UK Labour Income & Unemployment
This is a belated follow-up to Scott Sumner’s post last month on the puzzles in the UK macro data.
I tested Scott’s “musical chairs” model of the business cycle on the labour compensation data from GDP by income in the national accounts. Define “Labour Income” as the sum of total wage and salary compensation paid to employees (excluding employer social security contribution), plus household “mixed income”. The latter should be a good proxy for the income of those in self-employment, but is not perfect, I believe it also includes things like owner-occupiers’ imputed rents. (A topic for another post.)
Then we can find the ratio of mean hourly wages to per capita “labour income”. This shows the outcome of Scott’s “game of musical chairs”, which he explained here and also in his EconTalk podcast. (Briefly, the intuition is that when total income available to pay salaries falls below the expected trend path, total hours worked falls and unemployment rises, because hourly wages are sticky in nominal terms.)
1) per my previous post, the ONS data for mean hourly wages is not an official “national statistic”. The series was also incomplete, missing a single data point in 2001 Q1.
2) There appeared to be some seasonality in one of the series used for “Hourly Wage / Per Capita Labour Income”; all inputs except the population count are seasonally adjusted. I applied a 4 quarter moving average to smooth it.
Update: Ritwik Priya rightly took me to task on twitter for omitting the unit on the RHS, which is “£/hour / £000s/quarter/capita”.
Update 2: I very much appreciate Ritwik checking my data here. He noticed the RHS scale was still out by a factor of four in my original chart. I had carelessly used a 4Q moving total rather than a moving average to smooth the data. I have replaced the chart above with one which really does use a moving average. The original chart used in this post is here, for purposes of full disclosure; there is no difference apart from the scale on the RHS.
That seems like a remarkably good fit; perhaps (level) targeting per capita “labour income” on something like the definition above would be a good monetary policy for the UK.
As I’ve noted previously, when HMRC starts collecting real-time payroll data it should be easy for the government to estimate total current wage income; though income to the self-employed would still rely on surveys. A worthwhile initiative in “open government” would be for HMRC to publish aggregates of their payroll data in real-time.