Making policies work

GREAT insights Magazine

Market Integration and Border Effects in Eastern Africa

August 2012

Versailles, B. 2012. Market integration and border effects in Eastern Africa. GREAT Insights, Volume 1, Issue 6. August 2012. Maastricht: ECDPM.

Trading in East Africa can be difficult, both within and between countries. Between countries, the establishment of a customs union between the members of the East African Community (EAC) (1) in 2007, implies that, in principle, there should not be any impediments to trade goods across their borders. What do the numbers says on the cost of intra-regional trade in the EAC?

A customs union implies that participating countries cannot levy any tariffs on goods entering from another participating country. However, such tariffs are subject to exemptions and are not always uniformly applied (2). What is more, EAC member countries have in the past erected so-called non-tariff barriers (NTBs) to discourage imports. Such NTBs can take many forms, including outright import bans, imposing quality conditions on imports, packaging requirements, etc. 

Within countries, bad or non-existing roads, governance issues (e.g. security or corruption) or lack of market information can mean that it is prohibitively expensive for farmers and traders to bring produce to certain markets. Poor infrastructure plays a double-negative role for landlocked countries, as imports are dependent on the quality of their neighbours’ transport network. This motivates a search for finding good ways to measure the impact of international borders on market efficiency.

Quantifying price differences within the EAC member countries

My research tries to quantify the relative importance of these effects, by looking at price differences within and between all EAC countries, except Tanzania. The starting point is the Law of One Price (LOP) which says that identical goods should, absent any distortions, sell for the same price in different locations. The standard example is that if shop A on one side of town is selling bread for $0.80 a loaf and shop B on the other side is selling it for $1.00 a loaf, there is an arbitrage opportunity, with gains to be made from buying bread at $0.80 and selling it outside of shop B for $0.99. In an ideal setting this incentive to exploit differences in prices eventually drives those prices to equalize through competition.

Yet there are a lot of reasons why the LOP might not hold – one common one is the presence of transportation costs (in my example above the arbitrage opportunity vanishes if it costs me $0.25 to move a loaf of a bread between the shops). While the LOP is more of a theoretical ideal than an empirical reality, it is a reasonable benchmark for testing whether or not markets are working efficiently: if two identical goods are being sold for separate prices in otherwise identical settings, we know something is amiss. Looking at the prices of identical goods on either side of a border might tell us the extent to which borders are preventing prices from equalizing.

I use monthly price data between 2004 and 2008 from cities in Kenya, Uganda, Rwanda and Burundi to try and explain deviations to the LOP. Note that, except for Kenya, these are all landlocked countries. To disentangle the effect of a border from that of distance, I compare prices between cities both within and across these countries, while also trying to account for other factors that might drive a wedge between prices, such as changes in the exchange rate or non-tariff barriers to trade. I identify 24 goods that are deemed comparable across the 4 countries and 39 cities for which I have data. These goods are mainly food items that are widely consumed in the region.

Large price differences between EAC member countries

Even after allowing for these potentially-confounding factors, price differences are significantly larger (between 13 and 20%) between cities on opposite sides of a border. i.e. borders still appear to impede market integration. However, other variables also have an impact on price differentials between cities. A distance of 100 km between two cities allows for departures of around 13 percent from the LOP benchmark. From this estimate, we can calculate that the border effect is equivalent to differences in prices between two cities that are between 300 and 6,000 kilometers apart, depending on the specification. Compared to similar work done on the US-Canada border, distance plays a much bigger role in East Africa. What are the effects for cities that are more than one border apart? For example to get from Burundi to Uganda, one has to go through Rwanda. Our estimation shows that the border effect is additive, i.e. cities which are divided by more than one border see a larger divergence in prices than those with only one border in between them. 

The two other potential explanations for differences in prices that I look into, non-tariff barriers and the nominal exchange rate seem to be less important. The variable for NTBs is effectively an index where, using information from the World Bank’s Doing Business surveys, I combine information on (i) number of documents entrepreneurs/traders have to fill out for clearance before they can import or export a good, (ii) number of days it takes to import or export goods, and (iii) cost of fees for importing or exporting a 20 foot container. When explaining deviations from the LOP, the coefficient on NTBs is significant but very low. The way the variable is constructed however, makes it hard to interpret the size of the coefficient. The coefficient on the nominal exchange rate is also close to zero, implying a high pass-through from foreign to domestic prices, corroborating the conventional wisdom that pass-through is relatively rapid and complete in developing countries as opposed to 
more advanced economies.

The effect of landlocked countries shows up when we investigate the impact of the Kenyan political crisis at the beginning of 2008. The price data shows an increased divergence in prices for the first half of 2008, as Kenya is a transit country for many goods entering three aforementioned landlocked EAC countries.

Key lessons

The policy implications of this line of research are clear. First of all, there is a lot of work still to be done in low-income countries like the EAC member countries to improve the transport infrastructure, governance and market information within countries. Indeed, it is no surprise that the largest distance effects are found in Burundi, which was for a large part of the estimation period embroiled in a civil war. Even if EAC countries get their act together and further reduce border costs, there still needs to be more attention given to reducing transportation costs within countries, where poor roads and corruption remain a problem. A study by Rwanda’s Private Sector Federation found that the trek from Mombasa to Kigali required a whopping 36 stops and $864 in bribes, mostly at police checkpoints and weigh stations within countries, rather than at the border.

Second, the importance of the border effect brings home once again the message that reducing tariffs is only one in an array of complementary initiatives needed for reaping the benefits of economic integration. The EAC introduced a customs union in 2005 aimed at zeroing all tariffs between member countries, establishing a common external tariff and attempting to reduce non-tariff barriers as much as possible. While I do find that price differences between Kenyan and Ugandan cities (Rwanda and Burundi had yet to join) did fall ever so slightly after the customs union was adopted, the border effect persists, suggesting that integration still has a long way to go.

Third, more information is needed on exactly what impedes cross-border trade and more data is needed on non-tariff barriers. The construction of a non-tariff barrier index as an explanatory variable is a starting point, but would probably need to be disaggregated to get a better understanding of the underlying issues. 

Fourth, the research shows the importance of whether a country is landlocked in the transmission of economic shocks. The Kenya crisis at the end of 2008 severely disrupted markets in Burundi, Rwanda and Uganda and caused acute shortages in e.g. petrol products. As such governance issues in transit countries become a key issue for these landlocked countries, which is another reason for deeper regional integration.

This article summarises the findings of CSAE Working Paper WPS/2012-01 “Market Integration and Border Effects in Eastern Africa”, available at

Bruno Versailles is Economist at the IMF’s Fiscal Affairs department


  1. The EAC is made up of Kenya, Tanzania, Uganda, Burundi and Rwanda.
  2. For example, a recent publication found that on average tariffs within the EAC have gone down from  26.1 percent in 1994 to 9.2 percent in 2011.

This article was published in GREAT Insights Volume 1, Issue 6 (August 2012).

Share Button
Economic Transformation and TradeEast African Community (EAC)TradeBurundiEast AfricaKenyaRwandaUganda

External authors

Bruno Versailles