Major Numerical Problems in Tim Hudak’s Jobs Plan

When Ontario PC leader Tim Hudak kicked off the current election campaign with a plan to “create a million new jobs” in Ontario, he tried to dress up the platform launch with a certain scientific respectability.  The party released a “technical backgrounder” showing the precise composition of the million new jobs, along with two commissioned consultants’ reports that were said to justify the estimates contained in the plan.

I cannot find either the backgrounder or the consultant reports on the PC site (perhaps, as you read on, you will agree there is good reason for this), but they were circulated widely to reporters and others when the detailed plan was announced on May 14.  They have also been helpfully posted on-line by David Reevely of the Ottawa Citizen (who wrote about the plan).  Here are links to the technical backgrounder, the report from the Conference Board of Canada (estimating the job-creation effects of personal and corporate income tax cuts), and the report from U.S. economist Benjamin Zycher (from the right-wing American Enterprise Institute) estimating the job-creation effects of lower electricity prices, free trade with western Canada, corporate income tax cuts, and reduction of regulations.

There are many important economic and methodological issues with the policies proposed by the PCs, and the way they have used the consultant reports to try to justify their numbers.  These problems deserve a lot of further study, including:

• The apparent double-counting of fiscal savings (like the wage freeze on public servants).
• The failure to account for any employment downside from major public sector spending cuts.
• The failure to include the 100,000 direct public sector jobs that the PCs pledge to eliminate.
• The lack of any empirical support for the plan’s estimates of job gains from apprenticeship training or GTA transit initiatives.
• How they measure “regulatory reduction” and its impact on employment.  (Reevely also wrote a column on that topic, too.)

Ontarians can and should have a serious policy debate over whether Hudak’s proposals would have any stimulative effect on employment at all (let alone creating a million new positions).

However, those substantive policy debates may be swamped by an enormous arithmetic mistake which can be deduced by comparing the numbers in the PC technical backgrounder, with the supposedly supporting evidence contained in the two consultant’s reports.  Ontario Finance Minister Charles Sousa has critiqued this error.  Here are the details as I read them.  In essence, the two consultants generated estimates of the number of cumulative jobs that would be created from the simulated policies over several years.  But the PCs have assumed that those jobs would be created in each year of the 8-year plan.  They have thus exaggerated the number of jobs by several times (as many as eight-fold).

Let’s start with the Conference Board report on the employment effects of a 1-percentage-point corporate income tax cut.  Their results are reported in their Table 5.  That table (as it notes) shows the “level difference” of each variable versus the base case scenario: that is, it shows the running impact of the one-time policy change.  In the lower half of that table is a line titled “Employment.”  That shows how much the level of employment is higher in each year of the 10-year simulation, compared to what it would be without the 1-point corporate tax cut.  There are 2071 extra jobs in the first year, and the difference (compared to the base case) grows gradually over time (on the assumption that corporations respond to the lower tax rate with more investment spending over time — even though in practice that has not happened with past CIT cuts).  By the 10th year there are 5869 extra jobs.  In the 8th year (which is 2020 in their scenario, assuming the tax cut was implemented in 2013), there are 5323 extra jobs.  If you believed the Conference Board report, therefore, you should claim that a one-point CIT cut should create 5323 jobs by the eighth year after its implementation.

The Conservatives are promising a 3.5 point cut, so we should multiply that number by 3.5, giving a total of 18,631 new jobs.  This assumes the full 3.5 point reduction is implemented in the first year (whereas the PCs have said it will be phased in over time), so in reality even that number is too high.

Now, the PC backgrounder claims (referring to the Conference Board study for evidence) that the CIT cut will in fact produce 119,808 jobs over 8 years.  Where did that number come from?  A little forensic accounting can answer the question.  The last column of the Conference Board Table 5 reports something called “Cumulative Total.”  The cumulative total for the employment line (as explained in the text of the Conference Board report at the bottom of p.9) represents the total additional person-years of employment created over all 10 years by the 1-point tax cut.  Over the 10 years combined, there were 42,788 additional person-years of employment thanks to the tax cut.  There were only 5869 new jobs by the 10th year, but counting each year for each worker provides the number of person-years.

The PCs took the 10-year cumulative increase in person years, divided it by 10 (to get an “average annual increase in person-years,” a very odd concept), and then multiplied it by 3.5.  That equals 14,976 — the number listed in the PC technical backgrounder.  Then they assumed that many jobs would be created each year!  Multiplied by 8, that equals 119,808.  That is a gross and obvious mis-reading of the Conference Board’s own results.

Even if we believe the findings of the Conference Board study, therefore, the PCs counted over 100,000 jobs too many from a 3.5 point CIT cut.  The Conference Board estimated 18,631 jobs after 8 years, but the PCs claimed 119,808.

It gets better, because it is clear that the PC technical backgrounder made exactly the same mistake with the other one-time cumulative employment gains that were estimated by the two reports to result from other policy measures (including personal income tax cuts in the PCs’ dreamed-of second term, lower electricity prices, free trade with western Canada, and reduction of the “regulatory burden”).  Even if you accept the findings of those two studies (which I certainly do not, for good theoretical and methodological reasons), the Conservatives counted re-occurring gains in each year when their own consultants were projecting one-time cumulative job gains.

The Zycher report is not described as clearly or professionally as the Conference Board report, but it is still possible to ascertain the methodology he followed.  He estimated an econometric regression of the level of provincial GDP, based on numerous variables (including policy variables like tax rates, electricity prices, and whether or not a province was part of the western Canada free trade area).  He then estimated the one-time change in the level of GDP expected to result from changing those various input variables, as promised in the PC platform.  Those changes in GDP were then converted into expected changes in the level of employment on the basis of an employment multiplier (obtained from an employment regression which Zycher also estimated).  (Zycher follows a different and much stranger methodology to estimate the effects of reducing regulations; the end result is still measured as a change in the level of GDP and employment.)  It is clear that both the change in GDP and the corresponding change in employment are one-time cumulative effects resulting (after whatever period of adjustment is required — Zycher does not specify this) from the simulated policy change.  Yet the PC technical backgrounder has added in the job estimates from those changes (5048 from electricity price reductions, 199 from free trade with western Canada, and 10,600 from regulatory reduction) as occuring in each year of the 8-year timetable.

Across these 4 additional policy items (PIT cuts, regulatory reduction, electricity prices, and interprovincial free trade) the PC backgrounder has over-counted the job projections of their own consultants by a combined total of more than another 100,000 jobs.

The million jobs plan, therefore, now has a gaping 200,000 job hole in it, resulting from an obvious arithmetic error that throws into question the very competence of Mr. Hudak’s policy team.

On top of that, of course, is the fact he hasn’t accounted for his promised 100,000 public sector job losses.  And he claimed over 500,000 jobs as being part of his plan, which he acknowledged were going to be created anyway (and thus can hardly be ascribed to PC policies).

Therefore, the million promised jobs has shrunk to 200,000 jobs.  And when we start taking a hard look at the modeling methodology behind each of the remaining numbers in his plan (especially by carefully reviewing the approach used in Zycher’s paper, which in many respects is bizarre and ideological — stand by for more work on this), those remaining jobs are going to disappear pretty quickly too.

There have been many comparisons made between Tim Hudak’s campaign and that of his mentor, Mike Harris, who rode to power in 1995 on promise of a “Common Sense Revolution.”  Many observers have pointed out that Hudak’s current plan for public service cuts and deregulation is actually more extreme than Harris’s.  But there is another important difference between the two campaigns.

Mike Harris’s economic numbers were crunched in-house by a real economist, Mark Mullins, who left a career on Bay Street to work for the Conservatives (and later did a stint for the Fraser Institute).  Agree with him or not, Mullins was a highly competent economist who understood how forecasting worked, and prepared a fiscal and macroeconomic spreadsheet backing up Harris’s plan that was internally consistent, added up, and released in detail.  Mark Mullins would never have made a mistake as blatant and self-destructive as confusing jobs with person-years of employment.

This crew of PCs, in contrast, farmed out their analysis (largely to a U.S. economist long associated with extreme right-wing statements and causes), and couldn’t even correctly interpret the results that they themselves commissioned.  Not only is Tim Hudak’s policy platform more extreme than Mike Harris’s.  It is also clear that his policy team is not remotely as competent.

No wonder I can’t find those documents on the PC website.