The real price of gold is near the highest price it has been since the price was allowed to float in 1969. The last time gold was this high, the metal spent the next 12 years declining and did not hit bottom until it has lost 82% of its value in real terms. There is no really good reason to believe that the same exact pattern will be repeated this time, but we can be fairly certain that many of the factors that drove gold to the current heights are either unsustainable or that there exist other hedging instruments that are both cheaper and more effective. Gold might still go higher, but it is no longer a one way ticket. In the next decade, investors will need to hold a wider variety of alternative assets to hedge their traditional investments, and they will need to trade more frequently than in the past decade.
(click to enlarge)
Unlike other assets, gold has no intrinsic value. We can't estimate a fair value for gold the way we can for equities or bonds because there is no income derived from it, but the metal's price gyrations can still be quantified and explained by a number of macro-economic factors. Oxford Economics produced an interesting multi-factor model that is useful for discussion. I don't particularly like the model for forecasting because I think it includes a lot of factors that are not independent of each other, which introduces serious distortions, and it includes a lot of dummy variables that seem more convenient to the modeler than to the forecaster, but it is useful for starting the discussion because it includes practically everything. Details are available at Oxford Economics. The main drivers in their model include:
- Inflation
- The US Broad Exchange Rate
- Real Interest Rates
- Credit Default Premium
- Growth rate of the Fed Balance Sheet
I believe there are significantly better hedges for 1,2 and 4; that 2, 3, and possibly 1 are now significantly negative; and that 5 is entirely spurious. Let me explain.
Inflation. The presumed correlation to inflation is the most popular reason to buy gold, and yet, it's the most ridiculous reason of all. There are actually two reasons that this is ridiculous. The first is that it isn't even remotely true, and the second is that you don't need to hedge against inflation if you own stocks. Everyone who has ever studied the price of gold for a living knows that it is not correlated to inflation. Just look at the chart below, showing the relationship of changes in the price of gold to the changes in consumer prices. The correlation is effectively zero. Gold advocates have come up with a large number of ridiculous explanations for this. The most common is that even though gold is not correlated to inflation in the short-term, gold tends to hold its value over long periods of time. If you are a risk manager trying to explain that one to your boss in a multi-billion dollar hedge-fund, good luck keeping your job. If you need to hedge inflation, say in 2008, it isn't going to help you if gold holds its value in 2012. If something is sometimes correlated and sometimes not, then it is actually NOT correlated. It does not qualify as a hedge.
More importantly, there's nothing special about an asset holding its value against inflation. Anything that isn't perishable will increase in value against the dollar over time if the value of the dollar declines, which is what happens when there is inflation. A good investment should not just hold its value. It should increase in real value, which is what stocks do. If you own stocks, you don't need a hedge against inflation. Over the past 125 years, stocks have outperformed inflation by about 5% annually.
At any rate, Gold has achieved some of its greatest returns of all time in the past several years, at a time when both the actual inflation rate and the expected inflation rate have both moderated.
(click to enlarge)
The US Exchange Rate. This is not much different than the previous except that there have been numerous occasions in history when the value of the dollar has depreciated against a basket of other currencies without resulting in any subsequent bouts of inflation. In such cases, gold has always performed well against the dollar, but so have stocks and other currencies. US companies, especially large-cap stocks in the S&P 500 earn a substantial share of their profits from overseas operations, and any time the dollar weakens, not only do US companies become more competitive, but the earnings they generate overseas become more valuable in dollar terms. So although gold does well in these periods, you don't really need it. And even if you did, currency ETFs are now widely available and many are very liquid and nothing is more highly correlated to currency moves than currencies. My currency model currently suggests that the dollar is actually likely to be the currency of choice in 2012, which can be expressed with positions in UUP. The beauty of buying and selling currencies, is that it is a direct play on the currency itself, and can move independently of Gold and Stocks and Bonds.
(click to enlarge)
Real Interest Rates. In my view, this is the biggest problem for gold. As Warren Buffett has quipped, we dig gold out of the ground from some place in Africa and then we dig another hole somewhere else, burry it and pay a lot of people to stand around guarding it. It costs money to store gold, so there is a real opportunity cost to holding it, and on a couple of occasions, significant changes in real interest rates have coincided with very significant changes in the real price of gold. The usual caveat is required here: sometimes it works and sometimes it doesn't, but this appears to be the most plausible explanation for the doubling of the real price of gold in the past 3 years. Since 2008, the real interest rate on 10 year treasuries has dropped from nearly 6% to a negative 2%. The problem is that negative real interest rates have never been sustained for any significant length of time - they are always resolved in one of only two ways. Either inflation collapses, or nominal interest rates rebound. Neither of these outcomes can be expected to have a positive influence on the price of gold.
(click to enlarge)
Credit Default Premium. The spread between AAA and BBB bonds has had a significant correlation between the real price of gold, but there are several very serious problems with this. The first is that, as a rule, we don't own corporate bonds, so we usually couldn't care less what the default premium is. Corporate bonds don't protect us from deflation, they don't give us any growth, and occasionally they blow up. I understand that other people own them for other reasons and that's fine, but we really haven't ever had any good use for them. The credit spread has occasionally had a severe impact on stocks, and this is important to us, but unfortunately, during the two most important episodes of default risk contagion, during the 2000-2002 and 2007-08 bear markets, gold did not provide very good protection. Even when gold has seemed to be somewhat correlated to the default spread, the fact that gold consistently lead the default spread is concerning. You can't use the default spread to predict the real price of gold at any rate, and during both of these periods, there were significantly better hedges available. Whenever credit contagion is a global phenomenon, the dollar has been strong, and in part, this is exactly why gold hasn't worked as a hedge in these circumstances. Unfortunately, the currency trade is more of a trade than a hedge because it hasn't been volatile enough to compensate for the losses in stocks.
(click to enlarge)
Fed Balance Sheet. Milton Friedman is solely responsible for one of the most common myths in the market place today. His original explanation of inflation, namely that it is "always and everywhere a monetary phenomenon in the sense that it is and can be produced only by a more rapid increase in the quantity of money than in output," is so simple and self-evident that no one ever questions it, but unfortunately, it is simply not true. During the 1820s and 1830s, the money supply trebled in 15 years, but the CPI rose less than 1% annually throughout the period. Similarly, large swings in the money supply in the 1840s and 50s caused almost no change in general price levels. Similar examples have been recorded throughout the world in the record of historical prices of the past 800 years. Unless the newly produced money finds itself into circulation, it does not and cannot influence prices. In the past 4 years, the fed balance sheet has expanded 1373%, but there has been no meaningful change in the rate of inflation.
Gold bulls love to borrow the techniques of apocalyptic cults. Every time the predicted day of reckoning fails to materialize, they reschedule it. The argument goes something like: "If an x-fold expansion of the money base hasn't caused inflation, then surely some of that money will eventually find its way into the market and inflation will only be delayed." History has shown that there's no certainty in this statement at all.
The Fed's balance sheet appears to have had some correlation with the real price of gold between 1978 and 2004, but it is easier to ascribe this correlation to pure chance than to find a causal rational for this behavior, and at any rate, the model falters badly thereafter. If this were the only factor effecting the real price of gold, then gold would today be 8 times its current price. I am entirely uncomfortable with models that fail by this magnitude for even a short period of time, let alone 4 years running. It is entirely possible to explain the price of gold without this factor and probably best that we do.
(click to enlarge)
The Ultimate Utility of Gold
Warren Buffett recently stirred up a hornet's nest by pointing out that gold has no utility. Actually, he's been saying this at least since 1998, when he explained to an audience at the Harvard Club, "It gets dug out of the ground in Africa, or someplace. Then we melt it down, dig another hole, bury it again and pay people to stand around guarding it. It has no utility. Anyone watching from Mars would be scratching their head."
Since that statement was made, Gold has risen some 470%, while stocks rose only 22% with dividends reinvested, but even without the benefit of hindsight, it seems that Buffett is just wrong about gold. The ultimate utility of the metal is for investors in stocks and bonds because gold is correlated to neither, as the two charts below demonstrate. Any asset that is so uncorrelated to either of these two major asset classes has tremendous utility. Needless to say, however, gold is more useful as a hedge when it is both uncorrelated and also rising in real terms. Investors found cold comfort from gold's non-correlation between 1978 and 2000.
Since I do not currently own US Equities or bonds in any of my current portfolios, there is no obvious reason to own gold either. The question is, should I go short? Given my views that the US dollar will enter an appreciation phase against other currencies, that Europe is in a recession and the US is on the brink of one, that as a consequence of these, the probability of significant rises in consumer prices from here is very low, and at any rate, that the current real yield on government bonds is unsustainable, the only thing standing in the way of a decline in Gold prices would be the likely rise in credit spreads.
Even if we are correct in all of these assumptions, the market can still be annoyingly uncooperative for extraordinarily long periods of time. We always try to combine the best effort assessment of fundamental outlook with a rigorously back-tested technical model.
Our gold model is currently bearish. This model was developed using data between 2005 and 2010. We then tested the model on out-of-sample years 2010-2012 as well as out-of-sample years of 1970 to 2005. The total return over the entire period for gold was pretty impressive at 3,790%, but the model return was 96,842%!
(click to enlarge)
Most back-tested models that appear in public do not use such extensive out-of-sample tests, so they are often over-fitted for the given period tested, and this is why models can be so much less reliable when real money is applied than when they are tested. Because of the extremely long and multiple-periods of out-of-sample tests, there is very little possibility that this model is over-fitted to any unusual period of time. The model is also extremely simple, involving only one type of signal and is always either long or short. It has captured most of the gains from both of the bull markets in gold that have occurred in the last 50 years and has gained from most of the bear markets as well.
The only way the timing model can outperform a rising market is to use leverage. If you do not use leverage, then the model can only add value when it correctly forecasts bear markets because during correctly forecast bull markets, the returns will be identical to buy and hold. Buy and hold is of course superior during an incorrectly forecast period. If you are going to go to the effort to time a market, it makes no sense to handicap yourself this way, so we always use leverage in any timing model. It is unnecessary to use leverage on the short side, but we often do anyway.
The model has worked because gold has had a tendency to move in one direction for very long periods of time. If gold prices were for some reason to become more choppy, there is a risk that the model would go into a phase of continuous whipsaws without ever making any money. There is no way to be certain that this will not happen, but even if it does, the risk of loss is still fairly small. The Model went short on April 9, 2012 when the price of GLD was 159.72. The model is always long or short, never neutral, so we do not have to worry much about a run-away bull market in gold. My current estimate is that the model would turn positive if GLD were to rise to somewhere above $163, which would count as a 2% loss.
