Explaining the gold price after the Bretton Woods Agreement using independent variables. An ARIMA model approach


Master's Thesis, 2014
75 Pages, Grade: 1.7

Excerpt

Table of Contents

List of Tables

List of Figures

Executive Summary

Chapter 1 – Introduction
1.1 Background
1.2 Gold is different
1.3 The gold price since the end of Bretton Woods
1.4 Research questions

Chapter 2 – Literature Review and Theory
2.1 Theoretical framework: Explaining the movements of the gold price
2.2 Empirical findings: Independent variables correlating with the gold price
2.3 Conclusion

Chapter 3 – Data and Methods
3.1 The ARIMA model
3.2 Assumptions of an ARIMA model
3.3 Data collection and sources
3.4 Defining an ARIMA model to fit the gold price
3.5 Evaluation of the ARIMA model
3.6 Conclusion

Chapter 4 – Analysis and Results
4.1 Data description
4.2 The best fitting ARIMA model
4.3 ARIMA model fit during normal times and crises
4.4 Explaining divergences of the model fit during normal times and crises
4.5 Conclusion

Chapter 5 – Discussion and Conclusions
5.1 Summary
5.2 Implications
5.3 Limitations
5.4 Direction for Future Research
5.5 Reflections

References

Appendices

Project Proposal

List of Tables

Table 1: Data sources

Table 2: Descriptive statistics of dependent and independent variables: minimum, maximum, mean and standard deviation.

Table 3: Correlation between the gold price and the independent variables

Table 4: ARIMA model description

Table 5: ARIMA model statistics

Table 6: ARIMA model parameters contributing to the fit of the monthly price of gold

Table 7: Independent variables contributing to the quality of an ARIMA model during the complete period (March 1973 to December 2011), during normal times and during crises (January 1978 to January 1981)

Table 8: ARIMA model statistics during crises (January 1978 to January 1981 and August 2007 to December 2011).

Table 9: ARIMA model statistics during normal times (March 1973 to December 1977 and from February 1981 to July 2007).

Table 10: ARIMA model parameters contributing to the fit of the monthly price of gold during crises (January 1978 to January 1981 and August 2007 to December 2011).

Table 11: ARIMA model parameters contributing to the fit of the monthly price of gold during normal times (March 1973 to December 1977 and from February 1981 to July 2007).

List of Figures

Figure 1: Average gold prices in US dollars from 1973 to 2011 (rounded). Sources: Data for 1973 to 2008 taken from World Gold Council 2008 and for 2008 to 2011 taken from Kitco.com

Figure 2: Gold time-series made stationary by differencing once

Figure 3: Partial autocorrelation function (PACF) and autocorrelation function (ACF) of the model

Figure 4: Observed gold price and gold price model fit (MAPE for the model: 2.861%)

Executive Summary

To date, nobody has formulated a comprehensive theorem to determine gold valuation or precious metal prices. Until fairly recently, Eugene Fama’s Efficient Market Hypothesis was the predominant paradigm explaining asset markets but today it is widely acknowledged that markets can be irrational and investors are prone to act irrationally. When trying to explain gold market anomalies, behavioural science approaches can be useful. Phenomena such as herding (‘group think’), ‘safe value bias’ and investors’ ‘excessive extrapolation’ can help explain positive price performance over a certain time.

In this dissertation, the author investigates the applicability of a multivariate ARIMA (auto-regressive, integrated, moving average) model to help explain gold price movements from 1973 to 2011. This model uses the gold price and independent variables such as inflation, real interest rates, silver prices, the US dollar money supply (M2), oil prices, the MSCI World index and the S&P 500 as these are linked to gold and/or highly correlated with the gold price. The evaluation criteria were defined as R-squared, mean absolute percentage error (MAPE) and BIC. The model was calculated over so-called ‘normal times’ and times of crises (one political, one financial). The researcher used SPSS’ Expert Modeler to find the best-fitting ARIMA model and to identify the independent variables significantly contributing to the fit of the model. Remarkably, a multivariate ARIMA model using independent variables explained almost twice as much of the variability of the gold price as a univariate ARIMA model using only the gold price. Also, throughout the complete period and during normal times the model explained a much higher percentage of the variability of the gold price than during crises and comparably more of the independent variables contributed significantly to the fit of the model (5 vs. 2). This can be explained by investors’ tendencies to buy gold to preserve their assets (“safe value”), to follow the crowd (“herding”) and to extrapolate past price chart developments.

The results show that in an attempt to discern the cause of gold price movements, a multivariate ARIMA model outperforms a univariate ARIMA model significantly. The results of the study furthermore indicate researchers evaluating different methods to fit a time series should consider a multivariate ARIMA model, especially if the independent variables are highly correlated with the dependent variable.

Chapter 1 – Introduction

1.1 Background

“Gold – the different asset class” was the title of an article by Baumann and Sullivan (2011) published on Swiss bank Credit Suisse’s website in November 2011. It expressed the peculiarity of gold. Gold has been called a “zero-beta asset”, an “inflation hedge” and a “currency” (Fei and Adibe 2010, p. 1).

This study intends to develop a model to explain the movements in the price of gold since the end of the Bretton Woods era in March 1973 – and the reintroduction of free floating currencies – until 2011. It does so by testing those independent variables that researchers have found to show a strong correlation and/or connection with the gold price in US dollars. Given the data and the research goal, the research is based on ARIMA (auto-regressive, integrated, moving average, for model discussion see Chapter 3), the most frequently used model for time series or longitudinal type data (Adbullah 2012, p. 153). ARIMA models are often used to fit time series data as they comprise a robust family of models capable of correcting for autocorrelations, non-stationary data, and excessive volatility (SPSS 2012, p.3/4).

While univariate ARIMA models are limited to the information contained in the series itself to predict future data within the time series, multivariate ARIMA models include explanatory variables (De Gooijer and Hyndman 2006, p. 447). For the purposes of this dissertation, a multivariate ARIMA model included the following independent variables designed to illuminate gold price movements: inflation, the real interest rate, silver prices, the US dollar money supply (M2), oil prices, the MSCI World index, the S&P 500.

During crises, demand for gold tends to be unusually high as investors favour assets considered conservative. Therefore, for the purposes of this paper, research will include periods of so-called ‘normality’ and those of crises (Harmston 1998, p. 6/7). Between 1973 and 2011 the world experienced two major crises with global implications on the financial markets, oil and gold prices: a political crisis between 1978 and 1981 and a financial crisis that broke out in 2007. Oil price shocks will be considered by including the independent variable “oil price”; stock market crashes will be considered by including the independent variables MSCI World Index and the S&P 500.

To discuss the effect of certain crises on the gold price, the research focuses on two of the most significant periods (Amey 1998, p. 49/50):

- From January 1978 until the end of January 1981: Civil resistance against the monarchy in Iran intensified and culminated in the abdication of the Shah on February 11, 1979. Shortly afterwards, Iran held 52 US citizens hostage and that Christmas the Soviet Union invaded Afghanistan. The US hostages were released on January 20, 1981
- From 9th August 2007 until the end of the period under consideration, as the Eurozone was still struggling and the US was trying to exit a prolonged recession, a global financial crisis broke out after the French bank BNP Paribas announced it had ceased support for three hedge funds that specialised in US mortgage debt. This, in turn, was followed by the bankruptcy of the first major investment bank, Lehman Brothers, on September 15, 2008 (Elliott 2011 and Kingsley 2012).

1.2 Gold is different

Gold is not like other metals because its industrial use is negligible, which makes it different to other commodities such as zinc, copper or silver. This explains why the price of gold often moves differently than the price of other commodities during a recession or a depression and especially during periods of high inflation (World Gold Council 2011, p. 8). The gold supply is primarily absorbed in the production of jewellery, by central banks, investors and more recently by financial institutions offering gold ETFs (Shafiee and Topal 2010, p.178). Gold is also special because of its distinctive place in economic history and its use as a financial asset, in particular as a hedge against inflation and geopolitical and/or economic risk. Many individuals add gold to their portfolios as a risk diversifier (Dempster 2008, p. 5).

1.3 The gold price since the end of Bretton Woods

Before the introduction of the gold standard in 1900 (it was dropped in 1933), gold had been traded over the counter in London since the 17th century. In 1944, the Bretton Woods system of pegged exchange rates was introduced. It was named after an international conference held in the town of Bretton Woods in the US state of New Hampshire, which was designed to establish new global commercial rules following World War II. Between 1944 and 1971 the currencies of 43 nations were pegged to the US dollar, which in turn was fixed to the gold price (The Gold Standard). The price of an ounce of gold was fixed at 35 US Dollars (Hammes and Wills 2005, p. 504). The International Monetary Fund (IMF) was given the authority to intervene in case of an imbalance of payments (Karunagaran 2011, p. 5/6). The Bretton Woods system meant US dollars were fully convertible into gold at this fixed rate. Steadily growing trade volumes and enhanced world production meant increased demand for US dollars. At the same time gold holdings – U.S. gold holdings in particular – decreased. This caused the collapse of the system (Garber 1993, p. 461).

In 1968 a two-tier gold market was established, whereby central banks continued to trade gold among themselves at the official rate while the private sector traded at the market price (World Gold Council 2008, p. 1). During this time it became apparent the dollar was already overvalued, particularly given the significant increase in domestic spending resulting from President Lyndon Johnson’s ‘Great Society’ programmes. The cost of the Vietnam War to the American exchequer worsened the situation further. “By early 1971, the US dollar liabilities exceeded 70 billion, backed by only 12 billion US dollars” (Hammes and Wills 2005). In the same year, the US President Richard Nixon declared a temporary suspension of the dollar-gold convertibility. The system was finally dissolved in 1973 (IMF 2012). After March 1973, the former Bretton Woods currencies were floating freely again (Stephey 2008).

After the Bretton Woods system was dissolved, the gold price was floating freely for the first time in 250 years (Fei and Adibe 2010, p. 1 and Oxford Economics 2011, p. 5). From then on, the gold price steadily increased from its starting point of 35 US dollars an ounce and quickly rose to a high of 127 US dollars on July 6th, 1973 (World Gold Council 2008, p. 2). The gold price kept rising and peaked temporarily (in USD) during the second oil crisis in 1980. After the collapse of the Shah’s regime and the subsequent withdrawal of Iran’s oil from the world market, gold hit a record high of 850 US dollars an ounce (Hamilton 2011, p.16). In the following years, gold prices oscillated according to various economic and political crises but did not reach its 1979 heights again until 2008. During that decade, the gold price drifted and flattened until a low of 251.70 US dollars was reached in 1999. In that year, 15 European central banks agreed on limiting gold sales, which boosted the gold price to a two-year high of 338 US dollars an ounce in October, 1999. Since then, the gold price has risen inexorably because of a number of factors, including the 2003 Iraq war, a weakening dollar, relatively high oil prices, political tensions over Iran’s nuclear ambitions and worries about contagion of debt problems in the Eurozone. In 2011, it reached a new peak above 1900 US dollars an ounce, after a 650% rally (Popper 2013).

Since March 1973, gold prices can be viewed in three distinct periods:

(1) A bull market since the introduction of a freely floating gold price until a temporary peak in 1980
(2) A period of a rather flat to slightly falling gold prices between 1981 and 2000
(3) An increasing gold price since 2001 until the end of 2011 (see Figure 1). By the end of 2011, the price of an ounce of gold was 1575 US dollars (Kollewe 2010).

illustration not visible in this excerpt

Figure 1: Average gold prices in US dollars from 1973 to 2011 (rounded). Sources: Data for 1973 to 2008 taken from World Gold Council 2008 and for 2008 to 2011 taken from Kitco.com

Researchers found that the price of gold is influenced by some factors that are easy to quantify such as currency exchange rates, inflation, the price of crude oil, the price of silver and the US dollar money supply. However, other factors are far more intangible, such as political risk, official sector activity and central bank gold reserve sales.

1.4 Research questions

In an attempt to analyse how identifiable independent variables influence the gold price, the following questions will be addressed:

The central research question is: How well can the gold price since the end of Bretton Woods be explained using a multivariate ARIMA model that includes the following independent variables: inflation, real interest rates, silver prices, the US dollar money supply (M2), oil prices, the MSCI World index and the S&P 500 (evaluation criteria: R-squared, mean absolute percentage error (MAPE) and BIC)?

To answer the research question, the author will discuss the following:

A. How effective is the model including these independent variables in explaining gold price variations in times of so-called ‘normality’ and in times of crises?

B. What is the explanation for the differences in variability described by the model during times of so-called ‘normality’ and times of crises?

The following literature review comprises two sections. The first discusses theories explaining asset price behaviour. The second covers studies that tested determinants (independent variables) that supposedly have a significant effect on the gold price. If an independent variable can be expected to increase the quality of an ARIMA to explain the gold price, it will be included in the analysis.

In the following chapter, the research will discuss data collection, the ARIMA model and its assumptions. It will then provide analysis, interpretation and contextualisation of the results. Given the nature of the questions, providing conclusive answers to Question B will prove particularly challenging because it requires interpretation and examination through the application of a theoretical framework.

The rest of this dissertation is organised as follows: Chapter 2 discusses the existing literature and theory on the topic with particular emphasis on the connection between this paper and the existing theoretical framework; Chapter 3 covers data and methods: ARIMA and the approach applied will be discussed in detail; Chapter 4 presents and discusses the findings of the analysis; Chapter 5 concludes by summarising and listing separate sections on the implications and limitations of this study as well as on how the results might be used for further research.

Chapter 2 – Literature Review and Theory

The aim of Chapter Two is to review the relevant literature and theory in order to develop a coherent theoretical framework. This involves the discussion of various theories to explain the variations in asset prices that cannot be explained by the identified variables. The second part of the chapter provides examples of identifiable independent variables that are related to gold and/or correlate significantly with the gold price.

2.1 Theoretical framework: Explaining the movements of the gold price

In 2005, Faugère and van Erlach (p. 99) wrote that “assessing the fair value of gold largely remains a mystery in finance”. No comprehensive theory of gold valuation exists, they wrote, that is able to show how factors like inflation, exchange rates or other asset classes influence its value. This finding holds true given there is no widely accepted theory that explains the price of precious metals. Instead, the efficiency of the gold market will be discussed: how the gold price moves in times of crises and investor psychology that helps explain the gold price in normal times and times of crises.

Eugene Fama’s Efficient Market Hypothesis (EMH) was long considered to be the best description of securities’ price movements and was widely accepted. The main concept of EMH is that markets are efficient, that prices are unpredictable and that prices of securities reflect all available information at any time (Fama 1970, p.383). This view is no longer dominant. Many securities professionals and academics now agree markets are irrational at times and anomalies and irrational investors are at the root of mania and panics (Dieupart-Ruel et al. 2013, p. 129 and Yalcin 2010, p. 24). Kindleberger and Aliber (2005, p. 38) say the assumption of the always rational investor as defined by the EMH is unrealistic, citing the frequency of speculative manias and irrational exuberance.

The common understanding today is that the EMH fails to recognise that psychology plays an important part in our investment decisions (Dreman 1998, p. 4). In this sense, the gold market seems to be no exception. Solt and Swanson (1981) looked at the price of gold from January 1971 until the end of the decade. They found positive autocorrelations, including considerable heteroscedasticity in the variance and that the means of the price change is non-zero and non-stationary (p. 470). Overall, they conclude, the results are not consistent with gold market efficiency (p. 476/477).

Well-known phenomena postulated to explain market anomalies and manias are ‘herding’ (group think) and excessive extrapolation, which is the tendency of investors to extrapolate recent positive news and price developments into the future without the fundamentals changing.

Positive price performance over a certain time and “excessive extrapolation” often lead to market participants giving too rosy market forecasts (Utkus 2011, p. 6). Baur and Glover (2011, p. 7) argue the rapid price increase of gold from 400 US dollars to 1600 US dollars an ounce between 2005 and 2011 cannot be explained solely by a change in the fundamental value of the metal. This explanation is based on the assumption that the gold price primarily rises if the expected inflation increases, given the fact that gold pays neither interest nor dividends. Therefore, the somewhat speculative investing of chartists based on extrapolating past price trends must drive the price upwards. Utkus (2011, p. 3/7) writes that if buying based on the positive price development of the recent past goes on long enough and an ever larger percentage of investors start to join in the group-think, it can lead to the development of a bubble: understood as a situation when the current price of an asset substantially differs from its intrinsic value.

Demand for gold is particularly high during times of crises as investors are looking for means to protect their assets. The main triggers of such demand include financial instability, the decisions of central banks (Dieupart-Ruel et al. 2013, p. 129), political tensions, wars or distrust in the policies and prospects of nations (Nadler 2006, p. 56). The latter is linked to the expectation of rising inflation – as experienced during the Euro crisis and in the current era of monetary inflation – which encourages investors to invest in gold as they try to preserve their wealth (Dempster and Artigas 2010, p. 69). Times of crises also cause investors to become uncertain regarding the capital markets, which triggers the purchase of gold as a substitute investment (Cohen and Qadan 2010, p. 43). The extent to which fear can drive stock prices down and at the same time increase demand for gold can be observed during the Eurozone crisis. On November 16th 2011, prominent commentator Matthew Lynn wrote on marketwatch.com: “Gold is the only winner from the Euro crisis” as the Euro falls and “equities have struggled to make any progress all year”. On June 25th 2012 the New York Times reported: “Wall Street drops on Euro pessimism” as media commentators and the public doubted whether Europe could solve its debt crisis. Dieupart-Ruel et al. (2013) explored the extent of investor rationality and availability of information when making decisions. They evaluated if what they termed ‘the cognitive biases’ – ’anchoring bias’ and ‘safe value bias’ – influenced gold investors. They analysed the gold price between Q4 2003 and the end of 2012 and assumed three classes of investors:

(1) Rational informed agents (RIA) who take into account the fundamentals
(2) Irrational informed agents (IIA) who are informed and are prone to cognitive biases
(3) Non-informed agents (NIA) who take their information from observing the market. NIAs are influenced by the propositions of informed agents and by behavioural biases (p. 129/130).

Dieupart-Ruel et al. looked at the difference between a gold price they calculated based on global demand, which they termed the “fundamental price” and the actual gold price. The difference between the two – they assumed – would be accounted for by the influence of investor biases (p. 132). By considering anchoring bias and safe haven bias in their calculations, the estimation of the actual gold price could be improved significantly. To test the anchoring bias the researchers evaluated whether investors under-reacted to analysts’ predictions but found the effect to be rather small; to test the safe value bias they used the volatility indicator VIX. VIX measures the volatility of the S&P 500 and is based on the calculation of the average of calls and puts on the S&P 500 (p.130). Their analysis shows the safe value bias is much more important when trying to explain the gold price than the anchoring bias (p. 132). Interestingly, Dieupart-Ruel et al. found the importance of the safe value bias differs between uncertain times and “normal” times – just as could be expected. From 2004 and the latest financial crisis of 2008, volatility of the S&P 500 was weak and so was the effect of the safe value bias on the gold price. The calculated price during this period hardly differed from the real gold price. But from the beginning of the financial crisis onwards, volatility increased and the influence of the safe heaven bias kept pace with it (p. 132).

Having discussed various theories dealing with (commodities) markets and investor psychology, the discussion will now turn to the quantitative variables connected with the gold price.

2.2 Empirical findings: Independent variables correlating with the gold price

Researchers have found a number of variables that correlate with the gold price. First, gold is said to protect in times of inflation and/or serve as a hedge against future inflation, measured by the Consumer Price Index or the Producer Price Index (Dimson, Marsh and Staunton 2012, S. 9). Inflation erodes cash values but gold is generally considered a safe haven as its value tends to increase during periods of inflation. Inflation might drive up the gold price because investors expect prices to go even higher and therefore see the purchase power of their dollars deteriorate (Koutsoyiannis 1983, p. 571). O’Connor and Lucey (2012, p. 16) argue that because gold is primarily traded in dollars, if the dollar weakens then gold becomes cheaper when paid for in a foreign currency. This drives up its demand. If this logic holds true, we can expect a positive correlation between inflation and the gold price: the higher the inflation, the higher the gold price.

Several researchers have found a positive correlation between (expected) inflation and the gold price. Ghosh et al. (2004) write that because the dollar constantly lost purchasing power between January 1982 and December 1999 while during the same period gold lost 59% of its value, it did not serve as an inflation hedge in the short run (p. 2). In their own analysis, however, the researchers found that gold can be regarded as an efficient long-run inflation hedge. They analysed the monthly average spot dollar price for an ounce of gold between January 1976 and December 1999 and used cointegration regression techniques to analyse the relationship between and gold and inflation. When testing for the cointegration of the Retail Price Index and the price of gold, the hypothesis that these two variables are cointegrated could not be rejected. In the short run, they concluded, fluctuations in the price of gold are based on short-run influences such as changes in the gold lease rate, the real interest rate, default risk and the exchange rate of the dollar (Ghosh et al. 2004, p. 1/9/10/18).

Capie et al. (2005) agree that gold serves as a hedge over the long-run. They analysed the weekly gold prices between 1971 and 2004 in relation to the Sterling-Dollar and the Yen-Dollar exchange rates. Although the gold price peaked in 1980 and started increasing again from 2001, the yen-dollar exchange rate decreased steadily (thanks to a strengthening yen) during the entire period. The Sterling-Dollar exchange rate peaked in 1985 and was subsequently relatively stable. In 2004 it was slightly higher than 34 years earlier (p. 347). The scatterplots of the logarithm of the gold price against the logarithms of the two exchange rates led them to conclude that gold was a hedge against the dollar, but with several caveats. The researchers concluded that gold served as a hedge mainly because it is a homogenous asset that can easily be traded and because it cannot be produced by the same authorities that have control over the currencies. The researchers see three possible reasons for the varying quality of the hedge:

(1) Actors expected exchange-rate fluctuations to be temporary and decided to ride them out rather than rearrange their portfolios.
(2) Private sector investors might have been influenced by problems in gold-producing countries, i.e. they expected these problems to influence the gold supply in the future.
(3) States may constantly change their attitudes towards their gold holdings.

Given these insecurities, the authors cautioned that while gold has served as a dollar hedge, its price development remains highly insecure as it is influenced by the actions of individuals as well as political attitudes and unpredictable events (p. 351/352). While the exposure to political attitudes and unpredictable events is certainly true, it is not unique to gold but is also true for many, especially multinational corporations as they, too, are influenced by political decisions and unpredictable events such as natural disasters and wars. However, as states have direct control over their gold reserves and influence various factors that are claimed to influence the gold price (interest rates, foreign policy, and money supply), the influence of states on the gold price can be expected to be greater – bar exceptions like the nationalisation of corporations – than on securities.

Worthington and Pahlavani (2006) analysed two data sets, comprised of monthly data of the gold price in US dollars per ounce and the monthly US inflation rate. The researchers analysed two subsamples encompassing January 1945 to February 2006 and from January 1973 to February 2006. The second subsample starts after the Bretton Woods’ System of fixed exchange rates was dismantled. They used unit root tests for their analysis and had to consider two structural breaks in the gold price due to the oil crises in January 1973 and in December 1978 and two structural breaks in the inflation rate in February 1973 and January 1979. They concluded that gold served as a useful inflation hedge between 1945 and 1973 as in the period following the dismantling of the fixed exchange rates. They said a “strong cointegrating relationship exists” between gold and inflation from 1945 to 2006 (p. 260/261). Joy (2011) agreed with Chua and Woodward (1982) that gold serves as an inflation hedge. Joy applied a multivariate GARCH model using weekly data of the gold price and 16 exchange-rate pairings and found that gold served as an inflation hedge from 1986 to 2008 (p. 124/129); Chua and Woodward (1982) analysed whether gold served as an inflation hedge against the currencies of Canada, Germany, Japan, Switzerland, the United Kingdom and the United States between 1975 and 1980. They collected monthly data comprising gold prices and domestic consumer price indices (CPI).Inflation was computed by calculating the percentage change in the CPI. A simple regression model was used. The return from gold was positive solely for the US – and the US dollar – and the result significant, based on the data collected.

Not all researchers agree on the strong connection between inflation and the price of gold. Lawrence (2003, p. 2) denies any statistical correlation between the gold price and inflation (as well as between gold and GDP and gold and interest rates). His research is based on a time series analysis using quarterly data from January 1975 to December 2001. However, as his study was undertaken under the patronage of the World Gold Council it has to be taken with a pinch of salt. It must be expected that the sponsor is interested in positioning its “product” as attractively as possible and as a “safe haven”. Fisher (2011), a chartist, has also challenged the significant correlation of gold and the inflation rate. He concluded there is only a weak correlation, if any, between inflation and the gold price, claiming that the price of gold rises independently from inflation. He gives three indicators to underline his claim: From 1976 until January 1980, the gold price rose 523% while inflation increased 167%; between 1980 and April 2001 gold decreased 67% while the consumer price index advanced 226%; in the next bull market from 2001 until February 2011 the gold price went up 530%. In the same period the inflation rate was 125%.

Second, inflation in US dollar terms is very closely related to the US dollar money supply (M2), which is why – if the inflation rate is strongly connected to the gold price – it can be expected that the US dollar money supply (M2) is also strongly interrelated with the gold price. In November 2013, as the Federal Reserve was still buying 85 billion dollars’ worth of Treasuries and mortgage-bonds each month, thus steadily expanding the monetary base (the sum of US currency in circulation and bank reserves), so-called experts argued whether the expansion in the monetary base would someday inevitably cause inflation or not. While the Federal Reserve is convinced everything is under control as the “quantity of currency in circulation is entirely determined by demand from people and businesses” (Williams 2012), others like Huberts (2013, p. 3) argue that as soon as the velocity of money increases, so will inflation. As inflation is said to have an influence on the gold price, it is not surprising that many researchers investigated whether the US dollar supply influences the gold price. The logic behind this reasoning is thus: The US dollar is the most important reserve currency; many individuals and institutions are invested in US dollars; gold is traded against the US dollar. If the US dollar weakens, US dollar holders lose money. The weaker the US dollar, the bigger is the incentive to invest in another “reserve currency” – such as gold: If the dollar weakens, the demand for gold rises (Fei and Adibe 2010, p. 25).

Tully and Lucey (2006) agree on the relationship of the US dollar and gold and write that “gold appears to be the anti-dollar” (p. 317). Pukthuangthong and Roll (2011, p. 2070) share the same view but claim that the inverse relationship between a currency and the gold price holds not only for the US dollar but for any currency. In their own analysis, Tully and Lucey used a generalised autoregressive conditional heteroskedasticity model (GARCH) to investigate the macroeconomic influences on the gold price for the period 1983-2003. They researched such macroeconomic factors as the US dollar supply, the Pound Sterling supply, the British stock index FTSE 100, the UK consumer price index and US interest rates. The researchers came to the conclusion that among the variables considered, the US dollar money supply had the biggest significant impact on the gold price (p. 322/323).

Ismail, Yahya and Shabri (2009) found that the US dollar money supply (M1) was positively correlated with the gold price. The researchers developed a multiple linear regression model and used independent variables such as the inflation rate, the Commodity Research Bureau future price index, the US Dollar/Euro exchange rate, US dollar money supply (M1) and the NYSE and S&P’s 500 stock indices, employed SPSS and took the mean square error as the measure for the quality of the model’s forecast accuracy. Around 70% of the variance could be explained by a model using the variables that significantly influence the Commodity Research Bureau future price index, US Dollar/Euro exchange rates, the inflation rate and the US dollar money supply (M1). “M2 contains M1 plus certain other financial assets” such as savings, small denomination time deposits at all depository institutions, mutual funds, overnight Eurodollars and overnight repurchase agreements at commercial banks (Batten and Thornton 1983, p. 40). Given that M1 and M2 are very closely linked, research results that are found for M1 can also be expected to be valid for M2.

An interesting analysis on the connection between the US dollar money supply and the gold price comes from Artigas (2010). Using a multiple regression model, he analysed the correlations between the independent variables’ year-on-year growth in money supply (M1) of the US dollar, the Euro, the Indian Rupee and the Turkish Lira and the dependent variable year-on-year percentage changes in the price of gold for a given month. He found that an increase in the money supply of the US dollar does increase the gold price as much as an increase in the money supply of the other currencies. This confirms the findings of Pukthuangthong and Roll that increases in the money supply of other currencies have an effect on the gold price, too. According to Artigas, the highest correlation between supply increase and an effect on the gold price can be witnessed six months later (p. 8).

Third, the demand for gold is also influenced by the opportunity costs of capital. The higher the interest rates on government bonds and in bank accounts, the higher the opportunity cost of holding gold that pays no interest (Oxford Economics 2011, p. 7). If the nominal interest rate is lower than inflation, the real interest rate is negative. In such a situation gold is attractive for investors and demand is high, some researchers say. The inverse relationship between the real interest rate and the gold price seems to be widely confirmed by the findings of researchers and market analysts: Mickey (2009), Chief Investment Strategist of “Q1 Publishing”, an investment newsletter, calls the real interest rate the “main driver for gold prices”. Mitra (2011), another market analyst from Axis Bank, claims the existence of a negative relationship between real interest rates and the gold price: the more negative the real rate of interest, the higher the gold price. He found this tendency to be true for India, the US, Japan and China from 1998 through 2008. Academics like Barsky and Summers (1988) agree with this. They used an ARIMA model to analyse the relationship between the real interest rate in the US and the gold price between 1973 and 1984 and found a strong, significant correlation (pp. 543-545).

Fourth, several researchers and market analysts claim a positive correlation between gold and silver. According to the World Gold Council (2011), gold and silver showed a correlation of +0.67 between January 1991 and December 2010. Baur and Tran 2012 (p. 2) wrote that “gold and silver were substitutes for thousands of years suggesting that there is a long-run relationship between the two precious metals”. However, they also mention other factors that uncouple the prices of gold and silver from each other such as the industrial uses for silver and the use of gold for jewellery as well as central bank demand. Klapwijk (2011) writes that silver benefits from the attraction of gold as “for some, silver is a more economical alternative to gold”. The analysis of Tully and Lucey (2006), mentioned already on page 14, also discussed the gold-silver relationship from 1978 to 2002 and found that while the positive correlation between gold and silver holds in the long run, the relationship is weak or even broken in certain periods, in particular during the 1990s when it had been unstable.

Fifth, gold and the price of oil are also said to have a positive relationship and both tend to increase in price whenever there is a global (political) crisis, when there are tensions between nations or war breaks out. The correlation of oil and gold prices during the last 40 years was around 85% (Shafiee and Topal 2010, p. 180/181); Laidi (2008, p. 42) wrote that since 1972 the gold-oil-relationship has remained generally robust. The strong relationship between oil and gold is also confirmed by Simakova (2011), who analysed the relationship between oil and gold prices for the period from 1970 to 2010 and undertook a simple correlation analysis by using monthly data. The researcher confirmed the strong positive correlation of oil and gold over the entire period. However, during the financial crisis of 2008 the price of gold rose steeply while the price of oil fell along with the stock market (p. 656/657).

Le and Chang (2011) use seasonally adjusted monthly averages of oil and gold prices as well as inflation data from January 1986 to April 2011 to ascertain whether a rise in the oil price leads to a rise in the gold price. They try to answer this question by testing the following two hypotheses: A rise in the oil price generates inflation; inflation leads to a rise in the gold price. They find co-integrating, long-term relationships between the oil price and inflation and also between inflation and the gold price. A Granger causality analysis supports the suggested causality of oil and gold prices. They conclude that the oil price can be used to predict the gold price (pp. 13-19).

Sixth, to test how the gold price reacts in a crisis – assuming that the stock markets take a dip in a crisis – the MSCI World Standard (Large and mid-caps) and the Standard &Poor’s 500 (S&P 500) are included in the analysis, too. The MSCI World is a “common benchmark for global equity portfolios” that measures the development of equities’ markets of the developed world (Aon Hewitt 2012); the S&P 500 holds 500 leading US companies and covers roughly 75% of US equities (Federal Reserve Bank of St. Louis 2013). For the MSCI World as well as for the S&P 500 analysts and academics found that the correlation is sometimes positive and at other times negative. The correlation of the MSCI with the gold price fluctuated between 2002 and 2011 between -0.5 and +0.7 (Hindecapital 2012, p. 4). Compared to the previously discussed predictive variables, these two stock indices seem to be of lesser predictive quality. Other researchers that evaluated the relationship of the S&P 500 and the gold price found weak correlations. Duller and Barbee (2012, p. 3) found a positive, weak correlation of 0.12 between the S&P 500 and the gold price over the years 2007-2012 and an even feebler correlation of -0.06 between 1982-2012. Gault (2012, p. 18), however, used monthly data and looked at the correlation of the gold price and the S&P 500 and found a correlation of 0.313 from 1990 to the end of September 2011.

Last but not least, this study also considered the inclusion of the gold supply in the analysis but it was not possible to gather the required monthly data. Only yearly data is available, which is why gold supply was not included in the analysis. However, its effects are intuitive because like in any free market, the price of gold is an equilibrium price set by supply and demand. While demand is influenced by macroeconomic conditions, politics and special events such as a large-scale war (Bapna et al. 2012, p. 1), its supply increases constantly because of production and the fact it is non-perishable. “Unlike wheat, say, where most of the current supply comes from this year's crop, gold is storable and most of the supply comes from past production accumulated over centuries” (Haubrich 1998, p. 1). Shafiee and Topal (2010) claim that in the long-term, a reduction in gold production was one of three factors that contributed to a rise in the gold price, the other two being purchases from institutional and retail investors in uncertain times (“insurance”) and the facilitation of gold purchases through Exchange Traded Funds. However, Abken (1980, p. 12) claims supply is “relatively insignificant” when it comes to the price of gold because its annual production is dwarfed by the total amount of gold already on the market. Given the small academic evidence for a significant correlation of gold production and price and the fact only a small part of the annual gold on offer is actually newly produced, it would be no surprise if the gold price was not strongly influenced by variations in its production.

[...]

Excerpt out of 75 pages

Details

Title
Explaining the gold price after the Bretton Woods Agreement using independent variables. An ARIMA model approach
College
University of Leicester  (Center of Management)
Grade
1.7
Author
Year
2014
Pages
75
Catalog Number
V304522
ISBN (eBook)
9783668030930
ISBN (Book)
9783668030947
File size
783 KB
Language
English
Tags
explaining, bretton, woods, agreement, arima
Quote paper
Stefan Heini (Author), 2014, Explaining the gold price after the Bretton Woods Agreement using independent variables. An ARIMA model approach, Munich, GRIN Verlag, https://www.grin.com/document/304522

Comments

  • No comments yet.
Read the ebook
Title: Explaining the gold price after the Bretton Woods Agreement using independent variables. An ARIMA model approach


Upload papers

Your term paper / thesis:

- Publication as eBook and book
- High royalties for the sales
- Completely free - with ISBN
- It only takes five minutes
- Every paper finds readers

Publish now - it's free