Master thesis
Pablo Picasso
Student number: 1323458
Supervisor: Prof. Dr. R.A.H van der Meer
Program: MScBA Finance
Institute: University of Groningen
Date: December 12th, 2008
The Effect of Art
Exhibitions on Sales Prices
Abstract
Art exhibitions provide artists an exceptional opportunity to gain prominence. La Biennale di Venezia is the most historical exhibition for modern and contemporary art in the world and is organized once every two years. This paper assesses whether this exhibition has a significant positive influence on prices of paintings of invited artists, which is related to the reputation they obtained. A dataset consisting of 1209 auction records, acquired from 111 oil painters present on one of the exhibition years from 1995 to 2007 was compiled. Indices and geometric returns based on sale statistics were constructed and calculated. Hedonic regressions were ran to adjust sales prices for the painting‟s qualitative characteristics. From this, a price development of a „standard‟ painting is presented. The qualitative characteristics consist of the auction house and place, the surface of the painting, the presence of a signature, whether the painter is still alive and of the reputation of the painter. The annual return that resulted from the hedonic regression yielded -4.71% in the 3.5 year period before, 15.86% in the 3.5 years after the exhibitions, and 5.074% for the total period. Return analysis for a sample of only living artists indicate an even higher influence of the exhibitions on the selling prices. The general conclusion is that prices of paintings from invited artists show a substantial upward development after his/her presence on the exhibition--hence, making a very interesting factor in art investment strategy.
Acknowledgements
In order to write this thesis, I have made a tour along various interesting people who have been given their helping hand. The result of my thesis wouldn‟t have been this successful without the following persons and institutions:
Léon van der Maesen de Sombreff, who opened my door to the world of private banking where it all started. Fortis MeesPierson, that offered me an excellent internship. My both supervisors Robert van der Meer and Sylvia van de Kamp-Vergeer, who provided me from feedback and guidance during the whole period. Max Bevers, Veronique ten Baar and Jeannette ten Kate, who provided information and informed me about their contacts. Roman Kraeussl and Robin Logher, who gave essential input by introducing me into the economic world of art investing and presented their broad knowledge. Furthermore, I want to show my gratitude to Donna Wolf (founder of Deïska, an organization that provides business services to talented young artists), Kristie Wilson, and my uncle, Lex van der Sommen and his colleague Paul van Heekeren, who granted information about investments and investment funds in general.
Then my last, but definitely not least thanks go to my family and friends who greatly encouraged me during the whole process.
Thank you all for this great support and hope you enjoy reading this thesis.
Tessa van der Sommen
12 December 2008, Amsterdam
Preface
Table of Contents
I. INTRODUCTION ... 5
II. ART INVESTMENTS ... 9
II.1 ART VERSUS STOCKS ... 9
II.2 ART PRICE INDEX ... 10
II.3 ART FUNDS ... 12
II.4 ART EXHIBITIONS ... 15
III. DATA AND METHODOLOGY ... 19
III.1 DATA ... 19
III.1.1 ARTISTS ... 19
III.1.2 AUCTION RECORDS ... 20
III.1.3. TIME LINE ... 21
III.2 METHODOLOGY ... 22
III.2.1 SALES VOLUME AND PRICE ... 22
III.2.2 HEDONIC REGRESSION METHOD ... 23
IV. EMPIRICAL RESULTS ... 27
IV.1 SALES VOLUME AND PRICE EFFECT ... 27
IV.2 HEDONIC REGRESSION ... 30
IV.2.1 PSYCHICAL CHARACTERISTICS ... 30
IV.2.2 NON-PSYCHICAL CHARACTERISTICS ... 31
IV.2.3 PRICE INDEX... 32
IV.3 SUB INDICES ... 34
IV.4 ROBUSTNESS ANALYSIS ... 35
V. CONCLUSIONS AND RECOMMENDATIONS... 39
REFERENCES ... 43
APPENDICES ... 45
I. Introduction
In November 2006, Jackson Pollock‟s painting named No.5, 1948 was said to have been purchased for $140 (USD) million dollars, making it the highest recorded sale on a single piece of art. Second in ranking is a work of art from Willem de Kooning, Woman III, which was sold for $137.5 million dollars in 2006 to Steven Cohen. Both were private sales, whereas the number four on the list is sold at Sotheby‟s New York, and is therefore the most expensive painting sold at auction. This is a painting of Pablo Picasso (named Garçon à la pipe) and was sold for $104.1 million dollars1. These enormous amounts of money arouse interest for taking a closer look into the financial world of art.
The last ten years, the development in the art market has been followed by investors nearly as close as the stock market. Art is no longer seen just as an aesthetic object, instead, it has growing attention as an important alternative asset class. The value of the art market is over three trillion (USD) dollars and traders in this industry generate a yearly turnover of thirty billion dollars2. Even fine art indices exist measuring the performance of paintings over time based on the sales prices of works sold at auction houses. Although art has been debatable for quite some time, Mamarbachi et al (2008) illustrate that the market has performed and still performs well in the previous and present economic downturn. With the uncertainty of stock returns and extremely high interest rates, anxious investors are seeking alternative investment opportunities with long term stability. And when markets are volatile, people prefer to invest in a fixed and tangible asset. It is therefore not surprising that the value of contemporary art extremely increased over the last years. Not every segment of art, however, has experienced growth. Since art exists in various forms3 made by either the same or multiple artists, one should carefully study the art before making the investment. In this paper, art is synonymous to the paintings that will be studied.
The emergence of the art market has provoked a small but rising amount of research studies. The researches take the perspective of an investor as principle and primarily focus on the return art investments provide. They frequently investigate whether art is a good investment and make comparisons to the returns on stock and bonds. Other articles focus on the psychical returns on art market investments and compute correlation coefficients to other alternative asset classes like gold and private equity. The following part discusses the main studies with their obtained results regarding art investments.
1 The list of „top ten most expensive paintings sold‟ is available at www.theartwolf.com, and is last
updated in May 2007.
2 Source: Thomson financial data, 2006.
Baumol (1986) wrote the best known article that examines the return of investments in paintings and which consequently opened a door for future study. He found a real rate of return on paintings of 0.55% per year in the period from 1652 to 1961. This outcome is relatively low if one compares it to the 2% annual returns from bonds in the United States during the same time period. Buelens and Ginsburgh (1993) revisited Baumol‟s pessimistic conclusion and discovered that some segments of paintings did outperform the stock market over a time period of 20 to 40 years. They presented results that clearly illustrate how the performance of art prices was strongly heterogeneous. For instance, from the period of 1652-1961, the nominal rate of return for an investment of an English painting was 0.60% annually, but 3% for an Impressionist. Furthermore, Buelens and Ginsburgh illustrate that the price of an Impressionist type of painting significantly goes up after a painter‟s death, whereas English painters truly found the reverse. The articles from Pesando (1993), Pesando and Shum (1996), Barre et al (1996) Frey and Pommerehne (1989) and a more recently written article of Kraeussl and Schellart (2007) also calculated low returns up to only 1.5% per year. Results from Stein (1977), Chanel et al (1994) and Mei and Moses (2002) parallel with Buelens and Ginsburgh (1993), agreeing that the rate of return depended on the popularity of the style and/or artist. These studies calculated higher real rate of returns varying from 4.9% to 10.5% per year. Mok et al (1993) even found a nominal annual rate of return of 52.9% in the period from 1980 to 1990 when he studied the market of modern Chinese paintings.
Worthington and Higgs (2004) obtained a negative correlation between paintings and small US business stocks of -0.31, and concluded that paintings are a highly valuable alternative asset for an investment portfolio. Kraeussl and Van Esland (2008) and Hodgson and Vorkink (2003) also studied correlation coefficients of art with stocks and found coefficients of 0.19 and 0.44 respectively. Overall, these various levels of correlations can influence the investment portfolio heavily. In any case, results of the existing literature greatly vary and depend on the segment of the paintings, the time period and further specifications of the sample that is used. Generally, returns on art investments appear to be lower relative to the financial market.
Kraeussl and Ensing (2008) introduced a unique approach to studying paintings by exploring the influence of an exhibition on the sales volume and auction prices of paintings. They argued that the attendance of an artist on an exhibition can have a positive influence on the reputation of this painter and therefore on the valuation of his or her work4. To measure the effect, they made use of auction records from painters present on the Documenta exhibition that is a large exhibition for modern and contemporary art. A dataset of 4000
auction records acquired from 198 artists that were present on one of these exhibitions was used and records were obtained from the Art Sales Index in a time period from 1952 to 2006. Prices experienced an annual nominal geometric rate of return of 11.69% in the 3.5 year period after the exhibitions. Kraeussl and Ensing (2008) concluded that auction records from painters present on the Documenta exhibition experienced significantly positive effects by attending the show.
To find out whether this positive influence is also apparent on other comparable exhibitions, this paper studies the effect of the oldest and one of the most popular exhibitions for modern and contemporary art, known as La Biennale di Venezia5. The exhibition takes place once every two years from June until November in Venice, Italy. Founded in 1895, it attracted up to 319,332 visitors in 2007. Because the reputation of painters is directly linked to the prices of their works6 and the positive conclusion of Kraeussl and Ensing (2008), the main question in this research paper is stated as follows:
“Does ’la Biennale di Venezia’ have a significant positive effect on the reputation of the presented artists?”
In order to answer the main question, a dataset comprised of 111 oil painters (whose works were exhibited at least once from 1995 to 2007) was extracted from 1209 auctions records archived in the Price Database of Artnet. In line with the paper of Kraeussl and Ensing, auction records are measured in a sales window from 3.5 years before until 3.5 years after the exhibitions and split up in half-year intervals. The effect on sales prices and volumes is determined by computing the number of sales, the turnover, average and median prices within the sales window. A significantly higher turnover of $170,991,239 in the 3.5 years after the exhibitions is found in comparison to $73,745,748 in the 3.5 years before. The development of average prices for all auction records is determined by a calculation of annual geometric returns and include 41.64% after and -30.36% before the exhibitions. Furthermore, a hedonic regression is implemented to compensate for the qualitative characteristics that auction records bear, so that the development of the price of a „standard‟ painting can be examined. It carries out an Ordinary Least Squares (OLS) regression on the combined records from all time periods where an adjusted geometric return of -4.71% before and 15.86% after the exhibitions is obtained. The quality characteristics are separated in physical and non-physical characteristics. The physical characteristics include the auction houses, location, and the surface of the paintings. For example, the auction records obtained from auction house Finarte
are 24.53% higher than results from other auction houses. Sotheby‟s and Christie‟s obtain 6.14% and 2.6% higher selling prices respectively. Location wise , New York and London experienced the highest auction records. These are 83.90% and 74.07% higher than records from average other cities. The non-physical features can be the presence of a signature on the painting, whether the painter is still alive and the reputation of the painter. For the latter characteristic, a novel approach separating the higher ranked artists from the lower ranked is utilized. Results indicate that paintings with a presence of a signature are surprisingly sold with a discount of 6% and paintings from death artists are 38.74% higher priced than works from living artists.
II. Art investments
Any person serious about art should know that the investment in art requires distinct knowledge in the art of investing. This chapter therefore, aims to compare art to stock investments, to discuss the possibilities of creating an art price index, to argue various existing art funds with their characteristics and finally, to describe the world‟s largest art exhibitions.
II.1 Art versus Stocks
Unlike the financial market, where assets appear to be practically always liquid and investments can be made based on unbiased criteria, the art market can be biased on many factors.. One factor rests on a belief the price of art that is in part influenced from only a small number of significant auction houses. The other factor is verified from a complicated and personal set of values founded on prices (historic, current and prospective), individual feelings and varying market trends. There are other features of art investments drive the market to be inefficient. Such consist of the illiquidity of art investments, high transaction costs, low transparency and large differences in knowledge between customer and supplier6. In addition, art markets are critically incomplete with a small number of works from a certain artist sold each year7.Still, the main and most perceived characteristic is the physical return that art investments create. This return can be obtained from the individual and aesthetic appreciation and thus causes the asset to be much more valuable than if only a financial return is determined8.
In contradiction to the mechanism of securities, the supply of art investments consists of a number of heterogenic works whereby even two pieces of a specific artist are unsatisfactory substitutes. Furthermore, a painting is held by an individual, which suggests a monopoly on this work, while a particular stock is owned by various independent investors in an aggressive market. Another difference between art and stock investments is the frequency of sale moments. Trading of a given stock takes place in innumerable moments of time, whereas, the resale of a painting may not even happen once in hundred years. Additionally, the price that is paid for the art is popular among the persons directly engaged and is therefore not publicly accessible. Besides, when ultimately a price is paid, a person may not even know
6 See Worthington and Higgs (2002). 7
See Rubino (2001) for an overview of the drivers that cause the art market to be inefficient and the equilibration process weaker.
whether it is a forgery or not. In the situation of stocks, investors in principle know what its real value ought to be9. Because of the above-mentioned characteristics of art investments, the equilibration process on this market significantly weakens in comparison with other securities10.
II.2 Art Price Index
A price index was developed as a basis for calculating the return on paintings. There exist different approaches to obtain such an index and to find out which method can be selected best, this subsection compares the various methods. In practice and in existing literature regarding art trade, the focus lies on a construction of (1) naïve price indices, (2) the repeat sales regression and (3) the hedonic approach11.
The creation of naïve price indices that is mentioned first is rarely found in existing literature. This index is based on a computation of average and median prices and represents a general overview of the development of the auction records from the selected paintings. Often, a construction of the naïve price index is the first endeavor to make an art index and is subsequently followed up by the repeat sales regression or the hedonic approach. One reason for inaccuracy of this index is due to an unrealistic assumption that all paintings need to have the same quality12. When computing the naïve price index, a representative sample is selected and the performance is measured over time. The sample needs to be selected randomly or chosen by experts to obtain a representative outcome13.
The Repeat Sales Regression (RSR) is a different approach for creating an art index. The initial purpose of this regression was to measure the price development of real estate14. However, since real estate comprises comparable characteristics as paintings have, and this approach does take into account the heterogeneity of the objects, it is widely used for computing art indices. To compute the index, the method uses the purchase and selling price of separate paintings to measure the value changes over a specific time period15. This means, that only data from paintings that are sold more than once can be used, whereby constantly the price development of the same painting is observed. For every sales couple, a logarithm of the price combination is taken, which denotes a price change in terms of percentages.
9 See the article of William Baumol (1986), which is one of the most cited works concerning art
investments.
10
See Rubino (2001).
11 Existing literature discusses various approaches to obtain an art index, among others the articles of
Fase (2000) and Kraeusll and Ensing (2008).
12 Fase (2000) discusses all pros and cons of the different approaches. 13
See Kraeussl and Ensing (2008).
Enclosed, a regression is run on a set of dummy variables, with one dummy variable for each sale. An advantage of this approach is that an increase in value of one particular painting is measured, where through an adjustment for quality between paintings is not necessary. One shortcoming is that just a minor part of the available selling data can be used, through which possible value oscillations between sales moments do not become visible.
A third and last method mentioned here, is the hedonic regression method. When applying this method, the price is corrected for in quality, size, place of sale, etcetera. From a selected painting in the sample, the various objective characteristics are written and their implicit prices are estimated16. Values of these features are transposed into binary or dummy variables that need to exemplify the shadow prices of these features. Subsequently, these shadow prices are subtracted from the real price, such that a harmonized market price is originated. From the averages of these differences, price of a standard painting results such that the various works can be compared. Subsequently, the price index is calculated from these standard prices. This approach is applied most in earlier researches to measure the return on an investment in paintings17.
To determine the relative performance of the general art development, the obtained price index can be compared to other indices like the S&P 500, the MSCI World Index, 10-year treasury bonds and gold. If we compare the specific returns, risk and the correlation between these indices, an indication of the relative performance can be given18.
The fundamental principles of modern portfolio theory states the assumption that the investor strives for utility maximization of the assets that together form his capital19. The fortune will be invested such that risk and return occur in the most favorable relation. The distribution of the capital over the various assets totally depends on their relative returns, which reflect their liquidity as well. Great uncertainty, measured through the distribution of the return on each asset will increase the required return of the investor. His or her attitude towards risk can alleviate or just intensify this. Concrete analysis of this relation and degree of risk requires specification of the determined utility function of the investment portfolio, because the general formulation that appears in most textbooks no longer suffices. Jiangping Mei and Michael Moses exhibit the consequence of including art on the risk and return of the portfolio and provide optimal allocation percentages of the various assets. Figure 2.1
16 Chanel, Varet and Ginsburgh (1996) extensively discuss the construction of the hedonic price index. 17
The Hedonic Price Index is initiated by respectively Court (1939), Grilliches (1971) and Cramer & Kronnenberg (1974) and, in the beginning, used for car and house pricing. The regression formula such as applied in this paper is in principle implemented by Anderson (1974) and further developed by, among others, Buelens & Ginsburgh (1993).
18
Mei and Moses (2002) discuss and analyze correlation coefficients between art and other asset classes as an investment.
illustrates this effect. They make use of returns on the Mei Moses Family Art Index20, the S&P 500 Index, 10-year U.S. treasury notes, 90 day Bills and gold to figure the optimal asset allocation.
Figure 2.1: Portfolio optimization, risk and return
The figure is depicted from the website of Beautiful Asset Advisors®, and provides information about art as an alternative investment. It illustrates that the return of the overall portfolio increases for the same level of risk when art is included in the portfolio.
Source: Beautiful Asset Advisors – Mei Moses® Art Index
The dotted line represents the finest return attainable for certain levels of risk for the various assets over the last 50 years. The blue continuous line illustrates the effect of an alternative investment in the Mei Moses® All Art Index. The figure obviously shows that adding art reduces risk for an investment in the overall portfolio. The reason for this is the low correlation factor that art appears to have in comparison to other assets. The article of Mei and Moses (2002) illustrates correlation coefficients among real returns of the relevant assets. It also shows a coefficient of 0.04 and 0.03 in comparison to the S&P 500 and Dow Industrial Index respectively over a fifty-year period from 1950 to 1999. Correlation coefficients when art is compared to Government Bonds, Corporate Bonds and Treasury Bills even seem to be negative. These statistics demonstrate the diversification value of art investments--therefore making it an interesting alternative investment.
II.3 Art Funds
Art investments can be made by means of a purchase at an art gallery, via a trader, directly with the artist or a bid at an auction. Another alternative is to invest in an art fund, where deposits are invested on the art market. Such funds are comparable to mutual funds that invest in stocks or obligations. Since art investments are seen more and more as an alternative asset class, a number of art funds have been founded. It is because well-experienced art
20 This index has been created from a suitable database that includes over 12.000 buy and selling
purchases in general make better investments than inexperienced buyers, that an investment in these funds might be attractive. The strategies that are employed by existing art funds are prone to take two different forms. The first strategy is to trade well-known works, like paintings from masters as Pablo Picasso or Jasper Johns. Although they are more expensive, they have already proven their value and are therefore likely to be a relatively safe investment. The second strategy is to purchase, keep and then sell works that are more obviously undervalued. These works are made by artists who are known as possibly important but not of the primary rank. This is a riskier strategy due to the greater uncertainty about the future of these artists, but there is a potential to profit higher than the first strategy.
Suggestions for funds have been given a lot of media interest, but only a couple of them have ultimately started a business. One of the reasons is that funds need investors who are trustworthy and are willing to trust. On the contrary, there exist a number of disadvantages in investing in these funds , making it more difficult to contract investors. Table 1 presents an overview of the benefits and difficulties of investments in art funds.
Table 1: Overview of advantages and disadvantages of investing in art funds.
Advantages Disadvantages
Expertise fund manager Higher costs than 'private' art investments Diversification Long investment duration (often 10 year)
Convenience Illiquidity
No history (no tracking record) Limited information
Relatively high minimum deposit
Source: www.IEX.nl – art funds
The United States and the United Kingdom experienced most rises of art funds21. One of these is the Fine Art Fund of Philip Hoffman that convinced 10 billionaires to invest 70 million Euros in the fund. An example of a raised fund in the Netherlands is the fund of Jeanette ten Kate, named Sinopia, East Asia Fine Arts Collection. This was a pilot for the recently raised Sinopia East Fine Arts Collection II and more than doubled her initial deposits. Due to this success, a much larger successor has been founded last April. It requires a deposit of at least 50,000 Euros and includes duration of, in principle, five years. The Fund invests in various art works from established artists in China and other Asian countries. All artists experienced a verified tracking record, based on expositions, collections, and attendance of the specific artist in museums and other art exhibitions22. The purchased works
21
See http://www.sinopia.nl/media/nrc_apr_2008.pdf.
22 Information about the Sinopia Fund is obtained from the official brochure of Sinopia, Fine Arts
are bundled into one or more art collections, and will be sold as one collection at the end of the duration of the Fund. The various characteristics that art investments bear are processed in the way the fund is established and possibilities are created. One of these is of course the physical return. This characteristic is incorporated in a way that the fund focuses on the creation and trading of art collections and offers investors the opportunity to organize events around these art collections in consultation and cooperation with the trustee to public relations and other devotions during the term of the fund. Investors also have the opportunity to accompany the trustee during purchase trips to renown exhibitions like Art Basel and Hong Kong. These features make the investment in the fund more tangible and attractive. The liquidity point developed by investing the majority in paintings, sculptures and pictures from the established Chinese Contemporary art. In principle, these liquidities are not available for sale throughout the running time of the fund. There is a smaller part of the fund whose family of purchases from other Asian countries can be sold for liquidity purposes. Furthermore, since the success of the fund is, by a large extent, contingent on the expertise of its trustee, finding a qualified successor has always posed as a risk. Additional costs such as transportation, storage, insurances, maintenance, framing and archiving are discounted from the wealth of the fund.
The Amsterdam-based Deïska fund, managed by Dutch Eduard de Geer and American Donna Wolf, provides talented young artists assistance in exchange for their works of art and offers their investors all paintings on loan. The investors who are curious about the modern art can partake in meetings and open studio visits. This way, they become familiar with the Dutch art world by observing work in development; and if proven successful, they can expect a financial profit. The artists selected for this fund are expected to live in the Netherlands and to show substantial potential. A board of eight art experts makes the decision on whether an artist is finally selected Once the artist completes a work of art for Deïska, a monetary value that is based on the labor hours and painting size is assigned to it. In exchange, the artists receive delivery of qualified support customized to their requests. The investors then borrow the works of art for a fixed period of time. Deïska compels an investment of €6,750 and includes duration of five years. It is not possible to step out within this period. After five years, the fund collectively decides whether to go on and grow or to sell the entire collection.
however the Bank discovered the International offer of funds to be too limited at that time23. The objective was to invest in both the British Fine Art Fund and the China Art Fund, allowing them to focus on the secondary art market: private collections and auction houses. The British Fine Art Fund, however, demands a minimum amount of $250,000 from the investor and in turn purchases works of art that vary between 250,000 and 3 million dollar. A much easier and accessible fund is the Houses of Art24, which is also the first fund ever to be established in the Netherlands. This fund requires the same minimum investment as Sinopia, East Fine Arts Collection II, of €50,000. The invested art is valued between €2,500 and €50,000 indicating relatively lower-price works of art. In contradiction to the Fine Art Fund that focuses on artwork with a reputable sales history, the House of Art invests in starting artists who they expect to have a break through. Another difference is that the Houses of Art acquires the artwork directly from the artists such that no gallery or auction houses commissions need to be paid. The investments consist of 10 percent of avant-garde kind and therefore high risk facing 10 percent in „safe‟ artists that have demonstrated their skills. In between investments in young, local familiar artists and national stars are presented. To get knowledge of the quality of these works, the fund co-operates with a team of specialists that experience an insightful expertise in the art market. The investors are allowed to borrow a work of art for one year or to acquire the work at a discount. The purchased works are also presented for international exhibitions.
II.4 Art exhibitions
A superior reputation of traders and auctioneers frequently leads to a higher selling
price for the work of art. Art exhibitions
25offer publicity effectively for various
artworks
26. It is the fundamental place in the symbolic association among artists,
traders, auction houses, academics, reporters, museum keepers and customers
27. They
experience a direct influence on the market price of art through the way the work is
recognized and demonstrated. Large exhibitions for modern and contemporary art
include Art Basel, the Documenta and La Biennale di Venezia among others.
The ArtBasel Miami Beach exhibition, that acts as an artistic and social event, is the most renowned art fair in the United States. It presents a lineup of extraordinary showcases, presentations,
23 See http://www.elsevier.nl/web/artikel/beleggenartistiekepremie.htm. 24
Information about the Houses of Art is obtained from the authorized website of the Art Fund and linked articles.
25 Art exhibitions are conventionally the space in which art works meet an audience. The show is
commonly known to be for a provisional period.
26
The Art Newspaper listed last year almost 150 art exhibitions that are worth going to in North-America, Europe, Russia, China and Japan.
board debates, visits to personal compilations, and crossover affairs with film, music and architecture. The international show presents works from some of the most celebrated and important art-world characters and therefore attracts many thousands of guests annually. Last year alone, the show experienced record breaking attendance with 43,000 guests from each continent and 1,600 journalists worldwide. The exhibition is a sister of Art Basel in Switzerland, whose global selection of art comes from more than 220 top art galleries from North America, Latin America, Asia, Africa and Europe top galleries28. Another exhibition, the Documenta, compares to la Biennale di Venezia and is one of the world‟s most vital presentations for modern and contemporary art29. The exhibition takes place once every 5 years in Kassel and lasts 100 days. For that reason, it is called the „Museum of 100 days‟. Each exhibition, one director is chosen who delegates one specific showcase. The manager is responsible for choosing the artists and illustrating the universal themes involved with the showcase. The various works that are presented comprises of paintings, works on paper, prints, sculptures, miniatures, photographs and film. Since the first Documenta exhibition in 1955, the number of visitors has continually risen up to more than 750 thousand in 200730.
The remaining of this paper focuses on artists presented in the prominent exhibition called la Biennale di Venezia. Because this is the most significant, strategic and glamorously marketed art occasion for modern and contemporary art worldwide and its magnitude and class is comparable to the Documenta exhibition, it is chosen as main subject in this paper. It is therefore interesting to have some background information about the reputation and significance of this exhibition more thoroughly.
La Biennale di Venezia is organized once every 2 years and lasts for 5.5 months. For each showcase, a director is appointed who, together with an international team of experts, chooses which artists to invite. The featured artists depend on the selected themes of each exhibition. It is documented as the oldest artistic happening of its kind and attracts numerous people to each showcase31. The latest version of the Biennale, the 52nd International Art Exhibition, drew in its highest amount of 319,332 guests. While there are plenty of other important and recurrent expositions, none of them weigh against the Venice Biennale in terms of worldwide marketing and earning potential.
The Venice Biennale was founded in 1895 by a decision of the City Council on the 19th of April 1893. They suggested an origin of a national artistic exhibition once per 2 years
28 See http://www.artbaselmiamibeach.com/.
29 See http://www.documenta12.de/geschichte0.html?&L=1. 30
Kraeussl and Ensing (2008) investigated the price and volume effect of the Documenta exhibition and thereby extensively discuss this specific exhibition.
to honor the silver birthday of King Umberto and Margherita of Savoy. In 1895, the event occurred and not only the most important foreign and Italian artists attended, but also uninvited Italian painters and sculptors were present. All artists could join for a max of two works but they were never exhibited before in Italy. Antonio Fradeletto32 was chosen as General Secretary and turned out to be one of the most political powerful persons of that time. He was responsible for the choice of the artists, the set up of the exhibition, and in later years the creation of the overseas pavilions33. There were more than 200,000 visitors at the primary International Art Exhibition in Venice, which was later renamed the Biennale because it took place every two years.
In 1897, Mayor Filippo Grimani replaced Selvatico as the Head of the Venice Biennale and decided collectively with the Jury to set up a Critic‟s Prize. The intention of this prize was to improve publicity via editorials and reviews. This way, the condition of Italian Art comments improved and reached a record in the history of contemporary art reviews.
The Biennale paused during to the First World War from 1914 to 1920. After this period, the Biennale demonstrated an amplifying importance towards the major innovative artistic styles that had been interested in the Impressionists since 190834. In the years that followed, the Biennale established an international reputation by featuring more and more foreign artists. After the Second World War, the Biennale continued motion including French Impressionism, which was presented by Roberto Longhi in an impressive presentation. Rodolfo Pallucchini organized the first five exhibitions after this war as a general secretary. This period allowed him to exhibit a general picture of European avant-garde and in where he was successful in exposé contemporary art more reachable to the Italian community35.
The exhibitions in the 60s were turbulent and imposed modes and styles. Due to this variety and the great number of artists that were invited, the Biennale received many critics. It caused the Biennale to express an Informality movement. Attributable to these violent years, the Grand Prized normally handed out during the exhibition had to be vanished in the 70s. Next to it, the sales office was abolished because of its commercial objective36. In 1972, Mario Penelope organized a much more respected exhibition and was nominated as Special Commissioner for Fine Arts. The years that followed enclosed various themes and attracted a range of artists. The Biennale exhibitions performed worldwide and thereby received global
32 Antonio Fradeletto was a learned historian, art columnist, author, major politician and general
secretary of la Biennale di Venezia from 1895 until 1919.
33 See http://www.labiennale.org/en/art/history/origin/en/2954.html. 34
See http://www.labiennale.org/en/art/history/origin/en/2954.html.
fame. In the year of 1993, an exhibition that was curated by Achille Bonito Oliva, even included partaking from 45 countries37.
III. Data and Methodology
To measure the impact of the Venice Biennale exhibition on an invited artist‟s reputation, a sample of their auction records is selected and carefully studied. Furthermore, different methodologies are applied to make computations. This section discusses how and which data is selected and what methodologies are employed.
III.1 Data
A representative dataset has been selected using the official website of the Venice Biennale and related websites. From these sources, 2180 artists present on the Venice Biennale from 1995 to 2007 were selected and after applying various selecting criteria, 111 painters remain. Auction records from these artists are acquired from the Price Database of Artnet.
III.1.1 Artists
Artists present on the Venice Biennale exhibition that are utilized in this study meet the following selection criteria:
The artist is a painter.
The painter can be found in the Price Database of Artnet.41
The painter has no earlier presence on the Venice Biennale than the year 1995. The painter has paintings made by means of oil on canvas.
The painter had sales in the „sales-window‟.42
Since art is a synonym for paintings in this paper, the selected artist is a painter. To measure the effect of the exhibition, it is necessary to have auction records, which are obtained from the Price Database of Artnet. Therefore, the artists need to be present in this database. The criterion of no earlier presence on the Biennale is set because the possible price effect will not be correctly measured when an artist was already present on an earlier Venice Biennale exhibition. To find out whether these artists were present before 1995 as well, I use the database from the organization of the exhibition43, where artists present in one of the years from 1985 to 1993 are presented. After selecting the criterion that the paintings must have been made with oil on canvas, which I have chosen because oil on canvas is the most
41 The Price Database on www.artnet.com reflects actual auction records since 1985, which have been
compiled from over 500 auction internationally.
42
The sales-window represents the period where all sales are measured and is further explained in paragraph III.1.3.
forthcoming medium, 111 painters remain. This sample consists of 77 painters that are alive during the sales window and 44 painters who died before auction records in the sales window are measured. The painters that are alive during the period when sales are measured are segregated since in general, the price of a painting increases after the painter‟s death, inducing a smaller effect on the exhibition for deceased artists44. Table 2 exhibits an overview of artists present on exhibitions from 1995 to 2007.
Table 2: Descriptive Statistics of artist selection
This table exhibits descriptive statistics for the selection process for artists present on the Venice Biennale exhibition. It shows the number of artists present in the years 1995 to 2007. The number of artists left after selection is given in number and in percentage as part of the total artist present. Furthermore, artists present each year are shown as percentage of the sample. The last two columns give numbers and percentages of living artists each year.
Total artists Artists present As % As % of artists Living artists As % of specific
present after selection of specific year present after selection of specific year selection
2007 199 9 4.52% 8.11% 9 100.00% 2005 231 6 2.60% 5.41% 6 100.00% 2003 297 20 6.73% 18.02% 18 90.00% 2001 297 6 2.02% 5.41% 6 100.00% 1999 192 3 1.56% 2.70% 3 100.00% 1997 182 18 9.89% 16.22% 18 100.00% 1995 782 49 6.27% 44.14% 17 34.69% Total 2180 111 5.09% 100.00% 77 69.37%
Clearly, 1995 makes up the majority of the sample with 49 painters, followed by 2003 with 20, and finally 1997 with 18 painters. The year 1999 represents only 3 artists. . The last two columns indicate that the exhibition years 1995 and 2003 hold the same number of living artists. In total, 5.09 % of all artists present on the Venice Biennale exhibitions in the particular years have been selected. An overview of the names of the selected painters and their respective country of origin can be found in Appendix I. Italian artists emerged as the majority of the sample (22%), followed by France (19%), the United States (12%), Germany (9%) and finally China (5%). The remainder 33% of the total sample and encloses countries from the painter‟s origin with no more than 50 auction results.
III.1.2 Auction records
After selecting the criteria, I derived the following details from the auction records held by Price Database of Artnet: (1) Name of the painter, (2) Birth date (and death), (3) Painter‟s origin, (3) Medium used for the painting, (4) Surface, (5) Presence of signature or inscription, (6) Sale date, (7) Place of sale and auction house and (8) Sale Price (in USD).
The 111 selected painters in my sample provide in total 1209 auction records. An overview of the distribution of these results is presented in table 3.
44 Buelens & Ginsburgh (1993) and Itaya & Ursprung (2008) found a significant increase of selling
Table 3: Descriptive statistics of auction results
This table exhibits a summary of auction results for the Venice Biennale exhibitions from 1995 to 2007. It shows the number of auction records per year and as a percentage of total records. The last two columns provide an overview of the number of auction results of the living artists.
The table shows that the exhibition in 1995 holds by far the most auction results making up approximately 50 percent of the total sample, which is presented by 49 painters. This is in accordance with the number of artists present in this specific year and implies that there are no extreme numbers of auction results from one particular painter in this year in comparison to other years. The exhibition in 1999 holds the smallest amount with a number of 37 auction records. However, if we consider only living artists, then 1995 only makes up 26% of the sample. This percentage is provided by 17 living artists in that year as percentage of the living artists. The smallest number of auction results is obtained from the exhibition held in 1999, which is presented by only 3 painters in this sample. In Appendix II, a summary of the distribution from the auction results over the various auction houses and cities is available. Most auction results are obtained from Christie‟s auction house where 31% from the sample is acquired. Sotheby‟s follows with 28% and on the third place, Finarte with 10% of the records is presented. The other part (31%) consists of auction houses with no more than 100 auction records. In addition, an overview of the auction results per city is obtainable in Appendix II. It appears that New York includes the largest share of the auction results with 26 percent, followed by London (17%), Milan (16%), Paris (9%) and other cities (32%). All cities that are mentioned contain at least 100 auction records.
III.1.3. Time line
In the paper of Kraeussl and Ensing (2008)45, a sales-window46 from 3.5 years before to 3.5 year after presence on the Documenta47 exhibition is chosen to prevent possible influences of
45 Kraeussl and Ensing studied the impact on sales price and volume of artists present on the
Documenta exhibition.
46
The sales-window represents the period where all sales are measured.
47 The Documenta is a comparable art exhibition of modern and contemporary art that takes place every
five years in Kassel (Germany).
Nr of auction As % of Nr of auction records As % of sample
records sample of living artists living artists
a previous Documenta. That exhibition takes place once every four or five years, while the Venice Biennale exhibition occurs once every two years. However, since each exhibition and accompanying auction records of the relevant painters of that year in this study are taken separately I do not foresee any problems of influences of earlier Biennale exhibitions. Thereby would auction records from one year before to one year after the exposition not give a representative outcome due to the longer investment horizons that art investments have. Results in the research of Kraeussl and Ensing (2008) show that their sales window is large enough to obtain an overview of the impact of the exhibition, through which I decided to make use of a sales window that captures the same length around the exposition. Furthermore, I choose to split the window into half-year periods, because auctions occur mostly in April/May and in October/November48. The sales window taken for each exhibition is shown in figure 3.1. June is the month when every exhibition begins and is taken as time zero. The intervals before and after the beginning of a specific exposition are considered as half-year periods. They take place from either January to June or from July to December. This indicates that an interval does not need to take place in a specific year, but in a specific period measured before or after the exhibitions.
Figure 3.1: Time line for each Venice Biennale exhibition
-42 -36 -30 -24 -18 -12 -6 0 6 12 18 24 30 36 42
Venice → time in months
Biennale
III.2 Methodology
Auction results of the painters in this sample are primarily tested for an effect in sales volume and price. Thereafter, a regression model is implemented, becoming the focal point of this study.
III.2.1 Sales volume and price
The first sales volume effect is measured by adding the number of auction results in a particular half-year period within the sales window. As stated before, the sales-window consists of fifteen half-year periods from seven half-year intervals before until seven half-year intervals after the exhibition is held. Point zero represents the half-year period when the exhibition takes place. The price effect is tested by calculating the average prices for each half-year period, followed by determining the a geometric mean for various time intervals.
48 As stated by Hodgson and Vorkink (2004), spring auctions often take place in April and May and
To compute average prices, the total sum of all auction records within a specific half-year period before or after an exhibition are divided by the number of sales in this period. When averages are measured, the geometric mean can be computed by adding one to each periodic return, multiplying these values, taking the nth root of the product and subtracting one. The periodic return is the percentage change in the averages from one interval to another49. In formula: n n n n i
a
a
a
a
1 2...
/ 1 1 1
(1)Where
(
a
1,
a
2...
a
n)
are periodic returns and n the number of returns within the time interval. An advantage of computing geometric means is that all auction records can be used.III.2.2 Hedonic regression method
The hedonic approach was popular with the majority of earlier research. It implies that the value of a painting can be considered as a combination of various characteristics. Paintings are prized for the utility that these features bear. Hedonic prices are identified as the absolute prices of a set of characteristics and are econometrically estimated by regressing the sale prices on these hedonic variables50. The absolute prices are employed to adjust for quality change of a particular sales mix.
The hedonic approach makes use of quality characteristics aside from time dummy variables and carries out a single Ordinary Least Sales (OLS) regression on collected records from all time periods. The regression is represented by the following equation51 and is run in Eviews 5.0:
x j t t kt t t nkt j kta
X
C
P
1 1 0ln
(2)Where
ln
P
kt stands for the natural logarithm of the sale price52
of a painting k on time t, the
0
a
represents the intercept of the regression, the beta coefficient stands for the quality characteristic X,X
nkt is a symbol for the quality characteristic value of a particular painting
49 See http://riskinstitute.ch/00011599.htm. 50 See Chanel, Varet and Ginsburgh (1996). 51
See Diewert (2003).
52 Diewert (2003) illustrates that a logarithm of the price is preferred rather than the actual sales price
n,
t stands for the coefficient value for time dummy t,C
t represents the time period value equal to 0 or 1 and the last term
kt is a symbol for the disturbance term. For each painting k, the dummy variable C takes on the value “1” if it is sold in this period, and “0” otherwise.The quality characteristics selected for this model are subdivided into physical and non-physical characteristics, whereby the latter also includes the methodology applied for the artist variable53.
Two of the physical characteristics of a value is the auction house and city where the painting was sold. Christy‟s (AH_CHRISTIES) and Sotheby‟s (AH_SOTHEBYS) are the largest and most familiar auction houses and are expected to have higher prices on paintings than other auction houses. A third substantial auction house is Finarte (AH_FINARTE), which is also included in the regression. The coefficients are likely to be positive and significant, which is in line with results of earlier studies like Mei & Moses (2002) and Kraeussl & Schellart (2007). These coefficients are measured in comparison to smaller auction houses that are included in the term „other auction houses‟. This is the reference variable and therefore excluded from the equation. As is described in the data section, the selected cities are London (CITY_LONDON), Milan (CITY_MILAN), New York (CITY_NY) and Paris (CITY_PARIS). According to Renneboog and Van Houtte (2002) and Kraeussl and Roelofs (2008), London and New York are expected to have significant higher prices. Another physical characteristic includes the surface of a painting (SURFACE). This is the multiplication of the width and the length of the painting and its coefficient is expected to have a positive sign54. Corresponding to the logarithm of the sales price, the surface values are logged as well.
The non-physical characteristics that are used in the regression include: (1) the presence of a signature on the painting, (2) whether the painter is still alive or not and (3) the reputation variable. The reputation variable is what separates a higher ranked painter from a relatively lower ranked painter. The first variable, which measures the importance of a sign (SIGNED) of the artist on the painting, has a positive impact on the price. A buyer is more willing to pay a higher price for the painting if he or she is more certain of its authenticity. The dummy variable takes on the value “1” if it is signed, and “0” otherwise. The coefficient
53 Kraeussl and van Elsland (2008) developed a novel 2-step hedonic approach. Just as the average
price of art per year is adjusted for quality by using the hedonic approach, the average price of art per painter can be adjusted for quality the same way. The index yields the value of art per painter, in comparison to the base artist. The method consists of two steps, first, a new artistic value variable is created, by correcting the average price per painter for quality. Then the artist dummy variable is replaced by the new artistic value variable in the regression, and the regression is run again. See their article for a broad description of this method.
that includes the characteristic of a death or living painter is negative sign for a living artist, citing the article of Itaya and Ursprung (2008) which discusses a rise of the price of a painting after a painter‟s death and thus higher priced paintings for deceased artists. Like Buelens and Ginsburgh (1993), the dummy variable abides the value “1” if the painter is still alive (ALIVE) on the moment of sale and “0” if not, as is also stated by. The last variable takes account of the reputation of the artist (LN_REPUTATION). When people think that a well appreciated artist has created a painting, one would value this work higher than if a painter with a lower status produced an equal painting55. Other studies specify dummy variables for artistic qualities of each artist, but due to the large number of artists in my sample, I make use of a new approach that is developed by Kraeussl and Van Elsland (2008) which ranks the different artists by quality. This 2-step hedonic approach avoids an incredibly large number of dummy variables for each artist. The methodology for this approach is explained as follows:
First the single OLS regression, as mentioned earlier, is run. No artist dummies are included in this regression so far. Next, the following computation is completed manually in Excel:
x j n i m i r ij y ij j m i m r i x j n i m i r ij y ij j n i n y i ym
X
n
X
P
m
X
n
X
P
Index
1 1 1 , , 1 1 , 1 1 1 , , 1 1 ,exp
exp
(3)Where Py represents the sales price of a specific painter,
P
r stands for the sales price of the reference painter, n is a symbol for the number of paintings of the specific painter, m stands for the number of paintings of the reference painter,
represents the regression coefficient of a specific quality characteristic, Xij,y stands for the value of the quality characteristic for a particular painting of a specific artist andX
ij,r represents the value of the quality characteristic for a particular painting of the reference painter.To obtain the reputation index for each artist, the outcome of the formula is subsequently multiplied by 100. The reference artist is chosen based on the largest number of
sales of all artists present in the sample. Artists that are relatively lower valued will have a resulting index less than 100, higher valued artists remain index more than 100 and the value of the reputation index of the reference artists will exactly be 100. In my model, Massimo Campigli encloses the largest number of sales and is therefore chosen as reference artist56. The logarithm of this reputation index is taken in line with the logarithm function of the sales price and surface. Subsequently, the OLS regression with the reputation index is run again.
To measure the difference between the specific quality characteristic and the corresponding reference in percentages, the following calculation is applied57:
100
1
,
i a r ie
P
(4)Where Pi,r is a symbol for the percentage difference between the quality characteristic i, the accompanying reference r, and
a
i stands for the coefficienta
of the particular quality characteristic i. These percentages are computed for all quality characteristics except for the surface and reputation variable. Since these characteristics do not include dummy variables and have no benchmark, they can be interpreted directly from the regression.Next to the selected quality characteristics, I have also included time dummy variables in the regression. As is described earlier, the sales-window consists of 15 half-year periods from 7 periods before the exhibition, 7 periods after and 1 period during the exhibition. The dummy variable takes on the value “1” if the sale of the painting is in the specific period and “0” otherwise. The half-year period when the exhibition takes place, which is time zero, is taken as the reference variable. In this period, the price index follows a value of 100. Price indices after this period are expected to exceed this value if a positive price effect is present. The price index values of each period are calculated by applying the following formula:
t
index e
P 100 (5)
Where
stands for the coefficient of the particular time dummy variable t.
56 See Logher (2008).
IV. Empirical results
This part discusses the results whereby (1) the outcomes of the sales volume and price effect are argued and (2) the results of the hedonic regression are analyzed. The latter further elaborates on the importance of physical and non-physical characteristics of the painting.
IV.1 Sales volume and price effect
Figure 4.1: Overview of the turnover, average price and number of sales of the artists
The figure represents the total sale statistics per half-year period for all 111 artists present in one of the exhibition years 1995-2007 (1209 auction results). The middle point zero denotes the start of the exhibition on which paintings from the particular artists are presented.
0 20 40 60 80 100 120 0 50 100 150 200 250 300 350 400 -42 -36 -30 -24 -18 -12 -6 0 6 12 18 24 30 36 42 N u m b e r o f sa le s Tu r n o v e r a n d (x $ 1 ,0 0 0 ,0 0 0 ) a n d a v e r a g e p r ic e (x 1 0 ,0 0 0 ) Months
Total sale statistics per half year period
Turnover (x$1,000,000) Average price (x$10,000) Number of sales
Table 4: Geometric returns
This table exhibits the outcomes of the geometric returns over various time intervals. In the first column returns for all artists (111) and auction records (1209) are presented. The second column shows returns for only the living artists (77) and their auction results (795). Point zero corresponds to the beginning of the exhibition.
Figure 4.2 exhibits the development of the median price in the sales window. It shows a strong positive development from point zero and peaks 2.5 years after the beginning of the exhibition at a median price of $187,883 for all artists. The price is even higher for only living artists at $1,032,305. Overall, the figure shows a strong positive effect for the artists that are alive during the moment of sale than for all other artists included in the sample. This is in accordance with the theories presented in articles of Buelens and Ginsburgh (2003) and Itaya and Ursprung (2008). The median price is lowest for the total sample on 2.5 years before the beginning of the exhibitions where this price is $20,826, and lowest for only living painters two years before the exhibitions with a price of $12,925. Appendix III provides a more detailed overview of the development of the sales volume, turnover and average and median prices of just the living artists.
Figure 4.2: Overview of the median price development
This figure exhibits the development of the median price in the sales window from 3.5 years before until 3.5 years after the exhibition. Point zero represents the period of the beginning of the exhibition. Median prices are presented for auction results for all artists included in the sample (1209 records), and records for only living artists (787 records). 0 2 4 6 8 10 12 14 16 18 20 -42 -36 -30 -24 -18 -12 -6 0 6 12 18 24 30 36 42 M ed ian p ri ce (x$ 10 .00 0) Months
Median price per half year period
all artists living artists
interval
(months) semi-annual annual semi-annual annual
(-42,42) -0.36% -0.73% 6.10% 12.57%
(-24,24) 10.59% 22.30% 1.29% 2.59%
(-42,0) -16.55% -30.36% -14.24% -26.46%
(0,42) 19.01% 41.64% 30.30% 69.78%
(0,24) 35.58% 83.81% 29.68% 68.17%
In all computed sales volume and price results obtained in this subsection, the positive effect of the exhibitions is present. The positive effect on the average prices from all painters is specifically exhibited in the annualized geometric return from zero to two years after the exhibitions. This return includes 83.81% and is worthy of note from the viewpoint of the investor. The prediction was that the effect would be most visible among the sample of only the living artists due to an increase of the price of a painting after the painter‟s death. This forecast is justified if we look at the development of the median prices. These prices are much higher after the exhibitions for the sample of the living artists than for the total sample. Overall, the Biennale effect is most observable with the median prices for both samples.
IV.2 Hedonic Regression
The empirical results from the hedonic regression are divided into three main subsections: (1) The importance of the physical characteristics of the paintings is argued, (2) The impact of the non-psychical characteristics is analyzed and (3) The price development in the sales window is illustrated by means of a calculated price index.
IV.2.1 Psychical characteristics
added to the price of the work of art. Because both variables are logged, the outcome is analyzed in normal terms. This positive effect on the price is also not a surprising outcome58.
Table 5: Relative significance of the physical characteristics of a painting
This table exhibits an overview of the outcomes resulted from the hedonic regression. This regression is run in Eviews for all 1209 auction records in the sales window from (-42,42). It shows the importance of the psychical characteristics that paintings bear relative to their accompanying benchmark. The percentage change of the auction houses is measured relative to „other auction houses‟ that are included in the sample. These percentages are computed by implying the formula
100
e
ai
1
. The mentioned cities are measured against theirbenchmark, „other city‟s‟. The surface variable does not have a benchmark.
***/**/*, significance level at 1%, 5%, 10%
IV.2.2 Non-psychical characteristics
In table 6, the importance of the non-physical characteristics of the paintings is presented. The sign of the first variable, the presence of a signature on the painting, is surprisingly negative. From the regression results that signed paintings are sold at a discount of 6.01% in comparison to paintings without a signature. Although the variable seems to be insignificant, the outcome is in sharp contrast to earlier research and is rather difficult to explain. A possible explanation could be linked to a statement of Anderson (1974): “Many well known paintings are unsigned, because certain artists tend not to”. He presents an example of admired Italian painters who typically set no sign on their paintings. The second variable is measured relative to prices of paintings from a death painter and has the expected sign. The result implies that paintings from living artists are sold at a 38.74% discount over paintings from painters that are death. The coefficient of the reputation variable is most important and is in accordance with the findings of Kraeussl and Schellart (2007). The importance of this variable is also presented when the regression is run without this variable. The hedonic model has a R-squared of 74.8% when the reputation variable is included and 34.7% when this specific variable is excluded. Omitting this variable causes thus a drop of nearly 40%. The percentages indicate that 74.8% and 34.7% of the price of the painting is described by the hedonic model.
58 See Anderson (1974).
Variable Coefficient Std Error Probability Significance % change
