Working Paper

When electricity infrastructure adjusts slowly, expansion in one electricity-intensive industry can constrain subsequent local investment in another. We study this interaction between cannabis cultivation and data centers using close recreational cannabis ballots in a difference-in-discontinuities design and a new satellite measure of indoor cultivation. Legalization raises detected cultivation by 3.6 times the sample mean within five years. This increase implies about 22 megawatts of electricity demand from cultivation per county, 18 percent of the average county’s consumption. The active data center stock falls by 0.70 facilities per county and operating capacity by 16.2 megawatts. Retail electricity prices rise where cultivation expands, while total consumption does not move measurably. The decline in data centers begins in the ballot year and also appears in other-state counties on the same grid. We find little evidence of competition through shared labor markets. A discrete-choice model of data center siting implies that absent legalization, more capacity would be built in the narrowly passing states and annual electricity bills there would be 15.6 billion dollars lower.

How does an independent judiciary shape structural transformation? I study a Chinese reform that moved all lawsuits against local governments from the government’s own courts to external ones. Using difference-in-differences with household panel data, I find the reform raises rural off-farm work participation by 8.9 percent and household income by 7.3 percent. It strengthens rural land rights, raising compensation per unit of land lost, so households can take offfarm work without forfeiting their land. It also raises the demand from non-farm activities, where active businesses increase by 6.0 percent and incumbent private firms raise assets per employee and hiring.

Does a historically contingent change in who holds office affect long-run economic growth? I exploit China’s 1982–1984 top-down administrative reorganization, which replaced much of the Mao-era provincial leadership with Deng Xiaoping’s younger technocrats. Using a continuous-treatment difference-in-differences design, I find that a one-percentage-point increase in replacement intensity raises provincial GDP by 1.8%. The effect appears within a decade and is still as large four decades later. The evidence suggests that the reorganization changed local economic institutions in favor of the market economy, and that this shift was sustained by private-sector expansion and by later appointees with similar characteristics.

Do policy bundles built around government equity promote local high-tech industries? We study such bundles when Chinese prefecture governments invest their own capital in Government Industrial Guidance Funds. Using a staggered difference-in-differences design, we find that fund establishment increases firm numbers by 8.7 percent in targeted industry–prefecture cells within five years. These bundles add supportive regulations, subsidies and tax refunds that reach firms beyond fund portfolios. They mark costly-to-reverse government commitments that coordinate private investors and draw in their investments. The effects compound into local ecosystems, with knowledge, upstream–downstream and labor-market links, that persist after the funds stop investing. 

We study whether strengthening legal institutions can reduce civil conflict. We develop a simple conceptual framework linking legal capacity to conflict and test its predictions using the staggered roll-out of donor-funded judicial reforms across African countries in a difference-in-differences design. We find that judicial reforms reduced conflict by about 20%, with effects that grow over time and are concentrated in high-risk subnational regions. These impacts are driven by reforms that expand access to courts, while reforms that primarily improve judicial speed or quality have limited effect. Consistent with the model, reforms reduce civil–civil conflict most strongly where state enforcement is high and and reduce civil–state conflict only where judicial independence is strong. Using GDELT and survey data, we show that reforms increase trust in courts but not in other institutions. Economic growth effects cannot account for the decline in conflict, indicating that improved dispute resolution is the primary mechanism.

In economic history of China, there is a stigma attached to the traditional farming sector which has been commonly seen as the root course of technological backwardness, social inequality, mass poverty, class polarisation, social revolutions, and so forth. This study puts this stereotyped narrative to an acid test with reliable data from China’s first modern survey of the farming sector compiled by John Lossing Buck and his team. It re-examines the bona fide performance of China’s traditional agriculture in its ‘twilight stage’ before 1950. Thereafter, the Chinese Communists started radical programmes of collectivisation and state control over farming. Our results show that China’s traditional farming remained reasonable efficient in terms of agricultural productivity, farm wages and nutrition intake and certain growth potential remained for the sector.