Singapore’s Trade in Financial and Insurance Services and the Role of the Exchange Rate: An Asymmetric Analysis

ABSTRACT

Two studies have assessed the link between the exchange rate and trade in services. While one study concluded that the U.S. insurance and financial service trade is not affected by the exchange rate, another study reversed that finding by applying a non-linear method. We add to this literature by considering Singapore’s experience. After estimating the linear and non-linear ARDL models using Singapore’s imports and exports of insurance and financial service trade, we conclude that changes in the real effective value of the Singapore dollar have both short-run and long-run effects on Singapore’s trade in both services.

KEYWORDS:

• Asymmetric analysis
• real effective exchange rate
• Singapore
• trade in services

Acknowledgement

Valuable comments of two anonymous reviewers are greatly appreciated. Any remaining error, however, is our own.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ As mentioned above, most studies have assessed the impact of exchange rate changes on trade in goods, and Singapore is no exception. For a review of Singapore related studies on trade in goods, see Bahmani-Oskooee, Aftab, and Harvey (2016) and Bahmani-Oskooee and Harvey (2018).

² Bahmani-Oskooee and Ghodsi (2018) have demonstrated that these estimates are equal to the estimate of the coefficient attached to the lagged error-correction term in the Engle and Granger (1987) approach.

³ Since most macro variables are either I(0) or I(1), under this method, there is no need for pre-unit root testing.

⁴ For evidence of the asymmetric response of traded goods to exchange rate changes, see Bussiere (2013), Bahmani-Oskooee and Fariditavana (2016), and Arize, Malindretos, and Igwe (2017).

⁵ Note that the two partial sum variables are constructed as follows: $\mathit{PO}{S}_t=\sum _{\mathit{j}=1}^{t}max\left(\mathrm{\Delta }{E}_j,0\right)$, and $\mathit{NE}{G}_t=\sum _{\mathit{j}=1}^{t}min\left(\mathrm{\Delta }{E}_j,0\right).$

⁶ Shin, Yu, and Greenwood-Nimmo (2014, 291) even argue that critical values of the F test for cointegration should stay at the same high level when we move from a linear to nonlinear model.

⁷ For some other application of these methods, see Baghestani and Kherfi (2015), Gogas and Pragidis (2015), Nusair (2012, 2017), Baek (2020), and Hajilee and Niroomand (2019, 2021).

⁸ Note that by meaningful, we mean cointegration is supported either by the F test or the t-test reported in Panel C of the table.

⁹ Other diagnostics are like those in Table 1 and do not need to be repeated.