I. INTRODUCTION
The period immediately after World War II saw considerable shifts in emphasis in the economics discipline. The traditional approach with its focus on exposition and prose was being transformed by the likes of Jan Tinbergen in pushing forward more explicit quantification, and of Paul Samuelson in bringing the language of mathematics to the core of the subject. None of these changes were new in themselves. Stanley Jevons, Léon Walras, and others had been involved in encouraging more explicit formalization, but the early 1950s, in particular, saw a shift in the mainstream.
As the study of the aggregate had gained ground, Keynesian economics, with its novel concepts such as the consumption function and marginal efficiency of investment, cried out for more formalization and quantification. The inflations of the immediate post-World War II period highlighted a particular area where Keynesian theory was proving deficient: understanding the forces that cause rises in aggregate price levels. To help resolve these types of problems, electronic calculators and, later, mainframe computers began to provide the necessary powers of estimation; econometrics, the link between data and theory; and mathematics, the tools for formal hypothesis setting. The political desire to have more control over national economies provided the stimulus for systematic data collection. Institutionally, this led to a period of shifting “power” within the academy, with an overall rise in the perceived importance of economic research and education, resulting in new departments being formed, existing ones being expanded, and some established ones losing influence.
Against this background, this paper looks at a number of issues. At a broad level, the paper looks at some aspects of the nature of the transition process in economic methodology, and examines just how messy it was. It was neither a rapid change nor a very clean one, and certainly there was questioning of the speed at which some of the results were thrown up or were useful. To give context, the focus here is on the wage-rate change/unemployment relationship that was heavily discussed in the mid- to late 1950s and the approaches adopted to look at it.
More explicitly, individuals inevitably play a role in change and, within this context, I highlight the role of the Leeds University economist Arthur (A. J.) Brown in the debate surrounding wage-rate change and unemployment relationships, and the academic networks involved. The publication of Alban William (Bill) Phillips’s (1958) paper and the subsequent moniker of the “Phillips Curve” (hereinafter the “Curve”) have attracted considerable attention, but I focus on the extent to which Phillips’s findings were interlaced with those of Brown (Reference Brown1955), and indeed whether the prequel in a number of ways offers a superior piece of analysis.
There are not many economic concepts that are named after individuals; the Phillips Curve is one. Whether, as Bernard Corry (Reference Corry, Arestis, Desai and Dow2001) argues, the Curve is actually “Phillips’s Curve” or not, Phillips’s general formulation of a measurable, stable wage-rate change/unemployment relationship is still often found in textbooks and seen as a useful pedagogic device for policy discussions.Footnote 1
There are few rewards in academia, but recognition is one. What I consider is whether due credit has been given to Brown’s work on inflation and unemployment. Both Anthony Thirlwall (1972, p. 325), a student of Brown’s, and John Brothwell (1972, p. 57), a colleague, favored the “Brown Curve.” Away from Leeds, at Queen Mary College, Corry (2001, p. 171) posited that “a case can be made for referring to Brown–Phillips curves.” Louis Dicks-Mireaux and Christopher R. Dow (1959), Martin Bronfenbrenner and Franklyn Holzman (1963), and Edwin Kuh (Reference Kuh1967) also give Brown at least joint credit.Footnote 2
I am not concerned with claims made for prior developments of the Curve other than those of Brown; these are anyway largely dismissed in James Forder (Reference Forder2014). What I am concerned with is to offer a slightly variant view of the events of the mid-1950s than has been presented, and to position Brown’s work on inflation more centrally in a world when the academic focus was more geographically concentrated than today, and when there were major changes in the methodology of economic analysis taking place. In the latter context, it also provides something of a contrast between two alternative approaches to the same issue, and raises some questions about the current focus on hard, quantitative analysis, speed of publication, and the style of presentation.
II. THOSE INVOLVED
The academic world of the early 1950s was significantly different from that of today. Communications between academics was slower, generally conducted by mail or word of mouth; there were no photocopiers to mass-produce papers; computing was in its infancy; and there was no cheap air travel. Intellectually, networks were necessarily smaller and tighter. World War II had seen the university system disrupted both for students and academics as the wartime demands for intellectual inputs expanded. Most individuals of student age were enlisted into some form of national service, as were the bulk of university buildings.
The postwar World War II period saw a surge in demand for university places, partly met by expanding existing institutions but also by upgrading the status of former colleges. This afforded openings for younger faculty who had spent time away in the military or government to accelerate their careers. Importantly for Brown, this offered the opportunity to head the Department of Economics and Commerce at Leeds University, and the opportunity for Phillips to remain at the London School of Economics (LSE) after completing his PhD.
The general outcome of this was the formation of new academic networks within a fairly established communications system. Without the “advantages” of email, texting, and blogs, ideas spread more slowly and audiences tended to be more focused, given the challenges of joining an information network. Hence, those involved in the wage-rate change/unemployment debates of the 1950s were small in number, largely geographically concentrated, and personally acquainted. In this environment, personalities and context were important in understanding how ideas were developed, and which ideas ultimately gained most traction in economic policy debates.
The characters, backgrounds, and circumstances of Brown and Phillips were significantly different; given the large literature on Phillips, we focus more on Brown.
Brown was the archetypical high-flying academic of his period.Footnote 3 After winning a scholarship to Bradford Grammar School, he obtained a Hasting Scholarship in Natural Sciences at Queen’s College. He graduated in 1936 with first-class honors in philosophy, politics, and economics from Oxford University, obtaining the highest marks in over a decade. Within three years, and whilst working in Roy Harrod’s research group on prices and interest (Besomi Reference Besomi1998), and completing his DPhil at Oxford’s Institute of Statistics under the supervision of Jacob Marschak, he had published the first empirical analyses of liquidity preference in the UK, making the first use of confluence analysis (Brown Reference Brown1938, Reference Brown1939). And, in the meanwhile, he beat the future British prime minister, Harold Wilson, and future principal of the University College of Buckingham, Max Beloff, for a fellowship at Oxford’s All Souls College. He had a distinguished war service in the Foreign Research and Press Service, the Research Department of the Foreign Office, and in the Economic Section of the Cabinet Office, where, under the invitation of James Meade, he worked on the German rearmament and on the creation of the Government Statistical Service. He moved to Leeds University in 1947 as a professor of economics and head of the department at the age of thirty-three, the youngest professor in the subject of his generation.Footnote 4 He subsequently declined professorships at both Oxford and Cambridge.
By inclination, Brown was an applied economist, as can be seen in his subsequent economics textbook (Brown Reference Brown1959a), seeing his role as applying economics to assist in the formulation and assessment of policy. John Bowers (Reference Bowers2003) outlines Brown’s approach: “[H]aving explained the economic theory, Brown would produce a small slide-rule that he carried in his top pocket and carry out rapid calculations to give some quantitative basis for argument. In his hands this simple tool kit could be devastatingly effective.” By modern standards, this approach would likely be considered ad hoc, but it was used at a time before computers and associated software were widely available, and even electronic calculators were rare (Renfro Reference Renfro2004). In practice, it provided a powerful tool for policymakers of the day.
He did significant work in international trade, economic development, inflation, and regional economics. His 1959 book, Introduction to the World Economy, was a standard text for many years, and The Great Inflation, 1939–1951 has a claim to have presented the underlying concept of the Curve some four years before Phillips, given the dating of the Preface, and that, as Corry (2001, p. 168) says, it was “a long time in writing.”
Brown’s applied skills and drafting abilities made him a popular economist for government, and whilst at Leeds he continued with part-time public service. In the 1960s he was a member of the Raisman East Africa Economic and Fiscal Commission (1961) and the Secretary of State’s Advisory Group on Central Africa, and a member of the University Grants Committee, as well as serving on committees responsible for establishing the universities of Kent and Bradford. He was also the UK’s appointee to the UN Consultative Group on the Economic and Social Consequences of Disarmament (United Nations 1962), which also included Wassily Leontief and Oscar Lange. When the 1966 Wilson government launched an active regional policy, Brown was appointed to the committee looking into the intermediate areas (Hunt Reference Hunt1969). He was already directing a major project on regional policy based at the National Institute of Economic and Social Research and thus well prepared for the task (Brown Reference Brown1972). Later, he was a member of the Commission of the European Community’s (1977a, 1977b) MacDougall Group that looked at public finance issues that could arise from further European integration.
His extensive administrative acumen allowed him to build up a major academic department at Leeds, and to later serve as Pro-Vice-Chancellor of the university. Brown also had no adversity to econometrics, and was far from being non-technical, having initially intended to read science at Oxford, and appointed leading quantitative economists to his department. His contributions to public service were recognized by his being named President of Section F of the British Association for the Advancement of Science and an elected Fellow of the British Academy, being made a Commander of the Order of the British Empire, and receiving honorary doctorates from four universities. He was on the Council of the Royal Economic Society twice, and was its president from 1976 to 1978.
In contrast to the almost mandarin-style progression of Brown’s career, Phillips was more a man of physical action with entrepreneurial tendencies, who arrived in the realm of academia relatively late in life.Footnote 5 Brought up in New Zealand, he became an apprentice electrician, and then spent time traveling around Australia before moving to England in 1937, where he became a graduate of the Institute of Electrical Engineers. He signed up at the LSE, where he took courses in local government administration, geography, and banking, but then war intervened.
His wartime service in the air force saw him being made a Member of the Order of the British Empire (Military Division) for bravery, and spending three years as a prisoner of war. With peace, he returned to LSE, where he just managed to obtain a degree in sociology—after failing both the Applied Economics and Economic History examinations, and passing Economic Principles by a single mark—apparently with a little help from Lionel Robbins, Dennis Robertson, and Nicholas Kaldor, who saw considerable potential in him (Blyth Reference Blyth1975, p. 305).Footnote 6
In 1950 Phillips began, with Walter Newlyn, who moved from LSE to Leeds University in 1948, to build a physical hydraulic Keynesian model of the national economy, based on his work on economic dynamics (Phillips 1950; Newlyn Reference Newlyn and Robert Leeson2000). This was MONIAC (the Monetary National Income Analogue Computer). James Meade had him make one for LSE, and Abba Lerner, as his agent, got some sales in the US (Barr Reference Barr and Robert Leeson2000). Brown not only bought the prototype for Leeds University, he provided the funds from the Economic Department’s budget (£100 in prices at the time, or £3,200 in 2017 prices) for the initial hardware costs, and loaned the prototype to LSE for marketing purposes.Footnote 7 Brown also occasionally met Phillips when he came to work on the Leeds machine (Brown Reference Brown and Robert Leeson2000, Reference Brown2011). The “model” formed the basis for Phillips’s PhD thesis, which was supervised by Meade and for which Dennis Robertson was the external assessor.
The move by Meade to a professorship at Cambridge in 1957 could hardly have been anything but advantageous to Phillips, given the rather narrow nature of the economic networks of the period. Meade had been one of Phillips’s major champions at LSE and the person who, Basil Yamey indicates, brought the Curve paper to the attention of the editors of Economica, and, according to Allan Sleeman (2011, p. 228), “set about organizing his colleagues to do a little gentle lobbying.”
With Meade’s support, Phillips became an assistant lecturer at LSE in 1950, being promoted to Reader in 1954 and to the Tooke Professorship in 1958. The birth of the Phillips Curve, discussed in more detail below, was an untidy event variously described as “a rushed job” (Sleeman Reference Sleeman2011, p. 227), “quick and dirty” (Schwier Reference Schwier and Robert Leeson2000, p. 25), and “done in a weekend” (Gregory to Leeson, in Leeson Reference Leeson and Robert Leeson2000, p. 11), and possibly stimulated the enhancement of Phillips’s vitae prior to his seeking the Tooke Professorship. Following his publication (1958), Phillips did little else directly on the subject and, partly for personal reasons but also because of disenchantment with the situation in UK universities of the time, moved to a chair in economics at the Australian National University, where he spent most of his time studying the Chinese economy.Footnote 8
While Robert Leeson (2000, pp. 10–11) offers a number of broad possibilities regarding why Phillips’s interests changed, Kelvin Lancaster (1979, p. 634), focusing specifically on the Curve, argues that he became “increasingly aware of the difficulties of estimating the relationship [between wage changes and unemployment] he considered necessary for policy design and of the fact that the necessary techniques were beyond his grasp.”
For any concept to gain support, there is normally a set of influential people in the background, providing sounding boards, as well as networks facilitating dissemination. In this context most of these had LSE connections. While the London School of Economics and Political Science in the 1950s and 1960s was a powerhouse in economics, Leeds University was a “redbrick” institution steeped in the industrial needs of the textile industries of the region; indeed, it still attracts the highest level of industrial funding of any university in the UK. The Department of Economics and Commerce in the 1950s had little of the intellectual social science pedigree associated with the likes of Sidney Webb, Beatrice Webb, Graham Wallas, and George Bernard Shaw at the LSE.
Economists at LSE when Phillips was developing the Curve represented one of the strongest groupings in the UK, and with an international academic network to match. They included Robbins, from 1925 until 1961; Meade, from 1947 to 1957, when he moved to Cambridge; Yamey, who was editor of Economica when Phillips’s paper was published; Richard Lipsey, who subsequently offered a critique of Phillips’s work; and Ralph Turvey, who wrote a review of Brown’s book. Richard Sayers had been at LSE since 1947 as Cassel Professor of Economics, and later served on the Radcliffe Committee that examined the workings of the UK monetary system and to which Brown gave evidence.
Additionally, there was Henry Phelps Brown, who had been working on similar topics to Phillips since the late 1940s, and was an advocate of the cost-push theory of inflation.Footnote 9 He and Meade and Lipsey are acknowledged in a footnote by Phillips (1958, p. 285) for their “index of hourly wage rates.” Indeed, one possible factor in the story of the Curve, and certainly not a trivial one, was Phelps Brown’s supposed suggestion “one Friday” that Phillips could apply his data for wage rates and those of William Beveridge for unemployment to explore the price-income relationship (Lipsey Reference Lipsey and Robert Leeson2000, pp. 234–235).
There was also Frank Paish, the Sir Ernest Cassel Professor of Economics, who in the 1960s supported an extreme interpretation of the Curve when arguing that a small amount of spare capacity, involving a level of unemployment, was needed to prevent inflation (Paish Reference Paish1962). Added to this was Abba Lerner, who, besides being Phillips’s American marketing man for his “machine” (Barr 1988, pp. 317–318), was a graduate of the LSE, interested in links between wages and prices, and, importantly for the subsequent transatlantic uptake of the Phillips Curve, well known to Samuelson, not least because of their works on factor price equalizations. Kelvin Lancaster, who subsequently wrote a short biography of Phillips, arrived at LSE as an assistant lecturer in 1953, and was a contemporary of Phillips whilst completing his PhD.
In contrast, Brown was building up the Department of Economics and Commerce at Leeds, and although it had Newlyn (but who was away for extended periods working on African matters) and Denis Sargan, who subsequently anticipated the expectations-augmented Phillips Curve, they were at the beginnings of their careers in the early 1950s. There was also Hermann Hillmann, an expert on the German economy and who is explicitly acknowledged by Brown (1955, p. vii) in The Great Inflation, and John Brothwell, initially an industrial economist who subsequently became an expert on monetary studies, who joined the faculty in 1953, after graduating from Leeds. In sum, as Brown put it in his Introduction to The Great Inflation, “To the Royal Institute also I owe profound apology for the slowness with which the work proceeded, in competition with the rival claims of academic administration and teaching” (Brown Reference Brown1955, p. vii).Footnote 10 Brown was also a very meticulous writer, and took inordinate care over his work before it saw the light of day; there were no “rushed” papers.Footnote 11
III. PHILLIPS’S CURVE
The underlying question addressed is, “What were the additions of value that Phillips made to Brown’s analysis?” In many ways Phillips’s paper follows Brown’s Chapter 4. Both used similar data sets related to union-agreed wage rates, although Brown’s time frame was 1881 to 1951 and Phillips’s was from 1861 to 1957, and related to unemployment for their respective periods. Given the time frames, and particularly that prior to the 1920s, both the wage and the unemployment data inevitably embraced considerable synthesis of various time series, not to say inventiveness, in their construction (Wulwick Reference Wulwick1987). As Phillips pointed out, they covered only a small part of the labor force, and related to the standard (or “wages,” in the terminology of the day) and not the actual wage rates (“earnings”); the former often being negotiated to cover several years.Footnote 12 Additionally, the data on wages were not for the same industrial set as for unemployment.Footnote 13
Both looked at changes in wages and unemployment levels in the pre-World War I period (Brown to 1914 and Phillips to 1913), and related these to various later periods. Further, they had essentially the same quest; for brevity, I quote Brown (1955, p. 90):
The probability of a relationship of this kind [an inverse relation between wages/prices and unemployment] has … already played a considerable part in economic discussion.… It seems, however, that empirical investigation has not hitherto been directed towards verifying the existence of such a relation, or to finding the level of unemployment beneath which the price-wage spiral seems to come strongly into operation in any one economy.Footnote 14
Where they differ is in their approaches to the data and in some of their conclusions. As summarized by Malcolm Sawyer (1989, p. 102):
The approach of Brown can be contrasted to that of Phillips in three respects. First, Brown places the statistical relationship into a much fuller discussion of the process of inflation.… Second, Brown did not attempt to draw any curve through his data.… Third, Brown did not argue that the wage change-unemployment relationship observed for the pre-First World War period held thereafter.
Compared with Brown’s detailed examination of the causes of inflation, Phillips provides very limited explanation of what he finds.Footnote 15 While, in the “Hypothesis” opening section to his paper, there is a rather broad discussion of the forces that may underlay rates of change in the UK’s money wages, this is never really integrated into the statistical analysis, and there is limited discussion of the underlying rationale of his findings. James Tobin (1972, p. 9) later commented on the Curve that it was “an empirical finding in search of a theory,” while David Worswick (1979, p. 37), who was at Oxford in the 1950s and subsequently director of the National Institute of Economic and Social Research, described Phillips’s analysis as “stark.”Footnote 16 Forder (2010, p. 331) perhaps sums things up best: “The significance of Phillips’s finding is that it amounts to a claim that the relation is independent of any of the extensive institutional or political changes that took place over the period.”
Brown, in contrast, offers a more complete and integrated explanation for his findings, fitting them within an institutional and industrial organization framework, as well as a Keynesian explanation.Footnote 17 His basic view was that
though it may be possible to give a generally valid formal description of what constitutes inflation, the causal mechanisms by which it happens are various and their roots may go deep into the market institutions and the social and political structures of the communities concerned. For this reason it is not profitable to approach the events of the inflationary period treated in this book with a single theoretical scheme into which they must be fitted. (Brown Reference Brown1955, p. 17)
Put simply, he sees no Curve.
Given that data used by Brown and Phillips are remarkably similar, why did Phillips then see a consistent relationship between wage-rate changes and unemployment? First, Brown felt that any historical relationships prior to 1939 were not overly useful in explaining inflation between 1939 and 1951 (ibid., p. 90), the subject of The Great Inflation, given the scale of the military conflict from 1939 and its aftermath. Then there is the matter of the challenge of fitting the Curve.
As Meghnad Desai (Reference Desai1975) later points out, the fitting of the “relationship” would not be deemed very satisfactory, even allowing for the state of econometrics and the computing facilities available in the late 1950s.Footnote 18 The data were divided into six unequal tranches according to unemployment levels roughly corresponding to trade cycles. The average unemployment rate was used for each tranche to capture the rise and fall in unemployment over each cycle. A hyperbolic curve was fitted through the lower four points and close to the upper two. Phillips also looked at several trade cycles (1861 to 1868, 1868 to 1879, 1879 to 1886, 1886 to 1893, and 1893 to 1904, with two graphs offered for 1879 to 1886 using alternative data sets on wages) and found support for the general pattern of the Curve.
This procedure did not allow for any genuine statistical significance testing, and extrapolation into future periods saw deviations from the Curve explained in ad hoc terms. Nancy Wulwick (Reference Wulwick1989) did elicit from one of the “anonymous” referees of Phillips’s papers that the R2 for the four-point regression was 0.96, but this is not the same thing as the fit to the full data set, or even the full six averages that trace out the curve. The breaking down of the data by trade cycles was a common practice at the time to handle computation challenges and to cope with serial correlation, but testing for the latter is not reported, despite the work at LSE on the subject by James Durbin (one of Phillips’s PhD supervisors) and Geoffrey Watson (1951, 1952).
Subsequent fittings using more modern regression methods by Lipsey (Reference Lipsey1960) found that 64% of the variation in the wage growth was explained by variations in unemployment. Further, Lipsey found that fitting a curve to data for 1862 to 1913, as Phillips did, provides a somewhat different picture from that using 1923 to 1939 and 1948 to 1957 data—the latter being above the earlier period for unemployment rates under 3%, and below it for unemployment rates above 3%.
The methodology favored by Phillips also provided neither short- nor long-term equilibriums but rather, as Desai (1975, p. 2) puts it, “a locus of long-run equilibrium points for U and
$\dot{W}$ thus making the common distinction between a short-run and a long-run ’Phillips curve’ unnecessary and erroneous.” Brown, in contrast, sought to look for explanations of a temporal relationship between unemployment and wage-rate changes, because he was interested as much in the dispersion of results over time periods as in any sets of long-term equilibriums. It is in this context that Brown probably did not see the Phillips approach as useful.
Brown (1955, p. 90) sums up his position on statistical analysis of his data:
In between 1880 and 1914, the index of wage rates was strongly influenced by certain industries—most notably coal mining—in which … wages were related to the price of the product. This clearly reduces the probability that experience over these years will throw light on the present problem, but a glance at the data is, nevertheless, not entirely unwarranted.
Brown did not feel the quality of the data justified any efforts at detailed curve fitting. His approach to the pre-World War I data, besides reporting them, was to look for time series effects, and particularly for differences in business cycles. In doing this, he did find some consistency regarding the booms breaking in 1890, 1900, 1907, and 1913: “[M]oney wages usually continued to decline at a fairly steady rate of about 1 per cent. per annum until the onset of recovery, to the accompaniment of a steady increase of unemployment. During the upswing, the rate of increase of wages rose steadily as unemployment fell” (Brown Reference Brown1955, p. 90).
Brown also proffers an explanation for the quite dramatic changes after turning points—for example, especially after the booms of 1900 and 1907, when the rate of wage change became negative the following year. He introduces the importance of expectations, which is largely absent from Phillips’s paper, most notably in terms of its asymmetric implications for the wage-rate change/unemployment relationship in the upward and downward phases of a cycle.Footnote 19 In addition, he highlights the “floor” of wage changes that is observed in pre-World War I, although this is less pronounced in Phillips’s data. He argues (Brown Reference Brown1955, p. 92), “In a society in which labour is organized with any considerable degree of effectiveness, there is probably an absolute limit to the rate at which wage rates can be reduced—unless something remarkable is happening at the same time to the cost of living.” He does not, though, provide an explanation of what determines the critical level of a “considerable degree of effectiveness.”
In summary, Brown did not consider it worth expending much energy on the pre-World War I period as a guide to the time frame on which he was focusing. Relationships essentially change with circumstances. In a way, Phillips (1958, p. 298) also accepted this when he argued that the Curve breaks down on occasions, and most notably at times of major wars, and “in or immediately after those years in which there is a sufficiently rapid rise in import prices to off-set the tendency for increasing productivity to reduce the costs of living.”Footnote 20 This, however, he treats as irrelevant for a full analysis.
Regarding later periods, again, whilst Brown is content to just plot points, Phillips joins them and highlights the way they seem to fit the Curve. It is quite clear that the Curve breaks down between 1919 and 1923, a period Brown (1955, p. 93) describes as “quite unlike any experienced at other times.” While both agree that the pattern in 1919 to 1920 was due to cost of living increases, the fall in wages after 1920 was, following Brown, because “by the time the slump came, sliding scales had become so popular that more than half of the recorded changes in wages in the years 1921–2 were made automatically in accordance with them,” and one that Phillips (1958, p. 293) largely attributes to “a fall of 12.8 per cent in the cost of living, largely a result of falling import prices.”
For the subsequent period, Brown’s findings reflect a broad Phillips Curve, but even then only up to a point.
From 1924 until 1939 the relation between unemployment and rate of wage change was similar to that revealed in the cyclical movements of the generation before 1914; except that the level of unemployment throughout (as far as one can judge by comparing the post-war Unemployment Insurance statistics with the pre-war trade union records) was much higher. (Brown Reference Brown1955, p. 93)
Considering temporal patterns, this would imply that there had been a shift in the Curve, which Brown partly attributed to the weakening of trade union power after 1926, but mainly due to structural changes in the UK economy, whereas Phillips saw a strict adherence to the Curve despite the gyrations that occurred about it.Footnote 21 Brown’s position is clear, and relates both to changes in the types of industry and to their geographical concentration:
It seems likely that at least half the unemployment existing in 1937, for instance, was of a structural character, and if that is so, one may say that, had there been no more structural unemployment than before 1914, the critical level of unemployment in the later nineteen-thirties would have been less than it actually was by something like 4 per cent of the insured population. (Brown Reference Brown1955, p. 95)
Brown also informally introduced the concept of the famed “loops,” his spirals, essentially, as found later, associated with hysteresis in the labor market. Although he did not draw them, or at least offers no visuals in his book, he does talk of “roughly parallelogram-shaped cycles.” More specifically (Brown Reference Brown1955, p. 90), “The movement of both the variables were almost entirely of a cyclical nature, and the relation between them in their cyclical movement was fairly constant.”
There was some difference in Brown’s and Phillips’s interpretation of these loops, a fact the latter puts down to Brown’s not fully appreciating short-term lags in the system. Phillips, for example, finds that by introducing a lag of seven months into his analysis, the data for 1948 to 1957 fit well with the Curve (his Figure 11). He gives no explanation for the length of the lag; it simply seems to force the data onto the Curve. He does, however, discuss changes in the nature of wage agreements and bargaining, but in terms of the narrowing of his loops: “[T]hese agreements may have strengthened the relation between changes in wages rates and changes in unemployment in these industries [coal and steel]” (Phillips 1958, p. 292). Brown (1955, p. 93), in contrast, emphasizes the importance of sliding pay scales, albeit in terms of shifting the wage-rate change/unemployment function.
Brown also observed that from 1939 to 1951, “[i]n many of the countries most dependent upon international trade … the degrees of inflation experienced were not very different, though the stresses from which in the first place inflation arose—notably the stresses of war—bore very diversely upon them” (Brown Reference Brown1955, p. 251). Phillips focuses on the UK, which, a critic could argue, can offer little by way of a universal model, and just injects the idea that some of his spirals are influenced by import prices (Phillips 1958, pp. 296–297). Brown’s view that there had been a long-term shift in the wage-rate change/unemployment relationship after World War I was not subsequently changed by the publication of Phillips’s paper, since
the evidence is that the existence of unemployment alone, not accompanied by steady or falling costs of living, is not a very powerful reducer of wage-demands—or even of wage increases—in most modern economies. A generation or two ago when labour was less organized in most countries than it is now, and wages depended more upon competition between employers, it was somewhat more effective, but it seems that we must now reckon on rather a high rate of unemployment being required to produce a moderate reduction in the rate of wage-increase. (Brown Reference Brown1959a, pp. 77–78)
Where Brown and Phillips do agree, however, is with regard to the poor relationship between wage-rate changes and price inflation, or at least in the short run. This is clearly seen in Brown’s Figure 17 (where he does join observations) and discussed in his accompanying text, although, later, Robert Gordon (2011, p. 12) mistakenly gives precedence to Phillips: “Phillips surprisingly debunks a third possible correlation, that between the rate of change of wages and the retail inflation rate (‘working through cost of living adjustments’).” Equally, Gordon’s (ibid., p. 12) statement that “Phillips was already thinking of a world in which demand shocks … and supply shocks … both mattered in determining wages and price changes” seems to forget Brown’s position.
What has attracted most of the policy debate concerning the Curve, but one only scratched upon here (Forder Reference Forder2014 does a thorough job), is whether the Phillips Curve offered a menu of policy options involving trade-offs between inflation and unemployment. Brown’s essentially cost-push findings move more in the direction of prices and incomes policies, which he looks at in some length in his book (Chapter 7). The debate over whether the Phillips Curve is “Phillips’s Curve” is academically interesting and, depending on the interpretation, has policy implications, but really is a little beside the point when looking at the respective works of Brown and Phillips. The fundamental issue centers on whether the underlying causes of inflation are demand or cost pressures.
IV. A. J. BROWN AND INFLATION POLICY
Unlike Phillips, who rapidly moved away from the wage-rate/unemployment debate, Brown returned to the subject on several occasions, and very often in a policy context. Brown’s interest in the policy side of inflation had initially been stirred during World War II. He had contributed several papers to the Bulletin of International News on the subject, and most notably on wages and price control. His view on the subject then was that “the understanding of what [inflation] is (still less of how it is caused and how it can be avoided or remedied) has become as widespread as the fear of it” (Brown Reference Brown1945, p. 11).
Following the publication of The Great Inflation, Brown sought to explain its policy implications. On this, he spoke on the BBC (Brown Reference Brown1956), gave evidence to government commissions (Brown Reference Brown1958a, Reference Brown1959b), and took opportunities in his presidential addresses to Section F of the British Association for the Advancement of Science (Brown Reference Brown1958b) and the Royal Economic Society (1979) to reflect on the subject of anti-inflation policies.Footnote 22 He also extended his analysis in academic writings.
In terms of defending his analysis against Phillips’s work, Brown (Reference Brown, Harcourt and Riach1997b) raises questions about the quality of the data used in their analyses of pre-World War II periods, pointing particularly to the revisions of unemployment data by Angus Maddison (1964, p. 221): “For 1950–1960 we have raised the registration figures [which Phillips used] by [one percentage point] because they do not cover all categories of unemployment, and the discrepancy from the 1951 census was of this size.” As Brown (1997b, p. 55) points out, this would imply that subsequent use of Phillips’s calculations during what he calls the “Keynesian Golden Age” of the 1960s in the UK would have suggested that higher unemployment would have meant less wage inflation. The implication of this for Brown is that “it suggests a tendency towards local instability in the relevant part of the Phillips curve; that when unemployment is pushed down to some level at which recruitment difficulties are perceived or expected, the remaining available labour (or most of it) will be very quickly taken up for precautionary reasons.”
Brown’s later empirical work involving Jane Darby (1985) covered a range of developed and developing economies, and confirmed his view that any degree of wage inflation involves more than just demand factors and, in Milton Friedman’s sense, expectations. Explicitly in terms of the Curve, the authors emphasize the ordinal nature of their findings (ibid., p. 381): “The simple Phillips relation between the tightness of the market and the rate of change of hourly earnings remains as a fairly persistent, though not universal, constituent of wage equation.” And, regarding the larger matter of inflation (ibid., p. 382), “Institutional and political changes contribute heavily to the complicated story of wage push.”
While Phillips, excepting his inaugural lecture at LSE (Phillips 1962), eschewed any detailed discussion of policy, subsequent adherents and antagonists of the Curve generally have not been so reticent. Brown does venture into this, and particularly supports cost controls. His 1955 book contains a chapter on “Price Control” that reviews its prior applications and concludes that “in addition to price control and subsidies as measures of insulation against temporary excesses in demand, internal or in the world market, a wage policy is needed.”(Brown Reference Brown1955, p. 171). The theme is returned to in a more popular way in a speech on BBC Radio (Brown Reference Brown1956), when he argues for more public discussion and planning of “the proper absolute and relative sizes of wages, salaries, and profits” to prevent inflation. Later, his Presidential Address to the Royal Economic Society (Brown 1969, p. 11) pushes the point further: “[I]f I am right in believing that cost-push, and particularly wage- and salary-push, plays a central part in reducing industrial profitability and competitiveness then it follows that incomes policy is bound to be helpful in breaking the vicious circle.”
In one important sense, despite the widespread uptake of the term “Phillips Curve,” Brown’s work may have been more influential, at least in the short term. Phillips’s paper was timely in that the Committee on the Working of the Monetary System (1959)—especially since his LSE colleague Richard Sayer was a member, which cemented the notion of Keynesian and fiscal policy into the fabric of the UK’s macroeconomic policy—was conducting its enquiry at the time of its publication. But Phillips did not give evidence to the committee, whereas Brown (Reference Brown1958a, Reference Brown1959b) gave both written and oral evidence.
Brown, drawing on his book, spends time discussing monetary transmissions mechanisms:
[I]t does not follow that a restriction of money supply is an efficient method of stopping inflation. If the latter is caused by a tendency of wages and salaries to rise through the collective bargaining process even in circumstances, which would not otherwise raise prices, then monetary restriction may be … an inefficient instrument for checking it. It is inefficient in the sense that it succeeds only at a high cost both in unemployment of labor and equipment and in delay in application of technical improvement. (Brown Reference Brown1959b, p. 50)
But he also reflected on the wage-rate change/unemployment relationship in the later years of the interwar period:
The wage changes between the wars are odd in all sorts of ways. For one thing they are negatively correlated with the growth of industry; that is very strange and seems to me to suggest that they have more to do with bargaining power, union activity and so on, than with the level of effective demand for labor in a particular industry. This seems to me to be rather symptomatic on the whole for the changes that have happened since about 1914. (Brown Reference Brown1958a, p. 593)
Clearly, this is not a view consistent with Phillips’s findings, and the policy implications are more in line with the concept of a “social contract” brought in by James Callaghan’s UK government in the 1970s.
The academic uptake of the Curve foreseen by Brown was also seen in the US. Indeed, the term “Phillips Curve” was first coined there by Guy Routh (Reference Routh1959), but really gained in recognition in December 1959 when Samuelson and Robert Solow (1960) discussed it during a presentation to the American Economic Association.Footnote 23
Samuelson and Solow, however, made no mention of Brown’s work, despite the fact that his book contains a discussion, as well as a diagram, his Figure 15, virtually identical to their own Figure 1, and also uses hourly earnings data.Footnote 24 In fact, they seem to be unaware of it—“Strangely enough, no comparably careful study [as Phillips] has been made for the U.S” (Samuelson and Solow Reference Samuelson and Solow1960, p. 187)—despite Brown’s work, as we see below, having been reviewed in the major US academic journals.Footnote 25
Regarding the overall picture, there are also strong similarities among Brown and Samuelson and Solow. The former (1955, p. 95n1) finds, “It is … evident that the inter-war in the United States did not follow the same course as any of those in between 1880 and 1939.” Brown also isolates a shift in the US wage-rate change/unemployment relationship between 1921 to 1932 and 1935 to 1941, with Samuelson and Solow, some five years later noting a similar shift, albeit with 1933 to 1941 as the second period. Brown (1955, p. 96) attributes the change to “the New Deal, which greatly strengthened the power of United States labor, both by promoting its organization and by the introduction of relief and unemployment insurance,” with Samuelson and Solow (pp. 187–188) arguing, “One may attribute this to the workings of the New Deal … or alternatively one could argue that by 1933 much of the unemployment had become structural, insulated from the functioning of the labor market.” These latter arguments also have a remarkable similarity to Brown’s rationale cited earlier for shifts in the post-World War I UK’s wage-rate change/unemployment relationship.
V. WHY THE “PHILLIPS” CURVE AND NOT THE “BROWN–PHILLIPS” CURVE?
It is not unusual in economics to give dual attribution to advances that help move debates forward when elements emerge sequentially or independently—for example, the Marshall–Lerner conditions in international trade, the Hicks–Hansen synthesis for equilibrium in the closed macro-economy, and the Swann–Solow and Harrod–Domar models of economic growth, or even the Newlyn–Phillips machine.
In our context, this raises the question of why we have the Phillips Curve and not, as Corry favored, the Brown–Phillips Curve.Footnote 26 It was not as if there was an ocean between them; the works of Brown and Phillips were known to each other, and, of importance in the chronology of events, Phillips knew of Brown’s book. Indeed, Richard Sayers, Phillips’s colleague at LSE, read over Brown’s manuscript (Brown Reference Brown1955, p. vii).
There were also direct discussions between Brown and Phillips on the Curve. Brown (2000, pp. xii–xiv) reflects:
I cannot claim to have known Phillips as well as his LSE colleagues for instance do, but our acquaintance, which was connected with the hydraulic model and, more fleeting, with the curve, made a deep impression on me.… Before the publication of his famous article which launched the curve, knowing of my earlier work on the same subject he invited me to discuss with him the results recorded therein. I think our discussion turned on the difference made by our having used slightly different time lags in our respective plotting of largely similar data.
In a set of comments by Phillips (Bergstrom et al. Reference Bergstrom, Allan, Catt, Peston and Silverstone1978, p. xvi) is a similar report of the meeting: “[It] was a rush job, I had to go off on sabbatical leave to Melbourne; but in that case it was better for understanding to do it simply and not wait too long. A. J. Brown had almost got these results earlier, but failed to allow for the time lags.”Footnote 27
This latter statement, emphasizing the need for speed as a justification for not even acknowledging Brown’s work, seems rather strange, given the time frame involved.Footnote 28 But what is perhaps reflective of Brown’s personality, and also that he never seriously sought to claim precedence, is that he clearly held Phillips in high regard, to the extent of having a photograph of him and MONIAC hanging by his fireplace at his home (Brown Reference Brown2011, p. 33).
So why did Phillips’s paper attract the attention rather than Brown’s work? This is not an easy question to answer, and much of the subsequent discussion in the literature of the impact of Phillips’s work tends to treat it more or less in isolation.
Brown’s work was certainly not unknown amongst academics. As well as being discussed at Robbin’s LSE seminar, The Great Inflation was reviewed by the LSE economists Ralph Turvey (Reference Turvey1955) in Economica and Harry Johnson (Reference Johnson1956) in the Economic Journal, as well as by Phillip Cagan (Reference Cagan1955) in the Journal of Political Economy, John Gurley (Reference Gurley1956) in the American Economic Review, Victor Morgan (Reference Morgan1955) in the Economic History Review, Bronfenbrenner (Reference Bronfenbrenner1955) in the Journal of Economic History, and Edward Safarian (Reference Safarian1957) in the Canadian Journal of Economics and Political Science. This fact of the diversity of the review venues is perhaps, however, indicative of the book’s falling between too many stools in terms of the academic perception of its ethos.
One problem may be that Brown’s work on wage-rate changes and unemployment was also only part of a very much larger, international study of inflation, and thus it got lost in the overall exercise. Thus, his historical analysis of UK wage-rates changes and unemployment stood out much less than Phillips’s stand-alone, terse, and pointed article. Brown’s writing was also measured, and the work “evolved” with considerable feedback, rather than just “appearing.” Although the h-index and the like were well in the future, academic papers in journals were taking over from monographs; articles could be published more rapidly, and speed was certainly a feature of the publication of Phillips’s 1958 paper. It should also be noted that many journals were institution-based, and often a little parochial in their approach to content. Regarding Phillips’s paper, Yamey said:
I do not recall whether Bill gave me a copy of his paper for editorial consideration. I think I was given a copy by Lionel Robbins or James Meade. Anyway, before the day was over the paper had been enthusiastically recommended to me by several of the editorial board. I shared this enthusiasm after I had read the paper, and within a day of receiving it accepted the paper for publication in Economica—a speed of decision impossible for any refereed journal today. (Yamey Reference Yamey and Robert Leeson2000, p. 336)Footnote 29
The work of Brown was also probably too deeply grounded in the institutional literature (or too “descriptive,” to use Meade’s term regarding Brown’s wider approach to economics) to be acceptable for the more technical-orientation textbook writers from the mid-1950s. Indeed, the cover of The Great Inflation says, “No previous knowledge of economic theory is required by the reader.… It is hoped it will give the general reader some insight into the way in which the economist set about the analysis of current events.” In fact, reviewers of the book specifically commented on its readability; for example, Bronfenbrenner (1955, pp. 304–305) said, “Mr Brown has the gift, less rare among British than American economists, of disguising difficult and controversial propositions of economics as matters that any schoolboy can understand and all men of good will can accept.” Gurley (1956, p. 190) said, “Brown writes exceptionally well.” His later textbook, Introduction to the World Economy (Brown Reference Brown1959a), was written in a similar vein and manifestly diverges from the tomes of Samuelson and Lipsey that were then emerging. Clarity is, however, often not the best way to entrap readers.Footnote 30
The reception of Brown’s book also offers some indication of the thought processes regarding its key themes. Turvey (1955, p. 261) found, “The three chapters on the wage price spiral are excellent, particles of fact, theory, and commonsense combined in Brownsian movement.” Likewise, Victor Morgan (1955, p. 254) in his review of the book, said that “the task of handling so much material would have daunted most writers, but Professor Brown does it with consummate skill and the result is a remarkably interesting and useful book.” Indeed, the reviews in general were highly favorable, with critiques offering alternative perspectives, but certainly the book was taken very seriously. It was definitely not seen as trivial; it cleared the academic review fence.
Timing is another factor in the adoption of ideas, be they sound or not. The mid-1950s was a time when wartime recovery was almost complete and, as Worswick (1979, p. 28) puts it, “The Great Inflation … was published … just at the time when it seemed the unemployment problem had been solved.” By the later 1950s, however, inflation was becoming an issue again and the Curve offered a way of looking at economic trade-offs, irrespective of Phillips’s intent, for long-term policy formulation. This was something that was subsequently recognized by Brown: “I agree that [Phillips] cannot be blamed for the excessively literal meaning attached by others (like Frank Paish) to his curve as a “trade-off” indicator. But in the atmosphere of the time that sort of reaction to his work was inevitable” (cited in correspondence; Leeson Reference Leeson1997, p. 168n4).
VI. CONCLUSIONS
Mark Twain (1933, p. 162) said, “It takes a thousand men to invent a telegraph, or a steam engine, or a phonograph, or a telephone, or any other important thing—and the last man gets the credit and we forget the others. He added his little mite—that is all he did.” But in the case of the Curve, Brown’s contribution was hardly a “little mite”: his analysis was completed in advance of Phillips’s and had more by way of theoretical justification, and many of his conclusions have proven to be much more robust. But Brown’s The Great Inflation had neither good “visuals” nor an appropriate sound bite; Samuelson and Solow, if not originating the term, certainly helped Phillips by labeling the Curve. Also, as Napoleon so often liked to point out, luck is perhaps the most important attribute of all, and Phillips was lucky in his timing, his networks, his style of presentation, and in the changing fads in publication of his day.
The intention of this paper has not in any way been to belittle the work of Phillips—his analysis of the Curve, and perhaps more so of dynamic processes, was exceptional—but rather I offer advocacy for more attention being paid to Brown’s analysis of wage-rate changes and unemployment. In doing this I have largely steered clear of the debates surrounding the importance attached to the Curve by policymakers in the 1960s; this is still a contentious field and not directly relevant to assessing Brown’s pure academic contribution. I have taken the view advanced by David Laidler (Reference Laidler, Samuels and Biddle2002) that Phillips was not putting forward ideas of any menu of policies, but rather seeking stability conditions that could form the basis of reducing fluctuations in the economy. This would seem to be in line with Brown’s subsequent discussion of Phillips’s work.
