Do Unemployment Benefits Increase Unemployment?

Arthur Laffer thinks so. In yesterday's WSJ editorial page, he makes the case. It's mostly an argument based on the commonly stated idea that unemployment benefits discourage people from working, but he also tries to provide some real data to show this phenomenon happening in some place besides his own head. That's when things really fall apart:
He uses the above graph to argue "correlation", but his argument really hinges on the idea that rises in unemployment benefits cause rises in unemployment, not the other way about. Yet in the graph, there's a considerable time lag from sharp movements in the dark line (unemployment) to sharp movements in the light line (benefits), indicating the causation is actually, if anything, the other way.

And how surprising is this? Real benefits are calculated by multiplying payment amount per recipient by average duration of benefits. During slumps, it's harder to find a job, so the length of unemployment tends to increase, end of story.

Well, maybe not. If you look at the graph again, specifically at the points where the unemployment rate reverses course downward, Arthur Laffer's argument starts to look ever more absurd. There are 5 major peaks in unemployment shown, all of which hit their peak around the time unemployment benefits spike up sharply. Might this suggest that far from causing higher unemployment, increased benefits are causing sudden drops in unemployment?

Someone call Laffer Associates. I think we a problem.