VICIOUS AND VIRTUOUS CIRCLES-THE POLITICAL ECONOMY OF UNEMPLOYMENT IN INTERWAR UK AND USA

CARDIFF BUSINESS SCHOOL
WORKING PAPER SERIES
Cardiff Economics
Working Papers
Kent Matthews, Patrick Minford and Ruthira Naraidoo
Vicious and Virtuous Circles - The Political Economy of
Unemployment in Interwar UK and USA
E2006/25
Cardiff Business School
Cardiff University
Colum Drive
Cardiff CF10 3EU
United Kingdom
t: +44 (0)29 2087 4000
f: +44 (0)29 2087 4419
www.cardiff.ac.uk/carbs
ISSN 1749-6101
January 2006, updated November 2006
This working paper is produced for discussion purpose only. These working papers are expected to be published in
due course, in revised form, and should not be quoted or cited without the author’s written permission.
Cardiff Economics Working Papers are available online from: http://www.cardiff.ac.uk/carbs/econ/workingpapers
Enquiries: EconWP@cardiff.ac.uk

---

VICIOUS AND VIRTUOUS CIRCLES - THE POLITICAL ECONOMY OF
UNEMPLOYMENT IN INTERWAR UK AND USA
Kent Matthewsa∗, Patrick Minfordb, Ruthira Naraidooc
Revised November 2006
Abstract
This paper develops a political economy model of multiple unemployment equilibria to provide a theory
of an endogenous natural rate of unemployment. This model is applied to the UK and the US interwar period
which is remembered as the decade of mass unemployment. The theory here sees the natural rate and the
associated path of unemployment as a reaction to shocks (mainly demand in nature) and the institutional
structure of the economy. The channel through which these two forces feed on each other is a political
economy process whereby voters with limited information on the natural rate react to shocks by demanding
more or less social protection. The reduced form results obtained confirm a pattern of unemployment
behaviour in which unemployment moves between high and low equilibria in response to shocks.
Keywords: Equilibrium unemployment, political economy, “vicious” and “virtuous” circles, bootstrap-
ping, forecasting
JEL classification: E24; E27; P16
aCardiff Business School, Aberconway Building, Cardiff University, Column Drive, Cardiff, Wales, United Kingdom, CF 10
3EU. Phone +44-2920875855, Fax +44 (0)29 2087 4419. E-mail: MatthewsK@cardiff.ac.uk. ∗Corresponding author.
bCardiff Business School, Aberconway Building, Cardiff University, Column Drive, Cardiff, Wales, United Kingdom, CF 10
3EU. Phone + 44 (0)29 2087 5728, Fax +44 (0)29 2087 4419, E-mail MinfordP@cardiff.ac.uk.
cKeele University, Keele, ST5 5BG. Phone +44 (0) 1782 583107, Fax +44 (0) 1782 717577. E-mail: r.naraidoo@econ.keele.ac.uk.
The authors would like to thank Robin Bladen-Hovell, Damian Damianov, David Meenagh and workshop participants, and
in particular an anonymous referee, for helpful comments.
1

---

1. Introduction
Much of postwar unemployment policy has its roots in the experience of mass unemployment in the
interwar period. Unemployment in both the USA and UK reached levels that hitherto had never been
experienced in either economy. In the UK, the vision of the ‘army of one million unemployed ’ men after world
war one had powerfully etched itself into the collective memory of the electorate and the politico-economic
establishment. Not until the economic reforms of Mrs Thatcher did the UK economy, and importantly the
voting public, wean themselves off the postwar thinking of demand management. Between 1929 and 1933
in the USA, unemployment soared to 25 per cent of the workforce. The political economy that underscored
Roosevelt’s ‘New Deal’ remained with the economic establishment until the Reagan tax reforms of the 1980s.
The puzzle for the USA is that despite the best efforts of the New Deal, unemployment remained stubbornly
high and refused to fall below 15 per cent. The puzzle for the UK was that during the half century before
the First World War, real income grew at a rate of 2 per cent a year but unemployment averaged less than
4 per cent; from 1920 to 1938, real income continued to grow by about the same rate, but unemployment
averaged 14 per cent.
This paper argues that shocks causing sharp cyclical demand swings generate political reactions from
public opinion and vested interests, which in turn produce fiscal and monetary (demand policy) responses and
also changes in supply-side policy, i.e., policy affecting the equilibrium values of real variables or ‘natural
rates’. Specifically, bad demand shocks tend to produce supply policy that distorts the market because
these shocks generate demands for protection; these distortions in turn produce an equilibrium with a higher
natural rate of unemployment which in turn can reinforce the demands for yet more protection, until matters
are bad enough to create a political equilibrium where the bad effects cause enough opposition to yet more
distortions.
Vice versa, a good run of demand shocks produces more liberal supply-side policies as people are less ner-
vous about potential misfortune. This again is self-reinforcing so that the economy moves in a virtuous circle
to a low-unemployment high-output equilibrium. The dynamics of unemployment exhibit three equilibria-
one stable low unemployment equilibrium, one stable high unemployment equilibrium and a non-stable
intermediate unemployment equilibrium that lies between the two.
The model of political economy of supply-side policies in this paper is similar to Wright (1986). In
Wright (1986), workers have different unemployment risks, and unemployment benefits play the sole role of
an insurance against adverse shocks. The median voter will optimally determine unemployment benefits by
weighing their benefits (in terms of better insurance) against their costs (in terms of higher taxes). Thus,
the more exposed to unemployment is the median voter, the higher the political support for unemployment
2

---

benefits, much as in Meltzer and Richard (1981) the median voter’s support for redistributive taxation varies
with the state of the economy. However as noted by St-Paul (1996), this prediction neglects other effects
of unemployment benefits to the extent that higher wages lower job creation and harm the unemployed.
Therefore, a higher exposure of the employed to unemployment will tend to moderate the desire of the
median voter for a high benefit level. St-Paul also analyses how labour market institutions can affect the
median voter’s choice.
The political economy model of institutions in this paper extends the analysis of Meltzer and Richard
(1981), Wright (1986) and St-Paul (1996). Long-lasting shocks to the economy lead to demand for social
protection (identified with the benefit/wage -or replacement- ratio). However, we also take into account
the feedback distortionary effect of benefits on unemployment which progressively raises the chances of
unemployment for the median voter. We therefore address the missing channel of Wright (1986) of how
labour market institutions can affect the welfare of the decisive voter, as suggested by St-Paul (1996).
This paper joins a large literature on the creation and evolution of the institutions that favour or inhibit
capitalist growth (see, for example, Persson and Tabellini, 1994; Alesina and Roderick, 1993; Perotti, 1993;
Stokey and Rebelo, 1993, and Krusell et al, 1994). It also joins a set of other studies explaining the rise
in unemployment since the 1970s in terms of macroeconomic shocks interacting with institutional patterns
(see Blanchard and Wolfers, 2000; Fitoussi et al., 2000; Bertola et al., 2001; and Nickell et al., 2005).
Our contribution here (and in Minford and Naraidoo, 2004, for postwar unemployment) is to view these
interactions as the product of a political economy process.
This paper is organised along the following lines. The next section sketches the theoretical framework
that produces circles of a vicious or virtuous nature. Section 3 embeds the framework within the historical
context of the interwar period. Section 4 estimates the reduced form dynamic unemployment function which
is capable of producing a three equilibrium unemployment case consistent with the experience of the interwar
period. A bootstrapping procedure is included to test the prevalence of the 3-equilibrium case. In section
5 we test the reduced form model against a general model of hysteresis by providing out-of-sample forecast
accuracy tests. Section 6 concludes.
2. Vicious and Virtuous Circles
This section sketches the theoretical framework that produces vicious and virtuous circles in unemploy-
3

---

ment1. The basic idea is that the natural rate of unemployment is an endogenous variable that gets pushed
from a low position to a high position by the interaction of demand shocks and institutional responses. These
two forces are interlinked by a political economy mechanism whereby voters with limited information on the
natural rate of unemployment react to negative shocks by demanding protection which creates the vicious
circle.
Bad situations can generate political demands for reform which can cut into a vicious circle (one example
is Margaret Thatcher’s reforms in the UK during the eighties and in the USA the Carter deregulation of the
1970’s and the Reagan tax reforms of the 1980’s); and equally that very long periods of good fortune can
also generate a complacent attitude towards distortions whereby for example it is felt that ‘social justice’ can
be ‘afforded’ through higher social protection, worker rights and so on. These political developments act as
an additional source of shocks, over and above demand shocks, that can start a move towards respectively
a good or bad equilibrium. There are intimate linkages through political economy between the two sorts of
policies, namely, demand-management and supply side policies, and these links have the capacity to create
both vicious and virtuous circles of economic performance.
A good example is the USA interwar period. In the 1920s it was a high-performing economy with
full employment but in the 1930s it was in deep slump leading to heavy regulatory intervention and the
protectionism that made matters worse for all. Another case is Japan from the mid-1950s to the mid-1970s,
and then again from 1990 onwards. A miracle of rapid, unstoppable growth in the first and a hopeless
stagnation in the second. A further case is continental Europe which appeared to do nothing wrong in the
1960s and 1970s and yet, the role was reversed in the 1980s and 1990s.
Getting out of this state requires either a big positive shock to output and employment (for the USA; the
Second World War) or a determined broad-based reform programme (the UK in 1979), which revives popular
faith in market economics. Once popular confidence is built up, the demands for protection evaporate and
in their place, people and business demand freedom. Then, the virtuous circle operates and the economy
settles at a high-employment equilibrium.
The theory consists of two components: a model of the natural rate and a model of the political economy
of supply-side policy (particularly towards the central labour market). We discuss each in turn.
As outlined by Siebert (1997), starting from a simple notion of an equilibrium in a classically clearing
labour market, institutional arrangements can influence the clearing function of the labour market in basically
three ways: by weakening the demand for labour, making it less attractive to hire a worker by explicitly
pushing up wage costs or by introducing a negative shadow price for labour; by distorting labour supply; and
1For a fuller exposition see Minford and Naraidoo (2004)
4

---

by impairing the equilibrating function of the market mechanism (for instance, by influencing bargaining
behaviour). To these interventions may be added those of trade unions (see Blanchard and Katz, 1997).
Our set-up for the labour market follows that of Minford (1983). He took the classical labour supply
set-up and added the idea of a permanent unemployment benefit, payable without check on work availability.
The result was to tilt the labour supply curve so that the real wage offer never fell below the benefit level,
as shown in Fig. 1. Should the benefit level rise relative to productivity, unemployment will increase.
That is, people will voluntarily refuse to take available wage offers because benefits are preferable; they are
‘unemployed’ in the sense that they are not working but are ‘available for work’.
Real
Wages
Normal
Labour
Supply of
Force
Labour
Marginal
Product of
Real
Labour
Unemployment
Natural Rate of
Benefits
Unemployment
Employment

Figure 1: The Natural Rate of Unemployment
These ideas can be summarised in the following model of structural unemployment2. In our analysis we
focus purely on benefits, because this will be the choice variable for voters under our political economy model
below; such things as taxation and public expenditure, union power, and minimum wages are also potential
choice variables but for simplicity we leave them out of the explicit model,
2Later versions have proliferated; in the UK, Layard and Nickell (1986) estimated a similar model, and Bean et al. (1986)
attempted to extend it to other European countries which began to experience rising unemployment UK-style during the late
1980s and 1990s. It turns out that in each country there are substantial idiosyncracies in the social support mechanisms,
complicating effective modelling of the natural unemployment rate.
In more recent empirical works, Nickell(1997), Nickell and Layard(1998) and OECD (1999) suggest that structural unemploy-
ment in major OECD economies is associated with the following labour market features: (a) generous unemployment benefits
that are allowed to run on indefinitely, combined with little or no pressure on the unemployed to obtain work, (b) high union-
isation with wages bargained collectively and no coordination between either unions or employers in wage bargaining and (c)
high overall taxes impinging on labour or a combination of high minimum wages for young people associated with high payroll
taxes.
5

---

B
lnU
t
t = u0 + δln(
) + u1t + uc
W
t
(1)
t
where u0 is a constant, Bt is the real benefit rate , Wt is real wages (set it is assumed by productivity), uct
is cyclical unemployment, and u1t represents other persistent influences on unemployment, an error process
assumed therefore to display high serial correlation. Examples of such influences would be demographic
shifts (such as a rise in working age population), and sectoral shifts like a decline in manufacturing. These
influences will have no long-run effect on unemployment but their effect is assumed to be long drawn out.
This effect provokes a political economy response. In our model, the median voter holds some non-
human capital but nevertheless relies heavily on income from human capital. If this voter experiences
unemployment spells, unemployment benefits yield a much needed replacement of wage income. The higher
level of unemployment means that agents are more exposed to the risk of being unemployed which therefore
increases their desired benefit/wage ratio. However, we also take into account the feedback distortionary
effect of benefit on unemployment which increases the probability of unemployment for the median voter.
Hence,.as unemployment rises, the median voter’s demands for benefits rise but at a diminishing rate, as
these higher benefits progressively raise the chances of unemployment.
Suppose we define an index, Bt, as the level and duration (overall ‘generosity’) of real unemployment
benefits; for short we shall refer to it as ‘benefits’ (the same as in Eq. (1)). Our analysis (for details see
Appendix A) implies that the log of the benefit/wage ratio is a quadratic in the rate of unemployment, Ut,
or say :
B
ln(
t ) = B0 + ϕ(Ut
W
−1 − z) − β(Ut−1 − z)2 + ²t
(2)
t
Initially a rise in unemployment above some normal rate, z, would trigger demands for higher benefits,
but as unemployment rises, the rising chances of unemployment become an increasingly restraining factor.
In Eq. (2), B0 is a minimum benefit/wage ratio set in normal circumstances and ϕ and β are constants.
Combining Eqs. (1) and (2) leads to the following log-linear dynamic unemployment model:
lnUt = (u0 + δB0 − δϕz − δβz2) + (δϕ + 2δβz)Ut−1 − δβU2t−1 + u1t + uct + δ²t
(3)
6

---

or more compactly:
lnUt = a0 + a1Ut−1 + a2U2t−1 + ξt
(4)
3. Interwar Unemployment
The period between the two world wars has been the source of much contemporary economic, social
and policy lessons. The conventional view that aggregate demand deficiency was the principal cause of
unemployment in interwar Britain was first challenged by Benjamin and Kochin (1979) who argued that part
of the explanation for the high level of unemployment was due to the generosity of unemployment benefits
relative to wages. While Benjamin and Kochin’s explanation has been hotly contested in the academic
literature3 , a consensus view is that supply as well as demand shocks mattered to varying degrees4. More
recently Benjamin and Matthews (1992) have challenged the perceived differences between the UK and US
experiences of interwar unemployment. While demand factors mattered a great deal in the case of the USA,
supply factors in the form of public relief contributed to the high level of unemployment that persisted up
until the eve of the second world war5. This section reviews unemployment policy in both countries between
the wars with the aim of demonstrating the reinforcing reaction of supply policy to negative demand shocks.
The earliest plan for a national unemployment insurance scheme in the UK (National Insurance Act
1911) was based on the assumption that unemployment was caused by temporary maladjustments in the
labour market. Unemployment benefit grounded on the principles of insurance together with the savings
of individual worker was to provide a short-term tide-over until re-employment. In his presidential address
to the British Association in 1910, Sir Hubert Lewellyn-Smith, co-author with Sir William Beveridge of the
unemployment section of the NI Bill stated:
“The scheme must avoid encouraging unemployment and for this reason the rate of unemployment benefit
shall be relatively low. It would be fatal for any scheme to offer compensation for unemployment at a rate
approximating to that of ordinary wages”. But it was precisely on this principle that Bakke (1935) noted
that the battle between insurance and maintenance was fought. It was the results of this battle that led
Beveridge in his testimony to the Royal Commission on Unemployment Insurance in 1931 to describe the
3See for example the special issue of the Journal of Political Economy 2, 90, 1982 devoted to the critics and the reply
Benjamin and Kochin (1982)
4For example Matthews (1986) argues that both supply and demand shocks mattered with balance in favour of supply
shocks, whereas Broadberry (1986) argues the reverse.
5See for example Wallis and Benjamin (1981).
7

---

process as “a development from insurance by contract to relief by status” and whereas in the past he referred
to “insurance popularly miscalled the dole” he went on to refer to the “dole officially miscalled insurance”6.
The unemployment insurance system in the UK that developed after the first world war was fundamentally
different from its original inception. The Unemployment Act of 1920 removed many of the former system’s
safeguards and increased benefits by nearly 40 per cent and extended the coverage to a further 11 million
workers. In the following decade, the expansion of the insurance system was achieved by relaxation of the
contributory requirement, the extension of the duration and the allowance of additional weeks of special
benefits. Several forces combined to put pressure on the system to push the unemployment benefit system
from insurance to maintenance. The immediate post world war one system of unemployment relief for
uninsured workers was the local authority administered system of poor law relief. The financial burden of
providing unemployment relief brought many a local authority close to bankruptcy. In 1923 the Minister of
Labour reported that he was impressed with the plea not only from the Poor Law Guardians, and Workmans
Associations but also from employers, that benefit be extended to the uninsured. The political position of
both the T.U.C. and the Independent Labour Party was ‘work or maintenance’. In 1924, the Minister of
Health received a deputation from the Poor Law Authorities complaining that the burden of unemployment
relief could not and was never intended to be met from the local authority finances7. By 1931, ‘need’
had completely replaced ‘insurance’ as a criterion for transitional and unemployment benefits. The Royal
Commission of 1932 stated:
“Each successive government has made changes to the scheme, which have been determined less by the
need for careful balancing of income and expenditure than by a desire to attract, or do as little as possible
to repel electoral support.”8
The relaxation of the insurance rules was also accompanied with a continuous increase in the average level
of benefit. The increase in benefit was even more pronounced when the cost of living was taken into account.
In its recommendation for a reduction in the level of benefits the Royal Commission on Unemployment
Insurance 1931 stated:
“If a comparison be made between that rates of benefit paid in August 1924, in 1928 and now proposed
by us, with adjustments corresponding to changes in the cost of living index...it will be clear that the rates
which we propose are in effect higher than those in operation in the more prosperous years of 1924 and
1928”. Appendix B provides graphs of annual data on unemployment benefits and average earnings series
6See Bakke (1935) p.17
7Bakke (1935) p.147
8Royal Commission 1932, p.164
8

---

for the UK9 for the period 1920-38, together with an explanation on their patterns.
Turning to the case of the USA, there was no centralised system of unemployment relief at the end of the
first world war. Up until the 1930s, unemployment relief was local government based combined with private
charitable organisations. The unprecedented unemployment and financial hardship that followed the ‘Great
Depression’ brought forth a considerable expansion of relief activities. But it was not until 1932 with the
establishment of the Reconstruction Finance Corporation (RFC) that the federal government began to play
a role in the relief. However this system was largely ineffective as it disbursed funds to private businesses
rather than directly to unemployment persons and required the condition that all locally funded relief be
exhausted.
In 1933 the newly elected President Roosevelt supplanted the RFC with the Federal Emergency Relief
Administration (FERA), which was to spend $3 billion on unemployment relief. FERA represented a funda-
mental change in philosophy. Its principal aim was to put financial assistance directly into the hands of the
unemployment through cash grants and work relief. Recipients of FERA were occasionally required to take
part in public works programmes but it was far more commonplace for no work requirements be made. The
FERA was phased out in 1935 which was replaced by the Works Progress Administration (WPA). This was
a programme that explicitly established a ‘nation-wide’ system of work relief. Individual relief was based on
‘needs’ levels and local pay which determined the number of work hours contributed to WPA programmes10.
However, those who were either unable to work or did not have suitable skills for WPA programmes were
eligible for general relief which was an unconditional cash grant11. This combination of work relief and
general relief remained in existence until the onset of the second world war. The conclusion of Benjamin
and Matthews (1992) in their analysis of the unemployment relief in the USA was that the FERA and WPA
programmes did prolong the manifestations of the Great Depression. In the relief programme is the expla-
nation why by 1939 when per capita real income was only 4 per cent below its 1929 peak, unemployment
was 17.2 per cent compared with 3.2 per cent in 1929. Table 1 summarises the minimum and maximum
values of each unemployment series, together with their mean, standard deviation, skewness and kurtosis.
What is remarkable about the data is its similarity in the mean and the volatility within each unemployment
series. However, the higher standard deviation for the USA indicates the greater volatility in the economy
during the 1930s. A look at the graphs of the two variables in Appendix C shows distinctive jumps in the
9The only country for which we have been able to gather some data.
10So for example if a person’s need level was evaluated at $60 a month and the local pay for that person’s occupation was
$0.60 per hour, the recepient would work 100 hours on the WPA programme.
11Benjamin and Matthews (1992) estimate a demand for unemployment relief by State which depended on the size of the
population, the proportion of farm population, proportion of black population, the growth of per capita state income and the
level of benefits. They conclude that for every 10 people taken off the unemployment queue through WPA relief, a further 9
joined it.
9

---