RESEARCH
RESEARCH
WORKING PAPERS:
with Omer Bilgin (Oxford), Iason Gabriel (DeepMind) and Lewis Hammond (Cooperative AI Foundation and Oxford)
Autonomous AI agents, capable of complex planning and action, represent a significant technological evolution beyond current generative tools. As these systems become integrated into political and economic life, their distribution and capabilities will be highly consequential. This paper introduces and explores "agentic inequality" – the potential disparities in power, opportunity, and outcomes stemming from differential access to, and capabilities of, AI agents. We analyse the dual potential of this technology, exploring how agents could both exacerbate existing divides and, under the right conditions, serve as a powerful equalising force. To this end, the paper makes three primary contributions. First, it establishes an analytical framework by delineating the three core dimensions through which this inequality can manifest: disparities in the availability, quality, and quantity of agents. Second, it argues that agentic inequality is distinct from prior technological divides. Unlike tools that primarily augment human abilities, agents act as autonomous delegates, creating novel power asymmetries through scalable goal delegation and direct agent-to-agent competition that are poised to reshape outcomes across economic and socio-political spheres. Finally, it provides a systematic analysis of the technical and socioeconomic drivers – from model release strategies to market incentives – that will shape the distribution of agentic power, concluding with a research agenda for navigating the complex governance challenges ahead.
with Ole Teutloff (Oxford)
Advanced artificial intelligence has the potential to significantly reshape the labour market. Yet policymakers must navigate this uncertainty with incomplete, outdated evidence – an evidence gap that risks ill-informed interventions. This paper argues that frontier AI developers, with their privileged access to data, models, and technical insight, are uniquely positioned to help address this challenge. Building on nascent initiatives, we propose a framework forsystematic developer contributions. We first identify key knowledge gaps regarding AI’s labour market impacts. Second, we examine research methods needed to close these gaps focusing on structural barriers – related to data access, model availability, and resource constraints – that currently limit research by academic economists, government agencies, and other independent researchers. We conclude by presenting a strategic framework built on the principle of comparative advantage, structured around three pillars: (1) developing foundational data and privacy infrastructure; (2) systematically supporting external research by these independent actors; and (3) fostering multi-stakeholder dialogue and data harmonisation. By focusing on this enabling role, developers can help expand the evidence base needed to inform more effective policy responses to AI’s economic transformation.
Decoding Cross-Border Payment Flows in Sub-Saharan Africa: A High-Frequency Analysis of Granular Platform Data (2025) G20 Conference Paper
with Pedro Arnt (dLocal), Pardon Mujakachi (dLocal) and Samantha Torrance (Access Partnership)
This paper analyses cross-border payment flows in Sub-Saharan Africa (SSA) by leveraging high-frequency, granular transaction data from dLocal, a rapidly growing payment platform that connects global merchants with local payment methods across emerging markets. Focusing on pay-in transactions in Kenya, Nigeria, and South Africa from November 2023 to March 2025, we conduct a detailed economic analysis of both payment behaviours and the structure of digital consumption. The findings reveal deeply heterogeneous digital economies. In terms of payment methods, the data confirms well-established patterns: Kenya is defined by high-volume, low-value mobile money transactions, South Africa by a mature card-based system, and Nigeria by a dynamic mix of instant bank transfers and surging digital wallet adoption. This payment infrastructure facilitates distinct consumer spending patterns on the platform: South Africa's activity is dominated by e-commerce, Nigeria's by entertainment and streaming services, and Kenya's by foundational spending on telecoms and communication services. We further demonstrate how this data can capture nuanced behavioural responses to policy shocks and connectivity disruptions, illustrated by a detailed case study of Kenya, where proposed tax changes and internet disruptions in mid-2024 led to immediate, observable shifts in mobile money usage and a surge in VPN-related spending. Methodologically, this paper demonstrates the value of proprietary payments data in complementing official statistics, offering timely, granular insights that align in direction with broader system-level benchmarks while highlighting platform-specific nuances such as average transaction values and merchant category mixes. Substantively, the results provide actionable intelligence for policymakers and businesses seeking to foster a more resilient and inclusive digital economy in SSA.
"Minimum wages and economic shocks: Evidence from South Africa" (2023) IZA Working Paper
with Joshua Merfeld (KDI School & IZA) (submitted)
This paper studies whether a minimum wage changes how labour markets respond to economic shocks. Using data from South Africa, we show that an agricultural minimum wage leads to higher mean wages with no significant impacts on mean employment. However, these positive aggregate outcomes hide important heterogeneity: the imposition of the minimum wage leads to substantial declines in employment – especially overall hours – in the sector in the wake of negative weather-related economic shocks, which typically exert downward pressure on wages. The increased variance of employment across years in the post-law period suggests caution in interpreting the overall welfare impacts of minimum wage laws.
"Agglomeration economies in a developing country: Evidence from geo-coded micro-panel data in South Africa" (2020) Draft
There is a dearth of rigorous evidence on the productivity benefits of cities in developing countries. The few studies on developing countries to date have estimated much higher agglomeration elasticities than those found in developed countries, but these studies have generally been unable to control for sorting on unobservables or to work with the ideal geographic units. This paper estimates the effect of city population size on workers’ monthly wages in South Africa using a unique geo-coded panel micro-dataset where workers are tracked as they move across the country. Using individual fixed effects and an IV constructed from a novel dataset on historical population settlements, my preferred estimate for regional wage elasticity is approximately 0.03 (in line with estimates for developed countries). This estimate is robust to a large number of tests and considerably lower than existing estimates for other developing countries. I also find evidence that agglomeration externalities matter more than human capital externalities.
WORK-IN-PROGRESS (NEAR COMPLETION):
"Financing the AI triad: A framework for building capacity globally"
with Sumaya Nur Adan (Oxford), Luise Eder (Oxford), Daniela Muhaj (Georgetown), Robert Trager (Oxford) & Lucia Velasco (UN)
PUBLICATIONS:
"Revisiting the measurement of digital inclusion" (2023) World Bank Research Observer
As it becomes increasingly clear how central digital transformation is to development, the need for clarifying concepts and for coming up with standardized and accurate measures for digital inclusion becomes more urgent. Focusing on the internet as a foundational technology, this paper sets out a framework of core components of digital inclusion—including access/use, quality of access/use, affordability, and digital skills. The paper then surveys the ways these components are currently measured in household and firm surveys and by international organisations. Building on simple descriptive analysis of data from a wide range of sources, the paper highlights some of the often-overlooked weaknesses of current measures, and suggests possible improvements. The paper argues that (a) metrics for certain core components of digital inclusion—including quality of access/use and digital skills—are relatively underdeveloped, (b) some questions on technology use and skills may need to be adapted to developing country settings, (c) more attention should be paid to within-country inequalities in statistics reported by international organizations, (d) currently available digital inclusion indices are not very useful, and (e) there is much potential in using big data methods to measure digital inclusion.
"The labour market impacts of female internal migration: Evidence from the end of Apartheid" (2021) Regional Science and Urban Economics
Accepted version (ungated); Oxford CSAE Working Paper 2021:01
Women often migrate within developing countries for different reasons than men and female migrants tend to be very differently distributed across economic sectors as compared to male migrants. This paper provides some of the first evidence on the labour market impacts of female internal migration, examining effects in both the productive and household sectors. I merge large sample migration data from South African censuses with detailed labour force survey data, and exploit substantial time-variation in female migrant inflows into over 200 districts. To identify the causal effects of migration on labour market outcomes, I make use of the unique history of South Africa to construct a plausibly exogenous shift-share instrument for female migrant concentration based on earlier male migration flows from reserves during the Apartheid period. I firstly find that this migration increases the employment and hours worked of high-skilled women (but not men). I demonstrate that this effect is driven by substitution in household work as many female migrants find work as domestic helpers. I also find that female migration leads to a (short-term) reduction in the employment of low-skilled female non-migrants suggesting an increase in competition at the bottom of the economic ladder.
PRE-PHD RESEARCH:
"Analysing African LIC Labour Markets with a New Segmentation Model" (2017) DFID/IZA GLM/LIC Working Paper
with Haroon Bhorat (UCT), Morne Oosthuizen (UCT) and Kezia Lilenstein (UCT)
"Day labour and Xenophobia in South Africa: The need for mixed methods approaches in policy-oriented research" (2013) Urban Forum