Keynesian theory and AS/AD analysis

Keynesian theory and AS/AD analysis

Palley, Thomas I


In a recent paper, Dutt and Skott [1996] have provided a defense of the aggregate supply/aggregate demand (AS/AD) framework. They focus on the usefulness of the AS/AD framework for comparing alternative macroeconomic perspectives. This paper further investigates the AS/AD framework, and focuses on the different short-run adjustment dynamics implied by different perspectives.

Dutt and Skott also maintain that the neo-Keynesian (also termed neo-classical synthesis) and Kaleckian models represent different perspectives on macroeconomics. The current paper argues that the Kaleckian model is nested within the neoKeynesian model, and shares the same ouput dynamics. The Kaleckian model’s adoption of a constant marginal product of labor (MPL) is a special case of the neo-Keynesian model, as is the claim that nominal wage levels are irrelevant for macroeconomic outcomes.

The significant contribution of the Kaleckian paradigm is to emphasize differences in the propensity to consume out of wage and profit income. This pushes distributional concerns to the fore of macroeconomics, and highlights the significance of capital-labor conflict. However, these distributional concerns can be incorporated readily within the neo-Keynesian model. Since the Kaleckian model has the same analytical structure, output dynamics and equilibrium closure as the neo-Keynesian model, it is argued that the Kaleckian model should be considered a part of the Keynesian family of models rather than a separate theoretical entity.


Before turning to the issue of AS/AD adjustment dynamics, it is worth excavating the foundations of the AD schedule. This is a point that has recently been raised by Colander (1995), who argues that the AS/AD model promotes a conflation of the “AD function” and the “AD schedule”. The former provides a relation between price and quantity demanded that is predicated on a given output; the latter is a goods market clearing schedule which determines the price at which quantity demanded is equal to the level of output.

It therefore establishes the price level at which quantity demanded equals output. The AD “function,” which is drawn in price-quantity demanded space, shifts as output changes. It is the macroeconomic analogue of the Marshallian microeconomic demand function. The AD “schedule” is invariant to changes in output, and provides a locus of goods market clearing price-output combinations that is conditional on the exogenous factors affecting the AD function.

The AD function is invisible in the AS/AD diagram. This is because the system involves three endogenous variables — price, output, and quantity demanded – so that one must be dropped in a two-dimensional diagram.


Despite the problem of being restricted to two dimensions, the AS/AD framework is still highly effective for revealing the different short-run disequilibrium price-output dynamics implied by different perspectives. Figure 1 shows a conventional AS/ AD diagram, in which Eo corresponds to the initial equilibrium. The AD schedule yields the level of prices that would clear the goods market contingent on the existing configuration of factors affecting aggregate demand and the existing level of output. The AS schedule corresponds to the economy-wide marginal cost schedule, which is contingent on the level of nominal wages and the level of technology. These two schedules are given by

A positive autonomous demand shock shifts the entire AD schedule right. In a new classical world, prices and output are instantaneously flexible, and the economy immediately moves to a new short period equilibrium at E ^ sub 1^.

Figure 2 shows the conventional neo-Keynesian fix-price construction of output dynamics in which prices are a state variable that are fixed at any moment in time, but output is perfectly flexible. This construction corresponds to the fix-price general disequilibrium Keynesianism developed by Barro and Grossman [1971]. Output immediately jumps to E ^ sub 0^^sup 1^ , yielding a full multiplier effect. At this stage, prices are below marginal cost. Firms therefore begin reducing output, and the economy then slides up the AD schedule to E ^ sub 1^. However, this raises questions as to why profit maximizing firms initially increase production, given that prices are below marginal cost.

Figure 3 shows an alternative neo-Keynesian approach to adjustment which is predicated on the assumptions that prices are instantaneously flexible, but output is a state variable that adjusts sluggishly. Consequently, prices immediately increase to clear the goods market, and the economy moves to E ^ sub 0^^sup 1^ . The logic is that with unchanged output and increased aggregate demand, prices must rise to clear the goods market. At E ^ sub 0^^sup 1^ , prices exceed marginal cost as determined by the AS schedule, and firms therefore have an incentive to increase output. Thereafter, the economy slides down the AD schedule until it reaches E ^ sub 1^. Palley [1996, Chapter 4] provides a full account of these dynamics, and also provides microeconomic foundations for why dynamics of flexible prices and sluggish output adjustment are to be preferred. The existence of production scheduling costs and the fact that production is a time-consuming process both suggest that output is likely to be a sluggish variable, whereas instantaneous output adjustment and sluggish price adjustment results in a violation of profit maximizing behavior since firms produce in a region where prices are below marginal costs.

A final possibility, illustrated in Figure 4, is that prices and output are both sluggish. In this case, the economy slides along the AS schedule until it reaches the new equilibrium at E ^ sub 1^. Once again, this description is microeconomically problematic, since there is persistent unsatisfied excess demand along the adjustment path. This is because the market clearing price, as given by the AD schedule, lies above the price as determined by the AS schedule for the entire adjustment path.


The dynamics shown in Figures 2 and 3 are both consistent with the Kaleckian paradigm. Indeed, the Kaleckian model with its horizontal AS schedule has historically been represented in terms of Figure 2. Given this equivalence of dynamics, what distinguishes the neo-Keynesian and Kaleckian models?

One principle difference concerns the specification of the AD function. The neoKeynesian model assumes that propensity to consume out of wage and profit income is identical, whereas the Kaleckian model assumes that the propensity to consume out of wage income exceeds that of out of profit income. The neo-Keynesian AD function is therefore given by

The important implication is that the distribution of income between profits and wages now matters for the level of aggregate demand as wage and profit income enter as separate functional arguments.

The Kaleckian specification can be justified on a number of different grounds. First, households may adopt behavioral rules of thumb, and consume wage income while saving interest and profit income. A second “institutional” justification is that much saving is done through pension funds which automatically accumulate interest and profit income. A third “sociological” justification relies on a distinction between capitalist and worker households, the claim being that capitalist households have higher incomes and a lower marginal propensity to consume. Behind this claim lies the existence of a conventional Keynesian consumption function in which the marginal propensity to consume falls with income.

The standard neo-Keynesian specification rests on the claim that income is fungible, agents are rational, and all households have the same propensity to consume. Income fungibility implies that the source of income is irrelevant for the consumption decision, while rationality implies that agents recognize that pension funds are saving on their behalf and correspondingly reduce their own saving. Simultaneously, the permanent income theory of consumption implies that all households have the same propensity to consume: though levels of income may differ across households/classes, the propensity to consume does not, and this means the functional distribution of income has no implications for aggregate demand.

More refined neo-Keynesian models admit the possibility of distributional effects. One argument is that low income worker households are liquidity constrained and unable to borrow against future income. In this case, a shift in the distribution of income towards wages will increase aggregate demand by relaxing liquidity constraints. A second argument is that there are life cycle differences in consumption patterns. In this event, if older households have a lower propensity to consume and ownership is concentrated amongst these households, then the functional distribution of income will again impact aggregate demand.

A second distinction between the neo-Keynesian and Kaleckian models is the latter’s adoption of a constant MPL. However, this is a simplifying assumption rather than a necessary feature. To see this, consider the following model:

This last specification corresponds to the standard neo-Keynsian specification of the AS schedule.

These considerations reveal that the neo-Keynesian representation of the AS schedule as positively sloped is fully compatible with the Kaleckian model, and the assumption of constant MPL is a special case. It also shows that the Kaleckian nominal wage irrelevance theorem is unrelated to the behavior of the MPL. Instead, it rests on a particular sepecification of the AD function that excludes nominal wealth effects (about which more below), and the identical nominal wage irrelevance theorem applies to the neo-Keynesian model if the AD function excludes nominal wealth effects.

For short-run macroeconomic analysis, the differences between the neo-Keynesian and Kaleckian approaches therefore reduce to difference over (i) the degree of emphasis on and the particular justification for the functional distribution of income affecting aggregate demand, and (ii) the microeconomics of pricing. These are differences of “degree”. The Keynesian model is fully capable of incorporating distributional concerns through appropriate specification of the AD function, while the Kaleckian notion of the mark-up can be viewed as the incorporation of imperfect competition considerations into the macro model.

The substantive analytic contribution of the Kaleckian model is its implied focus on capital-labor conflict which follows from its “particular” construction of the macroeconomic effects of income distribution. However, even here the output effects are ambiguous since higher mark-ups will raise profits, and if investment responds strongly to profitability (i.e. investment is not exogenously determined), then AD and output could actually increase. This possibility has been identified by Bhaduri and Marglin [1990] with their distinction between “exhilarationist” and “stagnationist” regimes.


The previous section showed that the nominal wage irrelevance theorem rests on neither income distribution effects nor the presence or absence of diminishing returns to labor. Instead, it rests on the absence of nominal wealth effects on aggregate demand.

The conventional neo-Keynesian model includes a Pigou effect in the AD function. Thus, AD depends positively on real money balances, and a reduction in the price level brought about by a reduction in nominal wages will increase AD. Clearly, the same argument holds if a Pigou effect is included in the AD function of the Kaleckian model.

Some neo-Keynesians [Tobin, 1980] have argued that the Pigou effect may be dominated by the Fisher debt effect, and that price and nominal wage reductions reduce AD. The argument is that lower prices increase the wealth of creditors, but also increase the burden of debts on debtors. If debtors have a higher marginal propensity to consume than creditors, price level reductions may reduce aggregate demand.

This emphasis on the distinction between debtors and creditors complements the Kaleckian distinction between wage and profit income, which is itself frequently represented in terms of a distinction between worker and capitalist households. Indeed, since workers tend to be net debtors, the two sets are likely to be highly overlapping. The implications of including the Pigou and Fisher effects in a Kaleckian model can be readily seen from the following model.

***formula text omitted***

If the Fisher effect dominates the Pigou effect, the AD schedule shifts down to AD2, and output falls to Y ^ sub 2^. It is this type of outcome that leads Post Keynesians to challenge the conventional claim that Keynesian involuntary unemployment is the result of downward nominal wage rigidity, and would be cured by lower nominal wages [Palley, 1996].

Figures 5-8 underscore the usefulness of the AS/AD framework for analyzing competing representations of the macro economy. Figure 8 also shows that inclusion of the Pigou and Fisher effects means that nominal wages are no longer irrelevant for the Kaleckian model. If the Pigou effect dominates, nominal wage reductions will increase employment. The exact same economic logic and conclusions applies in the textbook neo-Keynesian model, the only difference between the two being that b ^ sub 0^ = b ^ sub 1^.


This paper has shown how the AS/AD framework can be used to illustrate the competing approaches to (i) the dynamics of short run price – output adjustment, and (ii) the employment and output effects of nominal wage reductions. The paper has also argued that the Kaleckian model represents a variation on the neo-Keynesian model, embodying the same equilibrium concept, the same pattern of output dynamics, and the same comparative statics in response to changes in the nominal wage. Nominal wages affect equilibrium output and employment if AD is subject to the Pigou and Fisher effects, and not otherwise. The Kaleckian model assumes a constant MPL, but this is merely a simplifying assumption. Once removed, the similarity between the neo-Keynesian and Kaleckian models is rendered transparent.

The principal contribution of the Kaleckian approach is a concern with the AD effects of the functional distribution of income. This introduces the mark-up as an important independent variable, and focuses attention on capital – labor conflict. This feature is an original and important contribution, and ultimately leads to a different theory of income distribution. However, when it comes to AS/AD analysis, the Kaleckian model is encompassed within the neo-Keynesian model.


1. **formula text omitted***

2.This corresponds to the standard Keynesian assumption that the marginal propensity to spend is less than unity. Thus, when output increases by one unit, AD increases by less than one unit, thereby necessitating a fall in prices to clear the goods market.


Barro, R., and Grossman, H. A General Disequilibrium Model of Income and Employment. American Economic Review 61, 1971, 82-93.

Bhaduri, A., and Marglin, S. Unemployment and the Real Wage: the Economic Basis of Competing Ideologies. Cambridge Journal of Economics 14, 1990, 375-95. Colander, D. The Stories We Tell: A Reconsideration of AS/AD Analysis. Journal of Economic Perspectives, Summer 1995, 169-88.

Dutt, A., and Skott, P. Keynesian Theory and the Aggregate Supply/Aggregate Demand Framework: A Defense. Eastern Economic Journal, Summer 1996, 313 -31. Palley, T.I. Post Keynesian Macroeconomics: Debt, Distribution and the Macroeconomy, New York: St.Martin’s Press, 1996.

Tobin, J. Asset Accumulation and Economic Activity, Oxford: Blackwell, 1980.

Thomas I. Palley


Copyright Eastern Economic Association Fall 1997

Provided by ProQuest Information and Learning Company. All rights Reserved