Market Intergration and Border Effects in Eastern Africa

Abstract

This paper studies border efects in Eastern Africa by exploiting a consumer price data set covering 4 out of 5 EAC member states, 39 cities and 24 goods over the period 2004-2008. The Law-of-One-Price (LOP) is tested by running level regressions on relative prices with city-pairs the unit of observation. Unsurprisingly, distance plays an important role in explaining relative price movements, both within and between countries. The border efect as measured by the coefficient on a border dummy is significant but smaller compared to what is usually found in the EngelRogers literature. Consequently, distance equivalents, at 300 to 6, 000 km, are much lower (and more reasonable) than what most papers working in the Engel-Rogers tradition find. Neither the significance nor the size of the border efect is reduced by allowing for nominal exchange rate variation or non-tarifbarriers. The nominal exchange rate pass through is very high across the region, but lower for country-pairs that include Kenya. The advent of the Customs Union in 2005 improved market integration, as measured by a reduced border efect between 2004 and 2008, but only between Kenya and Uganda. This makes sense as Rwanda and Burundi only joined the EAC in 2007. Further, larger departures from the LOP are found during the Kenyan political crisis as regional markets were disrupted. When splicing the data across goods, staple markets are by far the most integrated with the lowest border efects and the lowest distance coefficients. Non-Trade barriers are most important for fruits and vegetables, which could be linked to these goods being perishable. Non-food items show the largest departures from the LOP, which can be attributed to these goods (i) being less comparable, and (ii) being less present in the consumption baskets of the average Eastern African households (i.e. less deep markets).

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 13 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.