Unpacking interstate tomato trade in Nigeria: A Double-Hurdle PPML Approach

This paper contributes to the literature on agricultural market integration and domestic trade by moving beyond prices and trader-level behavior to analyze the actual movement of tomatoes through wholesale market networks in Nigeria. We treat wholesale markets as key intermediary nodes and develop a gravity-inspired framework in which interstate tomato flows reflect bilateral trade frictions, state-level demand and supply conditions, and the organizational capacity of markets. Using a field census of wholesale tomato markets in six major producing states, we construct a market-level dataset of import and export quantities between surveyed markets and all thirty-six states plus the Federal Capital Territory. Empirically, we estimate a two-stage model that jointly combines a probit selection equation with a Pseudo-Poisson Maximum Likelihood outcome equation on the selected subsample, allowing us to distinguish the extensive margin of trade link formation from the intensive margin of shipment size conditional on trade. A Monte Carlo simulation shows that this estimator outperforms both the log-transformed Heckman approach and naïve PPML when the data generating process features hurdle behavior and selection. Our central empirical finding is that interstate tomato trade is governed primarily by selection into trade relationships: distance robustly depresses the probability of forming import and export links, while its effect on shipment size conditional on trade is weaker and less stable across specifications. By contrast, market capacity, measured by the number of wholesalers, is the most robust predictor of conditional trade volume on both the import and export sides. Statelevel covariates such as GDP, tomato production, and nightlights also exhibit distinct effects across the extensive and intensive margins, showing that a single-equation approach masks important nonlinearities. These results contribute to the literature by showing that domestic movement of a perishable food commodity is best understood as a meso-level allocation process structured by trade frictions, partner selection, and intermediary capacity. The findings imply that policies to improve food-system performance should focus not only on lowering transport costs, but also on reducing the fixed and coordination costs of exchange and strengthening wholesale market capacity along the north-south tomato corridor.