Econometric Methodology

Legals

Definitions

PRICE LEVEL:

Measure of a good's average value over a given period of time. In the case of an artist, the price level corresponds to the price of a typical work in a specific medium.

To find an artist's price level is basically to calculate the value of an 'average' work reflecting the artist's output. We propose an average price without calculating the arithmetical average (this indicator is not significant for non-standard goods. It only works where each item is comparable): the typical hammer price paid is not the average of all prices paid. The value is expressed in monetary units (euros or US dollars).

Use: the price level provides an estimate of the average unit price of an artist's work and allows users to monitor change in this value over time. INDEX:

The rise in prices from a base date, generally given the value 100. If a price index (base 100 in 1997) is at 120 in 2000, prices have risen by 20% in the intervening period. An index at 80 in 2000 would mean prices had fallen by 20%.

Mathematically, the indices for all media combined are derived directly from price level (not the medium price level but another more abstract price level combining all media covered by the artist). They provide a guide to relative prices commanded in different years.

Use: the index tracks the market's trend for the artist in question. All artists are measured against the same price scale (base 100 in 1997). It therefore allows users to objectively compare the trends in the market value of different artists.

When interpreting the results, users should consult the benchmark if they want to know prices for a particular artist working in a particular medium and the index if they want to know their prices across all media.

Calculations

Econometrics:

the econometric model is a schematic way of representing a phenomenon using equations where the variables are economic data.

Such models aim to capture the key aspects of a real-life situation which can then be modelled in stylised form. The econometric model is thus a schematic and partial representation of a naturally more complex reality. The technique is to include only the data that is relevant in determining the targets set. The choice of data depends on the nature of the market, the type of decision to be made or the type of research being carried out.

To be valid, the models used must be based on one of several theories. Techniques vary depending on the type of information that is sought. We cannot apply the same methodology to estimate the total market, as for a particular segment or a particular artist. The number of data used in the alculations are highly variable and so the quality of the indicators is heavily affected by the choice of method. When calculating artist benchmarks, the most appropriate method is the

hedonic price method.

The hedonic price method. this is used for goods which are heterogenous (housing, wines, art, etc) where quality is the key determinant of price.

Underlying assumptions:

  • Works are non-standard, not only within a similar range (ie, of the same artist) but particularly between different artists.

  • The price of a work depends essentially on its quality. This last is to a great extent quantifiable. The explanatory variables being either intrinsic characteristics of the work (the author, size, format, technique, materials, date, etc) or related to the sale (auction house, location of sale, date of sale).

Technique:

the first step is to quantify the effect of quality on price. This analysis is not based on the work itself but on the various elements that go into it. Works of art are defined by a matrix of features. Using these estimates we deduce a set of equations that quantify the influence of each characteristic.

From these we can create synthetic works for each artist. Having estimated the parameters it is possible to find the average price of a composite work for each period.

All auctions relating to an artist are included in the calculation thus avoiding any mistakes from selectivity. The data universe is fully inclusive.