On the relationship between export and economic growth: A nonparametric causality-in-quantiles approach for Turkey

ABSTRACT

This study re-examines the dynamic causal relationship between exports and economic growth in Turkey for the period from 1960 to 2018. Unlike the previous studies ignoring the presence of the potential nonlinearities in the relationship between the series; we apply a novel nonparametric causality-in-quantile methodology that relaxes the restrictive assumption of linearity and provides more reliable and inclusive inference in the causal nexus between the variables. Using the nonlinearity test, we show that the absence of causality linkages based on the linear framework is subject to a misspecification error. However, by employing the causality-in-quantiles test, we find evidence of positive causation from economic growth to export growth at low- and high-quantile ranges of export growth. Our findings highlight the emphasis on the modelling of nonlinear interactions between the variables as well as the consideration of the entire conditional distribution to avoid the risk of misleading inferences on the causality analysis.

Keywords:

• Quantile causality
• export-led growth
• output-driven export

JEL Classifications:

• F43
• F4
• F
• O10
• O1
• O
• C22
• C2
• C

Acknowledgments

We would like to thank two anonymous referees for many helpful comments. However, any remaining errors are solely ours.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Pejman Bahramian  http://orcid.org/0000-0002-8917-6306

Andisheh Saliminezhad  http://orcid.org/0000-0002-3074-7522

Notes

1 A few studies have pointed out the possible nonlinear relationship between exports and GDP growth. They highlighted that the effect of exports growth on GDP growth relates to diminishing returns such that, a level of exports growth might have been outstripped before the realization of the enhancing effects of exports (Awokuse and Christopoulos 2009).

2 The nonlinear Granger causality test of Diks and Panchenko (2006) is employed in this study as a benchmark method for detecting the nonlinear causal linkages between the variables.

3 We use SIC lag-length selection as it is more consistent in the presence of an over-parametrization problem which commonly exists in the nonparametric methodologies (for details see Hurvich and Tsai 1989).

4 For brevity, we do not report the details of the unit root tests. However, the results are available upon request from the authors.

5 Although the BDS test results demonstrate the need for a nonlinear approach; however, as a robustness check, the nonlinearity test of Terasvirta (1993) has also been applied to provide more supporting evidence for our argument. The test is performed in a Lagrange multiplier (LM) test framework under the null of linearity (for details see Saliminezhad and Lisaniler 2018). Our finding shows the rejection of the null hypothesis at 1% level of significance (the value of the test statistic is 224.16, with 0.006 for the p-value of the test). This result also confirms the existence of nonlinear relations among the variables under examination.

6 However, using a more conservative significance level of 10%, we reject the null of no causality from economic growth to export for a wider quantile range (e.g. from 0.15 to 0.25 in low quantile ranges and from 0.65 to 0.85 in high quantile ranges).

7 It is freely available from the Comprehensive R Archive Network (CRAN) at http://cran.r-project.org

8 We also employ an orthogonal polynomial basis of degree 10 for constructing the series transformation. However, the inferences are not different from what we found by applying the cubic B-Spline basis. Results are available upon request.