Monday, September 12, 2011

Rising unemployment more puzzle than worry

The unemployment rate has risen by 0.2 percentage points for two months in a row. Taken at face value, that says the economy is rapidly heading into recession. But it's always a mistake to take economic statistics at face value and, fortunately, the truth is likely to be far more reassuring.

The trick to using economic indicators to understand what's happening in the economy is not to overreact to the latest reading from one indicator. The new figure has to be put into the context of the particular indicator's trajectory and the general message coming from all the indicators.

Some indicators are more important than others. The quarterly national accounts - the centrepiece of which is gross domestic product - are the most important because they constitute the summation of a host of ''partial indicators''. GDP may be a poor guide to the nation's overall well-being, but it's a good guide to the outlook for income and jobs.

The monthly figures for employment and unemployment are very important because that's one of the main things we expect the economy to do for us: generate jobs for all those who want them.

The next question to ask when you get a new reading from an indicator is: how reliable is it? The job figures are based on a rotating sample survey, meaning they're subject to sampling error (as well as a lot of opportunity for other, human errors).

They tend to bounce around from month to month for reasons you can never put your finger on, but which don't reflect the more stable reality of the labour market. The national accounts also bounce around and are subject to heavy revision as more reliable data come to hand.

So both the key indicators are a bit ropey, and economists often use one as a check on the other. We know from last week's national accounts that, though natural disasters caused the economy to go backwards in the March quarter, it bounced back strongly in the June quarter and will recover further over the rest of the year as flooded coalmines get working again.

Last week's jobs figures told us that national employment - which totals 11.4 million - fell by 4000 in July and 10,000 last month. These are trivial amounts; they're saying not that employment is falling, but just that it's not growing. Trouble is, we need employment to keep growing because the population of people wanting jobs keeps growing. We need employment to grow by about 10,000 a month just to hold the rate of unemployment steady. Fortunately, this is less than half the rate of employment growth we needed a year or two ago because the rate of growth in immigration is now so much lower.

Note, unemployment has risen not because people are losing their jobs, but because additional jobs aren't being created. As a general proposition, we need the economy to be growing steadily because that's what creates additional jobs. Stepping back to view a longer run of figures, we see that employment was growing very strongly until November, since when it's shown virtually no growth. Though the economy contracted sharply in the March quarter, this contraction was weather-related and concentrated heavily in mining. And, as we've seen, the economy grew strongly in the June quarter.

So the pattern of growth in employment isn't easily reconciled with the pattern of growth in production (GDP). We need to examine the jobs figures to see how robust they are.

One way to see if there may be problems with the rotating sampling process is to look at what's happening to the ''matched'' sample (the part of the sample that's unchanged from one month to the next). Kieran Davies, of the Royal Bank of Scotland, has done this and finds that ''smoothed matched-sample employment is growing at 17,000 a month, while headline employment is broadly flat''. Hmmm.

Examining the breakdown of the (headline) employment figures shows the weakness is heavily concentrated among men rather than women. It also shows the problem is concentrated in Queensland (where in two months the unemployment rate has risen 0.9 percentage points to 6.2 per cent) and Western Australia (where in one month the unemployment rate has risen 0.4 percentage points to 4.4 per cent).

I don't trust those figures. But if you take them literally, both they and the national accounts are saying the precise opposite to the conventional wisdom about the ''two-speed economy'': all the weakness is in mining and the mining states, while all the strength is in the so-called non-mining economy.

But here's another puzzle. While there's been no growth in the number of people employed this year, the total number of hours worked has been rising solidly, with average hours per worker rising from 34.7 to 35.6 hours a week.

As Davies has argued, this sort of behaviour by employers - where they work existing staff harder rather than employing more workers - is what often happens when the economy is recovering from a recession and the media is wringing its hands over ''jobless growth''. It may be that, fearing skilled labour shortages, employers stocked up with workers last year, but this year they're not hiring any more until they need to.

The forward indicators of employment (job ads, vacancies etc) are weaker than they were, but still not weak. And the outlook is for strengthening GDP growth over the rest of this year, with the Reserve Bank's forecast of 3.25 per cent growth over the year to December still in with a chance.

So whatever the job figures are telling us, it's not that we're sliding rapidly towards recession - even in the so-called non-mining economy.