University of Massachusetts Amherst

ScholarWorks@UMass Amherst

Economics Department Working Paper Series Economics

2005

Keynesian Theory and the AD-AS Framework: A Reconsideration

Amitava Krishna

Peter Skott
University of Massachusetts - Amherst

Follow this and additional works at: https://scholarworks.umass.edu/econ_workingpaper

Part of the Economics Commons

Recommended Citation
Krishna, Amitava and Skott, Peter, "Keynesian Theory and the AD-AS Framework: A Reconsideration"
(2005). Economics Department Working Paper Series. 47.
Retrieved from https://scholarworks.umass.edu/econ_workingpaper/47

This Article is brought to you for free and open access by the Economics at ScholarWorks@UMass Amherst. It has
been accepted for inclusion in Economics Department Working Paper Series by an authorized administrator of
ScholarWorks@UMass Amherst. For more information, please contact scholarworks@library.umass.edu.
DEPARTMENT OF ECONOMICS

Working Paper

Keynesian Theory and the AD-AS

Framework: A Reconsideration

by

Amitava Krishna Dutt and Peter Skott

Working Paper 2005-11

UNIVERSITY OF MASSACHUSETTS
AMHERST
 Keynesian Theory and the AD-AS Framework:
A Reconsideration

Amitava Krishna Dutt* and Peter Skott†

15 June 2005

Abstract

Contrary to what has been argued by a number of critics, the AD-AS framework is both
internally consistent and in conformity with Keynes’s own analysis. Moreover, the
eclectic approach to behavioral foundations allows models in this tradition to take into
account aggregation problems as well as evidence from behavioral economics.
Unencumbered by the straightjacket of optimizing microfoundations, the approach can
provide a useful starting point for the analysis of dynamic macroeconomic interactions.
In developing this analysis, the AD-AS approach can draw on insights from the Post
Keynesian, neo-Marxian and structuralist traditions, as well as from the burgeoning
literature on behavioral economics.

Key words: AD-AS, Keynes, New Keynesian theory, microeconomic foundations

JEL classification: E12, O11, B22, B41, B50.

*
Department of Economics and Policy Studies, Faculty of Economics, University of Notre Dame, Notre
Dame, IN 46556; email: adutt@nd.edu
†
Department of Economics, University of Massachusetts, Amherst, MA 01002; email:
pskott@econs.umass.edu.
1. Introduction

Several turn-of-the-century assessments of the state of macroeconomics regard the

discipline as healthy. There may have been fierce debates and controversies, but these

debates mainly served to highlight deficiencies of existing models and to stimulate the

creation of new improved hybrid models. The history of macroeconomics, according to

Blanchard (2000, p. 1375) is “one of a surprisingly steady accumulation of knowledge”,

and “progress in macroeconomics may well be the success story of twentieth century

economics”. Woodford’s (1999) assessment gives slightly more weight to the

disagreements and revolutions in the second half of the twentieth century. But Woodford

also sees convergence, and he concludes that “modern macroeconomic models are

intertemporal general equilibrium models derived from the same foundations of

optimizing behavior on the part of households and firms as are employed in other

branches of economics” (p. 31).

We disagree with these assessments. In our view, a large part of what has

happened in macroeconomics since the late 1960s has been a wasteful detour. A

generation of macroeconomists has grown up learning tools that may be sophisticated,

but the usefulness of these tools is questionable. Moreover, a great deal of damage may

be, and has been, done when the tools are applied to real-world situations.

In this paper we shall argue that, for all their limitations, the simple models of the

old Keynesian school using the Aggregate Demand-Aggregate Supply (AD-AS)

framework provide a better starting point for serious analysis than more recent models in

the New Keynesian (NK) or Real Business Cycle (RBC) traditions which have come to

dominate modern macroeconomics. The obsession with optimization and microeconomic

1
foundations has meant that a promising Keynesian research program has been largely

abandoned. Our preferred version of this program builds on the old Keynesian models,

but extends and modifies them to accommodate insights from the post Keynesian, neo-

Marxian, and structuralist traditions.

The main emphasis in this paper is on the usefulness and shortcomings of the

basic AD-AS framework, and on how the framework may be developed and improved.

Section 2 outlines a standard version of the AD-AS model and shows that it can be given

a logically consistent Marshallian interpretation. It also shows that the model does not, as

claimed by some critics, suffer from internal logical contradictions. Section 3 discusses

some alleged shortcomings of the model. Section 4 considers the NK alternative,

focusing on two main issues: microeconomic foundations and the treatment of stability.

Section 5 introduces post Keynesian and other arguments for the relevance of aggregate

demand, not just in the short run but also as an influence on real outcomes in the medium

and the long run. Section 6, finally, ends with a few concluding remarks.

2. The AD-AS framework

Following Keynes, the AD-AS approach visualizes the economy as a whole, that is, the

theory is ‘general’ rather than ‘partial’.1 Keynes’s (1936/1973) derivation of a fix-wage

general equilibrium in chapters 1-18 of The General Theory of Employment, Interest and

Money (GT) was an enormous intellectual achievement, and the one stressed by both

Blanchard and Woodford in their accounts of the Keynesian revolution. The AD-AS

framework gives a reasonable representation of the analytical skeleton behind this fix-

2
wage general equilibrium. The strength of the AD-AS apparatus is precisely the explicit

attempt to integrate the analysis of goods, labor and financial markets.

The AD-AS framework divides the economy into two parts – the ‘demand side’

and the ‘supply side’ – and examines their interaction using accounting identities,

equilibrium conditions and behavioral and institutional equations. The ‘demand side’

typically examines factors relating to the demand for goods and the demand and supply

of assets. The ‘supply side’ typically examines factors relating to output and pricing

decisions of producers, and factor markets. The framework ensures that neither demand

nor supply side factors are overlooked in the analysis and that macroeconomic outcomes

depend on the interaction between the different markets. The particular partitioning into

‘aggregate demand’ and ‘aggregate supply’ along with the choice of terminology may

provide the pedagogic advantage of making macroeconomic analysis possible in terms of

the same tools as the simplest microeconomic model of the market. But this advantage

comes at a high price. The aggregate demand and supply curves embody complex

interactions and are clearly not the same as the microeconomic curves which take a

partial view of the economy. The analogy therefore is spurious, and forgetting this has

led to a great deal of confusion in the literature, as briefly discussed later.

The basic AD-AS model is well-known, of course, but to ease the exposition it is

helpful to state a simple version of it explicitly. There are two equilibrium conditions

Y=C+I+G (1)

M/P = L, (2)

3
where, in standard notation, Y is real output, C, I and G, denote real consumption,

investment and government expenditure, M the supply of money, P the price level, and L

the real demand for money, and six behavioral or institutional equations

C = C(Y), (3)

I = I(r), (4)

L = L (Y, r), (5)

Y = F(N), (6)

W/Pe = F’(N) (7)

W = W0 (8)

where 0<C’<1, I’<0, L1>0, L2 < 0, F’>0 and F”<N, and where r is the rate of interest, N

the level of employment, W the money wage, and Pe is the price expected by firms.

Equations (3) through (6) are standard consumption, investment, money demand and

production functions. Since C, I and L are used to denote desired amounts in equations

(3) through (5), equations (1) and (2) are equilibrium conditions (rather than accounting

identities) showing that output is equal to the demand for it and that the money supply in

real terms is equal to the demand for it. Behind these equilibrium conditions lie dynamic

adjustment processes with excess demand for goods leading to an increase in P and

excess demand for money leading to an increase in r.2 Equation (6) is the profit

maximizing condition of firms that are assumed to be price takers in perfectly

competitive markets; since there is a production lag and firms make production plans

prior to knowing what price they will receive for their goods, the price that is relevant for

their production decision is the expected price. The levels of M, G and W are given

4
exogenously. To stress that this is the case for the money wage, equation (7) states that

the money wage is given at the exogenous level W0.

Our interpretation of the model is Marshallian and we examine the behavior of the

economy in two different ‘runs’. The expected price and the level of output are given in

the ‘market’ (or ‘ultra-short’) run. In the ‘short’ run expected price changes in response

to its deviations from the actual price, and this change is accompanied by changes in the

level of production; in a short-run equilibrium expectations are being met and the

expected and actual price coincide.

P P, Pe
IS
AS
LM AD

P1

Pe1

r
Y
Y1

Figure 1

In the market run, given Pe, and given W from equation (8), N is determined by

equation (7), and Y by equation (6). For this level of Y, substitution of equations (3) and

(4) into equation (1) yields a value of r which satisfies that equation, irrespective of the

price level. The IS curve in Figure 1, which shows equilibrium in the goods market in

(P,r)-space , is vertical at this level of r.3 The vertical arrows show the direction of price

5
adjustments when the economy is out of goods market equilibrium. Also for the given

level of Y, substitution of equation (5) into equation (2) yields a positive LM relation

between P and r, which represents money (and assets) market equilibrium. The

horizontal arrows show the direction of interest adjustments when the economy is out of

money-market equilibrium. The intersection of the IS and LM curves gives the market-

run equilibrium values of P and r. The equilibrium value of r is determined by the

position of the vertical IS curve, and the LM curve determines the value of P. With off-

equilibrium dynamics given by the equations in note 2, it is readily seen that the market-

run equilibrium is stable for a given Y.

In the short run Pe is allowed to change in response to unfulfilled expectations.

When Pe changes to a new level, firms adjust their employment and output levels. This

adjustment is captured by the AS curve, which shows the profit-maximizing level of

output produced by the firms for a given Pe. When Y changes, the IS and LM curves shift

in (r,P)-space and determine a new market-run equilibrium of r and P. The level of P

which clears goods and money markets for each level of Y is shown along the AD curve.

A higher level of Y increases the level of saving, so that goods market equilibrium

requires an increase in investment, a fall in r and hence a leftward shift of the IS curve.

A higher level of Y increases the real demand for money, so that money market

equilibrium requires a fall in P (or an increase in r), so that the LM curve shifts to the

right in (r,P)-space. Consequently, a higher Y implies a lower P for market-run

equilibrium, explaining the negative slope of the AD curve.

The short-run dynamics shown in Figure 1 can be described as follows. Starting

from an initial level of expected price, Pe1, output is determined at Y1 (as shown by the

6
AS curve) and price at P1 (as shown by the AD curve). Since P1 > Pe1, if firms revise

their price expectations adaptively, Pe rises, making Y expand along the AS curve and the

market-run equilibrium move along the AD curve (representing shifts in the IS and LM

curve) as shown by the arrow. This adjustment will continue till the economy arrives at

the short-run equilibrium at the intersection of the AD and AS curves, where P = Pe.4

Three comments about this model are in order. First, the Marshallian

interpretation of the model finds a great deal of exegetical support in Keynes’s own work

and in the writings of many Keynesians. Clower (1989), for instance, notes the

Marshallian aspects of Keynes’s GT, although not as precisely as done in our model (see

Dutt, 1992).

Second, the Marshallian interpretation is important for the internal consistency of

the economic argument. It has been argued by Barro (1994), Fields and Hart (1990),

Colander (1995) and Bhaduri, Laski and Riese (1999) that the AD-AS model embodies

two mutually-contradictory approaches to pricing and production by firms. According to

this criticism, the AD curve is based on IS and LM curves, but the analysis assumes that

firms fix the price (having the ability to do so) and that equilibrium levels of r and Y are

determined from equations (1) and (2), using (3) through (5). The story told is that firms

fix their price and adjust their output in response to changes in demand conditions. The

AS curve, on the other hand, assumes price taking behavior on the part of firms operating

in purely competitive markets with demand constraints, producing to maximize profits

given the money wage and the production function. While some textbook versions of the

AD-AS model do suffer from this inconsistency, our Marshallian model is free of it. The

equations of the model are similar to those of the standard textbook version,5 but in our

7
interpretation the AD and AS curves both embody profit maximization and price-taking

behavior: the AD curve in our interpretation shows equilibrium price for a given level of

output and not, as the standard AD curve, the equilibrium value of Y for different levels

of P. 6

Third, the model can easily be recast using Keynes’s own ‘AD-AS diagram’ with

employment and the value of output (price times quantity) on the axes (he did not

actually draw this diagram in GT, but described it in words in chapter 3).7 Keynes’s

Aggregate Supply function is given by W0F(N)/F’(N) and is derived from equations (6),

(7) and (8): its curve shows the expected value of output at each level of employment

consistent with profit maximizing behavior. The Aggregate Demand function is derived

from equations (1) through (6), and its curve shows the actual equilibrium proceeds (PY)

for any given level of N. The level of N determines Y from equation (6), and given this Y,

P is determined as shown in the IS-LM diagram of Figure 1, which determines the

equilibrium level of PY. The value of aggregate demand at the intersection between the

supply and demand curves defines “the effective demand” (GT, p. 25).

By construction expectations are being met at the point of effective demand. In

chapter 5 of GT, however, Keynes discusses the formation and revision of short-period

expectations, showing how firms produce a certain level of output with a certain level of

employment, given short period expectations, and then adjust these expectations if they

are not fulfilled. Though he does not explicitly analyze this process, we can do so by

using the expected proceeds curve, given by PeY = PeF(N), for a given Pe from equation

(6): it shows what firms expect the value of output to be for a given price expectation.

The intersection of this curve with the curve for the Aggregate Supply function

8
determines the market-run equilibrium level of employment since it satisfies equation (7).

For the market-run equilibrium employment level, one can read off actual proceeds from

the Aggregate Demand function. If actual proceeds are different from expected proceeds,

Pe will change, shifting the expected proceeds curve, till the economy arrives at short-run

equilibrium at the intersection of all three curves.8 For most of the GT, however, Keynes

confines attention to short-run equilibrium in which actual and expected price are equal,

thereby concealing the Marshallian adjustment process because it was not central to his

demonstration of the possibility of unemployment short-run equilibrium.9

3. Shortcomings

An AD-AS model of the type just described has many well-known weaknesses and

limitations, of which three are relevant for our purposes.

The criticisms that have received the most attention concern the alleged lack of

microeconomic foundations of the model. NKs (along with new classical economists and

RBC theorists), who have been vocal in this criticism, wish to supplant the model with

models based on explicit optimization. We shall take up the issue of optimizing

microfoundations in section 4 where we discuss the NK approach. But the behavioral

approach of the AD-AS model has also been criticized from another angle. Many post

Keynesian economists, but also some impeccably mainstream old Keynesians, have

suggested that the model is too mechanical and does not take into account uncertainty and

expectations in a serious manner.10 It is beyond the scope of the present paper to address

this important issue in any detail but in our view, ‘mechanical’ mathematical

formalization can be extremely useful. This formalization needs to be supplemented by

9
verbal descriptions and empirical analysis, and less formal discussions of possible

outcomes may also come into play if the relations determining the evolution of the

system are not capable of being formalized in a precise manner. Even this informal

discussion, however, will often benefit from using more formal analyses as points of

reference and by suggesting where and how the results of the models may need to be

modified.

A second set of criticisms claims that the AD-AS model omits many important

features of reality and that some of its implications are not consistent with empirical

observation. Assumptions of imperfect competition, for instance, should replace perfect

competition, and the money supply should not be treated as an exogenous variable in an

economy with modern monetary institutions.11 The consumption function should also

take into account income distributional effects on consumption, increases in aggregate

demand should provide a direct stimulus to investment, and the distinction between

nominal and real rates of interest may be critical (not least for the reactions of aggregate

demand to changes in money wages and the stability of full employment). These (and

other) modifications may complicate the model and affect some of its properties, but in

principle their introduction is quite straightforward and the resulting model can still be

depicted with AD and AS curves (see, for instance, Dutt and Skott, 1996). The

modifications, moreover, help to address some of the empirical criticisms of the AD-AS

model. The simple model, for instance, predicts a counter-cyclical movement of the real

wage. This implication, which finds little support in the data (as noted early on by

Dunlop, 1938, Tarshis, 1939), no longer holds in versions of the model that include

imperfect competition (perhaps with markup pricing à la Kalecki, 1971) and some

10
combination of non-diminishing returns to labor and/or a counter-cyclical pattern in the

markup.

A third set of problems with the AD-AS model concerns the unsatisfactory

treatment of dynamics. There is a lack of integration between the analysis of the short-run

and more long-term issues, and even when it comes to the treatment of the short run, the

analysis often relies on unstated or questionable assumptions concerning the process

leading to a short-run Keynesian equilibrium. Our own presentation above is quite

explicit in its assumptions (notes 2 and 4) but, perhaps unrealistically, it presumes that the

adjustment to market-run equilibrium is ‘very fast’ relative to the adjustments of price

expectations. The adjustment to market-run equilibrium could therefore be based on

given price expectations, and in the analysis of adjustments to short-run equilibrium it

could be assumed that there is continuous market equilibrium during the adjustment

process.12

The shortcomings of simple AD-AS models with respect to dynamics may be a

legacy of Keynes's own focus on short-run equilibria in GT. The assumption of fulfilled

expectations facilitated the presentation of the fix-wage general equilibrium.13

Unfortunately, it makes it hard to discuss the stability issues, and from today's

perspective – having before us a well-developed theory of general equilibrium – the truly

revolutionary and provocative message of the GT concerns the destabilizing effects of

money wage flexibility, rather than the existence of a fix-wage equilibrium with

unemployment.

The AD-AS model does not address the stability issue – it takes the money wage

as given - but can serve as a starting point. The model can be easily extended in a way

11
which makes it have the implications presented in the typical textbook: (i) that

unemployment can exist in the model because the money wage is exogenously fixed; (ii)

that if one allows the money wage to fall in response to the existence of unemployment,

the AS curve, given by P = F’(F-1(Y))/W is shifted downwards; and that (iii) this leads to

an expansion of output and employment along the negatively-sloped AD curve and

moves the economy to the ‘natural rate of unemployment’ (corresponding to the absence

of Keynesian involuntary unemployment). The mechanism behind this adjustment is the

‘Keynes effect’ by which a reduction in wage and price increases the real supply of

money, lowers the interest rate, and increases investment and aggregate demand. This

effect can be supplemented by the real balance effect by which the rise in real balances

directly stimulates the aggregate demand for goods.

This standard analysis is at odds with Keynes’s own argument in GT where, in

chapter 19, he insisted that involuntary unemployment would not be eliminated by

increased wage flexibility. Falling money wages will influence the economy in a number

of ways but, on balance, are unlikely to stimulate output.14 Keynes's analysis of the

effects of changes in money wages may have been sketchy, but the logic behind potential

instability is impeccable. The real balance effect was overlooked by Keynes, but has been

found to be empirically insignificant, and the expansionary effects of a decline in money

wages due to the Keynes effect may be more than offset by the adverse influences of debt

deflation, distributional shifts, and expectations of continuing reductions of wages and

prices. ‘Old Keynesians’ have been aware of these stability problems (see Tobin, 1975),

and post Keynesians have stressed additional problems arising from the role of

12
uncertainty, the financial situation of firms and the effects of an endogenous money

supply.

These complicating factors can be addressed by an informal discussion of the

diverse effects of money wage changes, using the AD-AS model as the starting point.

This is basically what Keynes did in chapter 19 of GT. The analysis and the destabilizing

effects can be illustrated using the AD-AS diagram (see Dutt and Amadeo, 1990). For

instance, debt deflation problems can make the AD curve upward-sloping and, in

addition, money wage reductions can shift the AD curve to the left (because of a higher

propensity to consume out of wage income than non-wage income), both of which

prevent the economy from converging to the ‘natural’ level of output. The analysis can

be made more precise by incorporating specific effects into more general Keynesian

models in order to formally examine the stability question, as done by Chiarella and

Flaschel (2000), among others. Their analysis demonstrates that the Keynesian models

can generate very complex dynamics and that local instability is a likely outcome for

plausible specifications.

4. The New Keynesian detour

The New Keynesian approach can be characterized as one which attempts to derive

Keynesian conclusions with respect to the existence of unemployment equilibrium and/or

the effectiveness of aggregate demand policy, while using a standard neoclassical

methodology.

Unemployment equilibrium can be explained in terms of the optimizing behavior

of agents in models that depart from Walrasian perfect competition by introducing

13
perceived demand curves for imperfectly competitive firms, asymmetric information,

efficiency wages, credit rationing, and the like.15 Some of these models are very

insightful, but they largely fail to address the issue of involuntary unemployment in

Keynes’s sense. Keynes explicitly defined ‘voluntary unemployment’ to include all

frictional and structural unemployment, that is, to include unemployment caused by

minimum wage legislation and excessive union wage demands, for instance. By

extension, Keynes’s notion of voluntary unemployment also includes structural

unemployment generated by the various departures from perfect competition that have

been invoked by NK. Structural unemployment of this kind may be theoretical interesting

and empirically significant, but it is not the kind of unemployment addressed by Keynes.

His involuntary unemployment is defined in terms of inadequate aggregate demand and

the failure of the market mechanism to ensure the adjustment of aggregate demand to the

level of aggregate supply associated with a structurally determined (minimum) rate of

unemployment. It is the deviation from a structural unemployment rate that makes

demand policy desirable.

In NK models the effectiveness of aggregate demand policy is confined to the

short run and derives from nominal wage and price rigidities. Some of the early NK

models were of the spanner-in-the-works variety which merely introduced nominal wage

and price rigidities into new classical or RBC models with rational expectations. But the

NK methodology requires that such rigidities be based on optimizing behavior: “rather

than postulating that prices and wages respond mechanically to some measure of market

disequilibrium, they are set optimally, that is, so as to best serve the interests of the

parties assumed to set them, according to the information available at the time”

14
(Woodford 2003, p.7). Thus, prices and wages are set in a forward-looking manner,

expectations are assumed to be rational, and preferences are regarded as structural and

invariant to changes in policy.

Our comments on the NK approach focus on two issues: the obsession with

microeconomic foundations based on explicit optimization, and the treatment of stability

issues. The two issues are related since the obsession with optimization stands in the way

of serious stability analysis.

4.1 Optimization

Optimization, in our view, can sometimes be very useful as a simple way of describing

goal-oriented behavior (indeed, both our simple AD-AS model and Keynes’s own

analysis included the assumption of profit maximizing firms). But insisting on

optimization can also result in problems. The problems with the optimization approach

are largely well-known and a brief summary of some of the main points will suffice.

The cognitive limitations and bounded rationality of all real-world decision

makers have been stressed by many authors, most notably perhaps by Simon, and a more

recent literature has documented the existence of systematic departures from optimizing

behavior (see Kahneman, 2000, and Camerer et al., 2004). From this perspective the NK

demand for optimizing microeconomic foundations is remarkable primarily because of

the highly restrictive form that it takes.16

Aggregation represents another problem for the optimizing approach. To obtain

definite results, any theory of the economy as a whole has to engage in aggregation.

Thus, there can be no attempt at full disaggregation in the agent space, as in Arrow-

Debreu models of general equilibrium, and it is well-known that even if all individual

15
agents were fully rational and maximized well-behaved utility functions subject to

standard constraints, the aggregate variables do not behave as if determined by an

optimizing representative agent (see, for instance, Kirman, 1992). Aggregation problems

therefore imply that the use of an optimizing representative agent in NK models has little

to recommend itself.

The existence of social norms and conventions provides a further reason to

eschew the mechanical application of optimization methods based on exogenously given

and constant preferences. The role of relative wages and norms of fairness in Keynes’s

GT analysis of wage formation presents an example of this perspective. The existence of

norms and conventions may be a source of ‘conditional stability’ in Keynesian models of

uncertainty (Crotty, 1994) but norms and conventions also change over time, both

endogenously and as a result of exogenous shocks. We shall return to these issues in

section 5 below.

A more subtle danger of the optimization approach is that it may predispose the

analysis to slide from individual ‘rationality’ to systemic ‘rationality’. Some economists

may view optimization is simply an organizing principle (see note 16), but countless

examples suggest that an optimization approach may generate (sometimes unconsciously)

a slippery slope in which individual optimization eventually leads to social optimality.

Sargent (1993), for instance, is able to assume bounded rationality and yet produce,

eventually, his unique, new classical equilibrium. As a second example, many of the

problems caused by efficiency wage considerations can be ‘solved’ when credit markets

function efficiently (again, with clever institutions). A history of how a focus on

16
individual optimization in neoclassical economics inexorably, albeit tortuously, has led to

presumptions of social optimality awaits an author, if one does not exist already.

A serious problem, finally, arises from the bounded rationality of the theorist.

Carrying the straightjacket of optimization – especially in its dynamic versions – reduces

the ability of the theory to incorporate many important aspects of reality in a tractable

manner, and therefore encourages the theorist to ignore them. One may insist on treating

all agents in a model as fully optimizing, but there is a cost to meeting this demand.

Simplifications then need to be made in other areas in order to keep the model tractable;

the number of distinct agents, for instance, may have to be kept very small and the nature

of the interaction between the agents very simple.

All useful models, of course, represent drastically stylized pictures of a complex

reality. The art of model building consists in choosing appropriate simplifying

assumptions, and in our view the insistence on fully optimizing behavior represents a

suboptimal ‘corner solution’ to the modeling problem: the gains from explicit

optimization are often minimal and the costs of the required simplifications in other areas

high. Thus, over the last 30 years macroeconomists have struggled to solve problems of

intertemporal optimization. These optimization problems grossly simplify real-world

decision problems, and the astounding implicit presumption has been that agents in the

real world solve (or act as if they had solved) these much more complex problems. The

neglect of aggregation problem and the use of representative agents in models that

purport to provide microeconomic foundations only serve to make the picture even more

bizarre. In fact, the contemporary approach with its sophisticated and perfectly rational

representative agents would seem to embody a good example of how not to use

17
mathematics: mathematical models arguably are useful primarily because they allow a

clear analysis of complex interactions between agents, each of whom may follow

relatively simple (but possibly changing) behavioral rules.

4.2 Stability and rational expectations

NK models may include non-clearing labor markets and allow for real effects of

aggregate demand policy. But it is assumed that, in the absence of shocks, the economy

converges to an equilibrium position, and cyclical fluctuations are generated by

introducing stochastic shocks into models with a stable equilibrium solution. If only

prices and wages were flexible, there would be no Keynesian problems of effective

demand.

The stability concerns that were at the centre of Keynes’s message have been

largely forgotten.17 Is there a NK answer to these stability concerns? Not really. Stability

is simply assumed in NK models. The models typically involve saddlepoints and jump

variables, and the presumption of stability is used to pin down the outcome in the short

run. Agents have rational expectations, and the jump variables seek out the stable

saddlepath. Thus, to the extent that there is an answer, it comes from the NK focus on

microfoundations and rational expectations, and from the implicit rejection of the old

Keynesian analysis because of its alleged deficiencies in these areas.

Rational expectations have been used before Muth and Lucas, although without

using that name. Keynes’s own GT approach of assuming that short-period expectations

are fulfilled is an example of rational expectations in the sense of perfect foresight, and

Harrod's (1939) warranted growth path also represents a rational expectations path. But

the extension of rational expectations to all models - and not just steady growth paths or

18
Robinsonian mythical ages - lacks both theoretical and empirical foundations. We

confine our attention to a few observations about theory.

The theoretical argument relies on the claim that the systematic deviations

characterizing other specifications would lead to changes in expectation formation. This

claim has some force and, indeed, changing expectations may be an important source of

instability (as suggested by the role of ‘animal spirits’ in Keynesian analyses). But the

claim this does not justify a focus on rational expectations. It has been notoriously

difficult to get convergence to rational expectations even in simple models of rational

learning, and the real-world learning process takes place within a complex overall

environment and one that is subject to constant and profound technical and institutional

change (Frydman and Phelps, 1983). These changes in the environment may lead to shifts

in expectations; indeed, some institutional or structural change is often invoked to justify

expectations that would otherwise seem unreasonable, viz. the appeal to a ‘new economy’

during the stock market boom of the 1990s. However, structural and institutional changes

of this kind count against rational expectations since the learning processes underlying

the claims in favor of rational expectations fare better in a stable environment.18

It should be noted, finally, that a dismissal of stability concerns cannot be justified

by reference to Walrasian general equilibrium theory. In fact, the realization that stability

had not and probably could not be established under reasonable assumptions may have

been a critical factor behind the virtual abandonment in microeconomics of all research

on Walrasian general equilibrium theory (Kirman, 1989, Katzner 2004).19

The use of individual optimization therefore does not imply that one can ignore

stability issues, and in fact not all contributions that can be called NK have ignored these

19
issues. A notable exception is the work of Hahn and Solow (1995), who develop an

overlapping-generations model and introduce real money balances using a variant of the

Clower constraint to show that wage-price flexibility can result in macroeconomic

instability. They also show that wage and price sluggishness as explained by standard NK

techniques can be stabilizing but also prevent the economy from attaining full

employment. However, unlike many of the other elegant theoretical contributions of

these authors, the model becomes extremely unwieldy, primarily due to its optimizing

assumptions (despite the artificial way money is introduced) and they have to resort to

simulation techniques to examine the behavior of the economy.

5. Post Keynesian, structuralist and neo-Marxian alternatives

The AD-AS tradition – including the recent work on ‘integrated Keynesian

disequilibrium dynamics’ by Chiarella and Flaschel and their associates – rightly stresses

the need to consider dynamic interactions across markets, and it is justifiably critical of

optimization methodology. But theories in the AD-AS tradition need to be developed not

just in terms of more advanced mathematical analysis of the dynamic interactions but

also in terms of a renewed attention to the behavioral assumptions and their implications

for the specification of the various equations.

The behavioral foundations, of course, have not been neglected in the Keynesian

literature, as is evident from even a cursory look at Keynes's own analysis or the efforts

of many old Keynesians. Nonetheless, some of the presumptions of the AD-AS tradition

seem questionable from a heterodox perspective. A post Keynesian approach questions

the limited role of aggregate demand in determining medium- and long-run growth

20
patterns in AD-AS models; a neo-Marxian approach suggests a greater focus on income

distribution and its interaction with the rate of accumulation and the movements in the

‘reserve army of labor’; a structuralist approach (see Taylor, 1991, 2004) emphasizes the

need to examine how the structural and institutional characteristics of economies

determine their dynamics.

It is beyond the scope of this paper to discuss the behavioral alternatives in any

detail. We shall confine ourselves to a couple of examples of what we have in mind. The

examples concern assumptions that affect the role of aggregate demand in the medium

and long run, and we shall focus on medium- and long-run steady states rather than the

stability of these steady states.

5.1 The medium run: Fairness and the ‘natural rate of unemployment’

The existence of a ‘natural rate of unemployment’ has been a mainstay of NK models,

and the extensions of the AD-AS models by Chiarella and Flaschel (2000) share this

feature; the natural rate of unemployment may not be asymptotically stable in their

models, but cycles take place around a structurally determined long-run equilibrium

(except for a brief sketch in their final chapter on the “road ahead”). The existence of a

natural rate of unemployment implies that aggregate demand plays (almost) no role in the

determination of the trend of output and the average long-run value of the unemployment

rate. We find this aspect of the models questionable, both empirically and theoretically.

Money wages may be sticky partly because workers care about relative wages (as

suggested by Keynes). This argument implies a rejection of a traditional view of

preferences as defined over the agent's own consumption. Instead, a notion of fairness

becomes central, and the behavioral literature has provided strong support for the role of

21
‘fairness’ in wage formation (see, for instance, Bewley, 1998, Fehr and Gächter, 2000,

Akerlof and Yellen, 1990). The literature also shows that changes in nominal wages are

relevant for the perceived fairness of the wage offer. The relevance of nominal changes

implies a kind of ‘money illusion’. As a result, there is no natural rate of unemployment.

Instead, a downward sloping Phillips curve emerges, and demand policies may affect real

output and employment in the medium and long run (Shafir et al., 1997, Akerlof et al.,

1996).

A more radical conclusion can be obtained if it is recognized that norms of

fairness may change over time and that the prevailing wage norms are strongly

influenced by the actual wage patterns in the past. Thus, according to Kahneman et al.

(1986, p. 730-1) notions of fairness tend to adjust gradually to actual outcomes:20

the reference transaction provides a basis for fairness judgments because it is normal, not because
it is just. Psychological studies of adaptation suggest that any stable state of affairs tends to
become accepted eventually, at least in the sense that alternatives to it no longer readily come to
mind. Terms of exchange that are initially seen as unfair may in time acquire the status of
reference transaction. Thus, the gap between the behavior that people consider fair and the
behavior that they expect in the market-place tends to be rather small.

Skott (1999, 2005) shows that this conventional aspect of wage norms may lead to

employment hysteresis, even in models that exclude money illusion of any kind.21 If

inflationary expectations are formed adaptively and adjustments in wage norms take a

simple linear form, the models generate a downward-sloping Phillips curve. In general,

however, aggregate demand policy will affect output in the medium run, but there will be

no well-behaved Phillips relation, vertical or downward–sloping, between employment

and the inflation rate.

These examples illustrate how lessons from behavioral economics may cast doubt

on the natural rate hypothesis.22 Theoretical doubts might not carry a lot of weight if the

22
empirical evidence was overwhelming, but this is not the case. Even strong supporters of

the framework concede that the applicability of the theory may be limited. Thus, Gordon

(1997, p. 28) concludes that

Within the postwar experience of the United States, the modest fluctuations in the NAIRU seem
plausible in magnitude and timing. When applied to Europe or to the United States in the Great
Depression, however, fluctuations in the NAIRU seem too large to be plausible and seem mainly
to mimic movements in the actual unemployment rate.

From a Popperian perspective, Gordon's reading of the evidence must imply that the

theory should be rejected.

5.2 The long run: Growth, accumulation and technological change

Models of the long run, which introduce capital accumulation, technological change and

labor supply growth, are generally of two varieties.

By far the more popular one is the one in which aggregate demand disappears

from the scene and aggregate supply determines growth. In fact, neoclassical growth

theory following Solow (1956), and new growth theory, following Romer (1986) and

others, abstracts entirely from the AD side, assuming perpetual full employment and

investment being determined identically by saving. The debate between neoclassical and

new growth theory revolves around whether or not the marginal product of the produced

factor of production, capital, falls to zero as the capital-labor ratio rises indefinitely and,

therefore, whether long-run growth is affected by the saving rate and other economic

variables. The neglect of AD is usually not explicitly explained in these models, but it is

implicitly assumed that wage and price flexibility will remove unemployment in the

medium run or, failing that, that government aggregate demand policy will do the job.

Thus, the long-run growth path is independent of AD factors.

23
 A less popular variety, with roots in the Keynesian theories of Harrod (1939),

Robinson (1962) and others, focuses on AD as determining growth. In these models

growth is determined by the interaction between aggregate demand and supply factors

(including, for instance, firms’ pricing decisions). Some work in this tradition has

included the labor market explicitly and linked the long-run rate of growth of output to

the growth of the labor supply in efficiency units (see, for instance, Kaldor 1957, Skott

1989, Dutt 1992a). Most models, however, do not impose the requirement that the

unemployment rate be constant in the long run but simply assume that the labor supply

does not constrain the rate of growth (see Marglin, 1984, Dutt, 1984, Taylor, 1991).

These models have many interesting implications, including the possibility that a more

equal distribution of income can increase the rate of growth and that technological

change can have immiserizing effects, and the assumption of no labor constraints can be

defended by pointing to the existence of large amounts of hidden unemployment in the

primary and tertiary sectors in most countries, developed as well as less developed, until

some time in the post World War II period. For the more recent period, however, the

hidden-unemployment argument may not be persuasive, at least for advanced industrial

countries. Most of the OECD economies arguably have become ‘mature’ in Kaldor’s

(1966) sense: they certainly have unemployment, both open and disguised, but it would

be misleading to treat the labor supply to the modern sector as perfectly elastic and to

disregard the labor constraints on the long-run rate of growth. Even under conditions of

maturity, however, the rate of growth may be influenced by aggregate demand.

As argued in section 5.1, the rate of employment can not be taken as independent

of the demand side, even in the medium run, and this dependence of employment on

24
aggregate demand opens up ways in which demand may also influence the rate of growth

in the long run.

One channel runs through migration. Even if a country has exhausted its domestic

reserves of hidden unemployment, the possibility of immigration provides an

international reserve army and, immigration laws permitting, the growth rate of the

country need not be limited by its labor supply. Immigration laws respond to economic

conditions (as evidenced, for instance, by the change in attitudes of European countries

between the 1960s and the more recent period), and the employment rate can therefore

have a significant effect on the rate of growth of the labor force.23

Induced technical progress represents a second possible channel. Labor shortages

provide an incentive for firms to seek out new labor saving techniques, and this

technology channel suggests that the rate of growth of the labor supply in efficiency units

may be positively related to the employment rate. Both the employment and technology

channels imply that insofar as aggregate demand policy influences the rate of

employment, it also affects the long-run rate of growth (Flaschel and Skott, 2005).24

A more radical approach is pursued by Dutt (2005) who considers a range of

models in which the rate of labor productivity growth responds to labor market

conditions, with tight labor markets speeding up labor-saving technological change. One

of the models makes the employment rate affect both changes in the ‘autonomous’

investment parameter (to capture the effects of unemployment and wage reductions on

aggregate demand through the Keynes effect) and the rate of labor productivity growth.

Since the same rate of employment makes investment and labor productivity growth

stationary, the result is a zero root model in which a change in the level of autonomous

25
demand (for instance, government expenditure) has a permanent effect on the long-run

rate of growth. The economy converges to its long-run rate of growth, at which the

economy grows with unemployment at its ‘natural’ rate, but the long-run rate of growth

itself is affected by aggregate demand. AD and AS grow at the same rate, but the growth

rate of the economy is not independent of factors determining AD.

6. Conclusion

We have argued in this paper that the older Keynesian tradition based on the aggregate

demand-aggregate supply framework provides a more suitable and promising framework

for building macroeconomics than the currently-dominant approach, including its New

Keynesian variant. This is so for a number of reasons.

Contrary to what has been argued by a number of critics, first, the traditional

aggregate demand-aggregate supply approach is internally consistent, at least in its

Marshallian interpretation, as well as consistent with Keynes’s own analysis.

Second, it has the strength of explicitly including the major markets and sectors of

the economy and examining their interactions. In this sense it is a general, rather than a

partial, theory. Walrasian general equilibrium theory may also be general in this sense,

but is different in several ways, including the perspective on behavioral foundations.

Third, the aggregate demand-aggregate supply approach does not insist on

optimizing microfoundations. The AD-AS model is not necessarily inconsistent with

optimizing behavior, but the approach is eclectic. It starts with some basic and

commonly-used accounting identities, adds rules of behavior of individuals or groups in

specific institutional settings, and examines their consequences for the performance and

26
evolution of the system. The theorist must be prepared to explain and defend the choice

of behavioral rules, but an appeal to optimization is neither necessary nor sufficient for a

successful defense. This eclecticism, we have argued, is a strength, and the New

Keynesian methodological position is flawed. New Keynesian macroeconomics has

produced interesting insights, but the insistence on optimizing microfoundations means

that these insights have come at the cost of neglecting a variety of important issues,

including the analysis of stability.

Fourth, it is true that a great deal of analysis using the aggregate demand–

aggregate supply framework is mechanical and fails to capture important aspects of

reality, and its extensions to medium- and long-run issues typically ignore the role of

aggregate demand. However, unencumbered by the straightjacket of optimizing

microfoundations, the approach provides a useful starting point for the analysis of

dynamic macroeconomic interactions. In developing this analysis, the approach can draw

on insights from the Post Keynesian, neo-Marxian and structuralist traditions, as well as

from the burgeoning literature on behavioral economics.

27
 REFERENCES

Akerlof, G.A. and Yellen, J.L. (1987). “Rational Models of Irrational Behavior”,
American Economic Review, May, 137-142.

Akerlof, G.A. and Yellen, J.L. (1990). “The Fair Wage-Effort Hypothesis and
Unemployment”, Quarterly Journal of Economics 105, 254-283.

Akerlof, G.A., Dickens, W.T. and Perry, G.L. (1996). “The Macroeconomics of Low
Inflation”, Brookings Papers on Economic Activity, 1-59.

Barro, R. J. (1994). “The Aggregate-Supply/Aggregate-Demand Model”, Eastern

Economic Journal, Winter, 1-6.

Bewley, T. (1998). “Why Not Cut Pay”, European Economic Review 42, 459-490.

Bhaduri,A., Laski, K. and Riese, M. (1999). “Effective Demand versus Profit

Maximization in Aggregate Demand/Supply Analysis from a Dynamic Perspective”,
Banca Nazionale del Lavoro Quaterly Review, 52 (210), 281-293.

Blanchard, O. (2000) “What Do We Know about Macroeconomics that Fisher and

Wicksell Did Not?” Quarterly Journal of Economics, 115, 1375-1409.

Boland, L. A. (1981). “On the Futility of Criticizing the Neoclassical Maximization

Hypothesis”, American Economic Review, 71(5), December, 1031-36.

Camerer, C.F., Loewenstein, G. and Rabin, M. (2004). Advances in Behavioral

Economics, Princeton: Princeton University Press.

Chiarella, C. and Flaschel, P. (2000) The Dynamics of Keynesian Monetary Growth.

Cambridge: Cambridge University Press.

Clower, R. W. (1989). “Keynes’s General Theory: The Marshallian connection”, in D. A

Walker, ed., Perspectives on the History of Economic Thought, Vol II, Aldershot: Edward
Elgar.

Colander, D. (1995). “The Stories We Tell: A Reconsideration of AS/AD Analysis”,

Journal of Economic Perspectives, Summer, 169-188.

Cross, R. ed. (1988). Hysteresis and the Natural Rate Hypothesis, Oxford: Basil
Blackwell.

Cross, R., ed. (1995). The Natural Rate of Unemployment. Reflections on 25 Years of the
Hypothesis, Cambridge: Cambridge University Press.

28
Crotty, J. (1994). “Are Keynesian Uncertainty and Macrotheory Compatible?
Conventional Decision Making, Institutional Structures, and Conditional Stability in
Keynesian Macromodels”. In G. Dymski and R. Pollin (eds), New Perspectives in
Monetary Macroeconomics: Explorations in the Tradition of Hyman Minsky, Ann Arbor:
University of Michigan Press, 105-142.

Dunlop, J. G. (1938). “The Movement of Real and Money Wage Rates”, Economic
Journal, 413-34.

Dutt, A. K. (1984). “Stagnation, Income Distribution and Monopoly Power”, Cambridge

Journal of Economics, 8(1), March, 25-40.

Dutt, A. K. (1987). “Keynes with a Perfectly Competitive Goods Market”, Australian

Economic Paper, 26, 275-93.

Dutt, A. K. (1992). “Keynes, Market Forms, and Competition”, in Bill Gerrard and John
Hillard, eds., The Philosophy and Economics of J M Keynes, Aldershot: Edward Elgar

Dutt, A.K. (1992a). “Conflict Inflation, Distribution, Cyclical Accumulation and Crises”,
European Journal of Political Economy 8, 579-597.

Dutt, A. K. (2002). “Aggregate Demand and Aggregate Supply: A History”, History of

Political Economy, Summer, 34 (2), 321-63.

Dutt, A. K. (2005). “Aggregate Demand, Aggregate Supply and Economic Growth”,

unpublished, Department of Economics and Policy Studies, University of Notre Dame.

Dutt, A.K. and Amadeo, E. J. (1990). Keynes’s Third Alternative? The neo-Ricardian
Keynesians and the Post Keynesians, Aldershot: Edward Elgar.

Dutt, A. K. and Skott, P. (1996). “Keynesian theory and the Aggregate-

Supply/Aggregate-Demand framework: A Defense”, Eastern Economic Journal, Vol. 22,
No. 3, Summer, 313-31.

Fehr, E. and Gächter, S. (2000). “Fairness and Retaliation: The Economics of

Reciprocity”, Journal of Economic Perspectives 14, 159-181.

Fields, T. W. and Hart, W. R. (1990). “Some Pitfalls in the Conventional Treatment of

Aggregate Demand”, Southern Economic Journal, 676-84.

Flaschel, P., Franke, R. and Semmler, W. (1997). Dynamic Macroeconomics Instability,

Fluctuations, and Growth in Monetary Economies, Cambridge, MA: MIT Press.

Flaschel, P. and Skott, P. (2005). “Steindlian Models of Growth and Stagnation”,

Metroeconomica, forthcoming.

29
Frydman, R. and Phelps, E. S. eds. (1983). Individual Forecasting and Aggregate
Outcomes: ‘Rational Expectations’ Examined, Cambridge: Cambridge University Press.

Gordon, R.J. (1997). “The Time-Varying NAIRU and Its Implications for Economic
Policy”, Journal of Economic Perspectives 11, 11-32.

Hahn, F. and Solow, R. M. (1995). A Critical Essay on Modern Macroeconomic Theory,

Cambridge, Mass.: MIT Press.

Harrod, R. F. (1939). “An Essay in Dynamic Theory”, Economic Journal, 49, 14-33.

Hicks, J. (1974). The Crisis in Keynesian Economics. Oxford: Blackwell.

Hicks, J. (1980-81). “IS-LM: An Explanation”, Journal of Post Keynesian Economics, 3,

Winter, 139-54.

Kahneman, D. (ed.) (2000) Choices, Values and Frames. Cambridge: Cambridge

University Press.

Kahneman, D., Knetsch, J.L. and Thaler, R. (1986). “Fairness as a Constraint on Profit
Seeking: Entitlements in the Market”, American Economic Review 76, 728-741.

Kaldor, N. (1957) “A Model of Economic Growth”, Economic Journal 67, 591-624.

Kaldor, N. (1966). Causes of the Slow Rate of Economic Growth in the United Kingdom,
Cambridge: Cambridge University Press.

Kalecki, M. (1971). Selected Essays on the Dynamics of the Capitalist Economy.

Cambridge: Cambridge University Press.

Katzner, D.W. (2004). “The Current Non-Status of General Equilibrium Theory”, UMass
Working Paper 2004-10

Keynes, J.M. (1930). A Treatise on Money, London and Basingstoke: Macmillan.

Keynes, J.M. (1930a). “The Question of High Wages”, The Political Quarterly.
Reprinted in Collected Writings, vol. 20: Activities 1929-1931, London and Basingstoke:
Macmillan, 1981.

Keynes, J. M. (1936 / 1973). The General Theory of Employment, Interest and Money.
London: Macmillan.

Keynes, J.M. (1973). Collected Writings, vol. 14: The General Theory and After - Part II
Defence and Development, London and Basingstoke: Macmillan.

30
Kirman, A.P. (1989). “The Intrinsic Limits of Modern Economic Theory: The Emperor
Has No Clothes,” Economic Journal 99, 126-139.

Kirman, A.P. (1992). “Whom or What Does the Representative Individual Represent?”,
Journal of Economic Perspectives 6, 117-136.

Krugman, P.R. (1998). “It’s Baaack: Japan’s Slump and the Return of the Liquidity
Trap”. Brookings Papers on Economic Activity 2:1998, 137-187.

Leijonhufvud A. (1968). On Keynesian Economics and the Economics of Keynes, Oxford

University Press.

Marglin S. A. (1984). Growth, Distribution and Prices, Cambridge, Mass.: Harvard

University Press.

Marshall, A. (1887). “A Fair Rate of Wages”. In A.C. Pigou (ed.), 1956, Memorials of
Alfred Marshall, New York: Kelley & Millman, 212-226.

Meltzer, A. H. (1988). Keynes’s Monetary Theory. A Different Interpretation,

Cambridge: Cambridge University Press.

Moore, B. J. (1988). Horizontalists and Verticalists, Cambridge: Cambridge University

Press.

Robinson, J. (1962). Essays in the Theory of Economic Growth, London: Macmillan.

Romer, D. (2000). “Keynesian Macroeconomics without the LM Curve”, Journal of

Economic Perspectives, 14(2), Spring, 149-69.

Romer, P. M. (1986).“Increasing Returns and Long-Run Growth”, Journal of Political

Economy, 94, 1102-37.

Sargent, T. J. (1993). Bounded Rationality in Macroeconomics, Oxford: Oxford

University Press.

Shafir, E., Diamond, P. and Tversky, A. (1997). “Money Illusion”, Quarterly Journal of
Economics 112, 341-374.

Skott, P. (1983). “An Essay on Keynes and General Equilibrium Theory”, Thames
Papers in Political Economy, Summer.

Skott, P. (1989). Conflict and Effective Demand in Economic Growth. Cambridge

University Press.

31
Skott, P. (1989a). “Effective Demand, Class Struggle and Cyclical Growth”,
International Economic Review 30, 231-247.

Skott, P. (1999). “Wage Formation and the (Non-) Existence of the NAIRU”, Economic
Issues 4, 77-92.

Skott, P. (2005). “Fairness as a Source of Hysteresis in Employment and Relative

Wages”, Journal of Economic Behavior and Organization 57, 305-331.

Skott, P. (2005a). “Equilibrium, Stability and Economic Growth”. In B. Gibson (ed.)

Joan Robinson: A Centennial Celebration, Northampton, MA: Edward Elgar, 175-196.

Solow, R. M. (1956). “A Contribution to the Theory of Economic Growth”, Quarterly

Journal of Economics, 70, 65-94.

Solow, R. (1990). The Labor Market as a Social Institution, Oxford: Blackwell.

Tarshis, L. (1939). Changes in Real and Money Wages. Economic Journal, 150-154.

Taylor, L. (1991). Distribution, Growth andIinflation. Lectures in Structuralist

Macroeconomics, Cambridge, Mass.: MIT Press.

Taylor, L. (2004). Reconstructing Macroeconomics, Cambridge, Mass.: Harvard

University Press.

Tobin, J. (1975) “Keynesian Models of Recession and Depression”. American Economic

Review, 65, 195-202.

Woodford, M. (1999) “Revolution and Evolution in Twentieth-Century

Macroeconomics”. Unpublished, Princeton University.

Woodford, M. (2003). Interest and Prices: Foundations of a Theory of Monetary Policy.

Princeton and Oxford: Princeton University Press.

32
 NOTES
1
According to the preface to the French edition of the GT, written three years after the English publication,
Keynes (1936/1973, p. xxxii) explains:

I have called my theory a general theory. I mean by this that I am chiefly concerned with the
behavior of the economic system as a whole, - with aggregate incomes, aggregate profits,
aggregate output, aggregate employment, aggregate investment, aggregate saving rather than with
the incomes, profits, output, employment, investment and saving of particular industries, firms and
individuals. And I argue that important mistakes have been made through extending to the system
as a whole conclusions which have been arrived at in respect of a part of it taken in isolation.
2
The dynamics can be explicitly formalized by the equations

dP/dt = βG [C+I+G-Y]

dr/dt = βA [L-(M/P)],

where t denotes time and 2i > 0 are speed of adjustment parameters for the goods and asset markets.
3
If we introduce real balance effects which make C (and, possibly, I) depend positively on M/P, the IS
curve would be negatively sloped rather than vertical. We abstract from this complication here, but refer to
it later.
4
The stability of short run equilibrium can be verified by representing the dynamics of expected price by
the equation

dPe/dt =βE [P – Pe ]

where 2E >0 is the speed of expectations adjustment parameter.

5
For a discussion of the history of the AD-AS model, including that of its emergence and spread in
macroeconomic textbooks, see Dutt (2002).
6
See Dutt and Skott (1996) for further discussion of the internal-consistency criticisms.
7
He probably used this type of diagram, rather than that in (P, Y) space, because aggregate price level and
real output were not in common use in his day, while value of output and total employment, involving
fewer aggregation problems, were.
8
Keynes’s Aggregate Demand function does not actually use the simultaneous equations approach to
solving P, focusing only on goods market equilibrium without taking into account asset markets explicitly.
An alternative formulation of the model, which focuses only on the goods market, but allows consumption
demand to respond to price changes due to either the real balance effect or distribution shifts, can easily be
developed. See Dutt (1987) for a version in which changes in price affect the value of output through
changes in income distribution between wages and profits.
9
The Treatise on Money had concentrated on the Marshallian ultra-short run (or market run) equilibrium:

My so-called `fundamental equations' were an instantaneous picture taken on the assumption of a

given output. They attempted to show how, assuming the given output, forces could develop
which involved a profit-disequilibrium, and thus required a change in the level of output. But the
dynamic development, as distinct from the instantaneous picture, was left incomplete and
extremely confused (Keynes, 1936/1973, p. xxii).

33
Skott (1989, 1989a) develops a model of cyclical growth using the Marshallian (or Keynes-of-the-Treatise)
ultra-short run equilibrium as the basic building block; see also Skott (1983) for a discussion of this
Marshallian approach and the relation between the Treatise on Money and the GT.
10
For a review of post Keynesian contributions see Dutt and Amadeo. For more mainstream discussions,
see Hicks (1980-81), Tobin (1975) and Meltzer (1988).
11
See, for instance, Moore (1988). New Keynesians have also abandoned the exogenous-money
assumption, but rather than stressing the nature of monetary institutions, they focus on the specific policy
rule adopted by the Central Bank in the US and elsewhere (e.g. Romer, 2000, Woodford 2003)).
12
In the context of our simple specification, however, it is easy to prove that local stability carries over to
the case where P, Pe and r are all treated as state variables, with their dynamics shown by the equations in
notes 2 and 4.
13
In a set of lecture notes from 1937, Keynes argues as follows:

When one is dealing with aggregates, aggregate effective demand at time A has no
corresponding aggregate income at time B. All one can compare is the expected and actual income
resulting to an entrepreneur from a particular decision. Actual investment may differ through
unintended stock changes, price changes, alteration of decision. The difference, if any, is due to a
mistake in the short-period expectation and the importance of the difference lies in the fact that
this difference will be one of the relevant factors in determining subsequent effective demand.
I began, as I have said, by regarding this difference as important. But eventually I felt it
to be of secondary importance, emphasis on it obscuring the real argument. For the theory of
effective demand is substantially the same if we assume that short-period expectations are always
fulfilled. (Keynes 1973, p. 181)
14
Hicks (1974) used the term Keynes’s ‘wage theorem’ to denote the benchmark result that variations in
money wages have no net effects on real output and employment in a closed economy'.
15
Some contributions are adventurous enough to depart from optimization to invoke ‘near’ rationality!
See Akerlof and Yellen (1987).
16
It can be argued that problems related to information gathering and computational ability need not
undermine the neoclassical optimizing hypothesis, because this hypothesis does not assume rationality in
an empirical sense (whatever that means), but simply uses the organizing framework of analyzing behavior
in terms of the optimization some objective function subject to some constraints (see Boland, 1981). This
argument, however, suggests that there is no overriding justification for insisting on the use of the
optimizing approach (for instance, based on some notion of the rationality of economic agents), and that a
non-optimizing approach need not be inferior to the neoclassical one.
17
The Japanese stagnation in the 1990s may have alerted the profession to some stability issues, and the
`liquidity trap' has made a comeback (e.g. Krugman 1998). The liquidity trap arises because of an inability
of monetary policy to reduce interest rates, that is, to change intertemporal prices. It seems to have escaped
attention, however, that the liquidity trap and the problem of intertemporal prices are indicative of the
general stability problem. Money wage reductions fail to solve the unemployment problem because
"[a]ccording to Keynes' diagnosis, it is fundamentally the intertemporal relative values observed or implicit
in the actual vector that are ‘wrong’", and, "although the most eye-catching symptom of maladjustment is
the great excess supply in the labor markets, ... the burden of adjustment should not be thrown on this
market.” (Leijonhufvud, 1968, p. 338 and 336; italics in original)
18
The learning argument is particularly vulnerable with respect to some of the key variables of
macroeconomic interest - saving for retirement, for instance, or educational choices (investment in human

34
capital) - where essentially each agent makes only a single decision. Parental background and experience
may be a critical influence on these choices but for backward rather than forward looking reasons.
19
Joan Robinson's criticism of tatonnement-based stability should have provided additional impetus for
this shift, but her criticism was not widely understood (e.g. Robinson (1962, pp. 23-29), Skott (2005a)).
20
The conventional aspect of fairness is implicit in many discussions of these issues. Keynes (1930a), for
instance, expressed his sympathy with the view that "there is a large arbitrary element in the relative rates
of remuneration, and the factors of production get what they do, not because in any strict sense they
precisely earn it, but because past events have led to these rates being customary and usual" (quoted from
Keynes 1981, p. 7). Marshall (1887) noted that fairness must be defined "with reference to the methods of
industry, the habits of life and the character of the people" (p.212). Fairness, he argues, requires that a
worker
ought to be paid for his work at the usual rate for his trade and neighbourhood; so that he may live
in that way to which he and his neighbours in his rank of life have been accustomed. (p. 213;
italics added)
Similar views have been advocated by Hicks (1974) and Solow (1990).
21
Here we use the term hysteresis in a broad sense to include zero-root models, and not just models with
‘remanence’ (see Cross, 1988).
22
Other theoretical and arguments against the natural rate hypothesis are discussed in, for instance, Cross
(1988, 1995).
23
This channel may be reinforced by the effects of unemployment on changes in the labor force
participation rate; women’s participation rate and the average retirement age, for instance, may respond
gradually to labor market conditions.
24
Verdoorn’s-law effects in which learning by doing generates a positive impact of the rate of growth of
output on productivity growth imply an additional stimulus from faster immigration to productivity growth.

35