Thinking about Welfare's Inefficiencies

There's a widespread popular belief that welfare provides for people better than a lot of minimum wage working. I don't deny the distorting effects that social cushioning can provide to a low-paid worker on the margin, but these distorting effects are probably much less than most people think:
http://www.nytimes.com/2007/05/01/us/01stamps.html?_r=2&scp=3&sq=Kulongoski&st=nyt

In Oregon, food stamps from welfare equal $21 per week to eat on.

However, everyone hears stories like this:

http://www.rantblogger.com/food-stamps-shrimp-and-lobster/#more-1361

This is allegedly a checkout person at a retail store who gets furious at the myriads of foodstamp recipients he sees buying "Delmonaco steaks and lobsters."

Talk to most people, and the amount of the food stamps is usually what they talk about. Liberals will typically talk for a long time about how inadequate welfare checks are for the people who need to survive on them, while conservatives will usually talk about how individuals on welfare receive more than many blue collar workers, and they shouldn't need so much help from the government and should get a job themselves.

What if (and I'm hoping someone with access to data can send it to me) the problem isn't the amount, but enforcement of the qualification threshold? Conservatives are almost certainly wrong in universally stating that people on welfare could do better for themselves if they tried; certainly, there are many cases of people who ARE trying and still aren't making it. However, liberals may not be fair in implying that every worker on welfare is suffering and needs welfare to survive. It is quite plausible that some of the recipients of welfare don't actually need it, and therefore use it to "cheat the system". This explains some of the anecdotal lavish purchases on welfare without making incorrect blanket statements about welfare recipients.

Therefore, perhaps qualifications for welfare are a better focus for people concerned about social spending. Better qualifications creates a more efficient system, with better efficacy per dollar spent. Individuals who need the help could receive more money, or at least it would cost less to give them the same amount. In a world with substantial opportunity costs of government spending in individual areas, that's an important goal.

On a side note, we can create a structural, instead of emotional or anecdotal, system of thinking about welfare, to make decisions and opinions about it more rational. If one gives both liberals and conservatives the benefit of the doubt and assume the most sophisticated version of their arguments, the liberal vs conservative argument for/against welfare is one of type I vs type II error. Both liberals and conservatives who are non-naive are forced to acknowledge that there are 4 kinds of people:

1) people who need welfare who qualify for welfare
2) people who need welfare who don't qualify for welfare
3) people who don't need welfare who qualify for welfare
4) people who don't need welfare who don't qualify for welfare

The relative sizes of these groups is up for debate, as is the severity of Type I errors (a large number of people in group 2) vs Type II errors (a large number of people in group 3). While intuition states that a large group 3 is less severe than a large group 2, this ignores the notion of opportunity cost; you may be better off with a large group 2 and using all the money you save to help people who fall in that group in different ways.

It seems to me that liberals misclassify members of group 3 as group 1, and therefore overemphasize group 2. Therefore, they try and minimize group 2 first, and then given the constraints imposed by minimizing 2, do their best to minimize group 3. Conservatives misclassify members of group 2 as group 4, and therefore overemphasize group 3. They try and minimize group 3 first, and then given the constraints imposed by minimizing 3, do their best to minimize 2.

It may be more sensible in a world with opportunity costs to try and minimize them jointly. Presented in the manner I just did, that seems to be a common sense middle ground. However, nobody on either side seems to be advocating for that; instead, we hear anecdotes and evidence about how we need to push further in one direction or the other.

Another decent case for mandating philosophy, economics and statistics for all college students; at least then, educated people will have some background in structural thinking and data analysis. Anecdote-based and evidence-based arguments are very important and certainly better than nothing, but they are more prominent in policymaking than they deserve to be relative to structural thinking. The dogmatic treatment of welfare by liberals and conservatives serves as a decent example of that.