*During the 2000s, investing in wine became very. *We observed an increase in the number of investment

Transcription

1 During the 2000s, investing in wine became very fashionable Low correlations with traditional assets Vintages 2009 and 2010 are highly rated Effect of Asian demand Turnover from sales of wine by major auction houses millions US$ millions US$ We observed an increase in the number of investment funds specialized in wine and also in the number of price indices as the Livex (London) The returns of wine can be exceptional: Sokolin (1998), Burton and Jacobsen (2001), Masset et al. (2010), Masset and Weisskopf (2010) However, studies show uneven estimations of returns whatever the type of wine: US Treasury Bonds Stock exchange indices Outperformance of wine Jaeger (1981), Burton and Jacobsen (2001), Jones and Storchmann (2001), Hadj Ali and Nauges (2003), Haeger and Storchmann (2006), Sanning et al. (2007, 2008), Devine and Lucey (2011). Duthy (1986), Byron and Ashenfelter (1995), Masset et al (2010), Masset and Weisskopf (2010, 2011). Underperformance of wine Krasker (1979), Ashenfelter et al. (1995), Wood and Anderson (2006), Devine and Lucey (2011). Hadj Ali and Nauges (2003), Masset and Henderson (2010). In this paper, we are looking for to test the significativity and the amplitude of impact that can have the exchange conditions on the formation of auction prices of wine. Inconsistency of information given to the market (high mortality of funds specialized in wine). The essential source of information bias is it in the structure of data used for calculation of indices?

2 From auction data of five first growth of Bordeaux, we evaluate the significativity of impacts on the average prices that can have different trading conditions: Trend of market (years of sales) and saisonality (month), Demand effect (number of sold bottles during an auction) Convention effect on the quality of exchanged product (evaluated by Parker note), Effect of reputation of product (château and vintages), General macroeconomic effect linked to the economic conjoncture (zones) or more localized (countries) Effect of financial markets (places of exchange) Transactional effect (auction companies and auction houses) if we consider that some companies use process of organization valorising better the products. About 80,000 transactions on the 5 first growths used in the Livex 50 (Lafite Rothschild, Mouton Rothschild, Margaux, HautBrion, Latour) 662,652 bottles of 750 ml are exchanged Total amount of 457 USD millions Vintages and sales auction companies: Acker Merrall, Bonhams, Christie's, Edward Roberts International, Hart Davis, Morrell and Company, Sotheby's, Zachy's. These companies operate in: 3 continents: Asia, North America and Europe 6 countries: China, France, Switzerland, UK, USA, Netherlands 10 places: Geneva, HongKong, Paris, Amsterdam, New York, Chicago, Los Angeles, Bordeaux, San Francisco, London VARIABLES MODALITIES In US$ (on exchange rate at the moment of the transaction) Unit price of a bottle of 750ml (P) non deflated in ln Number of sold bottles in each Expressed in ln transaction (Q) Evaluation of expert (E) Notes Robert Parker Haut Brion, Laffite Rothschild, Latour, Margaux, Mouton Châteaux (CH) Rothschild Vintages (V) Years (Y) Months (MONTH) On 12 months Zones (Z) Asia, European Union, USA Countries (C) China, USA, France, Netherlands, UK, Switzerland Amsterdam, Bordeaux, Chicago, Geneva, HongKong, Financial places of trading (PL) London, Los Angeles, Paris, New York, San Francisco Acker Merrall & Condit (New York), Bonhams (HongKong, Londres), Bonhams and Butterfields (New York), Christie's (Amsterdam, Geneva, HongKong, London, Los Angeles, New York, Paris, South Kensington), Edward Roberts Auction houses (AH) International (Chicago), Hart Davis Hart Wine Co. (Chicago), Morrell and Company (New York), Fine Wine Auctions (New York), Sotheby's (HongKong, London, New York), Zachy's Wine Auctions (New York), Zachy's/Wally's (Los Angeles), ZachysChristies (New York) Acker Merrall, Bonhams, Christie's, Edward Roberts Auction companies (AC) International, Hart Davis; Morrell and Company; Sotheby's, Zachy's

3 The spotting of zones of vintages or liquidity (according to auction volume) is realized by an algorithm of univaried partitioning of Fisher (for 3 classes). Period 1 ( ) Period 2 ( ) Classes Size Classes [28;502[ [502;1154[ [1154;1785[ [18;1477[ [1477;3402[ [3402;7077 [ Intraclasse 22695, , , , , variance 3 Objectsby classe Others in P Others in P The model includes all the variables identified in the database. The existence of multicolinearity implies to elimination of variables «countries» and «zones» for to retain only the variable «places», more relevant for integrate the characteristics of local financial markets. The model retained is on the following form: ln!" # " "! $% % & % The variables being categorial and quantitative (expressed in ln), the parameters are estimated by a ANCOVA with a general linear model which permits to test the impact of categorial variables on the price independantly of effects of the covariables. The estimation is realized on 4 different samples (all data, then by zone of liquidity). The estimation of normalised parameters will permit to compare the impact on the price of different conditions of sale. High significativity and explicativity of the models and mainly a contribution always significant of the variables. Global sample L1 L2 L3 (R²aj=0,781 ; PrF0,0001) (R²aj=0,858 ; PrF0,0001) (R²aj= 0,785; PrF0,0001) (R²aj=0,745 ; rf0,0001) Variables DDL F Pr > F DDL F Pr > F DDL F Pr > F DDL F Pr > F lnq 1 107,65 0, ,30 0, ,46 0, ,87 0,015 lne 1 510,03 0, ,24 0, ,54 0, ,36 0,0001 PL 6 345,81 0, ,22 0, ,63 0, ,32 0,0001 AC 6 264,70 0, ,57 0, ,79 0, ,217 0,0001 CH 4 V ,1 0 0, ,4 8 0, ,4 9 0, ,97 0, , ,4 6 0, ,52 0,0001 0, ,16 0,0001 Y ,2 9 0, ,1 5 0, ,0 9 0, ,64 0,0001 MONTH 11 63,43 0, ,85 0, ,05 0, ,51 0,0001

4 For answer to our problematic, we must evaluate for each sample the maximal impact that can have on the price the entry of the modality the most valuing in comparison to the entry of the modality the most penalizing. We want to show the risks of maximal distorsion (noted Diff) induced by a modification of transaction conditions which the price will be introduced later in an indice. Evaluation of the maximal distorsion of the transaction modalities ontheaveragelevelof pricesbysample. Whole sample Liquidity zone 1 Liquidity zone 2 Liquidity zone 3 Quantity 0,029 0,053 0,045 0,009 Evaluation 0,055 0,018 0,028 0,263 Min Max Diff Min Max Diff Min Max Diff Min Max Diff Vintages 0,477 0,049 0,526 0,453 0,078 0,531 0,229 0,42 0,649 0,263 0,204 0,467 Years 0,289 0,11 0,399 0,3 0,122 0,422 0,331 0,117 0,448 0,301 0,126 0,427 Châteaux 0 0,25 0,25 0,027 0,237 0, ,264 0, ,458 0,458 Places 0,021 0,07 0,091 0,062 0,036 0,098 0,017 0,073 0,09 0,013 0,09 0,103 Companies 0,043 0,03 0,073 0,045 0,041 0,086 0,047 0,028 0,075 0,041 0,029 0,07 Month 0,021 0,023 0,044 0,02 0,033 0,053 0,015 0,031 0,046 0,018 0,023 0,041 The order and the amplitude of the distorsion among the categorial variables are the same for the whole sample and the zones of liquidity L1 and L2. The maximal distorsion can result from the vintage, then of the year of sale, of the château, of the place of sale, of the auction companies and finaly of the month (except for the zone of liquidity 3 where the impact of the château is more important than that of the year). Except for the variable «month», the risk of distorsion is more important than the impact of the expert evaluation. This article shows that the system of composition of reference baskets of wine price indices, by neutralizing the conditions of price formation and the rotation of products, generates probably a bias in the appreciation of the asset returns on the wine market. This conclusion also lead us to recommend an expansion of the variables included in the hedonic functions and selection criteria in the repeat sales methods.

5 Annexes 1 : résultats des estimations de l ANCOVA Tableau 5a. Estimation des paramètres bruts des variables continues pour chaque échantillon Echantillon global L1 L2 L3 Source Valeur Pr > t Valeur Pr > t Valeur Pr > t Valeur Pr > t Constante 6,932 0,0001 7,144 0,0001 6,018 0,051 6,501 0,184 lnq 0,026 0,0001 0,047 0,0001 0,038 0,005 0,008 0,003 lnev 0,165 0,0001 0,028 0,002 0,078 0,009 2,860 0,040 Annexes 2 : résultats des estimations de l ANCOVA (suite) Tableau 5b. Estimation des paramètres normalisés par échantillon de l ensemble des variables lnq Echantillon global L1 L2 L3 0,029 0,0001 0,053 0,0001 0,045 0,0001 0,009 0,015 lnev 0,055 0,0001 0,018 0,002 0,028 0,0001 0,263 0,0001 Genève 0,001 0,693 0,018 0,006 0,000 0,936 0,004 0,267 Hong Kong 0,076 0,0001 0,036 0,0001 0,073 0,0001 0,090 0,0001 Paris 0,009 0,000 0,024 0,0001 0,007 0,148 0,009 0,007 Amsterdam 0,014 0,0001 0,042 0,0001 0,012 0,024 0,010 0,007 Londres 0,021 0,000 0,057 0,0001 0,017 0,110 0,013 0,107 New York 0,017 0,007 0,062 0,0001 0,017 0,163 0,010 0,284 Chicago 0,005 0,232 0,024 0,001 0,003 0,726 0,006 0,350 Los Angeles 0,000 0,000 0,000 0,000 Acker Merrall 0,031 0,0001 0,041 0,0001 0,028 0,0001 0,029 0,0001 Bonhams 0,043 0,0001 0,045 0,0001 0,047 0,0001 0,041 0,0001 Christie's 0,023 0,0001 0,007 0,302 0,030 0,0001 0,022 0,0001 Annexes 3 : résultats des estimations de l ANCOVA (suite) Tableau 5b. Suite Haut Brion 0,008 0,0001 0,066 0,0001 0,004 0,331 0,040 0,0001 Lafite Rothschild 0,254 0,0001 0,237 0,0001 0,264 0,0001 0,458 0,0001 Latour 0,114 0,0001 0,069 0,0001 0,107 0,0001 0,154 0,0001 Margaux 0,041 0,0001 0,027 0,0001 0,075 0,0001 0,065 0,0001 Mouton Rothschild 0,000 0,000 0,000 0,000 AN ,052 0,0001 0,152 0,0001 0,015 0,0001 AN ,092 0,0001 0,232 0,0001 0,073 0,0001 0,014 0,0001 AN ,104 0,0001 0,178 0,0001 0,157 0,0001 0,047 0,0001 AN ,232 0,0001 0,258 0,0001 0,267 0,0001 0,238 0,0001 AN ,289 0,0001 0,300 0,0001 0,331 0,0001 0,301 0,0001 AN ,251 0,0001 0,279 0,0001 0,281 0,0001 0,269 0,0001 AN ,174 0,0001 0,236 0,0001 0,215 0,0001 0,161 0,0001 AN ,067 0,0001 0,149 0,0001 0,105 0,0001 0,035 0,0001 AN ,075 0,0001 0,150 0,0001 0,115 0,0001 0,042 0,0001 AN ,087 0,0001 0,123 0,0001 0,105 0,0001 0,078 0,0001 AN ,052 0,0001 0,031 0,0001 0,039 0,0001 0,072 0,0001 AN ,115 0,0001 0,122 0,0001 0,117 0,0001 0,126 0,0001 AN ,000 0,000 0,000 0,000

6 Annexes 3 : résultats des estimations de l ANCOVA (fin) MOIS 1 0,021 0,0001 0,020 0,0001 0,015 0,000 0,018 0,0001 MOIS 2 0,006 0,004 0,005 0,177 0,006 0,141 0,011 0,000 MOIS 3 0,003 0,147 0,022 0,0001 0,005 0,230 0,001 0,752 MOIS 4 0,004 0,067 0,019 0,0001 0,002 0,563 0,009 0,006 MOIS 5 0,013 0,0001 0,029 0,0001 0,015 0,001 0,008 0,029 MOIS 6 0,011 0,0001 0,002 0,692 0,007 0,109 0,017 0,0001 MOIS 7 0,002 0,271 0,001 0,688 0,000 0,917 0,000 0,965 MOIS 8 0,001 0,419 0,002 0,510 0,002 0,450 0,002 0,499 MOIS 9 0,022 0,0001 0,029 0,0001 0,030 0,0001 0,020 0,0001 MOIS 10 0,023 0,0001 0,014 0,002 0,031 0,0001 0,023 0,0001 MOIS 11 0,013 0,0001 0,033 0,0001 0,020 0,0001 0,008 0,036 MOIS 12 0,000 0,000 0,000 0,000 MILLÉSIME 1946 MILLÉSIME 1947 MILLÉSIME 1948 MILLÉSIME 1949 MILLÉSIME 1950 MILLÉSIME 1951 MILLÉSIME 1952 L1 Ensembl e L2 Ensembl e L3 Ensemble 0,013 0,014 MILLÉSIME1945 0,420 0,049 MILLÉSIME1959 0,182 0,028 0,037 0,035 MILLÉSIME1955 0,079 0,092 MILLÉSIME1961 0,176 0,039 0,005 0,037 MILLÉSIME1966 0,093 0,220 MILLÉSIME1970 0,220 0,344 0,078 0,027 MILLÉSIME1975 0,205 0,292 MILLÉSIME1982 0,204 0,184 0,040 0,049 MILLÉSIME1978 0,189 0,267 MILLÉSIME1986 0,034 0,336 0,036 0,036 MILLÉSIME1979 0,178 0,239 MILLÉSIME1989 0,093 0,329 0,097 0,089 MILLÉSIME1981 0,168 0,239 MILLÉSIME1990 0,081 0,338