Download in Acrobat - University of Ottawa

file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
A COMPARATIVE ANALYSIS
OF THE POST-KEYNESIAN THEORY OF EMPLOYMENT
Marc Lavoie
Department of Economics
University of Ottawa
200 Wilbrod St.
Ottawa (Ontario)
CANADA K1N 6N5
FAX: (613) 562 5999
E-MAIL: [email protected]
December 1998
A sketch of this paper was presented at the seminar "El papel del empleo en la
teoria económica a finales del siglo", organized by Professor Clemente Ruiz
Durán, at the División de Posgrado de la Facultad de Economia, Universidad
Nacional Autónoma de Mexico, June 15-16, 1998. I am grateful for the comments
made during the seminar. I also received useful comments from Eduardo Loria Diaz
and Mario Seccareccia. The usual caveat applies.
A COMPARATIVE ANALYSIS
OF THE POST-KEYNESIAN THEORY OF EMPLOYMENT
Introduction
Widespread unemployment appears to be a major feature of the modern globalized
economy but, still, few economists seem to relate the actual levels of
unemployment to the well-accepted main cause of the Great Depression of the
1930s -- the lack of effective demand. For mainstream economists, high rates of
unemployment today are tracking the high rates of natural unemployment, or
rather as some prefer to put it, the high non-accelerating inflation rates of
unemployment (NAIRU). This high natural rate is said to be due to various
elements, but most signs point towards the behaviour of workers and the
intervention of government in the market place. Even in models that purport to
demonstrate that workers are not voluntarily unemployed, the failure of the
economic system to deliver full employment (net of frictional unemployment) can
be blamed on the behaviour of the workers, whose bargaining power on the labour
markets is reinforced whenever the employment situation improves. Hence,
ultimately, unemployment is caused by the lack of flexibility of nominal and
real wages.
By contrast, one of the crucial characteristics of post-Keynesian economics is
its emphasis on demand-led phenomena. For post-Keynesians, employment in the
labour market is essentially determined by effective demand on the goods market.
Employment depends on macroeconomic phenomena; microeconomic choices,
substitution effects and the like, are only of a second-order importance (see
Appelbaum 1979, Nell 1988, Seccareccia 1991a, Lavoie 1992 (ch. 5), Riach 1995,
King 1999).
Post-Keynesian economics shares many presuppositions with other heterodox
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (1 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
schools, when contrasted with the neoclassical school: the importance of
production (versus exchange and pure resource allocation in neoclassical
economics); a reasonable view of rationality, where knowledge is costly,
uncertain, and difficult to synthesize (versus the neoclassical view of absolute
rationality); an epistemology based on realism, whereby assumptions need to be
realistic and an explanatory story must be told (versus the instrumentalism of
neoclassical economics, whereby prediction is the main purpose); a methodology
based on some moderate form of organicism, where macroeconomics need not conform
to microeconomics and where class behaviour is relevant (versus methodological
individualism, based on the neoclassical representative agent) (Lavoie 1992, ch.
1).
Within this framework common to all heterodox schools, post-Keynesian economists
can be singled out for their emphasis on two features: the generality of
demand-led phenomena, as already pointed out, and the relevance of time. The
latter feature will not be underlined here, and will only be briefly mentioned.
Statements to the effect that the long period is but a succession of short
periods, as Michal Kalecki is often said to have claimed, are typical of this
post-Keynesian characteristic. Although all models constructed by
post-Keynesians do not abide by this dictum, their authors have usually built
long-run models under the understanding that they are only a first-step
approximation, until better models, that take historical time into
consideration, can be formalized. This implies that in post-Keynesian economics,
in contrast perhaps to some of the other heterodox schools, long-period
equilibria, if they exist, are known to be influenced by the path taken during
the traverse from one position to another. This means that disequilibria do have
an impact on the conceptual final equilibrium, in contrast to the standard
comparative static or dynamic analysis, where long-run equilibria only depend on
the values taken by the exogenous parameters, and where the reaction functions
during the transition have no impact on the long-run equilibrium that will
eventually be achieved (Setterfield 1995). This view of time and economics,
which was essentially argued but not formalized, until recently, by
Institutionalists and post-Keynesians, has now been promoted by some members of
the mainstream -- in particular one branch of the New Keynesian school,
sometimes called the post-Walrasian school (Colander 1996) -- under the notions
of multiple equilibria, path-dependence and hysteresis.
But even when hysteresis effects are taken for granted, these effects are
usually reduced to supply-side effects, with demand playing little or no role.
And if demand plays any role, it is restricted to the short run. By contrast,
post-Keynesians argue that demand-led economics is relevant, both in the
short-run and in the long run. This is in contrast to what is believed by most
New Keynesians and by many Marxists, who argue that insufficient aggregate
demand plays a role in the short period, but not in the long period, which would
remain supply-led (Duménil and Lévy, forthcoming).
These crucial characteristics of the post-Keynesian analysis are reflected in
its theory of employment. Mass unemployment is essentially seen as a problem
arising from a generalized deficiency of demand. The lack of employment,
according to post-Keynesian economics, has nothing to do with excessive real
wages or with a lack of wage flexibility. Indeed, in many versions of the
post-Keynesian model of employment, higher real wages are conducive to higher
levels of employment; and in the more straightforward models of growth and
distribution, they may be conducive to higher rates of economic growth.
The intent of the paper is to compare the various models of employment that can
be found in the orthodox and heterodox literature. The vital thread, that allows
for a comparison of all the models, is the role that real wages play in each
model. The survey only deals with models of employment in a closed economy. This
restriction can be justified on two grounds. First, open-economy models are
often built with the intention to duplicate the results obtained in the case of
the closed economy. It is therefore important to know whether the closed-economy
model is appropriate. Second, in a world of globalization, the world economy
taken as a whole may be considered to be a closed economy.
To highlight better the specific characteristics of the post-Keynesian models,
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (2 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
the first part of the survey deals with the standard neoclassical models of
employment. The second part presents post-Keynesian models with hybrid features
-- a Keynesian effective demand relation with a neoclassical production
function. The third part is devoted to the more radical post-Keynesian model -called the Kaleckian model -- with both demand-side and supply-side heterodox
features. In order to remain pedagogical, formal analysis will be highly
simplified. In the case of the neoclassical models, because they are usually
well-known, I will restrict myself mainly to a graphical analysis. In the case
of the post-Keynesian models, because their analyses are lesser known, some
simple equations will be provided to justify the results obtained. A table, with
a summary of the main features of the various models, is presented in appendix.
1. Mainstream analyses of aggregate employment
In mainstream economics, unemployment beyond frictional or temporary
unemployment is essentially due to some imperfection, to some rigidity or to
some lack of information, which sets the real wage at a level which is too high
compared to the equilibrium real wage that would provide for full employment.
There are various, sometimes similar, explanations of unemployment. We shall
look at each of them in turn: the new classical model, the post-Walrasian model,
the efficiency wage model of the New Keynesians, and the PS-WS
Layard-Nickell-Jackman model of wage bargaining.
1.1 The new classical model
Ever since students are taught principles of economics, a simple explanation of
unemployment is being offered. As in all other markets, there may be an excess
supply of some commodity when the price of that commodity is artificially set at
a level which is too high, i.e. when there is some price floor which exceeds the
equilibrium price. In the particular case of the labour market, an excess supply
of labour, and hence unemployment, arises when real wages are too high. This
occurs, we are told, because of the presence of labour unions or because of the
imposition by government of a minimum wage rate which impacts on the overall
real wage structure. Labour unions or minimum wage legislation, and more likely
a combination of both, drive the overall real wage to a value that exceeds its
equilibrium level. Generally speaking, the other more sophisticated mainstream
theories that explain unemployment attempt to recover the same relationship:
there is unemployment because real wages, for some reason or another, are too
high.
In the new classical model of aggregate supply, as came to be developed from the
work of Milton Friedman (1968) and Robert Lucas (1973), price expectations are
taken into consideration. The model is an instance of asymmetric information. It
is assumed that firms and entrepreneurs take their production and hiring
decisions on the basis of the prices that are actually realized on the product
market, whereas households and workers take their decisions about work versus
leisure on the basis of expected price. In the new classical model, unemployment
is fully voluntary. It arises when households and workers overestimate the price
level that will actually turn out to be realized on product markets. Because the
expected price level is overestimated, households ask for higher nominal wages
than those that would be demanded if the price level was correctly anticipated.
As a result, the realized real wage rate is above its full equilibrium level.
Still, the labour market, in some sense, is in equilibrium: firms hire less
labour because the realized real wage rate is too high, while households provide
less labour at this higher real wage rate, because, on the basis of their price
expectations, the perceived real wage rate is lower than the real wage rate that
would provide for full employment. Unemployment, in the new classical model,
arises from mistaken expectations, more specifically a price level which is
overestimated, which happens either because of changes in technology, as in the
real business cycles approach, or because of badly forecasted aggregate demand.
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (3 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
Figure 1
In Figure 1, the realized price level is P1, but the price level expected by
households is P2, where P2 > P1. The wage level that would be consistent with
full employment Lfe is thus w1, whereas the realized wage rate will be w2. The
realized real wage, w2/P1 is thus higher than the full-employment equilibrium
real wage w1/P1. However, at this higher realized real wage, more workers are
not forthcoming and willing to work, because workers perceive the real wage rate
to be only w2/P2, which is lower than the real wage w3/P2 that would be required
for a full amount Lfe of workers to offer their services. At the perceived real
wage being offered, only L2 workers wish to work. All of them will be hired, and
hence those not hired are voluntarily unemployed. Still, there is unemployment
because real wages are too high.
1.2 The post-Walrasian model
The post-Walrasian school has some similarities with the post-Keynesian school.
As defined by Colander (1996), according to post-Walrasians, the knowledge of
individual decisions is not sufficient to predict macroeconomic outcomes;
rationality is bounded; multiple equilibria, non-linearities, and path-dependent
equilibria are the norm. Problems of coordination are the main issue. Various
institutions, money and credit relations being one of them, are required to
provide the coordination that market mechanisms cannot generate. By providing
conventions and common expectations, sometimes with the help of so-called
imperfections and rigidities, these institutions also impose some stability in
an economic system that would otherwise be subject to unstable fluctuations. How
a society decides to provide this coordination is crucial to the resulting
economic output.
For modeling purposes, the Post Walrasian approach means there is an extra
component of the production function that might be called a coordination
component. This extra component eliminates the one to one relationship between
inputs and outputs that exists with the standard production function. Given
different degrees of coordination, the same physical inputs can yield quite
different aggregate outputs. (Colander 1996: 9).
Although post-Walrasians and heterodox economists share some concerns, the focus
of the former is on the supply side, with little attention devoted to effective
demand. Within the context of employment theory, a simple representation of the
post-Walrasian view is offered in the top part of Figure 2.(1) The top part of
the graph represents standard production functions. The upper production
function, labelled qnc, which can stand as either neoclassical or
no-coordination, represents the output possibilities that would exist, in the
abstract, if coordination was absolutely costless. The lower production
function, labelled qflex, represents possible output levels in a pure market
system where all prices would be flexible, when taking into account the costs
imposed by coordination requirements. The middle production function, labelled
qcon, represents possible output levels when conventions and institutions are
put in place to fulfil the coordination requirements. It is assumed here that
these institutions are more efficient to resolve the coordination problems than
the market left on its own.
Figure 2
The consequences for employment can be visualized in the bottom part of Figure
2. To each different production function corresponds a demand for labour curve,
based on the standard marginal productivity of labour. In the case of the
economy with conventions and rigidities, it is assumed that there are no forces
leading to full employment. Employment is only Lcon, below the full employment
level Lfe, but still, because its production frontier stands beyond that of the
pure market economy, the output level (and output per capita) is higher in the
economy with conventions and rigidities than in the pure market economy with
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (4 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
flexible prices (q*con vs q*flex). In the economy with conventions and
imperfections, full employment could be achieved if real wages could fall from
(w/p)con to (w/p)fe, but this is deemed to be improbable due to the existing
rigidities brought about by the presence of the coordinating institutions. To
summarize, in the post-Walrasian model shown here, the flexibility of prices and
real wages is detrimental to the economy, but ultimately, unemployment is
associated with real wages which are too high.(2) In addition, demand-led
phenomena are mainly left out of the analysis.
1.3 The New Keynesian model with efficiency wages
While the above description of the post-Walrasian model may seem somewhat devoid
of content, the presentation of a well-known New Keynesian model of coordination
failure may help to illustrate what is at stake. In the 1980s, New Keynesians
such as Shapiro and Stiglitz (1984) have rediscovered what Marxist authors had
long underlined, the fact that labour is not a commodity, and that one must
distinguish between labour and labour power (Green 1988). New Keynesians have
thus introduced work effort functions which depend usually on the expected cost
of job loss. This cost itself, at the macroeconomic level, depends on the
differential between the real wage (when employed) and income when unemployed,
and on the probability of losing one's job without finding another similar
one.(3) This probability is a function of the rate of unemployment. Hence, for
given social benefits and a given supply schedule of labour, effort depends
positively on the real wage and is an inverse function of the aggregate level of
employment.
Two kinds of effort function have been suggested in the literature. One is a
discrete effort function, of a binary sort, where effort is maximum provided the
real wage is sufficiently high. At that real wage, there is no shirking;
otherwise, at lower real wages, effort is zero and the worker is shirking on the
job all the time. It has been shown by critics of the efficiency wage approach,
that only this binary discrete effort function necessarily yields the results
that are sought. Whenever the effort level is a continuous function of aggregate
employment, bizarre constellations may arise (van Ees and Garretsen 1996;
Nistico and D'Orlando 1998). In all cases, however, an upward-sloping curve, in
the real wage and employment space is derived. This new curve represents the
no-shirking constraint (NSC) in the discrete case. In the continuous case, it
must be reinterpreted as the cost-minimizing real wage, i.e., the real wage that
maximizes the effort per worker at each level of expected aggregate employment.
Whether there are continuous or discrete effort functions, we shall denote them
as NSC curves. Along these curves, higher levels of expected aggregate
employment require higher cost-minimizing real wages.
When the effort function is discrete as described above, or when the continuous
effort function yields a well-behaved constellation, the aggregate labour demand
curve LD that results from the profit-maximizing behaviour of firms, taken in
the aggregate, is downward sloping, as shown in Figure 3. For simplicity, the
supply curve of labour is assumed to be vertical. Those three curves constitute
the standard New Keynesian efficiency wage apparatus. At the real wage (w/p)ew,
expected aggregate employment and realized aggregate employment are consistent
with each other. Point E -- the point of intersection between the aggregate
demand curve for labour and the no-shirking constraint curve -- is thus the
equilibrium point, the point where profit maximizing (LD) and cost-minimizing
(NSC) conditions are mutually consistent. At the equilibrium real wage rate
(w/p)ew, however, only Lew workers are being hired. For full employment Lfe to
be achieved, the real wage would have to fall to (w/p)fe, but this would be
inconsistent with the no-shirking constraint. Therefore, in this New Keynesian
model, ultimately, there is unemployment because real wages are too high. Still,
unemployment is involuntary: workers do not refuse to work, nor do they refuse
to let real wages fall. The firms are the ones who do not let real wages fall,
because they believe that workers hired at any lower real wage will be shirking.
Figure 3
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (5 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
Despite the elegance of the model and its appealing realistic features to
explain involuntary unemployment, there are several important logical flaws in
its conception, which lead some critics to argue that the model ultimately fails
to explain involuntary unemployment. As Sawyer (1998: 129) notes, this New
Keynesian model, like the generic post-Walrasian model of Figure 2, takes no
consideration whatsoever of effective demand constraints. It is assumed that
whatever firms have decided to produce, at employment level Lew, will be sold.
Introducing effective demand considerations would lead to over-determination.
But this could be considered to be an external critique, and thus may be of
little concern to New Keynesian authors. There are however more fundamental
critiques, related to the internal consistency of the model.
Secondly, the determinants of effort are usually such that the NSC curve is
asymptotic to the labour supply curve, as it has been drawn in Figure 3. As
Shapiro and Stiglitz (1984: 438) put it, "no shirking is inconsistent with full
employment". But then this means that near full employment, effort tends towards
zero. This feature of the analysis is explicitly developed by Bowles and Boyer
(1990), who clearly show that profitability falls below zero before the economy
reaches full employment. The immediate cause of this result is that effort is
assumed to be a function of the cost of job loss, this cost falling to zero as
full employment is approached. The difference between the actual wage rate and
the wage rate (or income) which is expected when laid off becomes nil, since all
workers would expect to be able to find another job right away. If the NSC curve
were not to be asymptotic (or nearly asymptotic) to the labour supply curve,
full employment equilibria could not be ruled out: the labour demand curve and
the NSC curve could intersect each other at full employment or even beyond full
employment! This possibility, which efficiency wage theorists wish to exclude by
definition, thus explains the analysis and the shape of the standard NSC curve,
as illustrated in Figure 3.
Thirdly, a related critique focuses on the stability of the equilibrium. One may
question whether point E in Figure 3 is a stable equilibrium. It has recently
been shown by Nistico et al. (1998) that, if the equilibrium of a New Keynesian
wage efficiency model is stable, then the labor demand corresponding to a zero
real wage rate is necessarily below the full employment level. The dynamics of
the model are examined on the assumption that the realized aggregate labour
demand of the previous period constitutes the aggregate employment level which
is expected by both workers and firms in the next period, when these groups of
agents define their effort functions and their profit-maximizing behaviour. In
the case of Figure 3, which is the standard representation of the wage
efficiency equilibrium, the labour demand curve cuts the horizontal axis beyond
the full employment level, thus implying that the E equilibrium is unstable. If
the aggregate employment level Lew is not expected from the start, it will never
be reached, with ever larger oscillations of the employment level.
On the other hand, in the case of Figure 4, the efficiency wage equilibrium is
stable. Suppose agents expect the aggregate employment level Le; as a result,
reading off the NSC curve, a real wage (w/p)1 will be imposed; this in turn,
reading it off the LD curve, will induce a realized employment level of LD1.
This level will now become the newly expected level of aggregate employment,
and, setting the new real wage accordingly, the new level of realized employment
will be somewhere between Lew and Le. This will become in turn the newly
expected level of aggregate employment, and it can be seen intuitively that the
expected and the realized values of employment are gravitating towards their
equilibrium value, given by Lew, and hence that the equilibrium real wage rate
(w/p)ew will eventually be realized. It has been shown that for this
configuration to occur, the labour demand curve must cut the horizontal axis
before the full employment level, as it does in Figure 4. This means, however,
that in the stable case, the real wage can never be low enough to provide for
full employment. It follows that, with such a stable efficiency wage
equilibrium, "unemployment is to be explained by referring to causes different
from those highlighted by the efficiency wage theory" (Nistico et al. 1998:
147).
Figure 4
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (6 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
A fourth and final critique of the efficiency wage model is that, in the case of
the continuous effort function, the aggregate labour demand curve need not be
downward sloping. While the no-shirking constraint curve keeps its
upward-sloping shape whatever, the aggregate labour demand curve could be upward
sloping, or have a segment which is upward sloping. This can be explained
intuitively in the following way. Higher expected aggregate employment levels
induce higher real wages, but they may also end up inducing a higher level of
effort, and hence a higher marginal product per worker. For each individual
firm, higher real wages (due to the expected higher aggregate employment) may
thus be associated with higher labour employment. That case is illustrated, and
mathematically demonstrated, by Nistico and D'Orlando (1998). Hence, when
aggregating over all firms, the aggregate labour demand curve may be
upward-sloping for some values of real wages. It should be mentioned however
that, as long as effort is based on the cost of job loss, part of the labour
demand curve will still be downward sloping, since expectations of near
full-employment will induce ever rising real wages and a labour effort that
approaches zero.
Figure 5
Figure 5 offers an example where the labour demand curve has some upward sloping
segment. As in the previous figures, point E is the only point where the
expectations about aggregate employment, as given by the NSC curve, are
consistent with the aggregate employment decisions of the firms, taken on the
basis of profit maximization. This equilibrium point, in addition, is stable.
Many other configurations could be possible, with either a unique equilibrium or
multiple equilibria, the two curves cutting each other more than once.(4) In the
configuration chosen here, excessive real wages are not the problem. If real
wages could be higher, at level (w/p)h instead of (w/p)ew for instance,
unemployment rates could be reduced and the level of employment could rise from
Lew to Lh. It is clear that the determination of real wages cannot be left to
market forces, for it generates an equilibrium with high unemployment and low
standards of living. Downward flexibility of real wages will not alleviate the
unemployment situation. On the contrary, what is needed is some institution that
would enforce high real wages as a national norm: this would lead to a
substantial reduction in unemployment. Note finally that, in this variant of the
model, as in the previous variant, the absence of full employment must be
explained by some factor other than variable effort, for full employment cannot
be achieved, whatever the real wage!
1.4 The wage bargaining model
The wage bargaining model, in its graphical form, yields results which are very
similar to those described by the well-behaved wage efficiency model. The wage
bargaining model, which comes from applied work on the determinants of
unemployment and the NAIRU, can be associated in particular with the work of
Layard, Nickell and Jackman (1991). Their model, also referred to as the PS-WS
model, consists of two equations, one determining the real wage w/p and the
other the margin of profit p/w. To these equations is sometimes associated a
third one -- the factor price frontier, involving the real rate of interest, the
real wage rate, and productivity growth.
The price-setting function, illustrated in Figure 6 as the PS curve, basically
represents the profit-maximizing behaviour of entrepreneurs, when deciding on
the number of workers they want to hire, at various real wages (or profit
margins). It is, as before, the labour demand curve, where the marginal
productivity of labour, or the marginal revenue product in the case of imperfect
competition, is equated to its real wage. Layard and his colleagues from the
London School of Economics provide a substantial amount of empirical evidence
regarding this PS relation, which leads them to argue that the rising rates of
unemployment in Europe, and the persistence of their high levels, must be
attributed to real wages which are excessive compared to productivity.
Anyadyke-Danes and Godley (1989) have however managed to replicate the results
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (7 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
obtained by Layard and his colleagues despite assuming, by construction, that
real wages and employment are independent random variables, fluctuating about
trends. In addition, it can be shown that the equations of the PS curve that are
used in econometric work are no different from those that can be derived from
the national accounts (Lavoie 1998a). It would thus seem that the empirical
evidence supporting the construction of the PS curve, interpreted as a
relationship arising from the profit-maximizing behaviour of entrepreneurs, as
constrained by production functions with diminishing returns, is at best of
dubious validity.
Figure 6
Then there is a wage-setting function, here the WS curve, which represents the
bargaining position of workers. The lower the rate of unemployment, the higher
the wage rate, and hence the real wage rate, that workers will fetch for, as a
result of their stronger bargaining position. A substantial amount of empirical
evidence regarding the slope of this curve has been uncovered by Blanchflower
and Oswald (1994). They show, notably by using cross-section data, that American
states with low rates of unemployment turn out to be those where real wages are
relatively high. The WS curve is thus upward-sloping in the real wage/employment
plane, as was the no-shirking constraint of the efficiency wage model. The two
models, in their well-behaved version, are thus almost identical, both yielding
the result that there is unemployment because real wages are too high. In the
PS-WS model, the point of intersection of the two curves yields an equilibrium,
where the profit margin required by firms is compatible with the real wage
demands of the workers or their labour unions.(5) What we have here is a
macroeconomic application of the microeconomic labour market, where labour
unions constitute an unavoidable monopsonist.
Models of the PS-WS variety usually pertain to demonstrate that higher real wage
demands, and hence higher equilibrium rates of unemployment (NAIRUs), are
associated with factors such as higher tax rates (the tax wedge) and lower rates
of growth of productivity. Parameters such as the level of support offered by
social programs, may also be taken into account. In the face of the lowering
explanatory power of some of these variables, an additional variable has
recently been taken into account, the real interest rate. It is argued that in
the long run the economy cannot but be on its factor price frontier, and hence
that, taking capital into account, the real wage that is allowable from the
standpoint of the firm is lower whenever the real rate of interest is increased.
Hence, with higher real rates of interest, the PS curve would be shifting down
and both the equilibrium real wage and the equilibrium level of employment would
be lowered, as illustrated by the dashed PS line. The more recent PS-WS models
thus attribute most of the increase in "natural" rates of unemployment during
the 1980s and 1990s to the high real rates of interest of that period (Cotis et
al. 1998). There is some irony here: heterodox economists have long been arguing
that high real rates of interest are detrimental to economic activity and
employment. But whereas heterodox economists attribute this result to the
negative impact of high interest rates on aggregate demand, PS-WS economists
link this to the supply side.(6)
2. Post-Keynesian analyses with hybrid features
2.1 The distinction between notional and effective labour demand
Whatever their merits, neither the wage-bargaining approach nor the efficiency
wage approach take aggregate demand into account. It may be possible to
introduce aggregate demand in such models, but as Sawyer (1995) showed in the
case of the PS-WS model, such an addition would render these models
over-determinate. Either one of the two equations has to be dropped, or an
additional variable has to be introduced. In the subsequent presentations, one
of the equations will be dropped and replaced an effective demand condition.
The purpose of the following model is to explicitly introduce aggregate demand
considerations, while still retaining the standard production function of
neoclassical analysis, with diminishing returns. We shall consider two variants
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (8 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
of this model: a flexible price variant, which can be attributed to Keynes
himself and to some of his most faithful followers, such as Paul Davidson
(1998); and a fixed price variant, which can be associated with the work of the
so-called French disequilibrium school. In both models, we will have to
distinguish between the notional demand for labour, and the effective demand for
labour, along the lines first developed by Patinkin (1965, ch. 13) and then by
Barro and Grossman (1971). The graphical representation of these latter authors
is however unsatisfactory, and as a result we will rely on the formulation of
Schefold (1983) and Fujimoto and Leslie (1983).
The production side of the model assumes a standard neoclassical production
function, with a diminishing marginal product of labour and decreasing returns.
Aggregate supply, in nominal terms, is thus given by:
AS = pq = pf(L)
where p is the price level, q is real output, and L is labour. As usual, we can
claim that the first derivative of this production function is positive, f'(L) >
0 , whereas the second derivative is negative, f''(L) < 0 , which implies that
the marginal product of labour is decreasing with respect to labour. The demand
for labour at different real wage rates is given by this f' function, assuming
as usual that firms attempt to maximize profit. This however should only be
considered as the notional demand for labour, in the terminology of Barro and
Grossman, because this labour demand curve takes no account of effective demand.
For a given real wage, the chosen level of employment only maximizes potential
profits, i.e., the profits that would be realized if all of production were
being sold. There is no assurance however that all of the goods being produced
will get sold. This is akin to the Marxist problem of profit realization. As
Keynes argued, one has to go beyond Say's law.
Thus, one has to take into account the effective demand constraint, i.e., the
constraint that aggregate supply needs to equal aggregate demand. In most
neoclassical models, this constraint has no relationship with real wages because
it is assumed that lower wages will immediately and automatically be compensated
by higher profits. In these models, aggregate demand only depends on factors
such as the stock of money supply and government expenditures. Here by contrast,
the distribution between wages and profits has an impact on aggregate demand. To
make the model as simple as possible, it is assumed that aggregate demand is
made up of only two components: wages, which are entirely consumed (the
propensity to consume out of wages is unity), and some autonomous expenditures,
which cover both investment expenditures and consumption on profits.(7) This
implies that consumption out of profits is assumed to depend on profits
collected in a previous period, as was often assumed by Kalecki (1971). We are
thus left with:
AD = wL + ap
where w is the average nominal wage rate and a is real autonomous expenditures
(and hence ap is nominal autonomous expenditures). It is assumed that investment
expenditures do not depend on real wages, a hypothesis which is not quite
coherent within a neoclassical framework, as demonstrated by Bhaduri (1983), but
we will leave it at that, on the supposition that in a post-Keynesian world
dominated by uncertainty, animal spirits are the principal determinant of
investment expenditures. The effective demand constraint is thus given by
equating aggregate supply with aggregate demand:
wL + ap = pf(L)
Solving for the real wage, we obtain the effective demand constraint, or the
effective labour demand curve:
w/p = [f(L) a]/L
Now it turns out that this function reaches its maximum when it equates the
notional demand for labour, i.e., the effective labour demand curve reaches its
highest point when it is intersected by the notional labour demand curve. This
can be seen by taking the first derivative of the effective demand for labour:
d(w/p)/dL = [f'(L).L (f(L) a)]/L2
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (9 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
and setting it to zero, which happens when:
f' (L) = [f(L) a]/L
The term on the left is the notional labour demand; the term on the right is the
effective labour demand. The second derivative of the effective labour demand
equation is negative when evaluated at its extremum, thus indicating that the
extremum is a maximum. We are then able to draw the two labour demand curves, as
shown in Figure 7. The plane can further be divided into three areas. Along the
LDeffective curve, AD = AS; above the curve, there is excess demand on the goods
market, AD > AS; and below the curve, there is excess supply, AD < AS. The
model, as constructed here, with its two distinct labour demand curves, is
common to both Keynes and the French disequilibrium school. What distinguishes
the two approaches are the hypotheses about market behaviour.
Figure 7
2.1 Keynes's flexible price model
Keynes's flexible price model in a competitive goods market has been put forth
by post-Keynesian authors such as Paul Davidson (1998) and Amitava Dutt (1987).
In Dutt's model of Keynes's effective demand model of the General Theory, firms
are assumed to be atomistic, believing they can sell any level of output at the
market price. These market prices, however, are not known until the end of the
market period, in contrast to what is assumed in the new classical model
presented above. To make their employment decisions, firms must have
expectations about the price level. It is assumed that firms set nominal wages
at the beginning of the market period, before they know the realized price
level. Thus, on the basis of the expected real wage, firms make their employment
decisions, in accordance with the notional demand curve. In Figure 7, the
expected real wage is (w/p)e, and hence aggregate employment is L1. With this
combination, however, there will be an excess demand for goods. Prices will
rise, until aggregate demand and supply are equated, i.e., until the economy is
back on the effective demand curve for labour. Market prices will thus be such
that the realized real wage rate has fallen to (w/p)1 in Figure 7.
Although the goods market clears, the situation as illustrated is inconsistent
with short-run equilibrium, since price expectations are not fulfilled. Firms
will thus revise their expectations. If these expectations are adaptative, the
expected price would be revised upwards. In the simplest case where the expected
price is the price realized in the previous period ("it is sensible for
producers to base their expectations on the assumption that the most recently
realised results will continue" (Keynes 1936: 51)), the newly expected real wage
would be given by (w/p)1, and hence, reading off the notional labour demand
curve, the level of employment by firms would be L2. At that combination, there
would be an excess supply for goods, and prices would fall somewhat, with the
realized real wage rate standing somewhere between (w/p)1 and (w/p)K. There will
thus be a succession of oscillations in employment. Intuitively, we can see that
point K will eventually be reached, where price expectations would be achieved.
The equilibrium is thus given by point K, where the notional and the effective
labour demand curves intersect. Point K is thus what Davidson (1998: 825) calls
the point of effective demand. Point K corresponds to rational expectations on
the part of the firms. If firms know how a demand-constrained economy works out,
and if they have perfect knowledge of both the notional and effective labour
demand curves, anticipations based on rational expectations should lead them
right through point K.
It also appears that Keynes and Davidson are right when they claim that the real
wage does not determine the level of employment, but rather that the level of
effective demand determines the real wage. The expected price level, which is a
function of the level of the expected aggregate demand, determines the level of
employment, read off the notional demand curve for labour, and this chosen level
of employment determines the realized real wage, given the realized aggregate
demand. The model is also consistent with Keynes's claim (1936, ch. 2), that
while firms are on their labour demand curve (the notional demand curve), thus
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (10 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
fulfilling what he calls the first postulate, workers are not on their labour
supply curve, which allows Keynes to reject the so-called second postulate.
In this model, however, the often made claim that there is unemployment because
real wages are too high is still partially valid. It is true that entrepreneurs
have no incentives and no means to lower real wages, given the existing
effective demand conditions, since real wages are determined by the point of
effective demand. As Davidson (1998: 822) claims, the point of effective demand,
as given by point K, "represents equilibrium in the goods market where the
expectations of sales by profit maximizing entrepreneurs are just being met by
the spending decisions of the buyers. At the point of effective demand there is
no reason (endogenous force) that causes entrepreneurs to alter their
production, pricing and hiring decisions as long as the determinants of
[aggregate demand and aggregate supply] remain unchanged".
Figure 8
It follows that real wages can fall, and full employment is restored, only if
effective demand increases, i.e., if the effective labour demand curve shifts
down, intersecting the notional demand curve at point W, as shown in Figure 8.
Point W is the Walrasian equilibrium, with both the goods and the labour markets
clearing. In the present simplified model, the downward shift of the effective
demand curve can only occur if real autonomous expenditures a increase. Two
mechanisms have been contemplated in the literature, both related to falling
nominal wages, accompanied by falling prices. The first mechanism is Keynes's
effect, whereby the fall in prices would diminish the demand for money needed
for transaction purposes, leading to a fall in interest rates and hence an
increase in the investment part of our parameter a. The second mechanism is
Pigou's effect, whereby the fall in prices would lead to an increase in real
money balances, given that the stock of money is exogenous, and hence to an
increase in the autonomous consumption part of parameter a. If neither automatic
mechanisms are credible, in particular when falling prices generate debt
deflation, bankruptcies, and adverse expectations, then we are back to Keynes's
position: the a parameter can be increased, and full employment can be restored,
only by government taking the discretionary decision to increase its autonomous
expenditures and fiscal deficit (Keynes 1936, ch. 19). Still, due to the
standard assumptions made about the production function, full employment must be
accompanied by a lower real wage -- here in Figure 8, at the level (w/p)fe. In
this model, there is a correlation between real wages and the level of
employment, but the causality runs from employment to real wages.
2.3 The neo-Keynesian model with fixed prices
In our description of Keynes's model, it was assumed that any discrepancy
between aggregate demand and aggregate supply would be quickly made good by a
variation in prices. In a world of imperfect competition, and also in most
markets of today, prices are set as goods are supplied, and hence they do not
adjust immediately to possible discrepancies between supply and demand. Most of
the adjustment is then carried by changes in inventories and in flow production.
This is precisely what is being assumed in the next model, where it is supposed
that wages and prices are fixed, at least in the short period. The model is
based on the equations presented above, and on the arguments offered by the
economists of the so-called French disequilibrium approach, or the neo-Keynesian
school, as can be found in the writings of Bénassy (1975) and Malinvaud (1977).
In their models, adjustments are made through quantities, rather than prices.
Neo-Keynesians consider that there are two kinds of unemployment: classical
unemployment and Keynesian unemployment. Classical unemployment occurs when real
wages are too high. This would be the case in Figure 8, if the real wage set by
the decisions of firms was anywhere above (w/p)K. Being above the effective
demand constraint, there is excess demand on the goods market, and since prices
are no more flexible, inventories are being depleted: this is the case of
so-called repressed inflation. There is classical unemployment because a fall in
real wages would allow employment to increase from, say L1 to LK. Note that, as
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (11 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
Kahn (1977) was quick to point out, the relevance of this sort of classical
unemployment appears unlikely: if there is excess demand on the goods market,
and with the excess supply in the labour market, one would expect that
eventually, prices would rise while nominal wages would drop, thus gradually
leading the economy to point K. In other words, classical unemployment here is
an unstable situation.
Increases in real wages may also allow for increases in employment, however.
Suppose that real wages are set at level (w/p)fe in Figure 8. Following a
profit-maximizing behaviour, firms would initially choose the level of
employment Lfe. At this combination of real wages and employment, however, the
goods market is not in equilibrium, since the economy is not on the LDeffective
curve. Two different levels of employment, at the real wage (w/p)fe, could clear
the goods markets: these are L1 and L2. But below the effective labour demand
curve, goods are in excess supply. Oligopolistic firms in a fixed-price world
will thus reduce production, at a constant real wage, until aggregate demand and
supply are equated, i.e., until the economy moves back horizontally to the
effective labour demand curve, and hence until employment is down to L1. In this
case, firms cannot be on their profit-maximizing notional labour demand curve.
For any real wage below (w/p)K, the effective labour demand curve is the
relevant constraint, and any increase in real wages will generate an increase in
employment, as firms respond to the new aggregate demand conditions by moving up
their effective labour demand curve.
Still, ultimately, to wipe out all of the unemployment, real wages ultimately
have to decrease. As in the previous model, an increase in autonomous demand a,
such as an increase in government expenditures, will be required, and this will
have to be accompanied by a fall in real wages, from (w/p)K to (w/p)fe. Thus, in
this model, as in Keynes's model, getting rid of all of unemployment requires
lower wages, unless the labour supply curve happens to pass through point K.
3. The post-Keynesian model based on Kaleckian foundations
3.1 The basic Kaleckian model of employment
A major source of controversy among post-Keynesians pertains to the appropriate
microeconomic foundations of the post-Keynesian political economy,
notwithstanding the fact that many post-Keynesians would consider macroeconomics
to constitute the foundations of microeconomics. Two views seem to emerge. Some
authors, mostly associated with Paul Davidson and Sidney Weintraub, endorse
Marshallian foundations, with diminishing returns and rising cost curves, as
well as conditions generally associated with pure competition. This group takes
the view that post-Keynesian economics is more general than neoclassical
economics, and hence that it is helpful to maintain these standard assumptions
which, as was just argued, are to be found in Keynes.(8) Other authors, closer
to Joan Robinson (1977), Nicholas Kaldor (1985) and Alfred Eichner (1991),
favour Kaleckian foundations, with administered cost-plus prices (simple
mark-up, normal-cost or target-return pricing) and constant or near-constant
unit direct costs, under conditions of imperfect competition. This latter group
sees post-Keynesian economics essentially as a more realist alternative to
neoclassical economics rather than an attempt at encompassing mainstream theory.
Thus the model of employment next to be presented, based on heterodox
microeconomic foundations, will be called the Kaleckian model.(9)
The aggregate demand part of the model is identical to what was presented in the
previous section. The new part lies with the supply side. Instead of the
standard neoclassical production function, Kaleckians assume that marginal costs
(and variable costs) are constant up to full capacity. When taking fixed costs
into account, such as capital depreciation allowances and fixed labour costs
associated with managerial and supervising duties, unit costs are decreasing,
also up to full capacity. It is also assumed by Kaleckians that firms generally
operate below full capacity, on the non-increasing portions of the marginal cost
and unit cost curves. Various reasons are invoked, such as the need to dispose
of capacity reserves that allow firms to preserve their share of the market by
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (12 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
being always able to respond to any sudden increase in demand. Such a model,
with overhead labour costs, has been initially proposed by Michal Kalecki (1971
p.44) and Joseph Steindl (1952), and it has been popularized by various
Kaleckian or Marxist authors (Harris 1974, Asimakopulos 1975, Rowthorn 1981,
Lavoie 1992). The firm thus hires two kinds of labour (L = Lv + Lf). Blue-collar
workers, denoted by Lv, are essentially the variable factor of production. They
vary directly with the level of output, so that we can write Lv = q/yv , where
yv is the constant productivity of variable labour. White-collar workers, are
essentially part of the overhead costs, since their numbers depend on the level
of full capacity output or on the number of existing plants. In the short run,
the number of plants is fixed, and hence so is overhead labour, denoted by Lf.
The utilization function, which is the post-Keynesian equivalent to the
neoclassical production function, and which relates total employment to
production when the level of capacity is given, may thus be written as
q = [L Lf)]yv
The aggregate supply function may now be rewritten as:
AS = pq = p[L Lf)]yv
In this model, there is no notional demand curve for labour, where the demand
for labour would depend on some profit-maximizing behaviour. Mark-up pricing
over unit variable costs is sometimes reinterpreted, both by New Keynesians and
some post-Keynesians, as trial and error profit maximization. There is evidence,
however, that normal cost pricing, where the price depends as well on some
normalized measure of unit fixed costs, is the most prevalent of the cost-plus
pricing procedures. This is clearly incompatible with marginal cost pricing and
profit maximization. In addition, empirical estimates of price elasticities in
oligopolistic industries yield below-unitary elasticities, which also implies
that profit maximization is not the pricing rule being followed (Koutsoyiannis
1984; Lavoie 1992: 137). As a result, we shall assume that the only constraint
on the demand for labour arises from the effective demand constraint, resulting
from the interaction between aggregate supply and aggregate demand. This
aggregate demand now becomes:
AD = wvLv + wfLf + ap
where wv and wf are the nominal wages of variable and overhead labour
respectively. In the simplest case, the wage rate of both kinds of labour would
be identical. We may also suppose that the wage rate of overhead labour is some
multiple of the wage rate of variable labour, and hence that wf = wv , with > 1.
Making this assumption, and combining the aggregate demand and supply functions,
one obtains a relationship between the basic real wage rate (wv/p) and the
overall level of employment. This relationship is the effective labour demand.
wv/p =[(L Lf)yv a]/[L + Lf( 1)]
It can be shown that the first derivative with respect to L of this equation is
positive. The effective labour demand curve is the upward-sloping curve shown in
Figure 9. The curve is asymptotic to the index of variable productivity yv. As
in the model of Figure 8, there is an excess demand of goods above the curve,
and there is an excess supply below the curve. In this model, as in the previous
neo-Keynesian model, firms are assumed to react to any disequilibrium in the
goods market by changing output levels. In short-run equilibrium, firms are thus
on the LDeffective curve. In this model, it is clear that an increase in the
basic real wage of variable workers, and hence in the real wage rate of overhead
labour, leads to higher employment. Full employment can always be achieved by
raising real wages (the only limit is that real wages should not exceed the
productivity of variable labour, otherwise firms are making losses). Looking at
things the other way, we can conclude from the above analysis that if real wages
were flexible downwards, this flexibility would be counter-productive. With
unemployment in the labour market, and a goods market in equilibrium, real wage
flexibility would bring about a fall in labour employment, thus worsening the
problem of unemployment.
Figure 9
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (13 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
3.2 Some extensions of the Kaleckian model
Now, a number of additional features could be added to this simple Kaleckian
model of employment. Before we do so, it can be noted from the above equation
that an increase in productivity, as measured by yv, leads to an upward shift of
the effective demand curve. This implies that, for given real wages, increased
productivity induces lower employment, a result which certainly makes sense in
the present context. Increased productivity must thus be accompanied by
increases in real wages, otherwise effective demand will fall off. On the other
hand, an increase in real autonomous demand a leads to a downward shift of the
effective labour demand curve, which implies, as Keynesians would guess, that an
increase in autonomous demand leads to an increase in employment, if real wages
remain constant.(10) Here falling real wages need not accompany increased
employment. However, in this case, although the real wages of each employed
worker remains the same, the average real wage decreases with the increased
level of employment, because the higher level of production (in the short
period) implies that a higher proportion of the labour force is made up of
variable workers, whose wage rate is lower than that of their supervisors
(Lavoie 1996-97). Here the negative relationship between average real wages and
aggregate employment is an artefact of aggregation.
In the Kaleckian model as presented, the real wage rate is an indeterminate
variable. One may wish to close the model by introducing either a wage
bargaining curve or a no-shirking constraint curve. This is shown in Figure 10,
where a curve labelled WS-NSC-LS has been added to the effective labour demand
curve. The wage bargaining curve WS, here interpreted as the idea that workers
manage to extort higher real wages from their employers when the level of
employment rises relative to full employment, is quite consistent with the
demand-led Kaleckian model. With the usual assumed shape of the wage bargaining
curve WS, a situation with multiple equilibria may easily arise. Suppose, as is
done in Figure 10, that workers and firms expect the level L0 of employment. At
this level, the WS curve is telling us that the real wage (w/p)0 would come out
of the bargaining process. But at that real wage rate, only LD0 workers would
end up being employed, on the basis of the effective demand constraint. Now
realizing that employment is much lower than previously expected, wage
bargaining would be revised, and a lower real wage would arise. This would lead
to still lower employment. In other words, the high employment and high real
wage equilibrium, given by Lh and (w/p)h in Figure 10, is not stable, whereas
the low employment and low real wage equilibrium, given by Ll and (w/p)l, is
stable. Again, this result just reasserts the need for government intervention,
to make sure that social conventions are such that the high real wage and low
unemployment equilibrium is enforced. Also, institutions should be set up to
ensure that the wage bargaining curve WS shifts out to the right, such that the
high employment equilibrium turns out to be consistent with full employment. As
can be imagined from the inspection of Figure 10, this would require an even
higher real wage.(11)
Figure 10
We could also suppose that labour supply, instead of being represented by a
vertical line, is given by the WS-NSC-LS curve. We would then be left with a
standard supply curve of labour, positively linked to the real wage, possibly as
a result of the standard neoclassical substitution effect (for that matter, the
curve could as well be backward-bending). The level of employments given by Ll
and Lh would this time represent two possible full-employment equilibria
(Seccareccia 1991b). In industrializing countries, the full-employment
equilibrium with high real wages could be interpreted as an equilibrium where
most workers would be employed in the formal sector, whereas in the
full-employment equilibrium with low real wages, most workers would be more or
less self-employed in the informal sector. Again, it is clear that the dynamics
of market forces are such that the low wage full-employment equilibrium is the
stable equilibrium. If market forces are left to themselves, low standards of
living will constitute the attractor of such an economy. Again, government
intervention and appropriate institutions are required to keep the economy
around the high wage full-employment equilibrium, or eventually to get it up
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (14 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
there. Market forces are unable to achieve this result.
Finally, there is the possibility of incorporating the efficiency wage framework
to the above Kaleckian model.(12) One may wonder how efficiency wage
considerations can be entered without taking profit maximization into account.
The reason is that, as long as we assume that "the level of the firm's output
does not alter the labour-extracting process, the firm may proceed sequentially:
first determine the cost-minimizing [real wage], and then determine the level of
output" (Bowles and Boyer 1990: 195). The determination of real wages and labour
employment is thus decomposable. First, firms and workers have some expectations
about aggregate employment, which may be determined by the realized level of the
previous period. This determines the cost-minimizing real wage and the level of
effort corresponding to it. This real wage, combined with the productivity level
associated with the chosen level of effort, then helps to determine the
effective demand conditions and hence the effective demand for labour. The model
is thus recursive, because it is assumed that the actual level of employment
does not have an impact on the current level of effort; it only has an impact on
effort and productivity of the subsequent period. In addition, it is assumed
that while prices, with respect to wage costs, are fixed according to
cost-minimization principles, on the basis of the no-shirking constraint,
Kaleckian firms set the level of employment in response to aggregate demand (in
line with the effective demand for labour). Kaleckian firms do not set the level
of employment on the basis of profit maximization: they do not make use of the
so-called notional demand for labour.
When taking effort into consideration, we are faced with some of the same
difficulties that marred the New Keynesian model of efficiency wages. Again,
these difficulties are alleviated if effort is considered as a dichotomic all or
nothing choice. Then the realized level of effort does not depend on the
aggregate level of employment; higher levels of aggregate employment will only
increase the real wage that is required to minimize unit labour costs. In this
case, the no-shirking constraint curve will be given by the WS-NSC-LS curve of
Figure 10, and the effective labour demand curve will remain unaffected by the
addition of effort or shirking considerations. Results similar to those of wage
bargaining will thus be attained, with again two possible equilibria, the one
with low employment and low real wages being once more the only stable
equilibrium.
3.3 Variable labour effort and the Kaleckian model
Things however could be different with a continuous variable effort function.
Because changes in effort will have an impact on productivity, as measured by yv
, they will have an effect on aggregate demand and the effective demand for
labour. This can best be seen by rewriting the effective labour demand equation
with L as the dependent variable. Let us do so by taking the simplest possible
Kaleckian model, with no overhead labour costs (Lf = 0). We then obtain that:
L = a/[yv (w/p)]
Because of the effort function, changes in the cost-minimizing real wages
(w/p)*, themselves induced by changes in expected aggregate employment, cannot
be made while assuming constant labour productivity yv. Changes in the real
wages set by firms are accompanied by changes in effort and in labour
productivity. This can be rewritten as:
yv = yv0 + (w/p)*
and therefore the equation of the effective labour demand curve can be rewritten
as:
L = a/[yv0 + (w/p)* (w/p)*] = a/[yv0 + ( 1)(w/p)*]
It follows that:
dL/d(w/p)* = (1 )/[yv0 + ( 1)(w/p)*]2
As we have seen, neoclassical specialists of the efficiency wage literature
usually assume that the induced higher real wages are accompanied by constant or
falling effort levels. In the present model, this is equivalent to claiming that
induced higher real wages are linked to lower levels of yv. With our notations,
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (15 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
this means that < 0. Checking on the above equation, this would imply that the
effects of higher real wages on aggregate employment would simply be reinforced.
This means that when the effort function is such that labour demand curves in
the neoclassical model are "well-behaved" and hence retain their negative slope,
the effective labour demand curve in the Kaleckian model retains its own
properties. Therefore, a "well-behaved" effort function will not modify the
upward-sloping shape of the effective labour demand curve in the Kaleckian
model. Figure 10 would still be a proper depiction of the economy with a
"well-behaved" effort function.
Indeed, such a depiction is proposed by Bowles and Boyer (1990: 205), in a model
which essentially incorporates a variable effort function with a Kaleckian model
of effective demand. In fact, with the standard New Keynesian and Marxist
assumption that cost-minimizing real wages need to rise abruptly near
full-employment, because otherwise effort would be falling sharply, the
effective labour demand curve would have to be upward-sloping at high real
wages. Under those conditions, as pointed out by Mason (1993), employment could
be restrained because of negative profitability -- a supply-side effect rather
than a demand-side one. In Figure 10, if by any chance employment happened to be
around Lh, the cost-minimizing real wage rate could exceed labour productivity:
negative profitability would arise, and this would induce firms to cut
production and employment.(13)
On the other hand, the upward-sloping effective labour demand curve of the
Kaleckian model could have a downward-sloping segment when increases in real
wages are accompanied by higher labour productivity. In the above equation, it
is clear that this awkward case, from the Kaleckian point of view, would occur
whenever > 1. The positive impact of higher real wages on effective demand would
then be counterbalanced by the negative impact of higher labour productivity on
the amount of labour required to produce the demanded output. If the elasticity
of labour productivity with respect to real wages is assumed to be constant,
such a result could only occur when this elasticity is higher than unity. In
such a case, the parameter could be smaller than one at low real wages, and it
could grow to be larger than one at higher real wages. Hence, with high real
wages, induced by high expectations regarding employment, higher real wages
could be accompanied by lower actual employment levels, as a result of the
growing productivity. The effective labour demand curve would have a segment
that is downward-sloping. Although some empirical results do not seem to support
this possibility (Seccareccia 1991b), it may be worthwhile to examine it
graphically.
The aggregate effective labour demand curve could then look like the one
depicted in Figure 11, exhibiting again the possibility of multiple equilibria.
Here, while keeping a continuous effort function, we have discarded the notion
that the expectation of a full employment economy would induce workers to offer
a zero level of effort. As a result, the NSC curve can cut across the labour
supply curve. Some cost-minimizing real wage is consistent with full employment.
This view is more in line with the approach taken by Akerlof (1982), where high
real wages act more as a carrot than as a stick. Workers work harder because
they are better paid and more satisfied about their job. Incidently, it can be
noted that the shape of the effective labour demand curve in Figure 11 is
somewhat similar to the shape of the aggregate demand constraint of Sawyer
(1995), and it is identical to that of the employment curve arising from a
Sraffian profit-maximizing choice of technique, when capital reversing is not
excluded by hypothesis (Garegnani 1990: 40).
Figure 11
In the graph, as shown, there would be three possible equilibria, where the
requirements of cost-minimization would be compatible with the effective demand
constraint. Point E1, with low employment and low real wages, would be stable.
The intermediate case, point E2 would be unstable. As to point E3, the behaviour
of the economy around that point would depend on the exact relative slopes of
the NSC and LDeffective curves. In any event, with such an effective labour
demand curve, full employment may be (temporarily) achieved at two different
real wages, as shown by the points FE1 and FE2 in Figure 11. If overall
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (16 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
autonomous expenditures were lower than shown on the graph, there could be no
such full employment real wage.
An interesting feature of this mixed model, which combines a variable effort
function to a Kaleckian model of employment, with an effective demand constraint
and decreasing unit costs, is that neoclassical results may reappear. If the
economy stands at point E3, it could be argued that there is unemployment
because real wages are too high, since lower real wages could bring the economy
to the full employment point FE2. Of course, such a statement sidesteps how,
given the behaviour of labour and firms, the economy could achieve and remain at
full employment. There is some irony, however. We have seen that the
introduction of the efficiency wage framework in the neoclassical model could
transform downward-sloping aggregate labour demand curves into upward-sloping
ones. On the other hand, the introduction of the efficiency wage framework in
the Kaleckian model can transform upward-sloping effective labour demand curves
into downward-sloping ones (Lavoie 1992: 253).
Concluding observations
In the simplest, as in the more sophisticated, neoclassical models of
employment, unemployment is essentially caused by excessive real wages. In most
of these neoclassical models, the constraint of effective demand is ignored, and
those, like the French disequilibrium school, who have taken effective demand
into account, have been pushed aside, away from the mainstream research fads. In
addition, when, in their eagerness to introduce some realism in their models,
either to respond to external critique or to squash new classical opponents, New
Keynesian and post-Walrasian economists try to explain involuntary mass
unemployment, they still retain the variants where full employment could be
achieved only if real wages turned out to be lower.
By contrast, post-Keynesian authors have emphasized the importance of effective
demand constraints. The models based on Kaleckian foundations essentially show
that high real wages are not an obstacle to full employment, and that quite the
contrary, high real wages may help to generate an economy with both high
standards of living and a small proportion of the labour force in the informal
sector. Essentially, what all the post-Walrasian and post-Keynesian models show
is that multiple equilibria should be considered as the rule, and single
equilibria as the exception. In addition, the better equilibria cannot generally
be achieved by laissez-faire and free market forces. They must be achieved with
the help of government intervention and national institutions. Designing the
proper institutions may be the real challenge, but clearly some obvious policy
lessons can be drawn, such as the potential positive effects of high minimum
wage legislation and the existence of a strong social net that would force
employers to offer decent wages, and that would sustain autonomous consumption
expenditures. Other lessons can be drawn, such as the fact that a reduction of
the work week, at a given weekly or monthly (real) wage, would have a favourable
impact on effective demand, since it would push up real wages.
Many aspects in the economics of labour have been left out, such as the notion
of segmented labour markets and the dual labour market hypothesis. This omission
is a reflection of my attempt to present various models within a single
framework. Segmented labour markets are without a doubt a crucial feature of all
economies, be they industrialized or semi-industrialized, and they are
considered as such by post-Keynesians, although most actual contributions to
this field have been made by authors working within other heterodox schools
(Seccareccia 1991, King 1999). In addition, some aspects of the relationship
between real wages and labour productivity have not been assessed, even within a
closed economy. In particular, little has been said beyond the short run. As
already mentioned, there is a large body of empirical evidence relating high
real wages to high employment (Blanchflower and Oswald 1994). Of course, this
evidence can be interpreted in two causal ways: (i) high real wages generate
high growth and high employment, as in the Kaleckian model based on effective
demand; or (ii) high growth and high employment induce higher real wages, either
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (17 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
because of a more powerful working class, or because of productivity effects
(Okun's law or more likely Verdoorn's law). In addition, both causal influences
may be true. There may also be wage efficiency effects based on good morale,
where high real wages induce workers to be more productive or to be more
cooperative when technical innovations are introduced. There is also the
possibility of X-efficiency effects, where the higher real wages force
businesses to reorganize more efficiently (Altman 1998). The high real wages may
also bankrupt the less efficient firms, yielding a larger share of the market to
the more efficient ones (the Webb effect). There is certainly an enormous amount
of room for research on this area, although post-Keynesians have certainly
contributed to this sort of literature, in particular with respect to Verdoorn's
law (McCombie and Thirlwall 1994).
As recalled in the introduction, while most heterodox economists believe that
high real wages and low savings rates are conducive to high income and high
employment in the short run, things are not so clear when the long run is taken
into consideration. I have not even tried to tackle the issue of the link
between real wages and the investment function, although I have elsewhere
(Lavoie 1995), as a response to authors such as Riach (1975), Taylor (1991), and
Bhaduri and Marglin (1990), who all argue that higher real wages may indeed
induce lower effective demand, on the account that high real wages diminish the
perceived normal rate of profit, thus inducing a decrease in investment that
would overwhelm the increase in consumption due to the higher wages. These
results are now mainly defended on account of an open economy and its balance of
payments constraint. But the open-economy theory of employment should be the
topic of another article! In the meantime, considering that models of effective
demand apply to the world economy taken as a whole, it can be concluded that low
wage policies, if pursued by most countries, can only lead to the impoverishment
of all.
Table A.1
Summary of the main features of the employment models
ModelKind of equilibriumConstraints involved
New classical modelReal wages are too high. Voluntary unemployment,
because the price level is overestimated.Labour supply curve, and
profit-maximizing downward-sloping labour demand curve, based on realized
prices.
Post-Walrasian modelInvoluntary unemployment, with multiple equilibria.
Real wages are too high.Coordination is the third factor in an otherwise
neoclassical production function.
New Keynesian model, with efficiency wagesInvoluntary unemployment.
Real wages are too high because of shirking and asymmetric
information.Cost-minimizing constraint with all or nothing labour effort,
and profit-maximizing downward-sloping notional labour demand curve.
Efficiency wages, revisited by criticsEquilibrium is stable only when
full-employment is impossible, even without shirking. Possibility of
multiple equilibria. Cost-minimizing constraint with variable labour
effort. The profit-maximizing notional labour demand curve may have an
upward-sloping segment.
PS-WS London School of Economics bargaining modelReal wages are too high,
as a result of labour union bargaining and monopoly power of
firms.Wage-setting curve of workers, and profit-maximizing
downward-sloping notional labour demand curve.
Neo-Keynesian model (French disequilibrium school)Real wages may be too
low or too high. There is a lack of effective demand. Prices are fixed,
and nominal wages are too rigid in the long run.Bell-shaped effective
labour demand curve, and profit-maximizing downward-sloping notional
labour demand curve.
Post-Keynesian model, à la KeynesInvoluntary unemployment.
There is a lack of effective demand. Higher employment causes lower real
wages. Prices are flexible.Bell-shaped effective labour demand curve, and
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (18 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
profit-maximizing downward-sloping notional labour demand curve.
Post-Keynesian model, à la KaleckiInvoluntary unemployment. Real wages are
too low. There is a lack of effective demand.Upward-sloping effective
labour demand curve, and exogenous real wage. Marginal costs are constant.
Kaleckian model, with efficiency wagesPossibility of multiple equilibria,
the low wage and low employment equilibrium being stable.Cost-minimizing
constraint, and effective labour demand curve, segments of which may be
downward-sloping with a variable labour effort function.
Neo-Marxist model, with efficiency wagesInvoluntary unemployment. Because
of shirking, real wages become too high near full employment. Due to
negative profitability, firms restrain production.Cost-minimizing
constraint, profitability constraint, and upward-sloping effective labour
demand curve.
References
Akerlof, G.A. (1982). "Labor contracts as partial gift exchange", Quarterly
Journal of Economics, 97 November, 543-569.
Altman, M. (1998). "A high-wage path to economic growth and development",
Challenge, 41 January/February, 91-104.
Anyadike-Danes, M. and Godley, W. (1989), "Real wages and employment: A
sceptical view of some recent econometric work", Manchester School, 57 June,
172-187.
Appelbaum, E. (1979). "The labor market", in A.S. Eichner (ed.), A Guide to
Post-Keynesian Economics, M.E. Sharpe, White Plains.
Asimakopulos, A. (1975). "A Kaleckian theory of income distribution", Canadian
Journal of Economics, 8 August, 313-333.
Barro, R.J. and Grossman, H.I. (1971). "A general disequilibrium model of income
and employment", American Economic Review, 61 , 82-93.
Bénassy, J.P. (1975). "Neo-Keynesian disequilibrium theory in a monetary
economy", Review of Economic Studies, 42, 503-523.
Bhaduri, A. (1983). "Multimarket classification of unemployment: a sceptical
note", Cambridge Journal of Economics, 7 September, 235-241.
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (19 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
Bhaduri, A. and Marglin, S.A. (1990). "Unemployment and the real wage: The
economic basis for contesting political ideologies", Cambridge Journal of
Economics, 14 December, 375-393.
Blanchflower, D.G. and Oswald, A.J. (1994). The Wage Curve, MIT Press,
Cambridge.
Bowles, S. and Boyer, R. (1990). "A wage-led employment regime: income
distribution, labour discipline, and aggregate demand in welfare capitalism", in
S.A. Marglin and J.B. Schor (eds), The Golden Age of Capitalism: Reinterpreting
the Postwar Experience, Clarendon Press, Oxford.
Carter, T.J. (1998). "Policy in a two-sector efficiency wage model: substituting
good jobs for bad ones", Journal of Post Keynesian Economics, 20 Spring,
445-462.
Colander, D. (ed.) (1996). Beyond Microfoundations: Post Walrasian
Macroeconomics, Cambridge University Press, Cambridge.
Cotis, J.P., Méary, R. and Sobczak, N. (1998). "Le chômage d'équilibre en
france", Revue économique, 49 May, 921-935.
Currie, M. and Steedman, I. (1993). "Taking effort seriously", Metroeconomica,
44 June, 134-145.
Davidson, P. (1998). "Post Keynesian employment analysis and the macroeconomics
of OECD employment", Economic Journal, 108 May, 817-831.
Duménil, G. and Lévy, D. (forthcoming). "Being Keynesian in the short term and
Classical in the long term: The traverse to Classical long-term equilibrium",
Manchester School of Economic and Social Studies.
Dutt, A.K. (1987). "Keynes with a perfectly competitive goods market",
Australian Economic Papers, December, 275-293.
Ees, H. van and Garretsen, H. (1996). "Endogenizing the natural rate of
unemployment: Phelps's structural slumps and the Post Walrasian framework", in
D. Colander (ed.), Beyond Microfoundations: Post Walrasian Macroeconomics,
Cambridge University Press, Cambridge.
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (20 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
Eichner, A.S. (1991). The Macrodynamics of Advanced Market Economies, M.E.
Sharpe, Armonk.
Fazzari, S.M., Ferri, P. and Greenberg, E. (1998). "Aggregate demand and firm
behavior: a new perspective on Keynesian microfoundations", Journal of Post
Keynesian Economics, 20 Summer, 527-559.
Friedman, M. (1968). "The role of monetary policy", American Economic Review, 58
March, 1-17.
Fujimoto, T. and Leslie, D. (1983). "A two-class model of Keynesian
unemployment", Metroeconomica, 35 February-June, 54-71.
Garegnani, P. (1990). "Quantity of capital", in J. Eatwell, M. Milgate and P.
Newman (eds), The New Palgrave: Capital Theory, Macmillan, London.
Garrison, C.B. (1991). "Cyclical variations in the capital utilization rate",
Journal of Post Keynesian Economics, 13 Spring, 431-435.
Green, F. (1988). "Neoclassical and Marxian conceptions of production",
Cambridge Journal of Economics, 12 September, 299-312.
Harris, D.J. (1974). "The price policy of firms, the level of employment and
distribution of income in the short run", Australian Economic Papers, 13 June,
144-151.
Kahn, R.F. (1977). "Malinvaud on Keynes", Cambridge Journal of Economics, 1
December, 375-388.
Kaldor, N. (1985). Economics without Equilibrium, M.E. Sharpe, Armonk.
Kalecki, M. (1971). Selected Essays in the Dynamics of the Capitalist Economy,
Cambridge University Press, Cambridge.
Keynes, J.M. (1936). The General Theory of Employment, Interest and Money,
Macmillan, London (1976 edition).
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (21 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
King, J.E. (1999). "Some elements of a Post Keynesian labour economics", in S.C.
Dow and J. Hillard (eds), Keynes, Uncertainty and the Global Economy: Beyond
Keynes, vol. 2, Edward, Elgar, Aldershot.
Koutsoyiannis, A. (1984) "Goals of oligopolistic firms", Southern Economic
Journal, 51 October, 540-567.
Lavoie, M. (1992). Foundations of Post-Keynesian Economic Analysis, Edward
Elgar, Aldershot.
Lavoie, M. (1995). "The Kaleckian model of growth and distribution and its
neo-Ricardian and neo-Marxian critiques", Cambridge Journal of Economics, 19
December, 789-818.
Lavoie, M. (1996-97). "Real wages, employment structure, and the aggregate
demand curve in a Kaleckian short-run model", Journal of Post Keynesian
Economics, 19 Winter, 275-288.
Lavoie, M. (1998a). "Le chômage d'équilibre: réalité ou artefact statistique?",
working paper, Department of Economics, University of Ottawa.
Lavoie, M. (1998b). "Efficiency wages and reproducibility conditions in
Kaleckian models of employment", working paper, Department of Economics,
University of Ottawa.
Layard, R., Nickell, S. and Jackman, R. (1991). Unemployment: Macroeconomic
Performance and the Labour Market, Oxford university Press, Oxford.
Lucas, R.E. (1973). "Some international evidence on output-inflation tradeoffs",
American Economic Review, 63 June, 326-334.
Malinvaud, E. (1977). The Theory of Unemployment Reconsidered, Basil Blackwell,
Oxford.
Mason, P.L. (1993). "Variable labor effort, involuntary unemployment, and
effective demand: irreconcilable concepts?", Journal of Post Keynesian
Economics, 15 Spring, 427-442.
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (22 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
McCombie, J.S.L. and Thirlwall, A.P. (1994). Economic Growth and the
Balance-of-Payments Constraint, Macmillan, London.
Nell, E.J. (1988). Prosperity and Public Spending: Transformational Growth and
the Role of Government, Unwin Hyman, Boston.
Nistico, S., D'Orlando, F. and Scoppola, B. (1998). "Some questions for New
Keynesians", in R.J. Rotheim (ed.), New Keynesian Economics/Post Keynesian
Alternatives, Routledge, London.
Patinkin, D. (1965). Money, Interest, and Prices, Harper and Row, New York.
Pivetti, M. (1985). "On the monetary explanation of distribution", Political
Economy, 1, 73-103.
Riach, P. A. (1995). "Wage-employment determination in a Post-Keynesian world",
in P. Arestis and M. Marshall (eds), The Political Economy of Full Employment,
Edward Elgar, Aldershot.
Robinson, J. (1977). "Michal Kalecki and the economics of capitalism", Oxford
Bulletin of Economics and Statistics, February, 7-17.
Rowthorn, R.E. (1977). "Conflict, inflation and money", Cambridge Journal of
Economics, September 1, 215-239.
Rowthorn, R.E. (1981). "Demand, real wages and economic growth", Thames Papers
in Political Economy, Autumn, 1-39.
Sawyer, M.C. (1985). Macro-economics in Question: The Keynesian-Monetarist
Orthodoxies and the Kaleckian Alternative, M.E. Sharpe, Armonk.
Sawyer, M.C. (1995). "Conflict and aggregate demand in post-Keynesian economics:
the problem of over-determinacy", in Unemployment, Imperfect Competition and
Macroeconomics: Essays in the Post Keynesian Tradition, Edward Elgar, Aldershot.
Sawyer, M.C. (1998). "New Keynesian macroeconomics and the determination of
employment and wages", in R.J. Rotheim (ed.), New Keynesian Economics/Post
Keynesian Alternatives, Routledge, London.
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (23 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
Schefold, B. (1983). "Kahn on Malinvaud", in J. Eatwell and M. Milgate (eds),
Keynes's Economics and the Theory of Value and Distribution, Oxford University
Press, Oxford.
Shapiro, C. and Stiglitz, J. (1984). "Equilibrium unemployment as a worker
discipline device", American Economic Review, 74 June, 433-444.
Seccareccia, M. (1991a). "An alternative to labour-market orthodoxy: the
post-Keynesian Institutionalist policy view", Review of Political Economy, 3
(1), 43-61.
Seccareccia, M. (1991b). "Salaire minimum, emploi et productivité dans une
perspective post-keynésienne", Actualité Economique, 67 June, 166-191.
Setterfield, M. (1995). "Historical time and economic theory", Review of
Political Economy, 7 January, 1-27.
Steindl, J. (1952). Maturity and Stagnation in American Capitalism, Oxford
University Press, Oxford.
1. The top part of Figure 2 was suggested to us by David Colander, during a
seminar at the University of Ottawa in 1997. I have added the bottom part.
2. For instance, Greenwald and Stiglitz, despite being possible post-Walrasian
representatives, and arguing that real rigidities rather than nominal rigidities
are at stake, end up concluding that "even if there were large shifts in the
demand curve for labor, if the real wage were flexible, demand and supply of
labor would equilibrate (1993: 38).
3. More sophisticated models mix efficiency wage theory with the dual market
hypothesis (Carter 1998).
4. Van Ees and Garretsen (1996: 196) give a graphical instance of multiple
equilibria, but their non-linear labour demand curve has a hill shape, which
seems hardly justified (there are two possible employment levels for the same
real wage).
5. The PS-WS model has some definite resemblances with Rowthorn's (1977)
inflation model of conflicting claims, and specially with the model developed by
Malcolm Sawyer (1985, 1995), whose w-curve and p-curve emulate the WS-PS model.
Although Sawyer's p-curve is based on the profit-maximizing behaviour of firms
in monopolistic markets, it does not necessarily have the standard
downward-sloping shape. While Sawyer assumes that firms face rising marginal
costs at high levels of employment and capacity utilization, he argues that they
initially face decreasing marginal costs. As a result, claims Sawyer, real wages
as determined in the goods market initially rise with output, only to fall
later. His p-curve thus has a bell shape, similar to the one to be shown in
Figure 7.
6. It should be pointed out however that the introduction of the real rate of
interest in the PS function resembles the Sraffian claim that real interest
rates determine the normal rate of profit. Pivetti (1985), for instance, has
long argued that a regime of high real rates of interest would induce firms to
set higher margins of profit, and hence to negotiate lower real wages. See also
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (24 of 25) [19/09/2002 8:06:57 PM]
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt
Lavoie (1992: 361-371).
7. The expenditures of the unemployed have been assumed away, but as shown by
Bowles and Boyer (1990) and Lavoie (1992: 270), introducing them would not
change the analysis.
8. Another recent graphic exposition is to be found in Fazzari et al. (1998),
where the authors however focus their attention on the case of imperfect
competition.
9. It should be pointed out however that, as early as 1937, following the
empirical evidence submitted by Dunlop, Kalecki and Tarshis, Keynes was quick to
accept the fact marginal costs could be constant rather than increasing, and
hence, it could be said that the Kaleckian model to be presented is consistent
with the Keynes of the post General Theory period (Keynes 1936: appendix 3).
10. It should be mentioned that Pigou's effect or Keynes's effect, even if they
exist, may not manage to achieve full employment in this model, in contrast to
the preceding model with neoclassical features. Here, if market forces are free
to act upon flexible wages and prices, the presence of unemployment should drive
down real wages, while nominal wages and prices are falling. The favourable
effects of falling prices on aggregate demand would then be counter-balanced by
the unfavourable effects of falling real wages on aggregate demand. As a result,
even in theory, free market forces may not provide for full employment.
11. Ironically, in the case of Figure 10, a higher level of employment
associated with the high real wage unstable equilibrium would require weaker
labour unions, that would accept lower real wages at each level of expected
aggregate employment. This paradoxical behaviour is of course due to the
instability of the high real wage equilibrium.
12. See Garrison (1991) and Mason (1993). Note however that some heterodox
authors believe that it is impossible to have a cardinal measure of effort, even
when it can be considered as a one-dimensional variable. They would thus be
quite critical of the models presented here (Curie and Steedman 1993).
13. In the model as presented, such an occurence cannot happen. One must
introduce the more sophisticated aggregate demand components introduced by
Bowles and Boyer (1990). For an analysis of all this, see Lavoie (1998b).
file:///D|/Work/2002_Contrat/UNIVERSITY/ENGLISH/A/A_10_MEXICO_ENGLISH_ww.txt (25 of 25) [19/09/2002 8:06:57 PM]