Index for all paintings: January 1906 - June 2004.
Indices for Flanders, France, Great Britain plus Ireland, and Holland.: January 1990 - June 2004.
Frequently Asked Question:
What is an art index?
In order to understand what an art index tells you it helps to first understand what a stock index tells you. Consider the Spiegel 2; an equally weighted index which contains two stocks. Call these stocks A and B. Day one begins with a portfolio valued at $100 that holds $50 worth of A and $50 worth of B. On day two A has risen to $53 and B to $51. The portfolio is now worth $104 and that is also the new value of the index. At this point the portfolio is rebalanced so that it owns $52 of each stock (i.e. sell $1 worth of A and buy $1 worth of B). Each day you do this produces a new index value that tells you how an equally weighted stock portfolio in A and B has fared over time.
An art index is designed to work similarly to a stock index and tell what the return has been to somebody owning a portfolio of paintings. Unfortunately, unlike stocks paintings do not sell every day or even every year. One solution is to create an index known as a "repeat sales index." Roughly, this index uses two sales on the same painting to calculate the return to holding that painting between sales. For example, if a painting sold for $10,000 in 1985 and then for $15,000 in 1995 the total return was 50% which yields an annual return of 4.1% (1.51/10). By calculating these returns for a large number of paintings one can then estimate what the average return was in any one year. Here is a simple example. Painting C sells in years 1 and 3 and during that time returns 5% per year. Painting D sells in years 2 and 3 and returns 3%. The repeat sales index would start by estimating the return from years 2 to 3 at 3% in order to perfectly fit the return on painting D. Given this it would then estimate the return to art from years 1 to 2 at 7%, which will then allow the model to also fit the return to painting C. The index would then have the following values:
Year 1: 100
Year 2: 107
Year 3: 110.21
This is a very simplified example and hides quite a bit of the mathematics involved when there are many more paintings. Interested readers can find the details for creating a repeat sales index in the February 1995 paper "Non-Temporal Components of Residential Real Estate Appreciation," Review of Economics and Statistics that I wrote with William Goetzmann. An application of the repeat sales methodology with art data can be found in the April 1995 article "Private Value Components, and The Winner's Curse in an Art Index," European Economic Review that I also wrote with William Goetzmann. That study uses a repeat sales art index to examine the degree to which investors might be willing "overpay" for art relative to its investment value.
