Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=ucha20

CHANCE

ISSN: 0933-2480 (Print) 1867-2280 (Online) Journal homepage: http://www.tandfonline.com/loi/ucha20

Bordeaux Wine Vintage Quality and the Weather

Orley Ashenfelter , David Ashmore & Robert Lalonde

To cite this article: Orley Ashenfelter , David Ashmore & Robert Lalonde (1995) Bordeaux WineVintage Quality and the Weather, CHANCE, 8:4, 7-14, DOI: 10.1080/09332480.1995.10542468

To link to this article: https://doi.org/10.1080/09332480.1995.10542468

Published online: 20 Sep 2012.

Submit your article to this journal

Article views: 97

Citing articles: 61 View citing articles

Statistical prediction of wine prices based on vintagegrowing-season characteristics produces consternationamong wine "experts."

Bordeaux Wine VintageQuality and the Weather

Orley Ashenfelter, David Ashmore, and Robert Lalonde

"Nor let thy vineyard bend toward the sunwhen setting."

-Virgil

In this article we show that thequality of the vintage for red Bor­deaux wines, as judged by theprices of mature wines, can bepredicted by the weather duringthe growing season that producedthe wines. Red Bordeaux wineshave been produced in the sameplace and in much the same wayfor hundreds of years. Whenyoung, the wines from the bestvineyards are astringent, andmany people find them unpleas­ant to consume. As these winesage, they lose their astringency,and many people find them verypleasant to consume. Because Bor­deaux wines taste better when

© Hl9S Spr-ingur-Vurlug New York, Inc.

they are older, there is an incen­tive to store the young wines untilthey are mature. As a result, thereis an active market in young wines(similar to "new issues" in the se­curities markets) and an activemarket in older wines (similar tothe secondary markets in securi­ties).

Surprisingly, the weather in­formation that is so useful in pre­dicting the prices of the maturewines plays little or no role insetting the prices of the youngwines. We show that youngwines are usually overpricedrelative to what we would predictbased on the weather and theprice of the old wines. As theyoung wines age, however, theirprices usually converge to ourpredictions. This implies that

"bad" vintages are overpricedwhen they are young and that"good" vintages may sometimesbe underpriced when they areyoung. Rational buyers shouldavoid bad vintages when they areyoung, but they may sometimeswish to purchase good vintages.

Although the evidence suggeststhat the market for older wines isrelatively efficient, it implies thatthe market for younger wines isvery inefficient. Why don't thepurchasers of young wines waitand buy them when they are ma­ture? Why do purchasers ignorethe weather that produced the vin­tage in making their decisions? Al­though there are no simple an­swers to these questions, wediscuss one possible explanationin the final section.

CHANCE 7

Vineyards

It has long been known thai the bestvineyards in Europe are usually lo­cated near a large body of water (todelay spring frosts, prolong fall rip­ening, and to reduce diurnal tem­perature fluctuations), on a slopewith a southern exposure to the sun(to enhance ripening), and in soilwith good drainage (to overcomethedilution of the fruit that accompaniesfall rains). The aspects of the

Vineyards and Vintages

The best wines of Bordeaux aremade from grapes grown on spe­cific plots of land and the wine isnamed after the property (or cha­teau) where the grapes are grown.In fact, knowledge of the chateauand vintage provides most of theinformation about the quality ofthe wine; that is, if we imagine 10vintages and 6 chateaux, there are.in principle. 60 different wines ofdifferent quality. Knowing the

weather during the growing seasonfor which a good vineyard site com­pensates are precisely those that,when absent, produce the growingseasons that make good vintages.An interesting research projectwould be to determine how much ofthe differences in the prices fetchedby the various Bordeaux chateauxmay be attributed to vineyard sitecharacteristics.

reputations of the 6 chateaux andthe 10 vintages, however, is suffi­cient to determine the quality ofall 60 wines; that is. good vintagesproduce good wines in all vine­yards and the best wines in eachvintage are usually produced bythe best vineyards (see "Vine­yards").

Although this point is some­times denied by those who pro­duce the wines, and especially bythose who sell the young wines. itis easy to establish its truth by ref-

erence to the prices of the maturewines. To demonstrate the point,Table 1 indicates the 1990-1991market price in London of 6 Bor­deaux chateaux from the 10 vin­tages from 1960 to 1969. Thesechateaux were selected becausethey are large producers and theirwines appear frequently in thesecondary (auction) markets. (Ablank in the table indicates thatthe wine has not appeared in themarket during the time period.Lower-quality vintages are typi­cally the first to leave the market.)The vintages from 1960 to 1969are selected because, by the 1990s,these wines are fully mature andthere is little remaining uncer­tainty about their quality.

From Table 1 it is obvious thatknowledge of the row means andthe column means is sufficient topredict most of the prices in thetable. (The explained variancefrom a regression of the logarithmof the price on chateau and vin­tage dummies is over 90%.) Aranking of the chateaux in order of

Table 1-London Auction Prices for Selected Mature Red Bordeaux Wines,1990-1991 (per dozen bottles in SUS)

Chateaux (Vineyards)

Vintage Lafite Latour Cheval Blanc Cos Montrose Pichon Averaged'Estournel Lalande

1960 494 464 486 4791961* 4335 5432 3534 1170 1125 1579 48841962* 889 1064 821 521 456 281 9n1963 340 471 251 4061964* 649 1114 1125 315 350 410 8821965 190 424 258 3071966* 1274 1537 1260 546 482 734 14061967* 374 530 441 213 236 243 4521968 223 365 274 2941969 251 319 123 84 152 285Average 1504 1935 1436 553 530 649

NOTE:Columnaveragesare calculatedonly for vintages indicatedby an asterisk,where a marketprice exists for everychateau.Rowaveragesare calculatedusingonly ChateauLafiteand ChateauLatour,wherepricesare availablefor all vintages. Sterlingpricesare convertedto dollarsat the exchangerate on November30, 1992.

SOURCE: Averageprices from LiquidAssets: The InternationalGuide to Fine WinesIssue8 (October1991)and Issue9 (Decem-ber 1992).

8 VOL. 8. NO.4. 1995

Q)u....t­o......oC

r-1

4.60517

o

o

o

o

o

o o

2.30259o o

o

oo o

1950 1955 1960 1965 1970year of vintage

1975 1980

Figure 1. Red bordeaux wines, price relative to 1961 vintage year.

quality based on their priceswould be Latour, Lafite, ChevalBlanc, Pichon-Lalande, Cosd'Estournel, and Montrose. In fact,as Edmund Penning-Rowsellpointed out in his classic bookThe Wines ofBordeaux (1985), thefamous 1855 classification of thechateaux of Bordeaux into qualitygrades was based on a similar as­sessment by price alone. Surpris­ingly, the 1855 classificationranks these chateaux in only aslightly different order-LafHe,Latour, Pichon-Lalande, Cosd'Estournel, and Montrose.(Cheval Blanc was not ranked in1855.) Likewise, a ranking of thequality of the vintages based onprice alone would be 1961, 1966,1962,1964, and 1967. The remain­ing vintages (1960, 1963, 1965,1968, and 1969) would be rankedinferior to these five, but, perhapsbecause of this fact, many of thewines from these inferior vintagesare no longer sold in the secon­dary market.

Real Rate of Return toHolding Bordeaux Wine

It is natural to ask why the pricesof mature wines from a single cha­teau, made in the same way fromgrapes grown in the same place bythe same winemaker, would differso dramatically from vintage tovintage, as is indicated by Table 1.There are two obvious explana­tions for this vintage variability.First, the older wines have beenheld longer, and so they must beara normal rate of return. This factalone would make the older winesmore expensive than the youngerones. Second, the quality of thewines of different vintages mayvary because the quality of thegrapes used to make the winesvaries.

Figure 1 provides a test of the hy­pothesis that the price of the winesvaries because of their age. In thisfigure (and throughout the remain­der of the article), we use as a meas­ure of the price of a vintage an in-

dex based on the wines of severalchateaux. The chateaux are delib­erately selected to represent themost expensive wines (Lafite, La­tour, Margaux, and Cheval Blanc)as well as a selection of wines thatare less expensive (Ducru Beaucail­lou, Leoville Las Cases, Palmer,Pichon Lalande, Beychevelle, Cosd'Estournel, Giscours, Gruaud­Larose, and Lynch-Bages). We con­struct the index of vintage pricefrom a regression of the logarithmof the price from several thousandauction sales on dummy variablesindicating the chateau and the vin­tages. (The precise composition ofthe sample has very little effect onthe results.) The regression coeffi­cients for the vintage dummies arethen used to construct the vintageindex. This provides a simple wayto construct a vintage index in thepresence of an unbalanced sampledesign. (We compute the antiloga­rithm of these coefficients and thenexpress the price relative to the in­dex price for 1961. This is merely

CHANCE 9

California's Climate for Growing Grapes

The desirability of California's cli­mate for grape growing and thereasons for it were well known inthe 19th century . In his report tothe California legislature , A.Haraszthy wrote (p. 143) in 1862that "The Californ ia climate is emi­nently adapted for the culture ofgrape-vines. The oldest inhabi­tants [of California] have no recol­lection of a failure in the crops ofgrapes. The production is tabu-

a convenient normalization and af­fects only the intercept in the re­gressions reported later.)

Figure 1 is a scatter diagram ofthe logarithm of the price of thewines of a vintage against the vin­tage year. (The vintages of 1954 and1956 are not plotted because thesewines are now rarely sold. Thesetwo vintages are generally consid­ered to be the poorest in their dec­ade .) It is apparent from the dia-

lous; and there is no doubt in mymind that before long there will belocalities discovered which will fur­nish as noble wines as Hungary ,Spain, France , or Germany everhave produced . In California, siteis not so material as in Europeancountries , especially where , duringthe summer season , a good dealof rain falls . Californ ia, having aneven temperature , is warm andwithout rains in summer ."

gram that there is a negatively in­clined relationship. The slope ofthe regression line through thesepoints is -0.035. This is an estimateof (the negative of) what econo­mists sometimes call the real prod­uct rate of return to holding Bor­deaux wines. A "real product rateof return" is a number such that itsreciprocal indicates how many bot­tles of wine one would have tokeep in the cellar to be able to con-

o

sume one bottle per year in perpe­tuity. These data indicate that itwould be necessary to have about28 bottles in a perpetual cellar thatwas intended to support the con­sumption of one bottle per year.With a cellar of this size, the pro­ceeds from the sale of the older vin­tages would be just sufficient to re­stock the cellar and provide theconsumption of one bottle. Be­cause it is denominated in bottlesof wine rather than dollars, thismeasure does not tell us what thereturn to holding wine denomi­nated in generalized purchasingpower (money) is.

We have analyzed the relation­ship between the (log) price ofBordeaux wine and its age formany individual chateaux. Solong as the sample includes atleast 20 vintages , we invariablyobtain a negative slope to this re­lationship of between - .02 and ­.04. It is notable that the study ofthe various vintages of wine pro­vides so reliable and simple a

. 6 1

. 78 . 6

080 . 53077 . 6 6 . 70

. 6 4

057 . 7 167 07314 .5

79 •0 5 6 55

072 0 6 3 . 52. 7 5

0 54 0 58 . 5 974

069

065

068 60

Harv 51

350 Rain

• Above Averag Price 0 Below Average Price I

18. 5

Summer

Temperature

Figure 2. Bordeaux: 1952-1980, summer temperature and harvest rain related to price.

10 VOL. 8. NO. 4. 1995

Table 2-Regresslons of the (Logarithm of) Priceof'Different Vintages of a Portfolio of Bordeaux

Chateau Wines on Weather Variables

NOTES: All regressions use as data the vintagesof 1952-1980, excluding the 1954and 1956vintages, whichare now rarelysold; all regressions containan intercept,whichis not reported; the data (anda readmefile) are also available by anonymousftp in the publwinedirectoryof irs.princeton.edu.

Standard errorsare in parentheses.

Vintages and the Weather

grapes planted there (see "Califor­nia's Climate for GrowingGrapes"). In Bordeaux. this some­times happens-but usually itdoes not. Great vintages for Bor­deaux wines correspond to theyears in which August and Sep­tember are dry, the growing sea­son is warm, and the previouswinter has been wet.

Figure 2 establishes that it is hot,dry summers that produce the vin­tages in which the mature winesobtain the higher prices. This fig­ure displays for each vintage thesummer temperature from low tohigh as you move from left to right.and the harvest rain from low tohigh as you move from top to bot­tom. Vintages that sell for an above­average price are displayed withblack triangles, and vintages thatsell for a below-average price aredisplayed in open squares. If theweather is the key determinant ofwine quality. then the dark pointsshould be in the northeast quadrant

of the diagram and the light pointsshould be in the southwest quad­rant of the diagram. It is apparentthat this is precisely the case. Evenanomalies. like the 1973 vintage,tend to corroborate the fact that theweather determines the quality ofthe wines. The wines of this vin­tage, which are of somewhat above­average quality, have always soldat relatively low prices; insidersknow that they are often bargains.

Table 2 contains a regression ofthe (log) price of the vintage on theage of the vintage and the weathervariables indicated. In practice,the weather variables are almostuncorrelated with each other andwith the age of the vintage. As aresult, the regression equation isremarkably robust to the additionof other variables. The second col­umn of the table contains the basic"Bordeaux equation," and thethird column shows the effect onthe regression of adding the tem­perature in September as an addi­tional variable. It is obvious thatthis variable does not have a sta­tistically significant coefficient,and, indeed, in further experimen­tation we have not found anyother statistically significant vari­ables to add to the regression.

It is possible, of course, to pre­dict the relative price at which thenew vintage should be sold as soonas the growing season is complete.In fact, we have been doing this forseveral years and publishing the re­sults in the newsletter Liquid As­sets: The International Guide toFine Wines. The basic idea forthese predictions is displayed inFig. 3. Here we have added to Fig.2 the data for the vintages from1981 to 1992. Two things are im­mediately apparent from the figure.First, all but one of these recentvintages (1986) was produced by agrowing season that was warmerthan what is historically "normal."It is no accident that many Europe­ans believe global warming may al­ready be here! This unusual run ofextraordinary weather has almostcertainly resulted in a huge quan-

0.00765(0.0565)

0.8280.293

0.0240(.00747)

0.608(0.116)

-0.00380(0.000950)

0.00115(0.000505)

0.8280.287

0.0238(0.00717)

0.616(0.0952)

-0.00386(0.00081)0.001173

(0.000482)

0.2120.575

0.0354(0.0137)

Age of vintage

Average temperatureover growing season(April-September)

Rain in September andAugust

Rain in the monthspreceeding thevintage(October-March)

Average temperaturein September

FPRoot mean squared

error

Independent variables

It is well known that the quality ofany fruit, in general, depends onthe weather during the growingseason that produced the fruit.What is not so widely understoodis that in some localities theweather will vary dramaticallyfrom one year to the next. In Cali­fornia, for example, it never rainsin the summer, and it is alwayswarm in the summer. There is asimple reason for this. In Califor­nia, a high-pressure weather sys­tem settles each summer over theCalifornia coast and produces awarm, dry growing season for the

measure of the real rate of return.As we shall see, most of the re­maining variation in the price ofthe wine of different vintages isdue to variation from vintage tovintage in the weather that pro­duced the grapes.

CHANCE 11

o

. 8 5 061078 06

. 8 8

080066 gn •• 90

077064 89

81057 071

14 .5 8 6 7 . 8 478 73 . 8 3

19.0

057 79 55

072 063 052 . 8 28 075 Summer

0547.P I 58 059 . 9 1 Temperature

065

069

068 I 60

Harvest

350 Rain

o 1952 .1980 • 1981 ·1991

076

Figure 3. Bordeaux: 1952-1991, summer temperature and harvest rain related to price.

tity of excellent but immature redBordeaux wine. Second, theweather that created the vintages of1989 and 1990 appears to be quiteexceptional by any standard. Is itappropriate to predict that thewines of these vintages will be ofoutstanding quality when the tem­perature that produced them is sofar outside the normal range? Be­fore making the prediction for1989, we did , in fact, turn to Lin­coln Moses (ofStanford University)for informal advice. Moses sug­gested two informal tests . (1)Would the last major "out-of-sam­pie " prediction have been correct?The idea here is to use the past toindirectly test the ability of the re­lationship to stretch beyond theavailable data. In fact, the last ma­jor out-of-sample prediction forwhich all uncertainty has been re­solved is the vintage of 1961,which had the lowest August-Sep­tember rainfall in Bordeaux his­tory. Just as the unusual weatherpredicted, the market (see Table 1),

12 VOL. 8, NO.4, 1995

and most wine lovers, have cometo consider this an outstanding vin­tage. (2) Is the warmth of the 1989and 1990 growing seasons in Bor­deaux greater than the normalwarmth in other places where simi­lar grapes are grown? The idea hereis to determine whether the tem­perature in Bordeaux is abnormalby comparison with grape-growingregions that may be even warmer.In fact, the temperature in 1989 or1990 in Bordeaux was no higherthan the average temperature in theBarossa Valley of South Australiaor the Napa Valley in California,places where high-quality redwines are made from similar grapetypes .

Based on these two informaltests, we are convinced that boththe 1989 and 1990 vintages in Bor­deaux are likely to be outstanding.Many wine writers have made thesame predictions in the trademagazines. Of course, it is still tooearly to determine whether thewines will fulfill their promise.

~arketlneflficiency

It is natural to inquire as to theprices at which the wines listed inTable 1 were sold when the wineswere first offered on the market. Inparticular, were the relative pricesof the young wines good forecastsof the relative prices the maturewines now fetch? It is difficult toanswer this question because theyoung wines were all sold in dif­ferent time periods and at pricesthat are not generally known. In­stead, we have explored a closelyrelated question: Were the relativeprices of the vintages when theywere first sold in the auction mar­ket good forecasts of the relativeprices of the mature wines? Werethe prices of the young wines ,viewed as forecasts of the pricesof the mature wines. as good as thepredictions made using the dataon the weather alone?

Table 3 reveals the answer toboth of these questions. In this ta­ble , we have listed, for each calen-

dar year from 1971 to 1989, theprice of the portfolio of winesfrom each vintage relative to theprice of the portfolio of winesfrom the 1961, 1962, 1964, and1966 vintages. The benchmarkvintage portfolio is a simple aver­age of the 1961, 1962, 1964, and1966 vintage indexes. The secondcolumn gives the value of thebenchmark portfolio in poundssterling in the year indicated andprovides a general measure of theoverall inflation in wine prices inthe London auction markets. Theentries for each of the vintages inthe remaining columns are simplyratios of the prices of the wines ineach vintage to the benchmark

portfolio in column 1 of the table.The 1961, 1962, 1964, and 1966vintages were selected for thebenchmark because the weatherdata in Fig. 2 predict they wouldbe good, and the wines from thesevintages are, no doubt as a conse­quence, still widely traded. Thevintages that are studied in the ta­ble include all those between 1961and 1972. Listed in the bottomrow of the table is the predictedrelative price of the vintage astaken from the "Bordeaux equa­tion" in Table 2.

The data in Table 3 confirm tworemarkable facts. First, most ofthese older vintages began theirlives in the auction markets at

prices that are far above what theywill ultimately fetch. For exam­ple, the bottom row of the tablesindicates that, based on theweather, the wines of a vintagelike 1967 would have been ex­pected to sell for about one-halfthe price of an average of thewines from the 1961, 1962, 1964,and 1966 vintages. In fact, thewines entered the auction marketsin 1972 at about 50% more thanexpected and slowly drifted downin relative price over the years.Second, the predicted prices fromthe "Bordeaux equation," whichis fit from an entirely different setof data, are remarkably good indi­cators of the prices at which the

Table 3-Price per Case of a Portfolio of Bordeaux Chateaux Relative to the Priceof the Portfolio for the Vintages of 1961, 1962, 1964, and 1966

Year Benchmarkof portfolio·sale Vintage

1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972

1971 £54 1.66 0.79 0.41 0.76 0.791972 £97 1.58 0.76 0.26 0.70 0.27 0.96 O.n 0.751973 £119 1.62 0.71 0.28 0.74 0.24 0.93 0.62 0.28 0.70 0.831974 £85 1.31 o.n 0.39 0.84 1.08 0.78 0.30 0.70 0.88 0.301975 £76 1.65 o.n 0.29 0.78 0.35 0.80 0.57 0.31 0.41 0.84 0.61 0.441976 £109 1.67 0.83 0.30 0.65 0.29 0.85 0.51 0.23 0.36 0.69 0.5419n £165 1.67 0.83 0.26 0.63 0.26 0.87 0.50 0.23 0.36 0.70 0.51 0.321978 £215 1.67 0.76 0.26 0.65 0.18 0.91 0.45 0.25 0.31 0.70 0.53 0.251979 £274 1.61 0.73 0.20 0.66 0.23 1.00 0.49 0.24 0.29 0.71 0.50 0.231981 £296 1.75 0.62 0.22 0.70 0.04 0.93 0.47 0.25 0.29 0.82 0.52 0.221982 £420 1.80 0.71 0.15 0.60 0.18 0.89 0.39 0.17 0.24 o.n 0.55 0.191983 £586 1.n 0.53 0.10 0.59 0.18 1.11 0.36 0.18 0.21 0.91 0.48 0.201985 £952 2.19 0.53 0.12 0.50 0.21 0.78 0.30 0.11 0.14 0.68 0.46 0.131986 £888 2.10 0.56 0.25· 0.54 0.17 0.80 0.30 0.15 0.19 0.65 0.46 0.141987 £901 2.11 0.56 0.53 0.80 0.32 0.19 0.20 0.64 0.49 0.181988 £854 2.01 0.56 0.21 0.61 0.14 0.82 0.34 0.23 0.20 0.67 0.58 0.171989 £1,048 2.09 0.61 0.28 0.53 0.19 o.n 0.27 0.24 0.18 0.66 0.43 0.15

P~edifted1.74 0.72 0.29 0.76 0.16 0.78 0.49 0.21 0.29 0.60 0.53 0.14pnce

NOTE: A portfolio(for a vintage)consistsof the wines from 15 leadingchateaux. The chateauxusedare the samefor all vintages.The "benchmark" portfolioconsistsof the portfolioof leadingchateauxand winefrom the vintagesof 1961,1962,1964,and 1966.

aAbsolute priceof benchmark portfolios.

bPredlcted relativepricefrom the "Bordeauxequation."

CHANCE 13

The Response of the Wine Press

The mere mention of the use of sta­tistical analyses to analyze the qual­ity of wine vintages has sent the winetrade press into a frenzy. The leaderin the New York Times (March 4,1995) was "Wine Equation PutsSome Noses Out of Joint," whichmay have been an understatement.Robert Parker, Jr., generally re­garded as the most influential winecritic in America, calls the approach,"a Neanderthal way of looking atwine." Britain's Wine magazine saidof the use of multiple regressionanalysis to predict wine quality that"the formula's self-evident sillinessinvited disrespect," Some of the re­actions in the wine press imply con­demnation of the entire approach for

mature wines will ultimatelytrade.

One interesting way to see theinefficiency in this market is tocompare the prices of the vintagesof 1962, 1964, 1967, and 1969 incalendar year 1972. As theweather data in Fig. 2 indicate,and the prediction in the bottomrow of Table 3 confirms, weshould have expected (in 1972)that the 1962 and 1964 vintageswould sell for considerably morethan the vintages of both 1967 and1969. In fact, in 1972 these fourvintages fetched nearly identicalprices, in sharp contrast to whatthe weather would have indi­cated. By around 1979, the pricesof the 1969s and 1967s had fallento around what would have beenpredicted from the weather.

It is apparent from Table 3 thatmost vintages are overpricedwhen the wines are first offered onthe auction market and that thisstate of affairs often persists for 10years or more following the yearof the vintage. The overpricing ofthe vintages is especially apparentfor those vintages that, from theweather, we would predict are thepoorest. This suggests that, inlarge measure, the ability of theweather to predict the quality of

14 VOL.8,NO,4,1995

reasons that most statisticians willfind humorous. For example, TheWine Spectatorsaid, "the theory de­pends for its persuasiveness on thematch between vintage quality aspredicted by climate data, and vin­tage price on the auction market. Butthe predictions come exactly trueonly 3 times in the 27 vintages since1961 that he's calculated, eventhough the formula was specificallydesigned to fit price data that alreadyexisted. The predicted prices areboth under and over the actualprices," Apparently the presence ofimperfect predictions in the regres­sion equation causes consternation,even if "they are both underand overthe actual prices,"

the wines is ignored by the earlypurchasers of the wines.

An interesting recent exampleof this phenomenon is the 1986vintage. As Fig. 3 indicates, this isa vintage that, based on theweather, we should expect to be"average." Compared to the othervintages of the last decade, thisvintage should fetch a consider­ably lower price. In fact, the vin­tage was launched with great fan­fare as among the finest twovintages of the decade. The wineswere sold at similar, and some­times higher prices to initial buy­ers than the wines of the other vin­tages of the past decade. Theenthusiasm for these wines hasdampened somewhat becausethey have not fetched auctionprices higher than those of theother vintages in the decade. Weshould expect that, in due course,the prices of these wines will de­cline relative to the prices of mostof the other vintages of the 1980s.

Conclusion

Why does the market for immaturered Bordeaux wines appear to be soinefficient when the market for ma­ture wines appears to be so effi-

cient? We think there may be sev­eral related explanations. The cur­rent Bordeaux marketing systemhas the character of an agriculturalincome stabilization system, andthis may be its purpose. Completeincome stabilization for the grow­ers would require that the price ofthe young wines be inversely re­lated to the quantity produced andindependent of the quality. Al­though the actual pricing of youngBordeaux wines falls short of thisideal, it is clearly closer to it thanwould occur if purchasers used theinformation available from theweather for determining the qualityof the wines. The producers do at­tempt to raise prices when cropsare small, despite the evidence thatthe quantity of the wines (deter­mined by the weather in the spring)is generally unrelated to the qualityof the wines. Moreover, it is com­mon for the proprietors to claimthat each vintage is a good one, in­dependent of the weather that pro­duced it. Indeed, there is no obvi­ous incentive for an individualproprietor to ever claim anythingelse!

A more fundamental questionarises about the motives of theearly purchasers of the wines.Why have they ignored the evi­dence that the weather during agrape-growing season is a funda­mental and easily measured deter­minant of the quality of the ma­ture wines? Will they continue todo so as the evidence for the pre­dictability of the quality of newvintages accumulates?

Additional Reading

Haraszthy, A. (1862), Grape Culture,Wines, and Wine-making, NewYork: Harper.

Liquid Assets: The InternationalGuide to Fine Wines, Princeton, NJ:Orley Ashenfelter, various issues.

Penning-Rowsell, E. (1985), TheWines of Bordeaux, 5th ed., SanFrancisco: The Wine AppreciationGuild.