Tải bản đầy đủ (.pdf) (948 trang)

The palgrave companion to LSE economics

Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (9.48 MB, 948 trang )


The Palgrave Companion to LSE Economics


Robert A. Cord
Editor

The Palgrave
Companion to LSE
Economics


Editor
Robert A. Cord
Researcher in Economics
London, UK

ISBN 978-1-137-58273-7
ISBN 978-1-137-58274-4  (eBook)
https://doi.org/10.1057/978-1-137-58274-4
Library of Congress Control Number: 2018943850
© The Editor(s) (if applicable) and The Author(s) 2018
The author(s) has/have asserted their right(s) to be identified as the author(s) of this work in accordance with
the Copyright, Designs and Patents Act 1988.
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or
dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does
not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective


laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, express or implied, with respect to the material contained herein or for any errors or
omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
Cover credit: Stephen Finn/Alamy Stock Photo
This Palgrave Macmillan imprint is published by the registered company Springer Nature Limited
The registered company address is: The Campus, 4 Crinan Street, London, N1 9XW, United Kingdom


For Helen and David


Introduction

This is a volume about the economics and economists associated with the
London School of Economics (LSE). It is the second in a series to be published by Palgrave examining the many and varied contributions made by
important centres of economics. With only a very few exceptions, the focus
of most history of economic thought studies, at least in terms of books,1
has been on schools of thought. Such an approach provides valuable insights
into how competing schools interact and how some come to predominate,
for whatever reason and length of time, while others fall out of fashion or
indeed never attain any particular notoriety. However, a key deficiency of
such a modus operandi is that it often fails to illuminate the many processes
and tensions that can and do occur at the level of the individual university,
the personnel of which may be fighting internal battles for supremacy while
at the same time trying to establish external hegemony.
Each volume in the series will consist of two parts. The first will contain
a set of chapters which will consider the contributions made by a centre

where these contributions are considered to be especially important, and
this subjects to a mixture of personal preferences and soundings from those
who know better. The second, longer part will be made up of chapters discussing the contributions of individual economists attached to a particular
centre. ‘Attached’ is the crucial word. Some economists are easy to identify
with a single institution as they may, for example, have spent their whole
academic careers at it. Those who have moved from institution to institution
1 Articles

are of course another matter.

vii


viii    
Introduction

are the more difficult case. One way forward in these instances is to place
an economist in the institution where they carried out their most important
work, although this, in its turn, carries with it the danger of disagreement
over what ‘their most important work’ was or is perceived to be and how
this has changed over time. Another factor perhaps worthy of consideration
is an economist’s education. Where such an education has been received at
the knee of a master, to what extent has this influenced the subsequent work
of the noted pupil and how should this be considered when that pupil has
flown the nest and settled at another institution? Issues of leadership style,
discipleship, loyalty, access to publication outlets and to financing also enter
the frame. Finally, there are issues of practicality, including space constraints
and unavailability of contributors, among others. Given this matrix of possibilities, disagreement about who should be in which volume is inevitable.
However, I hope that the outrage will not be too great given the overarching
goal of the series.

The next volume in the series will examine the University of Oxford.
Robert A. Cord


Contents

Part I  Themes in LSE Economics
1

LSE and Econometrics3
Jim Thomas

2

Economic History at the London School of Economics
and Political Science: A View from the Periphery35
Colin M. Lewis

3

Accounting and the Influence of Economics at LSE79
Christopher J. Napier

4

Business History at LSE: An Empiricist Voice113
Leslie Hannah

5


LSE’s Contributions to the Economics of Social Policy145
Howard Glennerster

6

Economica and LSE Economics165
Jim Thomas

ix


x    
Contents

Part II  Some LSE Economists
7

Edwin Cannan (1861–1935)197
Keith Tribe

8

Arthur Lyon Bowley (1869–1957)215
Adrian Darnell

9

William Henry Beveridge (1879–1963)239
Atsushi Komine


10 R.H. Tawney (1880–1962)263
Noel Thompson
11 Hugh Dalton (1887–1962)289
John E. King
12 Frank Walter Paish (1898–1988)311
Robert A. Cord
13 Arnold Plant (1898–1978)329
Robert A. Cord
14 Lionel Robbins (1898–1984)347
Susan Howson
15 Friedrich Hayek (1899–1992)373
Peter J. Boettke and Ennio E. Piano
16 Abba P. Lerner (1903–1982)399
Warren Young, Daniel Schiffman and Yaron Zelekha
17 John R. Hicks (1904–1989)431
Harald Hagemann
18 Henry Phelps Brown (1906–1994)463
Peter A. Riach
19 Evan Durbin (1906–1948)487
Catherine Ellis


Contents    
xi

20 R.G.D. Allen (1906–1983)515
Jim Thomas
21 Richard Sidney Sayers (1908–1989)533
Alec Cairncross and Charles Goodhart
22 Ronald H. Coase (1910–2013)555

Alain Marciano
23 A.W.H. Phillips (1914–1975)579
James Forder
24 Ezra J. Mishan (1917–2014)615
Euston Quah and Yew-Kwang Ng
25 James Durbin (1923–2012)631
Andrew Harvey and David Bartholomew
26 Michio Morishima (1923–2004)641
Naoki Matsuyama
27 John Denis Sargan (1924–1996)667
David F. Hendry and Peter C. B. Phillips
28 Ralph Turvey (1927–2012)697
Roger Middleton
29 Richard G. Lipsey (1928–)723
Max Steuer
30 Richard Layard (1934–)743
Richard Jackman
31 Charles Goodhart (1936–)765
Donald Kohn
32 Meghnad Desai (1940–)791
P. N. (Raja) Junankar


xii    
Contents

33 Nicholas Adrian Barr (1943–)805
Stuart Astill
34 Stephen J. Nickell (1944–)831
Jan C. van Ours

35 Christopher A. Pissarides (1948–)857
Etienne Wasmer
Notes on Contributors895
Index907


List of Tables

Chapter 6
Appendix 1Distribution of Articles and Reviews in Economica
(Old Series) 182
Appendix 2(a) List of Editors of Economica (New Series) by Date 183
Appendix 2(b) Alphabetical Listing of Editors of Economica186
Appendix 3Articles Published in Economica by LSE Economists
by Year (1931/1932–1997/1998) 187
Appendix 4
Ranking of Journals Based on Citations 188
Appendix 5
Journals Cited in LSE Economists’ CVs in 2017 189

Chapter 28
Table 1
Turvey (1951a): four types of inflationary process 702

xiii


Part I
Themes in LSE Economics



1
LSE and Econometrics
Jim Thomas

1Early Developments
Lionel Robbins, the head of the Economics Department at LSE from 1929
to 1961, was a sceptic when it came to the value of statistical estimation
in economics. In his Nature and Significance of Economic Science (Robbins
1935: 108–109), he satirised the efforts of ‘the wretched Blank’, who, in
1907–1908, estimated the elasticity of demand for the common herring
(Clupea harengus ) to be 1.3:
Rough computations of this sort are not really very difficult and may have
considerable utility for certain purposes. But what reason is there to suppose
that he was unearthing a constant law? No doubt the herring meets certain
physiological needs which are capable of fairly accurate description, although
it is by no means the only food capable of meeting these needs. The demand
for herring, however, is not a simple derivative of needs. It is, as it were, a
function of a great many apparently independent variables. It is a function of
fashion, and by fashion is meant something more than the ephemeral results
I am grateful to Olav Bjerkholt, Sue Donnelly, Charles Goodhart, Vassilis Hajivassiliou, Javier Hidalgo,
David Hendry, Sue Howson, Richard Layard, Peter Phillips, Steve Pischke, Peter Robinson, Nigel
Rogers and Marcia Schafgans for helpful comments. Since I did not follow all of their suggestions, I
remain responsible for all sins of commission and omission.

J. Thomas (*) 
London School of Economics, University of London, London, UK
e-mail: J.Thomas@lse.ac.uk
© The Author(s) 2018
R. A. Cord (ed.), The Palgrave Companion to LSE Economics,

https://doi.org/10.1057/978-1-137-58274-4_1

3


4    
J. Thomas

of an Eat British Herrings campaign, the demand for herrings might be substantially changed by a change in the theological views of the economic subjects entering the market. It is a function of the availability of other foods.
It is a function of the quantity and quality of the population. It is a function
of the distribution of income within the community and changes in the volume of money. Transport changes will alter the area of demand for herrings.
Discoveries in the art of cooking may change their relative desirability. Is it
possible reasonably to suppose that coefficients derived from the observation
of a particular herring market at a particular time and place have any permanent significance – save as Economic History? (ibid.; italics in original).1

Robbins’s answer was negative and it is perhaps not surprising that there
was little interest shown in econometrics by members of the Economics
Department at LSE for some considerable time. The early econometric developments at the School came from two members of the Statistics
Department: A.L. Bowley and R.G.D. Allen.2
The foundation of the Econometric Society resulted largely from the
efforts of Ragnar Frisch over a considerable period of time. In 1926, when
he sought support for his plans, one of the four colleagues he contacted was
Bowley (see Louçã and Terlica 2011: 59). When the conference to establish
the society was held in December 1930, Bowley was one of the ten men
elected to serve on its Council (see ibid.: 63). When the first Econometric
Society meeting was held in Lausanne in September 1931, Bowley was
scheduled to present a paper, though he was forced to cancel it and did
not attend the meeting (see Bjerkholt 2015a: 1160). However, Bowley was
responsible for organising some of the early European meetings of the society. A further link came from Bowley’s collaboration with Allen in a study
of family expenditure, in which the Preface states that of the three converging sources of the investigation, ‘One is an article in Econometrica, 1935,

“The Action of Economic Forces in Producing Frequency Distribution, etc.”
followed by an unpublished communication to the Econometric Society in
1934 on “The Variation of Expenditure”’ (Allen and Bowley 1935: v).
Allen’s links with the Econometric Society were also developing: in 1934,
he published two articles in Econometrica (Allen 1934a, b), and in 1935, he
was elected a Fellow of the society. Things progressed further and in 1937,

1The extent of Robbins’s knowledge of statistical theory is discussed in Thomas (2009: 411–412), where
the evidence suggests that he did not take the theoretical section of the compulsory course in statistics
which he should have attended as an undergraduate when studying at LSE.
2See Chapter 8 in this volume for an evaluation of Bowley and Chapter 20 for an assessment of Allen.


1  LSE and Econometrics    
5

when the Cowles Commission was looking for a new Director of Research,
Allen was proposed as one of the candidates:
The names considered during or after the [1937 Cowes Commission Research]
conference were in addition to Tinbergen, Marschak and Yntema, also
R.G.D. Allen, Oskar N. Anderson, Ernest H. Phelps Brown, and Costantino
Bresciani-Turroni; other names may also have been mentioned … It can be
noted there seemed to be a premium on Europeans for the position as CC
research director (Bjerkholt 2013, 2015b: 15).

Furthermore,
Frisch held…Allen in very high regard. In October he informed Cowles that
Allen had got a one-year Rockefeller fellowship and advised Cowles to get in
touch with Allen and persuade him to spend time at the Commission, adding
“the more I have thought the matter over the more I have come to the conclusion that Allen is the man to consider as a candidate for directorship taken

everything into account”.
Frisch’s overall record for later years shows perhaps that he did not always
hit the mark in personal assessments but he held sound opinions in 1937–
1938 … He tentatively concluded to Cowles that he placed Allen, Marschak,
and Tinbergen above the others. He also threw in the name of Phelps Brown
but the proposal was not pursued (ibid.: 16).

In July 1938, Allen and his wife attended the Cowles Commission
Annual Research Conference on Economics and Statistics in Colorado
Springs, where Allen presented two papers, the first entitled ‘Some Statistical
Measures of Labour Mobility in England’ and the second ‘Patterns of Family
Expenditure: The Effect of Social Level and Family Composition’. Among
those present were Harold Davis, Elmer Working, Abraham Wald, Gerhard
Tintner, Henry Schultz and Alfred Cowles.3
Meanwhile, in the LSE Calendar 1935/1936, Allen offered a course of 15
lectures under the title ‘Some Problems in Econometrics’. The details were:
SYLLABUS—The first part of the course consists of an account of the main
statistical methods used in the description and analysis of economic phenomena. The treatment is largely non-mathematical, and the essential mathematical notions are put as simply as possible.
3See

Cowles Commission for Research in Economics (1939). I am grateful to Sue Howson and Olav
Bjerkholt for bringing Allen’s connections with the Cowles Commission to my attention.


6    
J. Thomas

The second part is concerned with some particular problems in econometrics, with the testing of theoretical constructions and the evaluation of fundamental economic concepts. The topics considered include the deduction
of elasticities of demand and supply from market data, the analysis of family
budget collections and the measurement of the cost of living (LSE 1935: 112).


In 1936–1937, the same course was on offer, but now consisting of ten lectures, whereas it was back to 15 lectures in 1937–1938. In 1938–1939, it
was listed, but now as ten lectures to be given in 1939–1940. Finally, in
1939–1940, it had become nine lectures to be given in 1940–1941 and
there was a note that it was to be given in alternate years. As the war took
Allen off to official duties elsewhere, the course disappeared. However, the
Calendar 1939/1940 listed a new course to be taught by Allen:
The Econometric Approach to Business Cycle Problems, consisting of nine lectures. The details were:
SYLLABUS—The course will be concerned with an exposition of recent
work by Tinbergen, Frisch and others on econometric business cycle research.
The emphasis will be laid as much on the statistical methods used and the
nature of the ‘dynamic’ economic relations involved as on the conclusions
reached in the testing of theories of cyclical fluctuations.
BOOKS RECOMMENDED—Tinbergen, Econometric Approach to Business
Cycle Problems, ‘Einige Grundfragen der mathematischen Konjunkartheorie’
(Archiv für mathematische Wirtschafts-und Socialforschung, 1937), ‘On the
Theory of Business Cycle Control’ (Econometrica, 1938), A Statistical Testing
of Business Cycle Theories, Business Cycles in the U.S.A., 1919–1937; Frisch,
‘Propagation Problems and Impulse Problems in Dynamic Economics’
(Economic Essays in Honour of Gustav Cassel ) (Given in alternate years.)

The lectures were scheduled for October 1939.4
At the onset of the Second World War, LSE was evacuated to be accommodated at Peterhouse College, Cambridge, and many academics, including, as noted, Allen, left to undertake war service.5
4Despite the close links reported above between the statisticians at the Cowles Commission and the
Econometric Society, it seems that the first LSE academic to publish an article in Econometrica with
the word ‘econometric’ in the title was an economist, Victor Edelberg (Edelberg 1936). A second article with ‘econometric’ in the title was Edelberg (1940). Edelberg was appointed as an Assistant (what
would today be called a Teaching Assistant) in the Economics Department in 1935, but in the early
1940s he had a severe mental breakdown from which he never really recovered (see Howson 2011: 254
and Thomas, forthcoming).
5The outflow of academics from both institutions meant that teaching duties fell to a relatively small

number of teachers, and as a result, students from LSE and Cambridge shared courses. According to the
Calendar 1942/1943, one of the shared courses offered in the summer term by Rothbarth was a course
of ten lectures under the title ‘Introduction to Econometrics’. The course was offered for a second time


1  LSE and Econometrics    
7

Allen returned to LSE after the war, and in the 1946–1947 session, he
taught a ten-lecture course on ‘Problems of Econometrics’ that was essentially his 1939–1940 business cycle course with the addition of Leontief ’s
input–output analysis. The course was repeated in the 1947–1948 session,
but then discontinued and econometrics disappeared from the LSE syllabus
for a number of years.6
In the 1951–1952 session, a 10-lecture course appeared taught by
Geoffrey Penrice of the Statistics Department and entitled ‘Introduction to
Econometrics’. This course was provided as an option for students specialising in Statistics in the BSc(Econ), with the following syllabus: ‘Scope of
Econometrics. Derivation of Supply and Demand curves by regression analysis and simultaneous probability equations. Production and Consumption
functions. Problems of identification and aggregation. Connection between
micro-economic theory and macro-economic models. Problems of obtaining
suitable statistical data’ (LSE 1951: 353).
By the following session, the course had expanded to 20 lectures and
was being taught by Mr. Booker and Dr. Morton.7 The following session,
it expanded again to 24 lectures and Booker and Morton were joined by
W.J. Corlett, an econometrician from University College London (UCL).
The course now settled down and continued in this format until the 1959–
1960 session, though with game theory and linear programming appearing
in the syllabus.8 The following year, the teachers were listed as ‘Mr. Corlett
and others’ and the syllabus reverted to its original 1951 format, with game
theory and linear programming being dropped.9


in the summer term of the following session (LSE 1942, 1943). Erwin Rothbarth came to LSE with
one of the first bursaries for students who had been displaced from Germany, graduated in 1936 and
carried out research at the school until 1938, when he moved to Cambridge as a research assistant to
Keynes working on national income statistics. He returned to Cambridge from internment to teach
but volunteered for the British Army and was killed in Holland in November 1944. So far, I have
not been able to obtain any information on what was taught on the Rothbarth course. However, perhaps one may speculate from the fact that he reviewed the second volume of Tinbergen’s League of
Nations study (Rothbarth 1941) and, in an obituary of Rothbarth, the authors wrote: ‘But his most
striking achievement was perhaps that he made himself a master of the new mathematical technique of
writers of econometrics, such as Slutsky, Frisch, Koopmans and Tinbergen, and adapted their methods
for his own work’ (Champernowne and Kaldor 1945: 131). See also Cuyvers (1983, 1983–1984) and
Toporowski (2013: 122–124).
6For an account of Allen’s important contributions to economic theory, see Chapter 20 in this volume.
7In 1952, Penrice left LSE for a distinguished career in the UK Civil Service, the OECD and the IMF.
8Dr.

George Morton was a mathematician who taught game theory and linear programming. It is interesting to note that the proportion of the reading list devoted to those two topics expanded during the
period when he was involved in teaching the course.
9See Gilbert (1989) for a further discussion of the teaching of econometrics during this period.


8    
J. Thomas

Another important contribution from the Statistics Department was
made under the direction of Professor Maurice Kendal who was interested in
the analysis of time series and encouraged James Durbin to come to LSE.10
Durbin provided a number of courses that were important to the development of time series econometrics.

2The Contribution of A.W.H. ‘Bill’ Phillips
Bill Phillips played an important role in getting the Economics Department

to focus on the need to increase the mathematical and statistical content
in its provision of teaching and research. Coming from an engineering
background, he also brought a different approach to the interpretation of
macroeconomic models of the economy, by seeing the need to treat them
as dynamic systems, rather than analysing them through the conventional
comparative statistics. The Phillips Machine provided a dynamic simulation
of a macroeconomy that could trace out over time the actual paths of adjustment of key variables to changes in parameters, such as changes in interest
rates, the money supply or government expenditure.11
Phillips saw that the feedback mechanisms that engineers used to control
dynamic systems (proportional control—the action taken is proportional in
magnitude and opposite in sign to the error to be corrected; integral control—the action taken is proportional in magnitude and opposite in sign
to the cumulated error up to that time; and derivative control—the action
taken is proportional in magnitude and opposite in sign to the rate of
change of the variable to be controlled) had potential similarities to the economic policies used by the government and/or other agencies in an attempt
to control the economy. In two theoretical articles (Phillips 1954, 1957), he
explored the theory of optimal control in dynamic models. The innovations

10James Durbin and Denis Sargan were both undergraduates at St. John’s College, Cambridge, in the
1940s. Durbin was invited back to Cambridge in 1948 by Richard Stone to work in the Department
of Applied Economics. At the time, there was a good deal of research going on there on time series
problems, with Guy Orcutt and Donald Cochrane working on their test and transformation for dealing
with first-order (AR1) autocorrelated errors. Durbin worked on the problem with Geoffrey Watson,
and this led to the development of the Durbin–Watson test. His interest in time series problems led to
Durbin being appointed as an Assistant Lecturer in the Statistics Department at LSE and his involvement with econometrics teaching and research. For a full account of Durbin’s work at LSE, see Chapter
25 in this volume.
11See Chapter 23 for more information on Phillips and the Phillips Machine. For an excellent biography of Phillips and a non-technical account of his research and importance, see Bollard (2016). See also
Phillips (2000).


1  LSE and Econometrics    

9

were: (i) that policy should not be thought of in a static mode but rather in
a dynamic mode; (ii) policy is best thought of in terms of rules; (iii) it is very
hard to assess the interaction of policy and system dynamics; and (iv) some
useful observations about the nature of policy were presented as a result of
the simulations performed (see Pagan 2000: 130–131).
At the time, most econometricians who were interested in macroeconomic modelling followed the Cowles Commission’s methodology and were
estimating the parameters in a set of simultaneous equations using data
collected discretely. Phillips’s approach faced a new problem: Theoretical
models were formulated in continuous time, but the data were collected at
discrete points of time. This problem was addressed in a number of publications (see Phillips 1956, 1959; Phillips and Quenouille 1960), with this
work having important implications for later econometric developments at
the School.12
In addition to the gradual increase in the teaching of econometric theory and mathematical economics at LSE, there was one interesting development that took place with respect to applied econometrics in the form
of the establishment of a seminar to formulate models and submit them to
statistical testing.

3The Methodology, Measurement
and Testing (M2T) Seminar
The early development of applied econometrics at LSE resulted from a
negative reaction by a number of the younger members of the Economics
Department to the arguments put forward in Robbins’s Nature and
Significance. The reaction has been well summarised by the leader of the
group, Richard Lipsey (2009: 845).13
The first stage in the development of the group was methodological.
Group members spent time in discussions with Dr. Joseph Agassi, a junior
member of the Philosophy Department, concerning Karl Popper’s philosophy of science (see De Marchi 1988). The methodology that emerged from
these discussions was, in simple terms, that, to be taken seriously, models
12While continuous time modelling was largely ignored elsewhere, Rex Bergstrom was influenced by

Phillips in his work in this area and both Peter Phillips and Clifford Wymer worked on continuous
time modelling as PhD students at LSE (see Mizon 1995).
13Other members of the group were Chris Archibald, Bernard Corry, Kurt Klappholz, Kelvin Lancaster,
Maurice Peston and (later) Max Steuer.


10    
J. Thomas

should make testable predictions and that econometrics should be used to
test these predictions. This methodology led on to the two other components of the Seminar’s title: the need for relevant economic data to enable
testing (measurement) and then the testing itself. The group’s methodology
was basic not only in their own research but also in their reaction to the
research of other economists: a number of visitors to the Seminar who came
to present a model based on ‘plausible’ assumptions were nonplussed on
being given an M2T grilling and being told that the model being discussed
did not seem to predict anything that could be tested.
Not all members of the group were concerned with all three of these
components. For example, Kurt Klappholz was mainly interested in methodology (see Klappholz and Agassi 1959; Klappholz and Mishan 1962),
while Chris Archibald concentrated his research on investigating whether
a selection of mainstream models actually yielded testable predictions (see
Archibald 1960, 1961). Others carried out empirical analyses, notable examples being Dick Lipsey’s re-estimation of the Phillips curve (Lipsey 1960,
2000) and Lipsey and Max Steuer’s attempt to test a model of inflation proposed by Kaldor (see Kaldor 1959; Lipsey and Steuer 1961).14
The M2T Seminar had a gradual impact on the teaching of applied economics: whereas in the 1950s courses in applied economics had involved
plotting economic data against time and telling plausible stories about their
movements, courses now began to present the results of empirical testing
and discuss statistical significance and goodness of fit.
In its purest form, the M2T Seminar began to change in the mid1960s with the departure from LSE of a number of the original members:
Archibald and Lipsey moved to the new University of Essex, Bernard Corry
and Maurice Peston went to Queen Mary College to meet an expansion

in the teaching of economics within the University of London, and Kelvin
Lancaster was a visitor at Columbia University in New York before deciding
to stay there.15 The Seminar continued as a general meeting for the presentation of empirical research for a number of years under the chairmanship of
Max Steuer, but lost its relevance and was discontinued after the development of alternative seminars within the specialised branches of economics.
14Kelvin Lancaster was an economic theorist who explored the possibility of making testable predictions
in qualitative economics (see Lancaster 1962), while Bernard Corry published empirical studies of the
labour market (see Corry 1961).
15The departure of this group of younger members of the staff led to a sense of disappointment on the
part of some of the MSc students and was reflected satirically in one of a number of similar items performed in a cabaret at one of the Staff-Student Weekend Schools held in the 1960s: ‘Where have all the
Great Men gone?’, was the lament, with Robbins being blamed for driving them away.


1  LSE and Econometrics    
11

4The Coming of the Econometricians
The 1960s saw dramatic changes in the composition of the Economics
Department. Robbins’s attempt to become Chairman of the Financial Times
while continuing as an academic was rejected, and this, together with his
appointment to chair the Commission on Higher Education, led to his
retirement from LSE in 1961. A number of the long-serving senior members
of the Department also retired in the years that followed (dates in parentheses show the period the professor spent in the Economics Department):
Sir Arnold Plant (1930–1965), Frank Paish (1932–1965), Sir Henry Phelps
Brown (1947–1968) and Richard Sayers (1947–1968).16
Following Robbins’s retirement, there was a short period of confusion during which it was unclear who the head of the department was,17
but the situation was swiftly clarified by a reorganisation of the School’s
administration that led to the appointment of departmental Convenors.
Convenors were to be heads of departments, but on a rotating three-year
basis rather than for indefinite periods. The first Convenor of the Economics
Department was Ely Devons, and, having consulted widely among the staff

in the Department, he set out to attract new staff and was remarkably successful, so that by the late 1960s the Department had been joined by Rex
Bergstrom,18 Denis Sargan,19 Harry Johnson, Frank Hahn and Terence

16See Backhouse (1997) for a discussion of changes that took place at LSE in the context of developments at other UK universities. He notes (p. 44) that Birmingham University introduced an
MSocSc degree in 1952, which predated the developments at LSE. The driving forces at Birmingham
were Terence Gorman (there from 1949 to 1962), Frank Hahn (1948–1960) and Alan Walters
(1952–1968). They had already transformed undergraduate teaching at Birmingham, with courses in
‘Mathematical Economics and Econometrics’ and a compulsory individual ‘Quantitative Economics
Project’ for third-year students. All three later moved to LSE.
17My irreverent comment at the time was that it reminded me of the confusion in the Politburo after
the death of Stalin.
18Rex Bergstrom was an econometrician from New Zealand who came to LSE as a Reader in
Economics in 1962. He returned to Auckland University in 1964 as Professor of Economics, and, when
he returned to the UK in 1970, it was to the University of Essex, where he remained until 2005. While
at LSE he began his research on continuous time econometrics, but as most of his research in this area
was carried out after he left the School, it will not figure in this discussion of LSE’s contribution to
econometrics. See Phillips (1988a, 2010) for a full discussion of Bergstrom’s research and publications.
19Durbin played an important part in these appointments. In his ET Interview, he states that: ‘Bill
Phillips and I cooperated in getting two new posts at the Readership level at the school: one in the economics department and one in the statistics department. Rex Bergstrom took the post in the economics
department for a time and we persuaded Denis Sargan to come from Leeds to the post in the statistics department. Soon afterwards Bergstrom left and Denis migrated to the economics department as a
Professor of Econometrics’ (see Phillips 1988b: 135). For an evaluation of the influence of Bill Phillips
on econometrics, see Hendry and Mizon (2000).


12    
J. Thomas

Gorman.20 Sargan, who arrived at LSE in 1963, summarised the importance
of these appointments:
In the econometrics field, Gorman and Hahn were unique in moving to

LSE from Oxford and Cambridge, where they were already well established.
Gorman was already professor at Nuffield College, Oxford, in 1966, and
Hahn a leading member of Churchill College, Cambridge. They were friends
of long standing; they, Alan Walters and John Wise were sometimes referred
to as the “Birmingham mafia” as they had all been teachers or students at
Birmingham in the early 1950s. Undoubtedly, Hahn and Gorman only moved
to LSE to be reunited together in the hope of founding a pioneering school of
econometrics and mathematical economics, at a time when both Oxford and
Cambridge were uninterested in these fields. They, together with Bill Phillips
and myself, provided the critical mass sufficient to attract good students and
faculty from all over the world to take and teach our courses, particularly in
our new MSc in Econometrics (Sargan 2003: 433).

Further impetus for change came through changes in degrees at both the
undergraduate and graduate levels. In the process of revising the BSc(Econ)
degree, which had provided a limited number of courses in mathematical
economics and econometrics, a new degree, the BSc(Econometrics and
Mathematical Economics), was introduced that offered much more specialisation in those subjects. At the level of the MSc, LSE introduced new
one-year taught degrees in 1964, an MSc(Econ) and an MSc(Econometrics
and Mathematical Economics). There were conversion courses to enable
undergraduates in other disciplines to enter these new degrees, with the
result that there was soon a flow of students from science backgrounds with
advanced mathematics coming through into the MSc(Econometrics and
Mathematical Economics). The final change of interest was in the structure
of the PhD, where the Economics Department first persuaded the rest of
the School and then the University of London that, as an alternative to the
traditional doctorate in which the degree was awarded on the basis of a single magnum opus of 75,000 words, it was possible to submit a number of
shorter (and not necessarily connected) pieces of work that demonstrated the
required level of originality. These changes produced a flow of outstanding


20Gorman made an important contribution to the development of the BSc(Econometrics and
Mathematical Economics) degree by arguing for the inclusion of a compulsory individual ‘Quantitative
Economics Project’ for third-year students, as at Birmingham. He insisted on ‘Quantitative’ rather than
‘Econometric’, this allowing for a much wider range of statistical techniques to be used than simply
conventional econometric methods.


1  LSE and Econometrics    
13

PhD students in econometrics in the late 1960s. The excitement of the
period is described by David Hendry, the leading member of this group of
Sargan’s students:
The student rebellion at LSE was at its height in 1968–1969, and most of
Denis’s students worked on the computer at UCL, an ocean of calm. It was
a wonderful group to be with. Grayham Mizon wrote code for optimization
applied to investment equations, Pravin Trivedi for efficient Monte Carlo
methods and modelling inventories, Mike Feiner for “ratchet” models for
imports, and Ross Williams for nonlinear estimation of durables expenditure. Also, Cliff Wymer was working on continuous-time simultaneous systems, Ray Byron on systems of demand equations, and William Mikhail on
finite-sample approximation. We shared ideas and code, and Denis met with
us regularly in a workshop where each student presented his or her research.
Most theses involved econometric theory, computing, an empirical application, and perhaps a simulation study (Hendry 2003: 750–751).21

5Sargan, Hendry and the LSE Tradition
in Econometrics22
To understand the importance of the econometric methodology developed at
LSE, it is helpful to remember the poor quality of many of the applied econometric studies undertaken in the 1960s. Most of the models being investigated were single equations, to be estimated from time series data. There
were some data constraints imposed by short time series, but more important were the severe computational constraints of a pre-computer era in which
computation involved mastery of the Doolittle method of matrix inversion
using desktop electric calculating machines.23 A typical article might involve

Section 1: the discussion of an economic theory that leads to Eq. (1):
Yt = β0 + β1 X1t

(1)

with the prediction that 0 < β1 < 1. Section 2 involves the testing of the theory, so an error term is added to (1) to yield (2) and it is assumed that the
21The

recollections of another LSE econometrics student from this time are presented in Spanos (2014).
technical details, see Chapter 27 below, Gilbert (1986) and Hendry (2009). See also Hendry
(1980), Phillips (1985), Pagan (1987), and Hansen (1996).
23During the 1950s and early 1960s, LSE economists were fortunate to have available the skills of June
Wickens in this capacity, but when she married and moved to Bristol there was a considerable fall in the
productivity of the Economics Department.
22For


14    
J. Thomas

ut is independently and identically normally distributed with a mean of zero
and a constant variance:
Yt = β0 + β1 X1t + ut

(2)

Some time series data are now introduced with a perfunctory discussion
of their construction and relevance and (2) is estimated by ordinary least
squares.
The main emphasis in evaluating the results was on the goodness of fit

(high value of R 2), the correct signs and the statistical significance of the estimated coefficients. It was also a requirement that the researcher checked the
Durbin–Watson (DW) statistic to test for the presence of first-order (AR1)
autocorrelated errors. If the test suggested the presence of AR1 errors, then
the Cochrane–Orcutt transformation would sometimes deal with the problem, but in those cases where this was not possible, there was no available
alternative methodology to provide a solution. What tended to happen was
that the researcher would add more variables to the equation in the hope
that with enough ‘data mining’ an acceptable result might emerge that got
rid of the autocorrelation problem:
Yt = β0 + β1 X1t + β2 X2t + . . . + βk Xkt + ut

(3)

A non-technical description of the LSE econometric methodology is as
follows:
(1)In contrast to the implicit assumption in Eqs. (1) and (2) that the
economic processes adjust rapidly and are complete within a single
time period, the LSE alternative assumed that adjustment processes
might be slow and lagged values of the variables (both dependent
and independent) might be necessary. In addition, autocorrelation
was seen not as a purely statistical problem but rather the effect of
misspecification when lagged values of the variables were omitted.
(2)A second important difference is a respect for the economic time
series being used: rather than merely providing data for a ‘test’ of a
theory, the series contain important information on the process of
economic adjustment. Hence, the objective is to develop a model
that could have generated the data being observed.24

24This is the idea underlying the so-called data generating process (DGP). See Gilbert (1986) for the
technical details.



×