One of the enduring enigmas of investing in art is valuation. Unlike traditional financial assets that are largely priced according to expected future cash flows, fine art has no future cash flows except for the ultimate sale proceeds. In this regard art is similar to a zero coupon bond. Of course, the future selling price of any piece of art is unknowable in the present. This presents any would be financial analyst of the art world with a conundrum. A fair valuation - and by extension a reasonable performance estimate - cannot be made without knowing the future sale price of a work. Using expected future cash flows as a valuation method for art and antiques is therefore a doomed exercise in circular reasoning.
For a long time, this problem bedeviled me. Relative valuation, in comparison, is a much more approachable riddle in the world of antiques; items of higher quality and condition are generally worth more than those of lower quality and condition. But absolute valuations stumped me. How do I know that the antique I'm considering is priced fairly compared to stocks, bonds or other traditional financial assets?
But then it hit me. Maybe I was approaching this issue from the wrong angle. Rather than trying to determine valuation from unknowable future cash flows, maybe I should be looking at the amount of money available in an economy to buy art. In other words, I should look at the inputs (available money) rather than outputs (future sale prices). According to my hunch, inputs in this instance would be nominal (non-inflation adjusted) GDP (gross domestic product) - the value of all goods and services produced in an economy during a year.
So I started gathering data. Now historical price information for art and antiques is somewhere between difficult and impossible to find. However, after significant effort, I managed to put together a miniature index of U.S. type coins with price data back to 1950.
This index consists of three equally weighted constituents:
1) 1842 Liberty Seated silver dollar in XF condition (mintage: 184,618)
2) 1857 Flying Eagle cent in XF condition (mintage: 17,450,000)
3) 1908-D no motto $10 Indian Head gold eagle (mintage: 210,000)
These coins were very specifically chosen. They are all classic pieces that have been popular among U.S. coin connoisseurs for many decades. Although not plentiful, none of the pieces is a rare or key date. Only one coin is gold, limiting the impact of gold spot price fluctuations on the overall index. And because they are all U.S. coins, we can surmise that nearly all collector demand for them originates in the United States. This allows us to exclusively look at U.S. economic data in an attempt to find a meaningful correlation. In addition, our U.S. type coin index, possessing all the desirable attributes of investment grade antiques, should be an excellent proxy for the broader art and antiques market. So any relationship with economic data we find here will most likely carry over to that larger market.
This first chart shows our proprietary U.S. type coin index graphed from 1950 to 2015 against both nominal U.S. GDP and the U.S. inflation rate (CPI). You'll notice the very close relationship between the coin index and nominal GDP. The correlation coefficient between these two data sets is 0.943, a very high number indicating almost perfect correlation. To put this value in perspective, a correlation coefficient of -1 means two data sets are perfectly inversely correlated. A coefficient of 0 means two data sets are completely non-correlated, moving randomly in relation to each other. And a coefficient of 1 equals a perfect correlation. Inflation, useful as a reference baseline, lags far behind. The coin index and nominal GDP moving in almost perfect sync seems to confirm my theory that nominal GDP is the primary driver of collector demand. This makes a lot of sense, as art aficionados and antique collectors are constrained by their available resources and nominal GDP is a reasonable measure of those resources.
But we can take this data one step further. We can plot a ratio of our coin index versus nominal GDP to reveal periods of overvaluation and undervaluation. This is what our second graph shows. In this case, the long term average has been indexed to 1. So any number around 1 implies fair value while those significantly above signal overvaluation and those substantially below show undervaluation.
The results of our graph are intriguing. We can see that most of the period from the 1960s through the 1980s shows persistent overvaluation in the coin index. This period also broadly coincides with an elevated rate of inflation. So the story seems fairly straightforward. People who fear inflation flee underperforming stocks and bonds for the relative safety of tangible assets, including bullion, art and antiques. This phenomenon eventually pushes these tangible asset classes into overvalued territory. Our proprietary coin index almost reached double its fair value in 1975.
The period on our chart from 1995 to 2015 is equally illuminating. Once inflation receded and traditional financial assets began performing well again, people tended to abandon tangible asset en masse. This drove valuations well below fair value for the last 20 years. Although the coin index - and investment grade antiques in general - bounced off its year 2000 low of 0.57, tangibles are still substantially undervalued today. The latest 2015 data point for the coin index is 0.64, meaning a rally of more than 50% would be needed to return to our long term average valuation.
But how can we be sure our coin index data isn't a fluke, or perhaps just coin specific? Well, I also managed to get some historical data on the Antique Collector's Club Antique Furniture Price Index. This is an index started in 1968 by John Andrews that tracks the performance of 1400 types of commonly seen antique British furniture. It is divided into seven categories: oak, walnut, early mahogany, late mahogany, Regency, Victorian and country. Antique furniture is a radically different kind of market than coins. So if we find the same relationship to nominal GDP here that we found in the coin index we can be assured that it is meaningful. However, one caveat about antique furniture is that it is not, in my opinion, technically investment grade. This is due solely to the fact that furniture is not portable, a requirement for investability by my definition. But we work with the data we have, not the data we want.
The next chart shows the ACC Antique Furniture Price Index from 1968 to 2015 versus U.K. nominal GDP and the U.K. inflation rate (RPI). You can immediately see the tight relationship between the furniture index and U.K. nominal GDP from 1968 to 2002. At that point the relationship seems to break down. Looking at correlation coefficients confirms this analysis. From 1968 to the index's peak in 2002, the correlation coefficient is a near perfect 0.974, but if we look at it over the entire period, the correlation coefficient drops to 0.710. This is still a fairly good correlation, but not nearly as close as it had been until 2002.
What happened? Well, the antique furniture market has been undergoing a tumultuous period for the last 15 years or so. Specifically, living arrangements have been changing. Smaller houses, condos and apartments are now the norm. So a lot of massive antique furniture from bygone eras simply doesn't fit in the average home anymore. Also, many people are overly indebted. They struggle to make the monthly payments on their massive mortgages, car loans and credit cards. Paying top dollar for antique furniture isn't a top priority when you're barely making ends meet.
This leads us to our last graph. It shows the ratio of the ACC Antique Furniture Price Index to U.K. nominal GDP. And it shows a remarkably similar story to our coin index chart. A period of overvaluation in the 1970s through the early 1990s turns into extreme undervaluation for the most recent 10 years. As of 2015, the furniture index to GDP ratio rests at 0.40, indicating a 150% increase would be necessary to reach fair value.
Now I would take this indication of extreme undervaluation in the antique furniture market today with a grain of salt. While I do feel that antique furniture is undoubtedly selling too cheaply, lifestyles have changed in the last couple of decades - perhaps permanently. Fair value may be lower today than the (still evolving) long term data says it should be.
In any case, I think there are a few important points we can garner from this data. First, we can successfully measure the absolute value of art and antiques. Second, art and antiques as an asset class tend to appreciate at the same rate as nominal GDP growth over the long term. Third, inflation and the accompanying underperformance of paper assets lead to periods of overvaluation in antiques while booming stock markets and low inflation rates give birth to secular undervaluation. Fourth, all indicators show that tangible asset classes like art and antiques are somewhere between moderately undervalued and egregiously undervalued today. It is clear that there has rarely been a better time to become a connoisseur of fine art.
