Ross Douthat's column praising the budget policies of my adopted state of Texas has been pretty thoroughly critiqued. But I thought Ezra Klein's chart captured the essence of the problem with it. The upshot is that Texas's budget is in relatively better shape because there's not as much spending on social services.
It's worth remembering this chart during discussions of the "costs" of health coverage reform. By spending less, we're not really lowering costs so much as we are reallocating costs.
If health coverage reform fails, it's not like costs will go away. Instead, the costs will be borne by the uninsured, the underinsured, and people with rising premiums. And that's all before we consider the more intangible costs of fear and anxiety, along with the frustration (and social loss) of being stuck at a job you don't like in order to keep benefits.
The goal of reform is rationalize this system, and spread the costs around in a more efficient (and moral) way. (Hopefully, we'll actually lower total costs too). Things like the surtax shift these costs on to the people most able to bear them. Scaling back subsidies, by contrast, sounds "fiscally responsible" until you realize that all you're doing is reallocating costs to the working poor.
It's a bit more abstract, but you can also think of anti-poverty measures in these terms. Not spending money to eliminate poverty admittedly saves the government money. But it doesn't really reduce "costs." As Fallows explains (by quoting the Economist), there are costs to inaction on this front too, though some of them won't appear for a while:
[The Economist] ended up much more equivocal about the new Texas supremacy. Eg, "To begin with, that lean Texan model has its own problems. It has not invested enough in education, and many experts rightly worry about a 'lost generation' of mostly Hispanic Texans with insufficient skills for the demands of the knowledge economy."
And government spending can help this problem. In fact, it already has.