The recent dispute over figures used in Minister Leo Varadkar’s populist “Welfare frauds cheat us all campaign” shows that we need greater statistical literacy when it comes to numbers that are bandied around.
That’s nothing new – as an article from last year highlights.
In March 2016, Aaron McKenna of the free-market-focused Hibernia Forum wrote an article for The Journal with the provocative title, “We need to weed out bogus disability claims for the people who really need it.” It was also published on the Hibernia Forum’s website.
Since then, the statistics have been referenced from time to time in debates on social media, appearing as recently as May this year.
Also, for disability allowance alone, the figure for fraud given by the Department of Social Protection is 1.2 percent of expenditure, or 1.6 percent of claims.
The Journal article itself drew on three sources. Two of them, a report by the Comptroller and Auditor General and Department of Social Protection statistics, contradict the article itself.
The third “The Mystery of Disability” by the economist David McWilliams appears to be based on that author’s misreading of the census.
Let’s deal with McWilliams’ article first. It argues that there was a 37.7 percent increase in the number of people citing a physical disability as making them unable to work. That’s an increase of around 55,000 people, from 172,603 to 237,748 between 2006 and 2011.
Similarly, the claim is made that those citing mental-health problems as making them unfit to work increased by 44.6 percent from 61,499 to 88,898 over the same period. He stresses that he does not know what is behind this increase.
The data that McWilliams based his argument on appears to be drawn from the Central Statistics Office, which gives a census breakdown based on the years, categories, and numbers involved.
I drew out the data for “Population Aged 15 Years and Over by Disability Type, Sex, Principal Economic Status, Age Group and Census Year” for 2002, 2006 and 2011 – which has figures that match those used by McWilliams.
The results show that David McWilliams made a serious error. He has used the combined figure for all people with disabilities across all principle economic statuses, including (for example) people who are working, and people who are retired.
That’s a different –and much broader – group from those people who are outside the labour force due to illness or disability, in both “psychological or emotional condition” and “a condition that substantially limits one or more basic physical activities”.
Look at the complete table: who does it actually include? It includes those in the labour force: people at work, those unemployed and looking for a new, or their first, job.
It also includes those outside the labour force: those looking after the home or family, in education, and the retired, in addition to those unable to work due to disability.
If we look at the percentages of those participating in the labour force, we see that in 2002, it was 12.5 percent of all physically disabled people, in 2006, it was 13 percent and in 2011, it was 14.1 percent.
If we adjust for age by removing those above the retirement age of 65, those percentages are 25.4 percent, 26.9 percent and 28.7 percent. These rates are relatively consistent and moving in the right direction in each successive year.
A similar pattern can be seen in terms of those with a psychological impairment. Of this group, 26.8 percent were in the labour force in 2006 with this increasing to 34.9 percent if we adjust for age.
In 2011, 30.2 percent of all those claiming a psychological or emotional condition were either working or seeking work and, when we adjust for age, this is 37.4 percent.
What about those outside the labour force citing their impairment as making them unable to work?
When you adjust for age to remove those over the age of 65, you find that in 2002, 52.3 percent of those with physical impairments were unable to work citing disability as the reason. In 2006 it was 53.15 percent, and in 2011 it was 52.9 percent.
In terms of a psychological or emotional condition, again adjusting for age, the percentage drops from 46 percent in 2006 to 42.9 percent in 2011.
What we have seen is that while the overall numbers of disabled people have increased significantly, this has not impacted either the census labour-force rate or the rate of those unable to work due to disability.
Aaron McKenna’s other figures are easily debunked. In his column, he claims that: “The number of people on Disability Allowance alone has increased by 33,000 since the recession began.”
And continues: “Quite coincidentally, 33,000 is nearly the same number of people who were classed as being long term unemployed at the height of the boom.”
In reality, between 2008 and 2014, between the beginning of the recession and the last date available when the Hibernia Forum article was written, recipients of disability allowance increased by 16,343.
Maybe they were counting from 2007, the last full year of the boom? Nope. In that case, the figure would be 23,049, still 10,000 below the claimed increase in recipients.
In fact you would have to start from 2005 – years before the start of the recession – to hit around the 33,000 figure.
Looking at the figures on the graph on slide 14 of my presentation to the NERI conference, we see that prior to 2009 the number of claimants of disability allowance was increasing by around 4,000 to 6,000 each year.
However between 2009 and 2012, this rate was much lower; it fell to 1,000 to 2,000 and between 2011 and 2012, the number of claimants actually dropped from 102,866 to 101,784. The rate appears to have recovered to pre-crash levels after 2012.
What about the lower-back-pain study that McKenna also cites? In his summary, the Department of Social Protection sent 1,362 claimants for medical assessments, and after the doctors’ reports, 88 percent of the claims were terminated.
The study from 2003 by the Comptroller and Auditor General relates to short-term disability benefit (since renamed “illness benefit”). Its findings are a bit different to those presented by McKenna.
Rather than 88 percent of claims (1,200 out of 1,362) being disallowed for not meeting medical eligibility criteria – which, contrary to McKenna’s interpretation, does not necessarily mean fraud – 275 out of 1,360 (not 1,362) or 20.2 percent were found to be ineligible.
McKenna used a figure that included the 736 or 54.1 percent who had handed in their final certificate from their doctor ending their claim, after they were told about the upcoming assessment, and may have done so anyway.
While the report does say the numbers handing in their cert was higher than the norm, it needs to be remembered that the survey was prompted by a significant increase in claims of this type.
For this reason, it would be irresponsible to generalise from this survey of those claiming for a specific temporary injury (and the type of payment is a short-term claim) to those on disability schemes as a whole.