Stimulus is dangerous. The second-order costs of government spending are real, and we are very far from being able to understand or estimate them. Here are some second order costs:

1. Transfers of relative purchasing power from other citizens to the beneficiaries of government spending may call into question the legitimacy of the distribution of opportunity, wealth, and influence and of the government itself. Perceptions of make-work or corrupt contracting are deeply corrosive. Deficit spending commits government to future transfers that may come to seem undesirable or illegitimate.

2. Government spending choices may lead to lower quality uses of real resources than would have occurred if the government had not acted. Since economic activity is habit forming and temporary interventions become permanent, the cost of poor government choices can be high. It matters very much what work the government is paying for. Work must be well-tailored to the talents, interests, and future prospects of individuals. Employing people badly is much worse than just giving them money.

3. If funds are spent, directly or indirectly, on resources in scarce supply, prices may be harmfully propped or bid up. That might take the form of a general inflation, or a narrower effect on the prices of specific commodities or assets.

4. High levels of government debt may have a destabilizing effect on prices, increasing price volatility and impairing economic calculation even in the absence of a general inflation, or even in a deflation. Government obligations are liquid and hypothecable, and the availability of good collateral increases the degree to which subjective changes in relative valuation translate to changes in nominal pricing.

5. There exist theories of government solvency which suggest that the safety and value of currency is related to the indebtedness of the issuing government. Those theories may or may not be reasonable. They may or may not find support in the historical record. Regardless, to the degree they are widespread, they may be self-fulfilling. Whether sensible or sunspot, loss of confidence in a currency is possible. Currency crises represent a “tail risk” whose likelihood and cost are difficult to estimate.

There are second order benefits to stimulus as well as costs: multipliers, consumer confidence, etc. But these are also difficult to estimate.

This is from an author who affirms there are things the government should be doing, and describes his own view as skewing "activist". Not exactly the un-thinking deficit-hawk that Krugman and others like to hold up for scrutiny just so they can knock them down.