This is a very illuminating article on GDP numbers and shows how their very calculation skews the numbers toward more government spending:
Yet even this correction implicitly assumes that government spending is the source of all recovery. The logic, as with Bernanke's and Zandi's analyses, is that government spending cuts reduce overall demand in the economy, which affects growth and then employment. This argument ignores the fact that the government has to take its money out of the economy by raising taxes, borrowing from investors, or printing dollars. Each of these options can shrink the economy.
All these analysts also systematically ignore the fact that GDP numbers include government spending. When the federal government pumps trillions of dollars into the economy, it looks as if GDP is growing. When government cuts spending—even cuts within the most inefficient programs—aggregate GDP shrinks.
But that's misleading. If Washington spends $1 a year on a bureaucrat's salary, for example, GDP numbers will register growth of exactly $1, whether or not the employee has produced any value for that money. By contrast, if a firm pays an engineer $1, that $1 only shows up in the GDP if the engineer produces $1 worth of stuff to sell. This distinction biases GDP numbers—and the policies based on them—toward ever-increasing government spending.