As a conservative economist, I might be expected to oppose a stimulus plan. In fact, in October I declared my support for a stimulus. But the fiscal package now before Congress needs to be thoroughly revised. In its current form, it does too little to raise national spending and employment. It would be better for the Senate to delay legislation for a month, or even two, if that’s what it takes to produce a much better bill. We cannot afford an $800 billion mistake.

Start with the tax side. The plan is to give a tax cut of $500 a year for two years to each employed person. That’s not a good way to increase consumer spending. Experience shows that the money from such temporary, lump-sum tax cuts is largely saved or used to pay down debt. Only about 15 percent of last year’s tax rebates led to additional spending.

The proposed business tax cuts are also likely to do little to increase business investment and employment. The extended loss “carrybacks” are primarily lump-sum payments to selected companies. The bonus depreciation plan would do little to raise capital spending in the current environment of weak demand because the tax benefits in the early years would be recaptured later.

Instead, the tax changes should focus on providing incentives to households and businesses to increase current spending. Why not a temporary refundable tax credit to households that purchase cars or other major consumer durables, analogous to the investment tax credit for businesses? Or a temporary tax credit for home improvements? In that way, the same total tax reduction could produce much more spending and employment.

Postponing the scheduled increase in the tax on dividends and capital gains would raise share prices, leading to increased consumer spending and, by lowering the cost of capital, more business investment.

On the spending side, the stimulus package is full of well-intended items that, unfortunately, are not likely to do much for employment. Computerizing the medical records of every American over the next five years is desirable, but it is not a cost-effective way to create jobs. Has anyone gone through the (long) list of proposed appropriations and asked how many jobs each would create per dollar of increased national debt?

The largest proposed outlays amount to just writing unrestricted checks to state governments. Nearly $100 billion would result from increasing the “Medicaid matching rate,” a technique for reducing states’ Medicaid costs to free up state money for spending on anything governors and state legislators want. An additional $80 billion would be given out for “state fiscal relief.” Will these vast sums actually lead to additional spending, or will they merely finance state transfer payments or relieve state governments of the need for temporary tax hikes or bond issues?

The plan to finance health insurance premiums for the unemployed would actually increase unemployment by giving employers an incentive to lay off workers rather than pay health premiums during a time of weak demand. And this supposedly two-year program would create a precedent that could be hard to reverse.

A large fraction of the stimulus proposal is devoted to infrastructure projects that will spend out very slowly, not with the speed needed to help the economy in 2009 and 2010. The Congressional Budget Office estimates that less than one-fifth of the $50 billion of proposed spending on energy and water would occur by the end of 2010.

If rapid spending on things that need to be done is a criterion of choice, the plan should include higher defense outlays, including replacing and repairing supplies and equipment, needed after five years of fighting. The military can increase its level of procurement very rapidly. Yet the proposed spending plan includes less than $5 billion for defense, only about one-half of 1 percent of the total package.

Infrastructure spending on domestic military bases can also proceed more rapidly than infrastructure spending in the civilian economy. And military procurement overwhelmingly involves American-made products. Since much of this military spending will have to be done eventually, it makes sense to do it now, when there is substantial excess capacity in the manufacturing sector. In addition, a temporary increase in military recruiting and training would reduce unemployment directly, create a more skilled civilian workforce and expand the military reserves.

All new spending and tax changes should have explicit time limits that prevent ever-increasing additions to the national debt. Similarly, spending programs should not create political dynamics that will make them hard to end.

The problem with the current stimulus plan is not that it is too big but that it delivers too little extra employment and income for such a large fiscal deficit. It is worth taking the time to get it right.

Martin Feldstein, an economics professor at Harvard University, is president emeritus of the National Bureau of Economic Research.