Abstract
Short summary of the English working paper. Experimental variation in transport subsidies in Bogotá identifies a discrete choice model of travel demand. The model estimates welfare gains as consumer surplus: an average monthly increase of COP 48,085 (USD 12.84), or 3.1% of mean household income. Gains are much larger for non-work trips than for commuting, with differences linked to travel times and schedule flexibility.
Empirical strategy
The Bogotá experiment randomly offered monthly travel-card subsidies of COP 28,000 or COP 21,000, alongside a control group. Baseline and weekly mobility diaries record travel modes, costs, and times. A mixed logit model recovers individual preferences and model-implied consumer surplus. A two-stage difference-in-differences estimator addresses staggered subsidy use when estimating welfare changes. Separate fixed-effects and event-study specifications estimate travel-time responses.
These estimates combine experimental assignment with structural assumptions; results should not be read as general-equilibrium effects of a citywide policy.
Main equation
The random-utility model in Equation (2) is:
\[ U_{ijt}=\theta_i\,\mathrm{Price}_{jt} +\beta_i\,\mathrm{TravelTime}_{ijt} +\tau\,\mathrm{Treatment}_i +\Omega'X_{ijt}+\varepsilon_{ijt}. \]
Price and time coefficients vary across individuals. \(X_{ijt}\) captures observed individual and alternative attributes. The estimated utilities generate a log-sum consumer-surplus measure, expressed in monetary units.
Selected results
Monthly welfare gains
| Group | All participants | Work trips | Other trips |
|---|---|---|---|
| Pooled | 48,085 | 20,951 | 100,096 |
| A: COP 28,000 | 58,336 | 29,637 | 115,295 |
| B: COP 21,000 | 38,121 | 12,594 | 84,149 |
Travel-time reductions
| Group | Change in minutes | Standard error |
|---|---|---|
| Pooled | -10.06 | 3.686 |
| A: COP 28,000 | -18.66 | 7.425 |
| B: COP 21,000 | -1.605 | 4.283 |
- Welfare: the larger subsidy raises monthly consumer surplus by about 3.8% of mean household income; the smaller subsidy raises it by about 2.5%.
- Trip purpose: non-work gains are almost five times work-trip gains, consistent with the role of congestion and less flexible commuting schedules.
- Time: the pooled reduction is about ten minutes per trip. The larger subsidy drives the decline; the smaller subsidy’s estimate is not statistically significant.
Citation
The English draft is dated November 10, 2023. The linked Spanish CEDE/SSRN version is dated September 14, 2023; the results summarized above follow the English draft.
English working paper
@unpublished{OchoaRincon:2023Welfare,
title = {The welfare effects of subsidies: A case study of public transport},
author = {Ochoa Rincón, Jorge Luis},
year = {2023},
month = nov,
note = {English working paper, November 10, 2023},
url = {https://jorgeluis8ar.github.io/files/WP-Welfare-Effects-Subsidies-Case-Study-Public-Transport.pdf}
}Spanish CEDE version
@techreport{OchoaRincon:2023CEDE,
title = {Modelo de elección de modos de transporte: Evidencia de un experimento social controlado},
author = {Ochoa Rincón, Jorge Luis},
doi = {10.2139/ssrn.4575115},
institution = {Universidad de los Andes},
type = {Documento CEDE},
number = {30},
year = {2023},
month = sep,
url = {https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4575115}
}