Over the past month, welfare computer systems in both British Columbia and Ontario have crashed, paying thousands of recipients wrong amounts of money.
But it’s hardly the first time this has happened. Over the last 40-odd years, every new computer system for Ontario’s social assistance programs has had these same types of glitches.
In the early 1970s, officials discovered that the welfare mainframe could not pay a simple percentage rate increase.
In the 1980s, the notorious and much maligned Comprehensive Income Maintenance System (CIMS) took years to roll out and was so difficult to implement that many municipalities ran for the hills and kept a myriad of their own proprietary systems until finally forced to join the provincial system in the new millennium.
The Accenture system that replaced the 1980s model in early 2002 had so many setbacks, schedule and cost overruns that it was routinely lampooned in the media as a boondoggle. Workers scrambled to implement literally hundreds of “workarounds” effectively creating a manual backdoor system the computer system was originally designed to automate.
I know this well as I was there as part of a team working on user acceptance testing.
Then in 2004, then-minister Sandra Pupatello discovered that the new system had the same problem as the old mainframe system in the 1970s: it could not implement an automated rate increase. Officials at the time used an odd defence of the system by noting that it had to calculate assistance using more than 800 separate rules.
Fast forward to 2014 and it is déjà vu all over again.
But instead of worrying about fixing what we have, we should be worrying more about why it’s so hard to implement a computerized welfare system.
We should be less concerned about the alligators and starting to think about draining the swamp, starting with those 800 rules.
Ontario has 14 separate rules governing who is automatically ineligible for benefits, eight sets of benefit categories, and 12 sets of rules that define dependency.
The test of need includes complex rules surrounding liquid assets limits that vary by the presence of disabilities, family size and family structure with at least 42 separate rules that apply differently in three separate life situations.
Once an applicant or recipient’s level of liquid assets is below the prescribed limits, a test of income is applied. Some forms of income are exempted in their entirety (e.g. tax credits) while some are exempted in part (e.g. earnings). Still others reduce assistance on a dollar-for-dollar basis (e.g. child support).
Finally, a test of availability for work is applied in the Ontario Works program whereby recipients are required to participate in community and employment-related activities (unless exempted) under at least 25 separate rules.
Once all the tests are applied, an allowance structure based on basic, shelter and special needs varies by program, family size, family structure and accommodation type. The resulting amount is paid as assistance resulting in at least 25 separate rates of assistance. Special needs amounts are paid in addition to these allowances.
Add to this approximately 600 more rules defined in other regulations, directives and memoranda to administrators.
Then throw this large and unwieldy mass of rules at an army of computer code experts who have never delivered the program in the real world and expect thousands of social workers and clerks in 48 separate delivery centres to administer it and you have a recipe for problems and glitches that no amount of testing and rehearsal can predict.
So let’s work hard to fix the glitches and problems.
But let’s work harder on the preposterous over-complexity founded on the misguided idea that we have to parse every gradation of need in order to have a publicly acceptable welfare system.
- John Stapleton is Innovations Fellow at the Metcalf Foundation and a former social assistance policy analyst with the Ontario government