Microfinance and credit behaviour — Behavioural Econ Module 6

Microfinance — providing small-scale financial services to low-income people — has been one of the most-discussed development interventions of the past forty years. Initially heralded as a poverty-reduction breakthrough, the empirical record is more nuanced. This module covers the structure, the empirical evidence, and the recent transition from group-lending microcredit to algorithmic digital credit.

What microfinance is

The term covers multiple services for low-income clients:

Microcredit — small loans (typically $100-$5,000) to individuals or small businesses lacking access to commercial banks. The most-discussed component
Microsavings — small savings products (typically lock-savings, group savings) for low-income clients
Microinsurance — small-coverage insurance (covered in module 5)
Mobile money / payments — covered in the next module

Grameen and the group-lending model

Muhammad Yunus established Grameen Bank in Bangladesh starting 1976. The core innovation: lending to groups (typically 5-6 women) where the group was collectively responsible for repayment.

Joint liability — if one member defaulted, the group was on the hook for repayment. This created peer pressure to repay
Step-by-step lending — start small, demonstrate repayment, get larger loans
Regular meetings — weekly group meetings build social capital and reinforce repayment
Female focus — women were the primary clients; empirically showed better repayment than men in the target population
Doorstep service — staff visited borrowers rather than requiring trips to the bank

The model produced repayment rates of 95%+ in many programmes — extraordinary for credit to ultra-poor populations. Yunus received the Nobel Peace Prize in 2006 for the work.

Why group lending works (and why it sometimes doesn't)

Theoretical mechanisms:

Information advantage — group members know each other's circumstances better than the bank does. Bad-risk applicants are filtered out at the group-formation stage
Peer monitoring — members observe each other's behaviour and can flag problems early
Peer pressure — social cost of defaulting on the group exceeds the financial cost of repayment
Joint liability — explicit incentive for everyone to ensure everyone pays

Group-lending failure modes

The model's elegance hides real failure modes: • Correlated shocks — if all group members face the same shock (drought, market collapse), the group's joint-liability becomes meaningless; everyone defaults together • Free riding — sophisticated members extract value from the group while contributing less; group monitoring weakens • Punitive enforcement — peer pressure can become abusive in some contexts. Reports of intimidation, public shaming, social ostracism • Coordination cost — weekly meetings are costly to members in time and travel. Some members value the time at more than the access • Scale limits — group lending requires high-density social networks. Works in rural Bangladesh, less in urban contexts where social ties are weaker

The microcredit impact debate

The early enthusiasm assumed microcredit was transformational — poverty exit, women's empowerment, business growth. The empirical record has been less dramatic.

The six-country RCT (Banerjee et al. 2015)

Coordinated randomised evaluations of microcredit programmes in Bosnia, Ethiopia, India, Mexico, Mongolia, Morocco. The consistent finding:

Consumption effects — modest, typically 0-10% gains among borrower households
Business creation — small positive effect; perhaps 1-2 percentage points more business owners in treatment groups
Poverty exit — near zero effect across all six countries. Microcredit doesn't move households out of poverty
Women's empowerment — modestly positive in some metrics; not transformational
Heterogeneity — substantial heterogeneity in effects; some borrowers benefit substantially, others not at all

This wasn't the 'microcredit doesn't work' verdict the press reported. It was the 'microcredit doesn't do what early advocates claimed' verdict. Microcredit is a useful product for cash-flow smoothing and occasional business investment; it's not a poverty-reduction tool.

The shift to digital credit in East Africa

Starting around 2012-2014, mobile-money-linked credit transformed African lending markets. M-Shwari (Safaricom + NCBA, 2012), Tala, Branch, Fuliza, others have collectively issued tens of billions of dollars in instant loans.

Digital-credit mechanism

Mobile-data-based credit scoring — the lender analyses your mobile-money usage, phone data, voice/SMS patterns, and social-network features to score creditworthiness
Instant approval and disbursement — loans approved and disbursed in seconds via mobile money
Small ticket sizes — typically $5-$200 per loan
Short tenor — 7-30 days typical, occasionally up to 90 days
High interest rates — typically 7-15% per month or higher (90-180%+ annualised)
Automated collection — repayment auto-debited from mobile-money balance on due date

What digital credit gets right

Real access — millions of users who never had bank credit now have a credit option
Fast — solves the immediate cash-flow problem (medical emergency, school fees due) that informal credit also addresses
Trust through automation — predictable, transparent, automated systems
Scalability — the algorithmic approach can serve tens of millions without proportional human staff

What digital credit gets wrong

The digital-credit dark side

High default rates and consumer-harm patterns emerging: • Default rates 10-25% across the major Kenyan providers — much higher than traditional credit • Cycle of borrowing — many users borrow continuously, using new loans to repay old ones (effectively rolling-over debt) • Hidden costs — the daily-interest framing (covered in module 2 exercise) makes the true cost less salient • Mental-health impact — high-frequency notifications about repayment can trigger anxiety, particularly for people with mental-health vulnerabilities • Vulnerability targeting — algorithm may over-target customers who are likely to default but generate sufficient interest payments before doing so • Privacy concerns — extensive data collection on phone-data, social networks, location • Regulatory gaps — many digital lenders operate outside the regulated banking system, with weaker consumer protections The Central Bank of Kenya's Digital Credit Providers regulations 2022 brought most digital lenders into a regulatory framework. Impact is emerging.

Microcredit vs digital credit — comparison

text

                            Microcredit         Digital Credit
Typical ticket size         $300-$3,000         $5-$200
Typical tenor               6-24 months         7-30 days
Approval time               Weeks               Seconds
Interest rate               20-60%/year         90-180%+/year
Default rate                <5% (group lending) 10-25%
Loan officer relationship   Strong              None
Flexibility on default      Moderate            Aggressive
Use-case                    Investment          Cash-flow shock
Scale                       Tens of thousands   Tens of millions
Regulatory framework        Bank-like           Evolving

The future of African credit access

Hybrid models — digital lenders with longer-tenor products (3-12 months) at lower rates. Tala, Branch experimenting with this product structure
Bank-fintech partnerships — KCB, Equity, NCBA partnering with digital platforms. Brings credit scoring algorithm + regulatory framework + lower cost of capital
Open banking and consumer-controlled data — emerging standards for customers to share their financial data across providers for better credit scoring
Credit-bureau integration — Credit Reference Bureau Africa, TransUnion expanding coverage. Better credit scoring reduces risk pricing for good borrowers
Algorithmic regulation — emerging frameworks for ensuring credit-scoring algorithms don't produce discriminatory or harmful outcomes

Exercise

A Kenyan working professional regularly borrows from M-Shwari (KES 5,000 at 7.5% per 30 days), Fuliza (overdraft at 1.083% per day = 32.5% per month), and KCB-M-PESA (KES 3,000 at 8% per 30 days). She rolls over loans frequently, never holding zero balance for more than 3-4 days. (1) Compute the effective annual cost of borrowing across these products. (2) Estimate the cumulative annual cost as a share of borrowing. (3) Apply behavioural analysis: what mechanisms keep her in this borrowing cycle? (4) What advice would you give her to break the cycle?