The Masterpiece Effect, and the Market Efficiency of Art

In the empirical research regarding returns from art investments, an interesting phenomenon has been observed called the masterpiece effect. Intuitively, if art pieces are indeed believed to be “masterpieces”—or works of exceptional quality or renown—then we might expect the returns to investing in these art pieces to uniformly outperform the general portfolio and market. However, James Pesando, Jianping Mei, and Michael Moses (among others) have found that masterpieces tend to underperform the market and, in fact, provide lower cumulative returns than non-masterpieces.

Many economists have contributed their thoughts to explain this phenomenon. For example, some believe it’s due to overbidding followed by mean reversion. Thus, masterpieces outperform in one period—we could theorize the one in which their “masterpiece” status was originated or consolidated—and then underperform once they’re more established and change hands less frequently (presumably, these pieces would be coveted and thus not traded as often). Others suggest that masterpieces are less risky because they’re more liquid—they may not trade as often, but are definitely easy enough to sell in the market when they do enter it.

I tend to sympathize with this second theory the most (though the first also has its merits and, in reality, probably explains some portions of this effect as well). It is elementary intuition in financial economics that lower risk involves lower returns. To that extent, non-masterpieces would provide higher returns because they’re indeed riskier than the established pieces of art. For these riskier assets, you can buy at a low price and, with luck, sell at a much higher price later given changing art tastes (meaning, you’re lucky to buy “speculatively”—buy an emerging or obscure artist’s work in the hope she will catch on in the art market—and then sell when your prediction has come true). Yet, of course, this comes at the high risk that this lesser-known work will not in fact sell, or will not “catch on”.

But, masterpieces should in theory always be considered eminent, and as such are less risky. Thus, you’d buy at a high price and, technically, expect to sell at a similarly high price. Their very definition of masterpieces, after all, means they’re tried and tested works. Because tastes regarding established works don’t change much (after all, that’s why they’re “established” parts of the art canon), the only factor affecting increases in prices for these pieces should be inflation, allowing perhaps for slight changes in the interest of the artists at a given time (which reduces the investing game to simply having a sense for when an artist is being paid more attention to).

Given the above discussion, the masterpiece effect almost becomes a market efficiency question, in that masterpieces could be considered assets that trade “efficiently”, while non-masterpieces may not. Applying the concept of an efficient security to artwork, because of their very status as masterpieces we can presume we know most if not all possibly available information about these works and their artists, so nothing new or material (despite, perhaps, deterioration of the work itself or discovery as a fake, etc.) should ever come out about that work of art. Thus, because prices for an asset that trades efficiently should only adjust to new, material, and public information, we should expect prices for masterpieces to change very little over time and thus, these works to provide very low (if not zero) returns.

Because we may lack much more information on non-masterpieces, and because there is a higher likelihood that some particular investors or participants in the art market receive more or better information on them than others (a specification falling more under the “strong” form of market efficiency as defined by Eugene Fama), non-masterpieces may thus show inaccurate prices and allow for outsized returns that deviate from their true value, as compared to masterpieces.

Ashenfelter and Graddy summarize James Pesando’s discussion of the market efficiency question: when pieces trade efficiently, “the market should internalize the favorable properties of masterpieces into their prices, so that risk‑adjusted returns should not exceed that of other pieces.” Of course, here Pesando explains the masterpiece effect without relying on the inefficiency of non-masterpieces. In fact, for Pesando, it is because the market is efficient for both masterpieces and non-masterpieces that the former do not demonstrate higher returns than the latter (for non-masterpieces, there's simply more "new" (presumably good) information about them coming to the market, so there is more positive price adjustment for newer, non-established works as opposed to masterpieces). I might claim that the masterpiece effect is observed perhaps because market efficiency breaks down for non-eminent pieces of art (again, we can think that assets that are traded very infrequently, that are paid comparatively little attention, and for which there exists sparse information, would trade inefficiently compared to those better-known assets, in the context of the art market).

Of course, this entire discussion relies on some critical questions: firstly, empirical research into this topic requires us to appropriately control for survivorship bias. After all, as mentioned above, masterpieces are presumably much more liquid than non-masterpieces. As such, they are likely to sell much more often so that, when we consider all the non-masterpieces that don’t “survive” in the art market, the cumulative returns of these non-masterpieces may ultimately be below that of the more reliable masterpieces, making the latter ultimately still the better investment.

Secondly and more philosophically (but with much relevance to econometric models), how do we even define masterpieces? The results of any models will ultimately rely on what works are defined as masterpieces, be it through a dummy or other methods.

Lastly, when does an art asset actually trade efficiently? How do we go about showing that an art piece or certain sectors of the art market trade in an efficient way? The “weak” form of efficiency is easy enough to think about: after all, a “weakly” efficient asset is one whose returns cannot be predicted using standard time series methods (in this form, the information set is only past historical prices of the asset).  In the "weak" form, asset prices follow a random walk and only respond non-randomly to new information. Stronger forms of efficiency, however, don’t extend as easily to the art market. Conceptually, showing “semi-strong” efficiency would require us proving that prices for these pieces respond as expected to new, material information about these pieces. Yet, an art piece by definition shouldn’t really change much. Thus, there should be very infrequent new information about the piece to move the price. In turn, how would we go about conducting an event study of an art piece’s returns?

This lastly brings us to the most philosophical question of them all: why do prices change so dramatically for artworks at all? Other than inflation, why would a Picasso 50 years from now sell at a much higher price than today? The most obvious answer would simply be changing consumer preferences: perhaps 50 years from now, Picasso is even more popular than he is today. Yet, how do we calculate when the popularity of an artist has changed? How do we define that popularity to then measure and apply to the art market equivalent of an event study? How, at the end of the day, do we know where art prices should adjust—what the true, accurate value of an artwork should be—when art itself can transcend all attempts at understanding?

Noon: Rest from Work (after Millet), Vincent van Gogh (1890)

The Value of Art

A common topic of interest in the field of arts economics is that of art as an investment.  For a great survey on issues regarding art prices, returns, and, in turn, investment potential, see “Art Auctions” by Orley Ashenfelter and Kathryn Graddy in the Handbook of Economics of Art and Culture.

Among the many things this article covers is the topic of finding the value of art (in terms of prices), with the end goal of considering art returns, and how the auction process informs price formation.

When we think about valuing a piece of art (in this case, finding the price one should pay for ownership of that work), we could think of it as we do more “traditional assets.”  Thus, we could think of “comparable” works that have been involved in “past transactions” that would allow us to get an estimate for the value of the piece in question. Alternatively, one can conduct a valuation for a piece inspired by a type of discounted cash flow method. In other words, one could evaluate the kind of “cash flows” one would obtain from a given painting (and here, “cash flow” can be abstracted heavily to “value” in general—both monetary, cultural, and otherwise) and the risk involved with owning that painting, essentially valuing the piece of art as the sum of the discounted value you’d obtain in future time periods from that work.

Now, thinking of art valuation less traditionally, one can think of what is known as a hedonic model. This model, very simply, regresses observed prices on characteristics of the respective works. In short, it would give coefficients on different features of a piece of work. In this case, one would preferably estimate different models for different mediums, since different characteristics can have different impacts on the value of a painting based on the type of work—one can imagine that a dark red color would have a much more positive impact on a painting than on a marble statue. In turn, simply adding a dummy variable for different mediums to a regression that encompasses different kinds of arts may not result in correct coefficients (we can also think about the impact this could have on the standard errors of these coefficients, if perhaps some mediums have more widely dispersed price observations for the same variables than other mediums).

The way that a hedonic model would be applied then (if these models were to be used predictively or prescriptively—which both philosophically and economically may involve some issues) would be to input the value of the variables for each work in consideration to output an estimated price for that work. In short, an appraiser would add up the sums of the “values” of each characteristic of the work to reach the piece’s final value.

Of course, a hedonic model based on panel data could be adapted to a fixed effects model in order to control for the different perceived values of “quality” of each painting. In theory, though, paintings with identical values of the dependent variables should, by the definition of this model, have identical “quality”, a question that brings us quickly to the more transcendent and literally “price”-less dimensions of art. After all, what defines “quality,” and shouldn’t by definition the coefficients of the model capture it by defining the “value” or quality of a work given its characteristics? Why are some paintings that are objectively similar to others worth much more (or treated as much more higher-quality) than others?

Alternatively, another regression model to value art would be the repeat-sales model, which is frequently used in another “alternative” asset class: real-estate. This model is perhaps better suited to construct indices of art prices overall, as opposed to valuing individual pieces of art. It controls for the mix of art products being considered (in short, the quality of the works of art in given times) by only considering works that have been sold more than once (an index that does not do this could perceive an increase in the “price of art” that is actually only capturing the entry of new, high-quality or fashionable pieces of art that, because of their quality, would be increasing the index).

Of course, an issue endemic to the repeat-sales model (and generally most regression models observing prices at sale) is that of survivorship bias: a model may be overestimating prices of art because it only observes those pieces that were in fact sold (meaning, those that have “survived” in the market). By definition, a model that only looks at sales would not be accounting for the multitude of art pieces that failed to sell, so coefficients would likely be overestimated given this bias.

And of course, an even broader issue with valuing art is the major philosophical question:  can we even put a price on art? Clearly, auction houses, galleries, dealers, and independent artists have been putting price tags on their work, and this is mostly for good reason. Artists (and the market around them) deserve to make a living off of their labor and the value they add to culture and society.  So, of course, art should never really be “free”: it always adds some value. However, art can in many occasions be “priceless.” And this is where the major issue arises: will art valuation ever truly systematically, consistently, and accurately capture all the value an art piece offers to the world? More deeply, how can we even calculate the value an artwork provides? Beauty (and I use this term very generally, to encompass all forms and styles of “beauty”, including the grotesque and ugly, the conceptual and the performative), after all, is in the eye of the beholder. What kind of factors should we include and, by attempting to run models and quantify values are we not imposing a norm on what actually “counts” to price a piece of art?

Why should we limit and define the characteristics that have value to us humans when, often, art is transcendental and beyond the features we can visually discern (or perceive with different senses)? And who’s to say our calculated values should apply to all (or on that note, any) piece of art? That a blue is worth more than a red? Or a larger frame more than a smaller canvas? That a painting’s exposition in the Met or its more prestigious provenance or artist makes it a more “valuable” painting than the one by your grandfather, sitting over the mantle at home? These and many questions, like art itself, are perhaps beyond the limiting interpretations and assumptions of the human mind.

Life in Economic Consulting

About two months ago, I began my first full-time job as an analyst at an economic consulting firm. As a member of the team, I conduct economic analysis on a wide range of issues across the lifespan of commercial litigation and regulatory proceedings. Located at the unique intersection of the law, business, finance, and economics, economic consulting firms aid experts in their analysis of the case in question. Under the experts’ direction, we apply both quantitative and qualitative methods to uncover the economic phenomena at play in a given case. As such, these firms are a great primer for a career in research in any of the fields covered by practice areas of the firm.

So far, a job in economic consulting has allowed me to think much more deeply about applied issues in finance and economics. In particular, I’ve expanded my repertoire of research skills and approaches that should be useful for any field of study. For my preferred areas of Industrial Organization and Arts Economics, the kinds of cases economic consulting firms are retained on have already inspired me to think about the role of information in financial decision-making, and of the limits and boundaries of intellectual property rights.

For example, what do investors consider to be material information? When and how do they use information about a financial instrument? When are expectations about an instrument absorbed into its price, if at all? When an expectation becomes a confirmed (or a disconfirmed) fact, should we expect a movement in the price of that instrument? And of course, the classic question: are investors entirely rational?

As for intellectual property rights—and this relates to my previous post on this blog—what are the boundaries between creator and contributor to a given work? Particularly when it comes to the issue of copyrights for creative works—how small of a change in somebody else’s work is enough to grant the editor a claim over the transformed work? Can “concepts” ever be fully claimed by an “owner”? And how long should copyrights even last?

All in all, I’ve thoroughly enjoyed the work, the people, and the culture of the last two months in economic consulting. I’m eager to learn many more skills and gain an even better understanding of the quantitative and qualitative methods involved in researching topics in the law, finance, and economics. Economic consulting is highly recommended to those with some interest in any of these fields.

Any views expressed in this post regarding the work involved in economic consulting are mine only and do not necessarily reflect those of my peers or the firm I work for.

Intellectual Property in the Art Market

At this year’s Frieze Art Fair in May, Richard Prince sold unaltered screenshots of others’ Instagram pictures (without permission) for upwards of $90,000 each. His contribution?: some obscure, Instagram comments on the bottom of each screenshot from what appears to be an account he owns: richardprince1234. After a private exhibition at Gagosian earlier in September of 2014 that gave way to much criticism, Prince had the art market once again debating issues of authorship, authenticity, and originality.

This occasion joins many other instances where the line between artist and seller is incredibly blurred and contested. After all, whom does art belong to? Who transforms an ordinary object—such as Warhol’s soup cans, or Duchamp’s “Fountain”—into a piece of art? And then, who can lay claim to the authorship of that work and, even further, sell that piece of art to others (let alone at exorbitant prices)?

Like many other, more “traditional” products, art can be studied from the perspective of value added. After all, no one protests that the final products we purchase on retail go through what is usually a long, complex supply chain where subsequent producers, dealers, wholesalers, etc. add a feature, a service or a physical transformation to the intermediate good they acquired, to then sell on to the next producer in the chain and, eventually, the end-consumer. Yet, nobody protests because this is both established practice and, most importantly, an arguably fair practice. The intermediate producer gets paid by the supplier further downstream for the service he performed on the product. Leaving aside those strategies frequently studied in the vertical relationships section of an Industrial Organization course, we could argue that all producers who had a role to play in the production of a product get fairly compensated for their value added.

The question in this case—that of the art market—is how much value is enough to transform someone else’s work into one’s own? And how little value added (in this case, some comments) is needed to transform an Instagram picture into a $90,000 work? How should the $90,000 price of that artwork be broken down between Richard Prince, the end-supplier, and those Instagram users who added their value in creating and uploading their image to the website? Didn’t they, after all, add much more value than Prince himself did? Or are Richard Prince’s contacts with the art market—his ability to be exhibited at Gagosian and at the Frieze, to print those snapshots and have them set up, worth incredibly large amounts of money? Is his name, the service he’s adding in getting these snapshots a fine art exposure, worth thousands upon thousands of dollars (these same questions could perhaps be asked, interestingly enough, of luxury fashion retailers: does the brand add that much value)?

And ultimately, this can lead us to ask: if Richard Prince’s service and value added are really worth that much, why is art fetching such high valuations? I’m certainly not the first to question art prices—Qatar is reported to have purchased Cézanne’s The Card Players for upwards of $250 million in a private transaction. The literature is extensive and often, understandably, must rely on the philosophical: art, after all, appeals to a part of us as humans that is perhaps “irrational”, that is “priceless” and invaluable.”

Yet, if that is true, and if simply adding several comments is enough value added to fetch thousands of dollars based on other people’s works, maybe it’s time we start copyrighting and trademarking every single thing we do.