Figure 1. Conceptual framework linking production, market access and household welfare.

Figure 1 summarises the study’s conceptual transmission mechanism. Farm production and resource exposure influence market access and travel costs, which subsequently affect prices, wages, labour allocation and household consumption welfare. The integration channel operates through stronger competition, lower transaction costs and improved conversion of production into consumption. The pressure channel captures the possibility that local price increases, labour reallocation, congestion and unequal bargaining power may weaken or redistribute the welfare gains from improved connectivity.

Figure 2. Marginal association of log production with log consumption across market access and resource dependence.

Figure 2 plots the marginal association between agricultural production and household consumption over the observed market-access range at low, median and high levels of resource dependence. The three lines slope upward, indicating that better market access strengthens the conversion of production into consumption welfare. The slope is steepest under low resource dependence and weakest under high resource dependence, showing that resource dependence attenuates, but does not reverse, the production-access welfare premium. The shaded bands represent 95% confidence intervals.

Figure 3. Country-specific production-by-market-access estimates with 95% confidence intervals.

Figure 3 compares the production-by-market-access coefficient across Ghana, Mozambique, Nigeria, Tanzania, Uganda and Zambia. Every point estimate is positive, and every 95% confidence interval lies above zero, indicating a consistently positive production-access complementarity across the six countries. The estimated association is strongest in Mozambique and Tanzania and smallest in Nigeria and Ghana. The wider interval for Mozambique reflects lower statistical precision associated with its smaller national sample.

Sources of data

The empirical dataset combines harmonised household-survey observations with spatial-accessibility, geological, climatic, nighttime-lights, price and labour-market information. The complete analytical file contains 68,000 household observations from Ghana, Mozambique, Nigeria, Tanzania, Uganda and Zambia. Country survey names, waves, panel structures, observation counts and regional coverage are documented in the Survey_Metadata worksheet of the replication workbook.