One way to improve business training

The evidence on the effectiveness of business training is, at best, mixed (for an example, see my previous post on David McKenzie and Chris Woodruff’s artful review). As David and Chris point out, part of the problem was methods (esp. sample size). But even when the methods were good, the results were often lackluster, particularly for women.

A nice new paper by Erica Field, Seema Jayachandran, Rohini Pande, and Natalia Rigol, provides a counterpoint. And it’s a counterpoint that relies on friends. Field and co. evaluate a quite short business training program run by India’s SEWA Bank for its female clients and find that some significant impacts — all the way through to household expenditure.

The program that they are looking at is a two day training that covers some financial literacy basics and business practices (this is not only a short training, it’s also cheap at about $4 a head). The training focused on financial prudence, including savings. There was also some content aimed at aspirations — including a video of successful SEWA clients talking about how they got out of poverty. Finally, part of the training was for the women to do homework one night consisting of mapping out, on a worksheet, a financial goal they would like to achieve in the next six months. So far: a bit different, but not massively so, from previous programs of this ilk. But here’s the twist:  some of the participants were asked to come to the training with a friend, who would also get the training.

This, of course, was part of the experimental design. Field and co. are working with a fairly small sample: 212 women in the control group, 217 in training without bringing friends, and 207 in the training with friends arm. Field and co. collected baseline and a follow-up survey. They also make use of some of the data from the worksheets mentioned above, as well as the administrative records of the Bank. For estimates, Field and co. run the two training arms separately, but they also present pooled results — which makes sense given the small sample size. (Note that sample size was one of the issues that Chris and David’s review paper talked about (see this discussion) although in this case, I am guessing it was more a budget constraint than a shortage of entrepreneurs).

Before getting to their results, it’s worth discussing one interesting side note on their data structure. The evaluation is stratified by locality and, given implementation constraints, was also rolled out sequentially across locations rather than simultaneously. This means that to look at 4 month effects (which is their endline here), the follow-up surveys have to be spread out over time. To handle the issues this could create in the data, Field and co. assign the control respondents to treatment clusters — not for treatment but for the purposes of data collection. This allows them to cluster the standard errors by training session.

So what do they find? First of all, uptake is good — around 68 percent. The program shifts beneficiary financial behavior — those in the group that brought a friend are seven percentage points more likely to have taken out a loan (off of a control group mean of 5.7 percent) and this loan is more likely to be a business loan. There is nothing significant, in terms of borrowing, for the group that took the training without a friend. Elsewhere in financial behavior one surprising result, particularly given the emphasis on prudence and savings in the training, is that there is no significant impact on savings. Turning to how the business is run, Field and co. find a significant increase in hours worked, but only for the pooled sample. Sales were up, relative to a year ago, for the those who took the training with a friend, jumping by 10.7 percentage points. And those who trained with a friend were also significantly more likely to have plans for expanding their businesses. Putting these results together, Field and co. argue that those who trained with a friend were more likely to borrow for the business and put this loan to work in growing the business.

This business growth translates into better short-term welfare. Women who trained with a friend reported 11.6% higher household income and 15.6% higher expenditure (both significant at 10 percent). For those who trained alone, again, not much of anything. Finally, those who trained with a friend were also significantly (at 10 percent) less likely to report that they were a housewife — 4 percentage points less likely relative to a control group mean of 10 percent.  Caveating the overall welfare conclusions, Field and co. are careful to note that they don’t observe the full cycle of the loan, so they can’t say anything definitive about end welfare outcomes.

Nonetheless, it seems that this kind of training works much better with a friend. What’s going on with this? One of the arguments as to why female entrepreneurs do worse than males is that they have smaller and/or less diverse professional or peer networks (see, for example, chapter 5 in the World Development Report 2012). Part of this might be due mobility restrictions placed on them by their spouse or society. Field and co. tackle this issue with a summary index of restrictions (ranging from socializing to employment). When they interact this with treatment, the results are striking: the loan effects, the hours worked, and the seeing oneself as a housewife effects are much stronger for the restricted women who did the training with a friend. So this suggests that taking this training with a friend is particularly beneficial for women who face more restrictions to their mobility.

Is this simply an avenue to more contact? Field and co. asked the women if they discussed business on a daily basis. While there is an effect of the treatment on this, there is no significant difference for those who took the training with a friend. So it’s probably not working through this. Maybe then there was some kind of reinforcement (from the friend) of the aspirational messages. Field and co. measure confidence, but this shows no significant change.  Perhaps the friend effect is working through goal setting. Here Field and co. analyze the worksheets that the women did during the training. Unfortunately, while the coefficients here are suggestive that those who trained with a friend had different goals, Field and co. are hampered by the fact that they didn’t get data on very many worksheets and the sample is now too small to say anything for sure.

All in all, these are intriguing and exciting results. They suggest that taking into account and programming for the social context of business training may lead to a significant improvement in its effectiveness, at least for women.  Moreover, this type of training may work best for the women who are the most disadvantaged in the market due to the social constraints they face. I am looking forward to further work that explores this type of program in other contexts and also gives us additional insights into the mechanisms through which it works.

This post first appeared on The World Bank Blog

Author: Markus Goldstein is a development economist with experience working in Sub-Saharan Africa, East Asia, and South Asia.

Image: A vendor hands change back to a customer at her market stall with price tags marked in Cuban pesos in Havana October 23, 2013. REUTERS/Desmond Boylan