Gold and Oil have always had a central role within the international economy, and meet the interests of many investors, and in particular, speculators. The Euro introduction (1999) has added the Euro-Dollar exchange rate as a further main variable that the operators, investing on these commodities, have to consider when implementing their strategies. This paper analyzes the mutual relationship between commodities prices (gold and oil) and the Euro/Dollar exchange rate, within the time frame from 2004 to 2014, so to find which specific variable can give significant information on the expected variation of other variables and on which time horizon. This can support the of investors’ choices on taking more effective speculative positions. Results obtained by means of a VAR model show some significant statistical relationship between the three variables on the short term ( i.e. when considering daily data), but also some possible relationship on a longer term (monthly data), suggesting that oil prices can give significant information on the expected value of the Euro/Dollar exchange rate.
Gold and oil prices have always been two reference values for the international economy. Over the years their prices have been highly volatile. Consequently, gold and oil meet the interests of investors, but above all speculators.
From a theoretical point of view, gold price should increase during economic crises, and decrease in positive financial context. The opposite is expected for oil prices, as they are mainly linked to the industrial activity, thus positively correlated with the business cycle. The introduction of the euro in 1999 added the euro/dollar exchange rate as a further reference variable, and this induced the operators and speculators on international markets to include this variable in their models and in the implementation of their strategies.
The aim of this paper is to test for the existence of a long or short-term relationship between the prices of gold, oil and the euro/dollar exchange rate, so to give significant information for making profitable investments in commodities, simply observing the dynamics of the euro/dollar exchange or, conversely, taking positions on currencies, considering the price dynamics of the two commodities.
In this aim, the graphical and short-termed analysis can, in some cases, suggest a strong relationship. The graphs in
Actually, this evidence is misleading, because even if the three series have a similar ongoing only on the considered interval, the relationships are more complex to evaluate.
In fact, during the considered time period, the price of gold increased by 41.22%; Oil had an 80% price increase and the Euro/Dollar exchange rate increased by 14.87%.
The prices increase of the three variables is due to the turmoil in financial markets resulting from the subprime mortgage1 crisis, that has further weakened the dollar, pulled its value in 2008 to around 1.50 euro.
The Federal Reserve interventions, for limiting the huge deficit of the economy, implemented an expansionary monetary policy for sustaining the exports by means of a weaker currency vs. euro.
The oil price increase was due to two main reasons:
1) The dollar weakness. Demand stayed strong, despite the producing countries in-creased their oil production to compensate the lower profit due to the dollar weakness.
2) The international stock markets crisis. All commodities seemed to be a suitable in-vestment for speculative funds.
The US dollar weakness led the investors to take “long” positions on gold, pushing up its price. In other words, in this period the demand for gold as a refuge was particularly high. A second sub-period (April 2014-December 2014) (see
Specifically, between April 2014 and December 2014, the gold price fell by 7.59%, oil 41.89% and the exchange rate decreased by 7.53%.
The main elements linked to the decrease of the three goods prices are:
・ Regarding gold: gold price conveyed when Fed started the “tapering” of the QE. Moreover, the strong expectation of rising interest rates has recommended maintaining long positions in gold.
・ Regarding oil: the oversupply of the Arabian and the US (shale oil) countries, and a weak demand from Asian markets, led to a significant decline in oil prices, which fell from $102 in July 2014 to $59 in December 2014.
・ Regarding the exchange rate: the ECB started its QE program, that is still ongoing. The resulting lower or even negative interest rates led to a contraction in the euro from 1.35$ in July to 1.23$ in December.
These three short termed observations can suggest a strong correlation between the three variables. But the observation of the same variables on a longer term, clearly contradicts this hypothesis (see
It is, thus, fundamental to perform a quantitative analysis, based on appropriate methodologies, for finding the actual linkages between the dynamics of the
three considered variables.
The analysis here presented is developed by means of a VAR model on data referring to the time span from January 2004 to December 2014.
The reminder of the paper is structured as it follows: section 2 reports the literature review, section 3 and 4 report the econometric model and its results, section 5 discusses the estimation results and its conclusions.
This paper mainly refers to a literature stream that, through the analysis of prices of gold, oil and exchange rates, investigates over time the relationships between the variations of the considered variables. These studies mainly aim at verifying if the theoretical references on the expected relationships are actually met by market values. These findings always bring useful information for the investors and speculators operating in the foreign exchange market (Forex) and in the commodities market.
Indeed, no one of these studies includes the euro/dollar exchange rate as reference variable.
In the recent literature, the main references are the following:
・ [
・ [
・ [
・ [
・ [
・ [
These results are not univocal, and this is possibly due to the different modeling, and to the different datasets.
Nevertheless, the main results show significant linkages between the two commodities and the exchange rates on the US dollar on the short term (days), often not confirmed on longer terms (months).
The cited references also show that the euro/dollar exchange rate has not been considered in terms of its linkages to the main commodities. This can be due to the recent (1999) introduction of the European currency in financial markets, and to the small incidence on the European GDP of the gold and oil production, so that no significant theoretical effects of the two commodities prices variations are expected to induce any variations in the exchange rate, thus inducing no interest in testing these effects.
It has, instead, a significant interest for speculators, which are interested in having all the possible information on potential mutual relationship between the main commodities and exchange rates trends, so to implement more effective speculation strategies.
The main aim of this study is to test for possible linkages, on a short (days) and longer (months) term, of gold and oil prices and the euro/dollar exchange rate, and, in case, to have some references on which variable can help in having more precise expectations on the other variables variations.
The analysis is performed on the three variables time series, starting from daily values from January 2004 to December 2014.
The analysis is based on a vector autoregressive model, using data coming from Bloomberg, one of the main financial data providers, which include the daily closing prices of gold (dollars per ounce), oil WTI (dollars per barrel) and the euro/dollar exchange rate, and refer to the time span from January 2004 to December 2014. Overall the dataset is made up of 2870 daily observations for each variable.
The VAR models are based on time series, with multiple dynamic equations in which each variable is considered in its relation to the same and other lagged variables.
It is, thus, possible to summarize the dynamic relationships between the considered variables, with no distinction between endogenous and exogenous (or dependent and independent) variables, but the attention is on current (at time t) values, as a function of the previous (at time t-h) values, and the parameters are estimated so to have unbiased and consistent estimations.
As the VAR modeling requests the time series to be stationary, and the Dickey-Fuller test on oil and gold prices, as like on the euro/dollar exchange rate, estimated to be not stationary, the model is applied on the first differences, which resulted to be stationary as requested.
The optimal lag was then chosen on the base of 5 different indexes (LR, FPE, AIC, HQIC, SBIC, see
As it can be seen on
Lag | LL | LR | df | p | FPE | AIC | HQIC | SBIC |
---|---|---|---|---|---|---|---|---|
0 | −658.135 | 85.6099 | 12.9634 | 12.9947* | 13.0406* | |||
1 | −644.538 | 27.194* | 9 | 0.001 | 78.2397* | 12.8733* | 12.9983 | 13.1821 |
2 | −641.087 | 6.9021 | 9 | 0.647 | 87.2789 | 12.9821 | 13.2009 | 13.5225 |
3 | −635.277 | 11.619 | 9 | 0.236 | 93.0292 | 13.0446 | 13.3573 | 13.8167 |
4 | −631.979 | 6.5957 | 9 | 0.679 | 104.272 | 13.1565 | 13.5629 | 14.1601 |
Source: data processing on Stata output.
Lag | LL | LR | df | p | FPE | AIC | HQIC | SBIC |
---|---|---|---|---|---|---|---|---|
0 | −1108.06 | 0.021805 | 4.68803 | 4.69839* | 4.71436* | |||
1 | −102.69 | 10.735 | 9 | 0.294 | 0.022142 | 4.70335 | 4.74479 | 4.8087 |
2 | −1089.86 | 25.676* | 9 | 0.002 | 0.21787* | 4.68716* | 4.75967 | 4.87152 |
3 | −1082.22 | 15.283 | 9 | 0.083 | 0.021912 | 4.69289 | 4.79647 | 4.95626 |
4 | −1075.42 | 13.591 | 9 | 0.138 | 0.022117 | 4.7022 | 4.83685 | 5.04457 |
Source: data processing on Stata output.
Δ _Gold | Coeff. | Std. Err. | Z | P > |z| |
---|---|---|---|---|
Δ_Gold t-1 | 0.1162511 | 0.0921111 | 1.26 | 0.207 |
Δ_Oil t-1 | 0.2640494 | 0.6922552 | 0.38 | 0.703 |
Δ_EUR/USD t-1 | 39.96504 | 146.1948 | 0.27 | 0.785 |
constant | 4.116153 | 3.998092 | 1.03 | 0.301 |
Δ_Oil | Coeff. | Std. Err. | Z | P > |z| |
Δ_Gold t-1 | −0.011622 | 0.0127144 | −0.91 | 0.361 |
Δ_Oil t-1 | 0.3724225 | 0.0955545 | 3.90 | 0.000 |
Δ_EUR/USD t-1 | 26.02258 | 20.1798 | 1.29 | 0.197 |
constant | −0.0337552 | 0.5518713 | −0.06 | 0.951 |
Δ_EUR/USD | Coeff. | Std. Err. | Z | P > |z| |
Δ_Gold t-1 | −0.0000425 | 0.0000639 | −0.67 | 0.506 |
Δ_Oil t-1 | 0.0008917 | 0.00048 | 1.86 | 0.063 |
Δ_EUR/USD t-1 | 0.2080736 | 0.1013602 | 2.05 | 0.040 |
constant | −0.0010629 | 0.002772 | −0.38 | 0.701 |
Source: data processing on Stata output.
The Granger causality cannot be performed on these values, as the model is only estimated on lag 1.
As evident from
The estimations based on daily data show that gold prices are significantly affected by the oil prices with lag 1 and 2 (respectively with positive and negative correlation), and with the euro/dollar exchange rate with lag 1 (negative correlation).
Oil prices are instead linked to gold prices (lag 2) and the exchange rate results to be linked to gold (lag 2) and oil prices (lag 1) with a 5% significance level.
The Granger causality test, reported on
− Causality for gold prices as due to oil prices cannot be refused at 5% significance level;
− Causality for oil prices as induced by gold prices cannot be refused at 1% significance level;
− Causality for exchange rates as induced by gold and oil prices cannot be refused at, respectively, 10% and 5% significance level.
Δ_Gold | Coeff. | Std. Err. | Z | P > |z| |
---|---|---|---|---|
Δ_Gold t-1 | 0.0005361 | 0.0288984 | 0.02 | 0.985 |
Δ_Gold t-2 | 0.0353941 | 0.0288035 | 1.23 | 0.219 |
Δ_Oil t-1 | 0.3654271 | 0.2264416 | 1.61 | 0.107 |
Δ_Oil t-2 | −0.4051922 | 0.2153496 | −1.88 | 0.060 |
Δ_EUR/USD t-1 | −87.36052 | 48.3633 | −1.81 | 0.071 |
Δ_EUR/USD t-2 | −8.450506 | 47.97282 | −0.18 | 0.860 |
constant | 0.1703217 | 0.3435947 | 0.50 | 0.620 |
Δ_Oil | Coeff. | Std. Err. | Z | P > |z| |
Δ_Gold t-1 | −0.0003147 | 0.0034184 | −0.09 | 0.927 |
Δ_Gold t-2 | 0.0133602 | 0.0034072 | 3.92 | 0.000 |
Δ_Oil t-1 | −0.0214938 | 0.0267858 | −0.80 | 0.422 |
Δ_Oil t-2 | 0.0263361 | 0.0254738 | −1.03 | 0.301 |
Δ_EUR/USD t-1 | −1.141769 | 5.720905 | −0.20 | 0.842 |
Δ_EUR/USD t-2 | −0.7798057 | 5.674715 | −0.14 | 0.891 |
constant | 0.0427803 | 0.0406439 | 1.05 | 0.293 |
Δ_EUR/USD | Coeff. | Std. Err. | Z | P > |z| |
Δ_Gold t-1 | 0.0000201 | 0.0000177 | 1.14 | 0.256 |
Δ_Gold t-2 | 0.0000395 | 0.0000177 | 2.23 | 0.026 |
Δ_Oil t-1 | 0.000306 | 0.000139 | 2.20 | 0.028 |
Δ_Oil t-2 | 0.0000949 | 0.0001322 | 0.72 | 0.473 |
Δ_EUR/USD t-1 | −0.0106689 | 0.0296957 | −0.36 | 0.719 |
Δ_EUR/USD t-2 | −0.0247591 | 0.0294559 | −0.84 | 0.401 |
constant | 0.0001002 | 0.000211 | 0.48 | 0.635 |
Source: data processing on Stata output.
Equation | Excluded | chi2 | df | Prob> chi2 |
---|---|---|---|---|
ΔGold | ΔOil | 6.5857 | 2 | 0.037 |
ΔGold | ΔEUR/USD | 3.3154 | 2 | 0.191 |
ΔGold | ALL | 9.4601 | 4 | 0.051 |
ΔOil | ΔGold | 15.394 | 2 | 0.000 |
ΔOil | ΔEUR/USD | 0.06038 | 2 | 0.970 |
ΔOil | ALL | 16.233 | 4 | 0.003 |
ΔEUR/USD | ΔOil | 6.2113 | 2 | 0.045 |
ΔEUR/USD | ΔGold | 5.1681 | 2 | 0.075 |
ΔEUR/USD | ALL | 13.292 | 4 | 0.010 |
Source: data processing on Stata output.
For daily data, as the white noise test on residuals (reported in
The following
Variable | Portmanteau (Q) statistic | Prob. >chi2(40) |
---|---|---|
Gold | 34.5941 | 0.7117 |
Oil | 53.1062 | 0.0803 |
EUR/USD | 31.4806 | 0.8300 |
Source: data processing on Stata output.
The analysis of gold and oil prices and euro/dollar exchange rate performed on daily data from January 2004 to December 2014 shows some significant information.
The VAR estimations, like in [
The impulse-response functions show that only small variations are expected as a consequence of the price shocks, and that the euro/dollar exchange rate gives no significant in-formation on oil and gold expected trends. As regarding the monthly dataset, the euro/dollar exchange rate does not allow to take a financial position (long or short) on the examined commodities. Instead, the oil prices can give significant information on the expectances of the exchange rate, and suggest possibly effective speculative positions on the currency markets.
Overall, the analysis shows that the linkage between gold price and the exchange rate is significant only in the short term. For this reason investment strategies based on this link, should be implemented for a short time period (few days). These findings bring new information both from the academic perspective, aimed at understanding the actual linkages between the main currencies and commodities, and from the practical perspective of financial speculators, aimed at developing more effective investment strategies, which can have a technical advantage when using the information coming from these findings.
Patanè, M., Tedesco, M. and Zedda, S. (2017) Dynamic Relationship of Commodities Prices and EUR/USD Exchange Rate Trends in the Recent Past. Modern Economy, 8, 995-1004. https://doi.org/10.4236/me.2017.88069