The impact of digital credit in low-income countries

Article

Published 10.06.22

Digital credit does not appear to systematically improve lives, while the lack of transparency raises serious concerns about predatory lending

Digital credit has exploded in popularity in the past ten years. Digital credit generally refers to small, short-term loans that can be accessed instantly, automatically, and remotely, offering borrowers access to funds even without a formal credit history. These loans are often associated with the use of non-traditional data (such as mobile money transaction history) for credit scoring (Berg et al. 2018) – meaning decisions may be made by an algorithm rather than a loan officer – and operate in a range of financial systems and markets with limited regulatory oversight. The loans tend to have very high interest rates, especially when annualised. Yet, little is known about how these digital loans impact borrowers.

While the enormous demand for digital credit shows that millions of consumers need sources of easy liquidity, and though financial access is viewed as a crucial component of economic development (Fonseca and Matray 2021), many consumers are not aware of the loan terms and many end up repaying late (thus incurring fees) or defaulting (thus hurting their future ability to borrow). Other harmful impacts include debt traps, debt stress, data privacy, and coercive repayment tactics, which have led observers to call for regulation. In India, for example, the Reserve Bank banned hundreds of apps from the Google Play Store.

To examine this issue, the Center for Effective Global Action (CEGA) at the University of California, Berkeley, launched the Digital Credit Observatory in 2016, with support from the Bill & Melinda Gates Foundation, to generate rigorous evidence on the effects of digital credit, both positive and negative, on borrowers. To date, the evidence suggests that effects are modest and meet neither the worst fears discussed above nor the transformative effects touted by proponents. While the evidence base is still scant and comes from only a handful of completed studies, no study finds substantial negative effects, while one study finds improvements in household risk-coping ability and another finds modest effects on self-reported financial well-being.

There remains serious reason for concern, however. In Malawi, we find that consumers are uninformed about loan terms and regularly pay back late, incurring late fees. This lack of knowledge creates a clear opportunity for exploitative lending.

Welfare impacts

There have been three rigorous evaluations of the welfare impacts of digital credit. The first, Suri et al. (2021), studies Safaricom’s M-Shwari product in Kenya. M-Shwari offers loans with a 7.5% interest rate and a 30-day repayment period. The authors focus on how access to M-Shwari affects consumption smoothing. In these settings, shocks are common: 89% of people in the study sample reported experiencing a shock in the prior six months, and 68% of those individuals reduced expenses to cope. To evaluate the impact of M-Shwari, the authors use a regression discontinuity design and compare customers who have credit scores that just barely qualify them for loans to others whose scores keep them ineligible. The authors find a statistically significant improvement in one measurement of resilience: those over the threshold are 6.3 percentage points less likely to forego expenses in response to a shock. This effect is real but somewhat modest because it implies that 60% or more of people with access still must forego expenses to cope.

Our study (Brailovskaya, Dupas and Robinson 2021) is the second evaluation, and it investigates a similar product – Airtel’s Kutchova digital-credit product in Malawi. This loan product is even more expensive than M-Shwari, with an effective interest rate of 10% over just 15 days. We evaluated the product just as it was being launched. During this period, the lender only offered small loans, valued at about $1.30 (though borrowers could take out loans repeatedly, conditional on successful repayment). 

We use a strategy very similar to Suri et al. and compare people just above and below the credit scoring threshold. We find limited effects on most outcomes, including debt, savings, and measures of financial well-being. Unlike Suri et al., we find no effect on risk-coping. However, we collected some descriptive information on people’s perceptions of the product and found that the vast majority of respondents liked the product and do not regret taking out the loan.

The third completed impact study was conducted by Björkegren et al. (2021) in Nigeria. It differs from the other two in that it evaluates a digital smartphone lending app in which loans are made by a fintech firm rather than a bank. The interest rates vary between 1.5% and 20% per month. The authors used a randomised control trial with two arms. The first varied whether people would be offered any loan, by dropping the minimum credit score requirement for some new loan applicants. The second varied the size of the initial loan offered to new loan applicants, with values ranging from roughly $3 to $36. 

The authors find that the treatment group had an improvement in financial health, based on a standardised 14-question financial health index. The authors analysed effects along several welfare dimensions but found little evidence of significant positive or negative loan impacts (other than an increase in subjective well-being).

Consumer knowledge

While the effects appear muted, there is still cause for concern. In our study in Malawi, people appear to have very limited information on the details of the products. In our surveys, only 29% know the exact fee, only about half know when the loan is due, less than half know that there is a late fee, and over a third report that they do not know what would happen if they do not pay back on time. 

Repayment behaviour is also concerning because the majority of people pay back late: of new borrowers, only 38% pay back fully on time, while another 47% pay back fully, but late. These borrowers are harmed because they end up paying sizeable late fees: the average interest rate for these individuals is 27% in our data, compared to the 10% fee if paid back on time. Making matters worse, lenders, including Kutchova, change fees but do not transparently disclose the information to borrowers. In particular, the late fees charged were much higher than those posted on the terms and conditions webpage.

We conducted a randomised evaluation of a financial-literacy programme that used interactive voice response (IVR) technology to inform customers of the importance of understanding Kutchova loan terms. The gamified 15-minute IVR module led a listener through a hypothetical example of how to make borrowing decisions. The module was developed by the research team and based on the terms and conditions provided by Airtel at the July 2019 relaunch. The effect of the IVR module was compared to that of other interventions, including a much shorter ‘salience’ module that only informed customers of the existence of the product, a text message treatment, and pure control.

The results were surprising. While we expected the IVR module to reduce demand for digital credit (by making the expense of the product more prominent), we found the reverse: treatment respondents were more likely to take up the product. We attribute this to the reality that credit, in general, is expensive in Malawi (interest rates on other loans such as from savings groups are at similar rates), and so consumers are willing to pay such rates. In fact, about 67% of those surveyed after the IVR module reported that the loans were actually less expensive than they had previously thought, and 62% reported they were more likely to use Kutchova (whereas 20% reported they were less likely to use the product).

We also examined loan default behaviour and found that the treatment marginally improved loan repayment (i.e. decreased default), suggesting the intervention could be beneficial to the lender as well. However, the overall probability of at least one default increased; while people were less likely to default on any given loan, they took out more loans and thus were more likely to default on at least one.

Conclusion

The global expansion of digital credit is providing millions of historically unbanked individuals with access to formal loans for the first time. On the one hand, we might hope that this transformation could empower populations to make productive investments, smooth consumption in the face of income volatility, and generally give people more autonomy over their financial decisions. On the other, high interest rates and opaque loan terms could lead to systematic over-indebtedness and create financial distress.

The evidence so far is mixed. While the measured effects are small, the demand for the products suggests that these loans are providing a valuable financial service. We do not see evidence of transformative effects, but these loans may play a small beneficial role for people whose alternative source of credit is too expensive.

While these studies show no evidence of the worst fears around digital credit being realized, there is still reason for concern. First, these are only three studies with relatively reputable lenders; there is no doubt that there are many harmful lenders in the digital credit space. Second, there is overwhelming evidence that borrowers are unaware of terms and that lending terms are both opaque and regularly changed by lenders. There are clear opportunities for predatory lending in a market with high demand, poorly informed consumers, and little to no regulation (Garz et al. 2020). 

Finally, the promise of digital credit is far from being realised. Short-term high-interest consumer loans may serve some small purposes, but they have no role in financing education, business, home purchase, or any other major investment. Future product innovation and regulatory development will be necessary for digital credit to fulfil its promise.

Editors' note: This column first appeared on VoxEU.

References

Berg, T, V Burg, A Gombović and M Puri (2018), “Digital footprints and credit scoring”, VoxEU.org, 24 August.

Björkegren, D, J E Blumenstock, O Folajimi-Senjobi, J Mauro and S R Nair (2021), “Welfare impacts of digital credit: A randomized evaluation in Nigeria”, Center for Effective Global Action, unpublished.

Brailovskaya, V, P Dupas and J Robinson (2021), “Digital credit: Filling a hole, or digging a hole? Evidence from Malawi”, CEPR Discussion Paper 16848.

Garz, S, X Giné, D Karlan, R Mazer, C Sanford and J Zinman (2020), “Consumer protection for financial inclusion in low and middle income countries: Bridging regulator and academic perspectives”, Annual Review of Financial Economics 3, submitted.

Fonseca, J, and A Matray (2021), “The real effects of banking the poor: Evidence from Brazil”, VoxEU.org, 14 December.

Suri, T, P Bharadwaj and W Jack (2021), “Fintech and household resilience to shocks: Evidence from digital loans in Kenya”, Journal of Development Economics 153: 102697.