One of the oddities of debate in economics is that those who lose the war most often seem to win the peace. The famous victory of free market capitalism over communist central planning, for example, has simply given rise to a new generation of planning boosters in the free market states: not the ''bad'' planning that produces exploding TV sets and lineups for soap, but the ''good'' planning that carefully targets support to strategic, leading edge industries. The repudiation of the former serves only to vindicate the latter.

So it is with Keynesianism. Though it is often suggested Keynes is dead, his spirit lives on regardless, often in those keenest to pronounce his epitaph. This is most readily seen in those worthy editorialists who fret over the size of the federal deficit, on the grounds that this will constrain us from increasing it in the event of a recession.

Admittedly, this is unrepresentative of academic debate. Most ''neo- Keynesians'' now concede that deficits alone have no effect on economic growth, even in the short run. Either the deficit has to be financed through faster money growth - and hence inflation - or, if it is not money-financed, it drains capital markets and pushes up real interest rates - hence crowding out private investment. Or if it does not crowd out, it sucks in capital from abroad, and hence gives rise to balance of payments difficulties.

INSTRUMENT OF FINE-TUNING

The real fight, then, is not the old battle of fiscal vs monetary policy; it is whether monetary policy can or should be used as a discretionary instrument of ''fine-tuning,'' or should rather be held to a fixed-growth rule. Yet here the Keynesian ideal of demand management still holds sway. The yelping about high interest rates was bad enough when the economy was growing rapidly. As it appears it may now be starting to slow, we may expect the pressure to build on the Bank of Canada to loosen policy, amid accusations that it is deliberately engineering a recession.

Even the Bank's defenders typically speak of the need to slow the economy to prevent ''overheating.'' Both sides are united by the belief the central bank operates on inflation through the real side of the economy: growth, employment, real interest rates, real wages, as opposed to nominal variables like prices. And that is Keynes's legacy.

There is a certain irony in this. To ascribe almost total responsibility for output to monetary policy is the inverse of classic Keynesianism, in which financial markets were more or less an irrelevant sideshow to the real economy. But the essence of Keynesianism is not so much the choice of instruments, as it is the emphasis on aggregates: aggregate demand, total employment, and so on. Thus the policy response to a surplus of labor is not simply to increase the demand for labor, but to increase the demand for everything.

In the Keynesian view, growth, employment and inflation are not three separate questions, but one. Demand should sustain enough growth to lower unemployment, though not so much as to set off inflation; or at least, not too much inflation.

The whole aggregate model, in which growth, employment and inflation are but aspects of the same phenomenon of demand, can be analyzed into a causal chain. Monetary easing reduces interest rates; lower rates spur investment; more investment boosts growth; faster growth cuts unemployment; too-low unemployment sets off inflation. If any link fails, the model falls apart. In fact, the experience of recent years suggests every one of these links, if it ever was secure, has now been strained or broken.

Expanding the money supply cannot lower real rates if bondholders anticipate the inflation to come; investment may no longer be as sensitive to changes in interest rates, given changes in financial markets; growth in a service economy need not be as closely related to the rate of investment; rapid growth, we have seen, can have little or no impact on unemployment; and any amount of unemployment has been shown to be compatible with any amount of inflation.

The simple, mechanistic model of macroeconomics will have to be taken apart. We will have to treat each problem separately, with separate remedies. While inflation may be wholly a problem of aggregate demand, and growth partly so, unemployment must be seen for what it is: a mismatch of supply and demand in the labor market.

Yet everyone proceeds as if none of this had happened, as if all were aggregates. So we live in thrall, as Keynes observed, slaves to a defunct economist.