It would be silly to suggest that all of neoclassical economics is simply ‘wrong.’ I happen to think much of it is, sure, but some is right, and some may be merely incomplete. However there is another possibility, one I want to focus on in this post: some neoclassical theories are sound only if one defines for them a clear domain. In mathematics, a domain refers to the range of numbers one can feed into a function (what you ‘do’ to the number) and get non-nonsensical (sensical?) answers. Similar rules apply to many scientific theories: the perfect gas model is not appropriate for steam; Newton’s Laws do not apply at very large or very small levels, etc etc.
Economists do already use domains, to a limited extent. This is mostly done in theories of the firm, for which there are different ones depending on the number of firms ranging from perfect competiton to oligopoly through duopoly and finally monopoly. These theories are only supposed to hold in industries with the appropriate number of firms. However, even within these there are few criteria for distinguishing between when a firm will behave, for example, Cournot-y (varying quantity only) and/or Bertrand-y (varying price only), and hence which of these models is appropriate. So economists might still have a hard time knowing when to use which theory. DSGE has similar problems.
One area I’ve been thinking might be more sound if a specific domain – agriculture – were applied to it is marginalist economics: specifically, the much maligned perfectly competitive theory of the firm. It is perhaps no coincidence that economists are rather keen on using examples from agriculture in their parables about marginalist concepts: it’s the area where their analysis is most appropriate. There are a few reasons to believe this:
(1) Agriculture, for the most part, has perfectly divisible inputs and outputs. These are a core assumption of basic producer (and consumer theory), one which is blatantly unrealistic in most cases. However, it may be realistic in agriculture. Food and fertiliser are literally perfectly divisible, as they can always be cut down to smaller quantities; certainly at any level relevant for production. Livestock are not perfectly divisible when alive, but even so they are generally farmed in large quantities that can be continually adjusted, so perfect divisibility is at least a good approximation. Tractors, ploughs etc. are example of indivisibilities, but they are not often purchased and can be thought of as the exception to the rule, covered under ‘fixed capital.’
(2) Diminishing marginal returns. Agriculture is one of the few areas where we observe rising costs as output rises. This is partly because a major factor of production – land – is fixed. This is a standard assumption for the short run neoclassical theory of the firm; with land, it is also true in the long run, though some improvements in productivity can be made over the long run with the aforementioned fixed capital expansions.
(3) Perfect competition. Nothing better resembles the atomistic neoclassical ideal than many farmers competing on a single market with homogeneous goods like wheat, not having any discernible effect on price. With certain foods, some product differentiation (through quality) might be observed but even this would be captured by the theory of imperfect competition. Overall, a farmer is less likely to have discretion over the price of what they sell than, say, a retail store, or a lawyer.
(4) Lack of clustering or ‘QWERTY‘ effects. It is an obvious observation that firms in particular industries tend to cluster together geographically. Manufacturing requires a continuous stream of inputs, so firms at different stages in the supply chain will group together to minimise transaction costs. Manufacturing often – though not always, to be sure – requires workers with a particular set of skills, so employees and employers who best match together will tend to converge. Services, by their nature, requires face-to-face interaction, as well as even more specialised skills, so they too will group together. In both cases the easy transfer of knowledge around clusters also helps significantly. Clusters become self-enforcing: you set up shop in a cluster because everyone else in your industry is there. QWERTY effects create emergent properties that may suggest a role for government intervention.
However, agriculture, in most cases, does not exhibit QWERTY-like characteristics. First, agriculture requires large expanses of land so it is difficult to create ‘clusters.’ Second, most agricultural labour is not particularly specialised. Third, agriculture also follows an obvious harvesting cycle, so rather than a continuous stream of inputs, there are intermittent large purchases of supplies, making transportation costs less of a systemic issue. Fourth, agriculture does not really rely on information about new trends, management, techniques or what have you; it has followed similar techniques for centuries.
The reader might note that I’ve primarily been referring to extensive agriculture, rather than intensive agriculture – market gardens and so forth. Intensive agriculture does exhibit some characteristics similar to extensive farming: it produces the same type of goods, for a start, so much of the above still applies. Nevertheless, the use of technology and organisation is greater than extensive farming, and market gardens generally take up a smaller area, which suggests that the perfectly competitive market may not be appropriate. Modern market farming might be thought as a way to ‘capitalist-ise’ agriculture, hence rendering the perfectly competitive theory inappropriate.
So what are the implications of this, for extensive farming at least? Seemingly, our conclusions will align with the conclusions of basic economic theory. Price controls and subsidies are not advised under normal circumstances or in the name of long term policy goals; a monopoly would probably not be a result of innovation and would be unlikely to be superseded by technology, and so would be unambiguously bad.
Most of all, economists will be pleased to hear that their favourite theory, comparative advantage, is more directly applicable in the world of agriculture. This is for two main reasons. First, the most commonly used rationale for why a country might have ‘comparative advantage’ – resource endowments – is obviously applicable in agriculture: nobody questions why the UK doesn’t try to create a cocoa industry, or why New Jersey doesn’t grow as much wheat as Iowa. Fertility of soil and climate are determined by powers mostly beyond humanity’s control, and we must specialise according to this. Second, unlike manufacturing, short term losses in trade will not strengthen an industry to the point where it is more efficient in the long term.
This is basically a ‘market knows best’ mantra that may not sit well with my regular readers. To be sure, there will still be exceptions where governments might intervene: environmental concerns; ensuring national self sufficiency; emergencies; basic standards. Nevertheless, the disaster that is the CAP, with absurdities such as food mountains and paying farmers not to use their fields, as well as the effect it has on farmers in poor countries, seems to illustrate that if economist’s favourite creeds hold anywhere, it’s in agriculture.
Model-wise, there will still be issues with perfect competition even in agriculture, where it is at its most relevant. I fully expect superior, more comprehensive theories than the perfectly competitive firm can be (and have been) developed for agriculture. Nevertheless, insofar as perfect competition might apply to anything at all, it seems most suited here. It would at least be a start for economists to admit certain theories have only limited application, instead of extrapolating highly restrictive models onto situations where they don’t apply.