Asymmetric effects of exchange rate changes on the demand for money in China

ABSTRACT

In order to account for currency substitution, the majority of recent studies relating to the specification of the demand for money include the exchange rate as another determinant of the demand for money. However, those who have estimated the demand for money in China have been unable to find any significant effects of exchange rate changes on the demand for money by the Chinese. We show that this is due to the assumption that exchange rate changes have symmetric effects. Once depreciations are separated from appreciations of the yuan, those exchange rate changes are shown to have significant effects on the demand for money in China, but in an asymmetric manner.

KEYWORDS:

• China
• money demand
• asymmetry
• exchange rate
• non-linear ARDL approach

JEL CLASSIFICATION:

• E41
• F31

Acknowledgements

Valuable comments of two anonymous referees are greatly appreciated. Remaining errors, however, are ours.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Indeed, Laidler (1993, 175–6) was the first to point this out.

² For the exact normalization procedure, see Bahmani-Oskooee and Fariditavana (2015).

³ For some other applications of this approach, see De Vita and Kyaw (2008), Bahmani-Oskooee and Gelan (2009), and Hajillee and Al-Nasser (2014).

⁴ For similar arguments in the case of the demand for money in Iran, see Bahmani-Oskooee and Bahmani (2015) and for some other applications of the partial sum concept, see Apergis and Miller (2006), Delatte and Lopez-Villavicencio (2012), and Verheyen (2013).

⁵ In practice, POS at time t is the cumulative sum of positive changes in LnEX prior to time t (including t itself) where negative changes are replaced by zeroes. Similarly, NEG at time t is the cumulative sum of negative changes in LnEX, where positive changes are replaced by zeroes.

⁶ Note that the main reason to begin with 1996 is that the effective exchange rate of the yuan is only available from that date. See the Appendix for exact source.

⁷ The critical value comes from Pesaran et al. (2001, Table CI, Case III, p. 300).

⁸ For more details of this procedure, see Bahmani-Oskooee and Fariditavana (2015).

⁹ Note that Banerjee, Dolado, and Mestre (1998) have demonstrated that the t-ratio to judge the significance of ECM_t−1 is nonstandard. Hence, they tabulate new critical values. The critical value at the usual 5% significance level from their Table 1 is −3.82. This is for the case of k = 3 (three exogenous variables).

¹⁰ Note that in the non-linear model we have four regressors; hence, this critical value is for K = 4 and at the usual 5% significance level, again, from Pesaran et al. (2001, Table CI, Case III, p. 300).

¹¹ Note that, this time we must use the 5% critical value of –4.03 when K = 4. This comes again from Banerjee, Dolado, and Mestre (1998, Table 1).

Additional information

Funding

Dan Xi would like to acknowledge financial support by the Ministry of Education of the People’s Republic of China, project of Humanities and Social Science [grant number 14YJC790135].