SCALEX LABS | GIDE | BLOG
My journey from farm household economics to gender economics — and what it means for development evidence
Alfred Latigo
Founder and Chief Innovation Architect, ScaleX Labs and the Global Institute for Development Evidence (GIDE)
October 2026
My interest in gender economics began with household economics, studied within agricultural economics as an undergraduate. The starting point was the farm household: how it allocates land, labour, money, and time, and how those choices shape production and well-being.
Over a career at FAO, the UN Economic Commission for Africa (ECA), the World Bank, and the African Development Bank, that starting point led to a sharper question. When an economy grows, or a project succeeds on paper, who actually gains the ability to make economic choices? Artificial intelligence enters this story later, as a possible aid to better analysis. The economic reasoning must come first.
This is a story about the invisible half of the economic ledger — the work, time, and decisions that conventional measures too often overlook.
What gender economics is
Gender economics studies how gender shapes economic behaviour, opportunities, and outcomes — and how markets, institutions, and policies in turn reproduce or transform gender relations. It examines gender as a factor shaping the distribution of resources, work, and decision-making power, extending analysis beyond the disaggregation of data.
Its scope runs from paid and unpaid work, wages, and the care economy, through bargaining inside households and access to land, credit, and technology, to time use, taxation, public spending, and economy-wide modelling. Its distinctive contribution is to link constraints experienced by individuals and households to wider economic outcomes.
The discipline also requires care in reading differences. An earnings or yield gap is a starting point for investigation: it may reflect occupation, hours, care responsibilities, unequal opportunities, discrimination, or several of these together. Claudia Goldin’s research, recognised by the 2023 economics prize, shows how women’s labour market outcomes are tied to structural change, education, expectations, and family responsibilities [1].
The household is an economic institution
A smallholder family can be producer, consumer, employer, labour supplier, and bearer of risk at once. Agricultural household models showed why the effect of a farm policy depends on how families earn, spend, and allocate resources. Where credit, insurance, or labour markets are missing or imperfect, production and consumption decisions may become inseparable [2].
Gary Becker helped establish that households combine purchased goods with time, so time has an economic cost even when no wage is paid [3]. Consider two farmers attending the same training course. The fee is identical, but one must arrange childcare, walk further, and finish domestic tasks afterwards. Count only the fee, and her lower attendance may look like lower interest rather than a higher cost.
This is where gender enters. A household may look short of labour while a woman is already working very long hours on childcare, food preparation, water collection, and several plots. Ester Boserup’s pioneering 1970 work drew attention to women’s agricultural contribution and to development approaches that overlooked it [4].
Opening the household black box
Treating the family as one decision-maker with one purse can hide the mechanism a policy needs to change. Bargaining approaches ask how each member’s assets and alternatives — secure land, independent earnings, or a credible livelihood outside a relationship — shape outcomes. Collective models recognise multiple preferences and often assume that household allocations are efficient. Yet efficiency does not establish fairness, and even efficiency needs empirical examination.
Christopher Udry’s study in Burkina Faso found resources misallocated across plots controlled by different household members, costing about 6 percent of output in that setting [5]. The figure is specific to its context; the wider lesson is that looking inside the household can change the economic diagnosis.
Counting what national accounts leave out
At ECA, my work turned to a larger blind spot: the limited visibility of unpaid domestic and care services in national accounts and economic policy. Our 2004 issues paper, The Missing Link in Growth and Sustainable Development, prepared for the ADB/ECA symposium in Kampala, made the case for recognising household production and integrating it into economic analysis and policymaking [6]. Our subsequent work examined time-use studies as a foundation for that effort [7].
An accounting distinction matters here. National accounts already include some production by households, including goods produced for their own use and imputed housing services. Unpaid domestic and care services for a household’s own use are generally excluded from GDP. Satellite accounts can make these services visible while maintaining a clear relationship with the central accounts and avoiding double counting [17].
In South Africa, our work connected the 2000 Time Use Survey with economic accounts to develop a framework for household production and gender-aware economy-wide analysis. The historical satellite-account estimates put household production at about 44 percent of the extended economy, under the study’s definitions and valuation methods [8]. That estimate is neither a current national statistic nor an amount that can simply be added to GDP: overlapping components require adjustment.
The modelling also examined how trade liberalisation could produce different outcomes for women and men [8]. Bringing unpaid work and time allocation into the analysis made it possible to investigate effects that a model without those features could not identify.
In 2026, Omar Abdourahamani and I returned to this work, reconstructing the model from the published record and using AI-assisted methods to explore policy scenarios and sensitivity to assumptions [9]. Our reported simulations suggest that easing women’s time constraints needs to be accompanied by better access to remunerative employment if the gains are to translate into a stronger share of labour income. These are conditional model findings, based on historical data and specified assumptions, rather than measured effects of policies implemented today.
Valuing unpaid work involves choices. Replacement cost asks what an equivalent service would cost to purchase; opportunity cost asks what the person could otherwise have earned or done with that time. Prevailing wages can carry existing inequalities into the estimate. Valuation helps make resource use visible so that decisions about infrastructure, services, and the sharing of work are better informed.
Agriculture and climate make the mechanisms visible
FAO reports a 24 percent productivity gap between women and men farmers on farms of the same size [10]. Equal size does not mean equal land quality, inputs, or services, so the gap should not be interpreted as an inherent difference in ability. Imagine a project distributing improved seed. It records how many women receive seed and whether yields rise. The economic questions go further: who decides where it is planted, who supplies the extra labour, who sells the output, and who controls the proceeds?
Climate shocks sharpen the same questions. FAO’s The Unjust Climate found that rural female-headed households in the low- and middle-income countries studied suffered greater income losses from heat stress and floods than male-headed households [11]. Household headship does not, however, reveal how losses are distributed among individuals within a family.
A water investment may cut walking time and raise irrigated output. Whether the time saved becomes paid work, rest, schooling, or another unpaid task depends on access rules, household decisions, and available opportunities. Climate resilience therefore involves allocation as well as technology: who can use resources, manage risk, and sustain a livelihood when conditions change?
From diagnosis to evidence
This is where gender economics meets development practice. The Africa Climate Change Fund (ACCF), hosted by the African Development Bank, has supported a gender focus in climate adaptation and launched its fourth call on Gender-Transformative Methane Reduction in October 2024 [12, 13]. In my work on these approaches, the economic questions remain central: how does a project change who decides, who owns, and who benefits?
Good design starts with a diagnosis. If women-owned enterprises borrow less, the cause may be collateral, transaction costs, repayment schedules, weak demand, or risk. Each explanation implies a different instrument. A Theory of Change should make the mechanism explicit; a Logical Framework should distinguish outputs, such as training delivered, from economic changes, such as more secure control of assets or higher net income.
Measurement should follow the question. If the mechanism is time scarcity, collect time-use evidence. If it is control over returns, ask who decides how earnings are spent. The Women’s Empowerment in Agriculture Index offers a disciplined foundation for examining production decisions, resources, income, leadership, and time [14].
Evaluation must still separate observed change from attribution. Higher income after a project may also reflect rainfall, prices, or other interventions. As Esther Duflo cautions, neither growth nor empowerment automatically sets off a self-sustaining cycle that resolves inequality [15]. Deliberate policy choices and credible evidence remain necessary.
Where AI can help and where it cannot
Only after these foundations are clear does AI have a useful role. With appropriate validation, it can help reviewers examine a proposal against its diagnosis, Theory of Change, and Logical Framework — flagging, for instance, a promised income gain with no market pathway, or a labour-intensive activity that ignores care work. In monitoring, evaluation, and learning (MEL), it can help organise survey and interview evidence and identify conflicting accounts for human review.
Forecasting claims need particular care: validation on relevant data, uncertainty estimates, and checks across groups and settings. Prediction does not establish causation. NIST’s AI Risk Management Framework sets out relevant principles concerning validity, reliability, transparency, privacy, and the management of harmful bias [16]. Historical data can encode historical exclusion.
In our account of the reconstructed South Africa model, AI-assisted analysis supported approximately 6,000 model runs to explore how conclusions changed across policies and assumptions [9]. The economic simulations supplied the reported policy results; AI supported their exploration and analysis. Reproducing historical results and testing sensitivity are useful checks, but do not independently establish that the model predicts present-day outcomes accurately.
This is the direction informing GenTraQ™ at ScaleX Labs: connecting design, monitoring and evaluation, and the development of forecasting, while examining movement along the Gender Integration Continuum. Claims about its performance require appropriate validation. Professional judgement, contextual knowledge, and accountability remain with people. People lead. AI strengthens.
Balancing the ledger
The ledger runs through this story. Agricultural economics examined what the farm household produced; gender economics asks what that record left out — the unpaid hours, the decisions made inside the household, and the people who carried costs obscured by aggregate measures of success.
Balancing the ledger is more than an accounting exercise. A sound project or policy should name the constraints it seeks to change, explain how its resources will shift incentives and opportunities, and investigate who ultimately benefits. Responsible AI can help assemble and interrogate that evidence. The enduring task remains economic: to understand how scarce resources and decision-making power shape well-being, and to design institutions that advance gender equality.
References
[1] Royal Swedish Academy of Sciences (2023). The Prize in Economic Sciences 2023. Press release on Claudia Goldin’s research. Read source
[2] Singh, I., Squire, L., and Strauss, J. (1986). A Survey of Agricultural Household Models: Recent Findings and Policy Implications. The World Bank Economic Review, 1(1), 149–179. Read source
[3] Becker, G. S. (1965). A Theory of the Allocation of Time. The Economic Journal, 75(299), 493–517. Read source
[4] Boserup, E. (1970). Woman’s Role in Economic Development. London: George Allen and Unwin.
[5] Udry, C. (1996). Gender, Agricultural Production, and the Theory of the Household. Journal of Political Economy, 104(5), 1010–1046. Read source
[6] Latigo, A. A. R., and Ironmonger, D. S. (2004). The Missing Link in Growth and Sustainable Development: Closing the Gender Gap. Issues paper, ADB/ECA Symposium, Kampala, 24 May 2004. UN Economic Commission for Africa. Read source
[7] Latigo, A. A. R., and Mohammed, N. (2005). A New Round of Time-use Studies for Africa: Measuring Unpaid Work for Pro-poor Development Policies. UNDP and Levy Economics Institute conference paper.
[8] Latigo, A., Abdourahamani, O., Mohammed, N., Fofana, I., Decaluwé, B., and Cockburn, J. (2005). From National Satellite Accounts of Household Production to Macroeconomic Modeling in African Countries. Conference paper; reproduced on GIDE’s publications page in 2018. Read source
[9] Latigo, A., and Abdourahamani, O. (2026). AI-Powered Gender Transformative Macroeconomic Computable General Equilibrium (CGE) Model for Evaluating Impacts of Policies on Poverty Reduction in Africa: The Case for South Africa. Working paper, ScaleX Labs and GIDE, in preparation for journal submission. Public summary: Freed Time Must Meet Open Doors: Rebuilding a UNECA Gender-Aware CGE Model with Responsible AI. Read source
[10] FAO (2023). The Status of Women in Agrifood Systems. Rome. Read source
[11] FAO (2024). The Unjust Climate: Measuring the Impacts of Climate Change on Rural Poor, Women and Youth. Rome. Linked report summary. Read source
[12] Africa Climate Change Fund. Call for Proposals. Official overview of the third and fourth calls. African Development Bank. Read source
[13] Africa Climate Change Fund (2024). Fourth Call for Proposals: Gender Transformative Methane Reduction. African Development Bank. Read source
[14] IFPRI. About WEAI. Women’s Empowerment in Agriculture Index Resource Center. Read source
[15] Duflo, E. (2012). Women Empowerment and Economic Development. Journal of Economic Literature, 50(4), 1051–1079. Read source
[16] NIST (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0). NIST AI 100-1. Read source
[17] UNDP Human Development Report Office Research Team (2016). Valuing Care Work. 15 March. Read source
About ScaleX Labs and GIDE
ScaleX Labs is a Responsible AI platform company serving development professionals in the Global South, working with the Global Institute for Development Evidence (GIDE), an independent non-profit institute. To learn more or to partner with us, visit scalexlabs.io or write to info@scalexlabs.io.
