An X-Ray of the Risk Module within the momagri Model
Pr. Bertrand R. Munier,
Chief economist, momagri
The Historical and Scientific Foundations
The aim of the momagri team of economists since the early 2006, the time at which the construction of a model received a blueprint, was to give a better understanding and a more accurate image of agriculture throughout the world than the one reflected in the major models used in international organizations.
For the picture given by most often used models is quite strange. We all have been told for decades that agricultural prices would sometime stabilize around their market price. Yet, in the last thirty years, agricultural prices – unlike industrial prices - have been fluctuating more and more widely.
We have also been told that the reason why these prices have been fluctuating so wildly was only a temporary one. It is only because men have been unwise enough to refrain from getting rid of all tariffs, quotas and other type of barriers to free trade, including subsidies to farmers. But removing such barriers would open the road to stability. Yet, this type of barriers have been more and more removed in the last twenty-five years, and fluctuations of agricultural prices have gone the same puzzling way, in some cases frightening us.
Maybe the road opened is a very long road, but with such utterly long roads, won’t we all be dead before anything happens?
How to solve that puzzling situation?
I suggest that the clue to our amazements can best be found in contemporary history. In the early Thirties, most economists – in particular Roosevelt economists - came to consider that agriculture called for specific discretionary policies, even if these policies differed from a strictly orthodox set of rules like the ones we want to govern industrial and service companies. The effects on farmers’ production, if not unquestionable, were substantial.
After World War II, the issues had entirely changed. The problems from which people were suffering regarded by then exchange controls, inconvertibility of many currencies (including all major European currencies), trade barriers of all sorts which had been necessary to the war economies, lack of private initiative inherited from different pre-war organization in some countries, etc. In one word, abusive regulations were bringing poverty or low incomes and, at the same time, a profound need of economic development. That was the time of modernization, of growth at any rate, the one of the triumph of engineering, which led to huge public works without too many worries about environment or other collateral effects, the one, finally, during which economics students in western countries suddenly came to discover that there was a “third world”.
The focus of critical issues had thus completely changed. No wonder that the focus of solutions had entirely changed as well. Under such circumstances, how could one ever question that free markets were basically the pattern to aim at, except when some failure case appears, which economists carefully classify?
To be sure, the changing course was slow, some backlash took too often place, as the difficulty in removing exchange controls showed, as did the difficulty in France to get rid of some abusive corporations. Several problems were encountered and some more selective course of actions have been taken now and then. But the general conviction remained and was later strengthened by the crumbling dow of the Berlin Wall.
Jolly good! As for the specificity of agriculture, however, we entirely lost it along the way. I think our main models in use for international negotiations are products of that post-World War II movement of thought. There is the crux of the matter.
This is the reason why the elaboration of the momagri model was initiated. It follows an original modular architecture and innovating design guidelines, in particular with respect to risk modeling of the behavior of actors, be they producers, speculators or the government.
The Risk Module allows to model the strong volatility of agricultural prices on the international market, taking into account the specific market microstructure of the agricultural sector:
> Prices are way more volatile than their counterparts in industrial activities or in services, which makes them particularly difficult to anticipate
> Current production decisions (and not only investment decisions) taken at « t-1 » are virtually one hundred percent irreversible throughout until time « t ».
> The market is not only a spot market of physical goods, there are also forward and options markets, which allows covering or hedging, but also open positions for speculators of different types
> And agricultural productions are meanwhile subjects to natural hazards, wheher climatic or of other types.
It can be shown that, taking into account these four items leads to show that a large part of the risk exposures of farmers are of an endogenous type and constitute what i would term a natural market imperfection, using a simple analogy with the natural monopoly expression.
The momagri team, when elaborating the Risk Module, have produced suggestions regarding the treatment of risk and have followed a way of modeling speculation derived from the tradition of Frankel and Froot (1988), Kirman (1991) and several others since then (Brock and Hommes, 1997, etc.)
Rather than buying information in a rational way on the basis of what it costs and of the probability to find the best price, as is done in the « search » models - which could be adapted to speculators - we work here on the hypothesis that the further from the « fundamental» market equilibrium price, the more “fundamentalist” speculators jump on the market, on several grounds, in particular taking into account the opinions they observe among other actors. This variable has been stressed in contributions by several economists among which some of our own works (1991, 1994). This behavior is however taken here to be one of ‘ants’, as it has often come to be called. Meanwhile, speculators described inaccurately as “naïve” and even more inaccurately as “chartists” are modeled as anticipating purely and simply a continuation of the price trend. The respective proportions of these speculators are then endogenous and have an impact on the volatility of prices.
Expectations of agricultural producer are also based on a bounded rationality hypothesis in a risky environment. They derive both from direct observations made by members of the team made during the preparatory work as well as from an adaptive formula “à la Nerlove”. Direct observations led us to use a combination of the past prices faced by producers with their unitary “intakes” during the last three years (the model is thus a system endowed with “memory”). The formula uses a generalized concept of certainty equivalence (an extension of the rank dependent model and a simplified version specifically designed for agriculture) combined with an adaptive formula in the Allais-Friedman style. It is here appropriate to recall that the rank dependent model was initially produced by the Australian economist John Quiggin (1981) when working on farmers’ decisions. Quiggin said to have based his discovery on an idea originally formulated in the Fifties by French Nobel Prize winner Maurice Allais.
From one period to the other, demand parameters move in a specified way:
D’une période sur l’autre, les paramètres de la fonction de demande sont modifiés de deux façons :
> They keep the direct price elasticity a given level, which we determined on the basis of the best available data (derived from the GTAP1 data).
> They take into account in a simplified way the population growth.
Summing up, the ‘Risk’ module within the momagri model is based upon a bounded rationality behavior of several actors (producers, naïve and fundamentalist speculators). It comes close to some ‘evolutionary’ models, but extends the analysis to the uncertainty case, with a specific treatment of that case. This treatment is justified by the very nature of agricultural markets microstructure. It epitomizes the natural imperfection of these markets linked to the nature and origin of uncertainty.
From a more analytic point of view, the model is a non linear system, with feedbacks (hence the complexity, which could lead to chaos), even if the basic equations of the model are of a linear type (producers supply, total demand). Moreover, the system is “semi-open” to the extent that it changes its internal structure due to the information brought back from the outside, namely for the central module of the momagri model (which depends only in a limited way from the output of the Risk module in the preceding period, of course). Such systems are instable, as has been shown by a stream of research going back to Arrow and Nerlove (1958) and pursued by Grandmont, Brock and Hommes in particular since the early Eighties. In the Nineties, Boussard (1996) contributed to that tradition.
The instability is here “attenuated” to some extent by Risk aversion in the sense of Arrow and Pratt, and even more so by the suggested extension of the certainty equivalent based on the idea derived (Hilton, 1988) from the rank-dependent model, except in the cases where evolution is contrasted, in which case the notion of optimism – pessimism in the sense where it is used here can instead increase instability. This mechanism is one of the factors of turning around trends which appear in simulations. It aims at approximating as much as can be done agricultural markets volatile evolutions – without ever pretending to back-testing, even less so to predicting – these evolutions. Its significance is in predicting the kind of markets we can have in the future according to the policies pursued.
We hope to have thus shown that such evolutions are inherently linked to the very microstructure of agricultural markets, not to some natural risk – whether linked to the climate or to some other factor – as is generally accepted (Newbery and Stiglitz, 1981). For some values of the parameters, which we havn’t looked for here, for our main goal was different, such a market could produce chaotic results. It is known that the “chaotic” character of observable agricultural prices has been occasionally mentioned and that several scientific studies dealing with available data have concluded that their behavior was somehow closed to it (Jensen and Urban, 1984, Chatrath, Adranji and Dhanda, 2002).
1 GTAP (Global Trade Assistance and Production) is global data base describing bilateral trade patterns, production, consumption and intermediate use of commodities and services.