Title:	Tanzania Mercantile Exchange
DNr:	NAISS 2026/4-216
Project Type:	NAISS Small
Principal Investigator:	Mehdi Amani <mehdi.amani@su.se>
Affiliation:	Stockholms universitet
Duration:	2026-02-05 – 2027-03-01
Classification:	50201
Keywords:

Abstract

This research studies whether integrating agricultural export markets through an online auction platform improves price outcomes and competition for farmers. The empirical setting is the introduction and expansion of the Tanzania Mercantile Exchange (TMX), an online English auction system that began piloting in 2019 and has expanded across crops and districts over time. Traditionally, agricultural export commodities in Tanzania move through fragmented supply chains. Small farmers sell to local traders or cooperatives, often via sealed-bid auctions or bilateral negotiations, with limited transparency and geographically segmented buyer participation. These frictions restrict price discovery, reduce competition among buyers, and weaken farmers’ bargaining power. TMX alters this structure by allowing registered warehouses to sell standardized lots through centralized, online English auctions, where buyers from different regions can simultaneously participate. The first research question asks whether selling through TMX increases auction prices—prices that directly translate into farm-gate revenues. The hypothesis is that online transparency improves price discovery, lowers search and matching costs for buyers, and integrates geographically dispersed demand, thereby increasing competition. TMX also reduces non-price frictions such as travel costs and informal payments, potentially raising both efficiency and prices. To evaluate price effects, the study exploits the staggered roll-out of TMX across districts and crops using difference-in-differences designs. One specification compares treated and untreated district-crop pairs over time, while another uses treatment intensity, measured by the number of registered warehouses in a district. These approaches allow the analysis to separate changes driven by market integration from fixed district or crop-specific factors. A key challenge is that warehouse adoption is not random, which raises concerns about pre-trends and spillovers to nearby control districts. Beyond average prices, the research examines whether TMX reduces spatial price dispersion. If the platform integrates markets effectively, prices across districts for the same crop should converge. This is tested using district-pair regressions that relate absolute price differences to treatment exposure, controlling for pair fixed effects and crop-year shocks. A central mechanism explored is buyer competition. TMX data show a dramatic increase in the number of buyers over time, especially for commodities such as sesame, cashew, and chickpeas. To identify the causal effect of competition on prices, the study uses auction timing within a day as an instrument for the number of bidders in a lot, implementing a two-stage least squares framework. Earlier or later auction slots affect bidder participation but are plausibly unrelated to unobserved quality. Finally, the paper investigates whether farmers learn from past auctions—that is, whether price discovery improves outcomes within a season. Panel regressions at the warehouse level show that prices tend to decline over time for warehouses that sell repeatedly, suggesting limited scope for farmers to leverage past auction information to negotiate higher prices. Supervisor: Konrad Burchardi, Institute for International Economic Studies