BSc in Economics and Finance Faculty of Economics & Business
Bachelor thesis under the supervision of Ms. Rui Zhuo
The effect of exchange rate volatility on bilateral
trade flows between the United Kingdom and United
States
Jurriaan van Velthoven 10978569 June 2018
2 Statement of Originality
This document is written by Student Jurriaan van Velthoven who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 Table of contents 1. Introduction 5 2. Literature review 6 Theoretical research Empirical research 3. Methodology 10 Model Specification Description of the data Test procedure
4. Results 14
5. Discussion 16
6. Conclusion 17
4 Abstract
This paper contributes to the existing literature by empirically estimating the effect of exchange rate volatility on the export flows from the United Kingdom to the United States for a sample period of 1992-2017. Based on quarterly data the effect is analysed for a period that solely consist of a floating exchange rate regime. The foundation of this empirical research relies on the Engle Granger approach which includes long run modelling using non-stationary time series data. Empirical results of this paper confirm a significant negative effect of exchange rate volatility in the long run.
5
1. Introduction
On the 23th of June 2016 the Brexit referendum took place in the United Kingdom. The participating electorate won the referendum with 51.9% of the votes and with this slight margin of 1.9% the Brexit was confirmed. The announcement led to a shock in the global financial system and a series of economic consequences followed. One of those consequences was a drop of the pound against the dollar. The pound fell more than 10% against the dollar after the announcement which resulted in its lowest value since 1985 (Mackenzie & Platt, 2016). These fluctuations in the exchange rate are known as exchange rate volatility. In this paper the effect of exchange rate volatility on exports from United Kingdom to the United States are analysed based on a time period solely consisting of a floating exchange rate regime starting from the last quarter of 1992 till the last quarter of 2017.
The debate regarding the effect of exchange rate volatility started after the breakdown of the Bretton-Woods system. The Bretton-Woods agreement was established to rebuild the international economic system after the Second World War. In order to encourage and stabilize trade a fixed exchange rate regime was implemented; the participating nations pegged their exchange rate to the dollar. The system fell in 1973 leading to a new era where many nations acquired a floating exchange rate regime. The increase in risk that resulted from this floating exchange rate regime led to extensive research on the effect of exchange rate volatility. However, theoretical and empirical studies fail to establish a uniform hypothesis regarding the effect of exchange rate volatility on international trade flows.
The empirical research of this paper relies on the Engle Granger cointegration approach which allows for modelling the long run relationship of non-stationary time series data. The exchange rate volatility is based on the moving standard in this analysis. This measurement of exchange rate volatility will be implemented in the long run traditional export demand model to determine its effects.
This paper commences with a literature review in section 2 consisting of theoretical and empirical studies regarding the effect of exchange rate volatility. In section 3 the model and specific research method used to empirically analyse the effects of exchange rate volatility on the bilateral trade flows between the United Kingdom and United states are addressed. Section 4 provides the results of the empirical study and these findings will be discussed in section 5. Section 6 provides the conclusion of the paper.
6
2. Literature Review
Theoretical research
Many countries made the transition of a fixed exchange rate regime to a floating exchange rate regime after the breakdown of the Bretton-Woods system. Several theoretical models have been established to model the effects of exchange rate volatility in a floating exchange rate regime.
One of the first papers regarding the effect of exchange rate volatility is provided by Clark (1973). A firm level model is implemented in this analysis to determine the effect of exchange rate volatility on the volume of homogenous export goods. There are no imported goods included in this analyses and output is sold against a fixed foreign price under perfect market conditions. These assumptions are made to ensure the output decision doesn’t influence the price of the commodity or the exchange rate. There is a forward exchange market of ninety days which provides the only hedging facility. The long run output horizon of the firm is assumed to be longer than ninety days and therefore the firm will be exposed to exchange rate volatility. Clark (1973) infers exchange rate volatility has a negative effect on the volume of exports since firms are adversely affected by risk. However the assumptions made in this analyses are restrictive and do not provide realistic market conditions.
Ethier (1973) also conducted a theoretical approach to determine the effect of exchange rate volatility on trade. The study is based on a risk adverse importing firm and assumes revenues of the firm are influenced by the future exchange rate. Ethier (1973) concludes that the volume of trade is not influenced by exchange rate volatility. With the assumption of perfect forward markets the effect of exchange rate volatility only influences the amount of forward cover taken by the firm, therefore only the hedging position is affected.
The studies from Clark (1973) and Ethier (1973) have provided the first theoretical insights of the effect of exchange rate volatility after the collapse of the Bretton Woods system. In later studies performed by De Grauwe (1988) and Dellas & Ziberfarb (1993) the impact of exchange rate volatility results in an ambiguous effect for international trade flows. The debate regarding the effect of exchange rate volatility on international trade flows will be further analysed by empirical research.
7 Empirical research
Trade flows
The effect of exchange rate volatility can be derived for three groups of trade flows: aggregate, bilateral and sectoral. Aggregate refers to trade flows between multiple nations, bilateral to trade between two nations and sectoral to trade in commodity groups. The use of these different types of trade flows in empirical research will be further elaborated in this section.
Aggregate
Chowdhury (1993) examines the effect of exchange rate volatility on exports for the G-7 countries (United Kingdom, France, Germany, Italy, Japan, United States and Canada) based on aggregate time-series data for a sample period of 1973-1990. The results indicate that exchange rate volatility has a significant negative effect on exports. However implementing aggregate time series data could lead to biased results since no distinction is made in the effect of exchange rate volatility between the nations. This assumption of uniformity could lead to biased results if the effect of exchange rate volatility is not uniform between the nations studied.
Bilateral
To overcome biased results, many studies use bilateral time series data to determine the effect of exchange rate volatility on exports. Pozo (1992) uses this bilateral approach on trade flows between The United Kingdom and the United States for a sample period of 1900-1940. As in the study by Chowdhury (1993) equivalent economic fundamentals are implemented for the export model. The results indicate a negative relationship of exchange rate volatility on exports however do not rely on the implicit assumption of a uniform effect of exchange rate volatility as in the study by Chowdhury (1993). Thursby & Thursby (1987) also implement bilateral time series data for their analyses. The study relies on a gravity model to determine the effect of exchange rate volatility between seventeen countries based on a sample period of 1974-1982. In contrast to Chowdhury (1993) Thursby & Thursby (1987) use bilateral time series data for each country since pooling data is an inappropriate measurement according to their study. Exchange rate volatility results in a significant negative relationship in ten of the seventeen countries studied.
8
Sectoral
Belanger (1992) implements sectoral time series data to determine the effect of exchange rate volatility between the United States and Canada. The study is based on the following five sectors: food, industry supplies, vehicles, consumer goods and capital goods for a sample period of 1974-1987. Exchange rate volatility does not result in a significant negative relationship, except for a possible negative relationship in the sector capital goods. Of the forty estimates performed two coefficients result in a negative relationship. The use of sectoral data can therefore provide additional insights of the impact of exchange rate volatility among sectors. Klein (1990) implements sectoral time series data to test the effect on exchange rate volatility on nine sectors of exports. Exports from the United States to seven industrial countries (Canada, United Kingdom, the Netherlands, West Germany, Japan, France and Italy) are analysed for a period of 1978-1986. A significant effect of exchange rate volatility is determined in six sectors. From those sectors five indicate a positive effect and one a negative effect. These results indicate that the effect of exchange rate volatility could vary among sectors.
Non-stationary time series data
The effect of exchange rate volatility is extensively studied however only a few researchers focus on the characteristics of the data. In the study by Asseery & Peel (1992) a two stage Engle Granger procedure is implemented. The study emphasises the importance of potentially non-stationary integrated time series since neglecting this could lead to highly misleading results. Non-stationary time series have time changing statistics which should be taken into account when these series are used in a model. The Engle Granger procedure allows for modelling non-stationary time series which will be further elaborated in the methodology. In the study aggregate trade flows between Australia, Japan, The United Kingdom, The United States and West Germany are analysed for a sample period of 1972-1987. The results of the study indicate that exchange rate volatility has a significant positive effect on exports in the short run for all countries except the United Kingdom.
Chowdhurry (1993) implements a multivariate error correction model and analyses the short-run as well as the long-short-run effects. As in the study by Asseery & Peel (1992) the importance of possible non-stationary integrated time series are taken into consideration. For all countries a negative long run effect is found at 5% significance expect for Italy where a negative effect
9 at 10% significance is determined. Contradicting with the study by Asseery & Peel (1992) a significant negative effect of exchange rate volatility on exports is found for all countries in the short run. More recent studies by Arize (1995) and Doğanlar (2002) also emphasize the importance of non-stationary time series data. Taking all these findings into account, this paper will extend the existing literature by empirical research on the effect of exchange rate volatility.
10
3. Methodology
Model specification
A long run export demand model is used to determine the effect of exchange rate volatility on exports and defined as follows:
ln Xt = β0 + β1 lnYt + β2 lnRt + β3 Vt + εt (1)
The model relies on the long term behavioural model developed by Gotur (1985) which is used in several studies Chowdhury (1993), Asseery & Peel (1991) and Pozo (1992). The dependent variable Xt represents the nominal volume of exports from the United Kingdom to
the United States at time t. The first explanatory variable Yt represents the nominal gross
domestic product from the United States at time t. The second explanatory variable Rt
represents the bilateral real exchange rate between the United Kingdom and the United States at time t. The last explanatory variable Vt represents the moving average of exchange rate
volatility at time t. In this equation ‘’t’’ represents the time period, which consists of the last quarter of 1992 till the last quarter of 2017. As explained in the theoretical and empirical literature the effect on exports is ambiguous and will be determined in this study. Since the focus of the paper is on the effect of exchange rate volatility the measurement of this variable will be further elaborated.
Several studies have tested if real- or nominal exchange rate volatility should be used as a measurement for exchange rate volatility. Qian & Varangis (1994) state that the use of nominal or real exchange rate volatility should not significantly impact the results. This assumption has been analysed by Thursby & Thursby (1987), Mckenzie & Brooks (1997) by applying nominal as well as real exchange rates in their research. The results indicate no significant difference between the two measurements. Gotur (1985) states that the use of real exchange rates is a more appropriate measurement, since the price movements in the medium term will influence the proceeds of an exporting firm. These price movements are captured by the real exchange rate. Based on the reasoning of Gotur (1985) the real exchange rate will be implemented in this paper.
11 Many approaches have been used for modelling the volatility of the exchange rate since no generally accepted model is available. In this paper the moving standard deviation of real exchange rates will be implemented and is defined as follows:
Vt = [( 1 12) ∗ ∑ (𝐿𝑛𝑅𝑒𝑟𝑡−𝑖 12 𝑖=1 − 𝐿𝑛𝑅𝑒𝑟𝑡−𝑖−1)2] (12) (2)
In this equation Vt represents the quarterly real exchange rate volatility at time t. LnRer
represents the natural logarithm of the quarterly real exchange rate at time t. As in the study by Pozo (1992) a moving average of twelve will be used where ‘’t’’ represent the last quarter of 1992 till the last quarter of 2017.
Data sources
As mentioned by Thursby & Thursby (1987) using aggregate data is an inappropriate measurement and can lead to biased results since a uniform effect of exchange rate volatility is assumed among multiple nations. Therefore the effect of exchange rate volatility will be analysed using bilateral trade flows. A floating exchange rate regime for the period of 1992-2017 will be analysed based on quarterly data to determine the impact of exchange rate volatility on exports. The sample period solely makes use of a floating exchange rate period to omit the possibility of specification bias. This specification bias could occur due to the change of a fixed exchange rate regime to a floating exchange rate regime (Kenen & Rodrik, 1986). On October 1990 the United Kingdom entered the European exchange rate mechanism. The intention of this system was to reduce exchange rate volatility in Europe; the system set fluctuation boundaries in order to achieve this result. On 16 September 1992 the United Kingdom left the system, therefore the last quarter of 1992 till the last quarter of 2017 will be analysed to omit the possible specification bias (Werner & Burnham, 1996).
The data of this study is based on the four time series of the export demand model. The data for the exports from the United Kingdom to the United States is retrieved from the International Monetary Fund (IMF). Data regarding the GDP of the United States is retrieved from the Bureau of Economic Analyses. To obtain the data of the real exchange rate three datasets are used. First the data of the nominal exchange rate is retrieved from the Bank of England however since this is nominal data it has to be converted to real data. This is done by using the consumer price index (CPI) of the United Kingdom and the United States. Adjusting the nominal exchange rate with the price levels will result in the real exchange rate and
12 therefore the purchasing power of the two currencies are taken into account, the formula is defined as follows:
Real exchange rate = Nominal exchange rate (𝐶𝑃𝐼𝑈𝑘
𝐶𝑃𝐼𝑈𝑆)
(3)
The data of the consumer price index for the United Kingdom and the United States is retrieved from the Organisation for Economic Co-operation and Development (OECD). The data of the exchange rate volatility is computed by the moving average standard deviation as mentioned in the model specification.
Test procedure
The Engle Granger cointergration approach is used to derive the long run relationship between the exchange rate volatility and exports. If the variables are cointegrated they could temporarily deviate in the short run but will move together in the long run. The cointegration method developed by Engle & Granger (1987) is designed to test this long run relationship between non-stationary variables without causing a spurious regression. The residuals of the non-stationary variable should be tested for stationarity to determine if a long run relationship exists. If these residuals are stationary the variables are cointegrated and a long run relationship can be concluded.
Augmented Dickey Fuller
In order to perform the cointegration test by Engle & Granger (1987) all the time series variables should be tested for their integration order. Time series data can be integrated of order I(d) this implies the time series has to be differentiated ‘’d’’ times to become stationary. If the series are not integrated of the same order the method will conclude there is no cointegration between these time series variables. An Augmented Dickey Fuller test is performed to determine the order of integration for the time series variables. The Akaike information criterion is implemented to determine the minimum number of lags. The Augmented Dickey Fuller test is defined as follows:
∆𝑋𝑡= c + γ𝑋𝑡−1 + ∑𝑘𝑖=1θ𝑖 ∆𝑋𝑡−𝑖 + et (4)
In this section the model will be further elaborated on the basis of exports. In the equation ∆𝑋𝑡represents the first difference of the exports from the United Kingdom to the United
13 States. This first difference is obtained by measuring the change from one period to another. The variable ‘’c’’ represents a constant and γ represents the coefficient of the first lagged variable of exports 𝑋𝑡−1. This first lagged variable is the value of exports from the period: t-1. The summation sign represent the other lagged variables ∆𝑋𝑡−𝑖 and θ𝑖 is the coefficient of
these lagged variables. The last variable et represents the corresponding error term of the
model.
The Augmented Dickey Fuller test is performed to tests if the coefficient θ is equal to zero. If the coefficient is equal to zero a integration order of one is determined for the time series export. This will be tested with the following hypothesis: H0: γ = 0 versus H0: γ ≠ 0. If the
null hypothesis is rejected the time series exports will be differentiated ‘’d’’ times till their integration order I(d) is found in which the time series is stationary. This Augmented Dickey Fuller test will be performed for all the time series data of equation (1).
Cointegration
Cointegration among the time series variables will be analysed in order to determine the long term relationship. This cointegration analyses will only be further pursued if the time series are integrated of the same order. The cointegration analysis begins with estimating the export demand model from equation (1) by using a linear regression. The second step is to predict the residuals from the linear regression, this is defined as follows:
𝜀̂𝑡 = lnXt - β0 - β1 lnYt - β2 lnRt - β3 Vt (5)
The estimated residuals will be tested for stationary following the approach of Engle and Granger. However Davidson & MacKinnon (1993) point out that the critical values of the Augmented Dicky Fuller test should be corrected since the residuals are non-normally distributed. Therefore the critical values from Davidson & MacKinnon (1993) are implemented since these are corrected for the finite sample size. If the results indicate that the residuals 𝜀̂𝑡 are stationary, cointegration can be concluded. This implies that the exports and
explanatory variables have a long run relationship and the change in export volume from the United Kingdom to the United States is determined with the export demand model of equation (1). From this analysis the positive or negative impact of exchange rate volatility will be concluded.
14
4. Results
As mentioned in the test procedure the order of integration of the time series has to be defined before cointegration can be tested. Using the Augmented Dicky Fuller test the series are tested for integration at order I(0), this implies the series are not differentiated. The results of the test in Table 1 reveal this rejects the null hypothesis. From this result it can be concluded that the time series are non-stationary at level.
Variable AIC Critical value 5% T-statistic I(0)
Ln Xt 8 -2.896 -2.494 Rejected
Ln Yt 3 -2.892 -1.822 Rejected
Ln Rt 3 -2.892 -1.685 Rejected
Vt 2 -2.893 -2.547 Rejected
The variables are differentiated I(1) to determine if there integrated of order one. The results of the test in Table 2 reveal this accepts the null hypothesis. Therefore, it can be concluded that the time series are stationary at first order. The method of Engle and Granger will be further pursued since the series are integrated at the same order.
Variable AIC Critical value 5% T-statistic I(1)
Ln Xt 7 -2.896 -4.344 Accepted
Ln Yt 1 -2.892 -4.311 Accepted
Ln Rt 2 -2.892 -5.801 Accepted
Vt 1 -2.893 -5.621 Accepted
To obtain the residuals a linear regression of the model is performed. From this linear regression the residuals are obtained. These residuals are tested for stationary following the critical values of Davidson & MacKinnon (1993). The results in Table 3 indicate there is cointegration at 1% significance. This can be concluded since the value of the T-statistic: (-7.026) is lower than the 1% critical value: (-4.832). Therefore, significant cointegration among exports and the three explanatory variables is determined. The coefficient of the exchange rate volatility has a negative sign of (-1.61) and indicates that an increase in the exchange rate volatility has a negative impact on the exports from the United Kingdom to the United States in the long run. This result is consistent with the theory by Clark (1973) that states that the increase in exchange rate volatility will negatively affect exports volume since firms are adversely affected by risk.
Table 1: Augmented dicky fuller test, including lags based on the Akaike information criterion
15 Variable Critical value 1% T-statistic I(0)
𝜺̂ 𝒕 -4.832 -7.026 Accepted
Variable Coefficient Std. Error P-value
Ln Yt 0.9411 0.0282 0.000
Ln Rt -0.5720 0.0942 0.000
Vt -1.6101 0.5844 0.007
16
5. Discussion
Engle Granger two stage procedure
The use of the Engle Granger method has been adopted by several researchers to determine the possible cointegration of non-stationary time series as mentioned by Asseery & Peel (1992). The method by Engle Granger makes it possible to analyse these non-stationary time series however has some limitations. The main limitation of the method results in the assumption of only one cointegration relationship among the time series. In the case of more than two time series there may exist other cointegrating relationships. The possibility of more than one cointegration relationship among the four time series in this paper should therefore be further analysed in more extensive empirical research. Another limitation addresses the fact that the critical values of the Augmented Dicky Fuller test cannot be used to draw conclusions from the cointegration relationship. Therefore the critical values of the predicted residuals are corrected for the non-normal distribution by using the adjusted critical values from Davidson & MacKinnon (1993). Finally, Banerjee, Dolado & Smith (1986) addresses that small sample bias should be considered for the results of the cointegration approach. The Engle Granger method is a two-stage method to determine the long run as well as the short run effects of exchange rate volatility. In this paper only the first stage of the Engle Granger method is performed. The last stage of the Engle Granger method is based on an error correction model which includes the short run dynamics. Further research should be conducted to perform the second stage of the Engle Granger method.
Exchange rate variability
The paper makes use of the moving average standard deviation for modelling the exchange rate volatility since existing literature does not provide a general accepted model. The paper is limited to one model and could be extended by including different exchange rate volatility measurements. The use of these different estimating techniques and their effects on export should be further analysed.
17
6. Conclusion
In this study the impact of exchange rate volatility on the exports from the United Kingdom to the United States is examined using a sample period of 1992-2017. Using bilateral trade flows for a floating exchange rate period solely, overcomes possible bias in the estimation procedure. This estimation procedure is based on the Engle Granger cointegration method. The first stage in this method is to use the Augmented Dicky Fuller test to determine the integration order of the time series variables. The results indicate that the time series variables are all integrated of order one and the cointegration analyses established a significant long run relationship. In this long run relationship the effect of exchange rate volatility on exports is negative. This implies that exchange rate volatility will reduce export volume from the United Kingdom to the United States in the long run. However, it must be noted that further research has to be performed for encountering the shortcomings of the Engle Granger method. Finally, more extensive research should be conducted modelling exchange rate volatility since no general accepted measurement is available.
18
7. References
Arize, A. C., (1995). The Effects of Exchange-Rate Volatility on U.S. Exports: An Empirical Investigation. Southern Economic Journal, 62(1), 34-43.
Asseery, A., and Peel, D. A. (1991). The Effects of Exchange Rate Volatility on Exports.
Economics Letters, 37(2), 173-177.
Banerjee, A., Dolado, J., Hendry, D. and Smith, G. (1986). Exploring equilibrium relationships in econometrics through static models: some Monte-Carlo evidence.
Oxford Bulletin of Economics & Statistics, 48(3), 253-277.
Belanger, D., Gutierrez, S., Racette, D., and Raynauld, J. (1992). The Impact of Exchange Rate Variability on Trade Flows: Further results on sectoral U.S. imports from Canada. North American Journal of Economics and Finance, 3(1), 61-82.
Chowdhury, A. R., (1993). Does Exchange Rate Volatility Depress Trade Flows? Evidende form Error-Correction Models. The Review of Economics and Statistics, 75(4), 700-706.
Clark, P. B., (1973). Uncertainty, Exchange Risk, and the Level of International Trade.
Western Economic Journal, 11(3), 302-13.
Davidson, R., and MacKinnon, J. G. (1993). Estimation and Inference in Econometrics. United Kingdom: Oxford University Press.
De Grauwe, P. (1988). Exchange Rate Variability and the Slowdown in the Growth of International Trade. IMF Staff Papers, 35(1), 63-84.
Dellas, H., and Zilberfarb B. (1993). Real Exchange Rate Volatility and International Trade: A Reexamination of the Theory. Southern Economic Journal, 59(4), 641-647.
Doğanlar, M. (2002). Estimating the impact of exchange rate volatility on exports: evidence from Asian countries. Applied Economic Letters, 13(9), 859-863.
Engle, R. F., and Granger, C. W. (1987). Co-Integration and Error Correction: Representation, Estimation, and Testing. Econometrica, 55(2), 251-275.
19 Ethier, W. (1973). International Trade and the Forward Exchange Market. American
Economic Review, (63)3, 494-503.
Gotur, P. (1985). Effects of Exchange Rate Volatility on Trade: Some Further Evidence. IMF
Staff Papers, 32(3), 475-512.
Kenen, P., and Rodrik, D. (1986). Measuring and Analysing the Effects of Short-Term Volatility on Real Exchange Rater. Review of Economics and Statistics, 68(2), 311-15. Klein, M. W. (1990). Secoral Effects of Exchange Rate Volatility on United States Exports.
Journal of International Money and Finance, 9(3), 299-308.
Mackenzie, M. and Platt E. (2016, June 24). How global markets are reacting to UK’s Brexit vote. Financial Times. Retrieved from: https://www.ft.com/content/50436fde-39bb-11e6-9a05-82a9b15a8ee7
McKenzie, M., and Brooks, R. (1997). The Impact of Exchange Rate Volatility on German-U.S. Trade Flows. Journal of International Financial Markets, Institutions and Money, 7(1), 73-87.
Pozo, S. (1992). Conditional Exchange Rate Volatility and the Volume of International Trade: Evidence from the early 1900’s. The Review of Economics and Statistics, 74(2), 325-329.
Qian, Y., and Varangis, P. (1994). Does Exchange Rate Volatility Hinder Export Growth?.
Empirical Economics, 19(3), 371-96.
Thursby, M. C., and Thursby, J. G. (1987). Bilateral Trade Flows, the Linder Hypothesis and Exchange Risk. The Review of Economics and Statistics, 69(3), 488-495.
Watson, P. K., and Teelucksingh, S. S. (2011). A Practical Introduction to Econometric Methods: Classical and Modern. Jamaica: University Of The West Indies Press.
Werner, B. and Burnham, P. (1996). Britain and the Politics of the European Exchange Rate Mechanism 1990–1992. Capital and Class, 20(3), 5-38.
