Earlier Theoretical Work, Mostly on Developing Economies
David’s early work shows his commitment to the application of economic theory to problems of practical interest, the problems being largely those faced by developing countries. The genius of much of this work is that it starts from the fundamentals of microeconomic analysis and sees how far we can get in the face of the particular problem being addressed. David’s work demonstrates that neoclassical economics in practice is not—as it is so often characterised by its critics—about the blind application of textbook models in spite of real-world complexity. It is about providing a place to start in getting a conceptually tractable handle on what might otherwise be too easily (and lazily) characterised as a unique problem by those who dismiss microeconomic theory as being too abstract to have practical value.
David’s early theorising drew on his experiences in Tanzania combined with his interest in the advanced theory being produced by his colleagues, notably Tony Atkinson, Jim Mirrlees, and Ken Arrow. Thus in Newbery (1970), David offers an elegant two-page mathematical proof of a point made by Atkinson (1970) that the empirical measures of inequality (such as the Gini coefficient) are unreliable at the national level. David’s paper proves that ‘[t]here exists no additive utility function which ranks income distributions in the same order as the Gini coefficient’ (Newbery 1970: 264).
David’s lifelong interest in project appraisal is well demonstrated in his 1976 book (with Scott and MacArthur) on Project Appraisal in Practice: The Little-Mirrlees Method Applied in Kenya. Little and Mirrlees’ (1968) book on social cost-benefit analysis in developing countries made the argument that when conducting a project appraisal of costs and benefits in an underdeveloped economy, world (or border) market prices should be used for traded goods, not local market prices. This correctly reflects the true opportunity cost of goods traded in international markets, whatever their local valuation. However, the shadow price of labour, important for non-traded goods and services in a developing country, that should be used in social cost-benefit analysis is usually below the market wage. As a result, the shadow price that should be used for non-traded goods and services is somewhere between the local market price (evaluated at the official exchange rate) and the world market price. These values need to be estimated, and this is what the authors’ work does for Kenya. They argue that for developing countries, shadow prices are usually less than market prices at the official exchange rate because the official exchange rate is overvalued.
The book’s empirical calculations include an evaluation of whether it is NPV (net present value) positive to import grain to fatten cattle for export. The argument is that the world price of grain is low (due to agricultural subsidies), while the world price of beef is relatively high. Thus, moving up the value chain (to produce beef rather than grain) can be beneficial for developing countries, even though the opportunity cost of grain is its world market price. One can see the value of sound microeconomics to the practical problems of developing countries shining through this sort of argument. As Schafer (1979: 395) quotes in his excellent book review, the authors point out that social cost-benefit analysis is ‘an art which can only be learned by practising it’.
The influence of his Tanzanian experience is evident in another significant theoretical paper from David on ‘Risk Sharing, Sharecropping and Uncertain Labour Markets’ (from 1977). Sharecropping describes a contractual situation where a tenant farmer pays the landlord a fixed share of the crop. Traditionally, this arrangement was thought to be economically inefficient (by no less than Adam Smith himself) because it reduced the marginal incentives to add inputs on the part of the farmer. David’s paper sought to explain the circumstances under which sharecropping was actually an efficient arrangement. Like so much of good microeconomics, this paper is in the tradition of trying to explain why something so widespread—that looks like an example of inefficient underdevelopment—has actually a clear efficiency rationale. David’s argument looks at sharecropping as a means to spread production risk and as a means to spread input price risk. What David shows is that if wages are correlated with output, then sharecropping can be an efficient way of spreading labour (input) cost-related risk. Here we see a forerunner of David’s later work with Joe Stiglitz, in that he models a developing country world where formal risk markets for commodity producers do not exist and where other means have to be found by them to mitigate risk.
David’s most cited work is his 1981 book with Stiglitz on The Theory of Commodity Price Stabilization: A Study in the Economics of Risk. This is a quite brilliant attempt to extend general equilibrium thinking to incorporate risk, in situations where risk markets are incomplete. This book was written in the context of international (i.e. World Bank) concern about the volatility of commodity prices and the impact of this on developing countries. This had led to calls for commodity price stabilisation (through the holding of international buffer stocks) to stabilise the incomes of developing countries. However, Newbery and Stiglitz point out that it is not that simple and carefully attempt to analyse who gains and who loses and by how much. Rather cheekily, Dowie’s review (1983) suggests the book might have been better titled ‘The Economics of Risk: A Study of Commodity Price Stabilization’ (ibid.: 230).
I can still recall the intellectual thrill of learning (as an undergraduate at Cambridge) the following example which lies at the heart of their insights. Imagine there are two countries producing a single agricultural commodity. There are no insurance markets. Output is perfectly negatively correlated between the two countries. There is unit elasticity of demand in each country. Should we act to stabilise the price of the commodity? No, because if we do that, then we will stabilise the price but not the income of farmers in the country. It is income, not price, which enters utility functions. Indeed, the best thing for the farmers in the countries is that we leave the countries in agricultural autarky because in that case the prices will rise in inverse proportion to national agricultural output and hence agricultural incomes will be stable. This latter point is clearly made by Newbery and Stiglitz in their 1984 Review of Economic Studies paper.
The idea that price stabilisation will not be effective at stabilising income or consumption is at the heart of the book. The authors say it is written with three different audiences in mind—policy, agricultural, and general economists (whose various messages are nicely reviewed in Behrman 1985). As is characteristic of David, it starts with the general equilibrium model and relaxes its assumptions in important ways that reflect reality and sees where this goes. In this case, it provides an argument that a favoured intervention—price stabilisation through buffer stocks—is unlikely to work because the net gains are small, while the costs of speculative attacks if the operators of buffer stocks miscalculate are high.
David’s second most cited work is his 1982 American Economic Review paper with Richard Gilbert on ‘Preemptive Patenting and the Persistence of Monopoly’. This is another theoretical paper motivated by a real-world example—that of the alleged patent thicket of unused patents created by Xerox to prevent its rivals from competing with its products. In the paper, Gilbert and Newbery show that an incumbent monopolist has more incentive to come up with an unused blocking innovation than a rival firm. The paper characteristically starts with some classical theory and shows the circumstances in which it does not apply.
Arrow (1962) basically argues that incumbent monopolies have less incentive to innovate than a rival firm seeking to gain the monopoly. This is because for a cost-reducing innovation, the monopolist only gets the difference between the profit after the new innovation (n2) and its initial positive profit (П > 0; n < n2). This is less than a rival, who if they can get the monopoly after an innovation gets the same profit (n2) but a higher increase in profit than the incumbent (who only gets n2-nj). What Gilbert and Newbery brilliantly show (another warm glow comes over me as I recall reading this argument for the first time as a graduate student!) is that in a differentiated product market, the incentives to innovate are different. If the rival innovates, it gains less than half of the initial monopoly profit of the incumbent, because the market is now a duopoly. On the other hand, if the incumbent prevents the loss of its monopoly by innovating, its incentive to innovate is strictly greater than half of the monopoly profit, because it has stopped the market becoming a duopoly. QED: this suggests that there is a clear incentive for incumbents to create patent thickets if they can. However, in typical David style, the paper is careful to state that it is very difficult to decide in practice whether a particular R&D investment is pre-emptive.
While David was working at the World Bank, he was involved with a project which led to his 1987 co-edited book The Theory of Taxation for Developing Countries, with Nicholas Stern. This book focuses on the application of optimal tax theory to developing countries. As Toye (1988) points out in his review, this book (at 694 pages) is actually two books in one: the first co-authored between Stern and Newbery, and the second an edited conference volume. The idea behind the book is to apply Diamond and Mirrlees (1971a, b) thinking on optimal taxation to developing countries. The book consistently suggests that developing countries have very inefficient tax systems (when evaluated from a Diamond-Mirrlees viewpoint) and addresses the important issue of how their tax systems should be reformed. As is characteristic of David’s work, appeals to high theory when carefully applied rarely lead to definitive conclusions, though they can give clear guidance in particular cases. As Hines (1989) points out in his review, the editors manage to bring out a consistent message that optimal tax theory is an important reference point for tax reform. This is something very much reflected in David’s undergraduate lectures on applied welfare economics and links into the next batch of his work we review.
