2/2/2012: US Mint Sales for January - signaling return to fundamentals-driven demand?

January data is out for US Mint sales and time to update my semi-regular analysis. Here's the note. I am putting a disclaimer below - so the Irish stuffbrokers' community that somehow gets their facts wrong when no one is around to correct them breaths easier. Everything you read below is my personal opinion informed by my analysis of the official data from the US Mint.

January data from the US Mint on sales of gold coins presents an interesting picture, both in terms of seasonality and overall demand for the asset class.

Some background to start with. 

Gold prices have been moving sideways with some relatively moderate volatility in recent months. Between August 2011 - the monthly peak in US Dollar-quoted price and January 2012, price has fallen 4.55%, but in the last month, monthly move was 10.82% and year on year prices are up 30.4%. Crisis-period average price is now at USD1,154/oz and the standard deviation in prices is around 337 against the historical (1987-present) standard deviation of 330. In 2011 standard deviation for monthly prices stood at (small sample-adjusted) 144, well below historical volatility, due to a relatively established trend through August 2011. However, prices returned to elevated volatility in August 2011-January 2012.

These price dynamics would normally suggest rising caution and buyer demand reductions over time. And to some extent, this sub-trend was traceable in the data for US Mint sales in some recent months too. For example, unadjusted for seasonal variation, August 2011 sales of Mint coins peaked at 112,000 oz with relatively moderate 0.67 oz/coin sold gold content. By November 2011, sales slowed down to a relative trickle of 41,000 oz at 0.71 oz/coin sold. December sales came in at 65,000 oz with gold content on average of 1 oz per coin sold. Much media hullabaloo ensued with calls for catastrophic fall off in demand, the renewed claims that a gold bubble is now in action and the decline is coinage sales as evidence of that.

In reality, there was very little surprising in the sales trends overall.

Chart 1 below shows US Mint sales in terms of the number of coins sold. Care to spot any dramatic bubble-formation or bubble-deflation here? Not really. There is a gentle historical upward trend since January 1987. There is volatility around that trend in 2010 and far less of it in 2011. There is seasonality around the trend with Q1 sales uplifts in January, some Christmas season buying supports in early Q4 etc. There is also a slightly elevated sub-trend starting from early 2009 and continuing through today. More interestingly, the sub-trend is mean-reverting (heading down) which is - dynamically-speaking stabilizing, rather than 'bubble-expanding' or 'bubble-deflating'.

Chart 1

Source: US Mint and author own analysis

Now, January sales are strong in the historical context and within the sub-trend since 2009. January 2012 sales of US Mint coins came in at 127,000 oz with relatively low 0.50 oz/coin sales. So coinage sales in terms of oz weight are 95.4% up on December, but 4.9% down on January 2011. For comparison, 2011 average monthly sales were 83,292 and crisis-period average monthly sales were 94,745 all at least 0.5 standard deviations below January 2012 sales. As chart above clearly shows, sales are now well ahead of historical averages and above 6 months moving average.

However, as chart below shows, sales in January were well below the trend line for average coin weight for sold coins: oz per coin sold is down 50.5% mom and down 43.1% year on year. Significantly, smaller coins were sold in January this year than in 2011. 2011 average oz/coin sold was 1.0 and the latest sales are closer to 0.59 oz/coin historical average.

Chart 2

Source: Author own data and analysis based on underlying data from the US Mint

There is no panic in the overall trends in demand for coins when set against the price changes, with negative general trend in correlations between demand and gold price established in mid-2009 continuing unabated, as shown in Chart 3

CHART 3

 Source: US Mint, World Gold Council and author own analysis

However, when we look closer at the 12 months rolling correlations and 24 months rolling correlations, the picture that emerges for January is consistent with gentle negative correlation that has been present since the beginning of 2011. See Chart 4 below. January 2012 12mo rolling correlation between gold price and volume of gold sold via US Mint coins is +0.02, having reverted to the positive from -0.42 in December 2011. This is the first positive (albeit extremely low) monthly 12mo rolling correlation reading since July 2010. 24 mo rolling correlation in January 2012 stood at benign -0.30, slightly up on -0.34 in December 2012. Again, resilience if present in the longer term series and at shorter horizon there are no huge surprises either. Of course, in general, one can make a case, based on the recent data, that investors are simply turning back to the specific instrument after gold price corrected sufficiently enough. In this light, latest US Mint data would be consistent with fundamentals-supported firming of demand. But crucially, there is no evidence of either panic buying or selling.

CHART 4

Source: Author own analysis based on the data from US Mint

Lastly, let's take a look at seasonally-neutral like-for-like January sales. Chart below shows data for January sales, suppressing the huge spike at 1999. Clearly, sales are booming in terms of coins numbers sold. But recall that coins sold in January 2012 are smaller in gold content, so overall gold sold via US Mint coinage is marginally down on January 2011, making January 2012 sales the fourth highest on record.

CHART 5

 Source: Author own analysis based on the data from US Mint

The Table below shows summary of US Mint coins sales for 3 months November-January covering holidays periods sales, including the Chinese New Year sales. While January 2012 period shows healthy sales across all three parameters, there is still no sign of any panic buying by small retail investors anywhere in sight here. Sales are ticking nicely, in 2011 and 2012, well ahead of 2001-2008 levels (confirming lack of evidence that sustained price appreciation over the last 18 months has provided a signal to dampen retail demand), but behind 2009-2010 spikes (further supporting the view that 2011-2012 dynamics are those of potential moderation in the precautionary and flight-to-safety motives for demand, and more buying on long-term gold fundamentals).

TABLE: US Mint sales – 3 months through January

 Source: Author own analysis based on the data from US Mint

Welcome back to ‘normalcy’ in US Mint sales.

Disclaimer:

1) I am a non-executive member of the GoldCore Investment Committee
2) I am a Director and Head of Research with St.Columbanus AG, where we do not invest in any specific individual commodity
3) I am long gold in fixed amount over at least the last 5 years with my allocation being extremely moderate. I hold no assets linked to gold mining or processing companies.
4) I have done and am continuing doing academic work on gold as an asset class, but also on other asset classes. You can see my research on my ssrn page the link to which is provided on this blog front page.

5) Yes, you can find points (1)-(3) disclosed properly and permanently on my public profiles. 

6) I receive no compensation for anything that appears on this blog. Never did and not planning to start now either. Everything your read here is my own personal opinion and not the opinion of any of my employers, current, past or future.

07/11/2011: US Mint sales for October

In recent weeks there was some long-expected noises coming out of the gold 'bears' quick to pounce on the allegedly 'collapsing' sales of gold coins by the US mint. I resisted the temptation to make premature conclusions until the full monthly sales data for October is in. At last, we now can make some analytical observations.

The thesis advanced by the 'bears' is that October sales declines (for US Mint sales of new coins) are:

1. Profoundly deep
2. Consistent with 'gold bubble is bursting at last' environment and
3. Significantly out of line with previous trends, and
4. Changes are reflective of buyers exiting the market on the back of high gold prices

Let's take a look at the data:

First - sales. 

In absolute terms, number of coins sold by the US Mint in October has fallen to 65,000 from 115,500 in September. Mom, thus, volume of sales, measured in the number of coins is down 43.7% and yoy change is -45.6%. Significant declines. Latest sales are running below the historical trend and 6mo MA has hit the long term historical trendline. 

This suggests reversion to historical mean, as predicted by my previous note on this matter and is, in my view, a welcome sign of some 'froth' reduction in the speculative component of the market. The trend remains on the upside, and 6mo MA is still running ahead of pre-crisis averages. Historical average is at 98,329 coins with a massive standard deviation of 112,309. Crisis period average is 128,967 and smaller (but still substantial) standard deviation of 110,323. Now, for 10 months of 2011 so far, the average is 130,400 coins sold, but the standard deviation (imprecise estimate, of course) of 41,934 or roughly 1/3 of the volatility over entire history.

Thus, if anything, monthly movements along the elevated average trend for crisis period are now looking less volatile than in pre-crisis period, which suggests that gold is acting as a hedge during the crisis against prolonged risks in other asset classes and that this property is so far being reinforced by reduced volatility as well.

Chart above shows that when it comes to gold coinage sales in volume (oz) of gold content, October sales (50,000 oz) are well below September sales (91,000 oz) and are 46.8% behind October 2010 sales. Worried 'bears' are onto something here? Well, not exactly. As with coinage, volatility of the series historically runs at 52,985 against historical average sales of 55,768 oz. Crisis-period volatility is at 44,726 against crisis-period average sales of 95,859 oz. 10mo through october 2011 volatility is at 26,514 (1/2 of historical volatility) and average sales are 89,350 oz - below crisis period average. Again, there seems to be more stability in sales in terms of oz volume than before, which, surely, should be a good thing for a hedge instrument. The 6mo MA trend is on decline here since March-May 2011 and, again, this is not a bad thing, as it signals continued reversion to 'normal' trading conditions - i.e. potential reduction in speculative buying.

Next little thingy, volume of gold per coin sold on average now stands at 0.769 oz/coin in October, virtually bang on with September 0.788 oz/coin. Which too is a good thing. Average historical volume of gold per coin sold is 0.587 oz/coin (stdev of 0.2) and crisis-period average is 0.816 oz/coin (stdev of 0.191). Latest 10 months period average is 0.703 oz/coin (note - we are still well ahead of that in October) and stdev for the period is 0.126 - well below historical volatility. 

So no drama - in fact, much less drama - in October data. Upward trends remain, reversion to trend is ongoing nicely, volatility falling. I never make predictions about bubbles timing, but as far as 'bursting' explosions and profound changes - I don't really see them. At least not yet.

What about the fourth 'argument' listed above? Are buyers fleeing gold coins markets because prices are too high? Well, I don't know what buyers think, but correlation between price of gold and volume of gold sold via coins by the US Mint is evolving as follows:

Thus, in October, 12mo dynamic correlation has fallen to -0.24 from -0.06 in September. This looks dramatic, until we consider historical trends. Average historical correlation is at -0.09 with stdev of 0.397. Crisis period correlation averages at -0.151 with a standard deviation of 0.377. For the period of January 2011-October 2011, average correlation is at -0.205 and stdev at 0.151.

The above implies that while current negative correlation is not dramatic, the trend in 2011 is so far distinctly for deeper negative correlations between gold price and coins sales and for more stability along this trend line.

Is this a good thing? Nope. The opposite is true in my opinion. More negative correlation implies stronger reduction in speculative buying, leaving gold coins demand more dependent on long term hedging objectives and as the tool for preservation of wealth. In other words, less speculation, more long term demand. This is not what we should see in a bubble 'bursting' stages.

Once again, caution is due - I am not arguing if there is a bubble in gold markets overall. This is just analysis of the coins sales. I am simply suggesting that we are seeing a well-predicted reversion to the mean along upward trend in demand. We are also seeing, in my opinion, gold coins doing exactly what gold in general is expected to do - providing long term hedge instrument against risks associated with other asset classes.

Disclosure: I serve as non-executive member of the Investment Committee of GoldCore and I am long gold with stable unchanged allocation over the last 3 years. All of the above views are solely my own. 

11/09/2011: What gold coins sales tell us about the 'bubble'

Here is the extended version of the article published by Globe & Mail on the topic of US Mint sales of gold coins.

Of all asset classes in today's markets, gold is unique. And for a number of reasons(i).

Firstly it acts as a long-term hedge and a short-term flight to safety instrument against virtually all other asset classes.ii Secondly, it supports a wide range of instruments, including physical delivery (bullions, coins and jewellery), gold-linked legal tender, gold-based savings accounts, plain vanilla and synthetic ETFs, derivatives and producers-linked equities and funds. All of these are subject to diverse behavioural drivers of demand. Thirdly, gold is psychologically and analytically divisive, with media coverage oscillating between those who see gold as either a long-term risk management tool, or a speculative investment, a "barbaric relic" prone to "bubble"-formation.

In the latter context, it is interesting to look closer at the less-publicised instrument - gold coins, traditionally held by retail investors as portable units to store wealth. Due to this, plus demand from collectors, gold coins are less liquid and represent more of a pure 'store of value' than a speculative instrument. Lower liquidity of coins is not driven by shallow demand, but by reluctance of owners to sell them when prices change. Gold coins are economically-speaking "sticky" on the downside of prices - when price of gold falls, holders of coins are not usually rush-prone to sell as they perceive their coins holdings to be 'long-term accumulations', rather than speculative (or yield-sensitive) investments. They are also "sticky" on the upside of prices - while demand is impacted by price effects (with generally higher gold prices acting to discourage new accumulation of coins), holders of coins are not quick to sell to realize capital gains, again due to entirely different timing to the holdings motives. Think of a person setting aside few thousand dollars worth in gold coins to save for child's college fund.

In general markets, classical bubbles begin to arise when speculative motives (bets on continued and accelerating price appreciation) exceed fundamentals-driven motives for opening new long positions in the instrument. Bubbles blow up when these tendencies acquire wide-based support amongst retail investors.

In late stages of the bubble, we should, therefore, expect demand for gold coins to falter compared to the demand for financially instrumented gold (ETFs shares and options). In the mid-period of bubble evolution, however, as retail investors just begin to rush into the asset, we can expect demand for coins to rise in line with demand for jewellery and smaller bullion. But we do not expect a flood of gold coins into the secondary market (and hence a collapse in gold coins sales) until literally past the turn of the fundamentals-driven prices toward the stage of mid-cycle "bubble" collapse.

The US Mint data on sales of gold coins suggests that we are not in the last days of the "bubble". But there are warning signs to watch into the future.

August sales by the US Mint were up a whooping 170% year on year in terms of total number of coins sold, while the weight of coins sold is up 194%. On the surface, this gives some support to the theory of gold becoming short-term overbought by retail investors (see chart below), but it also contradicts longer-term view that gold coins sales should fall around bubble peak.

Source: US Mint and author own analysis

In part, monthly comparatives reflect huge degree of volatility in US Mint sales. Looking at the longer term horizon, since January 2008, US Mint sales volumes averaged 97,011 oz with average coin sold carrying 0.82 oz of gold, the standard deviation of these sales was 45,196 oz and 0.19 oz/coin, implying that August results comfortably fit within the statistical bounds of +/- 1/2 STDEV of the mean for the crisis period. Equally importantly, August results fit within +/-1 STDEV band of the historical mean since 1986 through today. In other words, current gold coinage sales do not even represent a 1-sigma event for the entire history of gold coins sales supplied by the US Mint and are within 0.5-sigma risk weighting for the crisis period since January 2008.

Neither is the current monthly increases in demand represent a significant uptick on previous months or years demand. At 112,000 oz of gold coins sold, August 2011 is only the 19th busiest month is sales since January 2008. It ranks as the 34th month in terms of the gold content per coin sold. Again, not dramatic by any possible metrics. Since January 1988 there were 87 months in which average gold content per coin exceeded August 2011 average and on 38 occasions, volumes of gold sold in the form of coins by the US Mint exceeded last month's volume.

Again, not dramatic by any possible metrics, especially once we recognize that in terms of risk-related fundamentals, August was an impressive month with US and EU debt crises boiling up and economic growth slowdown weighing on global equities markets.

Charts below illustrate these points.
Source: US Mint and author own analysis
Source: Author own calculations based on the US Mint data
Source: Author own calculations based on the US Mint data

The data also shows that physical demand for coins is largely independent of the spot price of gold. Historically, since 1986, average 12-months rolling correlation between the spot price of gold and the volumes of gold sold in US Mint coins is negative at -0.09. Since January 2008, the average correlation is -0.2. And over the last 3 years, the trend direction of gold spot price (up) and the volumes of gold sold in coinage (down) have actually diverged (see chart). The latter is, of course, concerning and will require closer tracking in months to come. The correlation between price of gold and volumes of gold sales through US Mint coins is now negative or zero for 13 consecutive months.

Source: Author own calculations based on the US Mint data

Source: Author own calculations based on the US Mint data

The chart above also highlights the fact that the current trend levels of US Mint sales are significantly elevated on previous periods, with exception of 1986-1987 and 1998-1999 demand spikes. Since the global economic crisis began, annual coinage sales rose 7-fold from just under 200,000 oz in 2007 to 1,435,000 oz in 2009, before falling back to 1,220,500 oz in 2010. Using data through August, I expect 2011 sales to remain at around 1,275,000 oz. This implies that the 2008-2011 average annual sales of US Mint coinage gold are likely run at slightly above 2 times the average annual rate of sales of coinage gold in the period 1988-2007.

Given the state of the US and other advanced economies around the world since January 2008, this is hardly a sign of dramatic over-buying of gold by masses of retail investors. Instead, we are witnessing two core divergent trends emerging from the coins markets:

1. Since, roughly-speaking, 2009, the trend in coins sales is moving counter to the trend in spot price for gold, implying that retail investors are not rushing into gold, as one would expect were gold to be a bubble, and
2. The levels of sales of US Mint coins remain elevated, on average, since the crisis began, implying that high demand for coins, relative to historic trends, is most likely being driven by fundamentals-underpinned demand for safety.
   

In short, there is no indication, in the data reviewed, of the bubble beginning to inflate (sharp rises in gold coins demand along the trend with prices) or close to deflating (sharp pull-back in demand for gold coins).

Of course, the evidence above does not imply any definitive conclusions as to whether gold is or is not a "bubble". Instead, it points to one particular aspect of demand for gold - the behaviorally anchored, longer-term demand for gold coins as wealth preservation tool for smaller retail investors. Given the state of the US and other advanced economies around the world since January 2008, US Mint data does not appear to support the view of a dramatic over-buying of gold by the fabled speculatively crazed retail investors that some media commentators are seeing nowdays.


Disclosure: I am long physical gold and hold no long or short positions in other gold instruments.


i These and other facts about gold are summarized in my recent presentation available at http://trueeconomics.blogspot.com/2011/08/20082011-yielding-to-fear-or-managing.html.

ii As shown in the recent research paper by Profs Brian Lucey, Cetin Ciner, and myself, covering the period of 1985-2009: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1679243