This is the 8th in our series of job market posts.
Governments across the world are investing trillions in public transit infrastructure to help workers reach jobs in city centers. But what if transit infrastructure reshaped economic opportunity itself and brought the job to you?
In my job market paper, I find that new transit stations in Delhi triggered a wave of firm entry near the stations themselves, creating new economic hubs in peripheral neighborhoods. The types of firms that moved in—larger retail and service stores—happened to be exactly the kinds of establishments that employ more women.
This matters because transit operates through two channels, not just one. Yes, it reduces commuting costs. But it also changes where firms locate and which types enter. When some Indian states tried giving women free bus travel, female employment barely budged, suggesting that mobility alone isn't enough if the composition of local job opportunities doesn't change.
Our Setting: Delhi, India—a megacity on the move
Delhi provides an ideal setting to study these patterns. By the early 2000s, the city had grown from 3 million to 15 million residents. Large firms were rare; fewer than 2% of establishments had 10+ workers. Most people commuted less than a mile to work. Female labor force participation sat at just 10%.
Into this context came the Delhi Metro. Planning began in the 1980s, the first line opened in 2002, and construction continued in phases through the present. Critically for identification, the final locations and timing of stations were determined by engineering constraints, land acquisition challenges, and funding availability—factors plausibly unrelated to future firm location decisions. The staggered rollout across phases allows me to compare areas receiving stations to not-yet-treated areas, while deviations from original plans provide planned but never-built stations as comparison groups.
The data challenge and breakthrough
Here's the problem: to study how transit reshapes local economies, you need neighborhood-level data. But firm data for Indian cities has traditionally only been available at the district level—far too coarse to detect localized effects around individual stations.
I solve this in two ways. First, I disaggregate the Economic Census using spatial harmonization and matching algorithms to create a 50-fold increase in granularity from 9 districts to 450 urban wards. This gives me snapshots of all formal and informal establishments in 1990, 2005, and 2013. Second, I construct an entirely new dataset by scraping and geocoding administrative records of establishment registrations. This yields the exact address and date of registration for close to a million firms since 2011, which I geocode up to the level of 1800 neighborhoods of roughly 500-meter radius. Additionally, I also use population census data to get employment and (aggregated) mobility of residents.
With these data, I document two patterns. First, female employment of residents rises 7.5 percentage points in neighborhoods within 1 kilometer of a station, echoing findings from Seki and Yamada (2025), as well as from transit expansions in Lima and South Korea. But second, firm entry spikes near transit station and sharply decays with distance from the station: 37% of new firms locate within 1km of a station, 25% within 1-2km, falling to just 13% beyond 5km. In a city where most workers commute less than a mile, local firm activity matters for employment.
The firm entry effect: More and bigger
To establish causality, I exploit the staggered timing of station openings, comparing neighborhoods within 1 kilometer of new stations (148 such neighborhoods by the end of the roll-out of period) to similar areas that could have received stations according to plans but did not (485 neighborhoods). 5km Conley standard errors are used to account for spatial correlation.
The effect is immediate and large: 9 additional firms enter per neighborhood in each 6-month period—a 150% increase over the baseline rate of 6 firms (Figure 1). And it's not just more firms; they're different. New entrants near stations are 19% larger on average and 94% more likely to be specialized operations where the manager and owner are different people. These findings are robust to alternate specifications of the comparison group and decay with distance from the stations; positive spillovers extend 1-2 kilometers from stations but approach zero beyond that.
Figure 1: Staggered differences-in-differences show significant and persistent increase in firm entry within months of transit station opening
What draws these firms? The answer reveals something fundamental about how transit reshapes opportunity. Of the 9 new firms entering near each station, 7 are in business-to-consumer industries: retail stores, diagnostic centers, personal services. Firms that serve other businesses or require high-skill labor show no response to transit access.
This points to consumer access as the key driver. Mobile phone location data confirm that foot traffic increases sharply around new stations. Larger, established brands tend to enter first, followed by smaller firms that benefit from customer spillovers, creating local commercial clusters—consistent with evidence from Kampala and Seoul, showing firms place high value on locations with dense consumer traffic. These 9 new firms translate into 56% more workers employed by new entrants near stations.
Employment effects: More jobs, especially for women
Do these new firms actually translate into more jobs? The Economic Census data reveal sustained growth: employment increases by 177 workers per 100 residents in neighborhoods within 1 kilometer of stations. Average firm size jumps 47%, and the share of firms with 10+ workers doubles.
Using market access based measure to account for general equilibrium effects, following Donaldson and Hornbeck (2016) and Tsivanidis (2023), I find that peripheral areas benefit most from connectivity gains, seeing the largest increases in employment and firm size, driven by larger retail and service establishments.
Here's where it gets interesting: of those 177 new workers per 100 residents, 28 are women. That might sound small—until you realize the baseline was just 1 female worker per 100 residents.
The effect shows up immediately in new firms. Female employment in newly registered establishments increases 98% versus 45% for men. More female-managed firms enter—a 180% increase. But women aren't siloed into "female" sectors. They're also gaining employment in male-managed firms (93% increase), large establishments with 10+ workers (115% increase), and specialized firms (82% increase).
What drives this? Firm size and industry composition. One might argue firms locate near metro stations expecting better access to female labor. But female-intensive manufacturing (textiles) and high-skill B2B services (call centers, BPOs) show no response to transit access—suggesting labor access isn't the key driver. Instead, Figure 2 shows industries locating closer to metro stations—jewelry stores, medical services, retail—already employed more women before expansion. These are consumer-facing businesses drawn by foot traffic, and they happen to employ more women. Size matters too, even within industries: larger firms consistently employ higher shares of women, whether due to reduced discrimination in competitive markets or women's preferences for formal establishments.
Figure 2: The types of firms that locate near transit stations also have a more female workforce ex-ante
Why this matters for development policy
The key insight: firm composition matters for who benefits from infrastructure. Transit doesn't just move workers to jobs, it also moves jobs to workers, and the types of firms that locate determine employment effects.
Where do these new workers come from? Together with co-author Karmini Sharma, I am conducting surveys at firms, households, and with commuters to trace this precisely. Pilot data suggest many employees at new establishments live within 1-2 kilometers—transit creates hyper-local employment.
One concern is firm relocation driving these localized effects. While most increases concentrate within 2km of stations, I find no evidence of local displacement at least—employment effects in areas 2-5km away are not significantly negative. However, we need to understand city-wide effects as well. I am developing a quantitative spatial model to capture broader distributional impacts. But here's the critical addition: standard spatial models assume firm boundaries don't matter—100 firms with 1 employee each should generate the same effects as 1 firm with 100 employees. This paper provides the first empirical evidence that this assumption might not apply when firm size matters for the distributional effects.
The policy implications are stark. Commuting interventions—fare subsidies, women-only transport—may fail to increase employment if they don't also create desirable jobs for women. On the infrastructure side, this suggests that transit-oriented development and mixed-use zoning around stations, could help move opportunities closer to those who are underrepresented in the workforce.
Akhila Kovvuri is a PhD student at Stanford University
Join the Conversation