Chapter 6

Possible gains from partial price stabilization in wine

Risk and stabilization

Wine is the result of the transformation of only one major input – grapes; its annual production is subject to extensive fluctuations caused by the occurrence of pests, diseases and major variations in weather conditions. No apparent cycle has been detected to date, therefore changes in output caused by these factors should be considered as random shifts (stochastic shocks), the effect of which are limited to the short term [I]. This is called risk in supply. Given a competitive market and a stable demand, any significant departure from the quantity usually supplied is matched by a rapid change in price (the market clears). The speed and the extent to which prices respond to random shifts occurring on the supply side has been particularly well illustrated in the course of the 1972/73 and 1979/80 marketing years, where prices soared, only to plunge afterwards, in both Italy and France. Similar, but less important disturbances may arise on the demand side, e.g. following drastic changes in taxation and promotion. This is a problem for most agricultural and primary commodities, whereby price movements are reckoned to be amplified by stochastic shocks occurring essentially on the supply side. Price elasticity of demand plays a crucial role, for the lower the elasticity, the larger the resulting price changes. The same applies for producers’ gross revenues, provided again that demand is price inelastic [2].

Assuming production costs are fairly constant (the only major variable cost is manpower at harvest time, which consists of casual labour mainly) then net producers’ income would also fluctuate in line with the price of wine. This is income risk. The very nature of the policies developed for wine in France and, later, in the Community reflect the presumption that, on the whole, producers would prefer a stable income to a risky one. Given the above, one may attempt to reduce risk in income by reducing fluctuations in price with a buffer stock programme. In stabilizing prices within a predetermined band, one contrives to weaken the pendulum of bargaining power between excess supply (a so-called buyer’s market) and excess demand (a seller’s market).

Regarding welfare, it has long been established that a situation where prices are stable is preferred to one of instability. This assumes that costless compensation takes place between those who end up better off and those who are adversely affected by stabilization. Suggested forms of compensation are a lump sum or tax, but indemnification is a concept that remains largely idealistic [3]. Analogous gains to those that arise from free trade could be achieved by moving goods not through space, but through time, from periods where the goods attract low prices to others where they are higher priced. Doing this is called arbitraging a commodity. Arbitrage has three major effects. It generates a cash benefit to those who store the commodity (arbitrage benefit), it redistributes income amongst agents (transfer benefit) and it reduces their risk (risk benefit) [4]. The precise level of the costs incurred in operating a buffer stock is difficult to determine a priori. It is thus common practice to assume that the programmes are costless to run. This greatly simplifies the valuation of the gains, but this also means that one does not carry out a proper cost‐benefit analysis, only an approximation of the welfare gains that would accrue to the agents situated at both ends of the market [5].

Welfare gains have traditionally been evaluated in terms of producers’ and consumers’ surpluses (Marshallian analysis) and the best-known framework for assessing the benefits is the Waugh-Oi-Massel (WOM) linear model [6]. Take a closed, competitive market which is in a permanent state of equilibrium and represent it with the following set of equations:

Q(t) = a0 + a1 P(t) + u(t)

D(t) = b1 – b1 P(t) + v(t)

D(t) = Q(t)

Supply (Q) and demand (D) are both linear functions of the current market price (P). The additive disturbance terms u and v are meant to capture stochastic shocks in supply and demand respectively, such as those caused by weather changes or, say, by a national advertising campaign (so-called additive risk). The third equation in the model is the market clearing condition. It is there to enforce equilibrium, so that both price and traded quantity are determined simultaneously. The model assumes that neither the supply side (producers) nor the demand side (consumers) will engage in any storage activity [7].

Suppose that an intervention agency wishes to do just that. Namely, it wants to reduce price variability by monitoring a buffer stock, which would compensate for the net effect of the two shocks. To achieve this, the agency would constantly put onto the market, or remove from it a quantity X(t) = v(t) – u(t), thereby ensuring that the nonstochastic (systematic) parts of supply and demand are equal at all times. The resulting price would be stabilized at its long run equilibrium value, the arithmetic mean in this very particular case. This is called perfect price stabilization. Whilst it is desirable in theory, it is totally impractical: running such a programme would require a prohibitively large number of transactions and, more generally, one has very little knowledge as to the exact nature of the supply and demand schedules. So, assume the agency were to engage in partial price stabilization, whereby it would only seek to reduce, not eliminate price variability. For instance, it could decide to intervene in accordance with the following rule :

X(t) = z{P(t) – P*(t)} + w(t)

In other words, the agency would buy or sell a quantity X(t) which would stand in direct proportion z to the departure of the current price P(t) from its target path P*(t). The scalar z measures the intensity of intervention and w(t) accounts for errors the agency would make, though one would require it to turn its arbitraging activity into a profitable business, by retrieving goods from the market when there is a surplus (buy cheap) and introducing them again when supply is relatively low (sell dear). Turnovsky (1978, pp. 134-7) stresses that chances of success for a stabilization scheme of this type would depend much on the agency’s performance as an arbitrageur. This point could hardly be overstated, for errors caused by undue intervention, which one assumes to be stochastic (i.e. with mean zero), could well result in the programme having a negative effect altogether. The notion of perverse effect is no recent addition to the interventionist’s vocabulary (see Milhau,

1935, p. 27). Rather, it is a constant reminder to policy makers that they may end up furthering what they tried to avoid in the first place. This is usually the result of them acting too late, on the basis of insufficient information, or under political pressure. Note, the target price P*(t) would have been previously identified by the agency as the price at which the nonstochastic parts of supply and demand would be equal, just as in the model for perfect price stabilization. P*(t) is the underlying long run equilibrium price, also called the systematic price, for it is the result of systematic market forces. The taxonomy is borrowed from Lee and Blandford (1980), who have proposed to estimate the systematic price by way of simulation, by calculating and removing the undesirable effect of stochastic shocks on price. Doing this allows one to be explicit in the distinction between stochastic and systematic sources of price variation and, at the same time, to view partial price stabilization as a mean-supply-preserving reduction in price dispersion. The less interested reader is invited to browse through the remainder of this chapter and Chapter Seven and to take up his reading at the beginning of Chapter Eight.

* * *


[1] Stochastic shocks are independent of each other. They are usually assumed to follow a Normal distribution, with mean 0 and a known variance (Box and Jenkins, 1976, pp. 8, 46-7).

[2] Since the quantity supplied each year, Q, is predetermined and, if demand is inelastic, prices P will fluctuate by a greater proportion for any given proportionate fluctuation in Q. Similarly, the more inelastic is demand, the greater the proportionate fluctuation in P for any given fluctuation in Q and, thus, the greater the fluctuation in revenue = PQ. In this chapter, one means prices in real terms, invariably.

[3] Boadway and Bruce (1984, p. 16).

[4] See Newbery and Stiglitz (1981, p. 252ff).

[5] A review of the different methods for evaluating the costs and benefits of buffer stock operations appears in Labys and Pollak (1984).

[6] The description and the results for the WOM model below are adapted from Turnovsky (1978).

[7] By producers one means wine producers, though the notion can be extended easily to include grape growers, as in the typical case of grape growers’ co-operative wineries. The term consumers is used in an even wider sense in this chapter, to simplify. It will thus fall on the reader to determine from the context, whether one is speaking of consumers in the strict sense, or of all the agents situated on the demand side (wholesalers, retailers and consumers taken together).

* * *


BOADWAY, R., and BRUCE, N. (1984) “Welfare Economics”. Basil Blackwell, Oxford.

BOX, G.E., and JENKINS, G.M. (1976) “Time Series Analysis: Forecasting and Control”. Holden-Day, San Francisco.

LABYS, WC, and POLLAK, P.K. (1984) “Commodity Models for Forecasting and Policy Analysis”. Croom Helm, London.

LEE, S., and BLANDFORD, D. (1980) “An Analysis of International Buffer Stocks for Cocoa and Copper Through Dynamic Optimization”. Journal of Policy Modeling, 2(3):371-88.

NEWBERY, D., and STIGLITZ, J. (1981) “The Theory of Commodity Price Stabilization – A Study in the Economics of Risk”. Clarendon, Oxford.

TURNOVSKY, SJ. (1978) “The Distribution of Welfare Gains from Price Stabilization: A Survey of some Theoretical Issues”, in F. G. Adams and S.A. Klein (eds) ’Stabilizing World Commodity Markets’. Heath, Lexington, 119‐48. © pierre spahni, 1988

© 1988 pierre spahni /

* * *