Thursday, October 13, 2011

Little Mercies

"The situation has remained terrible for so long that there is now a kind of defeatism that makes frustrated well-wishers eager to be thrilled by little mercies". Amartya Sen (see here)
Prof. Sen, in the quote above, is talking about Burmese politics.

Let's talk instead about unemployment in Australia, using the following two ABS charts (both coming directly from "Measures of Australia's Progress, 2010", ABS cat. no. 1370.0, here).

This chart tells a story quite familiar:
"For most of the last decade, the unemployment rate declined as a result of Australia's strong economic growth from a high of 6.8% in 2001 to a low of 4.2% in 2008. In the wake of the recent global financial crisis, the unemployment rate rose to 5.6% in 2009 before declining to 5.2% in 2010".
Not too bad, uh?

Let's now have a look at this other chart, from the very same ABS release:

Four things caught my eye in this chart:
  1. It includes older data (it starts in 1966);
  2. Unemployment during the first 6 years shown (the oldest data available) was under 2% (I kid you not: check by yourself!), and for the best part of first 12 years it was under 4%;
  3. It clearly shows that unemployment has a way of going up quick, but falling slow;
  4. The chart does not show the current fall in unemployment rate.
In fact, if the readers download Labour Force Historical Timeseries, Australia - Labour Force Status by Sex and Marital Status (ABS cat. no 6204), they'll find that it was only in February 1975 (well within the stagflation crisis of the 1970s) when unemployment reached 5.4% and thus became comparable with our current "relatively favorable" unemployment rates (5.6% average for 2009, 5.2% in 2010).

Interestingly, since 1975 the unemployment rate never returned to late 1960s/early 1970s levels.

This long-term picture isn't nearly as pretty.


One question comes to mind after seeing the second chart: if employment is the main source of income for a majority, what kind of effect should a higher unemployment have on incomes? A priori, it would be reasonable to assume that an increase in unemployment reduces incomes for the bottom 90% income earners.

Given that ABS does not provide much data on incomes (see here), I'll use information provided by the World Top Incomes Database (WTID, last year available: 2007). The flip side of the coin is that the database focuses on higher incomes: it does not provide details directly on lower income earners.

The following chart provides some insight on this, combining ABS unemployment rate data and bottom 90% average income, from WTID:

It might be an artifact of the data available, but from 1966 and until the first half of the 1970s is not clear that average annual employment rate and the bottom 90% income average move in different directions.

In fact, both average annual unemployment rate and the bottom 90% average income would seem to move together: one rises, the other rises; one falls, the other falls. In other words, both variables appear to be moving in the same direction.

However, from the second half of the 1970s on, both variables clearly move in opposite directions: increases in the average annual unemployment rate tend to be accompanied by falls in the bottom 90% average income; the inverse also holds: one observes falls in average annual unemployment rate tend to be accompanied by increases in the bottom 90% average income.

And here is a scatter plot showing these relationships:

The blue plus signs correspond to the average annual unemployment rate and bottom 90% income averages for the 1978-2007 time period; the red circles represent the same data, but for the 1966-1977 period. The blue dotted line is the second degree polynomial fitted to the 1978-2007 data; while the red discontinuous line is the same model, but fitted to the 1966-1977 data.

It should be noted that perhaps there are too few data points within the 1966-1977 data set, to conclude anything (as indicated by the low R^2). Therefore, the impression that income and unemployment move in the same direction during this period is not confirmed, although it seems evident that a different process underlies this data.

The fit for the 1977-2007 data set, however, seems much more conclusive: in effect, within the data range, it is clear that unemployment decreases incomes.

As can be seen, two apparently very different results.


Short-run data showing unemployment decreases only show improvements from a really bad starting point. That data is regularly released by the ABS (cat. Nos. 6291 and 6202, for instance).

These improvements overshadow the fact that unemployment remains at a much higher average level than it was during the 1960s. The 1960s data available also comes from ABS (cat. No. 6204).

However, short term changes get reported by the media, while the long-term picture, some academics excepted, is all but ignored. Thus, a fall of tenths of a percent in the unemployment rate tends to influence public opinion, even though the average level on unemployment is more than double what it could be.

This conclusion is reinforced by the fact that underemployment figures seldom receive any mention.

On a related matter: if the unemployment and income data available are reliable, something must have changed fundamentally in the Australian economy during the late 1970s and early 1980s to change the relationship between unemployment and incomes. From the data alone I cannot formulate any explicative hypothesis (any pertinent feedback is welcome)

However, given that currently unemployment and incomes move in opposite directions, the current higher unemployment figures (relative to the 1960s) lead to the conclusion that incomes for the bottom 90% income earners must have suffered, in comparison to what they could be.

In a previous post I showed how the top 1% average incomes diverged from the bottom 90% average incomes, pulling with it average incomes. The increase in inequality reported in that piece, which started in 1983 (at the beginning of the Hawke government), implies that average income is not an accurate indicator of economic well-being for the majority of Australians and that this systematic inaccuracy is increasing over time.

And this is particularly troubling, as average income is the main indicator of economic well-being in Australia, and income, with all its limitations, seems to have a relatively low priority within the ABS research agenda.

On the basis of these two observations, I must conclude that the ABS's Measures of Australia's Progress 2011 (see here) are of strictly limited value to assess the long-term improvement in living conditions for Australians, at least in what refers to income and employment.

No comments:

Post a Comment