On the issue comparisons and the JNLR, I am perhaps a lone voice on a particular topic- the comparisons. Every quarter the figures are issued and there is great emphasis placed on the falls and the gains. But they draw their data from two overlapping surveys. This means that the comparative surveys are using the exact same figures in both sets of data, which to my mind is hardly a comparison. Take a look at this graphically, as it save a lot of explanation:
So, this week we are comparing 2010/4 to 2011/3. Therefore they have the data for Q1, Q2 and Q3 of 2010 in both sets of data which is more doubling down than comparing.
Surely like with like is better: