UK-China Trade and the J-Curve: Asymmetric Evidence from 68 Industries

Abstract

Using monthly data over the period 2010M1-2018M12 we investigate the impact of the real yuna-pound rate on the trade balance of each of the 68 2-digit industries that trade between Britain and China. When a linear ARDL model was estimated, we found short-run symmetric effects in the trade balance of 45 industries that lasted into the long run only in 24 industries. Comparable numbers from the estimates of the nonlinear model were 64 and 39, respectively. The symmetric (asymmetric) J-curve effect received support in 17 (19) industries from the linear (nonlinear) model. Although the increase from the symmetric to asymmetric J-curve effect was not substantial, for the largest industry, i.e., industry coded 85 (Electrical machinery and equipment with 19.56% share of trade), while the symmetric J-curve was not supported, the asymmetric J-curve was. This highlights the significance of the nonlinear adjustment of the yuan-pound exchange rate.

Keywords:

• The U.K.;
• China;
• 68 industries;
• J-curve;
• asymmetry analysis

Notes

1 For more on Yuan valuation see Wang (2020) and on China’s trade policy see Liu (2020). For China’s trade balance with the U.S., see Tu and Zhang (2019).

2 It must be noted that available estimates of the Marshall-Lerner condition between U.K. and rest of the world by Goldstein and Khan (1978), Warner and Kreinin (1983), Marquez (1990), Rose (1991), Bahmani-Oskooee and Niroomand (1998), and Bahmani-Oskooee and Kara (2005) reveal that, indeed, the Marshall-Lerner condition is satisfied by most estimates. For a comprehensive review of the literature see Bahmani-Oskooee et al. (2013).

3 See Bahmani-Oskooee (2020) for more on recent developments.

4 For review of the entire literature, see Bahmani-Oskooee and Ratha (2004) and Bahmani-Oskooee and Hegerty (2010).

5 The approach in this paper closely follows Bahmani-Oskooee and Kanitpong (2019) in this journal.

6 Note that income elasticities in (1) could take opposite signs if increased economic activity is due to an increase in the production of import-substitute goods (Bahmani-Oskooee, 1986).

7 Note that Bahmani-Oskooee (2020) has demonstrated that estimate of η0 is exactly the same as estimate of the coefficient attached to the lagged error-correction term in Engle and Granger (1987) setting. The estimate must also be negative.

8 Shin et al. (2014, p. 291) even recommend treating the two partial sum variables as a single variable so that the critical values of the F test for cointegration stay at the same high level when we move from the linear to nonlinear model.

9 For some other application of these models see Apergis and Miller (2006), Gogas and Pragidis (2015), Durmaz (2015), Baghestani and Kherfi (2015), Nusair (2017), Aftab et al. (2017), Arize et al. (2017), Lucarelli et al. (2018), Istiak and Alam (2019), Hajilee and Niroomand (2019), Olaniyi (2019), and Bahmani-Oskooee and Nasir (2020).

10 Short-run estimates attached to the income variables are available upon request.

11 Note that by meaningful we mean cointegration is supported either by the F or t-test that are reported in the diagnostic Table 3.

12 Trade shares are reported in Table 2.

13 The 34 industries are coded as 04, 07, 17, 19, 23, 28, 33, 34, 35, 41, 42, 44, 51, 52, 57, 59, 60, 61, 62, 63, 64, 68, 69, 70, 72, 73, 75, 81, 83, 85, 90, 94, 96, and 97.

14 Once again industries in which the PAPR and PDEP variables take a negative coefficient are industries in which either British import demand or Chinese import demand are price inelastic. Other diagnostics statistics are similar to those of the linear models and need no repeat.