-- John Kemp is a Reuters columnist. The views expressed are his own --
By John Kemp
LONDON, June 30 (Reuters) - The idea that prices for dollar-denominated commodities should move inversely with changes in the dollar's value is beguiling but wrong.
If commodity prices have been moving more closely in line over the last four years, that is more likely to be accidental, or the result of traders believing the relationship is true -- a belief that becomes self-fulfilling if enough market participants act on it. There is no statistical evidence for a fundamental, causal link.
The belief commodity prices should move inversely with the dollar is a version of the quantity theory of money (MV=PT) and stems from the idea that "money is a veil" [ID:nLP411966]. The basic intuition is that the "real" price of a commodity should be set by physical supply and demand. It should not matter whether it is expressed in dollars, euros or conch shells.
Changes from one currency to another, or in the value of the measuring currency, alter the units, but not the fundamental price. So if the dollar is devalued and worth half as much as before, the price expressed in dollars should double, leaving the "real" price is unchanged. The theory is even supported by a causal mechanism:
* Commodity producers outside the United States and the dollar-zone are assumed to respond to a dollar decline by pushing for higher selling prices to compensate for lost purchasing power and continue to cover their (unchanged) costs for items such as labour and electricity in local currency.
* Consumers are assumed to respond by bringing forward or even increasing purchases of suddenly cheaper raw materials.
The incipient supply decline and demand increase raise selling prices until the whole of the exchange rate change is offset, and the "real" price is back to where it started.
NO FUNDAMENTAL LINK
It is a nice theory. Unfortunately there is no real evidence for it.
http://graphics.thomsonreuters.com/ce-insight/CURRENCIES-AND-COMMODITIES.pdf
(1) Most of the marginal demand for commodities such as oil, copper and iron ore comes from China and other emerging markets that have pegged their currencies to the dollar. Their purchasing decisions are unaffected by changes in the exchange rate.
(2) Only a small share of producers' costs is sensitive to exchange rate changes. Mining and oil production is a capital-intensive business. While operating costs (electricity and wages) may be denominated in local currency, they account for only a small share of total expenditure. Most of the borrowing to cover capital costs of exploration, development and infrastructure is done in international markets in dollars.
(3) Most international trade transactions are still invoiced in dollars, even when the United States is not involved as exporter or importer. Commodity producers complain about the loss of purchasing power when the dollar declines versus the euro or the yen, but it is largely notional rather than real, since invoicing prices for other imports are unlikely to change much in most cases.
(4) Correlations between commodity prices and the trade-weighted value of the dollar have varied over time and are simply not stable enough to have much explanatory power. Aluminium prices show a significant negative correlation with the dollar only after 2004. The correlation between oil prices and the dollar was actually positive (i.e. the opposite of what theory would predict) between 1990 and 2003 before turning negative from 2004 onwards.
(5) There is no sign of short-term, causal correlations between changes in commodity prices and changes in the dollar's value over one-day, one-month or even six-month horizons. Correlations are only really evident over periods of 12 months or more.
(6) Correlations with the dollar vary between commodities in ways that appear arbitrary. While most commodities have displayed strong negative correlations with the dollar in the last four years, prior to 2005 there were strong correlations for copper, lead and tin, but weaker ones for aluminium, zinc and oil, with no apparent reason for the difference.
(7) Dollar depreciation does not appear to affect commodity prices equally, even when all items are priced in the currency. Steep rises in the price of crude oil, iron ore and copper during 2005-2008 were not matched by rising prices for autos or computer chips.
(8) Differences in producers' pricing power are necessary and sufficient to explain why some commodity prices rose strongly in 2005-2008 (oil, iron ore, copper) while others did not (aluminium and computer chips). Exchange rate movements add nothing to the analysis. Using Occam's Law (which suggests simpler explanations should be preferred to more complex ones) exchange rate changes should largely be ignored when analysing commodity price movements.
(9) There is simply not enough pricing history to draw meaningful statistical conclusions. That might seem surprising since we have more than 4700 daily data points in the set (1990-2009), more than enough for a meaningful analysis. But the points are not truly independent. The period under examination consists of only one exchange-rate cycle and one commodity price cycle.
To draw strong conclusions we would need to analyse commodity prices over several cycles. Unfortunately, oil futures are only available from the early 1980s and base metals were not traded in dollars until the late 1980s.
SELF-FULFILLING TRADES
While there is no proof of a fundamental link, commodity prices have been moving much closely in line with changes in the dollar's value since 2005.
One explanation is that this is the period which corresponds to the dollar's plunge to historic lows. Casual links (such as producer costs) which had little or no effect at "moderate" levels for the dollar have become much stronger at extreme low levels. In other words, there is a fundamental correlation, but a non-linear one and it has more impact at extreme values.
Another is that all financial markets have become more "integrated" in recent years as a result of globalisation, improvements in financial and communications technology, and the influence of increased wealth management, hedge funds, pension funds and investment banks pursuing similar strategies in formerly separate markets.
The result has been a noticeable increase in correlations across all asset classes. Commodities are simply part of the trend as they have attracted increased interest from investors and increasing trade as financial instruments rather than physical raw materials.
Especially in the run up and aftermath of the banking crisis, all correlations seem to have gone to one, nullifying the hoped-for benefits of diversification and explaining why commodity correlations have risen so high in the last four years.
The last explanation (and consistent with the other two) is that if enough traders behave "as if" there is a link between commodity prices and the dollar, the belief will become self-fulfilling. Such self-fulfilling correlations are too powerful to ignore, and exert a strong influence on the market, until such time as they break down, to be replaced by an alternative proxy for traders to track and use. (Edited by David Evans)