Abstract. This paper examines methodological issues in labour economics, by examin- ing the claim that it is moving closer to notions of procedural rationality ...
Methodological Issues in Labour Economics: Procedures for Procedural Rationality?
Gianni Zappala Abstract. This paper examines methodological issues in labour economics, by examining the claim that it is moving closer to notions of procedural rationality advocated by Herbert Simon. A content analysis of the Journal of Labor Economics shows that the orthodox methodological perspective of deductivism has a strong hold in labour economics, and the nature of the empirical work is also allied to this perspective, casting doubts as to whether labour economics has moved towards procedural rationality. The paper then explores the possibilities of a bridge between labour economics and the schools of institutional labour economics and industrial relations. The paper argues that the former needs to borrow in a more constructive way from the latter in order to construct the procedures for procedural rationality.
I. Introduction A recent paper in this journal focussed on the history and theory of labour economics (Brunetta, 1991). This adds to other important contributions such as those by McNulty (1980)and Kaufman (1988).Less attention has been paid, however, to methodological issues in labour economics1 This is in contrast to a plethora of work examining and questioning the methodology and epistemology of economics in general,2and in specific areas such as econometrics (Granger, 1990; de Marchi and Gilbert, 1989; Darnel1 and Evans, 1990),macroeconomics (Dow, 1985) and game theory (Binmore, 1987; 1988; Hollis, 1990;
Gianni Zappala, Department of Applied Economics, University of Cambridge.
I thank Gay Meeks, Paul Ryan, Alessandra D’Acconti and Keith Whitfield for useful comments. The usual disclaimer applies. Received on August 4 and approved by the Editorial Board on December 1,1992. LABOUR 7 (1)209-231 (1993)
JEL B41
0 Fondazione Giacomo Brodolini 1993. Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 I JF, UK and 238 Main Streel,Cambridge, MA02142. USA.
210
Gianni Zappala
Sugden, 1991). This paper begins to fill this gap by providing a review of methodological issues in labour economics. Section I1 discusses why a focus on methodological issues in labour economics is needed. Section III summarises the main consequences of the adoption of the neoclassical paradigm for labour economics. Section IV discusses the methodological implications of this adoption by presenting the results of a content analysis of the Journal of Labor Economics. Section V discusses the nature of recent empirical work in labour economics. Section VI compares developments in labour economics with trends in institutional labour economics and industrial relations. The final section offers some suggestions for a bridge between these two fields and labour economics. 11. Why a Focus on Methodology in Labour Economics?
The lack of discussion of methodological issues in labour economics is surprising for several reasons. First, there is “labour economics ex~eptionalism”.~ Labour economics is often argued to be an exception to the issue being discussed, that it needs to differ methodologically from other fields in economics. However, the reasons for this difference, and what the appropriate methodology should be, is rarely pursued further. For instance, in a recent special issue of the Economic Journal on the future of economics, Baumol argued that the use of mathematics in economics was not appropriate to all fields in economics, “especially labour economics” (Baumol. 1991: 2). Similar examples of this exceptionalism can be found as early as Marshall (1946)and in Blaug (1980)and Berger (1989). Second, there are a number of recent criticisms of the theory and methodology of labour economics from eminent orthodox economists.4 For instance, Nobel laureate Robert Solow recently (1990) noted the inability of orthodox economic theory to deal with the reality of relative wage inflexibility in a world of high unemployment. Blaming the use of abstraction and unrealistic assumptions, he appealed to the importance of “common sense” in trying to understand the labour market. He urged economists to be more sensitive to the findings of their industrial relations colleagues. Does this appeal represent a plea for greater empiricism and induction in labour economics? Third, labour economists have recently become enamoured of the idea that models of the labour market should conform to the “stylized facts” (Oswald, 1987).5 What is meant by a stylized fact, however, is never made clear. Are they empirical descriptions based on observation? Are they assumptions based on what theory would predict? Are 0 Fondazione Giacomo Brodolini 1993.
21 1
Methodological Issues in Labour Economics
they typical? This concern with ensuring compatibility between theoretical models and reality by appeals to stylized facts may be a reflection of a deeper methodological malaise. Finally, in a recent collection of papers on the perceived methodological crisis in economics (Wiles and Routh, 1984), it was claimed that labour economics had been moving towards Herbert Simon’s notion of procedural rationality. Given the current debates and interest in notions of rationality in economics (Sugden, 1991; Meeks, 1991),the claim that labour economics has moved towards Simon’s notion of procedural rationality warrants closer examination, and will provide our framework for examining methodological trends in labour economics. 111. Labour Economics and the Embracing of Neoclassical Theory
The rise to dominance of neoclassical theory in labour economics has been well documented (Kerr, 1977; McNulty, 1980; Kaufman, 1988; Jacoby, 1990).hThis rise has not only had an impact upon the manner in which labour markets are conceptualised, but has also had important methodological consequences (Kerr, 1983). The main implications of the neoclassical resurgence in labour economics are therefore summarised before examining the methodological implications. The first consequence is methodological individualism. This is the principle that “all actions are performed by individuals”; that “a social collective has no existence and reality outside of the individual members’ actions” (von Mises (1949) quoted in Hodgson, 1986). The implications of methodological individualism are perhaps most extreme for labour economics (Boland, 1982: 31). Labour market behaviour is most often conducted by groups, the most obvious of which are trade unions. Yet these groups are considered to be no more than the sum of their individual members. Methodological individualists need to explain the status and behaviour of groups in terms of statements about interacting individuals and their preferences. The problem is that “while individual workers often have the freedom to break with groups they belong to, group membership itself is often not chosen as a means to some other end” (Marsden, 1984: 123).’ In other words, methodological individualism, when applied to the labour market, abstracts from the various cultural, historical and social factors which are important in explaining much workplace group behaviour. The second consequence is the a priori assumption of perfect competition in model construction. If “prevailing theories of the labour market are framed in terms of small scale private sector employers ... 0 Fvndazione Giacomo Brodvlini 19Y3.
212
Gianni Zappala
from a methodological point of view, the causes of the development and functions of labour market institutions tend to be sought in the working or failure of the competitive labour market rather than elsewhere” (Marsden, 1984: 121). He goes on to argue that the competitive labour market is only one form of labour market, and by no means the most typical. Of more importance is the existence of internal and occupational labour markets. A model of the competitive labour market may be useful for analysing some sectors of the labour market, but is by no means applicable to the whole economy. Third, and perhaps most important given the debate surrounding it,8 is the adoption of the orthodox notion of rationality in the context of labour market behaviour, defined as: “Given the set of available actions, the agent chooses rationally if there is no other action available to him the consequence of which he prefers to that of the chosen action” (Hahn and Hollis, 1979: 4). This concept of rationality has two main problems with respect to the labour market. First, at the level of the individual, the motivation of utility maximisation, and the global and objective cognitive abilities of agents, have not been borne out by industrial relations and social psychological studies of human behaviour at work (Kaufman, 1989). Second, following from the tenet of methodological individualism, it is assumed that the behaviour of collective groups is subject to the same constraints which would normally apply to individual utility maximisers (Arrow, 1987). Yet if group rationality differs from individual rationality, as has been shown by industrial relations research (Brown, 1973; Kerr, 1983; Marsden, 1984),the initial set of constraints and behaviour assumed under orthodox notions of rationality no longer apply.’ It has, however, been claimed by Marsden that the orthodox notion of rationality has been on the wane in labour economics, which “in recent years has experienced a rapid growth of interest in what Simon ( 1976) has described as ‘procedural rationality’ as opposed to ‘substantive rationality”’ (Marsden, 1984: 148). He was referring to certain theoretical developments in labour economics, such as theories of job search, signalling, discriminatory hiring practices and implicit contracts, arising from problems of imperfect information and uncertainty, which focus on how particular individual behaviour leads to particular institutional and market responses. At a more general level, however, Simon’s notion of procedural rationality requires a more radical change not only in the way economists conceptualise rationality, but in the way they do economics; that is, it requires a change in methodology. By procedural rationality, Simon was referring to the concept of rationality developed in psychology, that 0 Fondazione Giacomo Brodolini 1993.
Melhodological Issues in Labour Economics
213
“behaviour is procedurally rational when it is the outcome of appropriate deliberation” as opposed to the substantive rationality of economics which “is appropriate to the achievement of given goals within the limits imposed by given conditions and constraints” (Simon, 1976: 67-68). He argued that two conditions were necessary as economics moved towards the notion of procedural rationality. First, the “‘psychologizing’ of economics, the explicit adoption of the programme of economic behaviouralism” (p. 84). Second, economics needed to become less deductive and more empirical, since “The shift from theories of substantive rationality to theories of procedural rationality requires a basic shift in scientific style, f.om an emphasis on deductive reasoning within a tight system of axioms to an emphasis on detailed empirical exploration” (p. 84: emphasis added). To what extent has labour economics moved towards procedural rationality, as Marsden claims, by meeting these wider methodological conditions, and to what extent has the orthodox neoclassical concept of rationality retained its hold?
IV. The Methodological Implications: Empiricism and Deduction in Labour Economics Most recent reviews of methodology, and philosophy of science and economics, agree that the dominant methodological perspective in economics is deductivism (Hutchinson, 1984; Hausman, 1989; Gerrard, 1990).Deductivism can be defined as “the belief that the path to truth and certainty is provided by the application of the logical method of deduction to a set of axioms that are claimed to be true a priori” (Gerrard, 1990: 198). The consequence of deductivism for economics has been an attachment to reductionism, rationality and equilibrium (Loasby, 1989). Furthermore, it has meant the use of mathematics as the main language of economists, and the elevation of theoretical endeavours at the expense of empirical investigation. These consequences were demonstrated, for instance, by the reviews of Leontief ( 1982)and Morgan ( 1988). Reviewing papers published in the American Economic Review (AER)between 1972 and 1981, Leontief found that the majority could be grouped as “mathematical models without any data” (see Table 1). Updating Leontief ’s work, Morgan continued the analysis for the AER to 1986, and also used the same framework to compare papers published in the British Economic Journal (EJ).He found that there had been a decline from 50 to 42 per cent in the number of “mathematical Q Fondazione Giacorno Brodoiini 1YY3.
2 14
Gianni Zappala
Table 1. Articles Published in the American Economic Review Leontief data Mar. 1972 to Dec. ’76 (“A)
Theory - models, analysis, methodology Mathematical models without any data Analysis without any mathematial formulation and without data Statistical methodology Subtotal Empirical analysis Based on data generated by author’s initiative Based on data published or generated elsewhere Not using statistical inference Based on artificial simulations and experiments Subtotal Total
New data
Mar. 1977 Mar. 1982 to Dec. ’81 to Dee. ’86 (%)
(%)
50 21
54 12
42 7
1 -
1 -
72
66
1 50
1
1
5
21
23
39
5 1
7 2
1 6
34 100
51 100
28 100
models without any data” in the AER (the theory category as a whole declining from 72 to 50 per cent), with a subsequent increase in empirical analysis from 28 to 51 per cent. The bulk of the increase in this empirical work, however, was based on data published or generated elsewhere. Turning to the EJ, although Morgan found that the theory/empirical divide had remained constant (see Table 2), there had been dramatic change within the theory group, with “mathematical models without any data” rising from 34 to 52 per cent, with “analysis without mathematics or data” falling from 24 to 7 per cent. That is, the deductive approach had strengthened, at least as reflected by the EJ. Labour economics was previously a stronghold of institutional economics (McNulty, 1980; Ramstad, 1981), with a preference for inductive empiricism. For instance, taking a random sample of the American labour economics journal literature between 1915 and 1970, Ramstad estimated the contributions of institutional economists. The results varied according to the definition of institutionalist, but they Q Fondazionr Giacomo Brodolini 1993.
215
Methodological Issues in Labour Economics Table 2. Articles Published in the Economic Journal
Theory - models, analysis, methodology Mathematical models without any data Analysis without any mathematial formulation and without data Statistical methodology Subtotal Empirical analysis Based on data generated by author's initiative Based on data published or generated elsewhere Not using statistical inference Based on artificial simulations and experiments Subtotal Total
Mar. 1972 to Dec. '76
Mar. 1977 to Dee. '81
Mar. 1982 to Dec. '86
(W
(W
(%)
34 24
50 8
52
1
0
0 -
59
57
58
1
3
2
39
37
38
0 1
1 1
2
41 100
43 100
7
1
42 100
Source: Morgan(l988: 160-161)
began from a low of 41 per cent to a high of 60 per cent of the total (Ramstad, 1981). To investigate changes in labour economics methodology, I followed the approach of Leontief (1982) and Morgan (1988) in conducting a similar review of papers published in the American Journal of Labor Economics (JLE)and the results are presented in Table 3. The JLE was chosen as the best representative of scholarly mainstream work in labour economics.") Although a division based solely on an inductive or deductive approach may be considered philosophically simplistic, it was used to make possible comparisons with previous work.' As can be seen from Table 3, for the sub-discipline still considered the most applied within economics (Freeman, 1989), 40 per cent of articles published in its specialist journal were purely theoretical. Furthermore, the bulk of the theory category was composed of "mathematical models without any data", further reflecting the deductivist nature of research. As reflected by these data, labour economics is firmly rooted methodologically within the deductivist camp. If these data are a reflection of broader trends in labour economics, it would Q Fondazione Giacomo Brodolini 1YY3.
216
Gianni Zappala
Table 3. Articles published in the Journal of Labor Economics (January 1983-July 1991)* Theory - models, analysis, methodology Mathematical models without any data Analysis without any mathematical formulation and without any data Statistical methodology Subtotal Empirical analysis Based on data generated by author's initiative Based on data published or generated elsewhere Not using statistical inference Based on artificial simulations and experiments Subtotal Total
33 2 5
40 2 58
0 0** 60 100
'All data are rounded to the nearest percentage point. July 1991 was the latest issue when the content anaysis was conducted. Articles were only classified once; they were not divided between claws. This did not pose a problem for the pure theory papers, but there were many papers which were predominantly theoretical in nature, but had a small empirical section testing some of the propositions discussed. Although many of these would sit comfortably in the theory class, they were classed under empirical analysis in order to try to avoid subjective orderings of how much empirical there needed to be, and possible criticisms that the theory category had been artificially inflated. Therefore the 40% of articles for the theory category should be seen as a minimum, since a less strict classification (see Morgan, 1988: 160)would have resulted in a higher theory category. **There was one article based on an experiment, giving a percentage of 0.03: given the rounding of figures, however, this category appears empty.
appear that the first of Simon's criteria for a move towards procedural rationality has not been met.'* V. The Methodological Implications: The Nature of Empirical Work
The second of Simon's criteria was for economics to become more empirical and borrow from the work of cognitive psychologists. It is therefore apposite to comment upon the nature of the empirical work in labour economic^.'^ This empirical work is allied to the orthodox deductivist position, and this forces it to use notions of rationality that are non-behavioural rather than behavioural (Simon, 1957; 1976). Econometrics Concomitant with the rise of neoclassical theory in labour economics was the acceptance and use of econometrics as the dominant method to 0 Fondazione Giacorno Brodolini 1993.
Methodological Issues in Laboitr Economics
217
the exclusion of the more traditional techniques of the institutionalists such as participant observation, case studies, and researcher-devised questionnaires (Piore, 1979; Kerr, 1983). Noting this change in labour economics, a recent review of labour economics textbooks stated: “Labour economics has been transformed ... into a field which can be as technically demanding as any in the parent discipline and has even become a net exporter of econometric technology” (Drewes, 1989: 303). Is is interesting to note, for example, that five per cent of articles in the JLE were concerned with theoretical issues in statistical methodology, compared to only one per cent for the AER in each of the three time periods, and one and zero for the EJ. The greater use of econometrics in itself does not necessarily work against the notion of procedural rationality. Nor is the intention of this paper to provide a critique of econometric^.'^ Indeed, there may seem to be no necessary inconsistency between the greater use of econometrics and Simon’s call for greater empirical reality within economics. The use of econometrics in labour economics, however, is linked to the deductivist program, in that labour economists see its role as providing a test of the various theories derived from the various neoclassical axioms, hence “the process of econometrics is, therefore, a particular blend of inferential and deductive processes” (Darnell and Evans, 1990: 21). Econometrics is seen as the most appropriate way of falsifying economic theories (Lawson, 1985). Empirical research in labour economics therefore rests almost solely on econometric evidence as a means of testing the predictions or hypotheses of models derived from prior logical axioms.” This view of empirical work is consistent with the positivist view of knowledge, and requires an acceptance, if implicit, of the deductive approach to economics (Keat and Urry, 1975: 12). Although testing of theories may seem consistent with Simon’s call for empirical reality, the “unpersuasiveness” of econometric evidence in theory testing and the failure of the falsificationist program (Caldwell, 1982; Summers, 1989; Darnel1 and Evans, 1990), has acted as an immunising strategy for neoclassical economics, and its notions of rationality, rather than providing a more solid empirical base (Blaug, 1980, 1984; Summers, 1989). Secondary data Related to the growing use of econometrics is the increasing use of secondary data in the empirical work of labour economists. Only two per cent of papers were based on data generated by the author’s initiative. The majority of papers with empirical analysis in the JLE relied on government census and labour force surveys.I6 Of the 60 per cent of 0 Fondazione Giacomo Brodolini 1993
218
Gianni Zappala
papers classified as empirical analysis, there were no papers in the category “not using statistical inference” or based on “artificial simulations and experiments” (see ** for Table 3). Simon’s notion of procedural rationality was to be formulated partly by observing behaviour in experimental situations.” Although there has been some growth in the use of experiments in economics (Roth, 1988), there is little evidence of it in labour economics. The reliance on data published or generated elsewhere is similar to the findings for the AER and EJ. Leontief was critical of this reliance in his study as the government statistics mainly relied on in these papers were compiled for administrative and business use rather than research purposes (see also O’Brien, 1992). Although government surveys are often appropriate to labour issues to the extent that they are micro-data sets and highly representative (Oswald, 199 1 ), they often force economists to rely on poor proxies of variables in their econometric estimation (Kerr, 1983; Kaufman, 1989; Jacoby, 1990). This sole reliance on econometrics, combined with the use of secondary data, has led some to criticise empirical work in labour economics for its non-behavioural nature. The assumption in labour economics is that tastes are stable and do not differ systematically. Variations in labour market outcomes are almost always explained as the result of differences in environmental variables. Behavioural research, on the other hand, “emphasizes the independent and quantitatively significant influence that the intervening variables in the psychological field have in shaping and conditioning the direction of human behaviour” (Kaufman, 1989: 73).The non-behavioural nature of orthodox labour economics research is, therefore, in part conceptual, the result of using substantive rationality as a model for labour market actors, and in part methodological, the reliance on secondary data not intended for academic research, which does not collect information on behavioural variables important to using a notion of procedural or bounded rationality (Kaufman, 1989: 83). Summary From this review, the claim that labour economics is meeting the conditions for procedural rationality, at least in the wider methodological sense, cannot be countenanced according to Simon’s own criteria that economics needs to become less deductive and more empirically informed. Labour economics, as reflected by articles published in the JLE, is strongly deductive in orientation, and the nature of the empirical work is not compatible with notions of behavioural research needed for procedural rationality. 8 Fundazione Giacomo Brodolini 1993.
Methodological Issues in Labour Economics
219
VI. Institutional Labour Economics and Industrial Relations Two issues need to be examined with respect to institutional labour economics and industrial relations. First, the nature of the theoretical/ empirical divide in these two fields. Second, the nature of their empirical work; if different to that in mainstream labour economics, does it come closer to Simon’scriteria for procedural rationality? Theory and empiricism in institutional labour economics and industrial relations In contrast to labour economics, both schools fall firmly into the inductivist, as opposed to the deductivist, camp of orthodox economics. The central core of institutionalism is the repudiation of neoclassical economics and its deductivism (Dugger, 1979; Ramstad, 1981; Hodgson, 1988).The public policy concern of most institutionalists led to a call for a stronger factual and empirical orientation in economics. Accordingly, the two main criticisms of orthodox theory were that the predictions diverged from empirical reality, and that these discrepancies arose from a lack of realism in the model’s assumptions (Jacoby, 1990). Jacoby argues, however, that the institutionalists did not discard economic abstraction and the deductive method completely. Rather, the institutional labour economists’ methodology was one of building a wide range of eclectic middle range theories that bridged the abstractions of pure theory and the institutional realities of labour markets.lX However, despite this claim, institutionalism’s approach to the philosophy of science and methodology suffers from not being as clear cut as the positivist paradigm in orthodox economics and has subsequently suffered from the tag of being merely a critique of neoclassical economics (Bush, 1991).19 Industrial relations emerged as a field in the early 20th century due to the perceived failure of neoclassical economics to explain adequately the labour market (Marsden, 1982). Its epistemology was always influenced by empiricism (Marsden, 1982; Godard, 1989).*OIndustrial relations has therefore never been a highly deductive field, but arose out of the criticisms of neoclassical theory with respect to the labour market. More recently, Cappelli has argued that there has been a change in industrial relations research, from “theories and explanations generated inductively and focussed at the organizational level” to “deductive explanations generated from the social sciences and focussed on individual behaviour” (1985: 90). Similar to the orthodox neoclassical influence on the methodology of labour economics, this Q Fondazione Giacomo Brodolini 1993.
220
Gianni Zappala
change in industrial relations is traced by Cappelli to the influence of economics and organizational behaviour, both of which choose the individual as the unit of analysis. Nevertheless, despite recent attempts to mimic the deductivism of economics, it has largely remained wedded to an empiricist approach (Godard, 1989). This can be seen, for example, in the industrial rationalists’ criticisms of economic models of trade union behaviour. Most of the criticisms focus on the unrealism of assumptions and the inconsistency between the predicted behaviour of the models and observed behaviour (Turnbull, 1988; Mayhew and Turnbull, 1989). This attack on realism was rejected by Friedman ( 1953), but his challenge has had little impact on influencing the thinking of industrial relations and institutional labour economics on the importance of realistic assumptions with regard to the labour market (Kerr, 1983).2’ The nature of empirical research in institutional labour economics and industrial relations This differs from that in orthodox labour economics in two ways. In contrast to the orthodox notion of rationality in neoclassical economics, a model of behavioural man (sic)underlies industrial relations and institutional labour economics research (Dugger, 1979; Kaufman, 1989). Emphasis is placed on social/psychological factors, as well as the more traditional environmental variables. Also, in order to obtain data on these social/psychological characteristics, researchers rely mostly on primary data, which allows the inclusion of variables not normally found in secondary data sets (Wilber and Harrison, 1978; Kerr, 1983). Another characteristic of this research is that, although econometrics has been used increasingly,22use is also made of other research methods such as participant observation, case studies, historical sources, questionnaires, and pattern modelling (Dugger, 1979; Wilber and Harrison, 1978; Woodbury, 1987; Wisman and Rozansky, 1991). This is due, in part, to the belief that the problem being investigated determines the method to be used (Hodgson, 1988; Mari May and Sellers, 1988), and in part to the lack of a deductively derived model of behaviour which can be tested. The use of a range of research methods may encourage a focus on explanation, rather than theory testing and prediction (Wilber and Harrison, 1978; Jicks, 1979; Brewer and Hunter, 1989).23 Three factors characterise the industrial relations and institutional labour economics methodological approach to labour market research. First, in contrast to Friedmanites, the realism of theoretical assumptions matters (Kerr, 1983).Second, empiricism is not a dirty word, “knowing 0 Fondazione Giacomo Brodolini I993
Methodological Issues in Labour Economics
221
the facts about labor markets and keeping theory close to empirical reality are key parts of the ILE tradition that should be preserved” (Jacoby, 1990: 337). Third, an interdisciplinary perspective is adopted with regard to both theoretical conceptions of behaviour and research methods (Marsden, 1986), in particular, the borrowing from cognitive psychology of “behavioural” notions of rationality (Jacoby, 1990; Kaufman, 1989).
VII. Is a Bridge Possible?
From the above, it would seem that if labour economics is to adopt a broader notion of rationality that fulfills Simon’s criteria for procedural rationality, it would need to be more informed by the industrial relations and institutional labour economics perspective. Some would claim that this cross-fertilisation is already occurring, in what is known as the new labour economics; “An emerging synthesis of an institutional and descriptive-based industrial relations literature with neo-classical microeconomic theory and modem econometric techniques” (Hirsch and Addison, 1986: 2). Areas where this new methodology is claimed to be achieving success are in the economic modelling of trade unions (Oswald, 1987), efficiency wages and insider-outsider theories of pay determination (Jacoby, 1990; Carruth and Oswald, 1989; Solow, 1990; Brunetta, 1991).It is claimed that these theoretical developments represent a break with neoclassical theory (Jacoby, 1990; Brunetta, 1991); hence it can be argued that they represent a move towards procedural rationality at the conceptual level, in that the focus is not on equilibrium positions as such (substantive rationality), but how and why agents make decisions within a particular context where conditions of perfect knowledge and certainty are relaxed (bounded rationality). It is not universally accepted, however, that these developments constitute a break from the neoclassical paradigm and orthodox conceptions of rationality. For example, it has been shown by post-Keynesians that while theories such as efficiency wages may be non-market clearing, they do not depart from the traditional Walrasian conception of the labour market (Milgate, 1988; Seccareccia, 1991; Mongiovi, 1991).24 The concept of rationality can still be seen, in most cases, to be consistent with orthodox economic theory and a non-behavioural approach to research (Kaufman, 1989: 75).25This scepticism regarding the difference between new labour economics and orthodox approaches is also illustrated by the recent literature on economic models of trade unions. Although some have heralded this work as the best example of this soQ Fondazione Giacomo Brodolini 1993.
222
Gianni Zappala
called bridge between theoretical labour economics, and an empirically based industrial relations literature (Oswald, 1987), a closer examination reveals that the bridge has shaky foundations. To labour economists, the trade union “is like any other economic agent and can be thought of as attempting, subject to certain constraints, to maximise an objective function” (Oswald, 1982: 576). The modelling process therefore proceeds in a logical and deductive manner consistent with neoclassical axioms of rationality. The test of a good model is still consistent with Friedman’s (1953) notion of how well it predicts (Johnson, 1985), and notions of falsifiability, so that “an explanation which is inherently not falsifiable is unsatisfactory” (Oswald, 1987: 34). Indeed, the adherence to the positivist paradigm is seen as a virtue, despite recent developments in the philosophy of science which cast doubt on the validity of this methodological approach (Caldwell, 1982; Hausman, 1989, 1992; Lawson, 1989). So the proponents proudly boast that “the methodology ...is perfectly conventional, and the model of a trade union which emerges should fit in naturally with established economic theory” (Oswald, 1982: 576). Although there are various competing economic models of trade unions, and these models differ in their assumptions of union behaviour and predictions of economic outcomes (Creedy and McDonald, 1991; Kaufman and Martinez-Vazquez, 1990; Manning, 1992),26the modelling process typically follows five stages (Oswald, 1987). First, a theory of union preferences is formulated which allows the construction of a utility function. Second, the constraints on union behaviour are defined. Third, the model is solved mathematically. Fourth, estimating equations for empirical analysis (hypotheses)are generated, and fifth, the model is tested econometrically. This approach is said to differ from the orthodox approach in economics in the second and fifth steps. For instance, in identifying the appropriate constraints on union behaviour and in testing the models, use is made of information gathered by the “inductive empiricists” in industrial relations, and ensuring the models fit the stylized facts (Oswald, 1987: 39). Regarding stylized facts, the claim is that previous models ignore institutional features, for example, “lay-offs by seniority” written in collective agreements. These stylized facts are then used to construct assumptions about union behaviour within these constraints; that is, behaviour is, in a sense, bounded. Yet there is a difference between use of a casual stylized fact to construct somewhat more realistic assumptions within a conventional economic model and a move to a substantially more empirically based theory. Theory built upon a few casual stylized facts, rather than the thorough-going empiricism called for by Simon, is unlikely to escape from his conclusion that “the attempt to predict and prescribe human behaviour by 0 Fondazione Giacomo Brodolini 1993
Methodological Issues in Labour Economics
223
deductive inference from a small set of unchallengeable premises [stylized facts] must fail and has failed” (Simon, 1976: 83). Industrial relations scholars have also been critical of this use of stylized facts without the corresponding understanding of labour market behaviour which has come from detailed empirical investigation. Wiles, for instance, argues that stylized facts tend to be untrue and tailor-made to suit the theories. He adds: “a stylized fact is what the theorist finds it convenient to believe that his favourite empiricist has probably established” (1984: 3 11).*’ The industrial relations criticisms point to the indiscriminate use of stylized facts by economists which can lead to the construction of models based on erroneous assumptions. This can be seen in the belief that the written rules of collective agreements always translate into actual behaviour. This ignores, for example, the notion of custom and practice, crucial to industrial relations research and explanation (Brown, 1973). For example, although some collective agreements may make reference to seniority based lay-offs, the actual practice of redundancies is seldom conducted according to this rule (Brown, 1986; Turnbull, 1988). Similarly, many of these models rely on compulsory unionism, which is not a widespread practice (Mayhew and Turnbull, 1989). Although use of industrial relations empirical data to inform the construction of models ‘and bring assumptions of agent behaviour closer to reality is a welcome step, if it is used without an understanding of the corresponding industrial relations theory and context, “the potential for interdisciplinary research appears limited” (Turnbull, 1988: 114).Even Jacoby, who welcomes this new labour economics, cautions against equating stylized facts with good empirical work: “Often the ‘stylized facts’ that theorists work with are old or faulty ... too often NEO-ILE (and other) economists rush into print theories based on partial or incorrect factual knowledge; only later do the empiricists come to clean up the mess” (1990:338). In other words, the very nature of the empirical work of industrial relations cannot be seen in a vacuum, as a vat full of labour market facts, ready to be plugged into a model. This is falling prey to the discredited notion that facts somehow exist independently of their context. As Carr pointed out: “The facts are really not at all like fish on the fishmonger’s slab. They are like fish swimming out in a vast and sometimes inaccessible ocean; and what the historian catches will depend, partly on chance, but mainly on what part of the ocean he chooses to fish in and what tackle to use - these two factors being, of course, determined by the kind of fish he wants to catch (Carr, 1961: 23). The second criticism centres on the lack of primary data used in most of this work: the use of industrial relations evidence also requires the 8 Fmdazione Giacomo Brodolini 1993.
224
Gianni Zappala
use of research methods other than those presently used in labour economics. As Marsden concluded, “the relativization of the competitive model calls for a relativization of econometric among other research methods” ( 1984: 151).The criticism is not of econometrics per se, but of the use of econometrics, usually with secondary data, as the only valid test of various hypotheses (Kerr, 1983). This approach fits into what is called “naive falsificationism” (Darnel1and Evans, 1990). One alternative may be a stronger cross-fertilisationof ideas with recent methodological debates in econometrics (Granger, 1990).This healthy development in the econometrics literature, on such things as the appropriate modelling procedure, the appropriate role of theory, the merits of a falsificationist approach as opposed to an inductivist one (ibid, esp. pp. 1-29 and 97-121), may make the application of econometric techniques, together with other qualitative and primary data, a fruitful way of approaching procedural rationality at a conceptual and methodological level.28 A second, more innovative, alternative may be the use of a multimethodological approach, which again is more sensitive to theory construction than theory testing (Marsden, 1986; Brewer and Hunter, 1990).Both of these alternatives, however, involve a much greater role for industrial relations than that of simply supplying the stylized facts to be massaged into a conventional economic methodological framework.
VIII. Conclusions What then of the claim that labour economics has moved towards a notion of procedural rationality? Although it can be argued that this may be the case at the level of certain theoretical developments, it was noted that the notion of procedural rationality required changes at the wider methodological level, essentially changes towards a less deductive and more empirically informed economics. At this level, the claim is unconvincing for labour economics. It was seen that labour economics had become more deductive in approach, and the nature of the empirical work was allied to this approach, through the use of econometrics and secondary data. In contrast, an examination of industrial relations and institutional labour economics showed that the preference for inductive theorising, and the use of behavioural variables in empirical work, made them closer to using notions of procedural rationality. If it is agreed that the concept of substantive rationality is largely inappropriate to the labour market, these two traditions can therefore play a role in influencing 0 Fondazione Giacorno Brodolini 1993.
Methodological Issues in Labour Economics
225
future trends in labour economics and in meeting Simon’s criteria. Although there has been some movement in this direction, as illustrated by recent wage and union theories, it is used in a limited and distorting fashion. Labour economists must do more than merely pay lip service to the stylized facts. To continue Carr’s analogy, they must not ignore the sea from which they catch their fish. The procedures for procedural rationality are therefore two. At the micro level of theories, concepts of rationality must be empirically grounded in the work of industrial relations scholars and cognitive psychologists. In short, homo economicus must give way to what Kaufman has called “behavioural man” (Kaufman, 1989). At the wider methodological level, the present dominance of the deductive approach must borrow from the inductivist approach of industrial relations, and its requisite use of more primary data. If, as Oswald recently concluded, it is “time for economic theorists and industrial relations researchers to combine forces” (Oswald, 1987: 39), a bridge must be built to meet the expectations of both sides, one which is truly complementary. Only in this way can a genuine departure from the confines of the substantive rationality approach be made, and the procedures for procedural rationality discovered.
Notes I Although labour economics is made up of a number of heterogeneous schools of thought, the use of the term in this paper refers to mainstream or orthodox approaches to the subject. It is not possible to list all the recent contributions to the philosophy of science and economics. See for instance Redman (1989, 1991); Hausman (1989, 1992); Gerrard (1990); Caldwell (1991); Beed (1991); de Marchi (1988). There are now also two specialist journals dealing with philosophical and methodological issues, Economics and Philosophy (Cambridge University Press) and Methodus (Journal of the International Network for Economic Method). Labour economics exceptionalism, in the sense that the market for labour is not like the market for any inanimate commodity, has long been studied by classical as well as post-Keynesian economists: see Seccareccia (1991) for a recent review. The sense in which it is used here refers specifically to methodological concerns, although the two are related. It is important to distinguish these orthodox criticisms from the often stated methodological concerns of the institutionalists (Kerr, 1983; Kaufman, 1988; Jacoby, 1990). This appeal to the stylized facts is particularly prominent in the economics of unions literature. In a recent paper, a subsection of a few paragraphs entitled “Stylized Facts” consisted of the citation of a few case studies, or what are referred to as “casual studies”,
0 Fondazione Giacomo Brodolini 1993.
226
Gianni Zappala
which seemed to be consistent with the general argument. The rest of the paper was then devoted to the construction of a mathematical model based on deductive and logical premises (Clark, 1990). Kerr attributes the "new wave of scientific research in labor economics" to H. Gregg Lewis and the "Chicago School" ( 1983: 308). On the problems of methodological individualism in the economic analysis of trade unions, see also Turnbull, 1988; Mayhew and Turnbull, 1989. "Recent debates have focused on whether the orthodox concept of rationality is valid as well as how to make such a concept operational (see Hahn and Hollis, 1979; Sugden, 1991; and entries by Simon, Arrow and Sen in Eatwell er al., 1990). 'The best example of non-individual rational action in the labour market is informal rules or custom and practice: see Brown (1973). "'The JLE is published by The University of Chicago Press in conjunction with the Economics Research Center of NORC. Although relatively recent (the first issue was January 1983),it was established as an outlet for new ideas and the best papers in labour economics (editorial introduction to the first issue). I I This distinction is limited to two basic approaches considered acceptable by empiricist philosophers of science, deduction and induction. Recent realist approaches in the philosophy of science and economics, however, are critical of this distinction, arguing for the use of abduction or retroduction: see Lawson ( 1 989). '?The deductive orientation can also be seen from recent textbooks such as Fallon and Verry (1988) and field surveys such as Sapsford and Tzannatos (1990). I3Although a detailed classification of the empirical work was not conducted, for example, by issue or theory being examined, comments are based on random samples of articles and recent secondary sources analysing empirical work in labour economics: see references in the bibliography. l 4 Recent critiques of econometrics have come from within the discipline; for example, see Granger ( 1990);de Marchi and Gilbert ( 1989);Hoover ( 1990),as well as from other perspectives; see Summers (1991);Mirowski and Sklivas (1991);Lawson (1993). l JFor example, two leading economists writing on inter-industry wage differentials stated: "The competitive model ... has all the attributes of good theory. It offers clear predictions as to how firms and workers will behave given the constraints they face. ... The theory is specific enough to make falsifiable predictions while at the same time general enough to be applicable in a wide variety of settings ... competitive explanations of economic phenomana ... are to be preferred both on the grounds of simplicity, and because of the discipline the competitive model requires" (Krueger and Summers, 1987: 37; emphasis added). IhMost papers using empirical analysis in the JLE relied upon the U.S. Bureau of Census, Current Population Survey (CPS). For example, Simon recently concluded that "what distinguishes contemporary theories of bounded rationality from ad hoc and casual departures from the SEU model is that the former insist that models of human rationality must be derived from detailed and systematic empirical study of human decision-making behaviour in laboratory and real-world situations" (1987: 18). I" Jacoby (1990) argues that what he terms the new efficiency oriented institutional labour economics (NEO-ILE) has embraced a more deductivist methodology. With its roots in the ILE approach, the NEO-ILE draws more heavily on neoclassical theory, and is more rigorously analytical than ILE. Jacoby claims that NEO-ILEs have influenced mainstream labour economics and forced them to look at uncertainty, trust, behavioural realism, socialization, customs and norms. 0 Fondazionc: Gidcomo Brodolini I993
Methodological Issues in Labour Economics
227
‘’IAlthough there have been recent attempts to state the philosophical and theoretical basis of institutional economies more explicitly (Bush, 1991; Mari May and Sellers, 1988; Wisman and Rozansky, 1991), it is difficult to refute this criticism when papers conclude with statements like “there is no core of theory from which empirically testable hypotheses can be logically deduced ... what ties the school together is both a presumption of a high degree of ontological relativism and a set of preconceptions” (Wisman and Rozansky, 199 1 : 720). ’“This empiricism is mainly due to the influence of the founders of the discipline, Sidney and Beatrice Webb: see, for example, Webb (1975). ’I For instance, the following reads as though Friedman’s critique did not exist: “If the theoretical models that yield ‘reasonably’ accurate predictions make simplistic and naive assumptions about human beings (relative to what is known about human behaviour), then the models should be suspect” (Cummings, 1989: 96). It may be the case that naive assumptions should be replaced, but Friedman’s position needs to be acknowledged where models do predict well. However, the models often do not d o so, and prediction is not the only possible aim; for instance, realism may be more important for explanation (see Boland, 1982: chap. 9 for a more detailed discussion). ? ? Whitfield (1991), for example, found that 67 per cent of articles published in The Journal of Industrial Relations, British Journal of Industrial Relations and Industrial Relations used some form of univariate and multivariate statistical methods. ’7 It is somewhat ironical that much of the new labour economics has focused on testing theories of the labour market developed by the old institutional labour economists, (see Freeman, 1989).Wisman and Rozansky ( 1991) also note that “One of the comforting aspects of a highly developed body of theoretics is that once it is mastered, research is relatively easy: the theoretics provide a highly detailed map of the subject domain. For the researcher, once a problem is identified, the challenge is to generate or deduce hypotheses from these theoretics and check them through empirical testing”; for institutional economists, however, “there are no detailed templates of expected regularities that can be superimposed on a particular study. Researchers must ... immerse themselves in the specific case, searching for the patterns or regularities that are present therein”. For example, Mongiovi sees efficiency wage theory as “essentially marginalist optimizing calculus” ( 199 1: 34). Seccareccia ( 1 99 1 ) sees efficiency wage and insideroutsider theories as neoclassical, but introducing a number of sociological assumptions within a choice theoretic framework. It is interesting to note that although Jacoby (1990) sees aspects of the NEO-ILE as progressive, he argues it has much further to go, and needs to borrow more from the old ILE approach. 2s For instance, even Jacoby notes that although “cognitive and social psychology offer a rich body of empirical research that questions the strict rationality assumptions of the orthodox model ... only a bit of it trickled back into labor economics” (1990: 339). For example, some models include separate utility functions for the union leadership; some assume a homogeneous membership. Accordingly, the various models use different stylized facts consistent with their assumptions, such as what unions bargain over, whether layoffs are random or by seniority, the closed shop, union democracy, the nature of collective agreements, and so on. Irrespective of the theoretical assumptions of the models, however, the procedure is methodologically identical. For a more positive view of the role played by stylized facts, see Summers ( 199 1 ) and Lawson ( 1989). ?“ For example, a recent test of efficiency wage theory overcame many methodological problems of previous studies, by the authors collecting their own data at the firm level, for use with econometric techniques (Cappelli and Chauvin, 1991). ?‘I
’’
0 Fonddzione Giacomo Brodolini 1993.
228
Gianni Zappala
References Arrow K. J. ( 1987) “Economic Theory and the Hypothesis of Rationality”, in Eatwell J., Milgate M. and Newman P. (eds.) The New Palgrave Utility and Probability, Macmitlan, London, 1990. Baumol W. J. (1991) “Towards a Newer Economics: The Future Lies Ahead!” Economic Journal, 101 ( 4 0 4 1-8). Beed C. (1991) “Philosophy of Science and Contemporary Economics: An Overview”, Journal of Post Keynesian Economics, 13 (4):459-494). Berger L. A. (1989) “Economics and Hermeneutics”, Economics and Philosophy (5: 209-233). Binmore K . ( 1 987) “Modelling Rational Players”, Economics and Philosophy, 3 (1: 179-244). Binmore K. ( 1988) “Modelling Rational Players: Part II”, Economics and Philosophy, 4 (1: 9-68). Blaug M. (1980) The Methodology of Economics, Cambridge University Press, Cambridge. Blaug M. (1984) “Comment”, in Wiles P. and Routh G. (eds.) Economics in Disarray, Basil Blackwell, Oxford. Boland L. A. (1982) The Foundations of Economic Method, George Allen & Unwin, London. Boulding K. (1973) “Toward the Development of a Cultural Economics”, in Schneider L. and Bonjean C. (eds.) The Idea of Culture in the Social Sciences, Cambridge University Press, Cambridge. Brewer J. and Hunter A. ( 1 989) Multirnethod Research, A Synthesis of Styles, Sage, London. Brown W. (1973)Piecework Bargaining, Heinemann, London. Brown W. (1986)‘The Changing Role of Trade Unions in the Management of Labour”, British Journal of Industrial Relations, 24 (2: 16 1- 168). Brunetta R. (1991)“Labour Economics: History and Theory”, LABOUR, 5 (1: 75-100). Bush P. D. (1991) “Reflections on the ’henty-Fifth Anniversary of AFEE: Philosophical and Methodological Issues in Institutional Economics”, Journal of Economic Issues, 25 (2:321-346). Caldwell B. (1982)Beyond Positivism: Economic Methodology in the Twentieth Century, Unwin Hyman, London. Caldwell B. J. ( 1991 ) “Clarifying Popper”, Journal of Economic Literature (29: 1-33). Cappelli P. and Chauvin K. (1991) “An Interplant Test of the Efficiency Wage Hypothesis”, QuarterlyJournal of Economics, 106 (3:769-788). Cappelli P. (1985) “Theory Construction in IR and Some Implications for Research, Industrial Relations, 24 ( 1 : 90- 1 12). Carr E. H. (1961/1985) What is History? Penguin Books, Middlesex. Carruth A. A. and Oswald A. J. (1989) Pay Determination and Industrial Prosperity, Clarendon Press, Oxford. Clark A. (1990) “Efficient Bargains and the McDonald-Solow Conjecture”, Journal of Labor Economics, 8 (4: 502-528). Creedy J. and McDonald I. M. (1991)“Models of Trade Union Behaviour: A Synthesis”, The Economic Record, 67 (199: 346-359). Cummings L. L. (1989)“The Essential Nature of Models of Humans in Industrial Relations Research, Industrial and Labor Relations Review, 43 (1: 93-96). 0 Fondazione Giacomo Brodolini 1993
Methodological Issues in Labour Economics
229
Darnell A. C. and Evans J. L. (1990) The Limits of Econometrics, Edward Elgar, Aldershot. de Marchi N. and Gilbert C. (1989) History and Methodology of Econometrics, Clarendon Press, Oxford. de Marchi N. (ed.) (1988) The Popperian Legacy in Economics, Cambridge University Press, Cambridge. Dow S. C. (1985) Macroeconomic Thought:a MethodologicalApproach, Basil Blackwell, Oxford. Drewes T. (1989) “The Economics of Labour Markets: A Review of Current Texts”, Bulletin of Economic Research, 41 (4: 303-3 15). Dugger W. M. (1979) “Methodological Differences between Institutional and Neoclassical Economics”, Journal of Economic Issues, 13 (4: 899-909). Eichner A. S. (1979) “An Anthropogenic Approach to Labor Economics”, Eastern Economic Journal 5 (3: 349-366). Fallon P. and Verry D. (1988) The Economics of Labour Markets, Philip Allan, Oxford. Freeman R. B. (1989) Labor Markets in Action Essays in Empirical Economics, Harvester Wheatsheaf, New York. Friedman M. (1953) “The Methodology of Positive Economics”, in Friedman M. (ed.) Essays in Positive Economics, University of Chicago Press, Chicago. Gerrard B. (1990) “On Matters Methodological in Economics”, Journal of Economic Surveys, 4 (2: 197-2 19). Godard J. (1989) “Beyond Empiricism: Alternative Philosophies of Science and the Study of Industrial Relations”, Queens Papers in Industrial Relations, Queen’s University at Kingston. Granger C. W. J. (1990) Modelling Economic Series, Clarendon Press, Oxford. Hahn F. and Hollis M. (1979)“Introduction”, in Hahn F. and Hollis M. (eds.) Philosophy and Economic Theory,Oxford University Press, Oxford. Hausman D. M. (1989) “Economic Methodology in a Nutshell”, Journal of Economic Perspectives, 3 (2: 115-127). Hausman D. M. (1992) The Inexact and Separate Science of Economics, Cambridge University Press, Cambridge. Hirsch B. T. and Addison J. T. (1986) The Economic Amlysis of Unions: New Approaches and Evidence, Allen & Unwin, Boston. Hodgson G. ( 1986) “Behind Methodological Individualism”, Cambridge Journal of Economics( 10: 21 1-224). Hodgson G. M. (1988) Economics and Institutions,Polity Press, Oxford. Hollis M. (1990)“Moves and Motives in the Games We Play”, Analysis (50: 49-62). Hoover K. D. (1990) “The Logic of Causal Inference: Econometrics and the Conditional Analysis of Causation”, Economics and Philosophy, 6 (2: 207-234). Hutchinson T. (1984) “Our Methodological Crisis”, in Wiles P. and Routh G. (eds.) Economics in Disarray, Basil Blackwell, Oxford. Jacoby S. M. (1990) “The New Institutionalism: What Can It Learn from the Old?” IndustriaIRelations, 29 (2: 316-359). Jick T. ( 1 979) “Mixing Qualitative and Quantitative Methods: Triangulation in Action”, in Van Maanen J. (ed.) Qualitative Methodology, Sage, California. Johnson G. E. (1985) “Comment on A. Oswald, ‘The Economic Theory of Trade Unions - An Introductory Survey”’, Scandinavian Journal of Economics (87: 194-196). Kaufman B. E. (ed.) (1988) How Labor Markets Work: Reflections on Theory and Practice, Lexington Books, Lexington, MA. Q Fondazione Giacomo Brodolini 1993.
230
Gianni Zappala
Kaufman B. E. (1989) “Models of Man in Industrial Relations Research’, Industrial and Labor Relations Review, 43 (1: 72-88). Kaufman B. E. and Martinez-Vazquez J. ( I 990) “Monopoly, Efficient Contract, and Median Voter Models of Union Wage Determination: A Critical Comparison”, Journal ofLabor Research, 11 ( 440 1-424). Keat R. and Urry J. ( 1 975) Social Theory as Science, Routledge & Kegan Paul, London. Kerr C. (1977) Labor Markets and Wage Determination: The Balkanization of Labor Markets and Other Essays, University of California Press, Berkeley. Kerr C. (1983) “The Intellectual Role of the Neorealists in Labor Economics”, Industrial Relations 22 (2: 298-318). Krueger A. B. and Summers L. H. (1987) “Reflections on the Inter-Industry Wage Structure”, in Lang K. and Leonard J. S . (eds.) Unemployment and the Structure of Labor Markets, Basil Blackwell, Oxford. Lawson T. (1985) “Keynes, Prediction and Econometrics”, in Lawson T. and Pesaran H. (eds.) Keynes Economics MethodologicalIssues, Croom Helm, London. Lawson T. (1989) “Abstraction, Tendencies and Stylised Facts: A Realist Approach to Economic Analysis”, Cambridge Journal of Economics (13: 59-78). Lawson T. ( 1993)“Econometrics, The Limits Of ”,in Hodgson G., Tool M. and Samuels W. J. (eds.) Handbook of Evolutionary and Institutional Economics, Edward Elgar, (forthcoming). Leontief W. (1982) “Academic Economics”, Science (217: 104- 107). Loasby B. J. (1989) The Mind and Method of the Economist: A Critical Appraisal of Major Economists in the 20th Century,Edward Elgar, Hampshire. Manning A. (1992) “How Robust is the Microeconomic Theory of the Trade Union?” Discussion Paper no. 65, Centre for Economic Performance, London School of Economics. Man May A. and Sellers J. R. (1988) “Contemporary Philosophy of Science and Neo Institutional Thought”, Journalof Economic Issues, 22 (2: 397-405). Marsden D. ( 1984)‘Homo Economicus and the Labour Market”, in Wiles P. and Routh G. (eds.) Economics in Disarray, Basil Blackwell, Oxford. Marsden D. (1986) The End of Economic Man? Custom and Competition in Labour Markets, Wheatsheaf Books, Sussex. Marsden R. (1982) “Industrial Relations: A Critique of Empiricism”, Sociology (16: 232-250). Marshall A. (1946) Principles of Economics, 8th edn., Macmillan, London, (Appendix C, D). Mayhew K. and Turnbull P. ( 1989) “Models of Union Behaviour: A Critique of Recent Literature”, in Drago R. and Perlman R. (eds.) Microeconomic Issues in Labour Economics,Harvester Wheatsheaf, London. McNulty P. (1980) The Origins and Development of Labor Economics, MIT Press, Cambridge, MA. Meeks J. G. T. (ed.) (1991) Though@/ Economic Man: Essays on Rationality, Moral Rules and Benevolence,Cambridge Univesity Press, Cambridge. Milgate M. (1988) “Controversies in the Theory of Employment”, Contributions to Political Economy (7: 83-120). Mirowsky P. & Sklivas S . (199 1) “Why econometricians don’t replicate although they do reproduce”, Review of Political Economy, 3 (2: 146- 163). Mongiovi G. (1991) “Keynes, Sraffa and the Labour Market”, Review of Political Economy, 3 (1:25-42). Morgan T. (1988) “Theory versus Empiricism in Academic Economics: Update and Comparisons”, Journal ofEconomic Perspectives, 2 (4: 159- 164). 0 Fondazione Giacomo Brodolini 1993
MethodologiculIssues in Labour Economics
23 1
O’Brien D. P. ( 1992)“Economists and Data”, British Journal of Industrial Relations, 30 (2:253-285). Oswald A. J. (1982)“The Microeconomic Theory of the Trade Union”, Economic Journal(2: 576-595). Oswald A. J. (1987)“New Research on the Economics of Trade Unions and Labor Contracts”, Industrial Relations, 26 (1: 30-45). Oswald A. J. (1991) “Progress and Microeconomic Data”, Economic Journal, 101 (404: 75-80). Piore M. J. (1 979)“Qualitative Research Techniques in Economics”, in Van Maanen J. (ed.) Qualitative Methodology, Sage, California. Ramstad Y.( 1981)“Institutional Economics: How Prevalent in the Labor Literature?” Journal of Economic Issues, 15 (2:339-350). Redman D. A. (1 989)Economic Methodology:A Bibliography wirh References to Works in the Philosophy of Science (1860-1988), Greenwood Press, Conneticut. Redman D. A. (1991)Economics and rhe Philosophy of Science, Oxford University Press, Oxford. Roth A. E. (1988)“Laboratory Experiments in Economics: A Methodological Overview”, Economic Journal, 98 (393:974-1031). Sapsford D. and Tzannatos Z. (eds.) (1990)Current Issues in Labour Economics, Macmillan, London. Seccareccia M. (1991)“An Alternative to Labour-Market Orthodoxy: The PostKeynesian/Institutionalist Policy View”, Review of Political Economy, 3 (1: 43-61). Sen A. (1987)“Rational Behaviour”, in Eatwell J., Milgate M. and Newman P. (eds.) The New Palgrave Utility and Probability, Macmillan, London, 1990. Simon H. A. (1 957)Models of Man, Wiley & Sons, New York. Simon H. A. (1 976)“From Substantive to Procedural Rationality”, in Hahn F., Hollis M. (eds.) Philosophy and Economic Theory,Oxford University Press, Oxford, 1979. Simon H. A. (1 987)“Bounded Rationality”, in Eatwell J., Milgate M. and Newman P. (eds.) The New Palgrave Utility and Probability, Macmillan, London, 1990. Solow R. M. ( 1990)The Labor Marker as a Social Institution, Basil Blackwell, Oxford. Sugden R. (1991)“Rational Choice: A Survey of Contributions from Economics and Philosophy”, EconomicJournal, 101 (407:751-785). Summers L. H. (1991)“The Scientific Illusion in Empirical Macro-Economics”, Scandinavian Journal of Economics, 93 (2: 129-148). Turnbull P. J. (1988)“The Economic Theory of Trade Union Behaviour: a Critique”, British Journal of IndustrialRelations, 26 (1: 99-118). Webb S. and Webb B. (1932/1975)Methods of Social Study, Cambridge University Press, Cambridge. Whitfield K. (1991)“Statistical Methods”, in Kelly D. (ed.) Researching Industrial Relations: Methods & Methodology, Monograph No. 6, ACIRRT, University of Sydney. Wilber C. K. and Harrison R. S. (1978)“The Methodological Basis of Institutional Economics”, Journal of Economic Issues, 12 (1:61-90). Wiles P. and Routh G. (eds.)(1 984)Economics in Disarray, Basil Blackwell, Oxford. Wiles P. (1984)“Epilogue: The Role of Theory”, in Wiles P. and Routh G. (eds.) Economics in Disarray, Basil Blackwell, Oxford. Wisman J. D. and Rozansky J. (1991)“The Methodology of Institutionalism Revisited, Journal of Economic Issues, 25 (3:709-737). Woodbury S. A. (1987)“Power in the Labor Market: Institutionalist Approaches to Labor Problems”, Journal of Economic Issues, 21 (3:1781-1807). 0 Fondazione Giacomo Brodolini 1993.