Costing out the new Autism program. It looks like a 50–100M spending cut but, honestly, who knows?

I have been trying to cost out the Ontario government’s new autism program. Trying to cost out what this would look like during the transition is hard enough, but even costing out the fully-implemented version relative to the old one is hard enough, because there’s so much we don’t know.

Here’s my attempt to cost it out, with my requests for more information.

First, we need the counterfactual. What would the program have cost in future years if there were no changes? In this blog post, I detail why I believe it would be in the neighbourhood of 350M in future fiscal years. But that’s just a back-of-the-envelope guess.

Can the government tell us the counterfactual? Without these changes, what was the expected cost of the OAP for fiscal years 2019–20, 2020–21 and 2021–22?

Next, we need to cost the new program. First, we would need to know the maximum funding level for each child during each year. Surprisingly, we still don’t know that! The best guesses is that it’s $20,000/yr for kids aged 2–5 (a 4 year period) and $5,000/yr for kids aged 6–17 (a 12 year period) OR it’s $17,000/yr for kids aged 2–6 (a 5 year period) and $5,000/yr for kids aged 7–17 (a 11 year period). Either one would equal a maximum of $140,000 over 16 years. From a costing point of view, the 20/5 plan costs a bit less since the funding is more heavily weighted towards younger kids, and many kids don’t get diagnosed until they’re older.

In my costing, I’m going to assume the 20/5 plan, as it seems more realistic (governments like round numbers).

Can the government tell us how the funding works? Is it on a per-year basis? What is the maximum that kids can get at each age? What happens if a child is diagnosed at age 3.5? How much do they get for that year?

But remember, those budgets are maximums. The funding is means-tested, with families over $250,000/yr losing eligibility altogether. We’ve been told the phase-out starts at $55,000/yr income, but there’s no written confirmation of that anywhere. Nor do we have any idea out the phase-out formula works, and if there’s different formulas for couples vs. single-parents.

To cost this out, we need some idea of how much the average child would get, relative to the maximum. 80% seems like a generous estimate, so I will use that (I expect it’s significantly lower).

Can the government tell us how the means-testing formula works? At what income level does the clawback begin? Are there different formulas for couples vs. single-parents?

Next, we need to estimate how much of the budget kids would actually spend. In any program like this, there’s never 100% utilization. Some kids will receive more money than they need, some kids won’t be able to spend their money at all, because the therapy they need isn’t available. I’m going to assume a 90% utilization rate, which, again, I believe is generous.

Can the government tell us if they’ve forecasted utlization rates? If so, what are they projecting?

Finally, we need to know how many kids would be in the program and which age bracket they would be in. First, on number of children. The government has talked quite a bit about ~8,300 kids currently receiving treatment, and currently 23,000 on waiting lists. This would suggest they’re forecasting 31,000 or so children would be in the program.

Then we need to know their ages. If it is a 20/5 program, then it’s almost certain most kids would be in the older group, since the younger group only includes 4 years (ages 2–5) and the older group contains 12 years (6–17).

Assuming a 1:3 ratio would seem fair, but I’ll be generous and assume the proportion is more heavily weighted towards younger kids. So let’s assume there are 12,000 kids in the younger group and 20,000 in the older group, for 32,000 total.

Can the government tell us if they’ve forecasted how many kids will use the program, along with their ages?

I came up with a little costing tool in Google Sheets that anyone can use. It’s available here. I plugged in my parameter values and got the following:

Using these (I belive overly generous) parameters, I get that a new program cost of around $245 million. This is about $100M less than what I believe the old program would have cost.

I cannot state this enough: These are simply best guesses, in the absence of any information at all provided by the government. Let’s hope they provide more information so we have a better idea of what’s going on!

Of course, if you change the parameter values, you will get a different costing. And it is certainly possible to get a costing closer to 350M. Here’s one way to do it, while still keeping the 20/5 program:

  • Average budget as a % of maximum: 90% instead of 80%
  • Average utilization of budget: 95% instead of 90%
  • Number of younger kids: 15,000 instead of 12,000
  • Number of older kids: 25,000 instead of 10,000

Plugging those in, we get the following:

So now we have a program costing around 363M. Of course, we could move our parameters in the opposite direction by the same magnitude, and get a much smaller figure:

  • Average budget as a % of maximum: 70% instead of 80%
  • Average utilization of budget: 85% instead of 90%
  • Number of younger kids: 9,000 instead of 12,000
  • Number of older kids: 15,000 instead of 10,000

Now our estimated cost is much lower:

Given how sensitive the costing is to our choice of parameters, it’s really hard to know what the new program will cost. All the more reason why the government needs to be transparent, and give us more information!

All in all, I suspect we’re looking at a cut of 50–100M in program funding, but honestly, who knows?