Resources Policy 64 (2019) 101478

Contents lists available at ScienceDirect

Resources Policy
journal homepage: www.elsevier.com/locate/resourpol

The analysis of factors affecting global gold price T

a b a,∗
Yao Qian , Dan A. Ralescu , Bo Zhang
a
Department of Statistics, Renmin University of China, China
b
Department of Mathematical Sciences, University of Cincinnati, USA

ARTICLE INFO ABSTRACT

Keywords: As one of the most widely purchased investment products for global investors, the price of gold is increasingly
RSM attracting attention. This paper analyzes and summarizes the various factors that affect global gold price. A
Dollar index reverse process of response surface methodology(RSM) is first used to evaluate the impact of six different factors
Federal funds rate (namely, the dollar index, the federal funds rate, CPI, exchange rate, oil price and S&P500) on the gold price. The
CPI
results show that all factors have a negative impact on the price of gold except CPI. Furthermore, the impact of
Exchange rate
Oil price
CPI and Oil price on response variable is not significant at the 5% significance level, which also makes the
interaction effect involving these two factors more complicated.

1. Introduction rities and futures have increased rapidly. However, the price of gold
was showing an upward trend at the same time. The price of gold
Gold serves several functions in the world economy, and its link changed from $672 per ounce in August 2007 to $933 per ounce in
with financial and macroeconomic variables is well established March 2008. After 2012, the price of gold skyrocketed to $1700 per
(Pierdzioch et al., 2014a, b). It has monetary value and is sought after ounce due to the loose monetary policy adopted by many developed
by central banks to be part of their international reserves(Gupta et al., countries in order to stimulate the economy. In April 2013, the price of
2014). It has industrial uses and can be transformed into jewellery. international gold fell to $1481 per ounce and suffered the biggest
Gold has been traditionally used by investors as a hedge in portfolio weekly decline during December. After that, the gold price basically
diversification and a safe haven in times of extreme economic and went downward to $1059 per ounce by the end of November 2015. And
political turbulence and severe market turmoil (Baur and Lucey, 2010; then it continued to rise to $1350 per ounce in July 2016. Fig. 1 de-
Baur and McDermott, 2010; Lau et al., 2017; O'Connor et al., 2015). scribes the trend of gold price from 1990 to 2018, of which the data are
Gold is also a special commodity that has financial and monetary obtained from the WORLD GOLD COUNCIL. The large fluctuations in
functions. Although the monetary function of gold has been weakened the price of gold have caused widespread concern in the market. And
since the collapse of the Bretton Woods System in February 1973, the the current market has a wide divergence in the trend of future gold
financial asset function of gold has continued its importance (Wang prices. In the new economic situation, why is the price of gold changing
et al., 2016). Gold has also other distinguished characteristics. Its in this way? Motivated by this question, we first explore and select
supply is accumulated over the years, and its global annual physical representative influencing factors considering the attributes of gold,
production can be as small as 2% of total supply. So in contrast to other and then use the measurement method to empirically analyze these
commodities, its annual production may not sway its price as other factors.
factors do. Given the significance of gold in the modern world, the A myriad of different global variables affect the price of gold
ability to analyze the price of gold will be of utmost importance. (Pierdzioch et al., 2014a), which include the US dollar index, the in-
The global financial crisis that broke out at the end of 2007 has terest rate, the inflation rate, the exchange rate, the oil price and the
caused global capital markets to suffer, and the risks of various secu- stock market. The relationship between gold prices and major in-

∗
Corresponding author. 1036, Mingde Main Building, Renmin University of China, Haidian District, Beijing, China.
E-mail addresses: [email protected] (Y. Qian), [email protected] (D.A. Ralescu), [email protected] (B. Zhang).

https://doi.org/10.1016/j.resourpol.2019.101478
Received 12 June 2019; Received in revised form 21 August 2019; Accepted 22 August 2019
0301-4207/ © 2019 Elsevier Ltd. All rights reserved.
Y. Qian, et al. Resources Policy 64 (2019) 101478

Fig. 1. Global gold price trend from 1990 to 2018. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this
article.)

dividual variables is not stable and is time-varying. To the best of our power for detecting explosive behavior to approximate gold's funda-
knowledge, this is the first study to analyze the effects of six different mental value. Their studies demonstrate that inflation in a general
factors on gold price fluctuations based on response surface method commodity price index and gold ETF demand have a potential to ex-
(RSM) (Li et al., 2015; Ren et al., 2016; Bartley et al., 2016), which uses plain the price trajectory. Baruník et al. (2016) conduct a time-fre-
a reasonable experimental design method and obtains certain data quency analysis of dynamic correlations between pairs of key traded
through experiments. The multivariate quadratic regression equation is assets (gold, oil, and stocks) covering the period from 1987 to 2012 by
used to fit the functional relationship between the factors and the re- employing a wavelet approach. Their results show that heterogeneity is
sponse values. It is a statistical method for multivariate problems. a dominant feature in correlations across a number of investment
What is innovative is that we view our observations of gold price horizons between pairs of assets during times of economic downturn
and the value of factors that may influence the price of gold as the and financial turbulence for all three pairs of the assets under research.
results of multiple experiments. We restore the experimental design Gil-Alana et al. (2017) employ some recently developed techniques in
(Box–Behnken design,BBD; see e.g. Nam et al., 2018; Tak et al., 2015; time series analysis to deal with the relationship between oil prices and
Bakhtiari et al., 2016) from the experimental results, and then get the gold prices based on the concepts of fractional integration and coin-
quadratic regression relationship between the dependent variable and tegration. Their studies show that there is a fractionally cointegrated
the independent variable, the specific steps of which can be seen in relationship between the two variables, with an order of integration in
section 4. The remainder of the paper is organized as follows. Section 2 the long run relationship of about 0.46.
presents a brief review of the literature. Section 3 discusses the factors The impacts of foreign exchange rate and other factors on prices of
that influence the global gold price. Section 4 describes methodology gold have been also studied in the literature. For example, Aye et al.
and empirical results. Section 5 concludes the paper. (2015) construct several models to examine possible predictors of the
return of gold, which includes six global factors(business cycle, nom-
2. Literature review inal, interest rate, commodity, exchange rate and stock price) and two
uncertainty and stress indices (the Kansas City Fed's financial stres-
The literature on analyzing effects of different factors on the price of sindex and the U.S. economic policy uncertainty index). They find that
gold is not as generous as it is in analyzing gold return and volatility, all the predictors show strong predictive power at one time or another
and the methods employed are not as sophisticated either. O'Connor though at varying magnitudes, while the exchange rate factor and the
et al. (2015) provide a good review of relevant literature. Ewing and Kansas City Fed's financial stress index appear to be strong at almost all
Malik (2013) examine the volatility of gold and oil futures in- horizons and sub-periods. Beckmann et al. (2015) give a new per-
corporating structural breaks by employing univariate and bivariate spective on the link between gold prices and exchange rates. They
GARCH models. They provide strong evidence of significant transmis- conclude that exchange rate depreciations initially have a negative
sion of volatility between gold and oil returns under structural breaks. impact on the gold price after one day which turns out to be positive
Their findings support the idea of cross-market hedging and sharing of after two days in most of the cases. Besides, their results point to a
common information by financial market participants. Białkowski et al. specific role of the dollar in the context of gold-exchange rate re-
(2015) use several econometric models and apply a Markov regime lationships: volatility of exchange rates more frequently results in
switching Augmented Dickey–Fuller (ADF) test which has substantial strong hedging functions of gold prices. What is more, the gold price

2
Y. Qian, et al. Resources Policy 64 (2019) 101478

denominated in the US dollar tends to increase after a depreciation of

the dollar. Kang et al. (2017) employ the multivariate DECO-GARCH
model and the spillover index to investigate spillover effects among six
commodity futures markets – gold, silver, West Texas Intermediate
crude oil, corn, wheat, and rice. They conclude that gold and silver
apparently serve as sources of information transmission among the six
commodity futures markets, and investors demonstrate the flight-to-
quality phenomenon during financial crises. Besides, both gold and
silver are net information transmitters to the other commodity markets,
while the remaining four commodity futures (i.e. WTI, corn, wheat, and
rice) are net receivers of spillovers during the recent financial crises.
In view of the great attraction of gold to investors over periods of
high economic and political uncertainty, there exists an emerging in-
terest in studying the role of risk perceptions and uncertainty measures
on gold prices in the financial economics literature. Jones and Sackley
(2016) find that gold prices are positively related to EPU by in-
corporating the U.S. and European EPU indexes into a gold-pricing
model, of which the results show that the safe haven status of gold
induces an increase of its price in times of high uncertainty. Li and Fig. 2. Gold price and dollar index trend from January 1979 to November
Lucey (2017) examine time varying safe haven properties versus 2018. (For interpretation of the references to colour in this figure legend, the
equities and bonds of four precious metals (gold, silver, platinum and reader is referred to the Web version of this article.)
palladium) across eleven countries. They find that Economic Policy
Uncertainty(EPU) is a positive determinant of gold being an investment
safe haven. Bilgin et al. (2018) use the nonlinear Autoregressive-dis- theoretical level or the observed data.
tributed Lag (ARDL) model to analyze the asymmetric effect of four Interest rate is one of the tools for the state's macroeconomic reg-
uncertainty measures (namely, the volatility (VIX), skewness (SKEW), ulation and control. It is possible to adjust the supply and demand of
global economic policy uncertainty (EPU), and partisan conflict (PC) money through the adjustment of interest rate, thereby affecting the
indexes)on gold prices. The results show that rising economic policy macroeconomics. Interest rates can reflect the opportunity cost of
uncertainty contributes to increases in the price of gold. By contrast, holding gold to a certain extent. In general, gold prices and interest
gold prices are less likely to fall when economic policy conditions are rates are negatively correlated for it means that the return on monetary
improved. assets is higher when interest rates rise, so the demand for gold is
dropping. Conversely, as the interest rates fall, the attractiveness of
3. Factors influencing the global gold price gold will go up, and the price of gold will naturally rise. Apart from this,
the results of Erb and Harvey (2013) show that the correlation coeffi-
3.1. Data cient between US gold price and real interest rate in 1997–2012 is
−0.9. This article will choose the US federal funds rate to represent
The monthly data for the price of gold and the demand and supply interest rates since the United States is the global financial center and
of gold are obtained from the WORLD GOLD COUNCIL. The dollar the dollar has an incomparable superiority in the international mone-
index、the US federal funds rate are obtained from the Board of tary system. Fig. 3 is the line chart of the gold price and the federal
Governors of the Federal Reserve System. The monthly data for CPI are funds rates from December 1978 to January 2019. It can be inferred
acquired from the BUREAU OF LABOR STATISTICS. The original that there is indeed a strong negative correlation between them.
monthly exchange rate is also used in this study, and the data are ob-
tained from the Wind Database. The monthly data for the oil price are
obtained from U.S. Energy Information Administration. In addition, we
use the monthly data of the S&P 500 index which are acquired from
YAHOO FINANCE.

3.2. Analysis for different factors

The US dollar index is a comprehensive indicator used to measure

the exchange rate changes of the US dollar in the international foreign
exchange market. On the one hand, the international gold price is based
on the US dollar as the price scale. If the US dollar index falls, it means
that the dollar-priced commodity has appreciated indicating the price
of gold has risen; on the contrary, it means that dollar-priced goods are
depreciating if the dollar index rises, thus the price of gold is falling. On
the other hand, both the US dollar and gold are important international
official reserve assets. Therefore, there will inevitably be a certain de-
gree of substitution between the two from the perspective of invest-
ment. Furthermore, Fig. 2 plots the gold price from January 1979 to
November 2018, showing that there is a negative correlation between Fig. 3. Gold price and federal funds rate trend from December 1978 to January
the US dollar index and the gold price. In short, the US dollar index is 2019. (For interpretation of the references to colour in this figure legend, the
one of the most important factors in the price of gold whether from a reader is referred to the Web version of this article.)

3
Y. Qian, et al. Resources Policy 64 (2019) 101478

At present, the existing literature does not agree on the relationship correlated with the world's major currencies, of which the correlation
between gold price and inflation rate. Levin and Wright (2006) believe coefficient is low. Fig. 5 depicts the trend of USD/RMB exchange rate
that the price of gold is positively correlated with the US CPI while the and gold price from January 2000 to December 2018. The relationship
correlation between gold price and world inflation rate is not strong. between the two is difficult to understand from the figure, and sub-
But Erb and Harvey (2013) show that gold is not effective against both sequent verification is needed.
expected and unanticipated inflation. To analyze the confrontational As a major oil-consuming country, the economic development will
effect of gold on unanticipated inflation, it is necessary to examine the inevitably be affected when the oil price fluctuates, so as the price of
relationship between inflation rate changes and movements in gold commodities such as gold denominated in dollars. The fluctuations of
prices. The volatility of the price should be small if gold is an ideal asset Oil price have long been regarded as a signal for the rise and fall of gold
to cope with inflation in the short term. However, as can be seen from prices. Some scholars believe that the future gold trend can be judged
Fig. 4, the actual gold price fluctuates greatly with rises and falls. by the price ratio of gold/crude oil, which is the price of one ounce of
Therefore, the relationship between gold price and inflation rates re- gold and the price of one barrel of crude oil. A study by Sujit and Kumar
mains to be seen. (2011) found that the gold/crude oil price ratio was about 15.2 over the
Due to the influence of pricing factors, gold can hedge the risk of past 40 years. However, these views are still debatable. The US West
depreciation of the US dollar to some extent. Erb and Harvey (2013) Texas Light Crude Oil Index is regarded by many investors as the
studied the correlation between gold prices and major world currencies benchmark price of the international energy market, so I will use WTI
from 1975 to 2012. The results show that gold prices are negatively as an indicator to measure the price of oil. As can be seen from Table 1,
the highest price ratio of gold and crude oil in the past 35 years is
18.90, while the lowest value is only 9.51. Besides, Fig. 6 depicts the
trend of gold prices and oil prices from April 1983 to January 2019. It
can be deduced that the ratio of gold and crude oil prices fluctuated
greatly and it took a long time for coming back to the balance point.
From the historical experience, the short-term trend or long-term trend
of gold prices can be poorly predicted by the oil prices.
Stocks and gold are generally common assets that investors use to
construct portfolios, thus resulting in certain substitution between the

Table 1
Price ratio of Gold and oil from 1983 to 2017.
Period Gold price Oil price Ratio

1983–1987 379.67 24.09 15.76

1988–1992 379.45 20.41 18.59
1993–1997 369.7725 19.338 19.1
1998-2002- 287.14 23.16 12.39
2003–2007 508.85 53.50 9.51
2008–2012 1268 86.06 14.74
Fig. 4. Gold price and CPI trend from December 1978 to January 2019. (For 2013–2017 1262.79 66.80 18.90
interpretation of the references to colour in this figure legend, the reader is
referred to the Web version of this article.)

Fig. 5. USD/RMB exchange rate and gold price from January 2000 to December Fig. 6. The trend of gold prices and oil prices from April 1983 to January 2019.
2018. (For interpretation of the references to colour in this figure legend, the (For interpretation of the references to colour in this figure legend, the reader is
reader is referred to the Web version of this article.) referred to the Web version of this article.)

4
Y. Qian, et al. Resources Policy 64 (2019) 101478

during the period when the index changes greatly, suggesting the ne-
gative correlation between them.
It can be seen from Table 2 that the demand and supply for gold are
basically in a relatively stable state. We can expect that the demand of
gold and the supply of gold cannot be suddenly changed in a short
period of time, thus both of them will not have a significant impact on
the price of gold. In this article, we will ignore the impact of gold de-
mand and supply on the price of gold, for the balance of supply and
demand of gold will not change in the near future.

4. Data description and empirical methodologies

4.1. Data description

Our paper considers the price of gold as the dependent variable.

Factors we analyze in this section include the US dollar index, the US
federal funds rate, CPI, exchange rate, oil price and S&P500. The source
Fig. 7. Gold price and S&P500 index trend from December 1978 to January of the data is the same as described in 3.1, but the sample interval we
2019. (For interpretation of the references to colour in this figure legend, the select in the empirical study is from January 2000 to December 2018.
reader is referred to the Web version of this article.)
4.2. Box–Behnken design
two. The stock market is often used as judgments about future economic
trends. Stocks are more risky than gold, but they will produce higher Box–Behnken design (BBD), a commonly used form of RSM, was
returns than gold since risks and returns are proportional. If at the used to analyze the factors affecting global gold price. We consider 6
summit of economy, the stock market will be more prosperous than the factors including Dollar Index, Federal funds rate, CPI, Exchange rate,
gold market; otherwise, the situation is just the opposite. However, the Oil price and S&P500, every one of which has 3 different levels,-1,0,1.
price of gold is only significantly related to the representative and in- Factors and their levels for Box–Behnken Design are in Table 3:
fluential stock index. In this article, the S&P 500 index will be used as Based on the coded factor levels and the observed data, we get 94
an indicator to measure the stock market. From Fig. 7, we can see that different experimental value by using Design-Expert. Part of the ex-
the S&P 500 index and gold price change in totally opposite directions perimental design matrix and response are showed in Table 4:

Table 3
Table 2 Factors and their levels for Box–Behnken Design.
Global gold supply and demand from 2010 to 2018.
Variable Symbol Coded factor levels
Year Gold total supply Gold total demand
−1 0 1
2010 4313.90249 4219.854
2011 4519.00208 4702.046 Dollar Index A 80–90 90–100 100-
2012 4543.31187 4714.202 Federal funds rate B 0–2 2–4 4-
2013 4292.7189 4513.087 CPI C 0–200 200–230 230-
2014 4430.05819 4335.321 Exchange rate D 0–7 7–8 8-
2015 4340.97462 4302.166 Oil price E 0–50 50–90 90-
2016 4599.70304 4426.839 S&P500 F 700–1200 1200–1800 1800-
2017 4447.15434 4159.907
2018 4490.17675 4345.103

Table 4
Experimental design matrix and response.
Number Coded factor Response variable

A B C D E F Gold price

1 1.000 1.000 −1.000 1.000 −1.000 0.000 283.3

2 1.000 1.000 −1.000 1.000 −1.000 −1.000 257.7
3 1.000 0.000 −1.000 1.000 −1.000 0.000 270.6
4 1.000 0.000 −1.000 1.000 −1.000 −1.000 273
5 1.000 −1.000 −1.000 1.000 −1.000 −1.000 415.65
6 1.000 −1.000 −1.000 1.000 0.000 −1.000 425.55
7 0.000 1.000 0.000 1.000 0.000 0.000 653
8 0.000 1.000 0.000 0.000 0.000 0.000 789.5
9 0.000 −1.000 0.000 0.000 −1.000 −1.000 919.5
10 −1.000 −1.000 0.000 −1.000 0.000 0.000 1813.5

5
Y. Qian, et al. Resources Policy 64 (2019) 101478

Fig. 8. Actual and predicted values of the model.

5. Results and discussion

Table 5
We performed multiple linear regression and binomial fitting on the Analysis of variance (ANOVA) for response surface quadratic model.
experimental data in the above table by using Design Expert to obtain a Source Sum of squares Degree of Mean square F value Prob＞F
multivariate quadratic regression response surface model. And the final freedom
estimated response model equation for the gold price in terms of coded
Model 1.628E+007 13 1.252E+006 110.90 < 0.0001
factors is shown in the following equation: A 2.507E+005 1 2.507E+005 22.21 < 0.0001
B 4.504E+005 1 4.504E+005 39.90 < 0.0001
Y = 869.41 122.25A 181.25B + 55.68C 233.53D
C 17189.35 1 17189.35 1.52 0.2208
3.5E 73.41F + 151.17AE + 164.77BD + 436.66CD D 2.449E+005 1 2.449E+005 21.70 < 0.0001
59.53CE + 153.12A2 + 355.06C 2 124.84F 2 E 247.75 1 247.75 0.022 0.8826
F 53668.90 1 53668.90 4.75 0.0322
As can be clearly noted in Fig. 8, all of the predicted points were AE 3.051E+005 1 3.051E+005 27.03 < 0.0001
distributed uniformly and closely to the actual experimental points. BD 2.624E+005 1 2.624E+005 23.25 < 0.0001
Hence, the model sufficiently describes the correlation between the CD 5.403E+005 1 5.403E+005 47.86 < 0.0001
CE 48464.83 1 48464.83 4.29 0.0415
global gold price and the 6 factors we addressed. Besides, the signs of
A2 2.357E+005 1 2.357E+005 20.88 < 0.0001
the coefficient in Eq. (1) suggest that the corresponding factor has a C2 2.330E+005 1 2.330E+005 20.64 < 0.0001
positive or negative influence for the gold price. Specifically, CPI has a F2 1.274E+005 1 1.274E+005 11.28 0.0012
positive effect on the gold price which was also claimed by Levin and Residual 9.031E+005 80 11288.68
Wright (2006). As for Dollar Index、federal funds rate、exchange Lack of Fit 2.645E+005 29 9120.53 0.73 0.8190
Pure Error 6.386E+005 51 12521.56
rate、oil price and S&P500, they all have a negative effect on the re-
Cor Total 1.718E+007 93
sponse variable, which coordinates with Erb and Harvey (2013)and the R-Squared 0.9474 Adj R-Squared 0.9389
theoretical analysis.
In order to understand the statistical significance of the coefficients
above, we also performed analysis of variance(ANOVA). The sig-
nificance level was fixed at 5% to determine whether model terms are F value of 0.73 and the Prob > F value of 0.8190 stresses its insig-
significant or not to the response variable. The results of ANOVA are nificancy which further prove that the model properly fitted the data.
displayed in Table 5. The residual is the difference between the observed value and the
We can see that the model fits well due to the R-Squared value of predicted value. The residual analysis can determine the reliability of
0.9556 which indicates that 95.56% of the total variation in the gold the data, thus indicating the reliability of the model and whether the
price can be explained by the experimental variables studied. The response surface meets the assumptions of the ANOVA. Therefore, in
model F-value of 110.90 and Prob > F of less than 0.0001 emphasizes addition to the above, we also performed a residual analysis on the
that this model is significant (Table 5). In addition, factor C and factor E model. The results are shown in Fig. 9.
are insignificant terms while all other terms were significant to the It can be seen from Fig. 9 that the response values are basically
response which means that the CPI and oil price does not have a sig- attached to the residual line, indicating that the model conforms to the
nificant influence on gold price at 5% significance level. The lack-of-fit normal distribution. The internal and external studentized residuals are

6
Y. Qian, et al. Resources Policy 64 (2019) 101478

Fig. 9. Normal probability plot of the model residuals.

all distributed within ± 4.0, reflecting that there is no abnormal data or funds rate. The third panel illustrates the interactive effect between CPI
any hidden factors that influence the response surface factors. Besides, and exchange rate, the CPI seems to have an intricate influence on the
the random distribution of data points without any trend or abnormal response variable while the exchange rate is fixed. There is a turning
points further illustrate the reliability of the model. Hence, the response point of 210 for the CPI. when the CPI is below this point, it has a
surface polynomial model is highly significant and can effectively negative effect on the gold price which is adverse as the CPI exceeds
characterize the relation between the response variable and the factors 210. The last panel confirms the turning point of 210 for CPI, the effect
studied. of CPI on the response variable is totally opposite before and after the
So as to picture the interaction effects between the variables, three turning point. The oil price seems to have a smaller impact on the gold
dimensional response surfaces were drawn for each of the two targeted price than CPI as demonstrated by the curvature and density of the
variables while another variable remained constant at its corresponding contour line. On the whole, the interaction effects on the gold price
center level coded as zero. among the factors in Fig. 10 are significant although complicated due to
As can be seen from Fig. 10 and the ANOVA above, the interactive the range of some factors.
effect between the dollar index and oil price is significant. The gold
price decreases when dollar index goes up with fixed oil price, which is 6. Conclusion
the case of dollar index below 100. However, when the dollar index is
higher than 100, the scene becomes complicated. For a fixed dollar This paper begins with a concise introduction relating to the im-
index, the oil price has a negative effect on the gold price while the portance of gold in the world economy. Motivated by this, we in-
dollar index is less than 100. However, the effect turns out to be posi- vestigate the determinants of the price of gold with a particular atten-
tive as the dollar index surpasses 100. As for factor B and factor D, the tion on six different factors (i.e., the dollar index, federal funds rate,
situation becomes relatively simple. The gold price goes down as the CPI, the exchage rate, oil price and S&P500). In the empirical study, we
federal funds rate rises while the exchage rate is constant. Analogously, get a multivariate quadratic equation for the relationship between gold
The gold price declines as the exchage rate goes up with a fixed federal price and the six factors. The model fits well through the fitting figure

7
Y. Qian, et al. Resources Policy 64 (2019) 101478

Fig. 10. Interactive effect between factors.

between observed and predicted values, which can also be verified by not significant. However, the interaction between the federal funds rate
the results of ANOVA. We find that in addition to CPI, the other factors and the exchange rate is relatively simple. Finally, we find that the
have a negative impact on the price of gold. Furthermore, the impact of squared terms of dollar index, CPI and S&P 500 index have a significant
CPI and Oil price on response variable was not significant at the 5% impact on the price of gold. The findings of this paper offer great in-
significance level, which also has a decisive influence on the subsequent sights into factors affecting the price of gold and practices of the
interactions. Regarding the interaction, the results imply that the in- portfolio diversification.
teraction between the US dollar index and the oil price, the federal Future research on the determinants of the gold price should focus
funds rate and the exchange rate, the CPI and the exchange rate, the CPI on some uncertainty measures. For example, the volatility (VIX),
and the oil price have a significant impact on the price of gold. skewness (SKEW), global economic policy uncertainty (EPU), partisan
Specifically, the interaction effect between the dollar index and the oil conflict (PC) indexes and U.S. economic policy uncertainty, which can
price, the CPI and the exchange rate, the CPI and the oil price changes be considered as the potential drivers of the gold price. Besides, there
according to the range of factors in completely different directions. It are also some alternative ways of examining this issue. For instance,
can be speculated that this result is related to the CPI and the oil price, using standard cointegration techniques or fractional cointegration
because the influence of these two factors on the dependent variable is methods as those proposed in Johansen and Nielsen (2010, 2012).

8
Y. Qian, et al. Resources Policy 64 (2019) 101478

References Johansen, S., Nielsen, M.Ø., 2010. Likelihoodinference for a nonstationary fractional
autoregressive model. J. Econom. 158, 51–66.
Johansen, S., Nielsen, M.Ø., 2012. Likelihood inference for a fractionally cointegrated
Aye, G., Gupta, R., Hammoudeh, S., Kim, W.J., 2015. Forecasting the price of gold using vector autoregressive model. Econometrica 80 (6), 2667–2732.
dynamic model averaging. Int. Rev. Financ. Anal. 41, 257–266. Kang, S.H., McIver, R., Yoon, S.M., 2017. Dynamic spillover effects among crude oil,
Baur, D.G., Lucey, B.M., 2010. Is gold a hedge or a safe haven? An analysis of stocks, precious metal, and agricultural commodity futures markets. Energy Econ. 62, 19–32.
bonds and gold. Financ. Rev. 45 (2), 217–229. Lau, M.C.K., Vigne, S.A., Wang, S., Yarovaya, L., 2017. Return spillovers between white
Baur, D.G., McDermott, T.K., 2010. Is gold a safe haven? International evidence. J. Bank. precious metal ETFs: the role of oil, gold, and global equity. Int. Rev. Financ. Anal.
Financ. 34 (8), 1886–1898. 52, 316–332.
Białkowski, J., Bohl, M.T., Stephan, P.M., Wisniewski, T.P., 2015. The gold price in times Li, S., Lucey, B.M., 2017. Reassessing the role of precious metals as safe havens–What
of crisis. Int. Rev. Financ. Anal. 41, 329–339. colour is your haven and why? J. Commod Mark. 7, 1–14.
Baruník, J., Kočenda, E., Vácha, L., 2016. Gold, oil, and stocks: dynamic correlations. Int. Li, D.Q., Jiang, S.H., Cao, Z.J., Zhou, W., Zhou, C.B., Zhang, L.M., 2015. A multiple re-
Rev. Econ. Financ. 42, 186–201. sponse-surface method for slope reliability analysis considering spatial variability of
Beckmann, J., Czudaj, R., Pilbeam, K., 2015. Causality and volatility patterns between soil properties. Eng. Geol. 187, 60–72.
gold prices and exchange rates. N. Am. J. Econ. Finance 34, 292–300. Levin, E.J., Wright, R.E., 2006. Short-run and Long-Run Determinants of the Price of Gold
Bilgin, M.H., Gozgor, G., Lau, C.K.M., Sheng, X., 2018. The effects of uncertainty mea- [R]. World Gold Council.
sures on the price of gold. Int. Rev. Financ. Anal. 58, 1–7. Nam, S.N., Cho, H., Han, J., Her, N., Yoon, J., 2018. Photocatalytic degradation of ace-
Bartley, M.L., Boeing, W.J., Daniel, D., Dungan, B.N., Schaub, T., 2016. Optimization of sulfame K: optimization using the Box–Behnken design (BBD). Process Saf. Environ.
environmental parameters for Nannochloropsis salina growth and lipid content using Prot. 113, 10–21.
the response surface method and invading organisms. J. Appl. Phycol. 28 (1), 15–24. O'Connor, F.A., Lucey, B.M., Batten, J.A., Baur, D.G., 2015. The financial economics of
Bakhtiari, G., Abdouss, M., Bazmi, M., Royaee, S.J., 2016. Optimization of sulfur ad- gold—a survey. Int. Rev. Financ. Anal. 41, 186–205.
sorption over Ag-zeolite nanoadsorbent by experimental design method. Int. J. Pierdzioch, C., Risse, M., Rohloff, S., 2014a. On the efficiency of the gold market: results
Environ. Sci. Technol. 13 (3), 803–812. of a real-time forecasting approach. Int. Rev. Financ. Anal. 32, 95–108.
Ewing, B.T., Malik, F., 2013. Volatility transmission between gold and oil futures under Pierdzioch, C., Risse, M., Rohloff, S., 2014b. The international business cycle and gold
structural breaks. Int. Rev. Econ. Financ. 25, 113–121. -price fluctuations. Q. Rev. Econ. Financ. 54 (2), 292–305.
Erb, C.B., Harvey, C.R., 2013. The golden dilemma. Financ. Anal. J. 69 (4), 10–42. Ren, Z., Li, W., Billinton, R., Yan, W., 2016. Probabilistic power flow analysis based on the
Gil-Alana, L.A., Yaya, O.S., Awe, O.O., 2017. Time series analysis of co-movements in the stochastic response surface method. IEEE Trans. Power Syst. 31 (3), 2307–2315.
prices of gold and oil: fractional cointegration approach. Resour. Policy 53, 117–124. Sujit, K.S., Kumar, B.R., 2011. Study on dynamic relationship among gold price, oil price,
Gupta, R., Hammoudeh, S., Kim, W.J., Simo-Kengne, B.D., 2014. Forecasting China's exchange rate and stock market returns. Int. J. Appl. Bus. Econ. Res. 9 (2), 145–165.
foreign exchange reserves using dynamic model averaging: the roles of macro- Tak, B.Y., Tak, B.S., Kim, Y.J., Park, Y.J., Yoon, Y.H., Min, G.H., 2015. Optimization of
economic fundamentals, financial stress and economic uncertainty. N. Am. J. Econ. color and COD removal from livestock wastewater by electrocoagulation process:
Finance 28, 170–189. application of Box–Behnken design (BBD). J. Ind. Eng. Chem. 28, 307–315.
Jones, A.T., Sackley, W.H., 2016. An uncertain suggestion for gold-pricing models: the Wang, G.J., Xie, C., Jiang, Z.Q., Stanley, H.E., 2016. Extreme risk spillover effects in
effect of economic policy uncertainty on gold prices. J. Econ. Financ. 40 (2), world gold markets and the global financial crisis. Int. Rev. Econ. Financ. 46, 55–77.
367–379.