Tunisia: Selected Issues

This Selected Issues paper attempts to uncover the long-term determinants of the demand for foreign exchange reserves in Tunisia. It assesses the adequacy of current and projected reserves holdings in light of the country’s policy choices. The paper describes recent trends in foreign exchange reserves in Tunisia. Econometric evidence on the determinants of the demand for foreign reserves in Tunisia is presented. The results are used to forecast the desired level of reserves given Tunisia’s medium-term macroeconomic framework and to draw policy implications.

Abstract

This Selected Issues paper attempts to uncover the long-term determinants of the demand for foreign exchange reserves in Tunisia. It assesses the adequacy of current and projected reserves holdings in light of the country’s policy choices. The paper describes recent trends in foreign exchange reserves in Tunisia. Econometric evidence on the determinants of the demand for foreign reserves in Tunisia is presented. The results are used to forecast the desired level of reserves given Tunisia’s medium-term macroeconomic framework and to draw policy implications.

II. Is Tunisia Trading to its Potential8

A. Introduction

17. Tunisia’s trade performance is crucial to the country’s growth prospects. As one of the pioneers in the Barcelona process—the major integration initiative between the European Union (EU) and Southern and Eastern Mediterranean countries—Tunisia signed an Association Agreement with the EU (AAEU) in 1995. In 2004, the Agadir treaty was signed, which foresees a free trade zone between Egypt, Jordan, Morocco, and Tunisia. The Agadir treaty also has important provisions for rules of origin, allowing for accumulation of origin among all parties and the EU. Tunisia also has bilateral trade agreements with Kuwait, Libya and Syria and is negotiating another with Algeria.

18. However, while strong export growth has contributed to place Tunisia in the lead in the region in terms of economic performance, it has not been sufficient to fulfill Tunisia’s objectives of approaching lower-tier OECD income levels and significantly lower unemployment. To achieve this goal, further trade liberalization and trade facilitation will be important for at least two reasons:

  • To further improve its economic performance in a durable manner, Tunisia needs to accelerate productivity growth. Increased trade openness can contribute to productivity growth through a more efficient allocation of resources, technology transfers, access to a wider range of inputs, competitive pressure, and scale effects.

  • To ensure external sustainability and sufficient demand, Tunisia’s growth needs to be largely export-driven. Tunisia’s relatively high external debt (around 60 percent of GDP) constitutes a source of external vulnerability that would be substantially mitigated by continued strong export growth. In addition, given Tunisia’s small size, exports is likely to remain the most viable engine of growth.

19. This study attempts to quantify the scope for increasing Tunisia’s trade. The analysis uses a gravity model of bilateral trade applied to a database covering some 90 countries accounting for over 85 percent of total world trade. The gravity model predicts the “normal” levels of bilateral trade based on a number of characteristics of the countries involved. It can be considered as a global benchmark for trade patterns that would arise if all countries faced the same obstacles to trade (broadly defined) as the current “average” country. Tunisia’s actual level of trade can thus be compared to the benchmark to assess the potential to increase trade with individual countries. While the exact magnitude of such estimates should be treated with a degree of caution, they can provide guidance to (a) what are Tunisia’s prospects of significantly increasing trade over the medium to long term, given enabling structural reforms; and (b) which countries present the greatest untapped potential for increasing trade.

B. Estimating Tunisia’s Trade Potential

Analytical framework: the gravity model of bilateral trade

20. The gravity model is a tool often used to analyze bilateral trade patterns. Its simplicity and high level of statistical explanatory power have contributed to promoting its wide use. The basic gravity model relates some measure of bilateral trade (imports, exports, or both) to the economic size of two countries, and the geographical distance between them. Population (or GDP per capita), is often also included, along with other variables that could influence bilateral trade. The specification used here is (in logs, suppressing time subscripts for notational convenience):

Mij = αYi + βYj + λPi + δPj + ϕDij + ϕXij1 + uij

where Mij is the nominal value of imports to country i from country j. Yi and Yj are nominal GDP in country i and j respectively, Pi and Pj are population in country i and j, Dij is the distance between country i and j, Xij is a vector of variables describing either country i or j, or both. This vector includes the share of agriculture in GDP of the exporting country, the number of landlocked countries in the country pair (i.e., 0, 1, or 2), and dummy variables for trade between partners sharing the same language, for partners bordering each other, for partners where one colonized the other at some point in time, and for primary commodity exporters.9 To control for potential differences in north-north trade compared to trade involving developing countries (e.g. differences in the quality of infrastructure, human capital, labor and environmental standards etc.) a dummy for inter-industrial country trade is included. uij is described as (including time subscripts for clarity):

uijt = μij + εijt,

where μij are unobservable, time-invariant country pair-specific effects assumed to be uncorrelated with the other explanatory variables. εijt, is a normally distributed error term with zero mean.

21. Bilateral trade can be expected to depend positively on the size of the two economies, measured by GDP, and negatively on the distance between the countries. A large population is generally considered to relate negatively to trade, since this would imply a larger domestic market and a higher degree of auto-sufficiency. Moreover, for a given level of GDP, a larger population indicates a lower level of per capita income (a proxy for economic development) and hence generally a lower export capacity. It has, however, also been argued that a large population allows for scale effects and a more efficient division of labor and would therefore affect trade positively. Hence, the expected sign of the population of the exporting country is ambiguous in the model. Trade is likely to be higher between bordering countries10, countries sharing a common language, and countries with colonial ties, and lower for landlocked countries. The share of agriculture in GDP can be expected to correlate negatively to exports, since trade protection tends to be particularly high against agricultural products. Trade in primary commodities should ideally be excluded from the gravity model, given that terms of trade swings can cause significant volatility in the value of trade. However, IMF’s Direction of Trade Statistics (DOTS) provides data for total trade only, and the issue is instead addressed by including a dummy for commodity exporters.11 There is no strong a priory reason for this dummy to be positive or negative.

Empirical results

22. The econometric analysis is based on a panel dataset covering bilateral trade between 90 developing and industrialized countries for the period 1991-2002.12 The results from this global model are applied to actual data for Tunisia to determine that country’s benchmark level of bilateral trade. To smooth some of the volatility in the trade data, and limit the risk of calculating trade potentials based on unusual trade performances in a particular year, three-year averages are calculated for the periods 1991-93, 1994-96, 1997-99, and 2000-02. A random effects Tobit model is used for the estimations and pooled Tobit and OLS results are also reported for comparison (see the appendix for estimation results and a note on econometric issues). Time dummies are included in all regressions to control for time-specific events. The regressions yield reasonable results, with essentially all variables having the expected sign (perhaps with the exception of the dummy for north-north trade, which has a negative coefficient), nearly all significant at the 1 percent level. The results are robust to the inclusion or exclusion of intra-industrial (north-north) trade.

23. Tunisia’s trade potential is defined as the difference between the benchmark level of trade predicted by the model (2000–02 averages fitted to equation 1 in Table 1 in the appendix) and actual levels of bilateral trade. In other words, a positive trade potential indicates that Tunisia under-trades with the country in question, while negative potentials signify trade beyond predicted levels.13 These trade potentials are analyzed on a country-by-country basis rather than in the aggregate, since over-trading with one country cannot be seen as neutralizing a positive trade potential with another. Tunisia’s trade partners are divided into three regional groupings: the EU (pre-2004 accessions), Middle East and North Africa (MENA), and the rest of the world.14

Chart 1:
Chart 1:

Tunisian Exports

(average 1999-2002 in US$ million)

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A002

Source: IMF staff calculations
Chart 2:
Chart 2:

Tunisian Imports

(average 1999-2002 in US$ million)

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A002

Source: IMF staff calculations

24. Tunisia’s aggregate trade with the EU widely surpasses model predictions, but an important trade potential remains nonetheless. A closer look at EU-Tunisian trade reveals that the large wedge between predicted and actual trade levels is explained by significant over trading with a few countries, notably France and Italy. Meanwhile, Tunisia still under trades with half of the EU countries. There is even a substantial trade potential with the UK, estimated at nearly ½ percent of Tunisia’s GDP for exports and 1 percent for imports. Indeed, merchandise trade (imports plus exports) between Tunisia and the UK in 2002 was about one third less than, for example, trade with Belgium. This suggests that Tunisia could significantly increase its trade by targeting the UK.

Chart 3:
Chart 3:

Tunisia’s Trade Potential with the EU

(percent of GDP)

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A002

Source: IMF staff calculationsNote: Negative trade potentials indicate overtrade.

25. The wide disparity in Tunisia’s trade performance with the EU merits further analysis. Given that the same trade regime applies EU-wide, any trade-creating impact of EU-Tunisian integration should, in principle, affect Tunisia’s trade with all EU countries. Evidently, conventional tariff and nontariff barriers only explain part of Tunisia’s trade performance. Migration comes to mind as a potentially important factor that could facilitate the formation of business networks and other ties likely to enhance trade. Migration could not be included in the model due to data limitations but Tunisia’s largest “over-traders” in the EU also tend to be the countries with the largest Tunisian immigrant population.15 Part of Tunisia’s over-trade with France and Italy could possibly also be explained by intra-firm trade, although this can not be confirmed by the aggregate data used for this study. One could speculate that integration with the EU has affected trade primarily with countries with pre-existing ties to Tunisia, although further analysis of the evolution of trade patterns over time would need to confirm this, along with an analysis using more disaggregate data. It would also be useful to examine the role of the textile sector with regards to Tunisia’s overtrade with certain EU countries, to get an appreciation of the risks of lost trade associated with the expiration of the Multifiber Agreement (MFA) in 2005.16

26. Trade with non-EU countries broadly match predicted levels, reflecting over-trading with the MENA region roughly compensating for under-trading with the rest of the world. Total trade with non-EU, non-MENA countries fall somewhat short of predictions driven by weak trade with the US and Japan, more than offsetting over-trading with several other countries.17 Tunisia’s trade potential with the US alone (exports plus imports) is estimated at 2½ percent of GDP. As a comparison, Tunisia’s exports to the US in 2002 were only about one third higher than those to Iran and one third lower than those to India. Although a strong candidate, Switzerland is not included among the top trade potentials in the table below because data availability permits calculating only the export potential (¼ percent of GDP).

27. On the regional front, Algeria is one of Tunisia’s largest untapped sources for trade in the world, substantially surpassed by only the US, the UK, and Japan. This indicates that regional integration efforts are worthwhile, and that from Tunisia’s point of view, particular attention to Algeria could prove useful. Tunisia appears already well integrated with other countries in the region—even significantly over-trading with Libya and Morocco.18 Libya is Tunisia’s single largest trade partner outside the EU, with merchandise trade exceeding US$0.5 billion annually. To put Tunisia’s trade with Morocco in perspective, it is about double the trade with Portugal.

Chart 4:
Chart 4:

Tunisia’s Trade Potential (exports + imports), Top 5 Countries

(non EU/MENA, percent of GDP)

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A002

Chart 5:
Chart 5:

Tunisia’s Regional Trade Potential

(Percent of GDP)

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A002

Source: IMF staff calculationsNote: Negative trade potentials indicate overtrade

C. Conclusions and Policy Implications

28. There is significant potential to increase Tunisia’s trade. Despite the fact that Tunisia’s total trade substantially surpasses model predictions, large unexploited markets remain. Although the scope to increase trade within the EU (in particular the UK) is important, by far the largest potential lies outside the EU (the US and Japan). Tunisia’s total trade potential is estimated at 9 percent of GDP, one fourth of which with the EU. This underscores the need for Tunisia to accelerate trade liberalization on a multilateral basis, in parallel with the more advanced integration process with the EU. Within the region, Algeria presents a considerable potential to increase trade. Stepped up efforts to enhance regional integration may therefore also have a measurable impact on Tunisia’s total trade.19

29. Continued structural reform to improve competitiveness will be crucial to take advantage of new trade opportunities. An additional challenge will be to maintain Tunisia’s relatively large market share in certain EU countries (in particular France and Italy), especially with the imminent expiration of the MFA. Aside from trade liberalization, it will be important to create a business environment that is flexible and fully market-driven, so that entrepreneurs have the right incentives and the ability to exploit new trade opportunities. With Tunisia’s high unemployment and ready access to international capital markets, supply constraints are unlikely to be important hurdles in the medium to long term, assuming enabling structural reforms. One caveat, however, concerns the supply of specialized skilled labor: as Tunisia’s trade becomes more diversified, the labor force’s skill set will also need to be increasingly diversified. Ensuring that the formal education system matches the need for skills is therefore a priority.

30. Tunisia’s aggregate over trading with the EU should not be interpreted as a sign of serious trade distortions. Wide differences in the country’s trade performance with the EU countries suggest that factors other than conventional trade barriers need to be considered. Moreover, over-trading with a large number of non-EU countries casts some doubts on the idea that EU-Tunisian integration has caused substantial trade diversion, to the point were it would dominate the benefits of any trade creation with the EU. A more thorough analysis including changes in the trade performance over time would be useful before drawing firm conclusions.

APPENDIX Gravity Model Estimations

article image
Source: IMF staff estimates

Takes the value one if one of the partner countries is landlocked, two if both are landlocked, and zero otherwise.

Exporting country.

Note: Equations 3-6 are estimated using heteroscedasticity consistent standard errors.

***, **, *, and # indicate statistical significance at the 1, 5, 10, and 15 percent level, respectively.

Wald tests strongly rejected the pooled model in favor of the random effects (RE) model in Equations 1 and 2.

Testing between RE and fixed effects (FE) in the Tobit model was not feasible due the large number of parameters to estimate in the FE model (there are about 7000 country pairs).

Note on econometric issues:

Since the estimated model is expressed in logs, a solution is needed to deal with zero-value observations. Gravity model applications often deal with this issue by omitting zero-value observations. However, this truncates the joint distribution of the data, which introduces an estimation bias. This bias can potentially be sizeable, given that bilateral trade data typically include a large number of zero observations, particularly when developing countries are included. Other solutions include substituting zero observations with an arbitrary small number, or using nonlinear estimation techniques such as the Tobit model used here. The Tobit model explicitly incorporates information in the zero-value observations (for a discussion of the Tobit model in a panel setting, see e.g. B.H Baltagi, Econometric Analysis of Panel Data, John Wiley and sons, Chichester, 2001).

In any case, results turn out to be robust across estimation techniques, as illustrated by the chart below showing the estimated top and bottom 10 estimated trade potentials (divided between imports and exports). Although the magnitude of the estimated coefficients differ across estimation techniques, the net impact of these differences on the estimated trade potentials tends to be small. A significant impact on trade potentials is found only for a few large countries (mainly on the import side). The fact that large countries tend to be affected more than small ones is not surprising, given that the model is expressed in logs while the trade potentials are in absolute levels.

A02ufig01

Trade potential

(percent of GDP)

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A002

Source: IMF staff calculations.
8

Prepared by Ludvig Söderling.

9

All dummies take the value 1 or 0.

10

Although true in general, this is evidently not the case when cross-border traffic is restricted due to political or other conflicts.

11

The UN’s COMTRADE database does provide information on trade by major product group. Exploiting this could be useful for further research.

12

For another application on the same data, see Chapter II of the Selected Issues papers for the 2004 Article IV consultation for Morocco (IMF Country Report 04/164, 6/9//04).

13

A word of caution: while the methodology used here is by no means uncommon, it deserves to be mentioned that in-sample estimates of trade potentials have been criticized (see Peter Egger (2002), “An Econometric view on the Estimation of Gravity Models and the Calculation of Trade Potentials”, The World Economy, 25, pp. 297-312).

14

MENA countries are Algeria, Egypt, Jordan, Libya, Mauritania, Morocco, Syria, and Tunisia. Lebanon and the Palestinian Authority are excluded due to incomplete data.

15

According to the European Migration Centre, France had about 200,000 Tunisian residents in the early 1990s, Italy had 15,000, and Belgium over 6000. Data for other EU countries were not available.

16

The MFA gives Tunisia’s and other countries’ textile exports preferential access to the EU. The elimination of the MFA will greatly increase competition in textiles, in particular from China. This could potentially have significant consequences for Tunisia’s exports, one third of which is textiles.

17

The main non-EU, non-MENA countries trading with Tunisia beyond predicted levels are Turkey, Russia (imports only), India, Argentina (mainly imports), and Brazil.

18

It deserves to be mentioned, however, that Tunisia’s trade with the two other countries within the Fund’s Maghreb Initiative (Algeria and Morocco) represents only 2 percent of total Tunisian trade.

19

Regional integration is also important to mitigate potential “hub-and-spoke” effects arising from the EU’s separate trade agreements with Southern and Eastern Mediterranean countries.

Tunisia: Selected Issues
Author: International Monetary Fund