October 2019 newsletter: Improving the quality of French wine

It is well-known that the markets for wine, at many times and in many places, have been subject to fraudulent behaviour, providing a practical example of George Akerlof’s ‘lemons’ mechanism at work. In this article, Pierre Mérel and his colleagues1 show how the adoption of a law introducing the concept of appellation d’origine contrôlée (AOC) led to an increase in the price and quality of French wine in the areas so designated.

“An immense proportion of the Wine sold in England as Claret has nothing to do with the banks of the Garonne, save that harsh heavy vintages have been brought from Spain and Italy, and dried currants from Greece, there to be manipulated and re-shipped to England and the rest of the World, as Lafite, Larose, St Julien, and St Estephe.”

(The Telegraph, cited in the British monthly journal Ridley & Co’s Wine and Spirit Trade Circular, July 1887)

In the words of British wine critic John Michael Broadbent, pronounced in 1985 in front of an audience of wine connoisseurs gathered to celebrate the 50th anniversary of pioneering French legislation on the use of wine designations, ‘Fraud and wine have gone hand in hand for centuries.’ And not exclusively on the British market. Broadbent went on to quote 19th century British wine merchant and writer T G Shaw: ‘From time immemorial, all sorts of fraud have been committed, and this situation will endure as long as greed is allowed to prevail; cunning and disregard for truth will, time and again, thwart the most vigilant of governments.’2

The welfare consequences of fraud are, at least in theory, well understood by economists. Whenever quality is unknown to consumers at the time of purchase, heterogeneous goods may sell at the same price. If quality is costly to supply, unscrupulous producers may free ride on the quality signal sent by competitors by lowering the quality of their own product. This behavior lowers average quality and, to the extent that consumers experience quality after purchase, drives down price. The price decrease further deters production of high-quality goods. This vicious circle results in suboptimal quality being supplied in equilibrium and, in extreme cases, in a complete unraveling of the market. A simple variant of this phenomenon (whereby high-quality producers simply exit the market instead of lowering quality) was described by George Akerlof in a seminal contribution published in 1970 titled ‘The Markets for Lemons: Quality Uncertainty and the Market Mechanism’ Akerlof’s piece went on to become one of the most cited articles in the economic literature. Although the ‘lemons’ mechanism is well-recognized as a textbook case of market failure, empirical evidence of its economic harm remains scarce. Indeed, it is difficult to show that higher quality does not exist in a market because of imperfect information. Welfare impact estimates have recently been published in the context of insurance markets. They typically amount to a couple of percentage points of total market value.3  

In a new working paper,4 we show evidence of large welfare effects of the lemons mechanism using historical data for a highly differentiated market: the French wine market. We exploit a change in the informational setting caused by the adoption of a 1935 law that introduced the concept of appellation d’origine contrôlée (AOC) to designate wines produced in a particular geographical region, according to specific practices, and in verifiable ways. Prior to that pioneering piece of legislation, which later became the template for the European concept of ‘protected designation of origin’, the French system of geographical appellations (dating back to 1905) merely required wine grapes to originate in broadly defined administrative areas, with usually no further requirements regarding grape varietal, the type of terrain, or the appropriate winemaking practices.


A newly assembled historical dataset of wine prices and AOC eligibility

We pair longitudinal administrative data on wine prices received by producers and measured at the level of 76 French départements (hereafter departments) with a newly constructed and geographically detailed dataset on vineyard eligibility for AOC recognition.  Our dataset covers the years from 1907 (after the concept of appellation was first codified in French law) to 1969 (before the European common market for wine), providing a 30 year-span before and after the 1935 reform. Our analysis includes 277 different AOCs whose production requirements were codified between 1935 and 1969 through administrative decrees pursuant to the 1935 law.

A critical challenge is to determine the historical eligibility of French vineyards for AOC designation in order to construct a measure of each department’s ‘exposure’ to the new policy at any given point in time.  In our study, vineyard eligibility is proxied by crossing information from three main sources. First, we exploit recent cartographic data produced by INAO,5  the French institute in charge of the AOC system, which map current AOC delimitations. Second, we intersect these maps with remote sensing data indicating parcels in vineyards in order to exclude areas unsuitable for vine cultivation, such as roads or hedges. Third, we assemble all historical decrees that either created an AOC or modified a pre-existing one during the sample period. The decrees provide an exhaustive list of eligible municipalities for each AOC, sometimes with changes in eligibility over time. Combining these pieces of information, we estimate the share of acreage suitable for vine cultivation that is eligible for AOC designation (based solely on location), in each department and each year in the sample. Because areas eligible may not end up being used for AOC production, estimates obtained by regressing wine price on the eligible share of vineyards represent intention-to-treat effects.


The effect of the reform on wine prices

Our empirical strategy compares price trends between the pre- and post-reform periods across high-eligibility and low- or no-eligibility departments, and attributes any difference in trends to AOC eligibility. Specifically, we correlate the extent of vineyard eligibility within a department to the average departmental wine price, explicitly controlling for departmental wine production and implicitly controlling for common yearly shocks to departments located in the same wine region, and for department-level time invariant factors through the use of fixed effects.

Across a large series of model specifications, we find AOC eligibility to be associated with a marked increase in the average departmental wine price (calculated as an average across all wine sales). Our preferred estimates of the effect hover around 40 per cent, meaning that a shift from non-eligibility to full eligibility (as was the case, for instance, in appellation-rich departments like Gironde (bordeaux) or Marne (champagne)) caused a 40 per cent increase in the average wine price. Since our measure of price includes all wines produced in a department, including ordinary wine, and not all eligible vineyards ended up claiming an AOC, this average price increase masks a much larger markup for wines actually sold under AOC. A back-of-the-envelope calculation using supplemental information on AOC production indicates a staggering two-fold price increase for appellation wines claiming an AOC after the reform.

Of course, departments that became eligible for AOC recognition may have differed from non-AOC departments in ways that could correlate with trends in wine prices. Yet we find no evidence of this possibility when looking at the 30-year period leading to the reform. That is, eventual AOC eligibility pursuant to the 1935 reform does not explain price patterns during the pre-reform period. Our main effect also holds when we exclude departments with no AOC eligibility, that is, when we use the intensive margin of eligibility as the sole source of identifying variation.

The welfare interpretation of the results

Interpreting the observed relative price increase in eligible departments as having been caused by wine quality improvements — which the 1935 reform explicitly intended to incentivize — requires caution. First, the increase in wine price could have resulted from a reduction in supply. However, we show that neither acreage in vineyards nor yield declined in eligible departments relative to non-AOC departments. Therefore, the observed price movement did not occur along the demand curve, but rather as the result of a demand shift. Second, we rule out an alternative explanation according to which the price increase could have been caused by the denial of the AOC designation to certain wines sold under appellation prior to the reform, essentially forcing them into the ordinary wine market — an effect known as déclassement (demotion). This is an important point, as the reshuffling of wine quantities across market segments could, at least in principle, cause increases in the average wine price without any improvement in the objective quality of appellation (or ordinary) wine. To rule out reshuffling as the cause of the observed price increase, we use ancillary data to identify departments in which there was a decrease in the share of wine sold under appellation between the pre- and post-reform periods. When we exclude these departments from our analysis, the price effect stands. Having ruled out quantity decreases and mere wine reshuffling as plausible causes of the observed price increase, we conclude that the reform provided wine producers with incentives to supply quality ultimately valued by consumers.

To the extent that they are driven by quality improvements, what do these price changes tell us about economic efficiency in the French wine market? If market price increases, holding quantity constant, then surely consumers’ willingness to pay for wine must have risen due to the reform. We thus interpret the price change as an increase in gross welfare, where the ‘gross’ qualifier indicates that we do not account for the costs associated with quality improvements.

By multiplying the estimated relative price increase by the overall share of vineyards eligible for AOC at the end of the study period, we estimate a gross welfare increase of nearly 15 per cent across the entire French wine market. Of course, these consumption-related gains could have been entirely offset by increases in supply costs, so the size of effect of the reform on net welfare remains an open question (though it seems clear that the reform did not reduce welfare). Considering that, at the national level, more than half of AOC eligible acreage ended up being cultivated for AOC wine, it seems clear to us that the reform led to net welfare improvements.

Indeed, reflecting on ten years of recognition of French AOCs in the UK following its entry into the European Common Market, John Michael Broadbent concluded his speech in recognition of the 1935 French law as follows:

“But I am glad overall to convey to you Great Britain’s sentiment that we are, at last, conscious and hopefully respectful of the laws governing AOCs.”


1. P Mérel, Dept of Agricultural and Resource Economics, University of California, Davis, Davis CA. Email: merel@primal.ucdavis.edu;  A Ortiz-Bobea, Charles H Dyson School of Applied Economics and Management, Cornell University, Ithaca, NY. Email: ao332@cornell.edu; E Paroissien,INRA, Rennes, France. Email: Emmanuel.Paroissien@inra.fr

2. Translated from French by the authors.

3. Handel, B. R. (2013). Adverse Selection and Inertia in Health Insurance Markets: When Nudging Hurts. American Economic Review, 103(7):2643-2682

4. Mérel, P., Ortiz-Bobea, A. and Paroissien, E. ‘How Big is the Lemons Problem? Historical Evidence from French Appellation Wines’. Working paper, University of California Davis.

5. Institut National de l’Origine et de la Qualité.