Census raises questions about numbers out of work Comment Economics Editor

COMMENT:Misleading statistics on levels of employment have been lulling successive governments into complacency

ON APRIL 10th last year there were 4,588,252 souls alive in this State, an increase of more than one-third of a million on the last comprehensive count five years earlier.

The latest deluge of newly published census data yesterday showed that, despite rapid population growth (the most rapid in Europe, as it happens), the number of people at work fell by more than 120,000 between April 2006 and the same month in 2011.

Put another way, in 2006, for every 10 workers there were 12 non-workers, including kids, pensioners and the jobless. By last year every 10 workers had to support more than three additional non-workers, or 15.4 people. If those at work are wondering why they are paying so much extra tax, this is one of the main reasons.

If all that is bad, the unemployment rate in yesterday’s figures was doubly shocking. It shocked once by standing at a staggering 19 per cent and shocked again by being far above the 14.3 per cent rate the Central Statistics Office was reporting at the time.

And if that is not depressing enough, the difference between census figures and the rate gleaned from the regular Quarterly National Household Survey (QNHS) is particularly damning for young people.

In the 15- to 19-year-old category, the QNHS found unemployment was 40 per cent at the time of the last census. But the census figures put the rate at an incredible 58 per cent.

In the 20- to 24-year-old group unemployment was 28 per cent, according to QNHS, but 35 per cent according to the census. Which measure is more accurate?

The QNHS is based on a sample of citizens. Statisticians ask those surveyed a series of detailed questions. These questions are the same as their counterparts in other developed countries ask, based on a methodology devised by the International Labour Organisation (ILO), a large Geneva-based think tank.

The QNHS has the virtue of ensuring comparability across countries, but it puts the joblessness bar high. For instance, if you have been paid for as little as one hour of work in the week before being surveyed, the statistician asking you the questions will describe you as being gainfully employed.

The census is different. It is a self-assessment and, given that everyone fills it out, there are no sampling errors (although the QNHS sample is very large, so the sampling error is likely to be minuscule).

Interestingly, the census figures are very similar to the numbers claiming jobless benefits, and very much higher than those formally classified as jobless by the QNHS/ILO methodology.

Yesterday’s figures show that 425,000 people were unemployed when census forms were filled out. At that time, the live register showed that 440,000 people were on the dole.

The QNHS figures were radically different, showing just 305,000 people out of work.

In 2006, the census counted 180,000 jobless people while there were then 155,000 in receipt of jobless benefits. Again, the QNHS put the numbers out of work far below that, at 98,000.

The CSO insists that the QNHS figure is the best measure of joblessness. In light of the census unemployment figures being so much closer to those of the Live Register, this requires re-examination, not least because it is likely to have contributed to the complacency in addressing labour market failings when times were good.

It was obvious even then that the training system was as wasteful as it was rotten. It was clearer still that the benefits system – unchanged in decades – was utterly uninformed by the sort of well designed carrot-and-stick frameworks that exist in northern Europe.

But with the QNHS unemployment rate in 2006 at 4 per cent, as it had been since the turn of the century, the gains from fighting reform battles might have seemed limited (most economists consider such a rate to be consistent with full employment as there will always be a small fraction of workers who are between jobs, something known “frictional unemployment”).

But the census found that 8.5 per cent of the labour force was jobless in 2006, while 7 per cent of the labour force was then on unemployment benefits.

At a time of surging growth in jobs these much higher rates, if they had received more attention, might have caused alarm bells to ring and might even have made complacent politicians consider doing something – the economic and political cost/benefit calculus of reform would have been very different.

Statistics exist to illuminate. They are vital in highlighting where policy action is needed. The exclusive focus on the QNHS jobless rate obscured the scale of unemployment during the bubble. The CSO needs to reconsider the way it measures joblessness.