高盛黄金分析

使用大宗商品分析方法来预测金价走势

高盛研究

The monetary demand for gold has played a significant part in the large swings in gold prices of the past decade, with the selling of gold from government central bank reserves in the late 1990s depressing gold prices and the buying of gold for private investment—including the new gold exchange traded funds (gold-ETFs)—sending gold prices through $1,000/toz twice during the current financial crisis (Exhibit 1). Because the monetary demand for gold has had such a significant influence on gold prices over the past decade, it is tempting to view gold prices as driven simply by the vagaries of government policy and the investor’s view of financial distress; however, when viewed within the larger historical context, the past decade’s rise in gold prices falls within a longer cycle in gold prices, driven more by the economics of gold supply. This report introduces a framework for understanding the influence of both monetary demand and the economics of gold supply on the price of gold. Based on this framework, we present a new approach for forecasting the price of gold based on the monetary demand for gold (specifically, the buying by gold-ETFs and the selling by central banks) and on the crucial role of the real interest rate in determining the opportunity cost of mining.

过去十年中，对黄金的货币需求在金价浮动中起了十分重要的作用：政府央行在90年代后期抛售黄金储备引起了金价下跌，用于私人投资的黄金购买——包括新的黄金ETF——让金价在当前的金融危机中两次突破1000/toz（表1）。由于对黄金的货币需求在过去十年中对金价的影响如此之大，因此人们试图将金价看成纯粹是受无常的政府政策以及投资者对金融危机的看法的驱动；但是，如果从更大的历史背景来看，过去十年金价的上涨符合金价的长期规律，更多是受黄金供应的经济考虑。该报告介绍的框架是要更好地了解货币供给以及黄金供应对金价的影响。基于这个框架，我们提出了一种预测金价的新方法，即依据对黄金的货币需求（尤其是黄金ETF的买进与央行的卖出）和实际利率在决定采矿的机会成本中的重要作用进行预测。

表1：金价在过去十年上涨了十倍，去年更是有大幅涨跌

This “gold as a commodity” framework suggest s that gold prices have strong support at and above current price levels should the current low real interest rate environment persist. Specifically, assuming real interest rates stay near current levels and the buying from gold-ETFs slows to last year’s

pace, we would expect to see gold prices stay near $930/toz over the next six months, rising to $962/toz on a 12-month horizon. However, should real interest rates move lower or gold-ETF buying continue at its current torrid pace, the upside risk to gold prices would likely be significant.

“大宗商品分析法来预测黄金价格走势”的框架表明，如果当前低实际利率环境一直持续的话，黄金价格对当前和超过当前的价格水平具有有力支撑。具体地，如果实际利率一直接近当前水平，并且来自黄金ETF的购买降至去年的速度，那么我们可以预期黄金价格在未来六个月将会接近$930/toz，在未来12个月里会涨至$962/toz。然而，如果实际利率走低或者黄金ETF购买依然保持现在的高速度，则金价将面临较大的上涨风险。

This report is the first in our new Frameworks series. This series of occasional reports is intended to explore in greater detail the frameworks necessary for developing views and forecasts for commodity market fundamentals and prices. In this first Frameworks report, we outline an approach to forecasting go ld prices and fundamentals from the framework of “gold as a commodity”—based on the economic and financial determinants of gold supply, demand, and inventories. Historically, we have viewed gold more as a currency than a commodity (see our February 4, 2009 report “Gold: The currency of last resort”), valuing gold (in US dollars) in relation to the US dollar price (or exchange rate) of a basket of currencies. We view the approach in this report as complementing, not replacing, the currency approach, with each framework providing a valuable perspective on gold pricing.

此为我们新框架系列的首份报告。这一系列的临时报告是要更加详细地探究一些对发展中的观点和对商品市场基本面及价格进行预测所必需的一切框架。在这份框架报告中，我们大致描述了从“黄金作为商品”的框架对金价和基本面进行预测的方法——以黄金供应、需求、库存的经济和金融决定因素为基础。从历史上看来，我们更多地是将黄金看作一种货币而不是商品（见我们2009年2月4日的报告“黄金：最后的货币”），将黄金价格（以美元）和一揽子货币对美元的价格（或汇率）联系起来。我们将这个报告里的方法视为是对货币方法的补充而不是取代，每一种框架对黄金定价提供一种有价值的视角。

In terms of the approach to gold pricing, these frameworks can be described as follows:

至于黄金定价方法，这些框架如下所述：

?Currency framework: Gold priced in relation to the price of potential substitutes for use as a store of value and medium of exchange.

Key price drivers: Exchange rates, financial risk as measured by credit default swap rates on high-risk sovereigns and financials.

?Commodity framework: Gold priced in relation to the marginal cost of supply and to the marginal willingness of consumers to pay.

Key price drivers: Real interest rates, the overall price level as measured by the consumer price index (CPI), and movements in monetary demand for gold.

●货币框架：黄金价格与可以作为价值储备和交换媒介的潜在替代品的价格相关。

决定价格的主要因素：汇率，以CDS衡量的高风险主权债务风险和金融风险

●商品框架：黄金定价和供给的边际成本以及消费者的边际支付意愿相关。

决定价格的主要因素：实际利率，以cpi计算的总价格水平，以及对黄金的货币需求动向。

The “gold as a commodity” framework explains three key “stylized facts” of gold prices in terms of the influence of real interest rate s, inflation, and monetary demand for gold on the supply and demand for gold.

“将黄金作为商品”的框架解释了实际利率、通胀以及对黄金的货币需求对黄金的供应和需求的影响中黄金价格三个关键的‘程式化事实’：

Three stylized facts of gold prices:

黄金价格的三个程式化事实：

?Long-term stability of purchasing power: The real (inflation-adjusted) price of gold has been stable over extremely long periods of time (Exhibit 2). More specifically, over the past 100 years, the real price of gold (in 2008 dollars) has averaged roughly $420/ toz, with an ounce of gold having the same purchasing power in 2005 as it did in 1900.

?Negative correlation with real interest rates: While stable over extremely long periods of time, real gold prices tend to move in long cycles, which are negatively correlated with the level of real interest rates.

?Positive correlation with financial distress: Gold prices tend to spike upward during financial crises, as the demand for monetary gold increases.

●长期稳定的购买力：黄金的实际价格（经过通胀调整的）很长时间以来一直趋于稳定（表2）。

更具体地说，在过去的100年间，黄金实际价格（2008年美元）平均大概是$420/toz，一盎司黄金在2005年的购买力和1900年的相同。

●和实际利率负相关：金价在相当长时间内保持稳定的同时，在长期内有循环变化，这种循

环变化和实际利率水平负相关。

●和金融危机正相关：当对货币黄金的需求增加时，金价在金融危机中趋于上行。

表2：实际金价长期稳定，并与实际利率相关在长期内有循环

Interestingly, the historical behavior of supply, demand, and physical inventories in the gold market suggest that the first two “stylized facts” arise from the economics of gold supply, not gold demand, while the third arises from the monetary demand for gold. In particular, the following historical behavior is evident.

有趣的是，黄金市场的历史供应、需求和实物库存行为表明：前两个“程式化事实”源于黄金供应的经济性因素，而不是黄金需求；而第三个则是来自于对黄金的货币需求。下面的历史行为尤其显著。

Historical behavior of gold fundamentals:

黄金历史行为的基本原理：

?World gold mine production has historically moved in roughly 30-year cycles over the past century or more, with mine production growing at a faster pace during periods of high real interest rates and at a slower pace during periods of low real interest rates.

?Monetary demand for gold—the demand for physical inventories of gold bullion and coin—increases in periods of financial distress such as the current one.

?Non-monetary demand—the demand for gold to be used in the creation of jewelry, art, electronics, and dentistry, is relatively stable over time, reacting primarily to the movements in real gold prices, with increasing real gold prices decreasing the non-monetary demand for gold.

●世界金矿生产在过去的100多年里大约30年为一个周期，实际利率高的时期，金矿生产速

度加快；在实际利率低的时期，速度变慢。

●对黄金的货币需求——对金条、金币等实际库存的需求——在金融危机时期会增加，如当

前的这个金融危机。

●非货币需求——用于珠宝制造、艺术、电子产品和牙医方面的黄金需求——相对稳定，主

要与实际金价波动相关：上升的实际金价会降低对黄金的货币需求。

As discussed in detail be low, these three “stylized facts” of gold prices and the historical behavior of gold supply, demand, and physical inventories are consistent with—and can be seen as arising from—the following view of the economics of the gold market. On the supply-side, the marginal cost of extracting gold from the ground consists of the realized cost of extracting the gold plus the opportunity cost of not leaving the gold in the ground to be extracted in the future. The fact that real gold prices are stable over extremely long periods of time suggests that the realized cost of extraction increases with the price level of the overall economy. The fact that the rate of world gold mine production increases with the real interest rate suggests that, as the real interest rate increases, the opportunity cost of extracting gold declines, leading to greater gold extraction today. To take an extreme example, imagine that the real rate of interest is extremely high, such that one does not care at all about the future; then one would want to extract as much gold today as possible, given the realized cost of extraction.

正如下面将要详细讨论的，这三个黄金价格“程式化事实”以及黄金供应、需求和实际存货的历史行为与下面对黄金市场经济因素的观点是一致的，也可以看成是后者的生成结果。在供应端，将黄金从地下提取出来的边际成本包括提金的实现成本加上未来提取黄金的机会成本。实际金价长期稳定的事实表明：提金的实现成本随总经济的价格水平增长而增长。世界金矿生产的速度与实际利率的增长而增长的事实表明：当实际利率增加时，提金的机会成本就会下降，导致了如今大量的黄金提取。举一个极端的例子，想象一下对于给定的提金实现成本，实际利率高到极限、高到人们无需考虑未来，那么人们今天会提取尽可能多的黄金。

On the demand-side, monetary demand for physical inventories of gold bullion and coin increase with perceived financial distress. Increased monetary demand for gold must be met by increasing prices to motivate greater mine production or reduced non-monetary demand. Non-monetary demand—primarily jewelry demand—largely accommodates movements in gold supply and monetary demand, with higher real gold prices reducing non-monetary demand for gold and even inducing the scrapping of gold jewelry.

在需求端，对金条、金币实物库存的货币需求会因可察觉的金融危机而增加。对黄金增加的货币需求必须要有上升的价格，以刺激更大量的金矿生产或者减少非货币需求。非货币需求——主要是珠宝需求——为黄金供应和货币需求的走势提供了很大空间，更高的实际金价会减少对黄金的非货币需求，甚至导致黄金珠宝业的崩溃。

In the next section of this report, the historical behavior of gold prices and fundamentals outlined

above is discussed in more detail, providing the background for the conceptual framework of the economics of the gold market. In the section following, this conceptual framework is used as the basis for a new approach to forecasting gold prices. In the third and final section, the relationship between the frameworks of “gold as a currency” and “gold as a commodity” is discussed.

报告的下一章节将会更加详细的讨论上面大致描述的金价历史行为和基本原理，并为黄金市场经济因素的概念框架提供背景情况。在接下来的章节中，这个概念性框架被用做预测金价的新方法的基础。在第三也是最后的章节，讨论“黄金作为货币”框架和“黄金作为货币”框架的关系。

The historical behavior of supply, demand and physical inventories, and the economics of the gold market

黄金市场的供应、需求和实际存货的历史行为及经济因素

Gold mine supply: the economics of extraction and real interest rates

金矿供应：提金的经济因素和实际利率

Since the beginning of the twentieth century, the production of gold from the world’s mines has ebbed and flowed in long cycles extending over the span of decades. As seen in Exhibit 3, world gold production has exhibited a fairly regular cycle, with roughly 30 years between each peak. Interestingly, real gold prices have been strikingly stable over this more than a century of history, although they have tended to be low when world gold production is near a peak and high when world gold production is near a trough, suggesting that it is the movements in supply that are driving real gold prices over these long cycles. If supply is driving the price of gold, then the question becomes what is driving supply?

自20世纪初，世界金矿生产以超过十年的时间跨度为一周期上下起伏。从表3可以看出：世界黄金生产呈现出相对规律的周期，两个波峰间大约相距30年。有趣的是，实际金价在100多年的历史中令人瞩目地稳定，虽然曾经也在世界黄金生产接近波峰时走低，而在世界黄金生产落入低谷时走高，这表明：供应的走势影响了这些较长周期里的实际金价。如果供应决定金价，那么问题就是：什么决定供应？

As seen in Exhibit 4, changes in the rate of world gold mine production tend to move with the level of real interest rates, with high real interest rates leading to faster growth in gold mine production.

从表4可以看出，世界金矿生产的速度变化与实际利率水平趋同，高实际利率带来高速度的金矿生产。

表3：金矿生产在过去100多年以30年一个周期上下起伏表4：金矿生产随实际利率而增加

Exhibit 4 suggests that higher real interest rates increase the rate of gold mine production while lower real interest rates slow the rate of mine production. This is consistent with the economics of an extraction industry. The marginal cost of extracting today is not only the actual cost of mining, but the opportunity cost of not having the same gold to mine in the future. This suggests that higher interest rates lead one to discount the future more heavily, driving down the opportunity cost and leading one to extract at a faster rate. To take an extreme example, if one did not care at all about the future (i.e., an extremely high interest rate), one would extract the gold as quickly as possible.

表4表明：实际利率升高，金矿生产速度增加；实际利率下降，金矿生产速度减慢。这与提金工业的经济因子是一致的。现在，提金的边际成本不仅是采金的实际成本，还有未来没有相同的黄金可采的机会成本。这表明，高利率会导致未来预期严重打折，减少机会成本，从而导致提金业速度加快。举一个极端的例子，如果一个人完全不考虑未来（也就是，利率高到极端），他会尽可能快地提取黄金。

It is important to note, however, that the above economic behavior does not require that gold be viewed as an exhaustible resource. In fact, the economics of extraction can be seen as the mirror image of the economics of investment. It is conventional economic wisdom that high real interest rates discourage investment, as they increase the cost of capital. Consequently, one expects to see less investment in factories and manufacturing plants in high real interest rate environments. For a gold mine, however, extraction offers a second—more rapid—way to divest, by pulling the gold out of the ground today. Viewed in this way, the economics of extracting from an existing gold mine is the mirror image of the economics of investing in a new one. A high real interest rate reduces the motivation to invest and increases the motivation to extract. Consequently, higher real interest rates lower the marginal cost of extraction, which leads to greater supply and lower prices.

然而要注意的是，以上的经济行为并不要求将黄金看成是一种可耗竭性资源。实际上，提金经济学可以被看成是投资经济学的镜像。传统的经济理论认为，高实际利率会抑制投资，因为它们增加了资金成本。因此，在高实际利率环境中人们会预期看到工厂及制造业中的投资减少。然后对金矿来说，黄金提取提供了第二种——速度更快的——撤资方法，即今天将黄金从地下开采出来。从这方面看，从现存的金矿中提取黄金的经济学是投资新矿的经济学的镜像。高实际利率会减少投资动机，增加提金动机。因此，实际利率上升，会降低提金的边际成本，从而带来更大量的供应和更低的价格。

Monetary demand: Central bank reserves, gold-ETFs, and COMEX inventories

货币需求：央行储备、黄金ETF和COMEX库存

Once gold is extracted from the mine and refined, it can be used to satisfy both monetary and non-monetary forms of demand. The monetary demand for gold consists primarily of demand from the official sector for central bank gold reserves, from the private sector for holdings of gold bullion and coins (including those held by the gold-ETFs), and from the demand for warehouse stocks on the gold futures exchanges, principally the Commodity Exchange (COMEX).

一旦黄金被开采和提炼，它可以既满足货币需求，又可以满足非货币形式的需求。对于黄金的货币需求主要包括官方结构对央行黄金储备的需求、私人机构对持有金条和金币的需求（包括被黄金ETF所持有的）、黄金期货交易的库存需求，主要是COMEX交易所库存。

When discussing the monetary demand for gold, it is important to distinguish between the overall level of physical inventory being held and the strain on the current supply-demand balance from changes in the level of physical inventory being held. For example, holdings by gold-ETFs were 33 million toz, representing only 3.0% of all monetary gold holdings at the end of 2008 (Exhibit 5); however, while gold-ETF holdings of physical inventories are a fraction of the amount held by the official sector and by other private investors, the rapid growth in gold-ETFs over the past five years has exerted a meaningful pull on supply. For example, from the end of 2007 to the end of 2008 ETF holdings increase by 8.0 million toz, comparable to central bank sales and almost one-third the size of all the other private sector investment purchases (Exhibit 6).

在讨论黄金的货币需求时，一定要区别实际库存持有总水平和因实际库存持有变化而产生的当前供需平衡的压力。比如，黄金ETF持有3300万盎司，在2008年末仅占全部货币黄金持有的3%（表5）；但是，当实际库存的黄金ETF持有是官方储备以及其他私人投资者持有量的一部分时，黄金ETF在过去5年里的快速增长已经对黄金供应起到了有意义的拉动作用。比如，从2007年底到2008年底，ETF持有增加了800toz，和央行抛售大致相当，并几乎是其他私人部

门投资购买总量的三分之一（表6）。

表5：期货交易和黄金ETF库存在全部库存持有中只占一小部分表6：……近几年黄金ETF持有增长显著

Consequently, changes in gold-ETF and central bank holdings can—and historically have—exerted a meaningful influence on the underlying supply and demand balance. As seen in Exhibit 7, gold-ETF holdings have increased rapidly over the past five years, likely tightening the underlying supply-demand balance and exerting upward pressure on gold prices. Conversely, as shown in Exhibit 8, central bank sales increased rapidly in the late 1990s, weakening the underlying supply-demand balance and putting downward pressure on gold prices. The extent of the downward pressure on gold prices eventually led to the signing of the Washington Agreement in 1999, limiting the combined annual sales of signatories to 400 tonnes (12.9 million toz).

因此，黄金ETF和央行持有的变化可以——并且已经——对潜在的供需平衡产生了重要影响。表7可以看出，黄金ETF持有在过去5年增长迅速，拉紧了潜在供需平衡，并对金价产生上行压力。相反，如表8所示，央行卖盘在90年代后期增加迅速，削弱了潜在的供需平衡，使金价有下行压力。金价下行压力的程度最终导致了1999年的《华盛顿协议》，将签约国的绑定年卖盘限制在400吨（1290万盎司）。

表7：黄金ETF年需求在过去几年大幅增长表8：央行卖盘在90年代后期增长，导致了《华盛顿协定》

Relative to these other forms of monetary demand, COMEX warehouse inventories of gold are small both in terms of levels and changes. Consequently, they tend to have little direct impact on the supply-demand balance; however, COMEX inventories do play an important role in determining gold lease rates and, consequently, the shape of the gold futures price curve.

和其它形式货币需求相关，COMEX黄金库存水平低、变化小。因此，他们对供需平衡没有直接影响；但是COMEX库存在决定黄金租赁利率上起着十分重要的作用，影响着黄金期货价格曲线。

The gold lease rate is the interest that must be paid (in our case in US dollars) to lease, or borrow, physical gold for a specified period of time. Consequently, this can be viewed as the explicit cost of borrowing gold to hold for a period of time or the opportunity cost of holding one’s own gold and not lending it out to another. As seen in Exhibit 9, the one-year gold lease rate rose in the early 1990s to an average of between 1.50% and 2.00% by the late 1990s, with a spike to a high of 4.10% in October 1999 following the signing of the Washington Agreement, in which central banks agreed to limit sales of gold from their reserves. Gold lease rates then declined throughout the first part of the current decade, reaching a low of only 0.08% in September 2006, before rising with the current financial crisis. Throughout this time period the gold leas e rate moved inversely with the level of COMEX gold inventories see Exhibit 9), much as one would expect, given that the gold lease rate is the price of holding physical inventory.

黄金租赁利率是规定期限内租出、租借实物黄金所必需支付的利息。因此，这可以看成是借用黄金并持有一段时间的显性成本，或者是持有黄金不外借的机会成本。如表9所示，一年期黄金租赁利率在90年代早期呈上升趋势，到90年代默契平均在1.5%到2%之间，1999年10月达到峰值4.1%，紧接着就是《华盛顿协定》的签署，协定中，各中央银行同意限制他们储备中的黄金出售。黄金租赁利率随后在当前这个十年的前期呈下降趋势，2006年9月仅0.08%，然后随着当前金融危机的到来又一次上升。这个期间，如我们所预期，黄金租赁利率与COMEX黄金库存呈相反方向变化（见表9），因为黄金租赁利率就是持有实际黄金库存的价格。

表9：黄金租赁利率和COMEX黄金期货交易中已注册的库存呈相反变化

Further, Exhibit 10 illustrates a nearly text book “demand vs. price” relationship between the amount of physical gold inventories held on the COMEX and gold lease rates through August of 2007. The inventory-demand curve is quite stable and downward sloping, with less inventories of gold being

held on the CO MEX as gold lease rates increase. Following the onset of the current financial crisis in the second half of 2007, however, the gold lease rate began to climb to a much higher level than would have been expected, given the level of physical gold inventories. That is, the demand for physical gold appears to have shifted outward in response to the financial crisis.

另外，表10指出了2007年8月COMEX持有的实货黄金库存数量和黄金租赁利率间‘需求vs 价格’的关系。存货-需求曲线相当稳定并呈向下趋势，黄金租赁利率增加时，COMEX持有的黄金库存减少。但是，2007年下半年金融危机爆发，黄金租赁利率开始攀升到超出相对于此时库存预期的较高水平。也就是说，作为对金融危机的反馈，实物黄金需求曲线右移。

表10：高黄金租赁利率会减少对COMEX库存的要求表11：TED捕获的不断增加的金融风险会增加对黄金库存的需求。

The link between the current financial crisis and the increased demand for physical gold inventory can be seen explicitly by looking at the correlation between gold lease rates in the recent period and the TED spread or the difference in (one-year) interest rates between LIBOR and the US Treasury (Exhibit 11). Because the TED spread is the difference in borrowing rates between financial institutions and the US Treasury, it effectively captures the degree of financial counter-p arty risk priced into the mark et. As the degree of counter-party risk increases, so too does the demand for physical gold.

当前金融危机和对实货黄金库存增加的需求之间的联系可以从近期黄金租赁利率以及ted利差或者libor和美国国库债券（一年期）利率差中清楚地看到。由于TED利差是银行体系和美国国库债券间的拆借利率差，所以它有效捕捉了市场已消化的金融对手风险程度。对手风险程度增加的话，对实货黄金的需求也会增加。

The relationships in Exhibits 10 and 11 can be quantified through the use of a regression analysis. As seen in Exhibit 12, the gold lease rates are well-explained by the TED-spread and the level of COMEX registered gold inventories. Exhibit 13 reports the results of this regression analysis of the (log) one-year gold lease rate on COMEX registered gold inventories (in million troy oz) and the one- year TED-spread. The analysis was performed on the log of the gold lease rate in order to capture the non-linearity evident in Exhibit 10. As reported in Exhibit 13, a one million ounce increase in COMEX inventories decreases the gold lease rate by 0.68% while a 100 bp increase in the TED-spread increases the gold lease rate by 0.95%。

表10和11中的关系可以用回归分析进行量化。从表12可看出，黄金租赁利率可以通过TED 利差和COMEX注册黄金库存量进行有效解释。表13是对COMEX注册黄金库存（对数化调整）一年黄金租赁利率和一年ted利率进行还原分析的结果。分析是在黄金租赁利率上取对数进行的，为的是捕捉表10中显现的非线性特征。正如表13中显示的，COMEX库存每增加100万盎司，黄金租赁利率就下降0.68%；而ted利差每增加100bp，黄金租赁利率就增加0.95%。

表12：黄金租赁利率主要受COMEX注册库存和ted利差影响表13：黄金租金利率对ted利差上升而随COMEX库存下降

The gold lease rate is also the amount investors are willing to pay to hold physical gold as opposed to a gold futures contract, which helps explain why the TED-spread is an important driver of the demand for physical gold. Specifically, investors can gain exposure to the movements in gold prices by holding physical gold or futures contracts, but investors will be willing to pay more to hold physical gold in periods of high counter-party risk when they are less certain about the ability to collect on the futures contracts. In this manner, perceptions of greater financial risk increase the monetary demand for gold. To supply this increased monetary demand, physical gold needs to be either bid away from its competing uses or extracted from the mines.

黄金租赁利率也是投资者愿意支付持有实货黄金而不是黄金期货合同的数量，这可以解释为甚ted利差是实货黄金重要的驱动因素。具体地说，投资者可以通过持有实惠黄金或期货合同来建立净头寸，但在高对手风险期，投资者对收讫期货合同的能力信心减小，他们会更愿意支付持有实货黄金。这样，投资者看到更大的金融风险会增加对黄金的货币需求。为了满足这种增加的货币需求，实货黄金要不从竞争用途中得到，要不从金矿诶开采出来。

Non-monetary demand: jewelry demand and the real price of gold

非货币需求：珠宝需求和实际金价

It is com mon practice to divide the “demand” for gold into monetary and non-monetary demand. As discussed, the monetary demand for gold is the physical inventory of gold bullion and coins. Because gold is not consumed in use (as is, for example, petroleum), one could conceptualize the non-monetary demand for gold as the physical inventory of gold jewelry and art for example; however, the evidence suggests that the more relevant measure of non-monetary demand is not the physical inventory of jewelry and art but the demand for gold for the creation of new jewelry and art. That is, the change in the physical inventory of gold jewelry and art. Consequently, our concept of the gold market supply-demand balance would be that the change in amount of physical gold inventories held for monetary reasons equals supply less non-monetary demand.

将对黄金的‘需求’划分为货币和非货币需求时一般惯例。正如所讨论的，黄金的货币需求是金条、金币等实货库存。因为黄金不会在使用被消耗（如石油），所以我们可以将对黄金的非货币需求概念化为黄金珠宝和艺术品的实货库存；但是，有证据表明，和非货币需求更加相关的不是珠宝和艺术品的实货库存，而是用于制造新珠宝和艺术品的货币需求。也就是，黄金珠宝和艺术品实货库存中的变化。因此，我们关于黄金市场供需平衡的概念是指为了货币原因而持有的实货黄金库存量的变化等于为更少的非货币需求进行的供应。

Supporting the idea that the more relevant measure of non-monetary demand is not the physical inventory of jewelry and art but rather the demand for gold for the creation of new jewelry and art is

the fact that, once gold is embedded in high value jewelry and art, it is effectively locked up, much as it would be in a mine. Consequently, gold can either be held in a liquid monetary form, as inventory, or in illiquid jewelry. Gold for monetary holdings must either be sourced from the mine or from the scrapping of gold jewelry, both areas from which some expense is required to extract the gold.

与非货币供应更相关的因素不是珠宝、艺术品的实际库存，而是对用于生产新珠宝和艺术品的黄金的需求，支持这一观点的事实就是：一旦黄金被镶嵌于高价值的珠宝和艺术品中，它实际上是被锁定了，和在矿中差不多。因此，黄金可以以流动货币形式被持有，比如库存，或者是以非流动的珠宝形式。作为货币持有的黄金可以来源于金矿或者金饰品上刮擦下来，两种提金形式均需要一定的费用。

Perhaps because gold jewelry is purchased with the idea it will be held for long periods of time, non-monetary gold demand is remarkably stable over time. As seen in Exhibit 14, the apparent gold consumption per capita in the United States has seemed to simply fluctuate inversely over time with the real price of gold. This suggest s that, unlike many commodity markets, gold prices are not driven by non-monetary demand but that gold prices drive non-monetary demand.

或许人们购买黄金饰品是为了长期持有，所以非货币黄金需求长期内十分稳定。从表14可看出，美国的人均黄金显性消费在长期内似乎只和实际金价成反方向波动。这表明，和许多大宗商品市场不同，金价不是受非货币需求驱动，而是金价驱动非货币需求。

表14：珠宝对黄金的非货币需求起主导作用表15：历史上，非货币需求仅仅对价格有反应Highlighting the responsiveness of jewelry demand to gold prices is the fact that world jewelry demand fell in line with higher prices throughout the 1990s despite the rapid rise in jewelry demand from India (Exhibit 16). The strong growth in income in India, matched with a cultural proclivity for gold, has led India’s share of annual gol d jewelry demand to more than double over the past 15 years, increasing from 9.4% in 1993 to 23.0% in 2008 (Exhibit 17). Interestingly, Chinese jewelry demand was roughly flat over the same period, with its share of gold jewelry demand increasing only slightly, from 11.9% in 1993 to 12.6% in 2008.

强调珠宝需求对金价的反应是基于这样的事实：世界珠宝需求在90年代因价格升高而下降，尽管印度的黄金需求有快速增加（表16）。印度收入的有力增长和他们对黄金的文化倾向有关，这导致了过去15年中，印度在世界年黄金需求中所占的份额翻了一番有余，从1993年的9.4%增加到2008年的23%（表17）。有趣的是，中国珠宝需求在同时期大致持平，它在黄金珠宝需求中的份额略有增加，1993年11.9%到2008年的12.6%。

表16：世界珠宝需求在过去15年对价格持续作出反应表17：珠宝需求中心已经向印度转移

Assembling the evidence: a conceptual framework of gold as a commodity

整合证据：关于将黄金作为商品的概念化框架

The empirical evidence suggests the following conceptual framework for supply and demand in the gold market. Recalling Exhibits 15 and 16, one has the distinct impression that one is looking at price induced movements up and down a relatively stable demand curve (Exhibit 18). The origin of these price movements lie in shifts in the monetary demand for gold and from fluctuations in the cost of supply. Recalling Exhibit 4 suggests that most of the movements in supply seem to co me from fluctuations in real interest rates, with some role played by higher prices calling forth greater supply in the short run (Exhibit 19), while the shifts in monetary demand are driven primarily by central bank sales and private investment in monetary gold, notably by the gold-ETFs.

经验证据可以得出下面这个对于黄金市场供需的概念性框架。回顾表15和16，我们会有这样清楚的印象：我们可以将价格引发的上下波动看成是一个相对稳定的需求曲线（表18）。这些价格波动的根源在于对于黄金的货币需求的变化以及供应成本的起伏。回顾表4可以看出，供应的波动似乎大多是来自于实际利率的起伏，更高的价格往往会带来短期的供应上涨（表19），而货币需求的变化主要受货币黄金的央行卖出和私人投资——主要是黄金ETF——的驱动。

表18：非货币需求对价格波动的被动反应表19：实际利率驱动供应和价格涨跌

Forecasting gold prices

预测金价

From the framework of “gold as a commodity,” the evol ution of gold prices is determined by the movements in the underlying gold supply-demand balance. As in other commodity markets, such as copper and petroleum, inventories in the gold market act to smooth out the variability in the supply-demand balance. This role for inventories suggests that gold prices are driven by primarily two factors: physical gold inventory levels and the forward supply-demand balance. Like the other commodity markets, physical inventory levels drive the shape of the gold forward curve while movements in the forward balance are reflected in the movements in the prices of long-dated futures contracts.

从“黄金作为商品”框架可以看出，金价的变化决定于潜在的黄金供需平衡中的波动变化。和其他商品市场一样，比如铜和石油，黄金市场中的库存可以褽平供需平衡中的波动。库存的这个作用说明金价主要受两个因素驱动：实货黄金库存水平和未来供需平衡。和其他商品市场一样，实物库存水平影响黄金期货曲线的形状而未来平衡中的波动反映在远期期货合约的价格波动。

Consequently, a commodity pricing framework can be applied to forecasting gold prices by employing the following relationships.

因此，运用下列相关关系，商品定价框架可以用来预测金价：

?Gold forward price spreads: COMEX inventories and nominal interest rates drive the shape of the gold forward curve, specifically the price spread between near- and long-dated contracts. High inventory levels place downward pressure on near-dated prices relative to long-dated prices, as the market anticipates inventories returning to more normal levels over time.

?Long-run gold prices: Long-dated futures prices “look through” the near-term impact of inventories on prices to give a cleaner read on the long-run forward supply-demand balance and the long-run price of gold. The real long-run price of gold has been stable over the long term with deviations driven by real interest rates and monetary demand.

?COMEX inventories: COMEX inventories tend to be too small to exert a meaningful influence on the supply-demand balance. Further, COMEX inventories tend to move in anticipation of forward fundamentals, rather than the current supply-demand balance.

●黄金未来价格差：COMEX库存和名义利率影响黄金未来曲线形状，尤其是近期和远期合约

间的价格差。库存水平高会减小与远期价格相关的近期价格的压力，因为市场预测库存会随着时间推移转为比较正常的水平。

●长期黄金价格：远期期货价格体现了库存对价格的近期影响，给长期未来供需平衡和黄金

的远期价格给与更清楚的解读。黄金的实际远期价格在长期内保持稳定，但受实际利率和货币需求影响有偏差。

●COMEX库存：COMEX库存往往因为数量不大而对供需平衡影响不大。而且，COMEX库

存往往依据未来基本面变化，而不是当前的供需平衡。

In estimating these relationships, the role of speculative positions in capturing the forward balance of the market is crucial. Speculative positions convey information to the market on forward fundamentals. In this way, speculative positions can drive prices and weighted COMEX inventories. As expected, the information conveyed by speculative positions in the gold market primarily concerns the real interest rate, with net speculative length being inversely correlated with the real interest rate (Exhibit 20). Consequently, in employing these relationships to forecast gold prices, speculative positions are the channel through which forward fundamentals, notably real interest rates, drive gold prices and inventories.

在对这些关系进行估测时，投机头寸在捕捉市场未来平衡中的作用是十分关键的。投机头寸在

向市场传递了未来的基本面信息。通过这种方式，投机头寸可以驱动价格和加权COMEX库存。正如所预期的，投机头寸在黄金市场中传达的信息主要和实际利率有关，净投机头寸和实际利率反相关（表20）。因此，在应用这些关系预测黄金价格时，通过投机头寸这个渠道，未来基本面，尤其是实际利率会影响金价和库存。

表20：黄金净投机长度和实际利率反相关

Gold forward price spreads: COMEX inventories and interest rates drive the shape of the gold forward curve

黄金远期价差：COMEX库存和利率影响黄金远期曲线形状

The relationship between COMEX inventories and the gold lease rate implies a relationship between COMEX inventories and the shape of the gold forward curve. This is because gold can be leased indirectly through a set of futures market transactions, creating the potential for arbitrage between the gold lease rate and the gold forward curve. In order to lease gold indirectly through the gold futures market, an investor could engage in the following set of transactions. First, borrow funds at the interest rate to purchase physical gold. At the same time, sell short a futures contract (corresponding to the period of the lease). At the end of the lease period, deliver the purchased gold to close the short futures position, changing the purchase of the physical commodity into a lease. Because this set of transactions is equivalent to leasing gold directly, its cost must equal the gold lease rate. Consequently, the cost of leasing the gold must equal the price of physical gold plus the interest paid, less the price received on the short sale of the commodity futures contract. In short, the gold lease rate is inversely related to the carry in the gold forward curve.

COMEX库存和黄金租赁利率之间的关系暗示着COMEX库存和黄金远期曲线之间的关系。这是因为黄金可以通过一套期货市场交易被间接租赁，从而在黄金租赁利率和黄金远期曲线间创造套利。为了通过黄金期货市场间接租赁黄金，投资者可以运用下面这套交易。首先，以利率借款购买实货黄金。同时，卖空一份期货合约（与租约期限相符）。租约到期时，交付购买的黄金以结清卖空的期货头寸，将实货商品购买转变成租约。因为这一套交易等同于直接租赁黄金，因此其成本必须和黄金租赁利率相等。因此，租赁黄金的成本必须等于实货黄金价格加上已支付的利息，减去在商品期货合约卖空上收到的价格。总之，黄金租赁利率和黄金远期曲线持有收益反相关。

As implied from the discussion above and as seen in Exhibit 21, the shape of the gold price forward curve is well explained by nominal interest rates and COMEX inventories. Exhibit 22 reports the results of a regression analysis of the percentage spread between 60-month and 1-month forward gold prices on US Treasury yields and COMEX inventories, including registered and eligible. As reported in Exhibit 22, the carry in the gold forward price curve mirrors the carry in the US Treasury curve, with the coefficients on the 1-year and 10-year summing to five, reflecting the interest rate for the 5-year period of the gold price spread. The coefficients on the 1-year and 10-year US Treasuries reflect the weightings required to create a proxy for a 5-year zero coupon yield, with this relevant interest rate being roughly 20% the 1-year rate and 80% the 10-year rate over time. The carry in the forward curve also increases with the level of COMEX inventories, with a one million troy ounce increase in COMEX registered inventories increasing the carry by 1.5%.

从以上讨论和表21可以看出，金价远期曲线可以通过名义利率和COMEX库存加以解释。表22是美国国债收益和COMEX库存60个月和一个月未来金价的百分比差的回归分析的结果，包括注册的和其他可得库存。如表22所反映的，黄金远期价格曲线的持有收益反映了美国国债曲线的持有收益，1年期和10年期的系数共计为5，反映了5年期金价利差的利率。1年期和10年期美国国债系数反映了创造5年零息收益债券必须的权重，相关的利率大约是1年期20%、10年期80%。远期曲线中的持有收益同样随COMEX库存水平的增加而增加，COMEX注册库存增加100万盎司，持有收益增加1.5%。

表21：5年和1月远期金价差受利息和库存驱动表22：持有收益随着利率和COMEX库存水平的增加而增加

Interestingly, Exhibit 22 implies that both registered and eligible COMEX inventories affect the long-dated time spreads, while only registered COMEX inventories affect the lease rate (Exhibit 13). In other words, COMEX eligible inventories are increasingly important as one looks further out the forward curve, suggesting that the gold market is pricing them as available inventory, given time. Their effect on the time spreads is only two-thirds the effect of registered inventories, however. As will be discussed below, a weighted measure of the COMEX inventories equal to registered inventory plus two-thirds of the eligible inventories has the strongest connection to gold market fundamentals and prices. For now, note that because COMEX eligible gold inventories have tracked the growth in gold-ETF funds, this suggests that the growth of gold-ETFs will lead to lower registered inventories levels than have been observed historically in similar market environments (Exhibits 23 and 24).

有趣的是，表22暗示说注册和其他可得COMEX库存均会影响远期差价，但只有注册COMEX 库存影响租赁利率（表13）。换句话说，当人们更长远地注意远期曲线时，COMEX其他可得库存会越来越重要。这表明时间够长的话，黄金市场将它们作为可得库存定价的，但它们对跨期差价只有注册库存影响的2/3。正如下面将要讨论的，COMEX库存的一种加权方式等于注册库存加上2/3其他可得库存，这对黄金市场基本面和价格起着强大的联系作用。因为现在，COMEX

其他可得黄金库存已经能追踪出黄金etf基金的增长，这表明：黄金etf的增长将会导致注册库存水平在类似市场环境中比历史上已经观察到的水平要更低（表23和24）。

表23：注册库存在近期下降时其他可得库存却在增加……表24：其他可得库存和黄金etf的增长吻合

Long-dated gold prices: stable in real (inflation adjusted) terms over the long run but driven by real interest rates and monetary demand fluctuations in the medium term

远期金价：长期来说在实际（通胀调整）中趋于稳定，但在中等时间期限内受实际利率和货币需求波动的影响

The influence of COMEX inventories on the shape of the gold forward curve suggests that long-dated gold futures prices provide a clearer reflection of the forward supply-demand balance than do near-dated prices. Consequently, similar to the other commodity markets it is useful to decompose the prompt price of gold into the shape of the forward curve and the long-dated futures price. In this way, the present value of long-dated futures prices can be viewed as a proxy for the long-run gold price; however, in contrast to other commodity markets, the variation in the shape of the forward curve driven by inventories is very small, implying that the real long-dated price of gold is quite similar to the real price of prompt gold contracts (Exhibit 25). In other words, most of the variation in gold prices is driven by changes in the long-dated price of gold, not COMEX inventories.

COMEX库存对黄金远期曲线形状的影响清楚地反映了未来供需平衡而不是近期价格对远期价格影响更大。因此，和其他商品市场类似，将即时金价分解成远期曲线形状和远期期货价格是十分有用的。这样，远期期货价格的当前价值就能被看作是长期金价的代表；但是和其他商品市场相反的是，未来曲线形状受库存影响的变化很小，意味着黄金实际远期价格和即时黄金合约的实际价格十分接近（表25）。换句话说，金价中的多数变化都是受黄金远期价格变化的影响，而不是COMEX库存。

表25：历史实际黄金现货价格和估计的实际远期价格

Given the historical economic behavior of gold prices and fundamentals, one expects the real long-dated gold price to behave as follows.

?Real long-dated gold price is stable over long horizons.

?Over medium horizons, the real long-dated gold price is high when real interest rates are low and low when real interest rates are high, with the impact of real interest rates influencing it through the movements in net speculative long positions.

?Over shorter horizons, real long-dated gold price moves with changes in the forward fundamentals as reflected in changes in net speculative length as well as with gold-ETF buying and net central bank selling.

鉴于黄金价格和基本面的历史经济行为，我们可以预期实际远期金价行为如下：

●实际远期金价在长期内趋于稳定。

●中等长度时期内，实际利率低时实际远期金价就高；实际利率高时实际远期金价就低，实

际利率通过净投机多头的波动对其发生影响。

●短期内，实际远期金价随未来基本面的变化而变化，反映在净投机头寸的变化以及黄金etf

买进和净央行卖出。

We seek to capture these aspects of real long-dated gold prices by specifying that the percentage monthly change in the real long-dated price of gold is determined by the (log) real long-run gold price one-month earlier, the level of net speculative length one-month earlier, the change in the level of net speculative length, gold-ETF buying, and net central bank selling (all of which are measured in million toz). The results of this regression are reported in Exhibits 26 and 27. As seen in Exhibit 26, this model fits both the level and monthly changes in real long-dated gold prices quite well, capturing over 50% of the variation in monthly price changes. Further, as reported in Exhibit 27, each of the variables has a statistically significant impact on real long-dated gold prices, and in the expected direction.

我们想要通过详细说明实际远期金价的百分比月变化受一个月前实际远期金价、一个月前净投机多头水平、净投机多头水平的变化、黄金etf买入和净央行卖出（均以百万盎司计）等因素（经对数化调整）的决定来捕捉时机远期金价的这些方面。这种还原的结果见表26和27。如表26所示，这种模型对于实际远期金价水平和月变化都是适合的，捕捉了月价格变化中50%的变化。而且，如表27所示，按照预期的方向，每种变量对实际远期金价都有统计数据上的重要性。

表26：实际远期金价受净投机多头和货币需求的影响表27：回归结果

The model specified in Exhibit 27 suggests that there is a long-run equilibrium between the real long-dated price of gold and the level of net speculative length. This suggests that high levels of net speculative length (corresponding to low real interest rates) will lead to high real gold prices. Further, when real long- dated gold prices are high, or net speculative length is low, relative to this equilibrium, real long-dated gold prices will tend to fall, which restores the equilibrium. The relationship of real long-dated gold prices to the real interest rate is discussed in more detail in the text accompanying Exhibit 31 below.

表27中的模型表明实际远期金价和净投机多头之间一种长期平衡的关系。这表明净投机多头的高水平（与低实际利率相符）将导致高实际金价。而且，当实际远期金价高、或者净投机多头低的时候，和这种平衡相对应，实际远期金价将趋向下降，从而回归平衡。实际远期金价和实际利率的关系将在下面结合表31作进一步讨论。

The model specified in Exhibit 27 also suggests that, over shorter horizons, changes in gold market flows tend to move the real long-dated gold price relative to this equilibrium. In particular, a one million toz monthly increase in net speculative length, increase in gold-ETF purchases, or decrease in central bank sales raises real long-dated gold prices by 0.87% over the course of the month. This implies that a one million toz increase in the annual rate of gold-ETF purchases, for example, would raise the real long-dated price of gold by over 10.0% over one year.

表27中的模型同时还表明，在更短的时期内，黄金市场流动变化往往会推动实际远期金价和这种平衡相关。尤其，净投机多头、黄金etf买进每月上涨1million 盎司，或者央行卖出每月下降1 million盎司会使实际远期金价在一个月里上涨0.87%。这暗示说黄金etf买进年比率每增加1 million 盎司，将会使黄金实际远期价格在一年内上涨10%。

COMEX inventories: COMEX inventories driven by anticipated forward fundamentals, not the current supply-demand balance

COMEX库存：COMEX库存受预期的未来基本面而不是当前供需平衡的影响

The influence of COMEX inventories on the shape of the gold forward curve also implies that, in order to translate the real long-date d gold price back into a forecast of the spot price of gold, we need a forecast of COMEX inventories. Interestingly, COMEX inventories have shown a consistent positive correlation with the level of real long-dated gold prices over the history of the COMEX gold futures market (Exhibit 28).

COMEX对黄金远期曲线的影响同样暗示着：为了将实际远期金价转化成对黄金现货价格的预

测，我们需要对COMEX库存的预测。有趣的是，COMEX库存显示在COMEX黄金期货市场历史上与实际远期金价的正相关关系（表28）。

表28：COMEX实货黄金库存的高水平和高——不是低——金价相关

In commodity markets, high inventory levels typically coincide with low prices, with high levels of inventory representing an excess of supply that needs to be worked through the market. While this is typically the case, however, it is not always the case. The correlation between inventory levels and commodity prices is determined by whether the market is being driven by changes in the current supply-demand balance or the anticipated forward supply-demand balance. For example, suppose that the gold market anticipates a rise in the future marginal cost of supply, perhaps driven by lower real interest rates, speculators will likely begin bidding up longer-dated contracts, increasing the carry in the curve. The higher carry motivates investors to hold more inventories. Consequently, the anticipation of higher prices in the future leads to higher inventory levels today.

商品市场中，高库存水平与低价明显相合，库存的高水平代表着需要被市场消化后的剩余。虽然这是典型情况，但也不总是如此。库存水平和商品价格之间的相关决定于市场是否受当前供需平衡或者预期的未来供需平衡的影响。比如，假设黄金市场预测未来供应边际成本会上升，也许是受较低实际利率的驱动，那么投机者将会开始哄抬远期合约的价格，增加曲线中的持有收益。更高的持有收益会刺激投资者持有更多库存。因此，对未来较高价格的预测会导致当前库存水平增加。

This same phenomenon was observed in the oil market during the run-up in long-dated oil prices (see GS Energy Watch: “ Long-term shortages create near-term surpluses ” July 5, 2006). Interestingly, because the near-term gold supply and demand balance is relatively stable, leaving little need for gold inventories as a buffer stock, the role of inventories in anticipating price movements is far more evident.

在远期石油价格上涨时，同样的现象也出现在石油市场。有趣的是，因为近期黄金供需平衡相