Twenty-five years in economic inclusion and impact investment. One question that kept coming back, never with a satisfying answer: what does it actually cost to move one person from subsistence to economic security — and what does that move generate? This is my attempt to build a traceable, open, defensible answer.
I have spent most of my working life in the space where money meets poverty — financial inclusion, livelihoods, enterprise development, impact investment. I have sat in rooms with governments deciding where to allocate billions. I have sat with philanthropists deciding whether to fund a programme. I have sat with investors asking whether there is a return. And in almost every room, someone eventually asks the same question.
What is the return on investment of this work?
And in almost every room, the honest answer has been: we don't know. Or: it depends. Or: here is a study from one programme in one geography that measured one outcome variable for two years. None of which is a satisfying answer for anyone trying to make a decision that will affect tens of millions of people.
The evaluations that exist tend to stop at the programme boundary. Did income go up? Yes, by 23%, at 18 months. But then what? What did that income gain do to the household's health spending? What did it do to the next generation's education? Did the entrepreneur create jobs? Did those jobs last? Did the government programme that funded this become more efficient as a result of what was learned?
Nobody traces the full chain. And because nobody traces it, nobody can answer the question. And because nobody can answer the question, the argument for investing in economic inclusion — financial inclusion, livelihoods, enterprise development — has always been made on moral grounds rather than economic ones.
That is a weak position. And it is, I think, one reason these programmes remain chronically underfunded relative to their potential.
"If we could show that $1 invested in a nano-enterprise programme generates $X in economic and social value over Y years — with traceable assumptions and open methodology — the conversation with a government budget committee changes. With a pension fund. With a development bank. With a corporate CSR team."
This is my attempt to build that answer. Not a perfect answer — the assumptions are visible, the methodology is open, and I expect people to challenge it. That is the point. But a structured, traceable, adjustable framework that connects an investment to its downstream effects in a way that can be examined, updated, and improved.
The worked example is India — a country I know well, with a government system that already deploys enormous capital toward economic inclusion, with a rich evidence base from NRLM, MUDRA, and comparable programmes, and with an ambition (Viksit Bharat 2047) that makes the stakes of this question concrete. But the framework is designed to be adapted — to any geography, any programme type, any funder.
The framework traces four things simultaneously: what happens to one person, what happens to the people around her, what happens to the local economy, and what happens to the system that deployed the capital. Each is measurable. Each is sourced. Each can be set to zero.
Most programme evaluations measure one outcome at one point in time. This framework measures the full chain — income, jobs, education, and health resilience — across a multi-year horizon, with explicit persistence factors drawn from J-PAL graduation programme meta-analyses and BRAC long-run follow-up studies.
It also separates two distinct mechanisms that existing evaluations conflate. The first is direct: what does the investment produce for the people it reaches? The second is systemic: how does the same investment change the behaviour of government capital that is already flowing — making it better targeted, more effective, more likely to reach the people it was designed to reach?
Both are real. Both are measurable. They answer different questions for different audiences.
The test is simple: is there a credible, sourced way to measure it, and does the measurement produce a number that can be defended in a room with a sceptical economist? If yes, it is in. If not — however real and important the outcome — it is not.
Direct income gain at 18 months, with documented persistence factors across years 1–10. Sourced from BRAC, JEEViKA, J-PAL graduation meta-analyses.
Jobs created per entrepreneur at 18 months, survival rate to year 3, induced multiplier. BRAC (68–74%), J-PAL (65–72%), ILO skills programmes.
Share of income gain reinvested in schooling, lifetime earnings impact per additional school year. World Bank Montenegro & Patrinos (2014).
Annual shock probability, average cost, improved buffer rate. NITI Aayog / MoHFW. Measures households that can absorb a shock without catastrophic debt.
How much more effective does government capital become with better evidence, targeting, and playbooks? Measured as η — the percentage improvement in outcome per government dollar.
Women's agency, community resilience, environmental outcomes, intergenerational mobility beyond schooling. All real, all important. Not in the model because I don't have a defensible measurement approach for them yet.
The programme did not cause all of this. The government built the SHG network. The banks deployed the capital. The community mobilised. The attribution coefficient — the share assigned to the catalytic investment — is the most contested assumption in the model. I use 20% as the base case. The model lets you set it to 0%. If attribution is what you want to challenge, do that first.
Income + Jobs + Education + Health. Each persistence-adjusted across the programme horizon. Attributed at d_attr.
η = system effectiveness improvement. G = government capital already flowing. Measures additional capital reaching people more effectively — not a downstream multiplier.
Adjust investment, reach target, and attribution. See the return per dollar across all four value streams.
Use Story 2. Set government capital to your programme's deployment. See the effectiveness gain required to justify the investment.
The jobs stream and enterprise revenue data sit inside Story 1. The credit unlock data is in the India worked example.
Every assumption links to a source. Every parameter adjusts to your context. The methodology section explains every equation.
The collective income and GDP panels show purchasing power created at scale. 25M women each spending Rs 22,500 more per year is a consumer market.
All sources are listed. The model is open. If you have data that changes an assumption, the contact form exists for that reason.
This framework is deliberately incomplete. The four streams I measure are the ones I can defend. There are outcomes — women's agency, community resilience, intergenerational mobility beyond schooling — that I believe are real and larger than what is captured here. I have not included them because I do not yet have a credible measurement approach. If you do, I want to hear from you.
India deploys approximately $17.5 billion per year on rural women's livelihoods and enterprise finance — through NRLM's SHG bank linkage, MUDRA loans, PM Vishwakarma, SVANidhi, and bank MSME lending. One of the largest livelihoods programmes in the world. Also, by most measures, significantly under-performing relative to its potential.
The gap is not money. The gap is between the capital that is flowing and the women it is designed to reach. Poor targeting. Mismatched products. High delivery costs. Fragmented ministry mandates — agriculture, financial inclusion, women's enterprise, and digital infrastructure each operating in separate lanes, never integrating around the same household.
A catalytic investment — $20 million per year in demonstrations, evidence, TA, and playbooks — can improve the effectiveness of that $17.5 billion system by approximately 7%. That is the system story. The direct story runs alongside it: the same investment contributes to 25 million women having meaningfully higher incomes by 2030, 55 million by 2035, 100 million by 2047.
Bottom 40% = 520M · Per capita <$2,600
377M need to move Q1/Q2 → Q3+ by 2047
Priya is a composite drawn from programme data across NRLM, JEEViKA, and comparable livelihoods interventions. She is the median woman in a well-designed programme. Her journey is what the model's base case parameters produce.
Small food processing unit in Bihar. No bank account in her name. No formal credit. One health shock from a moneylender at 36%. Invisible to the systems designed to help her.
Income up 30%. Cashflow-based enterprise loan. Market linkage at 35% above local price. Input savings from collective buying. One part-time job created for another woman.
FPO board. Direct buyer relationship with an urban retailer. Rs 5,600 more per year in schooling. Household can absorb a Rs 50,000 health bill without a loan.
Trains others. Parametric insurance. The job she created has stabilised. Children staying in school past the point she would have pulled them out.
Rs 2–2.5 lakh. Bank Sakhi serving 200 households. Cooperative profit-share building family assets. The investment ended years ago. She is the system now.
The $4-per-woman figure requires no model and no assumptions. It is simply the catalytic investment divided by the number of women whose incomes improved. At $20M per year over 5 years — $100M total — divided by 25 million women: $4 per woman who moves from Rs 75,000 to Rs 97,500 and stays there.
For comparison: MGNREGS costs approximately Rs 18,000 per worker per year for temporary wage employment with no lasting enterprise. The catalytic investment costs $4 — once — for a permanent income trajectory that compounds for years after the investment ends.
I want this to be crowd-sourced, contested, and improved. The assumptions are visible. The methodology is open. If you have data that changes a number, a programme that belongs in the sources, or a perspective from implementation that the model does not capture — that is what the contact form is for.
The longer ambition is a framework and tool that anyone working in economic inclusion — a programme officer at a foundation, a budget committee in a state government, an impact investor evaluating a deal, a researcher designing an evaluation — can use to answer the question that has never had a satisfying answer. Not just for India. Not just for women's livelihoods. For any income-enhancing intervention, anywhere, where someone needs to explain to a room of sceptical people what their investment is actually worth.
The India rural women's case is the first worked example because it is the one I know best and where the evidence base is richest. It is meant to demonstrate that the framework produces defensible, tractable numbers — not to suggest that India is the only context that matters.
This is a working document. If you have data that changes an assumption, evidence that contradicts a finding, a programme that belongs in the sources, or a perspective from implementation the model doesn't capture — send it.
We'll be in touch if there's a reason to continue the conversation.
Five-constraint framework for women-led nano and micro enterprises. Coordinated ecosystem approach.
Nano-enterprise baseline: avg 7 years operating, 11% new to credit. $150B credit demand. Credit impact: household income +11%.
Growth-oriented vs subsistence segmentation. 7.7 crore nano-enterprises; small fraction has adequate formal credit.
Dvara ↗Sole proprietors: avg outstanding Rs 3.5L, flat for 2 years despite 7%+ GDP growth.
97.25% of MSMEs are micro enterprises. 6.3 crore+ total enterprises.
SHG-BLP: Rs 2.59L crore outstanding. ~Rs 80–90K crore annual disbursement.
NABARD ↗Income persistence and job survival at 18–36 months. Conservative persistence: 62% of gain at year 6.
68–74% job retention at 36 months. Long-run income trajectories to year 10. Source for φ_J = 0.70.
70–75% job survival for skills plus asset transfer programmes.
1 direct job per $4K deployed. Jobs-per-entrepreneur parameter basis.
+11.7% lifetime earnings per additional school year. Education stream present value basis.
Women reinvest 25–35% of income gains in children's education. Base case: 25%.
18% annual health shock probability. Average cost $600. Out-of-pocket expenditure 39.4%.
India's DPI doubled financial inclusion 2011–2021.
IMF ↗100M+ women in SHGs. 90L SHGs. FPO women 21.96L of 56.3L farmers. Sakhis: 48K BC + 3.5L Krishi/Pashu (Jun 2025).
VB targets: $30T economy, $15–18K per capita, zero poverty, 70% FLFPR. 377M need to move Q1/Q2 → Q3+.