The Sensitivity of U.S. Inpayments and Outpayments to Real Exchange Rate Changes: Asymmetric Evidence From Africa

ABSTRACT

A strand of the literature concentrates on assessing the impact of exchange rate changes on a country’s inpayments from and outpayments to its trading partners. Most studies have considered the U.S. experience with partners from OECD but not for those from Africa. We fill this gap by including 14 African partners in our study. We assess not only the symmetric effects of exchange rate changes on the U.S. inpayments from and outpayments to each African partner, but also the asymmetric effects which requires incorporating nonlinear adjustment of the exchange rate. While we found asymmetric effects in the trade with almost all partners in the short run, in a limited number of cases the short-run effects lasted into long-run asymmetric effects. All in all, our findings are partner specific, but they provide more support for using nonlinear models.

KEYWORDS:

• The U.S
• Africa
• inpayments
• outpayments
• asymmetry analysis

JEL CLASSIFICATION:

• F31

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1 Once again, a negative estimate of f will be an indication of the U.S. import demand being inelastic.

2 Note that once normalization takes place, we have $\frac{{\hat{\gamma }}_1}{-{\hat{\gamma }}_0}=\hat{b};\frac{{\hat{\gamma }}_2}{-{\hat{\gamma }}_0}=\hat{c};\frac{{\hat{\gamma }}_1^{\mathrm{\prime }}}{-{\hat{\gamma }}_0^{\mathrm{\prime }}}=\hat{e};and\frac{{\hat{\gamma }}_2^{\mathrm{\prime }}}{-{\hat{\gamma }}_0^{\mathrm{\prime }}}=\hat{f}.$

3 The critical values of Pesaran et al. (2001) are for large samples. For small samples such as ours, we use critical values from Narayan (2005). Note also that since critical values account for degree of integration of the variables, under this method variables must be a combination of I(0) and I(1) but not I(2). To verify this, we applied the ADF test to all second-differenced variables. Indeed, almost all second-differenced variables were stationary.

4 For example, on the asymmetric response of domestic prices to exchange rate changes, see Delatte and Lopez-Villavicencio (2012), and on the asymmetric response of import and export prices to exchange rate changes, see Bussiere (2013). The asymmetric response of the trade balance to exchange rate changes is evidenced in Bahmani-Oskooee and Fariditavana (2015), Arize, Malindretos, and Igwe (2017) and Nusair (2017).

5 Note that the two partial sum variables are constructed as $PO{S}_t=\sum _{i=1}^{t}max(\mathrm{\Delta }LnRE{X}_i,0),andNE{G}_t=\sum _{j=1}^{t}min(\mathrm{\Delta }LnRE{X}_i,0).$

6 For some other application of these methods see Verheyen (2013), Bahmani-Oskooee and Saha (2015), Gogas and Pragidis (2015), Durmaz (2015), Baghestani and Kherfi (2015), Pal and Mitra (2016), Al-Shayeb and Hatemi-J.(2016), Lima, Foffano Vasconcelos, Simão, and de Mendonça (2016), Arize et al. (2017), Aftab, Ahmad, I.Ismail, and Ahmed (2017) and Gregoriou (2017).

7 Other diagnostic statistics in Panel C reveal that residuals are autocorrelation free in most models and most of the optimum models are correctly specified. Coeffcient estimates are stable, and every nonlinear optimum model enjoys a good fit.