The following is an edited version of a speech made at the LBMA Precious Metals Conference in Edinburgh on 3rd November, 2009

As a central banker, I am not necessarily the most qualified person to speak about the public sector holdings of gold, for two reasons. First of all, because gold as an asset has been used since well before central banks were invented. The second reason is that we should of course not limit the public sector to the central banks. There are many other public sector entities that are holding gold. I will try to share some thoughts with you. One theme I think will appear in my presentation: despite the gold sales of the past decade gold remains, and will remain, an important asset for the central banks and, particularly, for the Eurosystem.

In my remarks today, I will concentrate on the Eurosystem and adopt a very conventional perspective: let’s talk about the past, then the present and, finally, the future. It is maybe a beaten track, but the topic does not really fit with audacious innovation.


Central banks and official international institutions have been major holders of gold for more than 300 years and are expected to continue to hold large stocks in the future. They currently hold about 20% of above-ground stocks. As I told you, gold has been used as a currency and as an asset since well before central banks were invented. Contrary to popular ideas, gold was not the first form of money used to avoid the difficulties and inconveniences of a barter economy. Many other goods have been used as money and some are quite exotic. I remember when I joined a central bank many years ago, I discovered that there were other forms of money than the traditional ones. For example, the shells used by the native Americans or whale’s teeth used in the Fiji Islands, or huge stone discs in the Pacific, or even tobacco. In Virginia in the US, tobacco had been the currency for 200 years – for a much longer period than the gold and the gold exchange standard put together. There were many forms of money, but the introduction of metallic money quite early contributed to overcoming some of the disadvantages associated with these forms of money: some were prone to decay, some were too heavy for transport, others were too easy to produce or reproduce. Among the metallic money, two forms were dominant: silver and gold. One remark aside: silver has probably been used on a much larger scale than gold as money.

The use of gold as money is reported before 1000 BC in China and around 700 BC in Lydia, introduced by the now forgotten king Gyges who was ruling the country, while one of his successors is better known: a certain Croesus. I don’t need to recall that the European exploration, to use a politically correct word, of the Americas was partly fuelled by the stories of the gold ornaments displayed by Native American peoples. You will notice, ‘en passant’, that gold played a role as money well before central banks where invented. But let’s not dwell too much on the history of money and allow me to come back to the central banks as gold holders.

It was only very late in history, in the last quarter of the 19th century, that central banks really started to accumulate gold as a reserve. This building-up of gold reserve was particularly important during the two world wars because of all the uncertainties in the world. It reached a peak in the 1960s and, at that moment, the official gold stocks reached a level of 38,000 tonnes, which accounted for about half of the above- ground stocks at that time. At that time, because of the fixed official price expressed in US dollars and its convertibility into dollars, gold was still considered to be the foundation of the international monetary system. Gold provided the ‘anchor’ to which all currencies of member countries in the system were linked, either directly, or via the dollar.

However, in contrast to the classical gold standard, at that moment, actually since the introduction of the gold exchange standard at the Bretton Woods conference, there was no longer a direct link between gold and the national money supplies. And as some central banks created more money than was consistent with stable prices, after several years of moderate but persistent inflation, the fixed official gold price became more and more unrealistic. Finally, in August 1971, US President Nixon suspended the convertibility of the US dollar into gold. The gold window was closed, and gold became one of the important reserve assets instead of being the centre of the system. That was the past.


So what do we do in the central banks with gold? As I said, it is no longer important from a pure monetary perspective, but it is important as an asset. Let me say a few words about the role of gold within the Eurosystem. The Eurosystem holds 10,800 tonnes of gold, while according to the World Gold Council, the total of official gold holdings amounts to a little bit less than 30,000. So the Eurosystem alone holds roughly one-third of all official gold reserves in the world. But this has evolved over the past 10 years. The share of gold in the Eurosystem balance sheet has decreased during the past 10 years, while the value of this investment actually increased due to the favourable price developments in the gold market.

When we started the ECB and the euro in 1999, the total balance sheet of the Eurosystem was €700 billion. In 11 years, it has increased three-fold. It reached a peak of €2.1 trillion at the end of 2008. During the same period of time, the share of gold in our portfolio diminished, although gold in our reserve doubled in value. It has now reached €220 billion. As the balance sheet increased even more quickly, the share of gold decreased from about 14-15% of the total balance sheet to 10%. This increase of the balance sheet of the Eurosystem has different sources. One is the evolution of banknotes in circulation. When we started with the euro, there were 370 billion banknotes in circulation. Now we have around 750-770 billion, more than double in 11 years, so automatically of course the balance sheet is bigger. A second element is that the reserve requirements we impose on banks have increased. Banks are obliged to maintain some deposits with the central banks and this is based on their short-term liabilities. As the balance sheets of the banks have increased quite dramatically over that period, automatically their deposits with the central banks have increased as well. The third reason for this big increase in the balance sheet of the Eurosystem of course has to do with those non-standard or non-conventional measures we have introduced in order to cope with the crisis that started on the 9th August 2007 and reached its peak immediately after the collapse of Lehmann Brothers.

Just to give you an order of magnitude, before the crisis, say around June 2007, the outstanding amount of lending we were providing the banking sector with was approximately €450 billion. We have reached, following the collapse of Lehmann, just less than €900 billion. The idea was to ensure that the banking sector didn’t collapse because of the absence of liquidity in the market simply because banks were reluctant to lend to each other. All these three elements, bank notes, reserve requirements and the non-standard measures explain this huge increase of the balance sheet of the Eurosystem from €700 billion in total to €2.1 trillion at the end of 2008. During 2009, it shrank slightly and now our balance sheet amounts to approximately €1,800 billion. This is because those exceptional measures are still present, but the total volume has been slightly reduced. Gold now represents again about 13-14% of our balance sheet. So we are back to the proportions we had in the past, although I repeat that the gold we have has doubled in value over the period. We now have €220 billion of gold in our portfolio.

Why do we have gold? There are voices that regularly say that it is not efficient, so why have it? Central banks, to justify their gold holdings, use different arguments and they don’t all emphasise the same points of view. There are four ideas behind those gold holdings: the economic security; the capacity to face unexpected needs; the question of confidence; and the risk diversification issue.

Economic security

Gold is a unique asset in many respects. In terms of physical and chemical properties, it is almost immune to decay, it has a very high density so that small volumes can represent large value, and it is a rather soft metal and can be easily manufactured. Also it is no one else’s liability. Therefore its value is not affected by potentially negative behaviour of the issuer. So this is a unique characteristic that is quite reassuring. In fact, gold has maintained its value in terms of real purchasing power in the long run, over history. Although, as regards the real purchasing power of gold, I hasten to add that, of course, past performance is no guarantee for future performance. We may see this next year in Berlin whether this is still true.

Unexpected needs

The second factor that is often used as an argument is the necessity to face unexpected needs. There are some circumstances that are extremely improbable and rare, but if they occur, they can be highly damaging. Those circumstances are, for example, wars or an unexpected surge of inflation (there are some countries that could witness this), or a generalised crisis that leads to the repudiation of foreign debts or even the international isolation of a country. Of course, those situations are rare, but we cannot say that they will never happen, and they are extremely difficult to anticipate or predict. In those circumstances, gold remains the ultimate and global means of payment that is still accepted and it is one of the reasons used by some central banks to justify gold holdings.

Gold can also be used as collateral for borrowing. For example, in 1974, the Banca d’Italia secured a loan that was granted by the Deutsche Bundesbank by providing gold as collateral. In 1991, India received a loan from the Bank of Japan that was also based on gold collateral.


The third element of justification is confidence. Of course, gold is no longer backing the currency in circulation. It was the case under the gold standard when gold holdings maintained the confidence in the otherwise worthless paper money, which was used for practical reasons only. Since the break-down of this system, gold lost its formal supporting role. However, people still instinctively trust gold, which is held for non- monetary purposes in both retail and institutional portfolios.

Under the gold standard, paper money had no value in itself but it was fully backed by gold, so its value was represented by the gold held by the central banks. This is no longer the case in modern central banking, but still one can argue that this notion of confidence is still there. What makes the value of money today is faith. You believe that the money you have is good. There is nothing else. Of course, this faith will be justified or not by the behaviour of the authorities and the central banks, but if there was a little bit of gold behind it, it could contribute to increase this faith in the currency. We are far away from the classic gold standard, but some argue that gold still plays a role and, for example, the IMF itself said recently that the gold holdings gave a fundamental strength to its balance sheet. We know that, for example, some rating agencies are taking comfort due to the presence of gold in the portfolio of either governments and/or central banks.

I recognise immediately that there is a great dose of psychology behind this. As I am neither a psychologist nor a psychiatrist, I will not elaborate further. Although, since the start of the crisis more than three years ago, I think we definitely need more psychologists and psychiatrists in the market, but that’s another story.

Risks diversification.

The fourth reason for having gold in the portfolio of central banks has to do with risk diversification. Gold presents very good diversification properties in this respect. It is clear that risk diversification is important for any investor, but it is probably even more valued in central banks. Central banks are usually quite conservative when they invest, and risk diversification is surely something extremely important in their choices. There are several studies that show that when you have gold in a portfolio, you have no systematic and no non-diversifiable risk.

There are a certain number of conditions, for example, a sufficiently long time horizon for the investment and so on, but this is accepted now.

My colleagues who are dealing with gold, or trading in gold, have also conducted some studies. We have three families of assets in terms of foreign exchange reserves. We have assets in US dollars, in Japanese yen and gold. The studies compared several portfolios with different compositions. All portfolios without gold systematically had a higher volatility than any portfolio where there was some gold, and this was statistically significant. It is also true, of course, that there are several optimum portfolios in this respect, but all those that did have good characteristics had at least some gold in their composition. These studies, which are based on a statistical and economical method that is called Standard Mean Variance Portfolio Optimisation, provided similar results to others conducted in other central banks, which reached the same conclusion. Of course, these statistical and econometric studies are not sufficient to determine how much gold we should have in our portfolio in the long term. We also need a more fundamental approach to determine exactly what we should have, but still there is this idea that gold contributes to a good risk diversification of the portfolio.

So those four factors, the economic security, unexpected needs, the notion of confidence and, finally, risk diversification explain why gold is important in the portfolio of the central banks.


I would like to look at the future and address two issues. The first one is the so-called CBGA (Central Bank Gold Agreement) and the second one is what will be the place of gold for central banks and authorities in general in the future.

First of all, the new CBGA was announced in August 2009 and is the third agreement of its type. In 1999, the signatories of the first Agreement (15 central banks: the Eurosystem, the Bank of England, the Swiss National Bank and Riksbank in Sweden) wanted to minimise the adverse impact of their own sales of gold. Therefore they decided to set up a formal and transparent agreement, and because the terms of the agreement were published, the market was well aware of the intention. Among many things, they communicated first the idea that gold would remain an important element of global monetary reserves. They said that they would sell gold, but it wasn’t their intention to dismiss gold.

The Agreement also stated that central banks would not enter into the market as sellers except for those sales already decided and, for these, there was an agreement not to exceed more than 400 tonnes per year, which was a total of 2,000 tonnes over the whole life of the Agreement, which was five years. At that time, the gold production was running at around 2,500 tonnes per year, so this additional supply via the sales of the central banks meant an increase of roughly 16%. One could have expected a negative reaction on the prices, but in the first three trading days after the Agreement had been announced, the spot price increased by 14%. So the perception was bullish rather than bearish, and the idea was simply that the market was better informed. Therefore there was suddenly much less uncertainty related to possible sales by central banks, and that contributed to the standardisation of the market.

During the five-year period, the signatories indeed sold those 2,000 tonnes. What was important also was the way these sales had been organised. The big sellers under the first Agreement were the central banks of Austria, Portugal, Switzerland and England. As an example, between 2002 and 2004, the Bank of England sold half of its stock of gold, roughly 300 tonnes. The others, for example, the Swiss National Bank, waited until the Bank of England had sold the lion’s share of this amount before entering itself very actively in the market. The Banco de Portugal waited until the very end of the Agreement before starting to sell. All this was co-ordinated and contributed to avoiding any shock on the market. In total, the reduction of gold held by those 15 central banks during those five years was approximately 2.5-3% per year. In 2004, the Agreement reached its end and it was decided to enter into a new Agreement, not exactly with the same central banks – there were a few changes – but the Agreement was reconducted. There was a difference. It was announced that the self- imposed ceiling would no longer be 400 tonnes but 500 tonnes, and therefore the total would be 2,500 instead of 2,000 tonnes over the five years. This could have led to some bearish reaction, but it was not the case, simply because the market now was reassured that those sales could be well organised and should not have a negative impact on the market.

I should say that the signatories sold only 1,875 tonnes at the end of this second Agreement, much less than was initially agreed. Over the five years of the second Agreement, the price of gold has doubled, which is not bad for the gold holders. It is amazing to observe this, because according to the statistics provided by the World Gold Council, in 1998, there were 33,500 tonnes of gold in the public sector holdings. At the end of 2008, there were 29,800 tonnes. So a reduction of 3,700 tonnes is the fastest and biggest decline ever observed in gold holdings. I think the CBGA contributed to that; in spite of this, the price of gold increased instead of declining.

In September 2009, the CBGA No. 2 reached its end. It has been decided to enter into a new Agreement – again for a five-year period – which will end in September 2014, again with a ceiling of 400 tonnes per year – 2,000 tonnes in total. Since the ceiling of the former Agreement was not used, it wouldn’t have made sense to maintain such a high ceiling. Another change is that the IMF has also announced that it will sell approximately 400 tonnes in the coming year. It has been stated in the Agreement that those sales can be accommodated within the ceiling decided by the signatories of the CBGA. Therefore, even if the IMF is not a signatory of this Agreement it will be treated similarly. This is surely an element that contributes to stabilising the gold market.

So what about all the central banks in the world? Will the amount of gold held by the official sectors globally continue to decline or not? It is very hard to make any forecast in this respect. Forecasts are always difficult, especially when they concern the future of gold, but there are some elements that can help us to make up our mind. First, it is clear that risk diversification is more and more an issue nowadays, and we have seen public sector authorities, mainly sovereign wealth funds, that have quite seriously changed their approach and diversified their investment, buying real estate or commodities. We have even seen some of them buying raw materials to secure the needs of their own country and so on. So, it is not excluded that gold can also benefit from that diversification. The CBGA signatories as a group have around 50-55% of their foreign reserves in gold. This ratio has increased in the past seven to eight years with the increase in gold price (which more than offset the effect of gold sales). In terms of volume, the Eurosystem, with 10,800 tonnes, holds a third of all gold held by the public sector. The top five European central banks in terms of gold holdings are the Deutsche Bundesbank with around 65% of its reserves in gold (based on World Gold Council and IMF statistics and end of month spot gold prices). Banca D’Italia and Banque de France are above 60%, the Nederlandsche Bank around 50% and the SNB around 30%. What those central banks will do throughout the duration of this CBGA is known, so there will probably be a small reduction in volume, but not in value.

So what about other reserve-holders in the world? Periodically, the biggest reserve- holders do not really have gold. China and Japan, for example, have only 2% of their foreign exchange reserves invested in gold. For Russia or Taiwan, the proportion is slightly higher but still well below the figures I mentioned for the signatories of the CBGA. So, I don’t think I take a huge risk if I anticipate that those countries are very unlikely to appear on the sales side of the market. We might observe a stabilisation, if not an increase of gold in the official reserve.

This leads to my conclusion. There are three elements that I wanted to make very clear. The first one is that the Eurosystem holds more gold in value now than 10 years ago, even if it has sold quite a lot of gold. The second is that even if gold prices can be volatile, gold really contributes to the diversification of risk in the portfolio, particularly that held by the official sector – even if some central banks will continue to sell gold and even if a new potential seller, the IMF, has emerged. I will not conclude with any kind of certainty that gold holdings will continue to decline in the coming years.

Thank you for your attention.

Paul Mercier is currently Principal Adviser in Market Operations at the European Central Bank after having served as Deputy Director General for 11 years. He was previously Head of Financial Markets of the European Monetary Institute (EMI). In this capacity, Paul Mercier participated in the design of the monetary policy and foreign exchange policy instruments of the European Central Bank.