significance. However, the improvement in food security does appear to be driven chiefly by the higher-NGO effort villages.

Beaman and co. then do a valiant job of trying to tease out the mechanisms by which the program is working. They can’t identify this for sure – as they aptly put it “this experiment focused on having sufficient power to detect impacts of the implementing partners preferred method of implementing SfC.” They can’t rule out or rule in time inconsistent preferences, as the variable they use isn’t well behaved (and they are thorough enough to actually check this and tell us). They do find some evidence that women in treatment villages, at end-line, might be somewhat more patient than those in the control villages. It doesn’t appear to be better risk sharing (remember, there is a lot of transparency and sharing of information as part of both the savings and lending processes in these groups) nor does it appear to be some sort of shift in intrahousehold bargaining power.

So, all in all, another point for savings. This is a fairly cheap intervention – it clocks in around $20 per household. With appreciative improvements in food security and household asset holdings, this might be a good deal.

Published in collaboration with The World Bank

Author: Markus Goldstein is a development economist with experience working in Sub-Saharan Africa, East Asia, and South Asia. He is currently the Gender Practice Leader in the Africa Region and a Lead Economist in the Research Group of the World Bank.

Image: Coffee beans are tossed in the air at harvest.

Don't miss any update on this topic

Create a free account and access your personalized content collection with our latest publications and analyses.

Sign up for free

License and Republishing

World Economic Forum articles may be republished in accordance with the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International Public License, and in accordance with our Terms of Use.

The views expressed in this article are those of the author alone and not the World Economic Forum.

Related topics:
Geographies in DepthFinancial and Monetary Systems
World Economic Forum logo
Global Agenda

The Agenda Weekly

A weekly update of the most important issues driving the global agenda

Subscribe today

You can unsubscribe at any time using the link in our emails. For more details, review our privacy policy.

Asia-Pacific: How the region is prioritizing a green economy

Kanni Wignaraja and Debora Comini

June 10, 2024

About Us



Partners & Members

  • Join Us

Language Editions

Privacy Policy & Terms of Service

© 2024 World Economic Forum