The experience of graduates from the early 1990s recession suggests that a slow career start isn’t necessarily fatal to long-term prospects. Employers seem willing to consider graduates who haven’t managed to quickly find employment. But this doesn’t rule out more pessimistic interpretations of recent graduate employment surveys.

There are several reasons why this might be the case. There could be some structural changes in the economy that reduce the quantity of or slow growth in professional jobs, to which graduates typically aspire. For example, there could be greater automation of tasks previously done by professionals, or more outsourcing to countries with lower labour costs. Or professional job growth could stay around its long-term trend, but the number of graduates increases more quickly.

The slide below shows the long-term trend in the number of jobs classed as professional. Annual growth is volatile, but I can’t see a structural slowdown. 2013 showed relatively low growth, which might help explain why that was a bad year for graduate outcomes compared to the immediately preceding years. But 2014 was a good mid-range result, with an estimated 85,000 additional professional jobs.

Source: ABS.

Completion numbers show less volatile growth than job numbers, but multi-year comparisons suggest that they are not (to date) growing at a much stronger rate than professional employment. However, new graduates are not the only flow into the graduate labour market. Existing graduates move in and out of the labour force, and there is migration, both from new migrants and expatriates returning to Australia. Permanent skilled migration in recent times has included 32,000 to 39,000 professionals a year, although there are larger numbers here on temporary visas.

From the ABS Education and Work survey and its predecessors we can construct a time series of graduates in professional and managerial jobs, both as a percentage of all graduates and of graduates with jobs. The slide is below. The trend is affected by occupational definitions, and is most meaningful within the periods in which a reasonably consistent classification system was used. Unfortunately Education and Work 2014 has some results that are hard to believe. I think they have over-sampled people with postgraduate qualifications and under-sampled people with bachelor degrees, and as a result the employment result is biased upwards. But the bias would not be large enough to turn the slight decline in graduate managerial and professional employment shown in the slide into a large decline.

Sources: ABS here and here.

The way Education and Work is conducted means that it would not show any fast consequences of a deteriorating new graduate labour market. The number of new bachelor-degree completions in 2013 was only about 5 per cent of the stock of professional jobs. Also, Education and Work is done as part of the broader labour force survey. Respondents to that survey are on an eight month cycle, with an eighth leaving the survey and being replaced each month. As Education and Work is conducted in May, only some of the people being captured as un- or under-employed in March or April by the Graduate Destination Survey could have been included.

Another issue is whether the Graduate Destination Survey, in investigating the employment situation of graduates so soon after completion, has always painted an overly-pessimistic picture of outcomes. Although many employers have graduate intakes structured around the cycle of university completions, normal economic growth isn’t going to to produce a sudden burst of professional jobs over December to April. It’s more likely that graduates will gradually be absorbed into the workforce.

One way of investigating this is the Beyond Graduation Survey, which looks at graduates three years out. It uses a sub-sample of the original Graduate Destination Survey. The respondents to the Beyond Graduation Survey report better employment outcomes four months out than did the full GDS sample, so the results are likely to biased upwards somewhat. What the slide shows is that full-time employment outcomes are trending downwards three years out for the cohorts that had bad outcomes four months out, for those who were new graduates in early 2009 and early 2010. The drop is about 4.5 percentage points on the peak year, but only 2 percentage points below what new graduates from early 2006 experienced three years out.

The BGS survey also lets us look at job quality. Of those who have full-time jobs, there is only a very small decline in the share of people who have managerial or professional jobs. There is a larger decline in people with full-time managerial or professional jobs as a proportion of all graduates. There has been a shift to full-time study and job searching.

The Beyond Graduation Survey is the clearest evidence of negative trends beyond the short-term employment outcomes, but the fairly small full-time employment declines the BGS finds are not disastrous. Taken in the context of the other ABS employment data reported in this post there is not yet enough evidence to say that there are major structural issues with graduate employment, although the relevant trend data needs to be watched carefully. The slowing growth in domestic undergraduate commencing student numbers is desirable. But it is still possible that the bad initial employment outcomes of recent years will end up being like the bad job figures of the early 1990s: slow career starts, but not career killers.

Poor recent graduate employment outcomes inevitably raise questions about whether this shows just a slow period of labour market adjustment, or whether it is a sign of something more serious. One theory is that early periods of unemployment or low-skill employment have a scarring effect on future employment. The basic theory is that during unemployment existing skills deteriorate and new skills that come with work experience are not developed. Either or both of these things happening or employers assuming from CVs that they may have happened compound the original employment problem. What could be a temporary setback is turned into a long-term disadvantage.

For graduates, the early 1990s recession provides an opportunity to look at potential scarring effects. There were three years of more than 25 per cent un- or under-employment from 1992-94, and 20 per cent plus for 1991 and 1995.

The ABS Learning and Work survey* has a question on when the respondent graduated. On a question asking what impact their qualification had in their working life in their first six months, those completing between 1990 and 1994 had the highest rate of saying ‘no impact’, 26 per cent. The next worst result was 23 per cent for those completed between 2000 and 2004. Unfortunately, the labour force results are hard to interpret due to sample size issues. The ABS says the margins of error are too high on all the unemployment results for them to be reliable. The not in the labour force results are higher for 1990-94 graduates than either 1985-1989 or 1995-99 graduates. However this is almost certainly due to women absent from the workforce for family reasons (if I break the results into male and female I get the expected outcome, but with the ABS again warning that the margins of error are too high).

Another option is to use the census, which has problems with people not answering all the questions but still has many respondents. While the census has no question on exactly when degrees were completed, as most students start bachelor degrees in their late teens we should be able to see any obvious scarring effects. My theory here is that people aged 39 to 41 years at the time of the 2011 census were likely to have graduated into the early 1990s recession. If there is a scarring effect, they should have worse outcomes than people who are a little younger or older. The slide below shows the results for being in work, for male bachelor degree holders only, as the female not in labour force results are too ambiguous.

What surprised me about this is how employment drops for men in their forties. While there is a slight increase in unemployment for the target 39-41 years group compared to younger men this looks like a life cycle effect. The same phenomenon is evident in the 2006 census. So overall I would say there is no strong evidence of a scarring effect on overall employment levels of graduating into the early 1990s recession.

Update 9 January: After yesterday’s post seemed more interesting for the activities of men in their forties than for employment scarring, I wondered if the issue might at least partly be residualisation of the bachelor-degree group. In other words, the more successful men go on to postgraduate study leaving the men with bachelor degrees who have given up looking for work as a larger share of the remaining people who say a bachelor degree is their highest qualification. As men get older, they do become slightly more likely to give a postgraduate qualification as their highest qualification (slide below).

However, this is only a partial explanation. When I separate the analysis into education levels, men with postgraduate qualifications also start leaving the labour force in their 40s, although at a lower rate (slide below).

I’ve had a quick look at some of the other characteristics of men with bachelor degrees who are not in the labour force. The affluent retired hypothesis has some truth but far from explains it. About 10 per cent of this group report a personal income of $1,500 a week or more, compared to more than 60 per cent of all men at this age and education level.

About 40 per cent report doing childcare, although this does not mean that they are the principal carers for their children. About a third have no live-in partner, so they are not obviously relying on someone else to pay the household bills.

Around 10 per cent of male bachelor degree holders who are not in the workforce report a ‘need for assistance with core activities’ compared to 0.2 per cent for those working full time. The cumulative effects of accidents and ill health are starting to show in this demographic.

* The results reported here are not available for free on the ABS website.