Two years ago this week a surprisingly controversial event occurred in the gold world. The first gold exchange-traded fund to trade in the States was launched. The StreetTracks Gold Shares gold ETF marked the dawn of a new era where investors can now directly buy a gold proxy from within their familiar stock trading accounts.

The November 18th, 2004 launch of GLD galvanized long-time gold investors like few other events I can remember. Hardcore gold investors have long believed in holding physical gold in their own immediate possessions, which is definitely very prudent. All portfolios ought to have a foundation of physical gold, the real deal, in their owners' own immediate possessions. Among these faithful gold investors, GLD spawned a schism.

In one camp were the radically paranoid, the tinfoil-hat crowd. They believed that GLD was a grand front by anti-gold forces. They truly thought the GLD trust would take investors' money, not buy any gold, and instead use the proceeds as margin to short gold and drive its price way down. This seems silly, but I am not making it up. I saved articles and forum comments surrounding the GLD launch because it was such a seminal event.

The second camp consisted of much more rational opposition. This generally swirled around the fact that GLD is merely "paper gold" since it isn't physical coins in fist. In the first couple weeks after GLD launched, a damning series of articles emerged questioning all kinds of highly technical aspects of the trust's operations. At the core of these attacks was the usual anti-paper-gold bias, "If it ain't physical then it ain't gold."

One particularly popular article right after launch, written by a GLD competitor in the digital-gold business, analyzed GLD's inventory of gold bars and pointed out alleged irregularities. Apparently some of GLD's reported gold bars in inventory were duplicates! The strong implication of this alarmist revelation was that GLD was a fraud. Published on a Sunday, the following Monday it spawned a minor panic of GLD selling.

Being a student of the markets and a CPA, these allegations fascinated me. So I went through the raw data myself. It turned out that out of 8306 gold bars GLD listed as inventory, there were 78 duplicate sets, or 156 bars total. Thus only 1.9% of GLD's gold bars were questionable. I dug deeper and found that in 100% of these 78 cases bars A and B with the same serial number had very different weights. So while they did have duplicate serial numbers in the GLD report, they weren't duplicate bars. I explained all this in depth in the 12.8.2004 Zeal Speculator.

At the time I concluded, "In addition to omitting the hugely important fact that the 78 bars in question were not identical but all have different weights, he cried "fire" in a crowded theater without sharing all the relevant facts. Other benign explanations are possible. All questioned bars have 4-digit numbers, so perhaps some IT guy at GLD truncated a digit on his import of raw data from GLD's custodians." My thesis? Someone creating the 160-page PDF file of GLD's gold-bar holdings just screwed up. No big deal.

A day later GLD issued a press release and it turned out I was right. There was a minor data problem in the very early exports of raw gold-bar data from GLD's custodian. I was willing to give the newly-launched GLD the benefit of the doubt because I was in the third camp, the hardcore gold faithful who took a rational and pragmatic approach to GLD. We believed that even though GLD was certainly no substitute for physical gold, any new way for non-traditional investors to buy gold indirectly had to be a good thing and expand the market.

As a physical-gold investor, I have never personally owned GLD and probably never will. That being said, there are vast legions of investors out there who cannot be bothered to go to a coin shop and buy real gold. They are pure stock-market investors and have little or no desire to expand their horizons. While GLD was irrelevant to the hardcore, it was a boon to the mainstream investors who wanted some gold exposure in their portfolios but would have never attained it if gold wasn't somehow tradable via their usual stock accounts.

Looking back now two years later, and still shaking my head at the anti-GLD circus run by pro-gold alarmists around GLD's launch, the GLD impact has indeed been very positive in hindsight. The day before the GLD ETF launched, gold closed at a bull-to-date high of $444. About 18 months after this launch in May 2006, gold had soared 62% higher to a new bull-to-date high of $720. Did GLD suppress the price of gold? Not so one would notice! Indeed GLD proved very helpful to our in-progress gold bull by broadening participation.

The day GLD launched, it had 8 metric tonnes of gold in the trust. No one knew if this stock-account-tradable gold proxy would be a hit or flop. A week later initial GLD demand was so strong despite the anti-GLD naysayers that it had swollen to 100 tonnes. This week, Thanksgiving 2006, GLD's holdings are up near 416 tonnes. Thus GLD has led to gold investment demand of over 400 tonnes, much of which might have never existed if mainstream investors and funds had no easy stock-account-tradable way to gain gold exposure.

Just how much gold is 400 tonnes? A crazy lot of it! In addition to being worth $8.3b, 416 tonnes of gold would rank the US GLD ETF alone as 13th in the world relative to central banks. In other words, there are only a dozen sovereign nations that hold more gold in their vaults than the vaults of the GLD trust! To go from zero to 400+ tonnes in two years is an incredible achievement and GLD deserves great kudos for pulling it off.

Regardless of where you stood in the past or stand today on gold ETFs in general, you have to admit that GLD's results today at its second birthday are impressive. And of course GLD is not the only gold ETF in the world, so in total gold ETFs have spawned much more gold demand than this one example alone. Yet since GLD is the US leader and it is so meticulous about accumulating data, examining it is a great way to get a feel for gold ETFs' impact as a whole.

In this essay I am going to examine the GLD trust's gold holdings, its daily trading volume, and its variance from the spot gold price over its two years of existence. This fascinating data provides an excellent window into the impact of the GLD ETF. After I reexamined all of this data this week, it really gave me a warm-and-fuzzy feeling on gold ETFs in general. Just as originally hoped, they are greatly expanding the gold market to a far wider pool of investors than would exist if physical gold coins were still the only game in town.

The relentless growth in GLD's gold holdings is fascinating. For the most part they have only gone in one direction, up. This surprises me as it defies the expectations I had before GLD launched. Four years ago when thinking about the theory of a gold ETF, I had assumed that its underlying gold holdings would fluctuate, in both directions, far more wildly than they have in reality. Before I get into that theory and why it didn't prove true, first some thoughts on this chart are in order.

GLD's gold holdings soared in the week it was launched. While gold ETFs had already been live for some time in other countries, GLD was the first such animal in the US and it generated a lot of press even among mainstream channels. Initially the GLD holdings continued to grow in early 2005 even when gold was weak. Investors, whether they were individuals or funds, were apparently still diversifying into GLD following its introduction.

Then gold itself was flat, oscillating above and below $425 for most of 2005. A sideways consolidating market is not exactly the type of environment that sparks excitement, yet GLD held its own. Even when the gold price dipped the total GLD holdings were stable. On balance during this consolidation, a low-demand period for gold, GLD holdings grew considerably. Even though gold hadn't done much of anything for a year, on its first birthday the GLD ETF's holdings were already well north of 200 metric tonnes of gold.

And then the greatest upleg of this gold bull began in autumn 2005. The first half of this upleg, into early February, witnessed a fairly orderly rise in the gold price. During this time GLD holdings soared with increasing demand as rising gold prices captured investors' attention. Interestingly, GLD holdings stabilized right near gold's initial interim top in early February of this year. But after that even when gold started consolidating lower for the next couple months, GLD holdings modestly grew.

Then, provocatively, during the second parabolic half of gold's latest upleg from late March to early May the GLD holdings were fairly stable. For some reason there wasn't a flood of unbalanced fresh demand for GLD so the trust didn't have to add to its gold holdings. While gold blasted from $550 to $720 in a very short period of time, the GLD gold holdings just grew modestly. And then when the parabola crashed the selloff in GLD wasn't all that big. There was just a trivial dip in GLD's holdings while the gold price plummeted into June.

Since its mid-June interim lows, the gold price spent most of the summer consolidating sideways. During phases when gold was rising like in late June and early July, GLD holdings grew. But during other times when gold was grinding lower or falling like from mid-July to early October, GLD holdings remained stable. Indeed overall GLD managed to grow in size during this latest high consolidation in gold. It is really quite remarkable.

Why? Per ETF theory, a gold ETF should act as a conduit between stock-investor capital and the physical gold world. A gold ETF needs to track the price of gold to be effective. And in order to track gold, the gold ETF actually has to shunt capital from stock investors into gold or from gold back to stock investors. In other words it really has to buy and sell actual gold. If there was not some mechanism to equalize supply/demand imbalances on either the gold or stock side, then the gold ETF would soon careen off track and not mirror gold at all.

So when the gold price is rising, the GLD ETF will be bid up in sympathy. But if stock investors are bidding up GLD faster than gold traders are bidding up gold, then the ETF custodians have a problem. The only way for them to keep GLD's price from decoupling from gold to the upside is to take the flood of stock capital buying GLD and pass it through directly to the gold market. They buy physical gold. So when GLD's physical-gold holdings are growing, it means that demand for GLD shares is growing at a faster pace than gold itself.

The GLD custodians take this excess capital and shunt it directly into gold. This helps stabilize the tracking of GLD to the real gold price. Excess GLD demand that would push GLD's price too far above gold's is funneled into the gold market via GLD trust purchases of gold. This forces the gold price to rise faster and keep pace with the GLD ETF demand from the stock side. Without this active supply/demand link, any ETF would be doomed to failure.

Obviously since the GLD ETF creates a conduit for stock capital to flow into gold, it should increase the volatility in gold. Before the ETF launched, I assumed this would be a two-way street. On the downside, if GLD shares were being sold faster than gold itself then the custodians would have no choice but to sell some of their physical gold to provide capital for redemptions. This would force excess GLD selling pressure into the gold market and ensure GLD tracked gold to the downside too. It would exacerbate gold volatility.

But amazingly, so far GLD only appears to have enhanced the upside volatility of gold. GLD holdings rise when gold is doing well, a testament to increasing GLD ETF demand. But even when gold is doing poorly, there are seldom any net redemptions, selling of GLD gold to keep the ETF tracking its underlying commodity. This means a couple things, both of which should be very exciting to all gold investors.

First, overall demand for GLD is growing with the price of gold. This means the pool of capital ready to chase gold is getting larger all the time. So as gold marches higher on balance, net net even during gold's sharp corrections there are more than enough new purchases of GLD shares to offset GLD sales. The GLD trust shunts this stock demand directly into gold by increasing its gold holdings. Overall GLD demand is growing with the rising gold price, a very bullish omen for gold.

Second, the stability of GLD's holdings even during the sharpest gold corrections such as the brutal May/June one strongly suggests that it is largely investors, not speculators, buying GLD. These strong-handed investors are probably buying GLD for portfolio diversification over the long haul, so they don't care if gold is up or down today. This validates the wildest dream of the gold ETF concept, that by providing an easy way for everyone to gain gold exposure it would dramatically increase the ranks of investors interested in holding gold.

If it was only speculators in GLD, then their manic buying and selling would cause the underlying GLD holdings to track gold much more closely. This includes falling sharply during gold corrections. But since GLD hasn't had to liquidate its gold when the gold price is falling, whoever owns GLD is holding it without regard for short-term gold psychology. GLD is helping to broaden overall long-term investor exposure to the gold market. The more long-term investors and capital involved, the longer and higher a bull will ultimately run.

The growth in GLD investment demand that isn't fleeing gold on weakness is also apparent in the GLD trust daily trading volume. Like a normal stock, GLD volume increases when the gold price is rising. And the higher gold gets in general as its bull progresses, the more interest grows in GLD and hence its trading volume rises over time.

Following an initial massive spike when the GLD ETF launched, volume stabilized at relatively low levels as gold consolidated sideways in 2005. And then as the latest gold upleg started marching higher in late 2005, GLD volume ramped up in sympathy. This volume growth was pretty orderly and it reversed as soon as gold appeared to hit an interim top in early February. GLD volume then waned with gold over the next couple months until gold's parabola started rocketing higher on sharp US dollar weakness.

As gold soared vertically, GLD trading volume followed. Since each share of GLD represents a tenth of an ounce of gold, near the May top the 21m shares of GLD being traded daily represented a staggering 64 tonnes of gold changing hands daily among stock investors. The GLD ETF had created an entirely new market where none had existed before since it was extremely unlikely the vast majority of the stock investors trading GLD would have gone and opened futures accounts if GLD hadn't existed.

Finally GLD volume hit an all-time high on a capitulation mega-spike on June 13th, the day gold plunged a staggering 7.3% to a new interim low. On this day alone an incredible 81 tonnes of gold effectively changed hands via the GLD ETF! Yet, despite this capitulation, apparently plenty of buyers of GLD were there to eagerly take the sellers' shares. How do I know this? GLD's gold holdings barely dipped during this sharp gold crash so the demand for and supply of GLD shares specifically was almost balanced.

Since the parabola and subsequent crash of gold earlier this year, daily GLD trading volume has once again stabilized. But it is really exciting to see that it is stabilizing at much higher levels. Where GLD was averaging around 2m shares a day for much of its first year, during the last five months average daily GLD volume has soared to nearly 5m shares! GLD is getting more popular, and seeing more trading action, as the gold price marches higher. This is great to see and is good news for all gold investors.

Now the purpose of GLD is not to take the place of physical-gold coins, but merely to track the gold price. As such, my final chart looks at how well GLD is tracking gold. The yellow GLD price is multiplied by ten to get to a one-ounce equivalent and is charted over the blue spot-gold price below. Then the volatile red line calculates the daily variance between where GLDx10 is trading compared to where gold itself is trading.

The white variance line does the same thing but compares GLDx10 to the London PM Fix in gold. This particular gold price is more relevant than spot gold to GLD's performance since it is this London PM Fix that the GLD custodians are using for their official tracking benchmark. Overall GLD's custodians have done an outstanding job of ensuring that their ETF closely tracks the underlying gold price just as it is supposed to.

Overall the daily correlation between GLD and the spot gold price is nearly perfect, running 0.9993. Initially GLD tracked gold incredibly tightly as the nearly identical yellow and blue lines in 2005 illustrate. For most of its first year the daily variance between GLDx10 and the London PM Fix was trivial and even its extremes were usually within +/- 1.2%. But during its second year of operation, since last November, GLD's variance has widened slightly.

Note that the oscillating white variance reading above is widening compared to the early days. Today the extremes in this measure of the effectiveness of GLD in accomplishing its gold-tracking mission are largely between +1.8% to -1.2%, a 3% extreme range. I suspect there are a couple reasons for this slightly looser tracking.

First, the underlying gold price itself has been far more volatile in the second year of GLD's life than in its first. During GLD's first year the gold price just pretty much ground sideways and didn't move around all that much. With gold being pretty sedate, it was probably easier for GLD to track it closely.

But during the past year gold has been all over the place. The wider and sharper the underlying swings in gold, the less likely day-to-day GLD supply and demand will match day-to-day gold supply and demand perfectly. As such, a wider variance should be expected with more underlying gold volatility.

Second, as the yellow and blue lines show GLD's tracking of gold is loosening. This is partially caused by the slow cannibalization of GLD. GLD, like any ETF, has to fund its own management and operation expenses. The way this is done is the same way it happens in a pure stock ETF, by gradually selling off pieces of itself. In GLD's case its annual expense ratio runs about 0.4%. Thus over each year of its existence 0.4% of GLD's holdings are gradually sold off to finance its operations. Thus today one share of GLD does not represent 0.1 ounces, but 0.8% less than that after two years or 0.0992 ounces.

This is why the yellow GLDx10 line is running lower than the blue gold line to a bigger degree as GLD ages. At year 3 it will be 1.2% lower and at year 10 it will be 4% lower. Now I know this concept bothered a lot of gold investors initially, but I have never had a problem with it. This is standard in the ETF business.

Every ETF cannibalizes itself gradually over time to cover its own management expenses, even the vaunted QQQQs. The expense ratio of the QQQQ NASDAQ 100 ETF is running about 0.2% per year. Not only is the gold ETF vastly smaller, but it is a heck of a lot more work to manage changing physical-gold inventories compared to changing virtual stocks inventories. The 0.4% GLD expense ratio is perfectly reasonable for this service.

Another reason why the gradual cannibalization of GLD doesn't bother me is because of the way the gold-coin market works. Commissions on buying and selling gold coins are absolutely gigantic compared to commissions for trading stocks. On just a single round-trip buy and sell, a gold-coin investor could very well pay 2% to 3% in commissions each way. So an effective 4% commission over a decade in the GLD ETF compared to 4% to 6% per each round-trip trade in physical gold coins makes the 0.4% per year GLD expense ratio look trivial. And indeed it is.

Today the GLD gold ETF, just turning two, is doing fantastically well. Its physical gold holdings have grown a staggering 50x in two years! And since these holdings seldom drop even during sharp corrections in gold, GLD holders as an aggregate are buying and holding. GLD demand from the stock-market-capital side is stable and growing despite the wild volatility in gold. This is a dream come true for the gold market.

The GLD ETF has created a conduit through which pure stock capital can be shunted into the gold world to broaden participation in the underlying gold bull. This benefits every gold investor, whether you own physical coins and gold stocks like us or whether you are a mainstream investor who only owns GLD. The bigger the GLD trust and the other gold ETFs grow, the higher the gold price will ultimately run in this bull market since these ETFs add to overall global gold demand. The more investors the merrier this party!

While we have never recommended GLD specifically at Zeal and probably never will, we are in the midst of a new deployment into elite gold stocks that are likely to thrive in the next major gold upleg. Increasing demand for GLD will help drive up gold prices which will in turn light a fire under the gold stocks. Please subscribe to our acclaimed monthly newsletter today so you don't miss the new trades to ride the next big gold surge higher!

The bottom line is GLD has been a runaway success in its first two years. Not only did it silence the naysayers and their crazy antics, but it exceeded the dreams of its proponents in broadening investor participation in the gold bull. The gold ETFs have opened up conduits through which the vast pools of pure-stock-market capital can indirectly bid up the physical-gold price. This is wonderful for all gold investors!

While the gold ETFs are not a replacement for physical gold for hardcore gold investors, they were never intended to be. The gold ETFs like GLD exist solely for mainstream stock investors who can't be bothered to buy physical gold directly, and for mutual funds which can't legally trade outside the stock markets. We ought to welcome these gold rookies into this bull with open arms since they are helping to drive up gold prices for all of us.