Angus Deaton Symposium – Introduction


The articles in this Feature are selected from among those presented at a conference held in honour of Angus Deaton in September 2009. The conference brought together current and former colleagues, co-authors and students. These participants had in common their shared admiration of Angus and his contributions to economics.

Angus earned his formal degrees at Cambridge and after spells working at the Bank of England, Cambridge and Bristol, he found his current home at Princeton University where the conference took place. Among Angus’s important early academic influences were W. M. (Terence) Gorman and Richard Stone; the latter was his mentor and the former his examiner.

Angus has contributed to so many areas of economics that it is difficult, in a brief appreciation like this, to do justice to his achievements. His early work focused primarily on consumer behaviour. His landmark book Economics and Consumer Behaviour with John Muellbauer brought the role of duality to a new generation and showed how it could be brought to the data (Deaton and Muellbauer, 1980a). Although the influence of Gorman and Stone is evident throughout the book, Angus and John took things much further. In the course of this work, they developed the Almost Ideal Demand System (Deaton and Muellbauer, 1980b). It offered a flexible and estimable functional form which had the desired theoretical properties and has subsequently been widely used in applied research on consumer demand.

Angus has made major contributions to the literature on saving behaviour. His article on intertemporal saving and labour supply with Martin Browning and Margaret Irish introduced the method of constructing true panel data from time-series of cross-sections (Deaton, 1985). This article and its successor are both widely cited and used in modern analysis (Deaton, 1985). In 1987, he also posed what has become known in the macro literature as the ‘Deaton paradox’ (Deaton, 1987). This starts from the realisation that most time-series analysts find that average personal income in the US (and elsewhere) is a unit root process with positive autocorrelation in first-differences. But if this is true, a unit shock to labour income should translate into a greater than unit shock in permanent income, so that the permanent income hypothesis predicts, not that consumption is smoother than income (which is what the data show) but that it should be less smooth than income.

Two other articles on saving have also had a large impact on subsequent literature. Angus’ article on buffer stock saving introduced a model that is now one of the basic workhorses of the analysis of consumption and saving (Deaton, 1991). The article characterises a stationary equilibrium in which consumers cannot borrow but can save. They bump along holding assets for short-term smoothing purposes but never accumulate substantial stocks of wealth. Another article written with Christina Paxson explores the relationship between intertemporal choice, insurance and inequality (Deaton and Paxson, 1994). This article has spawned a large and still-controversial literature in macroeconomics on consumption insurance. In 1991, Angus was asked to give the Clarendon Lectures. He used this as an opportunity to take stock of the literature, pulling together a whole body of his and others’ research. The resulting book, Understanding Consumption, provides an impressive account of the state of theory and evidence on consumption and saving (Deaton, 1992).

In the 1980s, Angus turned increasingly towards research in development economics. He joined and subsequently became the director of the Research Program in Development Studies at Princeton University. There, he built a group of colleagues and students interested in development issues, making Princeton one of the leading places for research in this field. Through his work with the World Bank, Angus contributed to the design of the Living Standards Measurement Surveys, which moved the field of development forward by creating a supply of household survey data from a range of countries. Angus used these and other data sets to produce influential research on topics such as intra-household discrimination in resource allocation between boys and girls (Deaton, 1989), the separation of quantity and quality effects when estimating price elasticities of demand (Deaton, 1988) and economies of scale and food consumption (Deaton and Paxson, 1998a). This programme of work culminated in another landmark book on household surveys in developing countries which quickly became a standard reference (Deaton, 1997).

In a series of articles with Guy Laroque, Deaton has examined the standard model of speculative storage for commodity prices. The world prices of commodities have profound effects on the economies of many of the poorest countries in the world, mainly but not exclusively in Africa, and a deeper understanding of price behaviour would likely have high payoffs for macroeconomic management in many poor countries. Their 1992 article has become the standard statement of the theory and of its implications although, as their subsequent work has shown, the model does a poor job explaining the data, a major disappointment (Deaton and Laroque, 1992, 1995, 1996).

Since the mid-1980s, Angus has been working on the measurement of poverty, in particular on how to adjust measured incomes for spatial differences in prices. This work – much of which has been published mostly in India – has recently been endorsed by the Government of India, whose poverty statistics will now be based on this methodology. He has also tried to understand why poverty rates around the world fall much less rapidly than might be expected given the growth rates that we have recently experienced, as well as how the construction of purchasing power parity exchange rates affects measures of global poverty and inequality (Deaton, 2005; 2010).

In recent years, much of Angus’s work has focused on health status and mortality, and their relationship with economic status. This work is new enough – and is still being produced at such a phenomenal pace – that it is too early to assess what its full impact will be in economics and the broader fields of health and epidemiology. However, Angus’s research in this area has already shaken up conventional wisdom. For example, he has provided convincing evidence that, contrary to a widely-held belief, inequality is not a health hazard (Deaton and Paxson, 1998b; Deaton and Lubotsky, 2003). This research, together with the other economic and epidemiological literatures, make up the subject matter of Deaton’s magisterial review ‘Health, inequality, and economic development’, in the 2003 Journal of Economic Literature (Deaton, 2003). In another important article, he shows that contrary to much work by economic historians, income and human height (often used as a summary measure of living standards) are only very weakly linked – and in some cases not even positively so (Height, health, and development, 2007). As a result, population heights are an unreliable indicator of living standards. Disease – especially disease in childhood – is the missing link. In a recent article with two students in Demography, Deaton has constructed a model of stunting and selection that provides new insights into how early childhood health affects adult mortality (Deaton et al., 2009). Even here, the story is complicated, and he has shown that in India, where net nutrition has certainly been improving and people are becoming taller, men are getting taller at three times the rate of women.

Many characteristic traits shine through the body of Angus’s work. He addresses issues of immense practical significance. He is second to no one in his concern for data quality and measurement issues. He is respectful of the need to ground empirical analysis in economic theory. And, his research is as likely to uncover puzzles – instances in which the data do not support the theory – as it is to provide confirmation of the models. The consequence is that his work has opened up major avenues of new research for others to explore.