The US labour market seems to be very weak. I do not think that it is weaker than this time last year – when the Fed undertook QE2. Bernanke has told us that is off the table for the minute – as inflation is too high. However, if the unemployment rate keeps rising, I would expect inflation’s pace to fall, and that would allow the Fed to ease policy further.
My assessment of the US labour market will not come as a surprise to those that take interest in the monthly payrolls report – however there is always some uncertainty about the state of the economy, given that everything which is sampled is sampled with noise. Also, with so many releases pertaining to the labour market, there is always something that seems okay.
To deal with this problem in a structured way, I quite like Principle Components Analysis — conceptually, it averages across the series, and therefore ought to cancel the noise.
Doing so, we find the that the June value of the index fell from +0.3 to -0.1, taking the 3mma down 0.2 to 0.19. The series runs from May’80 to Jun’11, the dots are the monthly values, and the 3mma is the red line (thanks to Manny C for help with the R chart!).
This is the worst monthly result since mid-2010. Both monthly results were weaker and the weak spot persisted for longer in mid 2010 – so the 3mma was markedly lower.
To manufacture the index, I have combined 49 indices pertaining to the labour market – including various establishment and household survey related indices, initial jobless claims, the regional Fed and ISM employment indices, business surveys, and monster job ads.
Looking at the matrix of weights, I was surprised to see that the regional Fed surveys contribute more to the (first) factor than the NAPM manufacturing ISM’s employment index. The regional Fed indices have a similar weighting to the BLS’s manufacturing employment diffusion index … so don’t feel too good about that firm manu ISM, as it’s inconsistent with the ‘better’ measured data.
So, what should we look at? The highest factor loading is on the Dallas Fed’s employment index. The non-manufacturing ISM’s employment index, and the BLS’s diffusion index (1m) score similarly high.
What do the first two rows of data frame look like?
Comments are closed.