Where do disability support pensioners live?Posted: November 15, 2014
Judging from the kind of commentary I read about the disability support pension—among the vanguard of the commentariat, the most unloved of Australian transfer payments—there is a fraudulent claimant in every second house on your street. Actually, only about 5 per cent of the working age population receive the payment. But I thought it might be interesting to look at whether people who receive the payment are distributed evenly across Australia or whether there tend to be more in some parts of Australia than in others.
In this post, I look at local government areas and the rate of DSP recipiency in the working age population. Recipiency rates vary from 0.54 per cent to over 20 per cent; or, to put it a different way, from about one in two hundred people on DSP in one LGA to one in five in another. This is—to put it mildly—a lot of variation.
Why might this be the case? Here are some reasons I can think of, just off the top of my head:
- The gap between the DSP rate and Newstart. The rates of these payments vary, so in places where unemployment is high, maybe there’s an incentive for unemployed people to try to qualify for the Disability Support Pension. This is a frequently rehearsed argument in the DSP discussion, and while I’m sure there’s something to it, there something I think we should bear in mind: when economic conditions turn bad, (either in a temporary or secular fashion), it’s likely that disabled people in the labour force may be more likely to lose their jobs than people who aren’t disabled. And there’s a reason that the DSP rate is higher than the Newstart rate: it’s more expensive to live with a disability than without. Accommodation, transport, food, medicine—having a disability can increase the cost of basic life things, even without thinking about specialist equipment and other things that are supposed to be covered by governments but to the payment of which people with disabilities are sometimes forced to contribute. So for disabled people who have some work capacity, we should think of the DSP as a special unemployment benefit; an unemployment benefit with a disability loading. In this sense, then, there may be reasons other than the Newstart–DSP gap itself that explain why unemployment and DSP recipiency rates might be correlated. I haven’t yet looked for estimates of unemployment rates in LGAs, but one proxy might be the recipiency of Newstart in an LGA, which does seem to be positively correlated with DSP recipiency rate
Such a correlation could be consistent with several of the stories I’ve told above, including the Newstart–DSP payment gap story (i.e., in areas with high unemployment, unemployed people also try to get on DSP because of the higher rate.) But then again, maybe this shows that the substitution between the two payments isn’t as strong as has been elsewhere suggested: after all, the slope of the fitted line is quite flat, and LGAs with very high unemployment rates don’t seem to have commensurately high DSP rates. Eligibility for DSP is, after all, medically assessed, and there is probably a soft upper limit to recipiency rates in any given area, no matter how high unemployment is.
- Age and disability are correlated. That is, as you get older, you’re more likely to have a disability. Areas where the working age population is older should, all other things being equal, have a high rate of disability prevalence, and hence a higher proportion of the population eligible for DSP.
- Price levels, especially for housing. The fortnightly DSP is not a whole lot of money to live on. And we know that the cost of living varies across the nation, while the DSP rate doesn’t. Someone living on DSP is unlikely to be able to afford to rent in the leafy green streets of Mosman: that’s probably why that suburb has one of the lowest DSP recipiency rates in Australia. I haven’t yet calculated the relationship between income in local government areas and DSP recipiency rates, but I have had a look at the relationship between income and core impairment (a census measure of physical disability) in small census tracts:
- The location of medical and government services. People with disabilities often need to access specialist services (therapists, say, or Centrelink). These aren’t always scattered uniformly across the country, so it may make sense for people on DSP to live in areas close to the things they need. A similar issue is jurisdictional: while the DSP is a federal government payment, some services provided to people with disabilities is provided by state governments. If a particular state government provides better or more generous services, then in border areas, people may move across state lines—say, from Wodonga in north-east Victoria across the Murray to Albury—to take advantage of better services. This could lead to ‘bunching’ of DSP recipients in LGAs on one side of border.
- The weather. Some disabilities—rheumatism, say—may react differently to different weather conditions. (From what I can make out, the scientific literature is inconclusive, although I have friends with joint conditions who say that cold weather exacerbates pain, and my arthritic relatives often travelled to warmer climes for relief.) In this case, people with these kinds of disabilities might cluster in certain geographical areas in order to alleviate some of the symptoms of their condition.
- Variation will be higher between smaller LGAs than larger ones. In local government areas with large populations, all the factors that contribute to differential recipiency rates are more likely to be ‘averaged out’ than in smaller LGAs. For example, the Brisbane LGA, being so big, is likely to contain relatively rich and relatively poor areas. These will tend to cancel each other out, and the overall recipiency rate is likely to be similar to that of the national average. (Which is around about 4-6 per cent; see one of my previous articles). This is easy to see in the following chart:
There are probably many more you can think of.
The least, and the most.
Let’s have a look at some actual numbers, which I’ve calculated by combining DSS data on the number of DSP recipients in local government areas and ABS projections of 2014 populations for each area, based on the 2011 census and fertility, deaths and other variables.
The ten LGAs with the lowest estimated rate of DSP recipiency in the working age population are an interesting mix of rural and well-to-do.
Ashburton, a council area in the Pilbara, is the LGA with the fewest disability pensioners in Australia, with an estimated 0.54 per cent of the working age population on the DSP. Mosman in Sydney is next, with 1 per cent, then Ku-ring-gai (Sydney; 1.02 per cent), Cambridge (in inner Perth; 1.06 per cent), Cottesloe (again, Perth; 1.13 per cent), Woollahra (Sydney; 1.14 per cent), Isaac, in rural Queensland (1.17 per cent), East Pilbara (1.36 per cent) and North Sydney (1.38 per cent).
What about the areas with many disability support pensioners? The LGA with the highest rate of DSP recipiency in the working age population is Peterborough, about three hours’ north of Adelaide, where 20 per cent of the working age population are on DSP. The Tasman LGA in Tasmania (which covers Port Arthur) comes next, with 16.7 per cent; Walgett Shire, in northern New South Wales, with 16.2 per cent; the Central Goldfields, in Victoria, with 15.8 percent; Kempsey Shire on the mid-north New South Wales coast, with 15.1 per cent; and the Tiwi Islands, with 15.02 per cent.
This map gives a very broad overview of the recipiency rates in Australia: blue areas have low fewer DSP recipients as a proportion of the working age population, while red areas have more. The grey areas are those for which there was insufficient data to calculate a proportion; these are usually remote areas with small populations, even though they may be large in area.
Now let’s take a closer look, state by state.
Let’s start in the Top End. Here, things are slightly complicated by the fact that DSS reports Darwin as one big local government area, so we don’t have much detail on a suburb-by-suburb basis. The area with the highest proportion of its working age population receiving the disability pension is the Tiwi Islands area, while the lowest is Litchfield.
Moving on to the Sunshine State, the LGA with the lowest recipiency rate is Isaac, southwest of Mackay; the LGA with the highest recipiency rate is Cherbourg, an Aboriginal community. We encounter here the same problem we had for Darwin: Queensland’s local governments can include a lot of people—especially the city of Brisbane, with about 800,000 people in the working age population alone.
Let’s also take a look at the Southeast. Brisbane City is the bright blue on the coast down the bottom.
New South Wales
In the first state of the Federation, we can see a definite reddening up towards the border with Queensland. Is this a confirmation of the services/jurisdiction hypothesis outlined above or is it just coincidence? I don’t know. Mosman has the lowest proportion of DSP recipients; Walgett the highest.
Here’s a look just at Sydney, by the way, where there doesn’t seem to be very large concentrations of people on the DSP as a proportion of the working age population:
My home state. There seems to be a big red patch in the centre west, upon whose causes I won’t speculate. In the Garden State, it’s among the well-manicured lawns of the City of Boorondara in the eastern suburbs of Melbourne that you’ll find the lowest DSP recipiency rate; the highest may be found in the Central Goldfields.
Once again, zooming in on the capital city gives us a soothing blue image:
The Apple Isle is a big dollop of red on our map:
Indeed, most of the state is above the national average.
Here’s Adelaide up close:
That ocean blue of the Adelaide Hills is the area in SA with the lowest recipiency rate; the highest is Peterborough (the highest rate, in fact, in Australia, as we saw above.)
Finally we arrive out west, where the highest recipiency rate is in Halls Creek and the lowest in Ashburton.
And let’s zoom in on the southwest:
Where to from here?
The distribution of DSP recipients across Australia is interesting enough as it is. But there’s more work that could be done. Looking at the hypotheses at the beginning of the post suggest some other data sources that could be joined with the payment data above: for example, census data on income and age, or estimates of unemployment rates in local government areas. (If you, dear reader, would like to give it a go, I’m happy to share data (CSV; ESRI) or code (R).)
I may have goofed up. That’s always possible when you’re doing fiddly data work, so, in the interests of transparency, here’s the basic outline of what I did.
First, I took the Department of Social Services’ June 2014 DSS demographic data, which lists disability support pensioners by local government area. It also gives numbers for postcodes and by federal electorate. In order to preserve privacy, though, DSS simply writes ‘less than 20’ for any postcode or LGA with fewer than 20 pensioners. This happens quite a lot with postcodes, and isn’t helpful when trying to calculate recipiency rates. On the other hand, federal electorates are quite large, and so don’t give allow us to drill down quite as far as we’d like into the numbers. I opted for the middle road, which was local government areas. This, though, begat its own problems: there aren’t LGAs in the ACT, for example, and local government sizes vary across the country. Nonetheless, for a first pass, I think this is an OK compromise. Repeating the exercise with both postcodes and federal electorates (or some hybrid approach, say, by substituting postcodes for LGAs in the ACT) would be a useful next step.
One problem I tackled on an ad hoc basis was the fact that the ABS and DSS definitions of local government areas aren’t always perfectly congruent. To take the Brisbane example again, the massive population size of this LGA means that while DSS only reports one DSP figure for this local government area, the ABS’s population projections give more finely detailed projection. In practice, this makes it hard to merge the data, since you have to add up all of the smaller area populations to get the denominator in the ratio. I did this half automatically with some R code. Each LGA has an ID with 5 digits, but the DSS data only goes to the first 4 digits—I think maybe the fifth digit is used for wards in local government areas or similar—so I split the ID code into the first 4 digits and the last digit, aggregated by the 4-digit part, and muliplied by 10 to get the DSS version, where all trailing digits are 0. This wasn’t sufficient to deal with Darwin, Brisbane, Townsville, Logan, Moreton Bay and Toowoomba, which all crossed over 4-digit ID codes, so I aggregated these ones by hand.
I then merged the resulting dataset with ABS projections of population in LGAs in 2014. Since these were made on the basis of the 2011 census, these aren’t perfect, but should give us a rough guide to the number of people living in each LGA. The DSP is a working age entitlement (15-64 years old) so I calculated the projected working age population, adding up single year projections, rather than simply using the total population provided.
There are LGAs that had fewer than 20 DSP recipients, since, as outlined above, these have been ‘censored’ by the ABS. This means that although we know that number of DSP recipients in these areas was lower than 20, we don’t know by how much. I simply deleted these cases; hence the grey areas on the map above. There were also some missing data where an area with a given population projection had no DSP data and vice versa; I deleted any case like this as well and these also come up grey.
I’m no specialist in mapping, R, or empirical welfare studies, so happy to get any suggestions for improvements I could make.