11

Click here to load reader

The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

  • Upload
    burcu

  • View
    213

  • Download
    1

Embed Size (px)

Citation preview

Page 1: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

THE PRICE OF GOLD AND THE EXCHANGE RATE:

EVIDENCE FROM THRESHOLD COINTEGRATION

AND THRESHOLD GRANGER CAUSALITY

ANALYSES FOR TURKEY

Burak GÜRÉÓ – Burcu KIRAN

(Received: January 24, 2012; revision received: February 24, 2012;

accepted: October 24, 2013)

This paper explores the relationship between gold prices and the US dollar/Turkish lira exchangerate between 1990–2011 by using cointegration and Granger causality analyses. The empirical find-ings indicate that there is a threshold cointegration relationship between the two variables. Thethreshold value obtained from the estimation of threshold vector error correction model equals–3.268. The Granger test indicates that there is evidence of a bi-directional causal relationship be-tween gold prices and the exchange rate, except when the threshold parameter exceeds the thresholdvalue in the exchange rate equation. According to these findings, gold price can be used as a hedgeagainst the exchange rate. However, since this relationship disappears above the threshold value,gold is only a weak hedge against exchange rate fluctuations.

Keywords: threshold cointegration, threshold vector error correction model, threshold Grangercausality, gold price, exchange rate, Turkey

JEL classification indices: C20, EO

INTRODUCTION

For centuries, gold has been the foundation of monetary systems. From 1900 to1971, with the global systems of gold standard and then the dollar standard, goldprice was regulated. Gold standard meant that the monetary authority held suffi-

0001-6373/$20.00 © 2014 Akadémiai Kiadó, Budapest

Acta Oeconomica, Vol. 64 (1) pp. 91–101 (2014)DOI: 10.1556/AOecon.64.2014.1.5

Burak Güréó (32), corresponding author. Associate Professor of Econometrics at Department ofEconometrics, Faculty of Economics, Istanbul University. E-mail: [email protected]

Burcu Kiran (30), Assistant Professor at the Department of Econometrics, Faculty of Economics,Istanbul University. E-mail: [email protected]

Page 2: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

cient gold reserves to convert all of the representative money into gold at a prom-ised exchange rate. In other words, it is a legal enactment to the effect that the le-gal tender of a country is convertible on demand into a specified quantity of gold(Jones 1933). Sachs – Larrain (1993) state that, under the gold standard, currencyunits were defined in terms of a given quantity of gold and were typically convert-ible into gold (Karunagaran 2011). Since 1972, gold has been disconnected fromthe American dollar (USD). In 1976, the International Monetary Fund (IMF),which was established to make arrangements for fixing exchange rates among na-tional currencies with some measure of flexibility, agreed upon the future of thegold standard and the international system. With the Jamaica Agreement, the IMFeliminated the pegging of gold to the USD and accepted managed floating ex-change rates. Since then, currencies are fiat money, not redeemable by gold; theo-retically, the money supply is infinitely expandable and gold price is determinedby market supply and demand only.

The demand for gold can be divided into two categories: the first one is the “usedemand” for gold. It can be used in the production of jewellery, medals, coins,electrical components and also in dentistry. The second one is the “asset demand”for gold where it is used by governments, fund managers and individuals as an in-vestment. The asset demand for gold is traditionally associated with the view thatgold is an effective “hedge”. In other words, gold represents a value store that in-vestors believe will insulate them against inflation and other forms of uncertainty(Ghosh et al. 2002). The gold price was called as a “very good indicator of futureinflation” by Alan Greenspan (Forbes 1991). The Federal Reserve (Fed) respondsto changes in the price of gold in the short run because it cannot afford to wait for alower frequency information concerning changes in the aggregate variable whosevalue it ultimately wishes to control. Thus, in the United States, the Fed is sup-posed to interpret a rising gold price as indicative of excess liquidity or inflation-ary expectations, in which case the Fed reduces bank reserves to stabilise the goldprice (Lastrapes – Selgin 1995). On the other hand, the reason for investors to in-clude gold in their portfolios is that gold is a precious metal with functions that acurrency has; among all the functions, the purchasing power parity is the most im-portant one. If it can be proved that gold could serve as an exchange rate hedge,then it will be more meaningful to add gold to the portfolio as a measure of assetvalue protection. In addition, if gold is a hedge against the exchange rate fluctua-tions, it has the ability to resist changes in the internal and external purchasingpower of the domestic currency (Wang – Lee 2011).

Many researchers have conducted studies to investigate the relationship be-tween gold prices, exchange rates and other macroeconomic variables, and foundmixed results. Dooley et al. (1995) search for the short- and long-run influences ofgold prices on exchange rates by means of multivariate vector autoregression and

Acta Oeconomica 64 (2014)

92 B. GÜRÉÓ – B. KIRAN

Page 3: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

cointegration modelling techniques. They test the hypothesis that exchange ratemovements reflect country-specific shocks by theorising that country-specificshocks should also be reflected systematically in the price of gold. Ghosh et al.(2002) investigate the contradiction between short-run and long-run movementsin the gold prices over the period 1976–1999 and find a long-run relationship be-tween the nominal gold price and the US retail price index. Twite (2002) studiesthe exposure of the stock prices of Australian gold mining firms to changes in goldprices and the valuation effects of the gold price exposure. Evidence from the be-haviour of stock price sensitivities suggests that gold mining firms can be repre-sented as a portfolio of gold assets and embedded real options. Capie et al. (2005)try to determine the extent to which gold acts as an exchange rate hedge, usingweekly data on the gold price, sterling dollar and yen dollar exchange rates. Ac-cording to their results, a negative, typically inelastic relationship is found be-tween gold and these exchange rates. Levin and Wright (2006) develop a theoreti-cal framework based on the simple economics of “supply and demand” that isconsistent with the view that gold is an inflation hedge in the long run. By usingcointegration and vector error correction model techniques over the period of1976–2005, they find important results. The first finding is that there is a long-runrelationship between the gold price and US price level. There is also a positive re-lationship among the changes of the gold price, the US inflation volatility and thecredit risk. Another finding of their paper is that there is a negative relationshipamong the gold price movements, the changes in the US dollar trade weighted ex-change rate and the gold lease rate. The final finding of the paper is that in the ma-jor gold consuming countries such as Turkey, India, Indonesia, Saudi Arabia andChina, gold acts effectively as a long-term hedge against inflation. Sjaastad(2008) examines the theoretical and empirical relationships between the major ex-change rates and the price of gold and finds that, since the dissolution of theBretton Woods international monetary system, floating exchange rates among themajor currencies have been a major source of price instability in the world goldmarket. Han et al. (2008) propose an interval method to explore the relationshipbetween the exchange rate of the Australian dollar against the US dollar and thegold price by using weekly, monthly and quarterly data. Their findings indicatethat the interval least squares estimates characterise well how the exchange raterelates to the gold price both in the long run and short run. Wang et al. (2010) ex-amine the relationship between stock exchange prices in Germany, Japan, Taiwanand China with oil prices, gold prices and various foreign exchange rates empiri-cally and determine a long term relationship between these variables. Wang – Lee(2011) investigate whether gold could be an exchange hedge in Japan over the1986–2007 period in the concept of nonlinear relationship between gold returnsand the exchange rate fluctuation of the Japanese yen. By using a threshold vector

Acta Oeconomica 64 (2014)

THE PRICE OF GOLD AND THE EXCHANGE RATE 93

Page 4: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

autoregressive model, they find that the effectiveness of gold as an exchange ratehedge depends on the depreciation rate of the yen.

This paper extends the current literature on the gold prices and exchange ratesfor Turkey. When looking at the Turkish economy, this relationship is very im-portant, especially after the economic stabilisation and reform program of Turkey,which included both financial and trade liberalisation. The main goal of the finan-cial and trade liberalisation program was to promote financial market develop-ment and to move towards a liberalised trade regime on the basis of export-ledgrowth strategy with a flexible exchange rate, respectively. International capitalmovements and foreign exchange operations were subject to tight controls before1980, they were entirely liberalised and the Turkish lira became convertible in1989. After the liberalisation of the financial sector, the Turkish economy wasable to provide enough foreign capital inflow, except during the Gulf crisis in1991 and the Russian crisis in 1998. In addition, the Turkish economy witnessedtwo financial crises in the near past. While the first crisis occurred in January1994, the second one occurred in February 2001. Both crises were followed byrapid reserve depreciations, a sharp and unexpected devaluation and a switch to anew exchange rate regime (Kasman – Ayhan 2008). At this point, it is importantto note that the price of gold increases or the purchasing power parity of gold goesup if the paper currency depreciates. Accordingly, one might prefer holding goldto protect against a loss in the purchasing power of the paper currency (Öztürk –Açékalén 2008).

In this paper, we explore the relationship between gold prices and the exchangerate of the US dollar vis-à-vis the Turkish liras (USD/TRY) between 1990 and2011 by considering the possibility of the nonlinearity and threshold effects in therelationship. Differing from the paper of Wang – Lee (2011), in our paper, the de-veloped threshold Granger causality test of Li (2006) is used in addition to thethreshold cointegration test of Hansen – Seo (2002).

The remainder of the paper is divided into 3 sections: Section 2 briefly de-scribes the threshold cointegration and threshold Granger causality methodology,Section 3 presents data and reports the empirical results. Finally, Section 4 con-cludes.

2. METHODOLOGY

In the literature, a commonly used approach to modelling the relationship be-tween gold prices and exchange rates is linear modelling. A linear vector errorcorrection (VEC) model assumes that the error correction mechanism is valid inevery time period. As a possibility of discrete adjustment, Balke – Fomby (1997)

Acta Oeconomica 64 (2014)

94 B. GÜRÉÓ – B. KIRAN

Page 5: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

combine nonlinearity and cointegration as the concept of threshold cointegration.By considering this possibility, in our analysis, we use the threshold cointegrationtest of Hansen – Seo (2002) and the threshold Granger causality test of Li (2006)in order to investigate the relationship between gold prices and exchange rates inTurkey. The brief descriptions of these tests are given below.

2.1. Threshold cointegration

In the concept of nonlinearity, a threshold VEC model takes the following form:

Db g

bx

A X u if w b

A X u if w bt

t t t

t t t

�� � �

� �� �

� �

1 1 1

2 1 1

( ) ( )

( ) ( ) �

� g

where

x

w

x

x

x

t

t

t

t

t

���

1

1

1

2

1

1

( )

( )

b

b

D

D

D

�������

and g is a threshold parameter, xt is a p-dimensional I(1) time series which iscointegrated with one p × 1 cointegrating vector b b b, ( )w xt t� � is the I(0) errorcorrection term, A1 and A2 are coefficient matrices, ut is an error term. A1 and A2

describe the dynamics in the regimes, depending on whether deviations from theequilibrium are above or below the threshold. Under the assumption that the errorterms are iid Gaussian, Hansen – Seo (2002) present an algorithm to estimate thethreshold VEC model by using the maximum likelihood estimation method. Test-ing for the presence of the threshold effect is an important stage of threshold VECmodel. Hansen – Seo (2002) suggest a test under the null hypothesis that there isno threshold, so the model reduces to a conventional linear VEC model. This teststatistic can be denoted as,

sup sup (~

, )LM LML U

� � �g g gb g

Acta Oeconomica 64 (2014)

THE PRICE OF GOLD AND THE EXCHANGE RATE 95

Page 6: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

where~b is the estimate of b. [gL, gU] is the search region set so that gL is the p0

percentile of ~w t�1 , and gU is the (1 – p0) percentile. Andrews (1993) suggests set-ting p0 between 0.05 and 0.15. In order to calculate the asymptotic critical valuesand p-values of the sup LM test, Hansen – Seo (2002) developed two bootstrapmethods.

2.2. Threshold Granger causality

Threshold model and Granger causality analysis have been combined by Li(2006) based on two regime threshold autoregressive distributed lagTADL(p,q,t,d) model which has the following model specification:

y a b y I c x It m mi t i mt mj t j mt

j

q

i

p

m

p

� � � ��

���� �

�����

111� � et

where I I yt t d1 � ��( )t and I I x x xt t t t kt2 1 11� � �, ( ,... , ) is a k × 1 vector at time t.

Li (2006) considers three null hypotheses given by,

H c c c cq q00

11 21 1 2� � � � �...

H c c q01

11 1� � �...

H c c q02

21 2� � �...

where H 00 implies that none of the covariates has predictive content in the two re-

gimes and H i

0 implies no predictive content in regime i, i = 1,2. According to Li

(2006), all hypotheses are tested based on the Wald statistic, written as:

� � � �W r R z z e z z z z R Rt t t t t t t� � � � � �� �� � �(

~) [ � ( ) ] (

~)u u

1 2 1 1

where R is the selection matrix for the null hypotheses, q is parameter estimates,z ft � � �( ) /u u , f E y t t� �( / )V 1 and �et is the OLS or nonlinear least squares(NLS) residuals. Li (2006) also shows thatW m~ ( )x

2 , where m is the number of

restrictions and standard asymptotic results are applicable.

Acta Oeconomica 64 (2014)

96 B. GÜRÉÓ – B. KIRAN

Page 7: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

3. DATA AND EMPIRICAL RESULTS

This paper uses monthly observations on gold prices per ounce (GOLD) quoted inTurkish liras and the USD/TRY rate over the period from January 1990 throughMarch 2011, as published by the Central Bank of the Republic of Turkey. Beforethe analysis, we convert the data into the natural logarithmic form and show theplots of the series (LNGOLD and LNER) in Figure 1.

As can be seen from the figure, the LNGOLD and LNER series have a verysimilar upward trend, which suggests that they may be cointegrated around a com-mon trend. Since the plots of the series are only suggestive of the existing behav-iour, we now turn to a formal analysis. As a first step, stationarity properties of theseries are investigated by using Augmented Dickey Fuller (ADF), Phillips–Perron (PP), Kwiatkowski–Phillips–Schmidt–Shin (KPSS) and Ng–Perron unitroot tests. It is important to note that the KPSS test differs from other unit roottests in that the series are assumed to be stationary under the null. The reason forusing the Ng–Perron unit root test additionally is that the test has some advan-tages: it has good size and power. It is considered to be more powerful because Ngand Perron use a generalised least square detrended procedure to develop four teststatistics1 for overcoming the problems of power distortions and poor size related

Acta Oeconomica 64 (2014)

THE PRICE OF GOLD AND THE EXCHANGE RATE 97

Figure 1. Plots of the LNGOLD and LNER series

1 Ng – Perron (2001) constructed four test statistics. These test statistics are modified forms ofPhilips and Perron Za AND Zt statistics, the Bhargava (1986) R1 statistic and the Elliot –Rotherberg – Stock (1996) point optimal statistics (see details in Ng – Perron 2001).

Page 8: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

to ADF. It is also particularly suitable for a small sample. The null hypothesis forthe MSB and MPT statistics of the Ng–Perron test is the same as the null hypothe-sis of the KPSS test. The results of the unit root tests can be seen in Table 1.

According to the results of Table 1, it can be seen that the series are non-stationary in level, but stationary after first differencing. In other words, they areintegrated of order one (I(1)). It is well known that nonstationary variables maypossess long-run equilibrium relationships and have a tendency to move togetherin the long run. In the light of this information, in the next step, we conduct athreshold VEC model. An important stage of this modelling is the selection of anappropriate lag length. In our analysis, the appropriate lag length is chosen as 1 ac-cording to Akaike information criteria (AIC) and Bayesian information criteria(BIC). By using this finding, the validity of the threshold cointegration versus lin-ear cointegration is investigated based on the Sup LM test proposed by Hansen –Seo (2002) and the results are presented in Table 2.

Acta Oeconomica 64 (2014)

98 B. GÜRÉÓ – B. KIRAN

Table 1

Unit root test results

ADF PP KPSS

LNGOLD –0.512 –0.598 0.493*LNER 0.037 0.184 0.507*DLNGOLD –14.271* –14.244* 0.078DLNER –10.028* –10.035* 0.114

Ng–Perron

MZ MZt MSB MPT

LNGOLD –0.168 –0.109 0.651* 89.585*LNER 0.388 0.275 0.710* 111.921*DLNGOLD –159.354* –8.926* 0.056 0.572DLNER –97.179* –6.960* 0.072 0.977

Note: * indicates rejection of the null hypothesis at the 1% significance level.

Table 2

Sup LM test results

Test statistic 20.196 p-value: 0.036

Critical values 1% 22.9755% 19.262

10% 17.602

Note: Bootstrap p-value is calculated from 5,000 replications.

Page 9: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

The results clearly indicate that the Sup LM test statistic is greater than the crit-ical values at the 5% significance level, implying the presence of a thresholdcointegration relationship between gold prices and exchange rate. It is well knownthat systems in which variables are cointegrated can be characterised by an errorcorrection model, which describes how the variables respond to deviations fromthe equilibrium. By using this knowledge, we conduct a threshold VEC model andpresent the results in Table 3.

It is seen from the table that the estimated value of the threshold parameterequals to –3.268. Based on this value, the threshold vector error correction modelis divided into two regimes. The first regime occurs when deviation from thelong-run equilibrium is below the threshold value and because this regime in-cludes 20.8% of observations, it can be called “unusual regime”. On the contrary,when the first lagged value of the error correction term is above the thresholdvalue, the second regime occurs. Because this regime includes 79.2% of observa-tions, it can be called “typical regime”. According to the results in the table, errorcorrection terms are found statistically significant. This implies that both goldprices and exchange rates respond to deviations from the long-run equilibrium.After finding the long-run relationship, in the next step of the analysis, we investi-gate the causal relationship between gold prices and exchange rates by using thethreshold Granger causality test proposed by Li (2006). The results are tabulatedin Table 4, indicating that there is a bi-directional causal relationship betweengold prices and exchange rate in all cases, except the second regime of exchangerate equation. In other words, gold prices and exchange rates are jointly deter-mined and affected, except when the threshold parameter exceeds the value of–3.268 in case the exchange rate is the dependent variable. In the light of these re-

Acta Oeconomica 64 (2014)

THE PRICE OF GOLD AND THE EXCHANGE RATE 99

Table 3

Estimation results of threshold VEC model

Variables DLNGOLD DLNER

wt � �1 g wt � �1 g wt � �1 g wt � �1 g

wt � 1 0.012*** –0.008* 0.031* –0.006**Constant 0.060* 0.007* 0.166* 0.026*DLNGOLDt–1 0.836* 0.121 0.799* –0.038DLNERt–1 0.391*** 0.332* 0.184 –0.026

g –3.268% of observations 20.8 79.2

Note: *, ** and *** indicate statistical significance at the 1%, 5% and 10% significance levels,respectively.

Page 10: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

sults, it can be said that gold price can be used as a hedge against the exchangerate, partially. But it is clear that the relationship between gold prices and ex-change rates disappears when the threshold parameter exceeds the thresholdvalue. According to this result, it can be concluded that gold price is only a weakhedge against the American dollars/Turkish liras exchange rate.

4. CONCLUSION

This paper investigated the relationship between gold prices quoted in Turkish li-ras and the USD/TRY exchange rate between 1990 and 2011 by extending thecurrent literature in the concept of nonlinearity and threshold effects. For this pur-pose, we use the recently developed threshold cointegration test of Hansen – Seo(2002) and the threshold Granger causality test of Li (2006). As a first step, theunit root properties of the series are tested by using ADF, PP, KPSS andNg–Perron tests. After finding that they are nonstationary in level, the long-runrelationship between these variables is considered. The validity of the thresholdcointegration versus linear cointegration is investigated by using the Sup LM testdeveloped by Hansen – Seo (2002). According to the results, it is found that thereis a threshold cointegration relationship between gold prices and exchange rate.The estimation of the threshold VEC model gives that the threshold parameterequals to –3.268. In order to investigate the causal relationship between goldprices and exchange rate, we also apply the threshold Granger causality test of Li(2006) and find that there is evidence of a bi-directional relationship between goldprices and exchange rate, except for the second regime of exchange rate equation,implying that gold price can be used as a hedge against the exchange rate, par-tially. Since this relationship disappears above the threshold value, it can be con-cluded that gold price is only a weak hedge against the American dollars/Turkishliras exchange rate.

Acta Oeconomica 64 (2014)

100 B. GÜRÉÓ – B. KIRAN

Table 4

Granger causality test results

H00 H0

1 H02

LNGOLD – LNER 61.154* 11.752* 49.402*LNER – LNGOLD 7.749* 7.565* 0.183

Note: * indicates statistical significance at the 1% significance level.

Page 11: The price of gold and the exchange rate: Evidence from threshold cointegration and threshold granger causality analyses for Turkey

REFERENCES

Andrews, D.W.K. (1993): Test for Parameter Instability and Structural Change with UnknownChange Point. Econometrica, 61(4): 821–856.

Balke, N.S. – Fomby, T.B. (1997): Threshold Cointegration. International Economic Review, 38:627–645.

Bhargava, A. (1986): On the Theory of Testing for Unit Roots in Observed Time Series. Review of

Economic Studies, 53(3): 369–384.Capie, F. – Mills, T.C. – Wood, G. (2005): Gold as a Hedge against the Dollar. Journal of Interna-

tional Financial Markets, Institutions and Money, 15(4): 343–352.Dooley, M.P. – Isard, P. – Taylor, M.P. (1995): Exchange Rates, Country-specific Shocks and Gold.

Applied Financial Economics, 5: 121–129.Elliot, G. – Rothenberg, T.J. – Stock, J.H. (1996): Efficient Tests for an Autoregressive Unit Root.

Econometrica, 64(4): 813–836.Forbes, M.S. Jr. (1991): Is Gold Running the Fed? Forbes, October 14, 25.Ghosh, D. – Levin, E.J. – Macmillan, P. – Wright, R.E. (2002): Gold as an Inflation Hedge? Discus-

sion Paper Series, No. 0021, Department of Economics, University of St. Andrews.Han, A. – Xu, S. – Wang, S. (2008): Australian Dollars Exchange Rate and Gold Prices: An Interval

Method Analysis. Proceedings of The 7th International Symposium on Operations Research

and Its Applications, pp. 46–52.Hansen, B.E. – Seo, B. (2002): Testing Two Regime Threshold Cointegration in Vector Error Cor-

rection Models. Journal of Econometrics, 110: 293–318.Jones, J.H. (1933): The Gold Standard. The Economic Journal. Blackwell Publishing for the Royal

Economic Society, 43: 551–574.Karunagaran, A. (2011): Recent Global Crisis and the Demand for Gold by Central Banks: An Ana-

lytical Perspective. Working Paper Series, 14. Reserve Bank of India.Kasman, A. – Ayhan, D. (2008): Foreign Exchange Reserves and Exchange Rates in Turkey: Struc-

tural Breaks, Unit Roots and Cointegration. Economic Modelling, 25: 83–92.Lastrapes, W.D. – Selgin, G. (1995): Gold Price Targeting by the FED. Working Paper. University

of Georgia.Levin, E.R. – Wright, R.E. (2006): Short-run and Long-run Determinants of the Price of Gold.

World Gold Council Research Study, No. 32. London.Li, J. (2006): Testing Granger Causality in Presence of Threshold Effects. International Journal of

Forecasting, 22: 771–780.Ng, S. – Perron, P. (2001): Lag Length Selection and the Construction of Unit Root Test with Good

Size and Power. Econometrica, 69(6): 1519–1554.Öztürk, F. – Açékalén, S. (2008): Is Gold a Hedge against Turkish Lira? South East European Jour-

nal of Economics and Business, 3: 35–40.Sachs, J.D. – Larrain, F.B. (1993): Macroeconomics in the Global Economy. Englewood Cliffs, NJ:

Prentice Hall.Sjaastad, L.A. (2008): The Price of Gold and the Exchange Rates: Once Again. Resources Policy,

33(2): 118–124.Twite, G. (2002): Gold Prices, Exchange Rates, Gold Stocks and the Gold Premium. Australian

Journal of Management, 27: 123–140.Wang, K.M. – Lee, Y.M. (2011): The Yen for Gold. Resources Policy, 36(1): 39–48.Wang, M.L. – Wang, C.P. – Huang, T.Y. (2010): Relationships among Oil Price, Gold Price, Ex-

change Rate and International Stock Markets. International Research Journal of Finance and

Economics, 47(September): 80–89.

Acta Oeconomica 64 (2014)

THE PRICE OF GOLD AND THE EXCHANGE RATE 101