Regarding the use of government data to demonstrate causality...
It's incredibly difficult, even impossible, to isolate the factors that contribute to any given result.
For example, suppose we're using government data to establish that raising the minimum wage leads to a rise in unemployment. Let's also suppose that when we graph those two events over time, we find that their respective trajectories are positively correlated.
Here's the problem: there are many unknown factors that are affecting unemployment. For that reason, proving causality with the data at hand is virtually impossible. And as a result, any argument based on the assumption that one event caused another event becomes vulnerable to debate.
Here's a hypothetical example...
Let's say someone produced a chart showing that a rise in the minimum wage was (surprise!) actually associated with a decline in unemployment over time. What are some of the factors that may have contributed to the decline that haven't been taken into account?
Is it possible that employers might receive subsidies for hiring select groups, thereby artificially suppressing the unemployment rate? Of course (see link below):
Connecticut Employers Can Receive Subsidies for Hiring Unemployed Veterans | Climbing Back
Is it possible that the federal government might tighten affirmative action laws to the same effect? (Note that such laws extend far beyond juicing staffing percentages.) Of course (see link below):
Uncle Sam Wants You ... to Hire More Vets, Women, and Minorities - Businessweek
Therein lies the problem with using data - i.e. empirical "evidence" - to demonstrate causality. It has been mentioned many times on WF. (For example, the fact that a rooster crows as the sun rises obviously doesn't mean the rooster's crowing has caused the sun to rise. Other factors are in play.)
But this same mistake - assuming causality - is made continuously when people discuss the economic effect of any given event.
Years ago, Hans Hoppe wrote about the need to approach economics as an a priori science rather than one based on empiricism. The link is below:
Austrian Method, Praxeology I
Here's a short video in which Hoppe discusses the same issue:
[ame=http://www.youtube.com/watch?v=tx7XUuPrbZo]The A priori of Argumention - YouTube[/ame]
There's another problem with demonstrating causality. Even if we can do so - and again, doing so is virtually impossible because of what we do not know - the long-term effects of any given event are usually ignored. Hazlitt mentioned this problem in the first chapter of his book Economics in One Lesson (see my previous post for a link to that book). He noted the following:
In addition to these endless pleadings of self-interest, there is a second main factor that spawns new economic fallacies every day. This is the persistent tendency of men to see only the immediate effects of a given policy, or its effects only on a special group, and to neglect to inquire what the long-run effects of that policy will be not only on that special group but on all groups. It is the fallacy of overlooking secondary consequences.
The problems outlined all but ensure that no one will give ground. The reason? Because everything based on empirical data is open to debate. The application of logic to prove causality is a much purer approach. But it's also one that is doomed to fail here.
Will raising the minimum wage to $15/hr. cause a long-term rise in employment within our current system - that is, one based on legislation and taxation? Impossible to know and impossible to prove. There are too many factors in play and too many fingers on buttons that manipulate those factors.