I have never been a big advocate of using fiscal policy to stimulate the economy in the short run. It doesn't work quickly enough when you need it, and it doesn't go away quickly enough when you don't need it. Trying to stimulate the economy through increased spending on infrastructure -- what once was called "public works" -- takes too long. At any given point in time, there aren't a lot of "shovel-ready" projects. Perhaps worst of all, spending increases tend to become permanent. Tax cuts aren't much better. Temporary tax cuts don't work; and with $19 trillion in federal government debt and a $500 billion deficit, we can't afford permanent tax cuts. I favor reforming the tax system to maximize long-term growth; I don't favor using it for short-term stimulus.
That said, avoiding bad macroeconomic policies is not a good reason for avoiding good microeconomic decisions. Infrastructure projects need to be evaluated on their individual merits using the same kind of net present value calculation that a profit-maximizing company would use. Other things equal, projects will be more attractive (and more likely to merit approval) when interest rates and the costs of labor and materials are low than when they are high. Consequently, spending on infrastructure projects should go up when the economy is weak, even if such projects are not explicitly done as part of a macroeconomic stimulus program.
Requirements that state (and local) governments balance their budgets might keep them from expanding infrastructure spending as much as proper investment evaluation suggests. If state and local governments can't borrow the money needed to make these investments, there may be a role for the federal government to borrow on their behalf, but this borrowing should be explicitly linked to these projects. Borrowing to fund investments with positive net present values makes sense; borrowing to fund a temporary tax cut or a larger increase in transfer payments than that provided under current law makes no sense at all.
The case for infrastructure spending can be strengthened and its macroeconomic impact enhanced by suspending or repealing prevailing-wage laws. For a weak economy to return to full employment, wages must fall to market-clearing levels. Holding wages paid to workers on government infrastructure projects at the level that prevailed before the economy slowed prevents that. And if your goal is putting the unemployed back to work, it makes more sense to employ four workers at $15/hour than to employ three at $20/hour.
I used to think that "macro" economics meant "big" economics; I now think it means "bad" economics. Many of my fellow macroeconomists forget what they learned in their microeconomics classes and believe things that simply make no sense. The best example of this is the Phillips Curve, which holds that there is an inverse relationship between unemployment and inflation. Those of us who remember our microeconomics know that low unemployment should raise real wages, not nominal prices. Yet the Phillips Curve remains an important part of most macroeconomic models and the current thinking of the Federal Reserve. More to the point of the current discussion, focusing on macroeconomic stimulus causes Keynesians to favor infrastructure projects that don't make economic sense and causes anti-Keynesians, in a knee-jerk fashion, to oppose projects that do make sense.
Those who expect an increase infrastructure spending to quickly boost the economy are likely to be disappointed, but that's not the point. A project that has a positive net present value, calculated using an appropriate discount rate, should be undertaken and will boost the economy in the long run. That bit of microeconomic theory holds true regardless of what your macroeconomic idealogy is telling you.