I'm reposting this from FT alphaville. If the images don't show up it most likely means they moved the article behind a pay wall. Sorry. You cannot fight progress. Incidentally I think the lead chart that suggests a sharp split between TIPS and gold is bogus. This is a chart I made a few minutes ago with slightly different axes:

A generous dose of pseudo-science produces this:

anyways I try to read with an open mind.


The following chart, we propose, has the potential to inspire a whole new way of looking at the gold and Treasury market:

It comes from Goldman’s latest research on gold and reflects the recent breakdown in the relationship between the 10-yr Treasury Inflation-Protected Securities real yield and the gold price. Importantly, it was also accompanied by a downside amendment to Goldman’s gold price forecast:

We lower our 3-, 6- and 12-mo gold price forecasts to $1,825/toz, $1,805/toz and $1,800/toz and introduce a $1,750/toz 2014 forecast. While we see potential for higher gold prices in early 2013, we see growing downside risks. As a result, we find that the risk-reward of holding a long gold position is diminishing and recommend rolling our long Dec-12 COMEX gold position into a long Apr-13 position and selling a $1,850/toz call to finance a $1,575/toz put to protect against a decline in gold prices. Since 2009, this strategy achieved a better Sharpe ratio than a long gold position.

The rationale for the move was as follows:

Positioning responds to the announced stock of purchases, pricing in the size of the purchase program at announcement while showing little response to the subsequent flow of purchases. This means that when the flow of Fed purchases is discontinued – but the stock of Fed asset holdings is unchanged – there should be little effect on gold prices. This sensitivity is in line with our US economists’ finding that it is the expected stock of Fed asset purchases and not the flow that matters for bond yields.

Stock announcements of asset purchases that require Fed balance sheet expansion push positioning higher. In turn, stock announcements that do not require balance sheet expansion push positioning lower, with this negative coefficient correcting for the decline in real rates that occurs around such announcements. This differentiated response stands in sharp contrast with the systematic decline in Treasury yields around QE announcements and suggests that gold prices “look through” easing that does not require potentially inflationary Fed balance sheet expansion. Conceptually, while Operation Twist had an impact on yields through an increase in Fed holding duration, it did not affect the monetary base as it did not require reserve creation.

What Goldman are essentially arguing is that when it comes to 10-year Tip yields tracking the gold growth rate (inversely) “not all QE is created equal”. Indeed, only the sort of QE that expands base money has the potential to lift gold prices.

All fine and logical.

But we can’t help feeling they are neglecting an important point: that The Fed’s unconventional liquidity measures have turned conventional Treasuries into a non-yielding gold type instrument in their own right. The dual protective qualities of Tips (protecting as they do against inflation and deflation) on the other hand have increased their superiority over both gold and conventionals, making them even more desirable than before.

Where we find ourselves consequently is a world where Treasuries and gold are becoming ever better substitutes for each other, since neither offer noteworthy income. Indeed, only the cost associated with holding gold and the ability to recover par value at maturity with Treasuries is now a differentiating factor. What’s more, as gold becomes increasingly accepted as industry collateral, even its liquidity and cash profile becomes akin to that of a Treasury.

Meaning only a burst of demand for gold futures — leading to an extremely steepened curve — has the potential to encourage flows out of Treasuries and into spot gold in a way that meaningfully lifts the clearing price of spot gold.

In short, the forward curve must compensate for more than the cost of money and storage if the yellow metal is to be considered preferable to a Treasury investment and see a rise in its spot price. Lacking that, the two assets should trade in tandem as they tug over the same volume of excess liquidity, while being bound by the zero level on the Treasury side.

This is because the zero level effectively caps gold’s desirability relative to Treasuries, since holding gold incurs a negative interest (due to the price of storage) whilst holding Treasuries doesn’t and protects principal. (That is unless new base money is created taking Treasury yields negative, and moving the gold clearing price higher.)

Indeed, you could say the moment short-term USTs began yielding next to nothing was the moment that inflows into gold and Treasuries came at the cost of each other. Like two different cola brands competing over the same customer base, despite peddling only a marginally different product:

No surprise then that net speculative length in gold futures — which is an expression of net open interest — plateaued from 2009 onwards, and then fell significantly with the onset of Operation Twist, as this chart from Goldman Sachs shows:

So why should Operation Twist have had such an impact on speculative length in gold futures?

Goldman explain it as follows:

To try to understand these interactions, we turn to the CFTC’s COMEX gold net speculative length data, as it has exhibited the most straightforward sensitivity to real rates over the past decades, rising with declines in real rates. This relationship started breaking down in 2009 with net speculative length increasing sharply around the announcement of QE1 and outperforming the decline in real rates. In turn, net speculative length declined sharply in the fall of 2011 around the announcement of Operation Twist (Exhibit 3). Understanding these dislocations is key to our gold outlook as net speculative length is the primary driver of our gold price forecast.

Their conclusion is once again related to OT not increasing monetary base.

While we agree that’s an important obvious factor at this stage we think they are under-appreciating OT’s influence on the gold choke price.

As Paul Krugman once commented (citing the Hotelling model):

Crucially, at least for tractability, there is a “choke price” — a price at which flow demand goes to zero. As we’ll see next, this price helps tie down the price path. So what determines the price of gold at any given point in time?

Hotelling models say that people are willing to hold onto an exhaustible resources because they are rewarded with a rising price. Abstracting from storage costs, this says that the real price must rise at a rate equal to the real rate of interest, so you get a price path that looks like this :

Which means the forward curve must be of a certain steepness if a gold investment is to pay off.

If the price of money falls to zero over, meanwhile, the gold curve flattens until it reflects only the cost of storage (chart via Krugman):

What’s more, since low interest rates generally make it more attractive to hoard resources — it makes even more sense to position oneself in physical gold rather than financial futures.

It should come as no surprise then that this is exactly the sort of flattening that struck the gold curve pre-Operation Twist. A flattening which persisted until the end of July 2011 (second chart):

Here’s another way of looking at it (the higher the line in the bottom chart the flatter the gold curve):

By June 2011, the flattening curve was indicating two things. One, that spot gold prices were approaching a possible choke price — in part a function of ever lower Treasury rates — and two, that until that choke price was reached people would continue to be incentivised to buy and hoard gold.

Everything crescendo-ed with gold’s final thrust higher in August 2011 — though this time arguably not because of a continuing approach to the choke price, but rather because changing interest rate expectations, on account of the US debt debacle, had shifted the choke price temporarily higher.

All of which — given other fundamentals (like still micro sized yields) had not budged — was the reflection of a mispriced futures curve and an over-estimated final choke price.

The net effect was a sudden and final burst of buying hoarding interest as people flocked to take advantage of a forward curve that was clearly over compensating for the cost of money and storage — a trend amplified by the dramatic yield compression that occurred in Treasuries at the same time.

Of course, once the debt debacle was resolved, it didn’t take long for the old pattern of curve flattening to resume, and for the choke price to reset at a level more reflective of current yields.

Indeed, if not for Operation Twist, it’s possible that a tug of war between Treasuries and gold would at this stage have set in and continued until either Treasuries traded at the same negative yield as gold — unleashing the detrimental effects of negative carry in the process — or gold spot prices traded over futures prices.

Operation Twist to the rescue

With Operation Twist, however, the Fed was able to transfer Treasury supply from the long-end to the short-end. The process released a veritable flood of short-term supply onto the market, with the immediate effect of supporting short-term rates.

From the point of view of safe-haven investors this was the equivalent of the Fed dishing out free carry, making the choice between Treasuries and gold suddenly a clear no-brainer. (Safe assets and yield, yes please!) (Not to mention the fact that higher rates generally disincentivise hoarding.)

The result was an immediate bearish turn for spot gold.

That we saw the return of a steeper curve, meanwhile, was prompted this time not so much by inflows into futures, but conversely outflows from the physical market. That the gold price didn’t collapse completely, meanwhile, was possibly linked to central bank buying, a move which eventually stabilized the curve and arguably encouraged the range-bound equilibrium we’re now in.

As for the breakdown in the relationship between Tips and gold speculative length… we propose that this is once again a gross mis-reading of the signals emanating from the Tips market. Rather than implying falling real interest rates — which would usually be associated with rising gold prices — we believe Tips may be reflecting greater appreciation of the principal protection option embedded with the securities.

Remember, a Tip security returns a set rate against a continuously adjusted principal on a cumulative basis.

However, in the event of deflation, while the Tip security benefits from principal protection — meaning the real value of the principal rises — the income stream continues to be priced off an adjusted principal (and during deflation, that means off an ever smaller number) . This means even if an investor’s principal ultimately ends up appreciating in real terms — unlike a conventional bond — the investor suffers a reduction in the size of the bond’s nominal income in the interim.

Another way of looking at it, if you invest in a Tips security and experience inflation, in 10 years time you are no worse or better off in real terms than you expected to be when you took the investment. Your principal retains exactly the same purchasing power it did 10 years ago, as do your coupons on a relative basis throughout the life of the investment.

In deflation, however, it’s a different story. The income retains exactly the same purchasing power, but the return of your principal ensures you are much better off in real terms at the end of the investment.

Thus, when it comes to the gold price vs the Tips real yield — as illustrated in the very first chart that so inspired us — it may not be that Tips are indicating a fall in real yields but rather the increasing value of their embedded principal option.