It’s the unemployment rate …

My friend Chris Joye has been having some fun with the benchmark population revisions. Today, however, he has backed the wrong horse in Bassanese.

The household survey is NOT designed to measure the number of jobs. It is designed to measure a bunch of key ratios, including the unemployment rate. Multiplication of population forecasts by the ratios yields the so-called jobs number. Population over- and under-estimates are corrected for by smoothing subsequent population estimates toward the revised estimate, and this may distort the derived levels. The following picture (from the ABS’s explanation) shows the population revision process.

Pretty much every professional Australian economist and policy maker i have come across understands the strengths and limitations of this data-set — and most wish the household survey had a larger sample, and that it was complemented by an establishment survey (that was designed to measure the number of jobs).

The ABS explained how it all works in last Thursday’s labour force publication. There are two sources of revision – the levels estimates are revised, as the population count is revised. And there can be small changes to the ratios, as new information about population composition changes the weighting of the survey results.

To see how population benchmarks drive employment level revisions, compare the two charts – above and below. The above shows the revisions to the labour force estimate, and the below chart shows the revisions to the employment level estimates. Notice how they have almost exactly the same profile.

Note that this does not mean ‘more folks in work’ it means ‘different weights applied to the survey’. Here is the ABS on the above chart:

The graph shows the largest revision was 156 thousand for September 2009. This does not mean that there were 156 thousand more people employed in September than was first estimated, it means in broad terms the weight assigned to each individual in sample in September was much more after the revisions were carried out.

There is another source of revision, which is due to the re-weighting of the survey results, as new information alters estimates of the population’s structure. This weighting is important, as the household sample doesn’t perfectly match the population’s structure, and different groups have different degrees of labour force engagement.

Most recently, we discovered that there were more 20 to 35 year olds in the population. This group has higher employment participation, so the result is a further increase in employment estimates. This second source of revision will also tend to increase employment to population ratios (see below) – and it is the only one you’d care about if you were forecasting inflation.

The RBA and Treasury are across all this, and I am confident that they have not been tricked into a policy mistake by ‘bad data’ – for they would not have been misusing the household survey in this way.

So Bassanese is wrong about the RBA not cutting at all if they had this data. These revisions would not have altered the RBA’s inflation forecast.

The RBA cut their policy rate because core inflation is expected to be in the lower half of the band for the entire forecast period (they have a 2% to 3% target range) and because growth is sub-trend and is not expected to be above trend until 2014 (their forecast is 3%y/y until June 2014, when growth accelerates to 3.5%y/y).

With these forecasts, the RBA’s unemployment rate forecast would have been rising, their employment to population forecast would have been declining, and these forecast innovations were why their inflation forecast had declined.

In response, the RBA cut by 50bps in May to get mortgage rates where they wanted them. Doing so allowed them to keep inflation on track for ~2.5%y/y over the period in which they have some influence on outcome.