When a new technology erodes an established source of revenue, firms may respond by expanding other parts of their business. I study this adjustment through the introduction of television and its effects on American radio stations. Using newly assembled station-level financial records and variation in television rollout associated with the 1948–1952 FCC licensing freeze, I find that television reduced annual major-network revenue by $8,634 per station while increasing local advertising revenue by $10,770. The estimated effect on total time-sales revenue is small but imprecise. These changes were concentrated among stations with preexisting major-network revenue. Stations with greater initial network dependence experienced larger network losses without correspondingly larger local advertising gains. Local advertising gains were also smaller in markets with competitors that received no major-network payments, even after accounting for population. The findings suggest that the ability to replace lost business varies with existing revenue sources and local competition.

Where and why labor unions first take hold may shape the later geography of organized labor. In this paper, we study the Amalgamated Society of Railway Servants, one of Britain’s earliest national railway unions. Using newly digitized union records, historical railway networks, and workplace accident reports, together with geographic variation in railway accidents that preceded the formation of early branches, we find that each additional infrastructure-failure accident is associated with an 8.9 percentage point higher probability that a district has an early union branch. Union geography was persistent: 82 percent of districts with an early union branch still had one at the end of our sample. Districts sharing railway lines with early unionized districts were also more likely to host branches later in the sample, although we cannot fully separate this from underlying railway activity. These findings suggest that where unions first formed influenced where they later existed, and that railway networks may have helped unionization spread.

Place-based industrial policies are meant to raise economic activity in targeted areas, but it is hard to tell how far their effects reach and where they are largest. In this paper, I study India’s expansion of Special Economic Zones (SEZs) following the 2005 SEZ Act, combining data on 244 geocoded zones with a panel of nearly 600,000 villages. Using a spatial difference-in-differences design, I find that villages within five kilometers of an SEZ see increases of 0.60 log points in nighttime lights, 0.33 log points in manufacturing employment, and 0.11 log points in firm counts. These effects decline with distance and largely disappear beyond 20 kilometers. A continuous measure of exposure that incorporates zone size and distance shows a similar pattern. The effects are also larger where local labor markets are tighter. These results suggest that SEZs raise economic activity mainly near the zone, and that the size of the gain varies with local labor-market conditions.