An Intro. to Gold Supply & Demand Analysis, Part 1

In Three Tools to Understand The Price of Gold, I mentioned the WGC’s Gold Supply and Demand Report. At that time, I believed that I understood how that report could be used to explain movements in the price of gold.

However, I have learned much more since I wrote that post, and I now want to explain the report in more detail while at the same correcting some misleading statements I made in that post.

Argument #1: Gold is money, not a commodity.

One of the arguments often made against using actual supply and demand data to understand the price of gold is that gold is a currency, or “money”, not a commodity. Therefore, this supposedly means that we can’t understand its price the same way we would understand the price of oil, or orange juice, or live cattle, etc.

Of course, gold is money. It isn’t like coffee or sugar or other types of goods that depreciate over time.

But unlike fiat currencies, gold is a real, physical thing. A certain amount of it exists above-ground and a certain amount of it is added to the above-ground stock each year. This is the “supply” of gold. At the same time, a certain amount of gold is held in government vaults or private collections or is bought in the form of jewelry, technology, coins and bars, ETF holdings, etc. each year. This is the “demand” for gold.

So, because it is a physical thing that must be dug out of the ground, it can be explained using supply and demand analysis.

Argument #2: Gold is never consumed.

A second argument one often hears is that supply and demand analysis can’t apply to gold because it never wears away or needs to be replaced.

This is of course true. Once you own an ounce of gold, you own it forever. At least, as long as you don’t sell it, lose it, or have it stolen from you. It never gets “used” and needs to be replaced.

But even though you never have to replace the gold you “use”, you still have to buy more. Unlike any other commodity, the desire of human beings for gold is never, ever satiated. Humanity’s store of gold is currently the largest it has ever been since the beginning of time. And yet, gold miners still have to dig up over 3,000 tonnes of it a year because current owners of it will not sell.

So the fact that gold never has to be replaced does not invalidate supply and demand analysis. It implies that gold’s supply is always growing, yes, but it’s demand is always growing as well. So it implies that what we will be paying attention to is the growth in supply vs. the growth in demand, not that we won’t be paying attention to supply and demand at all.

Argument #3: The above-ground gold stock is 80 times the size of annual mining output, rendering annual supply and demand statistics trivial in size.

A final argument often made against supply and demand analysis for gold is that annual mining output stats are trivial in size. This is presumably because the stocks-to-flows ratio for gold is 80+ years. The vast majority of the gold supply has already been mined. So, it is argued, mining output can’t affect the price very much.

The problem with this argument is that the demand for gold is also 80+ years larger than current annual demand. So if we add up the demand for gold over the past 6,000+ years and compare it to the supply of gold amassed over the past 6,000+ years, we get exactly zero deficit or surplus.

All of the gold ever mined in history is already owned by someone. And they aren’t selling. Or, to be more precise, some of them are selling and others are buying…but humanity as a whole is a net buyer.

For this reason, the annual supply and demand of gold is essentially the only thing that matters. Or, to put it another way (and to repeat what was said above) – what matters is the growth in the above-ground demand for gold and the growth in supply, not supply and demand itself.

Gold Supply And Demand.

Now that we’ve dispensed with these arguments, let’s see what we can learn about gold from supply and demand data.

The charts below were created using data from the Demand Trends Report of the World Gold Council.

However, I’ve made a few changes to it. First, I’ve added together demand for gold jewelry, dentistry, technology and small (retail) bullion bars and coins, and subtracted recycling of scrap gold to get a “physical demand” category. I’ve also subtracted scrap gold recycling from the “monetary demand deficit”, which I’ll explain in more detail below…but which is essentially jewelry, dentistry, and technology demand (or physical demand minus bars and coins).

Here is a chart showing the various categories of demand for gold products from 2002-2016. The full year of 2016 is estimated by averaging the numbers from the first three quarters and adding it to the total of these three quarters, since the fourth quarter data has not been released yet.


Remember once again that what we are looking at is the growth of demand and supply of gold. I’m going to be referring to this as the “demand” and “supply” of gold for the rest of this article. But it should always be in the back of your mind that I am talking about growth rather than absolute values.

We can learn a lot of interesting things about the gold market from this chart. For example, this chart shows that central banks were net sellers of up to 600 tons in 2002, but gradually became buyers over the next 14 years. We can also see that physical demand was above supply in the first few years of the 2000’s and that ETFs became sellers instead of buyers in 2013, but started buying again in 2016.

As interesting as this chart is though, it doesn’t tell us what we really want to know: What causes the price of gold to rise or fall?

This chart shows that supply was greater than total demand in 2005 and 2006, yet a glance at the price of gold for that time shows clearly that it was rising rapidly.

Total demand was also lower than supply in 2009. And yet, the price rose rapidly during this time period.

For reference, here is a chart with just total demand and supply compared to the price of gold.


As you can see, there seems to be no correlation at all between oversupply or under-supply of gold and a fall or rise in the price.

So what is going on here?

A Gold Warehouse?

In order to make sense of this data, we have to ask ourselves the question “who are the warehousers of gold?”

With a “normal” commodity (sugar, coffee, cattle, etc.), there is always a certain amount of the item that is not considered to be either “demand” or “supply” but is seen instead as “inventory”. When this inventory has increased, it means that there was too much of the commodity produced and not enough sold. When this inventory has decreased, it means that there was less produced than sold. Therefore, an increase in inventory usually means that the price fell and a decrease in inventory usually means that the price rose.

However, we do not see this happening with gold. The reason is because the “warehousers” are not really warehousers at all.

When gold is mined, it goes to refineries…and from there, to the bullion banks and exchanges. But the bullion banks and exchanges are not passive conduits through which gold flows. They do not merely hold gold until it can be sold. They have clients who themselves want to hold gold as a means of preserving their wealth. So the bullion banks really should be classified as “demand” rather than as a separate category.

And…if you look back at the history of the Gold Demand Trends Report, it is only recently that this category has been omitted from the report and treated like it is a “gap” between supply and demand rather than a part of demand itself. In the past, this “gap” between supply and demand was called “the balance”, “institutional investment” or “OTC and Derivatives”. So, until recently, there has always been some recognition that the bullion banks do hoard gold and affect its price.

Having made this recognition though, we are confronted with a problem. If we include bullion bank hoarding as part of “demand”, we will end up with the trivial conclusion that supply always equals demand. In other words, all of the gold that is dug up gets put somewhere. So the price of gold should never change. And yet it does, so there’s definitely something wrong with our theory.

So the question remains: Who are the warehousers of gold?

Is Non-monetary Gold A Warehouse?

Let’s consider what happens if we simply assume that holders of non-monetary gold (Jewelry, Technology, Dentistry) are the warehousers we are looking for.

In this case, “demand” for gold would simply be what the Demand Trends Report calls “investment demand” or what we might call “monetary demand” (central bank purchases, ETF inflows, bullion bar and coin sales, and bullion bank/exchange hoarding).

In this case, our chart would look like this (click to enlarge):


At the beginning of this chart, we see clearly that there was a large gap between supply and “total investment demand” in 2002. Over the course of 2002-2012, this gap shrank. Then, beginning in 2013, the gap widened again.

This roughly corresponds to the increase in price from 2002-2012 and the fall in price from 2013-2015.

For reference, here is a chart of just “total investment demand” and supply compared to the price of gold.


The only significant anomaly in this chart is 2016. In 2016, investment demand spiked up while the price rose only slightly. However, this anomaly would probably disappear if we had updated data from Q4 2016 – because I would expect that investment demand fell during that time period.

To make this even more clear, here is a chart of what I call the “monetary demand deficit” for gold compared to its dollar price. The monetary demand deficit is annual mining supply minus total investment demand. In other words, it’s the amount of gold “left over” that investors (or “monetary savers”) didn’t want. Consequently, this gold got turned into non-monetary uses such as jewelry or dentistry.

These lines are messy. So I’ve also had the charting program create trend lines to show the average movement in values over time.


In 2002, there were over 2,000 tonnes of gold dug out of the ground that were not desired by monetary holders of gold. This gold got melted down and turned into non-monetary uses.

Over the course of the next 14 years, this number gradually shrank. By 2016, it was 50% lower than it had been in 2002. This happened despite the fact that mining output increased by around 50% over this time period.

In other words, despite miners producing another 1,000 tons of gold per year, savers of monetary gold are hoarding another 2,000 tonnes per year than they did in the past. Thus they are absorbing all of the extra output plus more. As a result, only around 1,000 tonnes of gold are available each year for non-monetary uses.

This is why the price of gold is increasing. Gold is gradually returning in the world to its true nature as money.

If this trend continues, we will eventually become a society where 100% of mining output goes to monetary uses and all gold jewelry will be made from scrap.

So are jewelry and technology buyers really just holding gold until it becomes desired as money again?

Not quite. If only the answer was that simple. But I don’t think it’s that simple.

In part II, I’ll explain why. I’ll also develop a microeconomic model, complete with walrasian supply and demand curves, to explain what determines the price of gold.

I use GoldMoney to buy my gold at half a percent over spot.