Recent work on the microfoundations of macroeconomics had drawn heavily on the concept of imperfect competition and a very promising branch of this work consists of models which derive "Keynesian" or other types of macroeconomic results from microeconomic specifications which feature imperfect competition, as in the "Keynesian features" model due to Hart. Snower, Benassy and Dixon have also analysed models featuring imperfect competition using different specifications, but these have failed to reproduce Hart's results.

In this paper we re-examine a simpler version of Hart's original model. We introduce fiscal and monetary policy into the model, allowing us to elucidate the conditions under which monetary and fiscal policy have effects on real output in models of this class. This allows us to examine how differences in the model's specification affect the relationships between policy instruments and output, prices, and other variables of interest.

We assume that the supply of labour is fixed and completely inelastic, which is to say there is no "disutility" of work. This means there is an unambiguous "full employment" level of employment and output in the model and, under perfect competition, this will always be achieved. Imperfect competition is present in the model because of the presence of trade unions, which seek to maximize the total wage revenue of their members. There is no mystery about the unemployment which may result: it is due simply to monopolistic restriction by the sellers of labour. As such, it is "voluntary" from the viewpoint of a trade union, but "involuntary" from the viewpoint of an individual worker, who would always prefer to work more at the market real wage.

We distinguish between policy "ineffectiveness" and policy "neutrality". By "ineffectiveness" we mean the inability to influence real output, while "neutrality" is stronger and denotes the inability to influence any real variable. Only monetary policy can be "neutral" in this sense. Fiscal policy, which here refers to changes in government spending, cannot be "neutral", since a change in real government spending is itself a "real" change and must have some "real" consequences. Insofar as we find conditions under which monetary and fiscal policy are effective (i.e. can influence real output), we also find conditions under which there is no natural rate of unemployment (or of output or employment) since government can change this rate by changing its policy stance.

We examine policy effectiveness by comparing the Walrasian (perfect competition, market-clearing) equilibrium of the model with the equilibrium which arises under the assumption of oligopolistic behaviour. For monetary policy to be effective, then in the Walrasian version of the model, policy should already be non-neutral, i.e. monetary policy must affect some real variable in the Walrasian world if it is to affect output in the imperfectly competitive world. In other words, two deviations from the classical, Walrasian world are necessary: imperfect competition is not enough by itself.
In particular, if expected future prices are not unit-elastic with respect to current prices (i.e. a change in current prices does not lead to an equal percentage change in future prices), then monetary policy will be effective in our "Hartian" model. In Walrasian equilibrium, non-unit-elastic expectations cause a rise in the money supply to produce a rise in the price level which is not exactly proportional, but do not cause it to affect output, which is fixed at full employment. When combined with unemployment-creating imperfect competition, however, they ensure that monetary policy does affect output. Hart's model, appropriately reinterpreted, represents the extreme case where the price level becomes totally independent of the money supply, so that the latter affects output in a familiar Keynesian manner. Under these circumstances, monetary policy is non-neutral in the Walrasian and effective in the imperfectly competitive model.

We also examine fiscal policy and show that it will almost always affect output if there is unemployment, whatever the effectiveness of monetary policy. This justifies the conclusion that under unemployment-creating imperfect competition there is no natural rate. The most clearly "Keynesian" behaviour occurs when there are non-unit-elastic price expectations, causing the current price to become (either completely or partially) rigid, and causing monetary policy as well as fiscal policy to be effective. In the present model non-unit-elasticity, as a hypothesis considered in itself, is neither more nor less plausible than its alternative. Future work might usefully try to embed non-unit-elastic expectations in a model where they could prove to be "rational".