1 WHAT MODEL HAVE WE USED AND WHY?
The Index was constructed using the hedonic pricing model. Constructing an index for individually unique objects like artworks requires a different approach than used for an index of stocks, bonds and commodities. This is due to the infrequency of trading and differences in the characteristics of the objects that come to market from period to period. Thus, an index based on average prices over a period of time may be more dependent on the mix of objects that come to market, rather than changes in the underlying market itself.
Artworks are heterogeneous assets, with a variety of physical and non-physical characteristics that make an artwork unique, including artist reputation, the material used, the period during which it was produced and subjective traits like painting styles. The price of an artwork, to some extent, depends on these characteristics. The hedonic approach estimates (econometrically) the value attached to each one of the characteristics that are deemed to be significant in the determination of the price. The hedonic regression method controls for quality changes by attributing implicit prices to a set of value-adding characteristics. In other words, hedonic regressions strip the observable characteristics from the artworks to obtain an index reflecting the price of a “standard” artwork. Within this framework, time variables are treated as a sub-group of characteristics that influence the price of a painting. Actual price indices are then constructed from the associated estimated coefficients on the time variables. Thus, the hedonic approach circumvents the problem of heterogeneity and avoids the exclusion of single sale data.
2 WHY DID CITADEL DECIDE TO DEVELOP AN ART PRICE INDEX?
Many of our clients are avid collectors of art, both on an aesthetic level and, in some cases, as part of a diversified investment strategy. As Advisors, we realised there was no consolidated view of the industry as a whole, that could define art as a legitimate alternative asset class by allowing real objectivity, transparency, measurability and comparability over time. We thus developed the Citadel Art Price Index as a means of assessing the change in value of South African art and demystifying the asset class.
3 WHAT IS THE HEDONIC REGRESSION METHOD?
The hedonic regression method therefore controls for quality changes by attributing implicit prices to a set of value-adding characteristics. In this way, hedonic regressions strip the observable characteristics from the artworks to obtain an index reflecting the price of a ‘standard’ artwork. Within this framework, time variables are treated as a sub-group of characteristics that influence the price of a painting. Actual price indices are then constructed from the associated estimated coefficients on the time variables. The hedonic approach thus circumvents both problems of heterogeneity and exclusion of single sale data associated with the other methodologies.
4 WHAT IS THE MAIN BENEFIT OF DEVELOPING AN INDEX SUCH AS THIS?
It allows us to assess art as an alternative asset class and will also benefit the industry as a whole. The purpose of the index at this point is not to value specific artworks or artists, it is merely a view of the industry as a whole and the price fluctuations that occur over time.
5 ARE WE NOW ADVISING CLIENTS TO CONSIDER ART AS AN ALTERNATIVE ASSET CLASS IN THEIR DIVERSIFIED PORTFOLIO?
No, not at all. We understand that the purchase of art in many cases is still based on aesthetic preferences and individual taste. What we are saying is that if you are including art as an investment decision, ensure you have a means of assessing the industry and an understanding of what the asset class might do over time. This index will assist in determining the relative value of this asset class with other asset classes. It will also assist an investor with timing in the art market as periods of gross over-or undervaluation could be identified and transactions could be planned accordingly.
As this index builds up history and captures industry application, it will enable us to unravel the links between art and other asset classes. A deeper understanding of the timing and leverage of the wealth effect will enable us to produce much more in-depth research on the entire asset class.
6 WHO DID WE PARTNER WITH?
In order to compile the index, we partnered with AuctionVault who provided us with the input data – results from auctions captured since 2000. Construction, modelling and testing of the index was initially done by Econex, an independent economics consultancy based in Stellenbosch. The model was later handed over to Citadel’s investment team who updates the index twice a year.
7 WHAT DATA DO WE USE?
We use hammer prices of art sales achieved at a number of local (and one international) auction houses. Prices exclude commissions and VAT.
8 WHICH AUCTION HOUSES ARE INCLUDED?
We selected 5 of the top, most reputable, auction houses in South Africa as well as Bonhams and Christie’s in London. The auction houses are:
5th Avenue Auctioneers
Russell Kaplan Auctioneers
Strauss & Company
Stephan Welz & Company
They were chosen based on the regularity of their auctions as well as their stable and reliable provision of sales results.
9 WHY ARE GALLERY SALES NOT INCLUDED?
Reliable information is not readily available as gallery sales and prices achieved are not public.
10 WHY ARE ONLY HAMMER PRICES USED IN THE INDEX GENERATION AND COMMISSIONS EXCLUDED?
Hammer prices are the only constant prices that are published and therefore the most reliable indicators of price movements over time. While we model actual prices, only the model coefficients (and not the prices) are used to generate the index. Also (as with the CPI), we are not explicitly interested in the level of the index, but rather in changes in the index over time. If commissions have stayed roughly the same over time, it would not alter the index values.
If commissions vary by auction house, it would also have no effect on the index. Commissions would only become problematic if they have changed substantially over time. The same principle applies to VAT. Obviously if there are dramatic changes to the magnitude of these costs (such as, for example, VAT increasing by 5 percentage points) this will be reassessed.