Transcript Cantillon effects in the market for art
“Cantillon effects” in the market for art
Cameron M. Weber PhD Student, New School for Social Research Adjunct Professor, FIT/SUNY and St. John’s University
“Cantillon effects” in the market for art
i) Introduction to theory of Cantillon Effects ii) Our case study iii) The model iv) Results
“Cantillon effects” in the market for art
Given classical equation of exchange, MV = PQ Adding causality and non-neutrality: Endogenous money, PQ => MV (Hume, Wicksell, Marx) Exogenous money, MV => PQ (Keynes, Monetarist)
“Cantillon effects” in the market for art
Disaggregating “Q” into consumption goods and stages of production (time-variant intermediate goods) we get Austrian School use of deviation from ‘natural rates’ of interest and consequently malinvestment and prolonged business cycles
“Cantillon effects” in the market for art
Disaggregating and re-theorizing we get money supply effects on both output and assets Cantillon Effects describe increasing asset prices (asset bubbles) coinciding with an increasing exogenous (central bank) money supply Δ M => Δ Asset Prices
“Cantillon effects” in the market for art
An example of Cantillon Effects “But in November Naota Kan, who has since become Japan’s finance minister, made what he proudly call his ‘deflationary declaration’, urging the BOJ to redouble its efforts to combat falling prices. Days later the BOJ offered ¥10 trillion ($112 billion) of virtually interest-free liquidity to the banking system to fight deflation….As the yen came off its high, investors piled into shares. But since then the yen has rebounded with barely a squeak from the BOJ” (The Economist, January 30, 2010).
“Cantillon effects” in the market for art
Callahan and Garrison 2003 “We believe that events comprising the dot-com boom and bust can be illuminated by tracing the Cantillon effects as new money made its way from the Federal Reserve, through the banking system, and finally to the dot-com start-ups” (68).
“As a result, after increasing at a rate of less than 2.5 percent during the first three years of the Clinton administration, MZM (money zero maturity ) increased over the next three years (1996-1998) at an annualized rate of over 10%, rising during the last half of 1998 at a binge rate of almost 15 percent” (81).
Callahan, G. & Garrison R.W. (2003). Does Austrian business cycle theory help explain the dot-com boom and bust? Quarterly Journal of Austrian Economics, 6(2), 67-98.
“Cantillon effects” in the market for art
“Emerging signs of stronger economic activity and the Federal Open Market Committee (FOMC)’s second round of quantitative easing (QE2) have raised concern among some analysts that expansionary policy might be causing bubbles in financial and commodity markets— bubbles that might harm the economy if they burst. Prices for bonds, equities, and commodities have increased sharply since late August: The Reuters Jefferies/CRB weekly futures commodity price index increased by 22 percent (in U.S. dollars) through the week of November 9 (but fell sharply the following week), oil prices by 22 percent, the Economist food-price index by 20 percent, the Russell 2000 Index by 22 percent, and the broader S&P 500 Index by 15 percent (see charts). Given these increases, the concern over bubbles is reasonable, but it is difficult to distinguish beforehand the line between aggressive (“just right”) monetary policy and overly aggressive (“too hot”) monetary policy that generates bubbles. Rapid increases in commodity and financial market prices by themselves, however, are not reliable indicators of potential bubbles because such increases also occur as part of normal monetary policy.”
Anderson, R.G. (2011). Monetary Policy, Bubbles, and Goldilocks, Economic Synopsis, Federal Reserve Bank of St. Louis, January 13.
Our case study
The use of fine art might be an effective means to measure Cantillon Effects as art is removed from the capital structure of the economy, so we might be able to measure “pure” Cantillon Effects.
In other words, the “Q” value in the classical equation of exchange is missing all together for the causal chain, thus an increase in the money supply might be seen to directly effect the price of art.
Economic theory is that as money supply increases, the “time-preferences” of art investors decreases (art becomes cheaper relative to consumption goods) and/or inflationary expectations mean that art investors see price signals (“easy money”) encouraging investment in art.
“Cantillon effects” in the market for art
Art data based on 1,336 repeated sales from 1830 – 2007 on the London market. Data provided by Goetzmann et al 2010, data commonly used by cultural economists to measure returns to art versus other assets, updated by Goetzmann et al for error correction, contemporary art (1960s onward) and a more recent method for inflation indexing.
Goetzmann W., Renneboog L. & Spaenjers C. (2010). Art and money. NBER Working Paper 15502
“Cantillon effects” in the market for art
Money supply is “M1” based on methodology used by Bessler 1984 and Devadoss & Meyers 1987, in order to compare results of Cantillon Effects versus money supply effects on output prices.
Bessler, D.A. (1984). Relative prices and money: A vector autoregression on Brazilian data. American Journal of Agricultural Economics, 66 (1), 25-30.
Devadoss S. & Meyers W.H. (1987). Relative prices and money: Further results for the United States. American Journal of Agricultural Economics, 69 (4), 838 842.
“Cantillon effects” in the market for art
“Cantillon effects” in the market for art
We find two distinct periods of monetary history, leading to the need for data bifurcation between the pre-war Classical Gold Standard (1830 – 1913) and the post-war central-banking era (1946 - 2007).
“Cantillon effects” in the market for art
“Cantillon effects” in the market for art
H 0 : A change in money supply does not “Granger Cause” a change in the art index in the same in the direction H 1 : A change in money supply “Granger Causes” a change in the art index in the same in the direction
“Cantillon effects” in the market for art
Table 1
Descriptive Statistics Mean 1830-1913 (T = 81) ∆LnArt ∆LnM1 ∆LnFTSE 0.0420 0.0041 0.0042 1946-2007 (T = 58) ∆LnArt 0.0497 ∆LnM1 ∆LnFTSE 0.0592 0.0692 Min. -0.1600 -0.1100 -0.2528 Max. 0.2600 0.1600 0.1990 -0.2012 -0.0600 -0.8100 0.3247 0.1900 0.8600 Std. Dev. 0.0800 0.0456 0.0847 0.1005 0.0394 0.2180
“Cantillon effects” in the market for art
Long-term Relationships
Tables 2
Correlations Art M1 M1_1 M1_2 M1_3 FTSE FTSE_1 FTSE_2 FTSE_3 1830-1913 Art 1 M1 M1_1 M1_2 M1_3 FTSE FTSE_1 FTSE_2 FTSE_3 1946-2007 Art 1 M1 M1_1 M1_2 M1_3 FTSE FTSE_1 FTSE_2 0.876 1 0.952 1 0.783 0.845 1 0.953 0.999 1 0.734 0.7690 0.882 1 0.954 0.9990 0.999 1 0.707 0.7370 0.815 0.904 1 0.954 0.9980 0.999 0.999 1 0.819 0.812 0.672 0.615 0.577 1 0.954 0.952 0.952 0.953 1 0.803 0.787 0.833 0.724 0.673 0.897 1 0.952 0.95 0.947 0.947 0.984 1 0.786 0.781 0.804 0.849 0.759 0.812 0.909 1 0.949 0.946 0.944 0.941 0.959 0.983 1 0.769 0.783 0.791 0.821 0.864 0.738 0.832 0.918 1 0.948 0.945 0.942 0.939 0.931 0.955 0.982 FTSE_3 1 Notes: Table represents pair-wise correlations of each horizontal and vertical variable. Each correlation is significant at the 1% level. Art is the art index, M1 is the money supply in (billions of) pound sterling, and FTSE is the equities index. Both the art and the equity indexes were re-normalized for the 1946 2007 period. M1_1, for example, represents the money supply lagged one year.
“Cantillon effects” in the market for art
Short-term Relationships
Table 3
Comovement analysis 1 ∆Art 2 ∆Art 3 ∆Art 4 ∆Art 5 ∆Art 6 ∆Art 1830-1913 ∆M1 ∆M1_1 ∆FTSE ∆FTSE_1 0.025 (1.514) (1.512) (1.512) (1.520) 0.22 (1.520) 0.027 (1.524) -0.050 1946-2007 ∆M1 0.004 *** (1.135) (1.073) (1.051) ∆M1_1 -0.001 (1.235) ∆FTSE (1.030) ∆FTSE_1 (0.996) Notes: Comovement regressions use OLS and Newey-West (HAC) standard errors. ***, **, and * indicate significance of the regression coefficient at the 1%, 5% and 10% level, respectively. Following Goetzmann et al (2010) we conduct a comovement analysis to determine the “short-term”, year-to-year, relationships between the variables, to supplement the “long-term” correlations. We use first differences (∆) in each variable to capture the year to-year differences in the short-term analysis, although do not extend the analysis to lagged first differences beyond a single period following McCallum (2010) who recommends a one-lag model in case of “serious” autoregression.
“Cantillon effects” in the market for art
Regression Analysis Following McCallum 2010 we use a least squares regression, with independent variables of a one period lag for our general and robustness check models, and correct for heteroskedasticity to cleanse the error terms of autocorrelation which as noted in Table 2 (“Correlations”) we highly-suspect.
McCallum, B.T. (2010). Is the Spurious Regression Problem Spurious? NBER Working Paper 15690.
“Cantillon effects” in the market for art
Table 4
Regression analysis of Art as dependent variable Constant Art t-1 M1 t-1 FTSE t-1 1830-1913 General Model T = 84 Adj. R 2 = 0.987
d-w = 1.537 Robustness Check T = 84 Adj. R 2 = 0.987
durbin's h = 1.400 1946-2007 General Model T = 62 Adj. R 2 = 0.996
durbin's h = 3.980 0.199 2.235 (**) (***) (***) 0.943 (***) 0.001 Robustness Check 1.803 0.978 0.001 -0.001 T = 62 Adj. R 2 = 0.996
(***) durbin's h = 3.787 Notes: Least squares regression with heteroskedastic-corrected error terms. ***, **, and * indicate significance of the regression coefficient at the 1%, 5% and 10% level, respectively.
We are not able to reject the null hypothesis that there is no Granger causality between money supply changes and art price changes.
“Cantillon effects” in the market for art
Bessler 1984 and Devadoss & Meyers 1987 use a VAR log likelihood model to estimate the most likely lagtimes between a monetary change and effects on output prices. They find (using monthly data) that the most likely lagtime is 13 and 14 months with a rapidly dissipation of the price increase immediately following the most likely effect.
“Cantillon effects” in the market for art
Using the same methodology we find that the most likely effect of a money supply change on art prices is also between the 13 th and 23 rd month (e.g., during the 2 nd year using yearly data).
However, unlike the earlier work on output prices, we find that the effect on art prices does not decay rapidly after the monetary increase, showing perhaps that money changes are cumulative (and/or have momentum) and could lead to an ‘asset bubble’.
“Cantillon effects” in the market for art
Non-Decaying money effect on asset values
Table 6
VAR System Log Likelihood Ratio Test for Lagtimes Lags Loglik 1830-1913 1 2 3 4 5 6 83.18703 85.68235 85.7666 86.76375 86.81364 86.91812 7 8 89.10214 89.37293 1946-2007 1 2 3 4 5 53.92625 65.95571 66.25091 66.58107 66.71588 6 7 8 66.7168 67.86268 67.93534 AIC -2.110185 -2.149536* -2.125437 -2.125362 -2.100359 -2.076793 -2.107951 -2.088761 -2.294656* -2.268552 -2.243743 -2.211699 -2.174696 -2.180099 -2.145753 BIC -2.018182 -2.026865* -1.972099 -1.941356 -1.885686 -1.831452 -1.831943 -1.782086 -2.147324* -2.084387 -2.022745 -1.953868 -1.880032 -1.848602 -1.777423 length for the effect of LnM1 as exogenous variable on LnArt as endogenous variable is the second period (e.g., between the first and second year).
“Cantillon effects” in the market for art
Decaying money effect on output prices
Bessler, D.A. (1984). Relative prices and money: A vector autoregression on Brazilian data. American Journal of Agricultural Economics, 66 (1), 25-30.
“Cantillon effects” in the market for art
Thank you