Fine art has received a great amount of attention over the past year or two in sophisticated financial circles as the next tangible alternative asset, a viable portfolio diversification that reduces risk and increases expected return. Some enthusiasts have predicted that we will see up to 30 new art funds by the end of 2005. But the provocative question remains -- where is the money?

A recent Wall Street Journal article by Marcus Baram, titled "Art Funds Starved For Investors," describes the anemic state of capital raising that currently plagues the short list of pioneering art funds. The names in this article are familiar from earlier reports on the subject of art as investment: the Fine Art Fund, Fernwood Art Investments and ABN AMRO, plus a few lesser known players, including Falk Art Management and the Collector's Art Fund.

Mr. Baram cites several generally acknowledged problems with translating high levels of investor interest in art into meaningful shifts in investment capital -- a lack of transparency in the industry, high transaction commissions, and a reliance on auction records alone for market data. Add to this list the dearth of art fund managers with track records, and the increasingly bold display in art fund sales materials of indices and market measurement tools that have methodological weaknesses.

Among private investment fund insiders, especially on the institutional level, showing a track record of delivering superior returns to investors goes a long way toward attracting new capital. Track record is far more important than ingenious investment philosophies and strategies which, until proven, remain hypothetical. Most articles these days that fan the flames of the booming potential for art funds, and most private art fund prospectuses, cite the impressive 11.3% annually compounded rate of return generated by the British Rail Pension Fund from 1974 to 1999. Aside from the fact that only one percent of the Fund's 2,500 art objects ultimately accounted for the positive return, what's new since last century?

Private dealers have raised money for decades from collector clients and circles of friends to buy art, with varying degrees of success. Furthermore, a handful of hedge fund managers and smaller funds that are not generally on the investment media radar screen have enjoyed enviable returns from art investing. However, taking the fund business to the next level, and soliciting not only high-net-worth individuals beyond family and friends, but also pension funds and other forms of institutional capital, requires a formalization of presentation and packaging that includes a detailed history of successful investing.

Thus far, the Warren Buffet of art fund management has not emerged. Phillip Hoffman of the Fine Art Fund, perhaps the only highly visible art fund that has closed on substantial new investment capital, shows fresh evidence of art investment acuity. Recent announcements from that organization cite several profitable short-term transactions in art objects. However, until the dust settles over the life of any fund after fees and the carried interests of fund managers, an attractive internal rate of return is impossible to predict.

For the sake of credibility, aspiring art fund managers should not create synthetic track records and are therefore excused if a verifiable history of success does not exist. In turn, they can control the spin in sales materials and presentations. When a prospectus insinuates a level of development and sophistication in art market measurement tools that portray art as irresistibly attractive for investment, it may be doing more harm than good. Aping the use of price performance indices in sales presentations by professional fund managers in traditional securities (S&P 500, Lehman Bond Index, etc.), the new breed of art fund managers has taken to wearing art indices and market measurement tools on their sleeves.

Casting one's investment strategy in the context of market trends shows a command of available information, theoretically alleviating at least one concern of investors. However, all of these art indices have methodological weaknesses that institutional investors are perfectly capable of ferreting out, especially when the investment proposal involves substantial sums. The root of the problem lies in the heterogeneous nature of art market data when attempting to construct a true price performance index.

To put the issue in its simplest terms, one share of IBM common stock has a market value at any given moment because all IBM common shares are alike and they collectively represent the value of IBM. Because of this homogeneity, alert Finance 201 students can accurately generate an IBM price performance index. However, making sense of art transactional data, like real estate, involves comparing apples with not only oranges, but with a wide variety of exotic fruits and vegetables. In this realm of unlike objects, how does one reconcile the fact that at any given moment, one 36 x 24 inch Monet landscape will have a fair market value that is dramatically different from that of another 36 x 24 inch Monet with a similar motif from the same approximate period?

The subjective interpretation of variables affecting the value of unlike art objects, such as condition, provenance, scarcity and rarity, remains the leading obstacle to creating universally accepted art market indices and market measure tools.

Pioneers in unlike object art data interpretation have made impressive progress over the past several years. Repeat sales regression, the approach championed by Michael Moses and Jianping Mei and used in the Mei Moses All Art Index, has scientific validity. Repeat sales regression measures price changes by tracking individual artworks that appear and sell at auction more than once. Transactional "pairs" comprise a general art price performance index, and if enough data avails itself, performance indices for specific art categories such as Old Masters and Impressionist Art.

The primary problem with applying repeat sales regression to art data is the paucity of available information. The Mei Moses team has identified approximately 8,000 pairs since 1875 and from this sample of pairs projects a population estimate, proposing that art as an asset class has performed approximately as well as stocks and better than various other asset categories. Aside from what statisticians call potential selection bias problems, do 8,000 pairs provide any practical guidance for the art fund manager confronted with a myriad of unlike object buying options? The artnet Price Database has nearly three million auction records compiled since 1985, one indication of the vast amount of useful market information that repeat sales regression ignores.

The repeat sales regression method does, however, bring fine art to the attention of serious quantitative analysts and gatekeepers of institutional capital, providing strong evidence that art has positive expected returns. Once refined and possibly bundled with other specific art knowledge and financial techniques, it may enable further development of portfolio management tools that measure such things as volatility and correlations with other asset categories. Always eager for investment diversifications to enhance returns, institutions only require industry expertise, supported by reasonably correct analysis in their language of statistics.

Another popular approach, used by such online art information providers as Artprice and Art Market Research, averages prices of transactions in the public auction market. For example, a Picasso index will aggregate all Picasso sales in a given period and divide by the number of lots sold, producing an average price for Picasso art that oscillates through time. This method typically excludes outliers (the extraordinarily high- and low-priced transactions), and practitioners often use moving averages techniques to smooth the inherent dramatic volatility in auction market price movements.

The averaging approach, without addressing the apples to oranges issue, and especially in cases of excessive pruning of outliers and smoothing, borders on alchemy and irrelevance  as a true price performance index. Averaging is particularly dangerous in cases of sparse auction transaction data. The market may have tens of thousands of transactions in Picasso works -- perhaps the closest the art industry comes to an efficient market -- but what do average prices mean for artists with only a few auction transactions per year of pieces with wildly varying characteristics?

Faced with the difficulties of true fine art price performance index construction, a more straightforward description for an index using average prices relies on pure normalized information. Normalized information equates data of any kind to a base-year starting point (usually zero or 100) and then shows fluctuations through time as the underlying data changes. Without eliminating the outliers and without using smoothing, normalized data shows the real volatility in the art market. Normalizing multiple art auction data sources on the same graph, such as total sales, total lots and average prices, provides a visual snapshot of related market dynamics for analysts.

When art market data interpretation recognizes its limitations, it can be useful in conjunction with specific industry knowledge. Indeed, these normalized market indicators, especially if modified with a hedonic element or two, may become useful predictors of art price movements.

The hedonic method tops the wish list of most data miners attempting to develop widely accepted portfolio management tools for unlike art objects. As a statistical technique, hedonics applies qualitative judgments to certain defined attributes in each unlike art object. These range from the easily measurable -- such as size, medium and date -- to the more complex such as condition, provenance, scarcity and rarity. Individual pieces scoring extraordinarily high or low on an average scale for any of these attributes would be weighted accordingly in the construction of an art price performance index.

The hedonics methodology has the potential to claim the accuracy of repeat sales regression in comparing apples with apples, while enjoying a much broader utility of available information. To date, no think tank has gone into the level of detailed analysis of auction data applying hedonic techniques that warrants a breakthrough announcement.

Many believe that art investment, whether in fund form or more traditionally, has a bright future. The few successes in capital raising for art funds represent the tip of an iceberg as methodologies improve and track records become available. Art industry insiders have known for years how to buy low and sell high. The interpretation of art data is a new field but pioneers will apply creative approaches, mixing statistical techniques and curatorial-level knowledge to sharpen market measurement tools. Institutional capital, always interested in above-average returns, will embrace the proper approach.

KEVIN RADELL headed artnet's product development strategy for a new family of market research and fair market value reports for art collectors and investors. For more information, please feel free to call (212) 497-9700 ext. 383, or email .