Measuring the relationship between wealth and gold in selected countries of the One Road, One Belt initiative
Abraham Rodríguez Tostado
This proposal explores the relationship between the wealth of selected economies in the One Road, One Belt initiative and their respective gold demands. Wealth was measured by household expenditures per capita in the local currency of each economy. Three estimations were run, one for the overall gold demand and two for its main components: gold jewelry, and gold bar and coin. The estimation ran with the demand of gold jewelry as its dependent variable had the highest R2 (.926), which was expected as more than half of the annual gold demand is exclusively gold jewelry. All key variables are significant at the 1% level of significance. The results were not as expected in several cases, but they can be explained by referring to classical economic theory and cultural influences, such as religion. The most influential key variable was the local price of an ounce of gold with a negative relationship due to higher prices being associated with lower quantities demanded. Nonetheless, the aim of the proposal was achieved as it is demonstrated that there is a positive relationship between wealth and gold demand. In addition, gold was found to be a necessity and a normal good in the economies studied, which indicates its relevance in the selected emergent economies.
This paper could not have been done without the aid of my uncle, the advice of my father, and the encouragement of my mother.
In addition, without the insight of my friend Agustin with whom I held several discussions about this paper’s potential shortcoming and their respective solutions, this paper would have surely ended being of lower quality. Cheers, my friend, next year is your time to write a proposal…
If gold was a currency, this year (2016) it would have been the number two currency in the world, second only to the US dollar. It outperformed all other currencies when its compared in percent gains against the US dollars in such degree that its gains are 85% greater than the second place (Japanese Yen) and a 118% greater than the third place (Norwegian Krone).
However, gold is certainly not a currency but that does not mean that it cannot be used to build wealth. In the olden days, coins were minted in gold, silver, and copper. In contrast with the past, now gold can only be found in the form jewelry, bars, and coins (the coins do have a monetary value but it’s mostly a decoration, their value is entirely composed of their weight and the market premium specific to that coin). This proposal is going to explore the relationship between the wealth of selected countries of the One Belt, One Road initiative and their demand for gold.
Furthermore, this proposal is also going to be studying the demand for gold by its two main components: demand for gold jewelry, and demand for gold bars and coins. The former accounts for more than half of the overall gold demand. The latter is mostly used for investment purposes. Moreover, the framework of this proposal examines the impact of Islam in an economy’s overall demand for gold, which can only be measured by analyzing physical gold demand as Muslims are prohibited in investing in non-physical gold due to its speculative nature.
In addition, to studying the relationship of wealth and the demand for gold, this paper will also aim to prove the relevance of gold in the economies studied by analyzing the point elasticities of the demand for gold with regards to income and price.
- Literature Review
The relationship between gold and disposable income in certain emerging economies is an inverted u-shape curve, with the first half indicating a positive relationship between them, and the second half showing the contrary. Moreover, the inflection point of this relationship is the point of transition from lower middle class to upper middle class and beyond due to a change of consumer behavior from saving to consuming (Liu,2016). In other words, as individuals rise in economic status, their demand for gold will be lower. Parallel to this relationship, economies that become more developed will have a lower gold demand.
To better understand an individual’s transition between economic classes, they can be categorized into four different classes that will dictate their consuming behavior, especially the kind of goods and services they will purchase (Han et al., 2010). From the four groups, parvenu is the category that is essential to further understand the inverted-u relationship between gold demand and disposable income due to its members being lower middle class individuals who recently transition into the upper classes. Due to their recent advancement in the economic ladder, the parvenus have an urge to differentiate themselves from others of their former socio-economic class that is fulfilled by consuming luxury goods that will exhibit their elevation in status rather than continuing saving like their past fellows. Consequently, lower classes value saving more than upper classes due to the constant economic uncertainties that surrounds them (Sandmo, 1970). Thus, parvenus, who are new members of the upper classes, decrease their saving behavior in accordance to their now secure economic context.
In addition, this paper utilizes the gold to silver ratio, a financial tool that measures how many silver ounces are needed to buy a single gold ounce. This measure is currently considered a specious tool due to each metal having vastly different application (Ciner, 2001).
Moreover, religion is another factor that can affect a country’s demand for gold, as some religions do prohibit the usage of gold as jewelry. For example, the Quran states that it is haram (prohibited) for men to wear either gold or silk, because if they die wearing them Allah will deny them entrance to paradise. On the other hand, it is halal (allowed) for women to wear them (Al-Qaradawi, 1960). In the same manner, Islam does not prohibit the investment in gold but it is only allowed when gold is in its physical form. Thus, all gold demand that comes from Islamic economies must be physical, which is the kind of gold demand (gold jewelry, gold bars, and gold coins) this proposal studies, or Muslims that dare to held gold in any other non-physical form risk being unable to enter paradise.
This paper has two intertwined aims. The first is to prove that gold continues to be relevant in at least some of the current economies. The second is to support the existence of a n inverted-u relationship between wealth and gold demand in selected countries from the One Road, One Belt initiative, which will indicate that if these economies continue to be emergent ones their respective demands for gold should continue to increase (at a diminishing rate) until they become developed ones. If this relationship is support adequately, the relevance of gold in those economies should be demonstrated along with what factors have the greatest impact in its demand.
III. Empirical Framework
- Empirical Methodology
The relationship between gold demand and household expenditures per capita of selected countries of the One Road, One Belt initiative is being analyzed by an OLS regression for the years 2006 to 2014. The OLS regression is given by the following:
(1) Aurumit = B0 + B1Islamit + B2GtoSit + B3Gmintit + B4SPluxit + B5mHhExit + B6mLoCuOzit + B7Y2006 + B8Y2007 + B9Y2008 + B10Y2009 + B11Y2010 + B12Y2011 + B13Y2012 + B14Y2013 + B15Y2014 +B16China + B17Vietnam + B18Indonesia + B19Russia + B20Turkey + B21UnitedArabEmirates +B22Egypt + εit
Variables definitions and their respective sources can be found at Table 1. The regressions include year and country effects to account for China, Vietnam, Indonesia, Russia, Turkey, United Arab Emirates, and Egypt.
Furthermore, as the time interval of the regression (2006-2014) includes the Great Recession, which was an economic recession that resulted in unprecedented consequences such as negative interest rate and near-zero interest rates. This proposal studied the possibility of including the interest rates present in the selected countries during the years studied. However, the elasticity of saving rates with respect to changes in interest rates in emerging economies is miniscule (-0.1-0.2) (Balassa, 2013). Thus, the relevance of interest rates in this framework becomes non-existent.
Each of the following two regressions measures the effect of an increase in household expenditures per capita, a proxy for measuring the wealth of individuals of a country, in one of the two components of overall gold demand: demand for gold jewelry and demand for gold bars and coins.
(2) GDJit = B0 + B1Islamit + B2GtoSit + B3Gmintit + B4SPluxit + B5mHhExit + B6mLoCuOzit +δi + θt + εit
(3) GDBCit = B0 + B1Islamit + B2GtoSit + B3Gmintit + B4SPluxit + B5mHhExit + B6mLoCuOzit +δi + θt + εit
Household expenditures and the price of an ounce of gold are both measured in local currencies rather than US dollars to account for exchange rate fluctuations that can result in gold being relatively dearer or cheaper from one economy to another. The change in the relative price of gold is something that will affect the demand for gold of each economy, as gold is priced in US dollars but individuals earn and spend their respective local currencies. Furthermore, both variables are presented in their log forms due to range of their values.
SPlux, which accounts for the movements of S&P Global Luxury Index serves as proxy to measure if an economy is nearing the inflection point of the relationship between gold demand and household expenditures. Per the literature studied in this proposal, it is expected that this variable has an inverse relationship with gold demand due to luxury goods being a substitute for gold.
GtoS stands for a financial tool known as the gold and silver ratio, which measures how many ounces of silver are needed to acquire a single ounce of gold. When the ratio is over 80, gold tends to enter bullish territory, which means that individuals who have gold are most likely going to be profiting from their investments. An increase in the returns of gold can potentially increase its demand, as more people will want to benefit from its bullish run.
Gmint is a binary variable that accounts for the presence of a local mint, which should facilitate the acquisition of gold to locals. Moreover, the gold premium (a surcharge added to the price of gold when either gold bars or coins are purchased) should be lower if a mint is present, making gold even more attractive to consumers.
Islam is a binary variable that accounts for countries that have Islam as their main religion. In Islam, gold is a rather ambiguous commodity as it has a dual meaning associated with both goodness and temptation. Furthermore, only women can wear gold, as men who wear it are not to be granted entrance to paradise.
- Data and Descriptive Statistics
The data consists of the following eight countries of the One Road, One Belt initiative: China, Vietnam, Indonesia, Russia, Turkey, United Arab Emirates, and Egypt. Moreover, the yearly data was collected for years 2006 to 2014.
The White test was conducted in all three regressions, and the result for all three were the same a Chi-Square of 63 with a probability to reject of 0.4407, which results in null hypothesis (H0) that states the presence of homoscedasticity being rejected. Thus, all three equations have heteroscedasticity, an issue that was solved by running the regression with robust standard errors.
The range of the values of both variables HhEx and LoCuOz as can be seen in table 2 is too great, an issue that could result in bias in their respective coefficients. Log forms of the two variables were employed to deal with this issue. In table 2, the descriptive statistics of the modified variables can be observed.
In table 3, the only relevant correlation value (0.9616) is the one between mHhEX and mLoCuOz, but it is expected due to both variables being in local currencies and in log forms. They are not to be excluded from the regression as the direction of their relationship is in accordance with the literature, and because doing so may potentially result in omitted variable bias.
The results of the three estimations for the overall gold demand per country and its two components (demand for gold jewelry and demand for gold bars and coins) are shown in table 4. All three estimations explain the variance in each country’s demand for gold by at least 81.5%. Regression two has the most significance with an R2 of 92.6%. All three estimations have year fixed effects and country fixed effects. Furthermore, all key variables in all three regressions are significant at the 1% level.
Standard and Poor’s Global Luxury Index (SPlux) is the only variable that is measured in dollar terms rather than in the local currencies due to being a variable that measures the overall demand for luxury goods rather than each country’s specific demand. It was expected to have a negative coefficient, but it had a positive one. This unforeseen result could potentially be explained by the common usage of gold in certain luxury goods, as it helps to elevate a product status. Its inclusion is nevertheless reinforced by this unanticipated outcome, as it be seen in Table 4’s section of standardized betas that the influence of luxury goods in gold demand is actually higher than the influence of the gold to silver ratio, which is another variable studied in this approach that is also commonly used in the financial sector to see whether it is a good time to invest in gold or not. In other words, luxury brands are more influential than financial tools in regards to gold demand.
Mints (Gmint) are the institutions in charge of minting coins and bars in different materials, including gold. The presence of mint in an economy should benefit the local gold demand by providing a constant and accessible supply of gold in the forms of coins and bars. The expectation of the coefficient is for it to be positive due to the presence of a mint should allow for an easier access to gold bars and coins for a wider audience. However, the results were not as expected as the coefficient was negative. This can be explained by the scarcity principle, which presents the notion that goods that are hard to obtain become dearer and more appreciated by consumers (Hazlitt, 2010), and vice versa.
Islam is expected to have a negative coefficient due to how ambiguously gold is perceived in general by Muslims. Its results went in accordance to the expectations but there are more Islamic countries in the One Road, One Belt that could not be studied due to lack of data, and they should be studied for strengthening the validity of Islam and its negative relationship with gold demand.
Gold to silver ratio (GtoS) measures how many ounces of silver are equivalent to a single ounce of gold pricewise. Per literature, this tool was deemed as unreliable, but the opinion of experts in the field stated otherwise. In addition, its significance at the 1% in all three estimations endorses the opinion of the experts, but this may only be valid for countries in the One Belt, One Road initiative, which is an initiative promoted by the Chinese government that is equivalent to the American Trans-Pacific Partnership.
The price of gold in local currencies (mLoCuOz) is expected to have negative relationship with all three dependent variables as they account for demand. Price and demand in accordance to basic economic theory always have a negative relationship because people are expected to consume more of a good when its price is lower. This is the case for all normal goods, and gold is one of them at least in the economies studied, as it is shown in table 6 that the price elasticity for all kinds of gold demand is inelastic (-0.86-0.87).
On the other hand, household expenditures (mHhEx) is expected to have a positive sign due to an increase in wealth being associated with a higher gold demand in emergent economies at the early stage of their development (Liu, 2016). This is demonstrated in all three estimations, which serve to support the initial hypothesis that an increase in the wealth of a country is associated with an increase in that country’s gold demand.
Furthermore, the income elasticity for all three dependent variables is inelastic (0.29-0.31), which means that in the economies studied gold is regarded as a necessity. Thus, following economic theory that states that people will consume necessities regardless of their income, the demand of gold will be maintained constant. However, due to gold being also a normal good, its quantity demanded will also increase when income rises. Thus, gold results in being a good, which demands will either remain constant or increase in relation with income.
In table 5, the standardized betas offer us the opportunity to compare all key variables for determining which one has the most influence in the dependent variable(s). The price of gold (mLoCuOz) is the most influential variable, which was expected due to price and demand having an extremely close relationship. Second in influence was Islam, which is understandable due to the perceived nature of gold in Islam. Income does have a significant effect in the demand for gold and a positive relationship with it. Thus, the demand for gold will depend in a greater degree to the price of gold rather than the income of its consumers. This situation could result in a scenario in which the demand for gold is lower regardless of an increase in the wealth of individuals due to significantly high price.
The framework of this paper demonstrates that as wealth increases, so does its demand for gold. However, it was also proven that there are quite several factors that affect the demand for gold in greater degree than an increase in the wealth of individuals such as the price of an ounce of gold, Islam as the economy main religion, and whether the economy has a mint or not. Despite these factors, the only variable with both a positive coefficient and a significant degree of influence (the other variables with a positive coefficient have a standardized beta lower than one, which results in them having almost no influence even though they continue to be significant to at the 1% level) is household income per capita (proxy variable for wealth). Furthermore, all the countries studied in this proposal are emergent economies, which means that they are on their path towards becoming developed economies, and in that transition wealth must continue to rise until they become developed ones due to the nature of the relationship between wealth and the demand for gold being an inverted-u.
Gold was proven to be relevant in the economies studied due to being considered both a necessity and a normal good, but further research is needed to arrive to a general conclusion regarding its relevance in all countries of the One Belt, One Road initiative. Furthermore, the impact of Islam in the demand for gold needs to be also studied in a greater degree due to being close to having the same influence as the price of gold in demand for it.
Al-Qaradawi, Y. (1999). The Lawful and the Prohibited in Islam (al-halal wal haram fil Islam). American Trust Publications.
Balassa, B. (2013). The effects of interest rates on savings in developing countries. PSL Quarterly Review, 43(172).
Ciner, C. (2001). On the long run relationship between gold and silver prices A note. Global Finance Journal, 12(2), 299-303.
Han, Y. J., Nunes, J. C., & Drèze, X. (2010). Signaling status with luxury goods: The role of brand prominence. Journal of Marketing, 74(4), 15-30.
Hazlitt, H. (2010). Economics in one lesson: The shortest and surest way to understand basic economics. Crown Business.
Liu, J. (2016). Covered in Gold: Examining gold consumption by middle class consumers in emerging markets. International Business Review, 25(3), 739-747.
Sandmo, A. (1970). The effect of uncertainty on saving decisions. The Review of Economic Studies, 37(3), 353-360.
Table 1. Definition and Sources of Key Variables
|Aurum||The overall gold demand of a country for a given year in tons||The World Gold Council|
|GDJ||The demand of gold only for jewelry purposes of a country for a given year in tons||The World Gold Council|
|GDBC||The demand of gold solely for investment purposes of a country for a given year in tons||The World Gold Council|
|mHhEx||Household expenditures per capita in the local currency of each respective country for a given year. Log Form||The World Bank Data Bank|
|mLoCuOz||The price of a single ounce of gold in terms of the local currency of each respective country for a given year. Log Form||The World Bank Data Bank|
|Gmint||A dummy variable that equals one, if the country has a mint. Otherwise, it equals zero||Mints of the World|
|GtoS||A financial tool known as the gold to silver ratio. It measures how many silvers ounces are equal to a single ounce of gold pricewise||Author’s Calculations|
|SPlux||S&P Global Luxury Index in terms of US dollars||Standard and Poor’s Indices|
|Islam||A binary variable that equals one, if the main religion of the country is Islam. Otherwise, it equals zero||Infoplease|
Table 2. Descriptive Statistics (n=63)
|Aurum (Overall Physcal Gold Demand)||176.11||266.47||35||1463.3|
|Price of Gold||4959804||8850852||1139.813||3.41×107|
|Gold to Silver ratio||64.92||16.88||40.22||100.79|
|Standard and Poor’s Global Luxury Index||1433.31||433.31||794.6||2234.03|
Table 3. Descriptive Statistics of log variables (n=63)
|Household Expenditures (log)||12.15||3.17||9.1||17.3|
|Price of Gold (log)||11.05||3.63||7.03||17.34|
Table 4. Correlation Matrix
|Aurum (Overall Physical Gold Demand)||1|
|Price of Gold (log)||-0.2554||1|
|Household Expenditures (log)||-0.3597||.9616||1|
|Standard and Poor’s Global Luxury Index||0.1502||0.0067||0.0297||1|
|Gold to Silver ratio||-0.0463||-0.0139||-0.0101||-0.3952||0||1|
Table 5. Summary of All Three Estimations with the non-standardized betas and standardized betas
|Non-Standardized Betas||Standardized Betas|
|Dependent Variables||Aurum (Overall Physical Gold Demand)||Gold Jewelry Demand||Gold Bar and Coin Demand||Aurum (Overall Physical Gold Demand)||Gold Jewelry Demand||Gold Bar and Coin Demand|
|Standard & Poor’s Global Luxury Index||0.488***||0.356***||0.132***||0.8***||0.76***||0.78***|
|Gold to Silver ratio||11.005***||7.873***||3.132***||0.7***||0.66***||-73***|
|Household Expenditures (log)||423.730***||307.293***||116.437***||5.04***||4.8***||5.1***|
|Price of Gold (log)||-1,381.822***||-1,018.387***||-363.435***||-18.81***||-18.3***||-18.14***|
|Year Fixed Effects||Yes||Yes||Yes||Yes||Yes||Yes|
|Country Fixed Effects||Yes||Yes||Yes||Yes||Yes||Yes|
|Robust standard errors in brackets|
|*** p<0.01, ** p<0.05, * p<0.1|
Table 6. Point Elasticities
|Dependent Variable||Income Elasticity||Price Elasticity|
|Aurum (Overall Physical Gold Demand)||0.29||-0.87|
|Gold Jewelry Demand||0.29||-0.86|
|Gold Bar and Coin Demand||0.31||-0.87|