The question is then – does who buys the gold matter for its price? Do physical buyers and investor purchases have different effects on the gold price?
It is possible that the traditional buyers of gold, who kept buying through the period of falling prices, provided little or no support for gold prices. Their purchases are normally physical which implies that they are not planning to sell in the short term as there is a big transaction cost to trading in your jewellery or coins. Their purchases then seem to be a medium to long term withdrawal of liquidity from the gold market. A fall in liquidity should, all being equal, lead to an increase in the risk of holding gold. This would drive the price down to increase the returns from holding gold for prospective buyers.
The recently rising gold price has been associated with increases in ETF holdings. After 3 years of large outflows q1 2016 had a 363 tonne inflow, data from the WGC again. The largest annual inflows were in 2009 and 10 at 644 and 420 tones, so 2016 could be the biggest year for ETFs to date if that then holds. These purchasers are buying gold in a way that increases it’s liquidity and it’s possible that this is part of the reason for the price rise. Their reasons for buying are assumed to be related to golds safe have status. A summary of the research on gold ETFs is available here.
The effect of changes in physical demand on gold is under researched by academics. 3 of the authors of this blog are currently working on a paper and when we have something I will post it here. But it’s possible based on the above ideas that physical demand is not a price support in the same way as paper demand.
Both sets of tests point to the existence of periods where a bubble was present in the gold price, especially when variants of the bubbles tests that have been shown to be more powerful are utilised. Bubbles are shown to be frequent, if short lived, with only the mid 00’s lacking a bubble period for a few consecutive years.
A criticism of the use of lease rates as a fundamental determinant of golds value is available here on the blog.
Methodology: Markov-switching ADF tests
Data: Daily Am and PM Fixings and Gold lease Rates from 17th of July 1989 up to the 31st of July 2013
Abstract: We assess whether two classes of bubbles occur in the spot price of gold, rational speculative and periodically bursting bubbles, using gold’s lease rates for the first time in the literature as a measure of its fundamental value. This question is of particular significance as these are the only observable market measures of a yield that can be earned from gold. We use unit root and cointegration tests to look for rational speculative bubbles and Markov Switching Augmented Dickey-Fuller tests for periodically bursting bubbles. ADF and cointegration tests point to a rational speculative bubble. The more theoretically valid Markov Switching ADF test gives mixed evidence. No bubble is found to be present if we allow the variance to switch between regimes, the gold and its lease rate relationship is instead characterised by high and low variance periods. Imposing a constant variance gives evidence of a bubble for the 2, 3 and 12 month lease rates, but no bubble when we use the 1 and 6 month rates as determinants
A fundamentalist Central Banker would only be prepared to pay for gold what they believed they will receive from it in future. Gold’s lacks an available yield for an average investor which is why Warrant Buffet says he doesn’t invest in gold (“Gold does nothing but look at you”. But previous posts have discussed the idea that gold does have yield for central banks through the gold leasing market, a topic I wrote a paper on here with Prof. Brian Lucey. This post will uses lease rates and discount factors to estimate fundamental values for gold.
I will use 12 month lease rates as shorter lease rates frequently went negative over the past number of years and we will assume that central banks are long term investors in gold. Below are the annual dollar cashflows you could have earned by leasing gold. Following a period of unusually low leasing cashflows before the financial crisis there has been a recovery due, primarily to higher gold prices. On average you would have earned $4.72 from leasing one ounce of gold from 1989 to 2015. Pre-2000 the average was higher at $5.73 and post-2009 it averaged $5.95 per ounce.
A fundamentalist central bank would assess what they think gold is worth based on the annual cashflow they could receive divided by the relevant opportunity cost of capital. This blog estimated that gold’s Opportunity cost of Capital would be 0.25% if we assume it is a risk free investment, 1.25% if we assume that the central bank’s investment is as risky as an interbank loan (12 month LIBOR) and 0.98% if we treat gold’s alternatives through the Capital Asset Pricing Model. If we divide the above cash flows we get the below range of answers.
Discount Factors |
Leasing Cashflows | Average | ||
$4.72 |
$5.73 |
$5.95 |
||
0.25% | $1,888.00 | $2,292.00 | $2,380.00 | $2,186.67 |
0.98% | $481.63 | $584.69 | $607.14 | $557.82 |
1.25% | $377.60 | $458.40 | $476.00 | $437.33 |
Average | $915.74 | $1,111.70 | $1,154.38 |
$1,060.61 |
The table shows the various estimates based on the assumptions. If the central bank assumes that the average return is biased downwards by the slump in leasing cashflows in the mid 2000’s then they can take one of the higher estimates as their guess about future cashflows. The lower the risk the central bank perceives in leasing their gold the lower the discount factor they would choose.
This shows a very wide range of estimates from $377 to $2,380. The truth probably lies somewhere between these points as gold leasing cannot be as low risk as a three month loan to the US government. Assuming that cashflows stayed at current highs this implies that Central banks should be willing to pay a price of about $1,154.
The biggest problem with this analysis is that central banks do not buy gold so that they can lease it for yield. They only lease a tiny proportion of their combined gold holdings and anecdotally many central banks are not active in the lease rate market at all. Also central banks are not profit maximizing investors chasing yield as the above model assumes.
But since all attempts to value gold are at best guesses then this is as valid as any.
Finance 101 courses teach that the Opportunity Cost of Capital is the best alternative investment opportunity foregone, at a similar level of risk. So for a central bank if we will simplify and say they are investing in gold for leasing cashflows and capital appreciation, what investments are available at a similar level of risk?
First let’s look at alternative investments in bonds.
Some writers say that as gold can’t default it is a risk free investment. If we accepted this we would use a short term government interest rate, like the 3 or 6 month UST-bill rate. This perspective is based the fact on gold will always be gold no matter what happens to the economy. But here we are concerned with the value of gold in an investor’s portfolio and a brief look at a gold price chat over the last few years it should be clear that owning gold in your portfolio it is not risk free, $1800 gold wasn’t that long ago.
Gold is default risk free for average investors but only because it doesn’t promise to pay anything. Not too big of a claim to make. We can use these risk free rates to value gold but because they relate to a lower risk investment they will bias the results towards over valuation. Current 3m US T-Bills yield just over 0.25%.
LBMA documents indicate that the gold leasing market has never suffered a default so we can think of the probability of defaulting on a leasing payment as being very low, similar to the likelihood that major banks would default on loans to each other. So the London Interbank Offer Rate (LIBOR), despite its well publicised failings, seems the best candidate for an Opportunity Cost of Capital for gold if we think its like a bond of some sort. Todays 6 month Dollar LIBOR is around 0.75% and 12 month is just over 1.25%.
An issue with using LIBOR is that when there is a significant increase in the risk of an interbank default the rate rises and this would push the price of gold down in a discounted cashflow model – precisely when gold value would be expected to increase due to its safe haven characteristics..
We can also look at alternative investments in equities.
One way to calculate the opportunity cost of holding a stock is to get it’s Beta – it’s relationship with a market index calculated from the Capital Asset Pricing Model (another a Finance 101 idea). I calculate gold’s beta at about -0.11 which means that a 10% increase in the S&P500 index is associated with a -1.1% change in the US Dollar gold price and vice versa. This negative relationship is what we would expect and is shown below between 1989 and 2015 on a monthly average basis.
The issue is that this beta implies that gold is about 1/10^{th} as risky an investment as a diversified portfolio of stocks which doesn’t seem entirely credible. Both have very similar standard deviations and the chart shows big variations in both. It also shows long periods where gold prices and the S&P500 moved together broadly – 2003 to 2008 for example, its possible that Beta is being influenced strongly by safe haven periods where gold holds it’s value while stocks fall.
Beta, to me, seems a better indication of gold’s ability to diversify a portfolio rather than calculate a cost of capital to use in a valuation. But if we did accept it as a useful measure then the average annual return on the S&P500 over that period has been 6.9% so that the cost of capital for gold would be about 0.98%.
Both ways of measuring gold’s opportunity cost of capital are wrong in a pure sense as it isn’t really a bond or an equity. It doesn’t pay Par Value like a bond or a fixed coupon like a perpetuity. And unlike the company behind the share the likelihood that gold will stop existing in the future is not a significant one. But they do give an educated guess that the current cost of capital for gold is somewhere between 0.25% and 1.25% depending on the assumptions you make.
None of the models are able to forecast monthly gold prices over the long and the short run. Univariate models preformed better than the more complicated multivariate models developed in the paper. The most effective model was the Exponential Smoothing model. It reports the smallest forecast errors at 3, 4 5, 6, 7 and 8 months out. It outperforms the other models by betwee2n 22% and 84% based on the authors work.
It should be noted however that the best model to forecast the price next month in the Random Walk model. This means that over a short horizon gold prices are unpredictable.
The paper does not develop a trading rule to test the models. This would show if it is possible to make a profit trading gold based on any the forecasts in methods. Without this it is impossible to say whether this paper shows a way to beat the gold market.
It is also unclear why only monthly data is used as daily data is freely available for all of the metals used in the study.
Data: Monthly observations January 1972 – December 2013
Abstract:
This article seeks to evaluate the appropriateness of a variety of existing forecasting techniques (17 methods) at providing accurate and statistically significant forecasts for gold price. We report the results from the nine most competitive techniques. Special consideration is given to the ability of these techniques to provide forecasts which outperforms the random walk (RW) as we noticed that certain multivariate models (which included prices of silver, platinum, palladium and rhodium, besides gold) were also unable to outperform the RW in this case. Interestingly, the results show that none of the forecasting techniques are able to outperform the RW at horizons of 1 and 9 steps ahead, and on average, the exponential smoothing model is seen providing the best forecasts in terms of the lowest root mean squared error over the 24-month forecasting horizons. Moreover, we find that the univariate models used in this article are able to outperform the Bayesian autoregression and Bayesian vector autoregressive models, with exponential smoothing reporting statistically significant results in comparison with the former models, and classical autoregressive and the vector autoregressive models in most cases.
A zero coupon bond means an investor will receive a known par value at a known date in the future, so that if they hold it all the way to maturity the investor will know their nominal return in advance (assuming no default occurs). If you buy gold you are actually buying an asset much more like a zero coupon perpetuity. Just like a perpetuity the holder can sell their gold at any time but will be uncertain of the payoff when they do, as there is no promised par value.
For central banks, who can lease their gold more easily than ordinary investors, gold is different again. For them it is much more like a variable rate perpetuity. The graph below shows the difference between a buy and hold strategy; and a buy, lease and reinvest strategy for a central bank. The bank is assumed to own gold from 1989 when gold lease rate data starts and the chart shows the difference returns between the strategies if the bank were to sell their gold at any point after that. Long leases (12 months) were a better postion than a simple buy and hold over the whole period. Short term leasing (1 month) has been more variable and would have gone from profitable up until the turn of the century to mostly loss making since, owing to the fact that short gold lease rates are frequently negative .
But regardless of a central bank options, the fact that no one is promising to pay you back the principle in future when you buy gold makes it a very different asset to a zero coupon bond.
An increasing price of gold provides signals about the value of fiat money and can influence confidence in the financial and monetary system. Central banks therefore have an interest in gold price movements. This paper summarizes the main arguments for central bank gold price management, it replicates commonly used analyses to support the claims of price suppression and presents new statistical evidence. Conversely, we find no clear evidence of gold price suppression and provide explanations for apparent anomalies including liquidity, investor behaviour and the gold carry trade. Despite statistical evidence against central bank gold price manipulation, the opacity of the gold market and the role of gold as a thermometer for the value of fiat currency render it difficult to comprehensively dismiss claims of manipulation.
We begin with a review of how the gold markets operate, including the under researched leasing market; we proceed to examine research on physical gold demand and supply, gold mine economics and move onto analyses of gold as an investment. Additional sections provide context on gold market efficiency, the issue of gold market bubbles, gold’s relation to inflation and interest rates, and the very nascent literature on the behavioural aspects of gold.
A recent paper uses a new method, and suggests that New York, especially after market changes to make GLOBEX more effective, has become the dominant force. There is significant time variation however. The dominance is striking given the much larger volume of trading in London.
Reference: