However, I do think there is a significant structural change that reflects a little of what you are saying. There is an interesting question on Quora somewhat related: http://www.quora.com/Why-does-the-US-economy-take-increasing... and my answer http://www.quora.com/Why-does-the-US-economy-take-increasing...
[1] This is from 2010, but I can't find anything more recent: http://www.guardian.co.uk/business/2010/aug/24/us-home-sales...
If you push theories in which you you treat real estate separately from the rest of the economy (i.e., you assume construction workers don't immediately go out and become nurses), you are a structuralist rather than a Keynesian.
In any case, the Keynesian assumption that employment increases with production is simply not true of the modern economy. This is quite easy to establish empirically - just compare graphs of production to graphs of employment.
By the way, here is more data on employment across the economy.
http://crazybear.posterous.com/structural-shift-in-the-econo...
I think most Keynesian have moved on from that simplistic view. Stagflation in the 1970s showed that the inflation/unemployment relationship could move, and few would argue against structural change having a similar effect.
In any case, the Keynesian assumption that employment increases with production is simply not true of the modern economy.
That's an effect, not a cause. Traditionally it has happened like that, but anyone who argued it had to be that way would be stupid.
I think we are strongly agreeing about the importance of structural change? (And hence I agree with you about the lack of a strong relationship between production and employment). I don't agree with you that demand has recovered though. Your data (http://research.stlouisfed.org/fred2/series/DGORDER) shows it still at levels around that in 2004/05.
This "tight correlation" is a conclusion of Keynesian economics, and is strongly dependent on an assumed correlation between production and employment.
The logic goes: production and employment are highly correlated (i.e., you can't increase production much without increasing employment). Weak AD reduces production, therefore stimulating AD can increase production and employment.
As my data shows, the correlation between production and employment is quite weak, and we can therefore not use employment as a proxy for demand.