Performance of emotional assets : an optimal portfolio allocation analysis

(1)

Master Thesis:

Performance of Emotional Assets: an Optimal Portfolio Allocation Analysis

University of Amsterdam

MSc Business Economics, Finance Track

Author: Paul Negrila

Student number 10604553

Thesis supervisor: Dr. R.I. Todorov

(2)

1

ABSTRACT

This research is concerned with analyzing the benefits of including collectibles in financial portfolios. Firstly, the individual risk-return profile of art investment will be compared with that of traditional asset classes like stocks bonds and commodities using data over a period of 16 years. Additionally, efficient portfolios will be constructed using these assets to determine the optimal allocation to collectibles in a long-term portfolio. The results will show that art contributes to optimize long-term portfolios by reducing risk, mainly due to its negative correlation with traditional financial securities. Secondly, the performance of art before, during and after the financial crisis (2007-2009) will be closely analyzed in three-period study which will reveal that emotional assets lose their diversification benefits and have zero allocation in an efficient portfolio in times of crisis. Thirdly, using cross-sectional data of different art sectors, the study will replicate the investment portfolio of art funds, and will determine the return rate and volatility of indirect investment in art. Additionally, it will use the perspective of an art fund manager to estimate the optimal allocations to each emotional asset class or art sector aiming to maximize profitability. The results will reveal that various art forms perform differently as a consequence of trends and preferences of the art market, as well as cultural and historical reasons.

(3)

2

1. INTRODUCTION

At a time when conventional investment solutions are being challenged and financiers are questioning the investment merits of traditional asset classes, emotional assets are shaping up as a practical mainstream asset class. Emotional assets cover a wide range of collectibles, from fine art and photographies to prints, diamonds, stamps, coins and violins. With the present financial climate marked by high volatility and record-low interest rates, the poor performance of traditional financial securities has alerted investors in their search for viable alternatives. The aim is to identify different assets with low or even negative correlation with bond and equities that would offer high risk-adjusted returns from diversifying investment portfolios, and would store value during financial crises. Although the most high-profile alternative assets remain real estate and commodities, like gold or silver, collectibles are currently experiencing a rise to prominence powered by a more robust market infrastructure and ever-increasing investor interest. According to Artprice, 2013 was the best year ever in art auction history, with revenues of over $12 billion and a global demand rejuvenated by an influx of buyers from Asia, Middle East and Russia, who were instrumental in the market‟s fine performance.1

The idea of art as an investment form has been recently reinforced by the emergence of several funds specializing in art. More specifically, as Campbell (2008) notes, art funds “use a wide variety of trading strategies” and target hefty returns by “trading on the inefficiencies currently present in the art market, typically characterized by low liquidity”. As a result of this development, investors with an interest in the art market are faced with the dilemma of choosing between indirect investment - through art funds - and acquiring items directly. There is a growing academic literature on studying art as an investment, but none of these studies analyze closely the risk-return profile of art funds. This is mainly a result of the lack of available data, since funds deal in private transactions, and has left room for debate on whether these “mutual funds of the art world” are successful or not. However, recent developments in art market infrastructure have increased market transparency and enabled better access to art price information, which makes it possible to replicate an art fund investment portfolio using emotional assets‟ price indices. As a consequence, this paper will simulate an art fund‟s investment strategy by building efficient portfolios using data for various emotional assets and art sectors. The results do not only provide estimates for the return rate and volatility of this new investment solution, but also valuable information about which optimal allocation each emotional asset should have, depending on both the individual performance and correlation with other classes.

(5)

4 The paper at hand sets to take a closer look at the financial characteristics of art investments from the perspective of portfolio diversification and answers the question whether art‟s correlation with the traditional financial securities can benefit investors in the long run. The art market trades in real tangible assets with a unique financial profile, and in the empirical section of this paper I find that including collectibles in traditional portfolios improves the efficiency over long periods of time.

In order to conduct the empirical part of the research, time series data based on price indices of several emotional assets compiled by Artprice will be used, along with return data of four other traditional asset classes over a period of sixteen years. By handling this dataset, it becomes possible to estimate an average yearly return rate and standard deviation of global art investment and various individual emotional assets, as well as the variance-covariance matrices used in the optimization process.

In addition, this research also includes a close-up look at the crisis period and the extent to which it affected emotional assets‟ performance. Given the poor results of financial securities in that specific scenario, the importance of analyzing the reaction of art during financial turmoil needs to be emphasized. To conduct this analysis, a nine year interval centered around the 2007 Financial Crisis will be divided in three periods which will enable the author to estimate the risk-return characteristics of collectibles before, during and after the crisis. By employing this methodology, it becomes possible to assess the theory that emotional assets are a viable alternative investment in times of crisis, and whether art‟s long-term correlation with financial securities stays constant in this particular scenario.

The remainder of the paper is organized as follows: Section 2 provides an overview of several research papers concerned with methods for analyzing the dynamics and characteristics of collectibles and the art market. Section 3 offers a description of the dynamics of the art market and different channels of investing in it. Section 4 comprises an empirical study divided in three interrelated studies concerned with the financial performance of art investment and its role in portfolio optimization, and discusses the main findings. I find that due to low and even negative correlation with other financial securities, art is a contributing part of an efficient strategy to both minimize volatility and maximize risk-adjusted return. Finally, Section 5 provides some concluding remarks.

(6)

5

2. LITERATURE REVIEW

The next section reviews a number of academic papers which have been written on the topic of art investment from a financial perspective. I begin by reviewing papers concerned with the behavioral foundation of art investment, followed by the different methodologies employed to compute the return rate of collectibles, and, finally, selected literature that provides evidence of the financial performance of emotional assets.

2.1 Motivation & behavioral view for investing in art

As established by McAndrew (2010), the main reason for investing in collectibles is passion, with financial motivation playing only a secondary role. Even though the vast majority of collectors motivate their acquisitions by passion, their investment decision also incorporates financial reasons since they exhibit strict price criteria, for example. Various other studies have tried to analyze more profoundly the behavioral implications of investing in collectibles as well as the motivation behind this choice. Firstly, Burton and Jacobsen (1999) describe the art market as being dual to a certain extent; on one hand formed by a large number of collectors who are active due to non-monetary reasons (they enjoy art as consumption goods), on the other by a small number of speculators who enter the market for purely financial reasons and try to exploit inefficiencies in order to reap high returns. The reduced number of investors, added to the low liquidity, work as an incentive for the second group, because it becomes easier to manipulate the market. However, the authors‟ analysis is too simplified as it is unlikely to predict when a bubble is going to happen or which asset will outperform the rest. In addition, they fail to emphasize the importance of financial rationale for investors who invest in collectibles, as they are generally keen to buy and hold the assets which will generate the biggest returns they could get.2 However, because most of the benefits of owning art come from emotional and expressive aspects, collectibles have a large emotional component which makes the collectors susceptible to behavioral biases and heuristics. To name a few: the “endowment effect” (objects owned are subjectively evaluated higher than others), “availability bias” (drawing conclusions from information readily at hand), “anchoring” (price expectations linked to historical prices for similar items), “opportunity cost effect” (failure to consider alternatives before investing), etc.

(7)

6 It was Belk (1995) who introduced the notion of luxury consumption and explained that emotional assets are valued as unique and useless objects and that buyers seek the “inessential consumer goods that are removed from any functional capacity they may have had once”. This refers to the pleasure and gain in status generated by showing to others that you own valuable items that very few can afford or access. Koford and Tschoegl (1998) researched whether rarity in collectibles is positively linked to the value of the item, and based their study on rare coins. They ultimately found that if the asset is unique or very rare, the utility function rises and consequently increases the value of the object even though its physical quality remains unchanged.

The multi-attribute utility function introduced by Bollen (2007) models the utility of investors by compensating lower risk-adjusted performance with the additional utility derived from “investing in financial assets which adhere to their societal or personal objectives”. Campbell et al. (2008) extend this model to emotional assets, and propose that this additional utility be also a function of wealth, since the people who dispose of great financial resources are usually the ones acquiring collectibles. They also attribute an intrinsic value to collectibles, which corresponds to a form of aesthetic pleasure. Since this is independent of their monetary value, it constitutes a solid explanation as to why collectors agree to acquire emotional assets for more than their quoted value. The conclusion follows logically, that the emotional component is real and it is included in the price, and attempting to dissociate it from the monetary component is unfeasible, as it is impossible to quantify each part individually, especially because it differs for every single individual.

Alternatively, Corneo and Jeanne (1994) study conspicuous goods and the social consumption which enables the owner to signal and raise his social status, and gain “approval by the society”. They go on to define them as “goods that are chiefly purchased because of the demonstration effects that their consumption exerts on others”. Scott and Yelowitz (2010) add to this and claim conspicuous assets like diamonds can be consumed “not just for their intrinsic utility but also for the impression their consumption has on others”. According to their findings, collectors are disposed to pay significant premium in order to marginally increase the quality of the diamond and reach a higher threshold.

Ultimately, Mandel (2009) undertook a study to prove that the “determinants” of the value of art are different from those for other financial securities. He observes that “art offers no claim on an underlying flow of payments” and supply has zero elasticity since the market “is dominated by the masterstrokes of dead artists” and concludes that the main driver of art returns is the dynamic demand, radically different from any other financial assets.

(8)

7

2.2 Methods to compute collectibles’ rate of return

Unlike traditional assets like stocks or bonds, emotional assets present the problem of scarce information and limited transaction record, in addition to their heterogeneity. As a consequence, the methods used to compute historical returns for this asset class need to be tailored to address these issues. Aside from addressing the heterogeneous aspect of collectibles, Ginsburgh et al. (2006) recommend to “distinguish many different collecting categories since returns may vary dramatically” between them. As a result, the scientific literature has used a variety of methodologies in order to estimate collectibles‟ return rates based on historical data such as repeat-sale regression, hedonic regression or composite indices, which are discussed in more depth in the following part of this section.

Composite indices

The first method, as presented by Burton and Jacobsen (1999), is to build up a composite index using selected sample sets of items that vary over time. This method is somewhat popular among economists, as it is conceived to be easy to implement, but also presents a large number of drawbacks. The obvious one is the low accuracy due to the fact that different objects come up for sale at different moments in time, so the method is heavily reliant on the assumption that the quality standard of the basket is constant at all moments. Additionally, results can vary significantly and suffer from sample bias, depending on the assets which were traded during the studied time interval.

Hedonic regression

Another method to construct a price index is by hedonic regression, a model common in real estate valuation models. By regressing the price of the items on a series of characteristics, one can address the issues that arise from studying heterogeneous assets. Since each work of art is unique in its kind, this method is well suited to solve the heterogeneity problem. Commonly used features of collectibles in hedonic regressions are: purchase price, size, name artist, sale location, age, etc. Another use of hedonic regressions for emotional asset studies could be estimating the price gain caused by ageing in wine by interpreting the value of the coefficients. This particular method is a very demanding and powerful tool that requires a large number of factors and makes use of all recorded transactions. However, it is impossible to include all characteristics, so it becomes very important to justify which should be included at the expense of other. Since estimating qualitative features on a quantitative scale will always raise questions regarding the assumptions made, Triplett (2004) sanctions that hedonic regression‟s biggest liability is “the difficulty in introducing weights”. Then, as Collins et al. (2007) state, hedonic regression includes time-invariant drivers and therefore implies strong assumptions regarding the stability of the market in time. As previously mentioned, this method requires detailed information about a good‟s

(9)

8 attributes and its sale, and this could mean that many observations have to be left out of the sample due to lack of information.

Repeat sales methodology

A third approach is the repeat sales methodology, based on recording the price fluctuation of a certain asset over a fixed period of time. This method presents the obvious disadvantage of taking into account only assets that have been traded at least twice over the sample period, leaving a significant percentage of transactions outside the sample. This causes the results to suffer from sample selection bias, since it only uses the „successful‟ part of the sample (i.e. items that were successfully resold).It has been widely used in art research and is considered one of the most consistent and reliable methods because it “averts the need to deal with many issues associated with the heterogeneous nature of art”. This particular approach is better suited for some collectibles which are traded frequently, but less so for others like antiques for example. However, it overcomes a series of disadvantages associated with the two previous methods by controlling for quality by recording the price change of each item included and has been widely used to estimate the financial performance of emotional assets in the past, and is a common feature in the academic literature on emotional assets‟ performance.

Despite the diverging approaches of each method, Chanel et al. (1994) observe that over a long time period, results obtained via different approaches are strongly correlated. This study makes a strong case in defense of the hedonic approach for long-term analysis by stating that “returns can be computed using all sales and not resale only”. Alternatively, Fogarty and Jones (2010) compute returns to wine over a short period (1998-2000) by hedonic, repeat-sales and hybrid methodologies. Their results indicate that the repeat-sales regression yields considerably higher return estimates than the other methods, a conclusion that had been reached before by Ashenfelter and Graddy (2003). The study also signals that the hybrid methodology of regressing prices by hedonic criteria while identifying repeat sales is, in fact, the approach that generates the most precise estimates (lowest standard error). Locatelli-Biey and Zanola (2005) had also identified this approach as the most accurate by studying Picasso prints. On the downside, there is a certain difficulty in identifying variables which are time-variant and this makes combining the two methods problematic (Chanel et al., 1996).

In conclusion, the main drawbacks of all mentioned methodologies for computing returns of emotional assets is the need to constitute large samples and gain access to a large number of transactions to ensure

(10)

9 accuracy. As a result, a significant amount of work is needed, in addition to large sets of data for emotional assets, which are hard to come by. In addition, considering the need to study a large quantity of transactions, one can only find quarterly or bi-annual returns, and this makes comparisons with other traditional assets quite difficult. Ultimately, return calculations based on indices present other important limitation by only considering a part of the market since it does not include private transactions or primary sales. Furthermore, it only accounts for successful sales by ignoring items that have not been sold. This is called “survivorship bias”, stressing successful sales over unsuccessful and it leads to overly optimistic estimates. Moreover, samples consist of items with high demand that actually attract competitive bidders when they are sold, and ignore less demanded items that do not even make the auction floor. The indices also exclude transaction costs, which are considerable in case of collectibles and lack predictive power since they are constructed using historical prices. In spite of all this, these methods remain the most appropriate to analyze the historical performance of emotional asset classes.

2.3 Historical risk-return performance of art

Out of all emotional assets, the large majority of scientific literature has been concerned with studying the historical risk-return profile of paintings. Considering its extreme heterogeneity and diverse approaches employed, it is understandable that the results have varied considerably. Despite all these difficulties, art economists have managed to provide valuable insights into the rates of return on art markets.

During the 1970‟s, economists conducted the first analyses of the rate of return of art, and it is Anderson‟s (1974) study which stands out. Starting from the observation that art had had remarkable returns in the previous twenty years, he gathered data by going as far as 1780 in an attempt to estimate the long-term return of art. After estimating and including additional costs and fees in his calculation, he concludes that the real long-term return of art is much lower than the most recent results at the time and that “paintings are not particularly attractive investments unless one also includes the consumption value of art”. By using both the repeat-sales regression and the hedonic approach, he estimates a real return of 3.0% p.a. and 2.6% p.a. respectively and also finds that returns over the past two decades (1950-1970) were significantly higher than the long-term average. Anderson‟s results also reveal that modern paintings seem to reap higher returns than old masterpieces, a finding that is in accordance with the results of the thesis at hand. The author concludes that considering the reduced risk-adjusted performance of art, the only rationale for investing in paintings should be their consumption value, not the financial benefits.

(11)

10 Even less encouraging results were obtained by Baumol (1986), who conducted a comparable study using the repeat-sales method on a sample of transactions extending from 1652 to 1961 from which he excluded all the cases where there was a period of less than 20 years between the sales. However, his approach is susceptible to biases due to including sales from the distant past (17th century) when the returns were significantly lower. Moreover, by eliminating the cases with an interval of less than 20 years between the sales, it is very probable that many over-performing items (characterized by frequent sales) have been left out of the sample, which would have enhanced the estimates. The results show a very low real rate of return of 0.55% p.a. which can be justified by the vast sample period as well as the handling of it. The author also states that there may not exist an equilibrium level, so that prices of art objects may be strictly unnatural in the classical sense. Furthermore, he emphasizes the difference between a stock that is made up of a large number of homogenous securities, which are perfect substitutes for one another, and works of art, which are unique. Also, a particular stock is held by many individuals on a near perfectly competitive market, unlike owners of art pieces, who can be considered as holding a monopoly on that specific item. Although his conclusion, that investing in art solely for financial purposes is futile due to its low risk-adjusted return, one could assume that Baumol‟s results are underestimating the return rate of art investments in the current environment.

On a similar note, Goetzmann‟s (1993) study also concludes that investing in art is not justifiable from a purely financial perspective. He uses transaction prices of paintings brought to market at least twice over a period of 270 years (1715-1986) and constructs an art return index, used to determine the risk-return characteristics of art investment and compare painting price fluctuations to stock market movements over the studied interval. He interprets that the demand for art increases with the wealth of art investors, on the evidence of his art return index being strongly correlated (0.78) with an index of London Stock Exchange over the same time period. On the other hand, this finding implies that investing in collectibles as a hedge against stock market movements is not practical. Even though, in the latter half of the 20th century, art investment produced average nominal returns (17.5% p.a.) similar to those of the stock market, it still cannot justify the extraordinary risks it carries and the author concludes that a risk-adverse investor would not deem art to be an interesting purchase for financial gain alone. The article “differs from previous research in that it focuses on the time-series behavior of art and the stock and bond markets over very long periods”.

In the same year, Pesando (1993) studied art as an investment opportunity by estimating a semiannual index of prices for the period 1977-1992, but using repeat sales of modern prints instead of paintings. The infrequent sales of unique works of art and the hardship of tracking the prices of individual art objects over time makes it difficult to construct an art index that can be aligned with data on periodic returns of

(12)

11 other traditional financial assets. By focusing on prints (published in editions of 50-100), the author tries to circumvent this problem, as several exhibits of the same print can be offered for sale in the same season. This results in a dramatic increase in the number of repeat sales, 27,961 in the 16 year interval studied. He observes a real rate of return (1.51% p.a.) lower than stocks, US government bonds and T-bills, even though the risk is comparable to the risk of investing in stocks, and concludes that prints “do not compare favorably to traditional financial assets”. He also identifies a sharp spike in both nominal and real returns in 1990, the same year when the historical high of the analyzed period (400% increase compared to 1977) is achieved. Notably, the standard deviation of art (19.94) is practically in line with other traditional financial assets (22.47 for stocks and 21.83 for US government bonds) and prints prove to have a capacity to provide efficient diversification, once included in a portfolio of traditional financial assets. This is caused due to a fairly low correlation with stocks (0.3) and negative with US government bonds (-0.1); it can also be observed the sizeable difference between these results and Goetzmann‟s finding of a 0.7 correlation with LSE. The author also argues that a masterpiece portfolio will not outperform the market, being one of the first to provide evidence in that direction, and finds no evidence that certain artists command higher prices in certain countries. However, a puzzling find is that prices paid at certain auction houses are systematically higher than prices paid by buyers at other auction houses, with the example of two New York auction houses, Sotheby‟s and Christie‟s (14% higher prices for identical prints at Sotheby‟s compared to Christie‟s).

Mei and Moses (2002) signed an important contribution to the subject by constructing a new repeat-sales data set based on auction art price records for the American market, principally New York, between 1875 and 2000. They estimate an annual index of prices and try to address the issues of heterogeneity of art objects, as well as the infrequency of trading, problems also observed by J. Pesando in paintings before deciding to turn his attention towards modern prints. In addition, this study has a significant increase in the number of repeated sales (4,896) compared to the earlier studies of Baumol and Goetzmann. They are able to construct an annual art index as well as annual sub-indices for various painting schools (Impressionist, Old Masters, etc.), and find, contrary to some earlier studies, art to be a more rewarding investment than some fixed income securities, even though it under-performed stocks. To be more specific, they find a nominal rate of return of 4.9% p.a. over the whole period, and of 8.2% since 1950. Standard deviation is higher than that of stocks (S&P500), meaning that art compares unfavorably to stocks (lower returns and higher standard deviation), but one can observe an important decline in art‟s volatility when comparing the 1875-1999 period (0.428) to the 1950-1999 period (0.213). Furthermore, according to their study, the art index is weakly correlated with the S&P500 (0.04) and negatively correlated with government bonds (-0.15) and corporate bonds (-0.10). As a result, a well-diversified

(13)

12 portfolio of art works can play a more important role in efficient portfolio diversification than earlier studies show. In addition, they find strong evidence of underperformance of masterpieces, in accord with Pesando‟s paper which means expensive paintings tend to underperform the art market index. They also confirm Pesando‟s finding that „the law of one price‟ is violated in the New York auction market, by observing a small price difference between prices at Christie‟s and Sotheby‟s.

More recently, R. Campbell (2007) looked at art as an alternative financial investment, encouraged by the emergence of art funds as a new alternative investment opportunity that tries to exploit the negative correlation of collectibles with the traditional financial securities. After “desmoothing” the returns and including additional costs related to this type of investment, she finds an annual average return of 7.05% between 1980 and 2006. In addition, according to her study, art exhibits a low and even negative correlation with various asset classes like stocks, bonds and real estate. Her analysis reveals that the broad portfolio of a variety of emotional assets (art, wine, coins, stamps, clocks, atlases, books) can provide a sizeable contribution to the mean-variance portfolio, and also bring about a significant increase in the Sharpe ratio. Her conclusion is that the consumption value of collectibles is very important and needs to be taken into account, because investors are “willingly giving up a part of risk-adjusted returns by holding more art than wine and books”.

Other authors chose instead the hedonic approach, between them R. Agnello (2002) who included a significant number of variables to build the model‟s hedonic function, and applied it to an impressive database of 25,200 transactions. His results reveal a 4.20% p.a. average return for the 1971-1966 interval and a standard deviation of 23.10%, once again proving art compares unfavorably with the S&P500 index (11.60% return and 12.10% standard deviation). He also analyzes the correlation with the S&P500 (0.23) and Government bonds (0.07), reinforcing the idea that there are diversification advantages in art investment.

Similarly, Renneboog and Spajeners (2011) perform a traditional hedonic regression analysis, which relates transaction prices to a wide range of value-determining characteristics and year effects, to a dataset of a 50-year period, consisting of over 1 million transactions between 1957 and 2007. The resulted real annual return of art is 3.97% in real USD terms. This represents a return similar to that of corporate bonds, but at a much higher risk (0.19 compared to 0.09). However, for the 1982-2007 interval, the real return rises to 5.19% p.a., while applying a repeat-sales regression for the same subsample leads to almost identical estimates. Alternatively, the results show negative correlation of art with government and corporate bonds and even S&P500 (-0.03), while there is positive correlation with global stocks (0.20), gold (0.30) and commodities (0.44).

(14)

13

Table 1: Estimated returns to art from various studies

Authors Methodology Sample Time period Nominal

return Real return R. Anderson (1974) RSR Hedonic Paintings 1780-1970 1780-1960 3.70% 3.30% 3.00% 2.60% J. Stein (1977) Random sampling Paintings 1946-1968 10.5% W. Baumol (1986) RSR Paintings 1652-1961 0.60%

Frey and Pommerehne (1989) RSR Paintings 1635-1949 1.40%

W. Goetzmann (1993) RSR Paintings 1850-1986 6.20% 3.80%

J. Pesando (1993) RSR Modern Prints 1977-1991 1.51%

de la Barre et al. (1996) Hedonic Great

Impressionist 1962-1991 12.00% 5.00%

Chanel et al. (1996) Hedonic Paintings 1962-1988 4.90%

Pesando and Shum (1996) RSR Picasso Prints 1977-1993 12.00% 1.40%

C. Czujack (1997) Hedonic Picasso

paintings 1966-1994 8.30%

Mei and Moses (2001) RSR Paintings 1875-1999 4.90%

R. Agnello (2002) Hedonic Paintings 1971-1996 4.20%

R. Campbell (2005) Moving average Paintings 1980-2008 6.76%

Kräussl and van Elsland (2008)

Two-step

hedonic Paintings 1985-2007 7.3%

Renneboog and Spaenjers

(2013) Hedonic Paintings 1995-2007 4.00%

Note: Table 1 contains the prominent researches concerned with art’s financial return over the past 40 years, and summarizes the methodology chosen by the authors, the studied emotional assets, the time

(15)

14

3. THE MARKET FOR ART

3.1 Collectibles as an emerging asset class

The last decade has seen the market for art and collectibles experience a booming evolution, and art has gradually crystalized as a distinct asset class. On this backdrop, the art and finance industry has been constantly developing, in order to be able to better respond to the needs of investors interested in diversifying their portfolios of traditional assets by adding collectibles and art objects. The purpose of this section is to provide some context to the motivation for choosing this particular topic as a research object of this paper.

Besides being the object of interest for traditional passionate collectors, art is becoming increasingly interesting as an asset class in the wealth management community. One could talk about a shift in perspective, from regarding art as an emotional asset, with the chief purpose of being enjoyed for its aesthetical characteristics, to considering art objects as a tool for portfolio diversification and wealth preservation.

Following a period of recovery after the recent financial turmoil, increasingly more investors are looking to include in their portfolio assets that have low or even negative correlation with the traditional equities or fixed-income securities. Economic uncertainty has made investors shift their attention towards alternative investments with a more lasting value, with an increased focus on the objective of portfolio diversification.

There is a large number of collectibles being traded, ranging from fine art, wine and diamonds to stamps or violins. However, all of them possess many common features which make it possible to compare them with one another and to form a unique asset class. As opposed to the traditional asset classes which can generate cash flows or have an intrinsic value, collectibles are different and this is the reason why the valuation process of these items is more complicated. These “emotional" assets are not intended to be transformed over time, like commodities, nor do they generating any activity like companies or real estate do, but are intended to be enjoyed from an aesthetic point of view. Moreover, they possess value for the owner from a social perspective, to be watched or exposed to others (art, sculptures), since the only cash flow they can generate is in case of a sale. Wine is the only particular collectible that is also an

(16)

15 exhaustible asset, since once consumed, its value is lost. Moreover, the uniqueness of fine art works, or rarity in case of diamonds, is that kind of specific feature that enhances the value of treasure assets.3 Nevertheless, collectibles are traded like any other asset, albeit on a market with a less developed infrastructure, and that constitutes the reason why they should also be analyzed from a financial point of view. Collectibles have a financial life of their own, with returns, volatility and correlations that can be computed and used in forming portfolios, in spite of scarcity of information which represents the main drawback. Recent academic papers have provided evidencethat investing in a series of art sectors in order to hedge against market risk for a particular artist or category within an art portfolio is common practice among investors. More importantly, following the real estate crisis, economists have coined the term SWAG, as in Silver, Wine, Art and Gold, referring to real assets with enhanced performance potential, placing collectibles at the center stage of alternative investment solutions.

The market for art objects and collectibles is currently on a recovery path. In the last year, TEFAF4 estimates the global art market turnover at €47.4 billion (in total sales of art and antiques), close to its highest ever recorded total, and an increase of more than 150% in the last decade.

Figure 1: Evolution of global art trade

Source of data: TEFAF Art Market Report 2014

3_{Belk, R., 1995}

(17)

16 Looking back at the evolution of art trade in the last decade, the market experienced a boom from 2003 to 2007, with sales more than doubling in value. However, this positive trend turned in 2008, at the height of the global financial crisis, with aggregate values and the number of transactions both contracting by close to 40% within two years. Since 2010, the market for art has been on a recovery path, with 2013 turnover coming close to the maximum pre-crisis levels of 2007.

Figure 2: Global Art Market Share by Value in 2013

U.S. and China control nearly 70% of the art market in terms of sales volume. Although the records posted during the last couple of years suggest the art market has become increasingly excessive, it can also be concluded that it has been constantly evolving in spite of the various crises experimented, and that 2013 has been the best year in the history of the art market.5

(18)

17

3.2 Investing in emotional assets: Benefits and challenges

The reasons why collectibles are a smart investment can be easily identified. Apart from enabling investors to diversify their investment portfolio, these “hard” tangible assets offer long-term stability and, what some consider, a safe placement. However, the majority of those who acquire art have stated that their decision is motivated by pleasure (62% of the HNWI6 surveyed by Barclays invest in art motivated by enjoyment)7. One study, by professors Mei and Moses, finds that art has actually outperformed equity during the past decade, which if true will determine a large number of speculators to turn their attention towards the art market. The fact that corporations have also an important presence in the art market, deserves a mention; they account for 7% of sales in 2013, as seen in Figure 3.

Figure 3: Market Share of Sales by Buyer Group in 2013

Emotional assets may also prove more resilient during financial crises since they are not directly linked to any particular industrial activity, and since the motivation behind the investors‟ behavior is not purely

6_{High Net Worth Individuals} 7_{Barclays, 2013}

(19)

18 economic. This argument stands even though contractions of financial markets have been assumed to lead logically to lower cash resources for the wealthy collectors, which will in turn affect their investment in emotional assets. Nonetheless, this theory will be assessed in the empirical part of this paper, when art‟s performance during financial crisis will be studied. Another aspect that deserves a mention is that collectibles trigger important taxation advantages8, and this constitutes another reason to encourage investment in emotional assets.

However, there are quite many drawbacks linked to emotional assets that deserve a mention. Firstly, a major obstacle is the lack of regulated easy-to-trade markets, with collectibles being currently traded at auctions or on “over-the-counter” markets. In a study survey by Deloitte, 73% of private banks agree that the absence of a common and standardized market for art is the main hurdle standing in the way of its development as an asset class.9

Collectibles also demand high costs to be stored and preserved. Being “hard” tangible assets they need to be safely stored, and in case of art, even restored to ensure their shape and value is maintained. Like every highly-valued object, works of art also demand costly security measures. Since emotional assets are generally held over long time periods, storage costs are an important component which needs to be accounted for when computing real return rates of collectibles. Another well-known issue associated with the art market is illiquidity. Illiquidity implies a sizeable spread between bid and ask prices which can be seen as a hidden cost for trading in emotional assets. Moreover, acquiring treasure assets like paintings or diamonds represents a large investment, which in turn requires the portfolio to be large enough to benefit from the diversification. In conclusion, the high value of these assets combined with the low liquidity of the market enhance the risk undertook by the buyer, especially in case he wants to resell the item.

Weak transparency adds further to the difficulties of investing in collectibles since information on emotional assets is both scarce and costly. Data for the risk-return profile of emotional assets is not as easy to come by as for other asset classes, and accurate expertise is in short supply. However, investors can only rely on Mei & Moses Fine Art Index, artprice.com or Art Market Research, which are boutique indexes yet to be given the stamp of approval by any formal agency.10 As the nature of the art market only makes it possible to obtain quarterly or, at best, monthly data, the comparison with other assets is problematic and inaccurate. Mamarbachi et al. (2006) name other drawbacks common to emotional assets: Markets exhibit a weaker equilibrium process compared to other financial assets, and because the equilibrium prices cannot be determined, objective valuations are impossible. In many cases elasticity of

8_{McAndrew, C., 2010} 9_{Deloitte, 2013}

(20)

19 supply is low or even nil (i.e. in case of deceased artists). Owners of art works can be regarded as holding a monopoly on the respective items, which places any other investor willing to buy in a weak position to negotiate.

3.3 Art funds

The past decade has seen the emergence of a number of funds that specialize in art investment. These hedge funds of the art world appear to offer a valuable new form of investment or, at the very least, a way to hold wealth with security for the longer term. The main attraction of this particular investment opportunity lies in its highly beneficial diversification strategy in addition to extremely low correlation with the main financial securities. Direct investment in art has been common practice for wealthy collectors for centuries. Nevertheless, these innovative art funds offer investors structured solutions and the opportunity of buying a stake in a diversified art portfolio which actively trades in art works exclusively for financial profit.

The London-based Fine Art Fund Group has been launched in 2001 and is the most established fund operator with over $200 million in assets. Since then, ARTESTATE, Société Générale Asset Management and The Art Trading Fund have all raised sufficient capital to provide investors with indirect investments in the art market.11 An art fund is operated by a combined team of both art and financial professionals and offers customized financial vehicles and accounts, independent art consultancy and serves as a third party expert to major banking institutions worldwide. The key objectives are to generate long term capital growth and utilize its international art market and financial expertise for the benefits of its clients, as well as assisting them in diversifying their portfolios. Additionally, clients are offered the opportunity to borrow works of art held by the fund, for their private enjoyment. In order to enhance the provenance and value of the items they own, art funds lend artwork to museums and galleries for public viewing.

Referred to as the “hedge funds of the art world”, art funds are akin to those when it comes to the fees charged. Management fees range between 1-3% of all assets, and the fund also keeps 20% of the profits. There is also the matter of underlying costs of purchasing and selling the items (commissions to dealers and auctioneers), which together with taxes are reducing the fund‟s capital. Moreover, art funds may encounter specific issues like “facing resistance from the collecting world, and finding it difficult to

(21)

20 incentivize market insiders into sharing valuable information”12

, or debates between managers and investors over evaluation of performance. The illiquidity of the art market also translates into a serious problem, as most art funds are not able to sell a major part of their artworks from one day to the next, because they are usually held for at least 18 months.

In the recent years, these companies seem to have flourished as a consequence of newly available research, availability of new art indices and strong investor interest fueled by strong performance of the art market. Other benefits for investing in art funds are: access to insider knowledge of the market, art expertise, lower transaction costs than individual investors, due diligence services, professional advisory services and potential for increasing return rates. However statistical data on how these companies are performing is scarce and difficult to access, while some authors observed that there have been “very few examples of successful collectibles investment funds to date” (Horowitz, 2011). Motivated by these contradictory findings, this research is intended to shed light on the risk-return characteristics of a portfolio formed of art works or emotional assets, identical to those formed by art fund managers. Aside from the maximized return rate that can be achieved, it is also of interest to analyze the optimal allocations to each asset class which are strongly influenced by developments and trends in the art world. As reported by Deloitte, the global art fund industry recorded an asset under management increase of 69% in 2012, driven by growth in art investment trusts in China. Out of the 83 art funds and trusts that were in operation in 2012, 58 had been established in China since 2009.13

Additionally, the recently founded Alternative Investment Fund Managers Directive (AIFMD) provides a new standardized framework for all alternative investments and offers investors safeguards and controls not previously available. The necessity of introducing this measure derived from the lack of transparency which was one of the main causes of the 2008 financial crisis. The directive actually regulates the managers and not the products, by providing minimum requirements to ensure the adequate safekeeping and conservation of the assets, minimum requirements for the independent and transparent valuation of these assets as well as minimum requirements for the management of risks.

12_{Dimson, E., Spaenjers, C., 2010} 13_{Deloitte, 2013}

(22)

21

4. EMPIRICAL STUDY

In this section, I use time series data spanning from 1998 to 2013 to conduct three interrelated studies that will help answer the research question and offer additional insights into the financial side of emotional assets.

First study consists of estimating the risk-return characteristics of art over the sixteen years and comparing it to that of the common financial securities. Additionally, using the variance-covariance matrix of these assets, I determine the optimal allocation of art in an efficient portfolio and use the results to evaluate the diversification benefits of investing in emotional assets.

The second study is based on a subsample of nine years centered around the financial crisis of 2007 which will be divided in three periods in order to estimate the performance of art before during and after the crisis.

Finally, the third study uses price indices of five different emotional assets and five different art categories to build efficient portfolios that replicate the optimal investment strategy of an art fund manager. These results will make it possible to compare indirect with direct investment in art.

4.1 Methodology

The methodology of this empirical study consists of determining the yearly return rate and standard deviation for a series of emotional and traditional assets by applying Markowitz Mean-Variance Optimization to portfolios formed with these assets in order to determine their risk-return statistics, as well as their optimal allocations. In order to estimate the allocations, it is necessary to make an assumption about the expected return of all asset classes considered. In this case, the best prediction of future returns is the historical distribution, and these will be estimated by studying a 16 years interval starting 1998 until 2013. Additionally, there will be no assumptions regarding a maximum allocation limit for one particular asset, in order to ensure the portfolios lie on the efficient frontier.

(23)

22 I begin by defining the following variables:

(

)

(1)

where is the vector containing the expected individual returns for the n assets;

(

)

(2)

where is the portfolio weight vector of the n risky assets;

( )

(3)

where is the unit vector;

(

)

(4)

where is the covariance matrix.

The Markowitz efficient frontier is represented by the complete range of optimal portfolios that offer the highest level of expected return for a defined level of risk, but for the purpose of this study I will look closely at the portfolio with the lowest volatility (Minimum Variance Portfolio) and the one with the highest Sharpe ratio14 (Tangency Portfolio).

Minimum Variance Portfolio is subject to the following conditions:  minimize portfolio variance

(5)

Moreover, following restrictions on portfolio weights apply:

 sum of portfolio weights equals 100%

(6)

 short-selling is not possible

(7)

(24)

23

Tangency portfolio15 is subject to the following conditions:

 maximize portfolios Sharpe ratio

√

(8)

Restrictions on portfolio weights:

 sum of portfolio weights equals 100%

(6)

 short-selling is not possible

(7)

No additional constraints regarding a maximum weight limit for each asset have been included, since such a condition would mean the portfolios are not efficient. Nonetheless, in order to protect themselves from the risk of financial crises and unexpected devaluation of a certain asset class, investors can diversify their investment by limiting the allocations to certain assets, even if this means efficiency is partially lost.

4.2 Data

The fact that art funds have only recently been established means that data on their performance is scarce. Moreover, since two main characteristics of the art market are illiquidity and opacity, official data on the financial performance of collectibles is inexistent.As a consequence, this empirical analysis is based on estimated return rates derived from a merged database of art indices and return indices of traditional financial securities. This will offer a glimpse into the risk-return profile of art funds, which generally have to choose between investing in a variety of emotional assets or only in paintings, as well as offer a good indication of the performance level of a diversified portfolio including art.

The dataset spans over a period of 16 years and was acquired from Artprice, one of the leading databanks in art market information, and from Datastream. It consists of quarterly observations, starting from the first trimester of 1998 until the last trimester of 2013. Artprice constructs its indices using hammer prices before fees16 which include fine art sales and exclude antiques, furniture and anonymous cultural goods. The figures exclude taxes and buyer‟s premiums as well as all the private sales carried out by auction houses (unavailable to the public). The dataset includes indices for five different emotional assets classes

15_{Tangency portfolio is defined as the efficient portfolio at the point where the Capital Market Line is tangent to the} efficient frontier and is characterized by having the maximum Sharpe ratio.

(25)

24 (Paintings, Prints, Photographies, Sculptures and Drawings), five different painting periods (Old Masters, 19th Century, Modern Art, Post-war and Contemporary), as well as a Global Art Index17.

In order to determine the performance of the traditional financial securities (stocks, government bonds, corporate bonds and commodities), the following price indices have been considered: S&P500 Composite18, CGBI Trsy. 1-10Y19, CGBI Corp. 1-10Y20, GSCI Gold Total Return21.

The return rate of the market i has been calculated by employing the continuously compounded returns method:

(

)

(9)

where r is the return rate and p the auction price.

After determining the average quarterly return rate for every asset class in the database and annualizing the results, the following statistics have been obtained (see Table 2).

Table 2: Summary of the risk-return profile of collectibles and financial securities

17_{Indexed auction records are based on Fine Art and Design cataloged auctions (paintings, sculptures, drawings,} photographs, prints, watercolors, etc.) recorded by artprice.com, except antiques and furniture.

18

S&P 500 Price Index USD

19_{Citigroup 10 year Treasury Bonds Total Return Index USD} 20_{Citigroup 10 year Corporate Bonds Total Return Index USD} 21_{S&P GSCI Gold Total Return Index USD}

Paintings Prints Sculptures Drawings Photographies

Nominal return 2.75% 2.90% 2.79% 8.88% 4.86%

St. deviation 8.12% 8.37% 9.80% 14.00% 15.14%

(26)

25 Note: Table 2 displays the geometric mean nominal returns on art (Artprice Global Art Index), equities

(measured by the Standard and Poor’s 500 Index), gold (Standard and Poor’s world index for gold prices) corporate bonds (10 year U.S. corporate bonds) and government bonds (10 year U.S. treasury bonds). The second row displays the distribution of the returns measured by the standard deviation of the

corresponding distribution. The third row shows the Sharpe ratio of the corresponding asset class, computed as the arithmetic average excess return divided by its standard deviation.

Figure 4: Return indices of traditional assets and art (base year=1998)

Old Masters

19th Century

Modern

Art Post-War Contemporary

Nominal return 0.62% 0.50% 2.10% 5.90% 6.06%

St. deviation 13.94% 10.99% 7.98% 10.62% 14.01%

Sharpe ratio 0.04 0.05 0.26 0.56 0.43

Art S&P500 Gold Corporate

bonds Treasury bonds Nominal return 6.79% 4.75% 10.45% 6.20% 4.74% St. deviation 21.15% 14.16% 17.39% 4.95% 3.2% Sharpe ratio 0.32 0.34 0.60 1.25 1.48

(27)

26

4.3 Results and interpretations

4.3.1 Long-term diversification benefits from investing in art

Firstly, Table 2 reveals the individual risk-return profile of each asset class. Considered in yearly nominal terms, art (6.79%) has generated higher returns than stocks (4.75%) and treasury bonds (4.74%), and has been approximately on par with corporate bonds (6.20%), albeit considerably more volatile. Art also has been the most volatile (0.21) of all the five assets, even more than gold (0.17) which, however, has provided a very high level of return, in excess of 10% p.a. Corporate bonds (0.04) and treasury bonds (0.03) have been the least risky securities, with the latter proving to be the optimal choice of all with a Sharpe ratio of 1.48.

By analyzing the risk-return summary of the five different emotional assets contained in the dataset, one can observe that performance varies significantly between different collectibles categories. While Paintings (2.75%), Prints (2.90%) and Sculptures (2.79%) have generated approximately the same return rate and standard deviation (0.08 and 0.09), Photographies (4.86%) and Drawings (8.88%) have performed significantly better but have also been more volatile (0.15 and 0.14). Judging from the Sharpe ratio perspective, the stand-out emotional asset over this sixteen year period is Drawings (0.63), with the high risk it carries being justified by the considerable return rate. This result can be linked to a propensity to buy drawings in China, where they carry both a historical and cultural meaning.22

After studying art by different classes, one can draw the conclusion that some categories of art have secured a significantly higher rate of return than others over the last sixteen years. On one hand, items belonging to Old Masters (0.62%) and 19th Century (0.50%) have had an underwhelming performance, while Post-War (5.90%) and Contemporary (6.06%) paintings have secured a considerably higher return. Furthermore, the two underperforming classes (0.13 and 0.11) have been as volatile as the latter two (0.10 and 0.14), such as it can be concluded that they compare unfavorably to the latter two. This is obvious after analyzing the Sharpe ratios: Old Masters (0.04) and 19th Century (0.05) compared to Post-War (0.56) and Contemporary (0.43). The other category, Modern Art, offers a modest return (2.10%) but may prove attractive to risk-averse collectors as it is the least volatile of all (0.08).

(28)

27

Table 3: Correlation matrix of art and financial securities

S&P500 Gold Corporate

bonds Treasury bonds Art S&P500 100.00% Gold 6.43% 100.00% Corporate bonds 26.17% 39.44% 100.00% Treasury bonds -44.68% 20.66% 44.15% 100.00% Art 3.77% -10.80% -6.69% -24.76% 100.00%

Note: See Table 2 notes.

In order to evaluate the benefits of including art in a portfolio for diversification reasons, a correlation regression analysis was realized and the results are displayed in Table 3. The correlation matrix reveals that art as an asset class can play an important role in portfolio optimization as a diversification mean and to hedge risk. It is negatively correlated with treasury bonds 0.24), gold 0.10) and corporate bonds (-0.06) and there is also positive but reduced correlation with stocks (0.03). This tends to confirm the hypothesis that the reduced level of development of the collectibles market and scarcity of financial speculators is determinant for a low or even negative correlation with the commonly traded financial securities. In addition to this, the positive relation with the stock market reveals that a strong performance of stocks will also lead to better results of art. These results can be linked to the Mei and Moses‟ (2002) study which reveals both a low positive correlation with the S&P500 index (0.04) and a negative correlation with government bonds (-0.15), for a longer time interval (1950-1999). Moreover, their nominal return rate for art over this period (8.20%) is approximately close to that of this research (6.79%).

(29)

28

Table 4: Asset allocation and portfolios performance (art and traditional asset classes)

MVP TP S&P500 11.16% 10.91% Gold 0.00% 0.91% Corporate bonds 0.00% 0.00% Treasury bonds 84.67% 83.45% Art 4.17% 4.73% 100.00% 100.00% Return 4.74% 4.80% St. deviation 2.37% 2.39%

Note: Table 4 displays the optimal allocations to each asset class in Markowitz efficient portfolios formed over a long-term period of 16 years from 1998 to 2013, as well as the yearly nominal return rate and the

standard deviation of the portfolio. First column shows the allocations, return rate and standard deviation of the Minimum Variance Portfolio which is determined by minimizing the variance. Second column reveals the allocations, return rate and standard deviation of the Tangency Portfolio which is

determined by maximizing the Sharpe Ratio.

The allocations for the portfolios combining art with other financial securities are as follow: for the MVP portfolio art receives an allocation of 4.17% while treasury bonds (84.67%) and S&P500 (11.16%) make up for the rest of it. Although most of the weight is placed on government bonds, art contributes to the optimization process, while gold and corporate bonds do not receive any allocation at all. The results are in accordance with financial theory which considers treasury bonds as the least volatile security, and this portfolio minimizes overall volatility. The return of the MVP portfolio is 4.83% with a standard deviation of 2.37%.

In the second portfolio, based on Sharpe ratio maximization, the art allocation is 4.73%, while treasury bonds again hold the majority with 83.45%; stocks (10.91%) and gold (0.91%) are also included. Yet again, corporate bonds do not contribute and the weight on gold is minimal, despite having produced the highest return rate. The return of the Tangency portfolio is 4.89% with a standard deviation of 2.39%.

(30)

29

Figure 5: Risk Return Trade-Off and the Efficient Frontier

Note……….. Note: See Table 2 notes.

These results confirm that, even though art alone is not an attractive financial placement, taken into a portfolio with other financial securities, it has a role to play, mainly on account of its negative correlation with the other asset classes. By analyzing Figure 5, it can be observed that even though art as an individual investment is far from the efficient frontier, it receives an allocation in the optimal portfolios, while corporate bonds do not, in spite of being very close to the frontier. This emphasizes the importance of the correlation coefficients between assets and supports the claim that emotional assets offer diversification benefits to risk-averse investors looking to diversify their portfolios.

(31)

30

4.3.2 Optimal allocation to art investment during financial crises

The results from previous section have revealed that art can work as a potential hedge or diversification solution in the long-term. However, the turn of events in recent years makes it interesting to study how emotional assets perform in times of financial crisis. This analysis is of interest because consequences like poor stock returns and excessive volatility of financial markets have affected deeply the portfolios and wealth of financial investors. Given the disastrous performance of traditional assets, it becomes vital to determine what viable alternative investments exist, and in this particular case, whether collectibles perform differently from other securities during crises.

To properly analyze the effects of the crisis, this section will study a subsample of nine years, divided in three-year periods and centered around the crisis of 2007 as follows: the pre-crisis period (2004-2006), the crisis period (2007-2009) and the post-crisis period (2010-2012).

Table 5: Summary of the risk-return profile of art and financial securities by period

Art S&P500 Gold Corporate

bonds Treasury bonds Pre-crisis 6.08% 9.85% 19.51% 4.24% 2.73% Crisis -1.39% -5.27% 21.12% 6.51% 6.54% Post-crisis 11.65% 8.82% 16.82% 7.75% 4.19%

Note: Asset classes are explained in Table 2 notes. First row displays the average nominal return rate for each asset class during the period 2004-2006. Second row – period 2007-2009. Third row – period

2010-2012.

At a first glance, it can be observed that, like stocks, collectibles register a negative return rate during the crisis but not of the same magnitude. At the same time, art recovers better than the S&P 500 in the period after the crisis, generating a nominal return rate of 11.65% p.a., almost twice the rate before the crisis (6.08%). However, bonds and commodities have performed considerably better than collectibles during the crisis, giving a first indication that investing in art in times of financial turmoil is not an optimal decision.

(32)

31

Table 6: Correlation of art and financial securities by period

S&P500 Gold Corporate

bonds Treasury bonds Pre-crisis -0.41 -0.21 -0.44 -0.51 Crisis 0.17 -0.06 0.17 -0.10 Post-crisis -0.17 -0.24 -0.27 -0.26

Note: Table 6 displays the correlation coefficient of art with four traditional asset classes defined in the notes of Table 2. The rows represent the three periods described in the notes of Table 5.

By studying the correlation coefficients of art and the traditional asset classes during the three periods, it can be observed that art is negatively correlated with all the other four securities before the crisis. However, during the crisis, it becomes positively correlated with stocks and corporate bonds, and the correlation coefficients with gold and treasury bonds, although remaining negative, record a significant drop in absolute value. This represents a very strong indication that emotional assets lose their long-term diversification advantage during financial crisis, and proves that art is not an effective hedge in this particular scenario. To establish the authenticity of these results, this finding can be linked to the conclusion reached by Dimson and Spaenjers (2014) that high-end art prices dropped substantially during the time of the 2008 financial crisis as a consequence of “negative shocks to wealth or top income levels”.23

In the last phase of this analysis, results on the individual performance of art and the four traditional securities over the nine-year subsample, along with the correlation matrix, serve to determine optimal Markowitz portfolios for each period. The focus will be on Tangency portfolios which maximize the Sharpe ratio and results are displayed in Table 7.

(33)

32

Table 7: Asset allocation and portfolios performance (art and traditional asset classes) by period

Pre-crisis Crisis Post-crisis

S&P500 9.70% 0.52% 14.34% Gold 4.90% 0.00% 4.63% Corporate bonds 0.00% 15.32% 77.22% Treasury bonds 69.93% 84.16% 0.00% Art 15.48% 0.00% 3.81% 100.00% 100.00% 100.00% Return 4.71% 6.47% 8.46% St. deviation 2.01% 3.00% 2.20%

Note: Rows are defined in the notes of Table 2. Columns are defined in the notes of Table 5.

Results reveal that art has an optimal allocation of 15.48% before the crisis but it becomes zero during the crisis period, with treasury bonds and corporate bonds making up for 99.48% of the portfolio. After the crisis, art‟s allocation rises to 3.81%, while the largest part of the portfolio consists of corporate bonds (77.22%) and the S&P500 Index (14.34%). Although art appeared to have recovered better than other assets with an individual return rate of 11.65% (Table 5), its negative correlation with the other assets becomes weaker post-crisis, and consequently loses its diversification advantage.

To conclude, this section of the paper studies the performance of emotional assets, both individually and in relation to other securities, during different stages of financial crises. Firstly, the results reveal that art returns are deeply affected by the crisis, similarly to stock markets. Secondly, one can observe that collectibles are negatively correlated with traditional assets before and after the crisis, but the coefficients change drastically in the crisis period, revealing a positive correlation with the stock markets. Finally, the fact that art loses its diversification attributes during crisis is confirmed by studying optimal portfolios by period, with the allocation for art dropping to zero after year 2007. A logical interpretation for this development is that when stock markets collapse, the shock to the wealth level of High Net Worth Individuals reduces the demand for emotional assets, which results in illiquidity and depressed prices on the art market.

Performance of emotional assets : an optimal portfolio allocation analysis

Master Thesis: