Raj Chetty on Taxation
Chetty's one of Harvard's young superstar economists. Mankiw posted this on his blog, and I've been asked about it. What Chetty found is that in America in the range of taxes we have seen in the last 50 years, an X% increase in overall tax rates (of any kind) reduce the amount of pre-tax money earned by Americans by 0.5X%.
So if the government increases taxes by 10%, people (on a macro scale) reduce the amount of work they do, and thus the amount they earn pre-tax, by 5%. The reason for this is that if you're choosing between work and free time, and you make the return to an hour of work lower but you don't change how much fun you can have with your free time, you will switch to less work and more free time.
The implication of this is that on the margin, unless the government can create twice the impact of spending than individuals, raising taxes hurts the country. Some initiatives are clearly worth it (having a police department or a fire department or a military), some are clearly not (pork barrel spending), and many are arguable. However, given the variety of highly inefficient ways the government spends, it would be hard to argue that the current level of government spending shouldn't be reduced.