Investigating exchange rate shocks on agricultural trade balance: The case of Tunisia

ABSTRACT

The paper studies the impact of changes in Tunisia's exchange rate on the net external position of the agricultural sector. It shows that substitutability on production and consumption among domestically produced goods leads to an ambiguous impact for reasons that go beyond the Marshall-Lerner condition. Using cointegration techniques to disentangle the long and short-run impact of changes in the exchange rate on the net agricultural trade balance, we find that the depreciation of the domestic currency leads to a deterioration of the net external position of Tunisia's agricultural sector in the long-run.

KEYWORDS:

• Agricultural trade balance
• exchange rate
• Tunisia

JEL CODES:

• F14
• Q17
• C32

Acknowledgements

The authors would like to thank the editor and the reviewers for their valuable suggestions and feedbacks.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Houssem Eddine Chebbi http://orcid.org/0000-0001-6949-8677

Notes

1 Tunisia has gradually moved toward trade liberalization and much effort was deployed to facilitate regional integration. On October 2015, Tunisia and European Union launched the negotiations of a Deep and Comprehensive Free Trade Agreement (DCFTA). This DCFTA will build on the existing free trade area established since 1995 by the Euro-Mediterranean Association Agreement. Liberalization of trade in services and in agricultural goods will also be covered for the first time.

2 See for example the works of Boughzala and Sellaouti (2003) and Chebbi et al. (2011), for a review of the trade policy in Tunisia

3 See Rose and Yellen (1989) for a derivation in the general case when supply elasticities are not assumed to be perfectly elastic.

4 See also Bahmani-Oskooee and Wang (2007) who in a study of the impact of real exchange rate changes on the sector level net trade balance of Australia with the United States find that only around a third of the industries considered had a long-run positive and statistically significant coefficient. This suggests that the depreciation of the domestic currency leads to increases in the net sectoral trade balance.

5 However, the relationship between domestic currency depreciation and changes in the current account balance has been studied widely: Himarios, (1985); Bahmani-Oskooee, (1985); Rose (1990); Bahmani-Oskooee and Malixi, (1992); Bahmani-Oskooee and Alse, (1994); Backus et al., (1994); Boyd, Caporale, and Smith (2001); Bahmani-Oskooee and Ratha, (2004a) and (2004b).

6 Although, they conclude in their survey that in the long-run, generally, a depreciation of the domestic currency will lead to an improvement of the aggregate trade balance. But this tell us little about what happens at the sectoral level, and as the results by Yazici (2006), Ardalani and Bahmani-Oskooee (2007), and Bahmani-Oskooee and Wang (2007, 2008) suggest depreciations can well have a negative impact on sectoral trade balances in the long-run.

7 See Costamagna (2014) for a review of literature of the short and long-term effects of exchange rate behavior on the trade balance.

8 Please see the study of the World Bank (2014) for an analysis of the productivity growth in Tunisia and the reallocation of resources toward the most productive sectors.

9 For further details see the papers of Shabsigh and Domac (1999) and Sfia (2006).

10 This exchange rate policy, combined with prudent monetary and fiscal policies as part of the structural reforms, helped reduce inflation and establish a credible commitment to macroeconomic stability.

11 In 2004, a new Tunisian fiscal law provides for further capital account liberalization and exchange rate policy flexibility (World Bank 2010).

12 As mentioned by Lajmi and El Khadraoui (2014), the goal of this orientation, which was accompanied by an asymmetry in the response of the CBT to the change in the Euro-dollar parity, was twofold: First, to support Tunisian companies, including those benefiting from upgrade programs, to cope with the difficulties linked with the entry into force of the second phase of the Association Agreement between Tunisia and the European Union and the dismantling of the Multi-Fibre Arrangement). Second, to manage the situation of the tourism sector that was negatively affected by the 2002 terrorist attack.

13 To see this simply note that in the presence of a one tradable good, $p^A=P$, by definition. Then divide everywhere by the imported quantity, $m$, and note that if initially the trade balance is zero, then $x=\frac{x}{q}$. The last three terms become then the Marshall-Lerner condition.

14 In our empirical exercise, we use the official exchange rate ERO (units of dollar per units of Tunisian Dinar) since data for Real Effective Exchange Rate (REER) is only available from 1975 and this will limit the degrees of freedom we have when using the ERO. Importantly, the correlation coefficient between the two variables is 0.9 indicating a strong relationship between the two variables. (See Appendix 1 for a detailed description of the selected variables).

15 These authors modify the original agricultural trade model developed by Chambers (1981) to take into account the interaction between agricultural exports, imports and domestic macroeconomic factors.

16 Appendix 2 presents summary statistics for the variables used.

17 In April 1972, an off-shore regime for exports was created in order to encourage the emergence of industrial exports. The first off-shore companies law of 1972 marked the transition of Tunisia to economic liberalization after cooperativism.

18 Lanne, Lütkepohl, and Saikkonen (2003) extended Lanne, Lütkepohl, and Saikkonen (2002) non-linear break tests to the case of an unknown break date.

19 Johansen (1988, 1995) and Johansen and Juselius (1990, 1992, 1994).

20 In this case, the jointly determined variables do not contain deterministic trend but may have non-zero means. In addition, multivariate tests for autocorrelation (Godfrey 1988) and normality (Doornik and Hansen 1994) have been carried out to check for model statistical adequacy before applying the reduced rank tests. Results support the VAR model with two lags as a sufficient description of the data set.

21 Johansen cointegration methods may produce unreliable results for small sample. Therefore, for robustness check, we conduct also the trace statistic tests computed using the Bartlett small sample correction. For a discussion of the properties of both tests in small samples, see Lütkepohl, Saikkonen, and Trenkler (2001).

22 As suggested by the reviewers, we also estimate 2 models for each system, in order to check for the presence of structural break on the cointegration relationships (i.e., 1987: the date of the implementation of the program of structural reforms in Tunisia). The results of these estimations are available upon request and confirm the selected models we reported in our manuscript.

23 Our result is in line with the finding of the study conducted by the World Bank (2009). This study identifies at least four factors that have played a key role helping Tunisian industries to maintain competitiveness: First, Tunisian firms benefited from various support and upgrade programs. Second, the Tunisian dinar depreciated steadily vis-à-vis the Euro since 2001. Third, EU final manufacturing products, which enjoy better access to the Tunisian markets since 2000, are more expensive than locally manufactured goods. Particularly, product differentiation has allowed more affluent consumers in Tunisia to access better quality products while permitting domestic firms to continue securing a segment of the domestic market. Finally, the tariff dismantling is incomplete in many sectors and is quite recent for the “sensitive” sectors.

24 Granger-causality implies causality in the prediction (forecast) sense rather than in a structural sense.

25 Estimates from lagged endogenous term for the "AGRI-TRADE" system indicate that agricultural output is the only significant coefficient and seems to be more important than the exchange rate, agricultural prices, and per capita disposable income in explaining the variation in agricultural trade balance in the short run (see Appendix 3). Our result substantiate in part the findings of Gil, BenKaabia, and Chebbi (2009) that conclude that agricultural exports depend more on other factors than on the Tunisian agricultural output performance, for example, commercial agreements (i.e., most of the exported agro-food products are sent to the European Union and are subject to quantity restrictions) or decisions made by existing exporter lobbies in the most important export goods (olive oil, dates, citrus fruit, etc.).