44
Aggregate Demand-Aggregate Supply Analysis: A History Dutt, Amitava Krishna. History of Political Economy, Volume 34, Number 2, Summer 2002, pp. 321-363 (Article) Published by Duke University Press For additional information about this article Access Provided by Bangladesh University of Professionals at 05/03/11 6:49AM GMT http://muse.jhu.edu/journals/hpe/summary/v034/34.2dutt.html

Aggregate Supply Aggregate Demand Analysis a History

Embed Size (px)

Citation preview

Page 1: Aggregate Supply Aggregate Demand Analysis a History

Aggregate Demand-Aggregate Supply Analysis: A History

Dutt, Amitava Krishna.

History of Political Economy, Volume 34, Number 2, Summer2002, pp. 321-363 (Article)

Published by Duke University Press

For additional information about this article

Access Provided by Bangladesh University of Professionals at 05/03/11 6:49AM GMT

http://muse.jhu.edu/journals/hpe/summary/v034/34.2dutt.html

Page 2: Aggregate Supply Aggregate Demand Analysis a History

Aggregate Demand–Aggregate SupplyAnalysis: A History

Amitava Krishna Dutt

Aggregate demand–aggregate supply (AD-AS) analysis—which depictsthe economy using an aggregate demand curve and an aggregate sup-ply curve in a diagram with the price level and real output on the verti-cal and horizontal axes, and determines those variables at the intersec-tion of those curves1— has a curious status in economics. On the onehand, it has emerged in recent years as the preferred method of teachingmacroeconomics at the undergraduate level. It appears, and usually playsa central role, in almost all principles and intermediate macroeconomicstexts.2 On the other hand, it is severely criticized by economists in theirscholarly books and journals for being internally inconsistent, for failing

Correspondence may be addressed to Amitava Dutt, Department of Economics, University ofNotre Dame, Notre Dame, IN 46556. An earlier version of this paper was presented at thesession on aggregate demand–aggregate supply models at the History of Economics Societymeetings, Montreal, June 1998. I am grateful to Warren Young and to other participants inthe session, to Sandy Darity, and to two anonymous referees of this journal for their useful andinsightful comments and suggestions, as well as to Dennis Markov for able research assistance.

1. A different model, sometimes called the “dynamic” AD-AS model, which differs fromthe standard AD-AS framework by measuring the rate of inflation instead of the price level onthe vertical axis, is to be found in some textbooks—see, for instance, Dornbusch and Fischer1978, 402–29.Although this model is not examined in this paper, to the extent that this dynamicAD-ASmodel shares some features with the standardAD-AS framework, some of what followsapplies to it as well.

2. Principles texts are defined here as those that assume no prior study of economics on thepart of the reader, generally keep algebra to a minimum, and use simple graphs. Intermediatetexts usually assume that the reader has already been introduced to economic principles andmake use of algebra and more complicated graphical analysis.

History of Political Economy 34:2 © 2002 by Duke University Press.

Page 3: Aggregate Supply Aggregate Demand Analysis a History

322 History of Political Economy 34:2 (2002)

to adequately depict the operation of real economies (and perverting thetrue message of the Keynesian revolution), and for oversimplifying andleading to a conflation of macroeconomic and microeconomic concepts.In fact, a growing chorus of critics recommends jettisoning the frame-work as a teaching tool (see Barro 1994, Barens 1997, and McCormickand Rives 1998 for recent indictments). The resolution of this puzzle isan important order of business, since it has an obvious bearing not onlyon howmacroeconomics should be taught, but also on how the aggregatebehavior of the economy should be understood.

This article attempts to contribute toward this goal by examining theearly appearance of AD-AS analysis and its subsequent spread to in-termediate and principles textbooks. More specifically, its purpose is todocument the birth, diffusion, and growth in popularity of the analysis,and to trace when and how its alleged problems emerged. Before we turnto that, however, it is useful to review briefly the nature and criticisms ofAD-AS analysis.

AD-AS Analysis and Its Criticisms

Although theAD-AS framework appears in most principles and interme-diate economics texts, it does not do so in the same form everywhere.3

Here I present what is arguably the most popular version, which I call thebasic model, confining discussion of other versions to brief comments.4

I rely more on intermediate texts for an explicit derivation, since prin-ciples texts do not normally derive the AD and AS curves in detail. Theequations of the basic model are shown in table 1.

The AD curve shows the equilibrium levels of real output and pricelevel that clear goods and asset markets. It is derived from equations (1)through (5) by finding the level of output for different levels of the price.As the price level increases, the LM curve (derived from equations (4)and (5)) shifts upward in r-Y space, intersecting the IS curve (derived

3. I will use the term framework or analysis to refer to the set of all models that useAS andAD curves to determine Y and P , and AD-AS model to refer to a specific model that uses theAD-AS framework.

4. I will not cite individual works that present the more popular versions, referring to onlythose containing less common features. Also, throughout this essay I confine my attention tothe case of the closed economy. Models for the open economy contain, in addition to the ADand AS curves, a BP curve showing the relationship between price and output consistent withbalance-of-payments equilibrium. Our discussion clearly has relevance for many elements ofthe open-economy models.

Page 4: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 323

Table 1 Equations of the Basic Model

EquationsY = C + I + G (1)C = C(Y ), 0 < C′ < 1 (2)I = I (r), I ′ < 0, (3)L = L(Y, r), L1 > 0, L2 < 0 (4)M/P = L (5)Y = F(N), F ′ > 0 (6)W/P = F ′(N) (7)W = W0 (8)

SymbolsY = real output or incomeC = real consumptionI = real investmentr = rate of interestG = real government expenditure, exogenously givenL = demand for money in real termsM = nominal stock of money supply, exogenously givenP = price levelN = labor employedW = money wage, with W0 its exogenously given level

from equations (1) through (3)) at a lower level of output (and a higherinterest rate), implying a downward-sloping AD curve. A higher pricelevel reduces the real money supply with an exogenously given nominalsupply of money, increases the rate of interest, curtails investment de-mand, and, through the multiplier effect, reduces aggregate demand andreal output. This effect has come to be known as the Keynes effect. Al-though not all texts are explicit about it, those that discuss the dynamicsassume that departures from theAD curve cause firms to adjust output atgiven prices in response to excess demand and supply in the goods mar-ket (the money market always being assumed to clear). Some presenta-tions also include the real balance effect: a rise in the price level reducesthe real value of monetary assets, reduces consumption (and possibly in-vestment), and hence reduces aggregate demand. This can be representedby replacing equation (2) with

C = C(Y, M/P ), 0 < C1 < 1, C2 > 0. (2a)

Page 5: Aggregate Supply Aggregate Demand Analysis a History

324 History of Political Economy 34:2 (2002)

This modification makes the IS curve shift to the left when the price levelrises, so that the AD curve remains negatively sloped.

Although the AS curve is less standardized, one that is widely usedassumes profit-maximizing behavior by firms operating under perfectcompetition. Given the money wage, the stock of capital, and the pro-duction function (embodying given technologies), which reflects dimin-ishing returns to labor, a given price level is associated with a profit-maximizing level of output. The AS curve is obtained, given the moneywage, by finding the level of output firms produce at each price level: ahigher price level implies a higher level of employment and output givendiminishing returns to labor, so that the AS curve is upward-rising. For-mally, the curve is derived from equations (6) through (8). A variant,making the money wage a rising function of the level of employment,5

obtained by replacing equation (8) with

W = W(N), W ′ > 0, (8a)

also yields the upward-risingAS curve.Yet another version departs fromthe assumption of diminishing returns to labor with the assumption of aconstant labor-output ratio, a, so that

Y = N/a, (6a)

and from perfect competition by assuming that imperfectly competitivefirms practice markup pricing, so that

P = (1 + z)Wa, (7a)

where z is the exogenously given markup.6 If W is fixed in this formula-tion, so that (8) holds, the price level is fixed by the pricing equation (7a),and theAS curve is horizontal.7 If, instead, we assume that equation (8a)holds, a higher Y implies a higher N from equation (6a), which implies

5. This relation between the level of the money wage and the level of employment is some-times interpreted as an empirical relation (based on the Phillips curve) that shows how the levelof the money wage depends on the condition of the labor market and not a standard supplycurve of labor, which shows the quantity of labor supplied at each real wage (see Dornbuschand Fischer 1978). Some authors, such as Richard Froyen (1993, 225–27), interpret the curveas a supply curve, taking price expectation to be given and making labor supply a function ofthe money wage.

6. See Dornbusch and Fischer 1994 and Blanchard 1997.7. Horizontal (short-run) AS curves are used by Andrew Abel and Ben Bernanke (1992)

and N. Gregory Mankiw (1992), among others.

Page 6: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 325

P

AS

B C

E F

A

YY

P1

D

f

Figure 1 The AD-AS model

a higher W from equation (8a), which implies a higher P as shown inequation (7a), so that the AS curve is upward-rising.

As shown in figure 1, the intersection of the AD and AS curves (thelatter is drawn to be upward-rising) determines the equilibrium levelsof Y and P at point E. Two implications immediately follow. First, theequilibrium level of output may well be less than the full employmentlevel, shown in the figure by Yf (determined by labor demand and sup-ply curves and the production function; the labor supply curve has notbeen included in the model). Second, if the money wage falls when thereis unemployed labor,8 if the economy has unemployed labor the AS

curve will shift downward and to the right. As long as the AD curveis downward-sloping, the equilibrium level of output will increase untilthe economy attains the full employment level, as shown in figure 2.9

Thus, in the “short run” (when the money wage is rigid) we can haveunemployment, but in the “long run” (when money wage adjustmentsare completed) we have full employment.10

8. Or, in the case in which equation (8) is replaced by (8a), the wage function can be as-sumed to fall when there is unemployed labor.

9. Many recent presentations call the vertical Yf the long-runAS curve, reserving the termshort-run AS curve for the curve I have called the AS curve. The short-run AS curve is some-times assumed to be horizontal at low levels of output, then upward-rising, and finally verticalat the full employment level. Other presentations extend the short-run AS curve beyond Yf , asdone in the text, to take into account phenomena such as overtime work. These variations donot change the essence of the story.

10. I assume, as is done in most standard AD-AS models, that the level of wealth heldby wealth-holders (as well as the stock of capital) is held constant during this entire process.

Page 7: Aggregate Supply Aggregate Demand Analysis a History

326 History of Political Economy 34:2 (2002)

P

A

A

A'

Y

S

S'

L

Yf

Figure 2 Adjustment to full employment in the AD-AS model

Turning next to the criticisms of AD-AS analysis, A. A. Rabin andD. Birch (1982), T. Windsor Fields and William Hart (1990), RobertBarro (1994), David Colander (1995), and Amit Bhaduri, K. Laski, andM. Riese (1995) argue that the AD-AS model is internally inconsistentbecause it contains two very different theories of production and pric-ing. In their view, most treatments of the AD curve, which are derivedfrom the standard textbook IS-LM model, assume that firms produce tomeet demand by making quantity adjustments while keeping the pricefixed.11 In figure 1, with the price level at P1, this theory implies thatfirms produce at B. The AS curve, when it is derived in the standardway, assumes price-taking behavior by firms, which maximize profits.In the figure, with price P1, this theory implies that firms produce at C.Thus, for a given price level, theAD-AS model determines two differentlevels of output consistent with two different theories of production andpricing, making the model inconsistent.

It should be noted that this criticism does not apply to the version ofthe model with imperfect competition, since in that model price-taking

Possible shifts in the LM curve (and hence in the AD curve) due to changes in wealth areignored in this model.

11. This criticism is certainly not true of all versions of the IS-LM.William Darity andWar-renYoung (1995) examine many versions to show that there is a huge range of variations in themodels and, in particular, that not all of them take the price level to be given. It does, however,apply to those presentations—which, as we shall see below, are numerous—that explicitly takethe price level to be exogenously given and assume that firms adjust output to meet demand,as in fixed-price income-expenditure models and IS-LM models derived explicitly from them.

Page 8: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 327

behavior and profit maximization are not assumed, so that the secondtheory of pricing and production mentioned above is not invoked at all.Nor does it necessarily apply to the model with perfectly competitivepricing, since that model can be interpreted as follows. Given a pricelevel expected by firms, say P1 (figure 1), profit-maximizing, price-tak-ing firms produce at C, on the AS curve. Given this level of output, theprice level varies to clear the goods market, which means that the econ-omy will be at F , on the AD curve, where the goods and asset marketsclear. Since the price level at F is different from the expected price P1,firms will revise their expected price adaptively. This process of revisionand adjustment will continue until the economy is at E, where price ex-pectations are fulfilled and goods and asset markets clear.12 One can alsoassume rational expectations instead of adaptive expectations; the econ-omy then instantly arrives at the equilibriumE. This interpretation of themodel is not inconsistent since it does not use the fixed-price quantity-adjustment assumption. Thus, strictly speaking, only those versions thatexplicitly assume the fixed-price quantity-adjustment assumption in de-riving the AD curve can be faulted with inconsistency.13 However, sincepresentations that use the standard IS-LM model to derive the AD curvedo assume that the price level is given in deriving their AD curve (with-out clarifying what variable adjusts to clear the market), theymay also besaid to be possibly inconsistent. A consistent presentation would need toderive the AD curve by drawing IS-LM curves on r-P space for a givenY , and then finding different levels of P for different levels of Y . Ofcourse, the AD curve would formally be the same as the curve derivedwith P as the parameter.

A second criticism of AD-AS analysis is that it fails to portray boththe behavior of real economies and Keynes’s theory of output and em-ployment adequately. Although this criticism is made in many forms(for instance, the analysis does not adequately capture the roles of un-certainty, history, money, and financial institutions), perhaps the clearestway of expressing it is to note that the model implies that wage flexibilitynecessarily implies full employment, and that short-run unemploymentis due to wage rigidity (see Dutt 1986–87, Cottrell 1995, andMcCormickand Rives 1998). This property follows from the fact that the AD curveis negatively sloped and that it does not shift downward when the wage

12. See Dutt 1997 for this interpretation of the AD-AS model. The interpretation is consis-tent with the Marshallian interpretation of Keynes’s theory; see Clower 1989 and Dutt 1992.

13. Some individual contributions that are arguably guilty of error will be discussed below.

Page 9: Aggregate Supply Aggregate Demand Analysis a History

328 History of Political Economy 34:2 (2002)

falls. In actual economies, however, greater wage flexibility may in factincrease uncertainty and reduce aggregate demand, thereby exacerbat-ing the problem of unemployment, as Keynes (1936) explicitly notes.Moreover, as I shall discuss below, Keynes suggests that for a number ofreasons wage reductions may fail to stimulate, and instead actually de-press, aggregate demand. The role of uncertainty, money, and financialinstitutions in preventing wage flexibility and wage reductions in bring-ing about full employment has also been stressed by post-Keynesians(see Dutt and Amadeo 1990b), and similar ideas have been expressedby mainstream Keynesians such as James Tobin (1993) and Frank Hahnand Robert Solow (1995).

A third criticism is that in many presentations, especially in principlestexts, the AD-AS analysis is oversimplified and leads to a confusion ofmacro- and microeconomic concepts. David Colander and Peter Sephton(1998) refer to this problem as “dirty pedagogy.” Since principles textsdo not explicitly derive theAD curve from the IS-LM diagram and makethe goods and asset market interactions behind its inverse slope explicit,they often give the mistaken impression that the AD curve is similar tothe micro demand schedule, which shows quantity demanded at eachprice because of the substitution between goods and income effects ofprice changes (see Hansen, McCormick, and Rives 1985). These con-siderations should have no relevance for theAD curve, which deals withall goods and examines the effects of price changes given real income,and which shows equilibrium conditions in goods and asset markets.

Early History of AD-AS Analysis

Although Keynes 1936 did not develop or use the AD-AS frameworkdiscussed above, it is natural to start our narrative with this work, sincethe terms aggregate demand and aggregate supply functions appear in itand because the early macroeconomic models, including theAD-AS andothers related to it, were developed explicitly to depict Keynes’s analysis.In chapter 2 of The General Theory Keynes defines the aggregate supplyfunction as the relationship between the total value or aggregate supplyprice of the output, Z, which is “the expectation of proceeds which willjust make it worth the while of the entrepreneurs to give [a particularlevel of ] employment” and the level of employment, N , given technol-ogy, capital stock, degree of competition, and the money wage (24–25).Z rises with the level of N , since more output is produced with more

Page 10: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 329

employment and since higher N implies a higher price level, given di-minishing returns and the given value of the money wage.14 The aggre-gate demand function is the relationship between D, “proceeds whichentrepreneurs expect to receive from the employment of N men,” andN , and shows the sum of consumption and investment demand. Giventhe value of investment demand, which depends on the inducement toinvest (which in turn depends on the marginal efficiency of capital andon interest rates), the aggregate demand curve is upward-rising becausea rise in employment increases income and hence consumption. If D ex-ceeds (is less than) Z, N rises (falls), so that the level of employmentis determined by the intersection of the aggregate supply and aggregatedemand curves.

Although Keynes did not use the AD-AS framework (in the sensethat we are using it here, that is, with curves drawn in P -Y space), itis worth noting that Keynes’s analysis does not suffer from the three al-leged problems of the AD-AS framework discussed earlier. On the first,Keynes’s analysis assumes competitive (price-taking) profit-maximizingbehavior and a variable price level; hence there is no inconsistent useof two theories of pricing and production. On the second, although hedoes assume that the money wage is fixed, this assumption is made onlyin the first eighteen chapters of the book. In chapter 19 Keynes pointsout that a reduction in the money wage may quite possibly depress em-ployment. Arguing that the effect of a wage reduction in reducing theinterest rate, and thereby stimulating investment and employment (theKeynes effect), is likely to be weak, Keynes suggests that when wagesfall, aggregate demand can be reduced because (1) income redistributionfrom workers and from entrepreneurs to rentiers reduces the propensityto consume; (2) expectations of further reductions in wages reduce in-vestment; (3) workers may become discontented, which may reverse theoptimism found among entrepreneurs due to money wage falls; and (4)falling prices increase the debt burden of entrepreneurs and indeed leadsome of them to insolvency with a consequent adverse effect on invest-ment (266). Keynes also adds that a flexible-wage economy is not desir-able, since it results in price instability, which increases uncertainty inbusiness decision-making (268). Thus wage flexibility does not result infull employment. On the third, since Keynes does not visualize curves

14. This must be pieced together from chapter 2 and elsewhere; Keynes does not explainthe upward slope of the aggregate supply curve.

Page 11: Aggregate Supply Aggregate Demand Analysis a History

330 History of Political Economy 34:2 (2002)

on P -Y space, there is no of conflation of microeconomic and macroe-conomic concepts.

Since Keynes discussed the curves in Z-N—and not P -Y—space, itis not surprising that most of those who incorporated Keynes’s analysisdiagrammatically into textbooks in the early days, such as Dudley Dil-lard (1948), Sidney Weintraub (1951), Alvin Hansen (1953), and A. W.Stonier and D. C. Hague (1953), depicted the Keynesian system in thatspace. However, their diagrams do not play a major role in any of theirtexts, and their usually unclear discussion adds little to Keynes (see King1994). Indeed, several of these texts quickly switch to other geometrictreatments of the Keynesian system, as we shall see below.As J. E. King(1994, 14) notes, the subsequent controversy regarding this version ofthe aggregate supply curve in the mid-fifties in the Economic Journaldid little to reduce the confusion surrounding the diagram. In subsequentyears the diagram caught the fancy of American post-Keynesians likeWeintraub (1958), Paul Davidson (1972), and Paul Wells (1977), whoprovided clearer expositions of the model. However, this analysis seemsto have now disappeared from textbooks and even frommost of the workof the post-Keynesian school, with a few exceptions (see King 1994, 27–28).15

Of the geometric depictions of the Keynesian system popular today,the first two to appear were the income-expenditure or diagonal crossmodel, and what is now called the IS-LM diagram. The income-expen-diture model had its debut in Paul Samuelson’s 1939 paper on themultiplier-accelerator interaction theory of the business cycle; in that pa-per Samuelson draws the consumption function in expenditure-incomespace and determines equilibrium in the goods market at the point of in-tersection of the C + I line (with investment given autonomously) withthe 45◦ line.16 Although it is confined to a mere two pages in Samuel-son’s paper, in less than a decade this model became the backbone ofSamuelson’s 1948 principles text (see Pearce and Hoover 1995). The

15. The discussion of these models, which can be called the aggregate demand price and ag-gregate supply price models, would take us too far afield, since the relationship between thesemodels and the AD-AS framework discussed in this paper is unclear and shrouded in contro-versy. See, however, King 1994 for a history of these aggregate demand price and aggregatesupply price models.

16. Patinkin (1982) has argued that this diagrammatic apparatus not only represents theessence of Keynes’s theory of effective demand, but that it is implicit in The General Theoryand that it is verbally developed by Keynes himself in How To Pay for the War (1940). Be thatas it may, Samuelson’s paper predates Keynes 1940.

Page 12: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 331

diagram (measuring real or nominal output or income on one axis, andexpenditure on the other) also appears in Dillard 1948, Patinkin 1949,and Weintraub 1951. Because of its simplicity, this model with the axesmeasuring real income and expenditure subsequently became the popu-lar method of presenting the Keynesian theory of aggregate demand andunemployment in textbooks. It is usually presented at the beginning ofthe analysis of income determination but argued to be a first simplifica-tion with a given interest rate and price level;17 later chapters relax theseassumptions by bringing in asset market considerations and the effectsof wage and price changes.

The IS-LM model appeared in print even before the diagonal crossmodel in John Hicks’s 1937 paper, where it is called the SI-LL model.Although there are other contemporary versions of the model by JamesMeade, Roy Harrod, and others (see Young 1987 and Darity andYoung1995), Hicks’s model, which was the most influential (perhaps becauseit was presented geometrically), is worth examining in some detail be-cause of its close similarity with the AD-AS model presented in equa-tions (1) through (8), a fact that becomes clear by translating Hicks’stwo-sector (producing investment and consumption goods) schema intoa one-sector one (see figure 3).18 Following Hicks, we can derive theSI-LL in Z-r space given W as follows (where, as before, Z ≡ PY de-notes money income and output). Equations (1) through (3) give a now-familiar inverse relation between r and Y . Also, equations (6) through(8) imply that P rises with Y , given W . Since Y rises when r falls, andP rises when Y rises, we obtain an inverse relation between r and Z.This yields the SI curve, which combines the goods market equilibriumcondition and the supply-side relations. Next, starting with equilibriumin the money market shown by M = PL(Y, r), which combines equa-tions (4) and (5), an increase in Y implies a rise inL as well as a rise in P

from equations (6) and (7), which implies a rise in the nominal demandfor money, PL, and hence an excess demand for money. Since r has tobe increased to remove this and restore equilibrium, we get a positive re-lation between r and Z, producing equilibrium in the money market andsatisfying the supply-side equations (6) and (7); this yields theLL curve.The intersection of the two curves solves for the equilibrium values of Z

17. Since the price level is taken as given, the model is usually interpreted as one in whichfirms adjust output according to demand; since price-taking profit maximization is not as-sumed, there is no inconsistency.

18. See also Barens 1997 for a discussion of Hicks’s analysis in terms of one sector.

Page 13: Aggregate Supply Aggregate Demand Analysis a History

332 History of Political Economy 34:2 (2002)

L

I

r S

L

Z

Figure 3 Hicks’s SI-LL model

and r , which completes the SI-LL diagram. Once the equilibrium valueof Z is obtained, the equilibrium values of P and Y can be obtained us-ing the rectangular hyperbola Z = PY and the positive relation betweenP and Y obtained from equations (6) through (8).

Since the model is exactly the same as our AD-AS model presentedabove, despite their different geometric depictions, the properties ofthe models are identical.19 But what of the problems? Since Hicks’smodel is not fixed-price (but fixed-wage), and firms follow standardprofit-maximizing behavior as price takers, there is no inconsistencyabout pricing behavior. A fall in the exogenously given money wage,by implying a lower P for a given Y , shifts the SI curve to the left andthe LL curve to the right. The interest rate definitely falls and the effecton Z is uncertain, but equations (1) through (3) confirm that the lower r

must imply a higher equilibrium Y . Thus the model does imply that wageflexibility leads to full employment, which leaves it open to the secondcriticism. Finally, since Hicks does not draw curves in P -Y space, thereis no confusion of macro and micro concepts.

If Hicks’s version of the model is an AD-AS model in disguise (since

19. Apart from the two-sector formulation, there are some (for our purposes) inessentialdifferences with Hicks’s model due to the fact that in it (1) consumption and saving depend alsoon r , (2) government expenditure is omitted, (3) the demand-for-money equation is expressedin money, not real, terms, and (4) the money supply is at some times taken to be a positivefunction of the interest rate.

Page 14: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 333

it incorporates all of the equations of the AD-AS model discussed ear-lier, but is not present in P -Y space), by the time Hansen (1949, 1953)puts the IS-LM model into textbooks, it becomes the modern textbookversion in r-Y space without explicit consideration of the supply side ofthe economy, with an exogenous price level.20 Hansen (1949, 125) pointsout that a fall in the price will reduce the rate of interest by reducing thenominal demand for money, which will increase investment and real out-put, anticipating the AD curve. However, he argues that this method ofreaching full employment through wage reductions is “patently absurd,”and in fact he discusses several reasons why a wage reduction may notincrease employment, including income distributional shifts and an ab-sence of interest elasticity of investment. Hansen also points out that arise in aggregate demand and output will tend to increase the moneywage and the price level, thereby anticipating the AS side as well (135–36); however, he makes no explicit use of it.

We now turn to AD and AS curves drawn in P -Y space. We considerthese contributions in some detail—at the risk of making the narrativeread as if history was “just one damned thing after another”—to de-termine priority, to examine how the present versions emerged, and todetermine whether the early presentations suffered from the problemsdiscussed above.21

The earliest depiction of Keynesian macroeconomics in price-outputspace appears to be in Lorie Tarshis’s 1947 textbook. Tarshis derives thestandard supply curve of the price-taking, profit-maximizing firm fromits marginal cost curve, then aggregates to obtain the industry supplycurve, and finally generalizes this to the AS curve (Tarshis 1947, 448–50) drawn in P -Y space. Tarshis explains that this curve takes moneywage and technology as given, and that it is relatively flat at low levels ofoutput (since the strength of diminishing returns is weak), then upward-rising, and then vertical at full employment, due to the scarcity of la-bor. Tarshis does not have an explicit AD curve, but he determines theequilibrium level of output by equating it to the demand for goods andservices (a demand that depends mainly on consumption and investmentdemand and that is taken to be independent of the price level); the price

20. Another change was that models also mostly assumed that there is one sector in theeconomy, departing fromHicks’s analysis (as pointed out above), which considered two sectors.

21. In some cases in which it is obvious, I will not mention whether a model suffers fromeach of the three problems. See, however, table 2 for a summary of the major contributions andtheir problems.

Page 15: Aggregate Supply Aggregate Demand Analysis a History

334 History of Political Economy 34:2 (2002)

level is therefore determined by the AS curve and the equilibrium levelof output, with the latter implicitly giving the vertical AD curve (452).22

Although Tarshis does invoke the Keynes effect in analyzing the effectsof a rise in the money supply (453), he also argues that a fall in themoney wage may increase or reduce investment spending, dependingon how expectations and industrial relations are affected (although thereis a good chance that investment spending will increase if profitabilityis increased and the interest rate falls), and most likely will reduce thepropensity to consume due to a redistribution away from labor, so thatthe combined effect on aggregate demand and hence output is indeter-minate (555–64).

Both AD and AS curves in P -Y space appear first in Kenneth Bould-ing’s 1948 textbook and O. H. Brownlee’s 1948 contribution to a text-book on applied economics.23 Boulding does not use the curves to de-termine the price level and aggregate output, but rather to reveal thedangers of aggregative thinking, arguing that “no solution is possible”(262) with such aggregate curves. Brownlee, however, attempts to showhow such an analysis is possible.Assuming that the interest rate is given,that the demand for money depends on nominal income, and that themoney supply is exogenously given, he draws a downward-sloping curve(of the formMV =PY , where V is the given income velocity of money)in P -Y space to depict money market equilibrium. Assuming a produc-tion function with diminishing returns, profit maximization by firms,and a fixed money wage, Brownlee uses a four-quadrant diagram to de-rive a positive relation between real output and the price level, a relationthat gives his labor market schedule; with real output measured on thevertical axis, the curve becomes horizontal at full employment. The in-tersection of the money market equilibrium and labor market schedules(which we can call the AD and AS) determines equilibrium output andthe price level.

22. King (1994) interprets Tarshis as having a downward-sloping AD curve, its slope ex-plained by the fact that higher prices reduce demand, especially among lower income groups. Inthe pages on which King finds this explanation (475–77), Tarshis discusses the fall in consumerdemand, which makes possible a lower allocation of GNP to the private sector due to a rise inmilitary spending by the government. However, Tarshis does not incorporate this analysis intohis AS diagram.

23. The textbook on applied economics is an edited volume that is used in courses on eco-nomic principles and problems. It is intended for use in the second of a two-semester sequenceof courses; students are assumed to have had an introduction to economics and to economicanalysis already; it can thus be called an intermediate text.

Page 16: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 335

Turning to the problems, Brownlee’s model is not inconsistent, sinceit has only one theory of pricing, in which firms are profit-maximizingprice takers. On the effects of money wage changes, a fall in the moneywage shifts the AS curve to the left (with price measured on the hori-zontal axis) and leaves the AD curve unchanged, so the price level fallsand output and hence employment rise. Thus, the analysis is open to thesecond criticism. However, Brownlee (1948, 247) notes that a fall in themoney wage also reduces workers’ income and their demand for com-modities and the demand for money, so that the AD curve can also shiftto the left, so that it is not necessarily true that a fall in the wage increasesemployment.

Despite determining equilibrium using AD and AS curves, Brown-lee’s claim to presenting the first AD-AS model is compromised by thefact that his AD curve does not explicitly incorporate the goods mar-ket. As noted above, this procedure is made possible by assuming thatthe interest rate is fixed. Although Brownlee does not provide a reasonfor this assumption, it is problematic if it is fixed by the monetary au-thorities who adjust money supply, for then it is no longer permissibleto assume that money supply is given, as his presentation does. WhenBrownlee discusses the case with a variable interest rate, which affectsthe level of investment, he uses the 45◦ income-expenditure diagram toshow how nominal output is determined by goods market equilibrium,but this analysis is not incorporated into the AD-AS framework.

Within two years, however, in a paper published in the Journal of Po-litical Economy, Brownlee (1950) achieves the distinction of being thefirst to present a published version of a complete AD-AS in P -Y space.Without referring to any earlier work on the IS-LM model, Brownleeuses goods and money market schedules in Y -r space to determine equi-librium output for a given price level. Since a higher price level shiftsthe money market schedule upward, equilibrium output falls, so that thedownward-slopingAD curve—which Brownlee calls the ZZ′ curve—isobtained in a quadrant measuring Y on the vertical axis and P on thehorizontal. Turning next to the supply side, Brownlee notes that with aperfectly flexible money wage, there is a unique level of employment atfull employment, given capital and technology, and, correspondingly, aunique level of real output independent of the price level, which is shownby a horizontal YY ′ curve. If the money wage is fixed, however, the re-lation between real output and the price level that will fulfill “the con-dition of equilibrium in the labor market” (implying that employment is

Page 17: Aggregate Supply Aggregate Demand Analysis a History

336 History of Political Economy 34:2 (2002)

determined by labor demand, but without labor market clearing) is anupward-rising W line that becomes horizontal when it reaches the YY ′curve (as in his earlier presentation). The intersection of the ZZ′ and W

curves determines the levels of real income and the price. With a givenmoney wage, the equilibrium level of real income or output may well bebelow the full employment level, and Brownlee terms this the “Keyne-sian” solution.

Brownlee does not take the price level to be the exogenous variablein the sense of assuming quantity adjustment in the goods market for agiven price; rather, theAD curve is explicitly defined as showing combi-nations of P and Y showing goods and asset market equilibrium. Thusthere is no evidence of inconsistency in his presentation and no confu-sion of micro and macro concepts. If the money wage falls, the W curveshifts to the right, so that equilibrium output and employment rises. Butno positive wage that ensures full employment may exist—if the interestdoes not affect the aggregate demand for goods or if the interest rate can-not fall. If, however, a price reduction increases the real value of mon-etary assets, which in turn increases consumption, wage reductions willincrease the level of real output even in the cases in which the Keyneseffect does not work. Thus, the model’s logic does imply that wage flex-ibility leads to full employment, despite Brownlee’s cautionary remarks.

Closely on the heels of Brownlee, JacobMarschak’s (1951) publishedlecture notes provide the first full textbook treatment of the AD-ASmodel. In lecture 2 Marschak presents the same model as in Brownlee1948, although without citing him or anyone else. Subsequent lecturesdevelop the Keynesian income-expenditure analysis in which moneyoutput is determined by the consumption function (which makes thelevel of money consumption depend on the levels of money income andmoney supply) and the exogenously fixed level of money investment.Variations in both physical output and price can equilibrate demand tosupply. Extensions of this model make money investment depend posi-tively on money income and the interest rate, and money consumptiondepend on the interest rate in addition to the variables mentioned ear-lier. In the money market, nominal money demand is taken to depend onthe interest rate and the level of money income, and exogenous moneysupply is set equal to money demand. Money and goods market equi-librium is then presented in interest rate and money income space withcurves that look like IS and LM, although money income (as in Hicks’spresentation) is measured on the horizontal axis. In this discussion there

Page 18: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 337

is no fixed-price assumption, so there is no inconsistent use of two pricetheories. Lecture 18 moves to a space with price and real income on theaxes in which he derives a downward-sloping curve showing “the de-mand curve for all goods” (DD) drawn on a quadrant measuring Y andP on the vertical and horizontal axes. The curve embodies both the in-terest rate effects on investment and consumption and the real balanceeffect on consumption. The “supply curve for all goods” (SS) is drawnusing the labor demand condition, which sets the real wage equal to themarginal product of labor; the production function, which relates realoutput to employment; and a supply curve of labor, which makes themoney wage depend positively on the level of employment (reflectingwhat Marschak calls “money illusion”);24 demand and supply of laborare assumed to be equal in deriving SS. DD and SS are drawn on a di-agram measuring real income on the vertical axis and the price level onthe horizontal axis, and the two curves intersect to simultaneously deter-mine the two variables (pp. 56–57).25 If there is no money illusion (sothat the supply of labor demands on the real wage and the labor mar-ket clears), the SS curve is horizontal; this is the full employment case.Marschak also presents a case similar to Brownlee’s (1950) in which themoney wage is fixed (due to collective bargaining) and there is no labormarket clearing.

Two other early AD-AS models may be briefly mentioned. Wein-traub’s 1951 textbook uses demand and supply curves in a space in whichprice and real output are on the two axes to show equilibrium. The sup-ply curve is based on the marginal cost curves of firms as in the contribu-tions just discussed, but theAD curve (called the income-demand curve),drawn for a given level of investment, is downward-sloping without anyexplanation. Since Weintraub provides no explanation for the negativeslope of his macroeconomic AD curve, it is possible that he conflatesmacroeconomic and microeconomic considerations, but if so, it is theonlyAD curve from this period surveyed here that does so. Bent Hansen(1951), whose treatise aims to develop equations of motion for inflationrather than determine equilibrium output at the intersection of AD andAS curves, derives the curves on P/W -Y space. The AS curve is de-rived in the standard way, given the money wage, using the assumptionsof diminishing returns, profit maximization, and perfect competition.

24. The equation is the same as equation (8a) in the present text, interpreted as a supplycurve.

25. This is essentially our earlier AD-AS model but with equation (8) replaced by (8a).

Page 19: Aggregate Supply Aggregate Demand Analysis a History

338 History of Political Economy 34:2 (2002)

However, Hansen explicitly refers to expectations, stating that the sup-ply curve shows for any given price expectation the level of output thatwill be produced in the economy. The demand side has two features thatdistinguish it from the other contributions of the period. First, Hansen’scurve shows quantity of aggregate demand for goods, not equilibriumin goods and asset markets. Second, aggregate demand falls when P/W

rises because of a fall in consumption demand due to a shift in incomedistribution from workers to capitalists, who have a lower propensity toconsume, rather than any effect on real investment, which is assumed tobe given by a fixed level of the interest rate.

AD-AS Analysis in Intermediate Textbooks

After its initial appearance, theAD-AS analysis took a few years to makeits way into standard textbooks. The story of its appearance and spreadin intermediate texts is rather different from that in principles texts; Itherefore tell them separately.

The intermediate textbook by JosephMcKenna (1955)26 appears to bethe first standard textbook to contain the AD-AS model,27 if we do notinclude Brownlee 1948 and Marschak 1951 in that category. McKennafollows Marschak in starting with the diagonal cross model (taking theprice level and real investment to be given), moving next to the IS-LMmodel (with a given price level), and then deriving the downward-slop-ing AD curve by examining the effects of the change in price due to theKeynes effect. TheAS curve is derived both for the case of perfect com-petition (as in earlier presentations) and of monopoly with a constantprice elasticity of demand for goods and an informal aggregation proce-dure, with a given money wage when unemployed labor exists. At fullemployment, with perfect wage flexibility, the curve becomes vertical.McKenna refers to Hicks 1937 for the IS-LM framework and Brownlee1948 for the AS side of the AD-AS framework. Drawing the curves ina diagram with P and Y on the vertical and horizontal axes, McKenna

26. McKenna does not require students to have any prior knowledge of economics. Butsince it assumes familiarity with algebra, I classify it as an intermediate-level text.

27. Since—to the best of my knowledge—this is the first attempt to write the history ofthe AD-AS approach in price-output space, I have had to rely on my own review of textbooks.Although I have tried to cast my net as widely as I could, it is possible that some books haveescaped my attention. This explains my use of the phrase “appears to be” quite frequently inthe following narrative.

Page 20: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 339

explains that their intersection determines equilibrium P and Y , at whichunemployment may prevail.

McKenna nowhere states that the AD curve is derived from quantity-adjusting behavior by firms at a given price level. The curve merelyshows combinations of P and Y at which goods and assets markets arein equilibrium. Without the fixed-price interpretation for the AD curve,there is no necessary inconsistency between the AD and AS sides, al-though since he does not explicitly say that variations in the price levelclear the goods market there may be one. McKenna points out that ifthe money wage is flexible, the wage will fall, shifting AS downward,thereby reducing unemployment. He states that this has led some to ar-gue that Keynesian unemployment is possible if the wage is rigid, butargues that Keynes’s analysis was not based on wage rigidity since wageflexibility may not be effective in increasing employment. Indeed, hestates that the liquidity trap will make the AD curve vertical, preventingoutput increases when the wage falls; but he continues that the Pigou orreal balance effect will make it downward-sloping again. Thus althoughMcKenna repeats and apparently endorses Keynes’s views on wage flex-ibility, his AD-AS model with both Keynes and real balance effects sug-gests that the economy goes to full employment with flexible wages.

No other textbook seems to have used the framework until McKenna’ssecond edition appeared in 1965. In fact, up to the late 1970s only twoother textbooks—those by John Lindauer (1968) and D. C. Rowan(1968)—appear to have used it, the rest confining their attention tothe income-expenditure and IS-LM models. Lindauer mostly followsMcKenna, although he draws on Keynes to additionally discuss a num-ber of reasons why price and wage reductions need not take the economyto full employment. The significance of this book will become clear inthe next section. The significance of Rowan’s work lies in its attempt toreformulate the AD-AS model. Although his AS curve is the standardone with perfect competition and a given money wage, his AD curveis different from the earlier one, which, being derived from the IS-LMequilibrium output for different price levels, is a schedule showing goodsand assets market equilibrium rather than a real AD curve. Rowan in-stead takes a given price, finds the corresponding output supply fromthe AS curve, then finds the levels of consumption and investment de-manded at that output level and the corresponding interest rate for assetmarket equilibrium, and finally adds them together to arrive at aggre-gate demand. TheAD schedule traces out aggregate demand at different

Page 21: Aggregate Supply Aggregate Demand Analysis a History

340 History of Political Economy 34:2 (2002)

price levels and shows “how much businessmen will sell if they pro-duced their planned output at a given price level” (328).28 This curve isupward-rising up to the price level at which full employment is reached(since higher price implies higher levels of output and hence consump-tion demand), but above that price it becomes backward-bending, sincethere is higher price but no higher output. In its positively sloped seg-ment it is steeper than the AS curve. Output and price are determined atthe intersection of the AD and AS schedules. Rowan’s model is an earlytextbook presentation of theWalrasian disequilibrium approach with ra-tioning at a (temporarily) fixed price, and like other fixed-price disequi-librium models with perfect competition, the model suffers from the in-consistency of having price-taking firms that cannot sell all they wishat the going price.29 The model has the additional problem that it doesnot make consistent use of the short-side principle in calculating effec-tive demand, taking the level of output supplied notionally, rather thanthe level of output actually produced (see Dalziel 1993). Rowan’s laterbook, Rowan and Mayer 1972, removes the AD-AS apparatus, return-ing instead to a brief treatment of the AD-AS in Z-N space in additionto the income-expenditure and IS-LM models. Despite the fact that hefailed in his endeavor, Rowan’s attempt to derive the AD curve as a truedemand schedule for all goods (like Bent Hansen) rather than an equi-librium schedule for the goods and asset markets, is an interesting one,if only to point out that the AD curve is not a demand curve at all.

The two popular new textbooks that appeared in the closing years ofthe 1970s—Dornbusch and Fischer 1978 and Gordon 1978—and an-other widely used textbook in its second edition—Branson 1979—alluse the AD-AS framework. After presenting the income-expenditureand IS-LM models with a constant price, they derive the AD curve by

28. In terms of the equations of the model in table 1, Rowan’s diagram is derived as follows.The AS curve shows the value of Y at each level of P from equations (6) through (7) in theusual way. The curve, however, becomes vertical at the full employment level (given a standardsupply curve of labor, which shows labor supplied at each real wage). TheAD curve is derivedas follows. We start from a given P . From equations (7) and (8) we find N , which yields thevalue of Y from equation (6). This output is restricted to not exceed the full employment level,as in the AS curve. The P and Y just found are inserted into equations (4) and (5) to find theasset market clearing level of r . These values of r and Y are substituted into equations (2) and(3) to find the levels of C and I . These values, together with the exogenously given level of G,are summed up to get the aggregate demand at the given P . The AD curve is found by findingthis level of aggregate demand (C + I + G) for each P .

29. See, for instance, Dutt 1987 for an elaboration of this point.

Page 22: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 341

determining real output from the model for different price levels, follow-ing earlier presentations; Gordon also introduces the real balance effect.

They depart from earlier tradition in their treatment of the AS side,however. Dornbusch and Fischer develop the AS curve under the as-sumption of imperfect competition, with firms charging a markup overwage costs.30 Given a Phillips curve, which shows that a lower unem-ployment rate implies a higher rate of wage increase, a positive relationbetween the output level and the price level, given last period’s wage,is derived: a higher output level implies a higher employment level, alower unemployment rate, a higher rate of wage change, a higher wage(given last period’s wage), and hence a higher price from the markupequation. This is the short-run AS curve, which extends beyond the po-tential or full employment output level. Gordon and Branson, on theother hand, consider a Friedman-Phelps supply curve in which work-ers make labor supply decisions given the expected real wage, firmsmake labor demand decisions under perfectly competitive assumptionsand diminishing returns, given the real wage (since, unlike workers, theyare assumed to know the price of the product),31 and the labor marketclears through money wage variations. The AS curve is obtained for agiven price expectation: starting from an initial position of labor mar-ket equilibrium, a higher price level implies that firms demand more la-bor at a given money wage while workers supply no more (since ex-pected price does not change), there is excess demand for labor, andthe money wage and employment levels are higher, so that output ishigher.32 Gordon and Branson therefore verbally tell a Keynesian storybut use a Friedman-Phelps labor market model with perpetual labormarket clearing in which output deviates from “the” full employmentlevel only due to worker misperceptions regarding the price.33 The

30. Successive editions of Dornbusch and Fischer have changed the nature of the pricingrule. The earlier editions assumed diminishing returns but set the price as a markup on the wageor a constant multiple of the wage (the multiple being determined by the labor-output ratio atexpected output). Later editions have assumed a fixed-coefficient technology with a constantlabor-output ratio, and assumed that the price is a constant markup on unit labor costs. Themarkup equation has also beenmodified to take into account nonlabor material costs to examinethe effects of oil price shocks.

31. Branson also considers the case of imperfect competition.32. For a given price expectation of workers, the curve is the same as Marschak’s (1951)

curve. However, Marschak’s model assumes money illusion rather than expectational errorswhich can lead to the revision of expectations—see below.

33. In a later edition Gordon (1981) also presents a version with unemployment due to arigid money wage in the short run, as in his earlier presentation. In a still later edition Gordon

Page 23: Aggregate Supply Aggregate Demand Analysis a History

342 History of Political Economy 34:2 (2002)

introduction of rational expectations into this model would convert itcompletely into a new classical model!

Output and the price level are determined at the intersection of theADand the AS in all three presentations, but this equilibrium is a short-runone. Dornbusch and Fischer are the most explicit on this point, analyz-ing also the longer-run dynamics of the model explicitly. If the levels ofoutput and employment are lower than their potential levels, the moneywage will be lower in this period than in the previous period, so that lastperiod’s wage will be lower in the next period than in the current period.This implies that the short-run AS curve will move downward. Giventhat the AD curve is downward-sloping, the economy must converge toa long-run equilibrium at which the short-runAS curve and theAD curveintersect each other at the potential level of output, which is denoted bya vertical long-run AS curve at which the money wage does not changeover time.34 Thus we obtain the result that in the short run there maybe unemployment (or overemployment), but in long-run equilibrium theeconomywill be at full employment. This appears to be the first textbookneoclassical synthesis model that explicitly uses AD and AS curves andthat has Keynesian unemployment in the short run (at the intersectionof the AD and short-run AS curves) and classical full employment (atthe intersection of the AD and long-run AS curves) in the long run (al-though the idea was present in other presentations that did not use formalAD-AS models, as in Samuelson’s principles text from the third editiononward). Gordon’s AS curve also moves when we go beyond the shortrun as firms revise their short-run price expectations adaptively. Whenprice expectations are fulfilled in long-run equilibrium, the AS curve isvertical and the economy is at its potential level of output. Despite thisimplication of his model, Gordon does not appear very convinced thatthis self-adjusting mechanism will work, due to a liquidity trap, con-sumers’ expecting prices to keep falling when there is deflation, and theredistribution of income from debtors with a high propensity to spend tocreditors with a lower propensity to spend, ideas that are left out of hisformal model. Branson does not explicitly develop a long-runAS curve,

(1984) removes this Friedman-Phelps version of the supply curve and bases it entirely on theassumption of a rigid money wage.

34. The process works in reverse if output exceeds the potential level, with the short-runAScurve shifting upward until the economy arrives at long-run equilibrium at the potential levelof output.

Page 24: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 343

but since his long-run Phillips curve is not vertical, it appears that hedoes not expect full employment to exist in the long run.

Most subsequent intermediate texts adopted some form of the AD-AS model. Surveys of intermediate texts show that out of twenty re-viewed, only two—Auerbach and Kotlikoff 1995 and Barro 1990—donot contain the AD-AS model (Dutt 1997; Grieve 1998).35 The AD-ASmodels in the texts that do contain them come in four main versions.The most common is the one that bases the AD curve on the IS-LMmodel by examining equilibrium output levels for different price lev-els, and the AS curve on nominal wage rigidity, perfect competition,and profit maximization, as in McKenna 1955. However, almost all ofthem allow the short-run supply curve to have a positive slope for lev-els of output above its potential or full employment level (that is, theoutput level shown by the vertical long-run supply curve), and allowthe short-run supply curve to shift in the long run, so that in long-runequilibrium the economy is at the potential level of output. Quite a fewof these presentations—although not all—explicitly assume quantity ad-justment at given prices in deriving the AD curve and can therefore becharged with inconsistency (see Dutt 1997). A second version—as inHall and Taylor 1991 and Mankiw 1992—assumes that the short-run ag-gregate supply curve is horizontal due to short-run nominal price rigid-ity, but, in the long run, price adjustment and movements in the short-runaggregate supply curve take the economy to the potential level of out-put. A third version—as in Dornbusch and Fischer (as mentioned above)and Blanchard 1997—explicitly introduces imperfect competition us-ing a markup-pricing equation and has an upward-rising short-run sup-ply curve based on the Phillips curve; over time the short-run AS curvemoves due to changes in the money wage, and in the long run the econ-omy is at the potential level of output. These second and third versions donot assume price-taking behavior for firms in deriving the AS curve, sothat the charge of inconsistency cannot be sustained.A final form—as inParkin 1984—does not allow for involuntary unemployment even in theshort run but derives the short-run AS curve, given the price expectedby suppliers of labor (as in the earlier models of Gordon and Bransondiscussed above); in the long run the economy is at the potential levelof output. To the extent that these presentations derive the AD curve by

35. Auerbach and Kotlikoff (1995) present most of their analysis using the IS-LM model,while Barro (1990) devotes only one chapter—the last—to Keynesian models and there usesonly the IS-LM model, the rest containing new classical models.

Page 25: Aggregate Supply Aggregate Demand Analysis a History

344 History of Political Economy 34:2 (2002)

changing the exogenously given price, they may be charged with incon-sistency, although one can make the more charitable interpretation thatthe curve is derived in this way for heuristic purposes and could havebeen derived with output as the independent variable.

The late dispersal of the AD-AS apparatus in intermediate textbooksin the later 1970s onward may be somewhat puzzling since, as we sawearlier, it had already appeared in the literature in the late 1940s andearly 1950s and at least one intermediate textbook had adopted it by themid-1950s. The timing of dispersal arguably has to do with real-worldeconomic issues and theoretical developments within macroeconomics.In terms of real-world issues, the supply shocks of the 1970s and thehigh inflation rates of the 1970s made it inconvenient for textbooks tocontinue using income-expenditure and IS-LM models, which focusedon the demand side and assumed the price level to be given (see Kennedy1998). This is confirmed by a perusal of the prefaces of textbooks thatadopted the framework in the late 1970s. Dornbusch and Fischer (1978,v) state that real-world events have made them “go beyond the standardcoverage in presenting also the theory of aggregate supply, [and] the vi-tally important topics of inflation and unemployment.” Branson (1979)also states that one of the reasons for introducing the model in his secondedition was the inflation of the 1970s. In terms of theoretical develop-ments, Keynesian economics was by this time under attack from mon-etarist and new classical economists who focused on the supply side ofthe economy and in whose analysis the price level played an importantrole. The AD-AS model allowed a clearer comparison of the Keynesiansystem and the monetarist and new classical systems in common termsthat highlighted both demand and supply sides of the macroeconomy(Kennedy 1998). Branson (1979) explicitly gives this reason in his pref-ace. Also in terms of theory, the AD-AS model can be thought of as themost complete of a sequence of macroeconomic models. The income-expenditure model is usually interpreted as taking the interest rate andthe price level to be given; the IS-LM model, as relaxing the assump-tion about the fixed interest rate but continuing to assume a given pricelevel; and, finally, the AD-AS model, as endogenizing even the pricelevel. This sequence made it appear that the AD-AS model is the log-ical culmination of this sequence of models and hence the most com-plete and general one for analyzing the economy. This reason—whichincidentally was already apparent in Marschak’s (1951) and McKenna’s(1955) presentations—may have had some role in the rapid spread of

Page 26: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 345

the model, once it was already in a number of textbooks. Finally, con-formism (more on which later, in the discussion on principles texts) dueto competition among intermediate texts arguably had an important rolein this propagation.

AD-AS Analysis in Principles Textbooks

Principles textbooks were slower in adopting the AD-AS framework,and in fact they were slower in adopting Keynes’s analysis in any form.36

Garver and Hansen 1937, Benham 1938, Fairchild, Furniss, and Buck1939, and Meyer 1941, which all appeared after The General Theory,do not discuss Keynes’s theory of output at all and explain the price levelin terms of the quantity theory of money with full employment. If at allthey are mentioned, plant closings, output reductions, and business cy-cles are discussed in microeconomic terms. In later editions, Fairchild,Furniss, and Buck in 1948 and Benham in 1955 continued with this ap-proach. Garver and Hansen’s 1947 edition and Meyer’s 1948 edition in-troduce the Keynesian national accounting scheme and allow underem-ployment equilibrium, but have no Keynesian theory, the latter not evendiscussing the consumption function and the multiplier. The only realexception in these early days is the Keynes-inspired textbook by Meade(1936), which analyzes the problem of unemployment in terms of aggre-gate demand and states that reducing unemployment requires increasedspending. Meade does not, however, use equations or diagrams to ex-plain how output is determined.

It is not until the appearance of the new textbooks by Tarshis (1947),Samuelson (1948), and Boulding (1948) that the Keynesian model ofincome determination begins to play a central role in principles texts.Samuelson’s text, of course, became the introductory textbook ofeconomics, going through eleven editions by 1980.37 Understandably,Samuelson’s dominant position allowed him to set the tone for otherprinciples authors, which makes it important to discuss his version ofthe Keynesian model. Samuelson’s exposition is centered around the the-ory of income determination using an income-expenditure model that—as mentioned earlier—he introduced in Samuelson 1939. Although herelates Keynes’s theory of liquidity preference, he does not integrate it

36. See also Pearce and Hoover 1995.37. See Elzinga 1992 and Pearce and Hoover 1995.

Page 27: Aggregate Supply Aggregate Demand Analysis a History

346 History of Political Economy 34:2 (2002)

with his model of income determination, presumably since in his viewmonetary policy is ineffective (see Pearce and Hoover 1995, 197) be-cause of the existence of excess reserves, which allow financial institu-tions to absorb the effect of policy changes without changing interestrates. Later editions give a more prominent role to money, but continuewith the emphasis on the income-expenditure model. Except in discus-sions of the inflationary gap, the price level and wage determination arealso not incorporated into the formal apparatus, but described indepen-dently of it. This may be explained at least in part by Samuelson’s beliefthat the Phillips curve is unstable (see Pearce and Hoover 1995, 206).

Subsequent textbook writers, such as George Bach (1954), JohnGuthrie (1957), and Campbell McConnell (1960), follow Samuelson infocusing on the income-expenditure analysis, generally calling the ag-gregate expenditure curve the aggregate demand curve, and the 45◦ line(sometimes up to the full employment line, at which it becomes verti-cal) as the aggregate supply curve. These texts generally assume thatthe price level is given, but sometimes note that as output increased tothe full employment level (before full employment is reached), the pricelevel increased due to increases in the price of materials, wages, or sec-toral bottlenecks.

The first principles texts to use the AD-AS analysis appear to be Lin-dauer 1977 and Baumol and Blinder 1979.As noted earlier, Lindauer hadintroduced it into his intermediate text almost a decade earlier, where hehad derived the AD curve from the IS-LM model and the AS curve withwage rigidity. The principles text, however, provides a superficial treat-ment, drawing theAD curve with a downward slope without a proper ex-planation beyond noting that at a lower price customers buy more (596),and the AS curve is drawn with a rising part that is not clearly explainedand a vertical part at full employment. Lindauer is clearly choosing anincorrect but simple story over the correct but complicated one of his in-termediate text. Upward shifts in AD and AS curves are used to explaininflation of the demand-pull and cost-push types, respectively, but theincome-expenditure model remains the mainstay of the book. Baumoland Blinder provide a fuller discussion.Although the initial discussion oftheAD curve sounds too much like that for a micro demand curve—thatis, a higher price, given income, implies a lower quantity demanded—the later discussion makes it clear that the downward slope is due to theKeynes effect. The AS curve is explained as having three regions: thefirst fairly flat, the second rising, and the last virtually vertical. The first

Page 28: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 347

is argued to represent the Keynesian approach and the last the monetaristone. Changes in the money wage are argued to shift the AS curve: withunemployment, the money wage falls, shifting the AS curve downwardand expanding output, given the downward-sloping AD curve. The au-thors note that this self-correcting mechanism takes the economy to fullemployment in the long run, but they point out that shifts in the curve aremuchmore efficient at eliminating inflationary gaps than unemployment.

The late arrival of the AD-AS curves into principles texts can partlybe explained by the fact that these curves, especially theAD, are difficultto explain to students who are just being introduced to economics us-ing simple verbal exposition. Intermediate texts using the IS-LM frame-work could easily explain the downward-sloping AD curve in terms ofshifts in the LM curve and the Keynes effect, and the real balance effectcould also be easily introduced through shifts in the IS curve. But princi-ples texts, which almost all used only the income-expenditure diagram,found this harder to do, which is probably why Lindauer eschews sucha discussion despite the fact that his intermediate text contains such ananalysis. Another reason can be traced to the nature of the economicsprinciples textbook market. The market was dominated for a long timeby Samuelson’s book, which went through many editions over the years.Samuelson, and others who were popular and who therefore publishedseveral editions, arguably found it inconvenient and time-consuming torestructure their texts to incorporate the newly developedAD-ASmodel.This inertia among the market leaders, who set the tone for other books,probably slowed down the adoption of the model. It is not surprising,then, that when the AD-AS model did end up in texts it did so in newbooks (as in the case of Baumol and Blinder 1979) or when substantialrevisions took place with the addition of new coauthors in the case ofestablished texts. Samuelson’s text switched to using the AD-AS onlywhen a coauthor was added in the twelfth edition of the book (Samuel-son and Nordhaus 1985). McConnell’s popular text, which went throughnumerous editions since it first appeared in 1960, also went through itsmost extensive revision (which also involved the AD-AS model) in itseleventh edition, which added a coauthor (McConnell and Brue 1990).

An additional feature of the market was the strong need to conform.Even Baumol and Blinder (1979, vi), one of the first to introduce theAD-AS model in principles texts, write in their preface that they have“by and large tried to avoid novelty,” and they first introduce the modelalmost through the backdoor as a digression (but subsequently go on to

Page 29: Aggregate Supply Aggregate Demand Analysis a History

348 History of Political Economy 34:2 (2002)

use it extensively later in the book). These characteristics are probablyalso present in the intermediate text market (note that the Dornbusch-Fischer and Gordon texts were new ones), but arguably not as importantin them because of the smaller size of the market and the absence of adominant text comparable to Samuelson’s.

Nevertheless, the need to introduce the analysis was quite strong, forsome of the same reasons why the analysis became popular in the in-termediate textbooks, mainly, the need to explain the supply shocks andhigh inflation rates of the 1970s. A look at the prefaces of the pioneer-ing books confirms this. Although Lindauer (1977) provides no expla-nations, Baumol and Blinder (1979, vi) state that while many textbooksinitially treat the price level to be given, they from the very beginning“analyze the world as it really is, with prices all too readily driven up-wards.” It is worth noting that Blinder (1979) also used the diagram asthe theoretical underpinning to explain the stagflation of the 1970s at arelatively nontechnical level. Samuelson and Nordhaus (1985) also usethe need to explain the concurrence of unemployment and inflation astheir reason for adopting the model. But their preface provides an ad-ditional reason, that is, to integrate different schools of thought, includ-ing “Keynesian, classical, monetarist, supply side, rational expectationsand modern mainstream macroeconomics” (viii). McConnell and Brue(1990) also give the need to explain differences between competing per-spectives as one of the reasons for using the model.

In the following years other principles texts began to adopt the frame-work. By 1985, Hansen, McCormick, and Rives (1985) show that outof twenty principles texts they survey, fourteen had the AD-AS frame-work and another one included it in a supplement to the text—implyinga sample adoption rate of 75 percent. While inflation continued to makethe analysis relevant, two other reasons may be given to explain thisrapid spread. The first is that principles texts began to treat the AD-ASframework as just another application of the standard demand-supplyanalysis. The typical textbook would introduce that analysis early in thetextbook and then use it in both the macro and micro parts of the book,irrespective of which was dealt with first. As we have already seen, thiscreated some problems in distinguishing the micro demand curve fromthe macro one, but most textbooks started distinguishing the two curvesmore carefully, pointing out that the macro curve’s negative slope wasdue to the Keynes, real balance, or some other effect. The second has todo with intense competition in the principles text market: the appearance

Page 30: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 349

of a large number of textbooks vying for a larger share in the large mar-ket made each one try to do at least as much as its competitors and more.Since a few adopted the AD-AS model, others were induced to do thesame. Conformism, which delayed the appearance of the model in prin-ciples texts, now speeded up its diffusion. Even Samuelson’s textbook,with coauthor William Nordhaus, eventually bowed to the trend: everychapter in the macroeconomics part of the book is explicitly related tothe AD-AS model!

Virtually every principles text now contains the AD-AS model, al-though those texts usually also contain the old workhorse income-expen-diture model. Of the eighteen such texts that are currently in existence orhad several editions by the 1980s that I reviewed,38 only Heilbroner andGalbraith 1987 and Fusfeld 1988 did not contain the model, implying anincrease in the adoption rate of close to 90 percent (in the sample).39

The texts containing the AD-AS analysis have the following com-mon features. First, the AD is shown to be downward-sloping. All ofthem explain this in terms of the Keynes effect and the real balanceeffect, most also introduce the foreign trade effect due to substitutionbetween domestic and foreign goods, and a few introduce tax effectsand intertemporal price effects (see McCormick and Rives 1998). Thedownward-slopingAD curve is not normally explicitly derived from theIS-LM diagram. Few textbooks discuss that model, and the handful thatdo, do so informally, usually in a chapter appendix. There is thus lit-tle evidence of inconsistency in terms of pricing theories; the presenta-tions are not specific enough for that. Second, despite this vagueness andreference to the similar shapes of microeconomic demand curves andthe AD curve, most recent texts are careful to point out the logic of theAD curve as an equilibrium schedule. Some, such as Parkin 1998, how-ever, leave room open for confusion by defining the curve to show the

38. The books reviewed wereAmacher and Ulbrich 1995, Baumol and Blinder 1997, Bron-fenbrenner, Sichel, and Gardner 1984, Case and Fair 1996, Colander 1998, Fischer, Dornbusch,and Schmalensee 1988, Fusfeld 1988, Heilbroner and Galbraith 1987, Lipsey, Steiner, andPurvis 1984, Mankiw 1997, Mansfield 1992, McConnell and Brue 1996, Miller 1997, Parkin1993, Ruffin and Gregory 1997, Samuelson and Nordhaus 1995, Stiglitz 1997, and Tresch1994.

39. The first of these is the eighth edition of the Heilbroner-Galbraith text, which is at amore elementary level than the other texts and contains only the income-expenditure modeland a discussion of inflation in terms of the Phillips curve. Fusfeld’s is somewhat idiosyncratic,using explicit Marxian terminology and being the only relatively recent text using Keynes’sconcepts of aggregate demand and supply price (although in Z-Y rather than the earlier Z-Nspace).

Page 31: Aggregate Supply Aggregate Demand Analysis a History

350 History of Political Economy 34:2 (2002)

relationship between the quantity demanded and price (and by present-ing the curve before discussing goods market equilibrium in the income-expenditure diagram; see p. 521). One textbook—Colander 1998—triesto deal with this problem with a lengthy discussion distinguishing be-tween demand and supply curves and the macro curves used in the text-books, and by renaming theAD curve the aggregate equilibrium demand(AED) curve and the AS curve the aggregate supply path (ASP). How-ever, in a more recent edition, Colander (2001) returns to theAD-AS ter-minology.40 Third, the short-run AS curve is drawn to be upward-rising,although most textbooks argue that it is nearly horizontal at low levelsof output, then rising, and then vertical at some point beyond the fullemployment level. The upward slope is explained in terms of (1) someinput prices (such as the wage) being relatively fixed in the short run,so that firms facing higher prices produce more; (2) increases in somecosts (including wages) as output rises, which get translated into higherprices; and (3) the appearance of supply bottlenecks in some sectors asoutput increases, which pushes up the prices of certain goods. None ofthe texts, not even Parkin’s (1993)—unlike his intermediate text—usesthe price misperceptions story. The discussion is not formal, and onlyone textbook derives it explicitly from profit-maximizing, price-takingbehavior for a given wage. Third, all texts make the AS curve shift dueto wage changes and argue that the short-run AS curve shifts to inter-sect theAD curve at the full employment level of output and the verticallong-runAS curve in the long run. Competition seems to have led to notonly the adoption of the framework, but also to near complete conver-gence in its presentation!

Conclusion

I conclude by summarizing the analysis of the foregoing sections to dis-cuss in turn the appearance and spread of theAD-AS framework, its dif-ferent forms, and its criticisms, and by commenting on the recent callsfor abandoning it. A tabular summary of the main contributions is alsoprovided in table 2.41

40. Colander (2001, ix–x) writes in his preface that his alternative methodology “involvedsimply too much analytics for students to learn.”

41. The table follows the text in defining the AD-AS framework as one that shows macro-economic equilibrium using aggregate demand and supply curves in P -Y space. The table doesnot discuss the problem of “dirty pedagogy,” which is to be found only in some of the principlestexts—such as Lindauer’s (1977) and several of the later ones.

Page 32: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 351

Table 2 Summary of Main Economics Textbooks

Effect ofAuthor, Date, WageType of Work AD curve AS curve Inconsistency? Reduction

Keynes 1936; No. Keynes effect No. Analysis No Ambiguousscholarly present, but based on perfectbook possibly offset by competition and

other effects. fixed W.

Hicks 1937; No. But equations No. But No Rise inscholarly present. Keynes equations output andjournal article effect. present, based on employment

perfectcompetition andfixed W.

Tarshis 1947; No. Implicitly Upward-rising; No Ambiguoustextbook vertical. based on perfect

competition andfixed W.

Brownlee Money market Upward-rising; No Ambiguous1948; equilibrium with based on perfectarticle in fixed r competition andtextbook fixed W.

Brownlee Downward- Upward-rising; No Rise in1950; sloping due to based on perfect output andscholarly Keynes and real competition and employment,journal article balance effects. fixed W. although

Derived from IS- barriers areLM. noted.

Marschak Downward- Upward-rising; No Not1951; sloping due to based on perfect discussed,textbook Keynes and real competition and but implicitly

balance effects. money illusion in an increaseDerived from IS- labor supply in output andLM. function or fixed employment.

W.

Weintraub Downward- Upward-rising; No Not1951; sloping; not based on perfect discussedtextbook explained. competition and

fixed W.

Page 33: Aggregate Supply Aggregate Demand Analysis a History

352 History of Political Economy 34:2 (2002)

Table 2 continued

Effect ofAuthor, Date, WageType of Work AD curve AS curve Inconsistency? Reduction

B. Hansen Downward- Upward-rising; No Not1951; sloping demand based on perfect discussed;scholarly schedule due to competition and presumablybook differential fixed W. negative due

consumption to shift inpropensities. income

distribution.

McKenna Downward- First flat, then Maybe Formal1955; sloping; derived upward-rising; modelintermediate from IS-LM; based on impliestext Keynes and real monopoly and increase in

balance effects. then perfect output. Butcompetition with discussesfixed and problems.variable W.

Lindauer Downward- Upward-rising; Maybe Formal1968; sloping; derived based on perfect model showsintermediate from IS-LM; competition and increase intext Keynes and real fixed W. output. Cites

balance effects. Keynes topoint outambiguities.

Rowan 1968; Partly upward Upward-rising; Yes Not discussedintermediate rising and partly based on perfecttext downward competition and

sloping; uses AS fixed W.curve in drawingcurve to findaggregatedemand forgoods at eachprice level; notan equilibriumschedule.

Page 34: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 353

Table 2 continued

Effect ofAuthor, Date, WageType of Work AD curve AS curve Inconsistency? Reduction

Dornbusch Downward- Short-run rising No Increase inand Fischer sloping; derived with markup output with1978; from IS-LM; pricing and fullintermediate Keynes effect. Phillips curve; employmenttext long-run in long-run

vertical. equilibrium

Gordon 1978; Downward- Friedman-Phelps Maybe Same asintermediate sloping; derived expectation above;text from IS-LM; errors; drawn informally

Keynes effect with given price argues thatand real balance expectations for this iseffects. workers and unlikely in

labor market reality.clearing.

Branson Downward- Same as Gordon Maybe Not discussed1979; sloping; derived 1978.intermediate from IS-LM;text Keynes effect.

Lindauer At lower prices Partially upward Not explicit Not discussed1977; consumers buy rising, but enough toprinciples text more; conflating explanation reveal

micro and macro unclear. inconsistency.concepts.

Baumol and Unclear at first, Has upward- Same as Rise inBlinder 1979; then downward- rising part Lindauer output andprinciples text sloping due to because of wage 1977. employment.

real balance and rigidity. Process notKeynes effects. very

effective.

Samuelson Downward- Has upward- Same as Rise inand Nordhaus sloping due to rising part due to Lindauer output and1985; real balance and the rigidity of 1977. employmentprinciples text Keynes effects. some prices and

wages.

Page 35: Aggregate Supply Aggregate Demand Analysis a History

354 History of Political Economy 34:2 (2002)

1. The AD-AS framework initially appeared as a formalization ofthe Keynesian system of output and price determination. However, itwas preceded by a number of alternative formalizations, including thediagonal-cross income-expenditure model, the IS-LM model, and theaggregate demand price–aggregate supply price model in employmentand value of output space. Hicks’s original IS-LM model (which hecalled SI-LL) is virtually the same in structure as the AD-AS model,but it was shunted in a different direction by his popularizers.

2. From its first appearance in full form in a journal article by Brown-lee in 1950 and in the published lecture notes of Marschak in 1951, theAD-AS framework has come a long way: it is now to be found in vir-tually every textbook at the principles and intermediate levels. It wasadopted by McKenna in an intermediate text in the mid-1950s, but didnot spread beyond a couple of such texts until the late 1970s, when it wasused by a number of leading texts. It was only then that its rapid disper-sal started. Its adoption in principles texts began even later, only in thelate 1970s, but it was widely adopted in the 1980s and even more so inthe 1990s. Today, although textbooks still use the income-expendituremodel and the IS-LM model (the latter appearing mostly in intermedi-ate texts), the AD-AS model receives the pride of place as the completedepiction of output and price determination.

3. Although the framework has been known since the early 1950s andappeared in a few intermediate textbooks after that, it spread extensivelyin the later 1970s only after inflationary pressures led economists to giveincreasing attention to the price level and the supply side, and with thegrowing challenge of monetarist and new classical alternatives to theKeynesian approach. The adoption by principles texts was delayed bythe perceived complexity of the framework, but simplifications of theframework (and its identification with microeconomic supply and de-mand curves), real-world and theoretical developments in economics,and intense competition among different textbooks led to its subsequentrapid spread. The story for the delayed appearance and later rapid prolif-eration of the framework provides an interesting case study of the spreadof ideas in textbooks and of how this spread is affected by real-worldevents, theoretical developments in the discipline, and conformism.

4. Our history shows that the AD-AS framework has taken manyforms. It was initially developed to portray the Keynesian system witha fixed wage and perfect competition, but variants also dealt with moneyillusion, some wage flexibility, and imperfect competition remaining

Page 36: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 355

within the Keynesian system. Some writers also represented full em-ployment equilibrium with perfect wage flexibility with it to compareit with the Keynesian case. Subsequent versions of the framework haveeven dealt with the monetarist price misperceptions story with full em-ployment. While most versions have interpreted the AD curve to showgoods and asset market equilibrium and explained its slope in terms ofthe Keynes effect and the real balance effect, some have interpreted it asa demand schedule for goods or explained its slope in terms of incomedistributional shifts. At present, however, this diversity has been dimin-ished by an amazing degree of convergence in interpretation, especiallyin principles texts.

5. The charge of inconsistency between the AD and AS curves can-not be sustained as far as the early developers of AD-AS analysis areconcerned. Brownlee, for instance, did not take the price level to be theexogenous variable in the sense of assuming quantity adjustment in thegoods market for a given price, but explicitly defined the AD curve asshowing combinations of P and Y showing goods and asset market equi-librium. Marschak did not assume a fixed price in his treatment of ei-ther the income-expenditure or the IS-LM model. Even Hicks’s SI-LLmodel is not inconsistent. In subsequent intermediate texts, the incon-sistency between the AS and AD sides can be argued to be present onlyin those texts that explicitly develop the AD curve from the fixed-priceincome-expenditure or IS-LM model and the AS curve from the price-taking, profit-maximization model; inconsistency is not present in thosetexts that explicitly assume imperfect competition or those that assumea given price in the short run. Moreover, even those texts that do use thefixed-price model to motivate the AD curve and use the perfectly com-petitive underpinning for the AS curve may not be really inconsistent,since the AD curve can be interpreted as showing equilibrium combina-tions of price and output that equilibrate goods and asset markets—notnecessarily showing the equilibrium output for a given price (see Dutt1997 and Kennedy 1998). Since most principles texts do not explicitlydiscuss the derivation of the AD curve in terms of fixed-price IS-LMmodels, and do not explicitly introduce price-taking behavior in deriv-ing theAS curves, it is difficult to argue that the presentations suffer frominternal inconsistency; the vagueness of their discussions protects themfrom such a charge. In sum, the charge of internal inconsistency mayhave some merit for some current intermediate texts, and perhaps the

Page 37: Aggregate Supply Aggregate Demand Analysis a History

356 History of Political Economy 34:2 (2002)

fault may originate in the early treatments of McKenna and Lindauer,but it cannot be sustained for all texts.

6. As we have seen, almost all textbook presentations of the AD-ASframework today have the implication that unemployment (or in the caseof the monetarist model, deviations from the potential level of output) istemporary and self-defeating due to wage variations (or expectationalchanges in the monetarist model),42 in contradiction to Keynes’s ideas.Although faithfulness to Keynes’s ideas should not be taken to be anacid test for the validity of a model, it is quite possible—as discussedearlier—that the result ignores a number of important characteristics ofreal economies and hence makes the model unsuitable for understandingtheir behavior.

Three comments should be made about this criticism, however. First,as our history has shown, the early contributors to AD-AS analysis suchas Tarshis and Brownlee explicitly pointed out that a wage reductionneed not increase employment. Early presenters of the framework intextbooks such as McKenna’s and, especially, Lindauer’s, and in somelater texts such as Gordon’s, also make it clear that Keynes’s view wasthat wage reductions need not take the economy to full employment (al-though the formal models developed by all of them do imply full em-ployment in the long run due to the Keynes and real balance effects).The neoclassical synthesis model with long-run full employment ap-pears explicitly only in the late 1970s. Second, the wage rigidity viewof Keynes, which states that without such rigidity the economy goes tofull employment and the economy is therefore self-correcting, is a viewthat has spread widely and is now accepted by most—but by no meansall—mainstream Keynesians, especially the new Keynesians (see Duttand Amadeo 1990a). However, the propagation of this view is a phe-nomenon that is to a large extent independent of the development ofthe AD-AS framework, although it is arguable that the framework mayhave aided it. Hicks’s early SI-LL model and subsequent IS-LM mod-els imply the result. Franco Modigliani (1944) analyzed a Keynesianmodel in which the money wage is rigid below full employment andargued that the distinctive feature of Keynes’s analysis—a feature thatenabled him to get unemployment at equilibrium—is the fixed moneywage assumption. Modigliani admitted that if the liquidity trap existed,

42. For the monetarist price misperceptions model, this should be translated to read that de-viations from the potential level of output are temporary and self-defeating due to expectationalchanges.

Page 38: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 357

unemployment could exist without a rigid money wage (he could toohave added that this could happen also if investment is interest inelas-tic); otherwise a reduction in the money wage will reduce the price and,through the Keynes effect, expand output and employment. A. C. Pigou(1943), Gottfried Haberler (1946), and Don Patinkin (1956) went furtherto point out that money wage rigidity was essential for Keynes’s conclu-sion of unemployment equilibrium—even if the interest elasticity of in-vestment was zero and the economy was in the liquidity trap—becauseof the real balance effect, a point that was also made in Brownlee 1950.43

Third, since the result depends on having a negatively sloped AD curvethat does not shift when the money wage is changed, it is possible toavoid it by taking the curve to be upward-rising and by assuming that itmoves to the left when the money wage falls, thereby taking into accountsome of the arguments and the other critics of the AD-AS framework(see Dutt 1986–87). Thus, the negative slope for the AD curve assumedin almost all presentations does not invalidate the AD-AS apparatus.

7. The problem of vagueness and “dirty pedagogy” arguably affectedthe early principles texts that appeared in the late 1970s, but not the in-termediate texts, which were quite explicit in their treatment of the ADcurve. Even principles writers—who all too often slip into the practiceof defining the AD curve as showing the aggregate quantity of goodsand services demanded at each price—have subsequently become morecareful about the problem and provide clearer expositions of the curve asan equilibrium schedule.44 The problem is therefore not insurmountable.

8. The foregoing discussion allows us to conclude that the historyof AD-AS analysis does not provide us with a case for jettisoning theframework. The criticisms of internal inconsistency, lack of realism, and“dirty pedagogy” may apply to some versions of the analysis, but not tomany, and the problems have in some cases arisen recently. In any case,they do not imply insurmountable obstacles.

But the failure of these criticisms to stick may not be enough to con-vince us that the AD-AS framework should be retained. The history ofAD-AS has shown us that the spread of the framework was related to

43. Meir Kohn (1981) has argued that even without the real balance effect, money wagerigidity is necessary for unemployment equilibrium, since investment depends on the real in-terest rate, which is not fixed by the liquidity trap.

44. Indeed, Colander, one of the early critics of “dirty pedagogy,” writes in the preface of thelatest edition of his principles text that “all principles books now do a better job presenting theAD/AS model, and distinguishing it from a micro supply/demand model” (Colander 2001, x).

Page 39: Aggregate Supply Aggregate Demand Analysis a History

358 History of Political Economy 34:2 (2002)

the rise in importance of price level changes and inflationary pressuresin the economy and to the rise of alternative models that stress the supplyside of the economy. We may be led to argue that inflation is no longeran important issue in present-day economies and the world may in factreveal deflationary tendencies.

However, in addition to the fact that the claim that inflation has beentamed may be premature, the need for the analysis of both the pricelevel and output arises because stable prices are worthy of explanation,and the effects of deflationary tendencies need to be understood, forwhich the AD-AS framework continues to be a useful tool. Moreover,macroeconomists also do not agree on a single model of the economy,as evidenced by the presence of a variety of models in textbooks andin the scholarly literature. By virtue of its flexibility in portraying anumber of different models of the economy in price-output space, theAD-AS framework can be used to understand these different modelsand to compare them using a common framework. Indeed our historyof the framework has shown that different forms of the framework ex-ist in the literature, and many textbook writers have used the frame-work because of its ability to compare alternative approaches andmodelsin macroeconomics. These different models include not only standardmonetarist/new classical and new Keynesian models, but also those thatincorporate imperfect competition, income distributional issues, and theadverse effects of wage reductions stressed by post-Keynesian econo-mists and other more mainstream Keynesians. It is in this that the mainstrength of the framework lies, making it a more useful device for teach-ing macroeconomics than alternative presentations such as the income-expenditure and IS-LM models, which do not explicitly incorporate theaggregate supply side.

References

Abel, Andrew B., and Ben S. Bernanke. 1992. Macroeconomics. Reading, Mass.:Addison-Wesley.

Amacher, Ryan C., and H. H. Ulbrich. 1995. Economic Principles and Policies. 6thed. Cincinnati: South-Western College Publishing.

Auerbach, Alan, and Laurence Kotlikoff. 1995. Macroeconomics: An Integrated Ap-proach. Cincinnati: South-Western College Publishing.

Bach, George L. 1954. Economics: An Introduction to Analysis and Policy. Engle-wood Cliffs, N.J.: Prentice-Hall.

Page 40: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 359

Barens, Ingo. 1997. What Went Wrong with IS-LM/AS-AD—and Why? EasternEconomic Journal 23.1:89–99.

Barro, Robert J. 1990. Macroeconomics. 3rd ed. NewYork: Wiley.. 1994. The Aggregate-Supply/Aggregate-Demand Model. Eastern Eco-

nomic Journal 20 (winter): 1–6.Baumol,William J., andAlan Blinder. 1979. Economics: Principles and Policy. New

York: Harcourt Brace Jovanovich.. 1997. Economics: Principles and Policy. 7th ed. NewYork: Harcourt Brace

Jovanovich.Benham, Frederick. 1938. Economics: A General Introduction. London: Sir Isaac

Pitman.. 1955. Economics: A General Introduction. 5th ed. London: Sir Isaac Pit-

man.Bhaduri, Amit, K. Laski, and M. Riese. 1995. Making Sense of the Aggregate

Demand-Aggregate Supply Model. Working paper, Vienna Institute for Compar-ative Economic Studies.

Blanchard, Olivier J. 1997. Macroeconomics. Upper Saddle River, N.J.: PrenticeHall.

Blinder, Alan S. 1979. Economic Policy and the Great Stagflation. NewYork: Aca-demic Press.

Boulding, Kenneth E. 1948. Economic Analysis. Rev. ed. NewYork: Harper.Branson, William H. 1979. Macroeconomic Theory and Policy. 2d ed. New York:

Harper.Bronfenbrenner, Martin, Werner Sichel, and Wayland Gardner. 1984. Economics.

Boston: Houghton Mifflin.Brownlee, O. H. 1948. Money, Credit, and Monetary Policy. In Applied Economic

Analysis, edited by Francis M. Boddy. NewYork: Pitman.. 1950. The Theory of Employment and Stabilization Policy. Journal of Po-

litical Economy 58:412–24.Case, Karl E., and Ray C. Fair. 1996. Principles of Economics. 4th ed. Upper Saddle

River, N.J.: Prentice Hall.Clower, Robert W. 1989. Keynes’s General Theory: The Marshallian Connection.

In vol. 2 of Perspectives on the History of Economic Thought, edited by D. A.Walker. Aldershot, U.K.: Edward Elgar.

Colander, David. 1995. The Stories We Tell: A Reconsideration of AS/ADAnalysis.Journal of Economic Perspectives 9.3:169–88.

. 1998. Economics. 3d ed. Boston: Irwin/McGraw Hill.

. 2001. Economics. 4th ed. Boston: Irwin/McGraw Hill.Colander, David, and Peter Sephton. 1998. Acceptable and Unacceptable Dirty Ped-

agogy: The Case of ADAS. In Aggregate Demand and Supply: A Critique ofOrthodox Macroeconomic Modelling, edited by B. Bhaskara Rao. NewYork: St.Martin’s Press.

Page 41: Aggregate Supply Aggregate Demand Analysis a History

360 History of Political Economy 34:2 (2002)

Cottrell, Allin F. 1995. Comment. In New Perspectives on Keynes, edited by Allin F.Cottrell and Michael S. Lawlor. HOPE 27 (supplement): 217–22.

Dalziel, Paul. 1993. Classical and Keynesian Unemployment in a Simple Disequi-libriumAS-AD Framework. Australian Economic Papers 32:40–52.

Darity, WilliamA. Jr., andWarrenYoung. 1995. IS-LM: An Inquest. HOPE 27.1:1–41.

Davidson, Paul. 1972. Money and the Real World. 2d ed. London: Macmillan.Dillard, Dudley. 1948. The Economics of John Maynard Keynes. NewYork: Prentice

Hall.Dornbusch, Rudiger, and Stanley Fischer. 1978. Macroeconomics. New York: Mc-

Graw Hill.. 1994. Macroeconomics. 6th ed. NewYork: McGraw Hill.

Dutt, Amitava K. 1986–87. Wage Rigidity and Unemployment: The Simple Dia-grammatics of Two Views. Journal of Post Keynesian Economics 9.2.

. 1987. Keynes with a Perfectly Competitive Goods Market. Australian Eco-nomic Papers 26 (December): 275–93.

. 1992. Keynes, Market Forms, and Competition. In New Perspectives onKeynes, edited by Bill Gerrard and John Hillard, 129–48. Aldershot, U.K.: Ed-ward Elgar.

. 1997. On an Alleged Inconsistency in Aggregate-Supply/Aggregate-Demand Analysis. Eastern Economic Journal 23.4:469–76.

Dutt, Amitava K., and Edward J. Amadeo. 1990a. Keynes’s Dichotomy and Wage-Rigidity Keynesianism: A Puzzle in Keynesian Thought. In Keynes, Macroeco-nomics, and Method. Vol. 4 of Perspectives on the History of Economic Thought,edited by D. E. Moggridge. Aldershot, U.K.: Edward Elgar.

. 1990b. Keynes’s Third Alternative? The Neo-Ricardian Keynesians and thePost Keynesians. Aldershot, U.K.: Edward Elgar.

Elzinga, Kenneth G. 1992. The Eleven Principles of Economics. Southern EconomicJournal 58.4:861–79.

Fairchild, Fred Rogers, Edgar Stevenson Furniss, and Norman Sydney Buck. 1939.Elementary Economics. 4th ed. NewYork: Macmillan.

. 1948. Elementary Economics. 5th ed. NewYork: Macmillan.Fields, T. Windsor, and William R. Hart. 1990. Some Pitfalls in the Conventional

Treatment ofAggregate Demand. Southern Economic Journal 56 (January): 676–84.

Fischer, Stanley, Rudiger Dornbusch, and Richard Schmalensee. 1988. Economics.2d ed. NewYork: McGraw Hill.

Froyen, Richard T. 1993.Macroeconomics: Theories and Policies. 4th ed. NewYork:Macmillan.

Fusfeld, Daniel. 1988. Economics: Principles of Political Economy. Glenview, Ill.:Scott Foresman.

Garver, Frederic, and Alvin H. Hansen. 1937. Principles of Economics. 2d ed.Boston: Ginn.

Page 42: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 361

. 1947. Principles of Economics. 3d ed. Boston: Ginn.Gordon, Robert J. 1978. Macroeconomics. Boston: Little, Brown.

. 1981. Macroeconomics. 2d ed. Boston: Little, Brown.

. 1984. Macroeconomics. 3d ed. Boston: Little, Brown.Grieve, Roy H. 1998. The ADAS Model: Two into One Won’t Go. In Aggregate

Demand and Supply: A Critique of Orthodox Macroeconomic Modelling, editedby B. Bhaskara Rao. NewYork: St. Martin’s Press.

Guthrie, John A. 1957. Economics. Homewood, Ill.: Richard D. Irwin.Haberler, Gottfried. 1946. The Place of The General Theory of Employment, Inter-

est, and Money in the History of Economic Thought. Review of Economics andStatistics 28 (November): 187–94.

Hahn, Frank, and Robert M. Solow. 1995. A Critical Essay on Modern Macroeco-nomic Theory. Cambridge, Mass.: MIT Press.

Hall, Robert E., and John B. Taylor. 1991. Macroeconomics. 3d ed. NewYork: Nor-ton.

Hansen, Alvin H. 1949. Monetary Theory and Fiscal Policy. New York: McGrawHill.

. 1953. A Guide to Keynes. NewYork: McGraw Hill.Hansen, Bent. 1951. A Study in the Theory of Inflation. NewYork: Rinehart.Hansen, R. B., K. McCormick, and J. Rives. 1985. The Aggregate Demand Curve

and Its Proper Interpretation. Journal of Economic Education 16 (fall): 287–96.Heilbroner, Robert L., and James K. Galbraith. 1987. The Economic Problem. 8th

ed. Englewood Cliffs, N.J.: Prentice Hall.Hicks, John R. 1937. Mr. Keynes and the Classics: A Suggested Interpretation.

Econometrica 5 (April): 147–59.Kennedy, Paul. 1998. Defending ADAS: A Perspective on the ADAS Controversy.

In Aggregate Demand and Supply: A Critique of Orthodox Macroeconomic Mod-elling, edited by B. Bhaskara Rao. NewYork: St. Martin’s Press.

Keynes, J. M. 1936. The General Theory of Employment, Interest, and Money. Lon-don: Macmillan.

. 1940. How To Pay for the War. London: Macmillan.King, J. E. 1994. Aggregate Supply and Demand Analysis since Keynes. Journal of

Post Keynesian Economics 17.1:3–31.Kohn, Meir. 1981. A Loanable Funds Theory of Unemployment and Monetary Dis-

equilibrium. American Economic Review 71.5:859–79.Lindauer, John. 1968. Macroeconomics. NewYork: Wiley.

. 1977. Economics: A Modern View. Philadelphia: W. B. Saunders.Lipsey, Richard G., Peter O. Steiner, and Douglas D. Purvis. 1984. Economics. 7th

ed. NewYork: Harper.Mankiw, N. Gregory. 1992. Macroeconomics. NewYork: Worth.

. 1997. Principles of Economics. Fort Worth, Texas: Dryden.Mansfield, Edwin. 1992. Economics. 7th ed. NewYork: Norton.

Page 43: Aggregate Supply Aggregate Demand Analysis a History

362 History of Political Economy 34:2 (2002)

Marschak, Jacob. 1951. Income, Employment, and the Price Level. NewYork: Kel-ley.

McConnell, Campbell. 1960. Economics: Principles, Problems, and Policies. NewYork: McGraw Hill.

McConnell, Campbell R., and S. L. Brue. 1990. Economics: Principles, Problems,and Policies. 11th ed. NewYork: McGraw Hill.

. 1996. Economics: Principles, Problems, and Policies. 13th ed. NewYork:McGraw Hill.

McCormick, Ken, and Janet M. Rives. 1998. Aggregate Demand in Principles Text-books. In Aggregate Demand and Supply: A Critique of Orthodox Macroeco-nomic Modelling, edited by B. Bhaskara Rao. NewYork: St. Martin’s Press.

McKenna, Joseph P. 1955. Aggregate Economic Analysis. NewYork: Holt, Rinehartand Winston.

. 1965. Aggregate Economic Analysis. 2d ed. NewYork: Holt, Rinehart andWinston.

Meade, James. A. 1936. An Introduction to Economic Analysis and Policy. Oxford:Clarendon Press.

Meyer, Albert L. 1941. Elements of Modern Economics. 2d ed. NewYork: Prentice-Hall.

. 1948. Elements of Modern Economics. 3d ed. NewYork: Prentice-Hall.Miller, Roger Leroy. 1997. Economics Today: The Macro View. 9th ed. Reading,

Mass.: Addison-Wesley.Modigliani, Franco. 1944. Liquidity Preference and the Theory of Interest and

Money. Econometrica 12 (January): 45–88.Parkin, Michael. 1984. Macroeconomics. Englewood Cliffs, N. J.: Prentice-Hall.

. 1993. Economics. 2d ed. Reading, Mass.: Addison-Wesley.Patinkin, Don. 1949. Involuntary Unemployment and the Keynesian Supply Func-

tion. Economic Journal 59 (September): 360–83.. 1956. Money, Interest, and Prices. NewYork: Harper.. 1982. Anticipations of the General Theory? And Other Essays on Keynes.

Chicago: University of Chicago Press.Pearce, KerryA., and Kevin D. Hoover. 1995. After the Revolution: Paul Samuelson

and the Textbook Keynesian Model. In New Perspectives on Keynes, edited byAllin F. Cottrell and Michael S. Lawlor. HOPE 27 (supplement): 183–216.

Pigou, Arthur C. 1943. The Classical Stationary State. Economic Journal 53 (De-cember): 343–51.

Rabin, A. A., and D. Birch. 1982. A Clarification of the IS Curve and the AggregateDemand Curve. Journal of Macroeconomics 4 (spring): 233–38.

Rowan, D. C. 1968. Output, Inflation, and Growth: An Introduction to Macroeco-nomics. London: Macmillan.

Rowan, D. C., and Thomas Mayer. 1972. Intermediate Macroeconomics: Output,Inflation, and Growth. NewYork: Norton.

Page 44: Aggregate Supply Aggregate Demand Analysis a History

Dutt / AD-AS Analysis 363

Ruffin, Roy J., and Paul R. Gregory. 1997. Principles of Economics. 6th ed. Glen-view, Ill.: Scott Foresman.

Samuelson, PaulA. 1939. Interactions between the Multiplier Analysis and the Prin-ciple of Acceleration. Review of Economics and Statistics 21 (May): 75–78.

. 1948. Economics. NewYork: McGraw Hill.Samuelson, PaulA., andWilliamD. Nordhaus. 1985. Economics. 12th ed. NewYork:

McGraw Hill.. 1995. Economics. 15th ed. NewYork: McGraw Hill.

Stiglitz, Joseph. 1997. Economics. NewYork: Norton.Stonier, A. W., and D. C. Hague. 1953. A Textbook of Economic Theory. London:

Longman.Tarshis, Lorie. 1947. The Elements of Economics: An Introduction to the Theory of

Price and Employment. Boston: Houghton Mifflin.Tobin, James. 1993. Price Flexibility and Output Stability: An Old Keynesian View.

Journal of Economic Perspectives 7 (winter).Tresch, Richard W. 1994. Principles of Economics. Minneapolis-St. Paul: West.Weintraub, Sidney. 1951. Income and Employment Analysis. NewYork: Pitman.

. 1958. An Approach to the Theory of Income Distribution. Philadelphia:Chilton.

Wells, Paul. 1977. Keynes’s Disequilibrium Theory of Employment. InModern Eco-nomic Thought, edited by S.Weintraub. Philadelphia: University of PennsylvaniaPress.

Young, Warren. 1987. Interpreting Mr. Keynes: The IS-LM Enigma. Oxford: PolityPress.