Journal Issue
Share
Chapter

4. Scope for Further Integration? An Analysis Based on a Gravity Model

Author(s):
Céline Allard, Jorge Canales Kriljenko, Jesus Gonzalez-Garcia, Emmanouil Kitsios, Juan Trevino, and Wenjie Chen
Published Date:
March 2016
Share
  • ShareShare
Show Summary Details

To assess the underlying factors that drive trade integration and to estimate the order of magnitude of a potential “trade gap” for sub-Saharan Africa, we then use a gravity model approach. In general, trade between two countries tends to be more intense the closer the two countries are both geographically and culturally—such as sharing a similar language or past colonial ties. In addition, the size and level of development of the trading economies are important parameters influencing trade flows. A common way in the literature to assess the relative size of such flows is to estimate “gravity models,” linking the magnitude of bilateral trade flows to these very characteristics of the trading countries (Head and Mayer 2014).

We estimate such a gravity model using the IMF’s Direction of Trade Statistics (DOTS) database. Our sample covers 167 countries for the 1980–2013 period. While the DOTS database lacks data on services trade, it provides the most extensive panel data set of worldwide bilateral trade flows currently available. Our empirical specifications can be summarized in the following equation:

In this equation, the exports from exporting country i to importing country j in year t, xijt, are conditioned on Mit1Ex and Mjt1Im, which denote the vectors of the attributes of exporter i and importer j in year t – 1. Factors that affect trade costs between i and j are represented by Dijt − 1 and uijt denotes the unobserved bilateral trade cost determinants. To avoid potential biases from reverse and simultaneous causation, we condition on the one-year lagged values of the regressors and we control for global common shocks by including a year fixed effect, at.

Table 1 shows the gravity equation estimates of the determinants of bilateral trade patterns. The standard errors reported in the regressions are robust and clustered at the country pair level to account for bilateral trade correlation across time and to allow for different variance across the pairs.

Table 1.Gravity Model Estimates
(1)(2)(3)(4)(5)
ln (Exports)ln (Exports)ln (Exports)ln (Exports)ln (Exports)
Exporter ln (population) (lag1)1.063***1.043***1.042***1.059***1.319***
(0.008)(0.008)(0.008)(0.008)(0.012)
Importer ln (population) (lag1)0.966***0.981***0.980***0.962***1.087***
(0.008)(0.008)(0.008)(0.008)(0.012)
Exporter ln (GDP per capita) (lag1)0.946***0.854***0.854***0.907***0.827***
(0.011)(0.013)(0.013)(0.012)(0.023)
Importer ln (GDP per capita)0.703***0.712***0.712***0.665***0.651***
(0.010)(0.011)(0.011)(0.011)(0.021)
Log of distance (lag1)−1.393***−1.374***−1.360***−1.368***−1.398***
(0.016)(0.024)(0.024)(0.017)(0.021)
Common official language (lag1)0.498***0.554***0.561***0.482***0.474***
(0.065)(0.063)(0.063)(0.064)(0.096)
Common language (lag1)0.337***0.497***0.486***0.515***0.521***
(0.066)(0.064)(0.064)(0.065)(0.099)
Common colonizer (lag1)0.579***0.690***0.676***0.632***0.674***
(0.054)(0.054)(0.054)(0.053)(0.084)
Exporter landlocked (lag1)−0.756***−0.562***−0.565***−0.651***−0.631***
(0.038)(0.037)(0.037)(0.037)(0.056)
Importer landlocked (lag1)−0.811***−0.785***−0.787***−0.735***−0.758***
(0.037)(0.035)(0.035)(0.036)(0.051)
Both Asia and Pacific (lag1)1.889***1.963***
(0.109)(0.110)
Both Europe (lag1)1.672***1.758***
(0.089)(0.092)
Both Middle East and Central Asia (lag1)0.0060.091
(0.110)(0.112)
Both North and Latin America (lag1)1.071***1.151***
(0.092)(0.094)
Both CEMAC (lag1)0.508
(0.373)
Both EAC (lag1)1.607***
(0.419)
Both SACU (lag1)−0.061
(0.536)
Both WAEMU (lag1)1.097***
(0.290)
Both sub-Saharan Africa (lag1)−0.328***
(0.072)
None sub-Saharan Africa (lag1)0.727***
(0.033)
Exporter rule of law (lag 1)0.364***
(0.037)
Importer rule of law (lag1)0.153***
(0.035)
Exporter infrastructure (lag1)0.226***
(0.021)
Importer infrastructure (lag1)0.165***
(0.021)
Exporter ln (tariff) (lag1)−0.112***
(0.010)
Importer ln (tariff) (lag1)−0.057***
(0.011)
Exporter ln (domestic credit) (lag1)0.302***
(0.033)
Importer ln (domestic credit) (lag1)0.187***
(0.029)
Observations48459548459548459548459554997
Time fixed effectsYesYesYesYesYes
Country fixed effectsNoNoNoNoNo
R-Squared0.6240.63520.63550.62440.7271
Source: IMF staff calculations.Note: Robust standard errors are shown in parentheses; * indicates significance at 10 percent, ** at 5 percent, and *** at 1 percent.

Column 1 controls for exporter and importer attributes such as size (population) and development (GDP per capita), as well as trade cost measures (bilateral distances, common language dummies, common colonizer dummies, and dummies representing landlocked countries).4 To perform intraregional bilateral trade comparisons across regions, we use the group of sub-Saharan African countries as the comparison group in column 2 and introduce regional dummies for regional trade occurring within other regions.5 Similarly, column 3 allows for intraregional comparisons between sub-Saharan African countries that have formed monetary and trading unions and those that have not. To compare trade flows emanating from sub-Saharan Africa to trade occurring elsewhere in the world, in column 4 we use as the baseline comparison group the group in which either the exporter or the importer is a sub-Saharan African country, and introduce dummies for trade flows where none of the trade partners are from sub-Saharan Africa (for completeness, we also account via a second dummy for sub-Saharan Africa’s intraregional trade). Column 5 additionally includes estimates for institutional and policy-related variables.6 The average values of these institutional and policy-related variables for sub-Saharan African countries and the rest of the countries in our sample are provided in Table 2.

Table 2.Output Drops and Decelerations (1990–2013)
Sub-Saharan AfricaRest of the World
Tariffs7.11.6
Infrastructure2.84.6
Rule of law−0.50.5
Domestic credit24.168.8
Sources: IMF, World Economic Outlook database; and World Economic Forum.

The overall analysis suggests that exports and imports from SSA are significantly lower than trade flows elsewhere in the world. Of course, this partially reflects lower levels of income in sub-Saharan Africa, as well as relatively longer distances and a higher number of landlocked countries in the region, as accounted for in the determinants of the gravity model equation. But even after accounting for these determinants, the dummy for trade occurring elsewhere in the world in column 4 of Table 1 still comes out significant. More specifically, column 4’s estimation suggests that bilateral trade flows from sub-Saharan Africa tend to be, on average, 50 percent lower than trade flows elsewhere in the world, even after accounting for economic and other determinants (Figure 11). Likewise, the dummies for trade occurring in other regions in column 2 are also significant, with the exception of the Middle East and Central Asia, suggesting that sub-Saharan-African regional trade is much smaller than regional trade in most other regions in the world—85 percent lower than in South and East Asia, 80 percent lower than in Europe, and 65 percent lower than in Northern and Latin America.7 It is noteworthy that sub-Saharan African regional trade exhibits such substantial gaps despite the existence of numerous intraregional trade agreements—possibly because their overlapping groupings greatly reduce their effectiveness.

Figure 11.Trade Flows Compared with Other Regions1

Sources: IMF, World Economic Outlook database; World Economic Forum; and IMF staff calculations.

1 Sub-Saharan Africa trade compared with trade of other regions, after controlling for size, level of development, cultural ties, and geographical conditions.

What explains these substantial gaps? To shed light on that question, the gravity model described previously is augmented in column 5 of Table 1 to include determinants such as the rule of law, tariff levels, quality of infrastructure, and level of credit to the private sector, as is frequently done in the literature (see, for example, Nordås and Piermartini 2004). These factors are found to play a significant role in further explaining the extent of bilateral trade flows at the global level. All else equal, a more supportive business environment, lower tariffs, better infrastructure, and easier access to credit all favor larger trade flows. And these factors are substantially less conducive to trade in sub-Saharan Africa, with the quality of infrastructure about 50 percent lower in the region than elsewhere in the world, credit-to-GDP ratios about 25 percent lower, and tariffs on average four times higher than elsewhere (Figure 12).

Figure 12.Potential Increase in Trade1

Sources: IMF, World Economic Outlook database; World Economic Forum; and IMF staff calculations.

1 Percent increase in sub-Saharan Africa‘s trade if the variable moves from the average for sub-Saharan Africa to the average for the rest of the world.

More specifically:

  • Infrastructure appears as the most important impediment to trade for the region. In fact, bringing infrastructure to the average level of quality at the global level would help enhance sub-Saharan African trade by as much as 42 percent, as this would substantially lower the cost of cross-border movements of goods. Indeed, efforts to fill the infrastructure gaps are currently under way in the region.

  • Further efforts to improve governance and the business climate would also have a very favorable effect: raising the index of rule of law to the average level elsewhere in the world would generate another 28 percent increase in sub-Saharan African trade flows. In particular, measures to lower nontariff impediments to trade—export taxes and duties, but also corruption, regulatory requirements, and delays in clearing customs that all add up to extra costs—would greatly improve prospects for trade, especially at the regional level.

  • Likewise, access to credit for the private sector plays a paramount role for the region’s trade. Further financial deepening to the level observed elsewhere in the world would support an expansion of trade by as much as 29 percent. Such expansion would need, however, to be accompanied with adequate macroprudential frameworks to carefully manage the corresponding risks (IMF 2012).

  • Finally, continuing to work toward lowering tariffs in the region would further support the development of both international and regional trade. On average, bringing tariffs to the average global level could yield about 14 percent additional trade. One consideration, though, is that taxes on trade still represent a substantial source of fiscal revenues for many countries in the region, and policies to lower tariffs need to be accompanied by continued efforts to increase revenue mobilization from other sources.

  • At the regional level, deepening existing customs unions with further economic integration would help, as the examples of the EAC and WAEMU illustrate in column 3 of Table 1: all else equal, cross-border exchanges within the EAC are found to be five times larger than average regional trade flows within sub-Saharan Africa; in the WAEMU, they are about three times larger. But having a single currency by itself is not enough, as evidenced in the Central African Economic and Monetary Community (CEMAC), where intracurrency union trade flows are not found to be significantly higher than regional flows outside the currency union.

    Other Resources Citing This Publication