Skip to content

Fieldwork and model building in economics— Part 2

↓ Jump to responses

Download the newsletter ›

By Karim Errouaki2

[Editor’s note: Part 1 is in Issue 6-6 HERE]

Fieldwork, Conceptual Analysis in Economics

Part I argued that fieldwork means finding out what people actually do, how they actually think and behave, and what they mean when they say something. Part II will be devoted to the examination of fieldwork in economics. The main thesis of the note is that to understand and sometimes even to discover the truths of reason, it is necessary to investigate the world, and especi­ally, perhaps, to investigate investigating. Our approach is a striking parallel to Marschak’s call for interviews with businessmen in order to clarify specification of the investment function.

Fieldwork in Economics

Nell (1998) argued that fieldwork has not been widely discussed or widely employed in economics – but it has been there right from the beginning. Adam Smith visited a pin factory, and observed it closely. This led him to explain how the division of labour worked. But, in general, economists have not done much fieldwork.

In view of the importance of Adam Smith’s example, why are economists reluctant to give prominence to fieldwork? There are exceptions: the intuitionalists did it; and much industrial organization is based on fieldwork, as is a good deal of labour economics (Andrews, 1949; Bewley, 1999; Blinder, 1998; Commons, 1968; Edwards, 1979; Florence, 1972). Work on the ‘informal economy’ provides a good contemporary example (Portes et al., 1989). Surveys of consumer confidence (Oxford surveys, conference board, NBER/Sloan Foundation, INSEE) reflect fieldwork, but most so-called empirical work today is based on number-crunching (Nell, 1998, p. 101; Nell and Errouaki, 2013, Ch. 10).

More recently, Swann (2008, p. ix) proposed a new direction and a new attitude to applied economics, what he calls “vernacular knowledge of the economy, knowledge of the economy gathered by ordinary people from their everyday interactions with markets.” He argued that “such vernacular knowledge may sit uncomfortably with the formal models of economists […] But no wise economist should discard the vernacular, because it offers insights that can never be found in formal analysis alone.”

Fieldwork calls for participation: to know the meaning of a social practice, it is necessary to experience it in some way. It may be possible to gain an understanding imaginatively, or through discussions with participants; and it is certainly not necessary to participate in every aspect. But participation ensures that the observer directly experiences the social practice and can check the meaning and appreciate the nuances by asking other participants. The object is to get beneath the surface, to contrast actual behaviour with the ‘official’ view, and to relate language and description to behaviour (McCloskey, 1985). It draws on the method of ‘Verstehen’, a method that economists tend to regard with suspicion although it was central to the work of the German historical school. Indeed, this suspicion seems unwarranted; there is widespread appreciation for realism among economists – at least those who reject Friedman’s extreme position. Even Blaug (see Nell, 1998, Chs. 3 and 4) refers with approval to realism, for example in his comments on Hicks, who regarded it as central. Yet ‘realism’ can be verified only by fieldwork.

Without fieldwork, our numbers and therefore our statistics will give us a distorted picture of the world. Without fieldwork, we cannot know the operating rules in our economic institutions, or the true motivations of agents.

Mayer (1993) gives the example of time inconsistency theory, in which a game theoretic analysis demonstrates the case for a rule-based rather than a discretionary monetary policy. In this approach, the central bank is assumed to generate inflation in order to trick agents into overestimating their real wages and therefore work effort. As Mayer points out (ibid., pp.64–5), the statistical evidence suggests strongly that Fed policy has been anti-inflationary during most of its existence. The only exceptions were during wartime. This could be supported even more strongly by reading the records of meetings of the board of governors and the open market committee. Further, even if the Fed had an inflationary bias, the reason for this bias might be quite different than that assumed by time inconsistency theory. That theory rests on an attribution of intentions to an institution, the Fed, an attribution made without considering the available evidence, or doing the fieldwork necessary to gather and evaluate new or better evidence.

A different but even more extreme case is provided by Lucas’s (in)famous claim that:

‘involuntary unemployment is not a fact or phenomenon which it is the task of theorists to explain. It is a theoretical construct which Keynes introduced in the hope that it would be helpful in discovering a correct explanation for a genuine phenomenon: large-scale fluctuations in measured, total employment’ (Lucas, 1987, p. 354; see also the commentary in Rosenberg, 1992, pp. 77–8).

Even minimal fieldwork will establish that ‘involuntary unemployment’, in the normal sense of the term, is a fact, and, moreover, one in need of explanation. Further (historical) fieldwork will show that the character of employment in leading industrial countries changed from before 1914 to after 1945. The legal, regulatory and institutional arrangements changed.

The books by Alan Blinder (1998) and Truman Bewley (1999) are good illustrations of smart fieldwork in economics that Nell has advocated since the publication of his (1998) book. Blinder of Princeton University and his graduate students visited 200 American companies to find out why managers are slow to raise and lower prices. However, Bewley’s (1999) study grew from small beginnings. Seeking inspiration for theoretical models of wage rigidity, in 1992 he arranged a few interviews with businesspeople.

Susan Helper (2000) thinks that fieldwork allows exploration of areas with little pre-existing data or theory. Indeed, she wrote:

I started my dissertation research thinking I would look at automakers’ make/buy decisions. But when I started interviewing and reading trade journals, I realized that important changes – not reflected in the existing literature – were occurring on the ‘buy’ side. US automakers were moving from adversarial deals to ‘voice’ relationships in which they worked with suppliers to improve performance. (See

Helper (2000) observes that, because of fears about the unreliability of field methods, some economists get ideas from the field but do not discuss their fieldwork in their published articles. But understanding the setting can help explain differences in findings between cases by making clear the mechanism by which variables are linked.

Furthermore, Udry (2003, p. 1) noted that development economics has benefited from a rich tradition of field research. Within this broad tradition there is a huge variety of methods, from short qualitative studies to large-scale surveys:

Typically, empirical work in economics relies on existing data. However, it is becoming more common in development economics to complement existing data with relatively short, often less structured visits to the field site in order to clarify aspects of the data, to better define the economic environment, or to collect limited amounts of complementary data.

The NBER Project on Industrial Technology and Productivity was begun in 1994 with funding from the Alfred P. Sloan Foundation. It has three intertwined objectives. First, it seeks to foster research on the fundamental determinants of productivity improvement. Second, it encourages economists studying these issues to supplement their traditional theoretical and empirical research methods with direct observation of business firms and conversations with managers and workers. Finally, the project provides a framework for communication among economists, researchers from other academic disciplines, and policy-makers.

These are recent studies, but recognition of the need for this sort of work goes back a long way, to Jevons, Marshall, and to the founders of econometrics, namely, Frisch, Leontief, Marschak, and Tinbergen.

Conceptual Analysis and Fieldwork

Neoclassical models analyse behaviour in specific ways. Instead of drawing on fieldwork to define motivation and set the problems of choice in well-described institutional context, agents are considered abstractly and presumed to be rational and to choose freely. This, then, leads to models that exhibit a particular kind of market behaviour, which we can call a ‘stimulus-response’ pattern.

Nell and Errouaki (2013, Ch.1) argued that these models are strongly behavioural, paying little attention to structure. The context of action is abstract; the questions concern what an agent, usually a ‘household’ or a ‘firm’, would normally do, acting under the influence of an assumed motivation and calculating rationally, when presented with various stimuli. It is assumed that the actions in response to stimuli are successful – a harmless assumption when it is households making purchases, but question-begging when it is investors introducing a new technology. Given the behavioural assumptions, reaction patterns to such hypothetical stimuli are constructed, and from these sets market functions are aggregated. Equilibrium market positions are then determined by solving the market equations on the hypothesis that behaviour will be adjusted as stimuli move, until the markets are cleared.

The conclusions of a rational choice model have an extraordinary power. They represent what ought to be done in the given conditions – not what should be done morally, but rationally. The model tells us the right, proper, sensible, best thing to do in the circumstances. Agents in the given conditions who do not act in accordance with the model may be considered foolish.

There is an important difference in focus here compared to neoclassical thinking. Both are concerned with intangibles, but the latter’s concern is with states of mind that are properly ascribed to individuals, whereas structural models relate to features of institutions. This calls for a focus on roles, duties, and norms rather than preferences, wants, and desires.

Nell and Errouaki (2013, Ch. 10, p. 358) argued that:

“the methodology of scientific economics adopted the traditional empiricist’s view of the mind as the passive recipient of sense impressions, organized by definitions and analytic truths. Sense data provided the basis of our understanding of the external world, the building blocks out of which the edifice of knowledge was constructed. These were classified and manipulated by means of analytic truths, such as those of mathematics, forming the building blocks into patterns and structures which pictured the world – that is, were isomorphic to it.”

They went on to argue that sense data were passively recorded; the structures were built to conform to external reality – the structure of knowledge, even the logical structure of propositions, mimicked the structure of the world. Knowledge was recorded, it was not created. Nell and Errouaki (2013, Ch. 5) argued that in the picture of the economy sketched by neoclassical theory, the minds of economic agents play no role. They (ibid, Ch. 1) also argued that the formulae follow from the axioms of rationality – the axioms, in turn, are taken as given. This vision of the passive mind, however, is no longer acceptable philosophically. The underlying theory of perception has been shown to be inadequate. In economics, in particular, truths of reason provide us with a map of the relationships between agents and the material world – in economic terms, between rational choice and production.

The argument is twofold. First, conceptual truths provide a basic framework for understanding the structure of human social systems. Such a structure, in turn, provides the setting in which behaviour takes place, a setting that limits and conditions behaviour. Rationality then guides behaviour, but ration­ality works through, and must be understood in terms of, conceptual truths. In economics such truths provide a frame­work, a set of guidelines, telling us how to construct theory and to build models to picture the world adequately. Second, conceptual analysis based on fieldwork will provide the essential assumptions and definitions on which model-building should be based. In order to construct the kinds of models that will enable economists to understand the way the system works, we need to start from conceptual truths, fleshed out by understanding from the inside, and then to develop stylized facts by interpreting statistics in the light of the fieldwork.

In particular, such analysis allows us to understand the relationships between agents, institutions and the material world in an economic system, providing an account of structure. Structure, in turn, is the setting for behaviour; behaviour has to be seen in a context that defines not only opportunities and limitations, but also commitments and expectations. With these in place, the role of rationality for the individual agent can be addressed. One aspect is instrumental: the rational agent seeks to choose the most advantageous option among those available. But another is procedural: the rational agent carries out his or her commitments in the most appropriate way. And finally, rationality can be both critical and imaginative with respect to ends and objectives.

Nell (1998, pp. 96-7) argued that:

“conceptual theorizing must be based on and embody empirical work (here in the sense of fieldwork), which will tell us the identifying characteristics of the objects under study. The common belief that conceptual truths are supposed to make it possible to understand the world by just thinking about it has the true relationship exactly backwards. On the contrary, to do pure thinking, to theorize about the world, it is also necessary to investigate the world.”

Indeed, conceptual analysis of fieldwork can put together the real patterns of behaviour and motivation in the context of the available and actually operating technology, ways of working, making and doing things. Such conceptual analysis may be concerned with ‘deconstruction’, a literary analysis taking apart the reported picture, discovering concealed meanings and hidden agendas, on the part of both the observers and the observed. An important part of this will be uncovering the presuppositions of the concepts and activities reported by fieldwork. Or – the programme of economics – it may accept the picture, and set out to construct models that will show how the system works in various ways, including how it may fail to work and break down.

Conceptual analysis based on fieldwork will provide the essential assumptions and definitions on which model building should be based. In order to construct the kinds of models that will enable economists to understand the way the system works, we need to start from conceptual truths, fleshed out by understanding from the inside, and then to develop stylized facts by interpreting statistics in the light of fieldwork (Nell, 1998, ch.3).


Many economists remain skeptical of qualitative research, fearing that it is not objective, replicable or generalizable. Econometricians ask what the standards for good fieldwork are, saying that, in econometrics, they know to look for identification and specification issues, but what are the analogues in fieldwork? How is it different from journalism? The trouble is, they have not read the literature on fieldwork. Furthermore, there is a tendency to think that while econometrics requires years of training, fieldwork research is easy. It’s not. It’s just as important to pay as much attention to careful research design and sample selection as to quantitative research.

Nobel laureates Haavelmo (1958, 1989), Stone (1978) and Klein (1982, 1987), and leading British econometrician Johnston (1963 [1984]) hinted implicitly at the relevance of the fieldwork approach in econometrics. An econometrician coming cold to a study would run the risk of very slow progress with much searching through inappropriate formulations. The aforementioned authors emphasized the importance of knowledge of the institutional realities, and suggested that developing institutional realities (obtained through fieldwork) into well-grounded formulations of economic relationships and refinements of basic data sets would contribute much more to the improvement of empirical results than more elaborate methods of statistical inference.

These can be further developed on the basis of published statistics (adjusted in the light of information uncovered in fieldwork), and the models can be tested, revised, and so forth. Verification and falsification have a place here; not a privileged place, but a role to play nevertheless. They are not decisive, but they are useful (see Nell, 1998, part II). Many fieldwork insights can be translated into the language of econometrics or theory. It is possible that economists using only those methods could have generated the same insights, but in fact, they didn’t. Fieldwork offers a new source of inspiration, one that is complementary to more conventional methods.

To claim that there can be a priori knowledge of the world does not imply that we can sit in our armchairs and figure out the ways African markets differ from those in Latin America. Such specific matters are never a priori. Truths of reason provide direction to research; they tell us where to look and what kinds of things to look for. They tell us about the shape of the world; they don’t give us facts – they outline the possibilities and the limits. A priori knowledge of the world requires examining the world, too. Just because knowledge is a priori, does not mean that anyone has privileged access to it, or that the conclusions cannot be criticized, disputed, or revised.

Nell and Errouaki (2013, Ch. 10) argued that all three levels – conceptual analysis, fieldwork and model-building – interact. Each can help to extend and develop the others. No single criterion governs all. Each draws on precepts and practical maxims peculiar to itself, but each provides assistance to the others, and in some measure each is necessary to the others. By using fieldwork in conjunction with conceptual analysis in economic model building, we hope to avoid what Friedman (1991, p. 36) expressed elegantly when he observed that “the use of mathematics and econometrics in economics had progressed beyond diminishing returns to ‘vanishing returns’.”



Andrews, P.W.S (1949), Manufacturing Business, London: Macmillan

Bewley, T. (1999) Why Wages Don’t Fall during a Recession?, Cambridge, MA: Harvard University Press

Blinder, A.S. (1998), Asking About Prices: A New Approach to Understanding Price Stickiness, Russell Sage Foundation

Commons, R.J. (1968), Legal Foundations of Capitalism, Madison: University of Wisconsin

Edwards, R. (1979), Contested Terrain: The Transformation of the Workplace in the Twentieth Century. New York: Basic Books

Florence, P.S. (1972), The Logic of British and American Industry, London: Routledge and Kegan Paul

Friedman, M. (1991), ‘Old Wine in New Bottles’, Economic Journal, 101 (404), 33-40

Haavelmo, T. (1957), ‘Econometric Analysis of the Savings Survey Data’, Bulletin of the Oxford University Institute of Statistics 19, 145-149

Haavelmo, T. (1958), ‘The Role of the Econometrician in the Advancement of Economic Theory’, Econometrica 26, 351-35

Haavelmo, T. (1989), ‘Econometrics and the Welfare State’, Nobel Price Lecture, reprinted in American Economic Review, 1997, 87(6), 13-15

Helper, S. (2000), ‘Economics and Field Research: You can Observe a Lot Just by Watching’, American Economic Review Papers and Proceedings, 90, 228-32

Johnston, J. (1963/1984), Econometric Methods, New York: McGraw-Hill

Klein, L. R. (1982), ‘Economic Theoretic Restrictions in Econometrics’, in G.C. Chow and P. Corsi (eds.), Evaluation the Reliability of Macroeconomic Models, New York: Willey

Klein, L. R. (1987), Interview with Mariano, R. , ‘The ET Interview: Professor L.R. Klein’, Econometric Theory, 3, 1987, 409-460

Lucas, R.E. (1987), Models of Business Cycles, Oxford : Basil Blackwell

Mayer, T. (1993), Truth vs. Precision. Aldershot: Edward Elgar

McCloskey, D.N. (1985), The Rhetoric of Economics, Madison: University of Wisconsin

Nell, E.J (1996), Making Sense of a Changing Economy, London and New York: Routledge

Nell, E.J (1998), General Theory of Transformational Growth, Cambridge, UK: Cambridge University Press

Nell, E.J and Errouaki, K. (2013), Rational Econometric Man, Cheltenham, UK: E. Elgar

Portes, A., M. Castells and L. Benton (1989), The Informal Economy: Studies in Advanced and Less Developed Countries, Baltimore: John Hopkins University Press

Pratt, J. and R. Schlaifer (1988), ‘On the Interpretation and Observation of Laws’, Journal of Econometrics, 39, 23-52

Rosenberg, A. (1992), Economics: Mathematical Politics or Science of Diminishing Returns, Chicago: University of Chicago Press

Smith, A. (1985), An Inquiry into the Nature and Causes of the Wealth of Nations, New York: Modern Library

Stone, R. (1978), ‘Keynes, Political Arithmetic and Econometrics’, Proceedings of the British Academy, 64, 55-92

Swann, P.G.M. (2008), Putting Econometrics in its Place, Cheltenham, UK: E. Elgar

Udry, Ch. (2003), ‘Fieldwork, Economic Theory and Research on Institutions in Developing Countries’, UM, Yale University

From: pp.6-9 of WEA Commentaries 7(1), February 2017

Download WEA newsletter Volume 7, Issue No. 1, February 2017 ›

1 response

  • Emmanuel Tweneboah Senzu says:

    That is a wonderful essay; for the economic profession to be mimic, stem from the fact of higher concentration of rationalization under mathematical model as a course for beauty sake but irrelevant to address issues in material world per it propounded theories. This never indicate most of the theories are in error but wrong descriptions to behavioral dynamics of agents or specimen in the material world to put life into the accurate functioning of theories. Which a good field work would have answered this problems

Respond to this article

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Please note that your email address will not be published.