Exchange rate nonlinearities in India’s exports to the USA

Abstract

Purpose

This paper aims to investigate the determinants of bilateral exports of India to the USA by taking the non-linearity issue in export demand equations which is neglected so far in the empirical work. The study tries to know the reaction of change in exports to exchange rate changes in a non-liner fashion. For this purpose, non-linear autoregressive distributed lag (NARDL) bounds testing approach of Shin et al. (2011) has been used. This approach allows testing for non-linearities both in the short and long run, which might give indications of strategic pricing and non-linearities in exchange rate. The empirical analysis is carried out for bilateral export demand relationships of India with the USA for the period from January 1993 until December 2013. The overall results show that exports are determined in the long run by foreign demand, exchange rates and relative prices. The assumed linearity in export demand functions might be too restrictive. Thereby, the one threshold model that distinguishes exchange rate effects between appreciations and depreciations delivers plausible results. If exchange rate non-linearities are detected, it would seem that exports respond stronger to appreciations than to depreciations. A reason for this might be that firms perform strategic pricing in international trade to gain or maintain market shares.

Design/methodology/approach

The paper uses the newly developed non-linear ARDL framework of Shin et al. (2011) to investigate whether there are non-linearities with respect to the exchange rate for India’s exports to the USA. One of the important features of this framework is that it is free from unit root pre-testing and can be applied regardless of whether variables are I(0) or I(1). In addition, ARDL and NARDL technique efficiently determines the cointegrating relation in small sample. The short-run and long-run parameters with appropriate asymptotic inferences can be obtained by applying OLS to NARDL with an appropriate lag length. Following is the NARDL representation of equation 4(a) and 4(b). For brevity, this is illustrated for 4(a) only, where is the first difference operator, P is the drift component and it is the white noise residual, the coefficients ?_1 to ?_4 represent the long-run relationship, whereas remaining expressions with summation sign represent the short-term dynamics of the model. This equation nests the linear ARDL model presented in Pesarean et al. (2001) for the case of d_k^+=d_k^-and ?_2=?_3for all k. Thus, equation is less restrictive than a linear model. For this test, as its distribution is non-standard, Pesarean et al. (2001) tabulate the critical values. The bound test is used to examine the existence of the long-run relationship among the variables in the system. This test is based on Wald/F-statistic and follows a non-standard distribution. To check whether a cointegrating relationship exists, one has to test the null hypothesis ?_1=?_2=?_3=?_4 = 0 in the equation. Pesarean et al. (2001) provide two sets of critical values in which lower critical bound assumes that all the variables in the ARDL are I(0) and upper critical bound assumes I(1). The null hypothesis of cointegration is rejected if the calculated F-statistics is greater than the upper bound critical values. If the F-statistics is below than the lower critical bound, then null hypothesis cannot be rejected; this indicates no cointegration among the variables. If it lies within the lower and upper bounds, the result is inconclusive. After examining the cointegration, long-run coefficients are calculated by estimating the model with the appropriate lag orders based on the Schwarz Information Criteria (SIC). Further, the short-run dynamics of the model is also analyzed by using unrestricted error correction model based on the assumption made by Pesarean et al. (2001). Thus, the error correction version of the NARDL model pertaining to the central export equation can be expressed as: 10; 10, where ? is the speed of adjustment parameter, and EC is the residuals that are obtained from the estimated cointegration model of equation 4(a). The EC term is expressed as 10; 10, where are the OLS estimators obtained from the equation (5a). The coefficients of the lagged variables provide the short-run dynamics of the model covering the equilibrium path. The error correction coefficient ( ) is expected to be less than zero, and its significant value implies the cointegration relation among the variables. Finally, various tests such as serial correlation, functional form, normality and heteroskedasticity have been conducted to check the performance of the model.

Findings

Many empirical studies have estimated the elasticities of different final export demand components with respect to the exports because of their importance in trade policy formulation. But all the work has accounted only linearity in the exchange rate in export demand equation. Hence, in this paper, we tried to estimate non-linearities in export demand equation. The study fills the gap in the literature by improving on existing literature with the incorporation of the newly developed NARDL approach of Shin et al. (2011). This approach allows testing for non-linearities both in the short- and in the long run which might give indications of strategic pricing and non-linearities in exchange rate. The empirical analysis is carried out for bilateral export demand relationships of India with the USA for the period from January 1993 until December 2013. The bound test shows that there exists cointegration among the variables. Results show that exports are determined in the long run by foreign demand, exchange rates and relative prices. The long-run coefficients have got the expected sign and are of reasonable magnitude and statistically significant. Regarding non-linearities, the results show that assuming linearity in export demand functions might be too restrictive. Thereby, the one threshold model that distinguishes exchange rate effects between appreciations and depreciations deliver plausible results. If exchange rate non-linearities are detected, it seems that exports respond stronger to appreciations than to depreciations. A reason for this might be that firms perform strategic pricing in international trade to gain or maintain market shares.

Originality/value

The originality of this paper lies in the fact that it applies NARDL approach to Indian trade data (export demand) and analyzes the asymmetrical and non-linear impact of exchange rate changes on Indian exports.

Citation

Mathur, S.K. and Shekhawat, A. (2021), "Exchange rate nonlinearities in India’s exports to the USA", Studies in Economics and Finance, Vol. 38 No. 1, pp. 1-12. https://doi.org/10.1108/SEF-07-2015-0179