Macro is not having a good day

Over at the Stumbling and Mumbling blog Chris Dillow writes,

Paul Walker wonders whether we need macroeconomics at all. He says: “Maybe aggregate economics just doesn't work, we lose too much valuable information in the process.”
I agree. Today’s labour market numbers show the point.
They show that, in the three months ending in February, aggregate hours worked rose by 1.2%, thanks to a 1.1% rise in full-time employees. However, the NIESR estimates (pdf) that in this period, real GDP grew just 0.1%. This means that productivity fell sharply in the quarter, and rose by just 0.2% in the last 12 months.
Conventional macroeconomics says this shouldn’t have happened. Productivity fell in the recession because firms hoarded labour. But they should have unwound this in the upturn with the result that productivity should have risen, with a smaller rise in employment.
So why hasn’t this happened? One possibility is that the aggregate data hides sectoral shifts.

See here for more on Chris's example.