If off-year elections were reliable referenda on great issues of economic policy, Reaganomics would lose on Nov. 2, despite the Republican appeal that voters "stay the course."

The president's program would get a vote of no-confidence, not because voters are nostalgic for Tip O'Neill's alternatives but because it has fallen well shy of its own goals and promises.

You don't have to be much of an economic theorist to grasp the nature of the failure. Across- the-board income-tax cuts generated no wonderful burst of productive investment or jobs.

Of course, no conventional economists expected that. There was no body of historical experience to support the expectation that a great give-back of disposable income to high-bracket taxpayers would cause a boom. Instead, the old Keynesian expectation, rashly scoffed at by "supply-side" economic theorists, held true. "Rentiers" -- those who get their money from investments -- do not lead booms in demand. And without demand, inventories pile up.

What the Reagan tax cuts have accomplished, when combined with rising defense budgets and fixed expenses in the social budget, is huge and growing federal deficits. The resulting pressure of federal borrowing has been such that the Federal Reserve Board was, until recently, reluctant to ease credit for fear of resurgent inflation.

The Fed's anti-inflation priority -- the right one, in my view, and the one officially approved at the White House -- has had a painful cost. The high cost of borrowing has driven the industrial slump to new depths and the unemployment rate to its highest point since before World War II.

The theoretical basis of Reaganomics, at least among its most publicized architects, was Prof. Arthur Laffer's famous "curve." It purported to show that there is an ideal relationship between tax rates and total revenues. But now another famous curve has returned to haunt the scene. It is the so-called Phillips curve, a concoction in the late 1950s of the British economist A. W. H. Phillips. It suggests a predictable connection between employment levels and inflation. It also contradicts the simple-minded belief that there are painless, even magical, ways to employ everyone who wants work without endangering price stability -- and, conversely, that you can stabilize prices without a cost in jobs.

I once heard the Phillips curve dismissed as "hot air," and maybe it is. But it is certainly no airier than the Laffer curve.

In gross terms, both curves express economic truisms. Laffer says that at some level of excessive taxation you so discourage taxable work that revenues fall. Phillips says that somewhere between zero and "full" employment, you risk generating excess demand and, with it, damaging inflation.

The trouble with both these elegant and enticing theories is that economic policy defies graphs. It is a matter of middle ranges and narrow choices, where sweeping theories are seldom helpful.

There was a time, in the mid- and late-1970s era of "stagflation" when the Phillips curve seemed invalidated by events. Surely, it was said, if Phillips had found anything like a "law" of economics, rising unemployment should be dampening inflation, though it wasn't.

Two years of Reaganomics have breathed new life into the Phillips curve. The return toward price stability seems to have been purchased only at the cost of a slump and rising joblessness.

The voters may for that reason repudiate Reaganomics on Nov. 2. But they need to remember, all the same, that the Phillips curve works both ways. If curing inflation is depressing, curing joblessness can be inflationary. No economist has yet invented the free lunch.