Ontario HST Study RejectedA debate has broken out over the question of whether $2 billion per year in tax savings for businesses in BC ($4.2 billion in Ontario) are passed-through in the form of lower prices for consumers. Some economists argue that families are confused over the high visibility of the HST, which hits them in the pocketbook on everything from vitamins to soy milk-substitute (which is not a “basic grocery”). Ignore the anecdotal evidence they say, trust us that a lot of small price cuts are reducing your pain. I say trust no one, least of all economists, and base your referendum vote on how the HST has affected you.
The argument from economists Jon Kesselman and Michael Smart is that although some prices rose initially because the new tax is on an expanded tax base (most services and some previously PST exempt goods like bicycles and vitamins), that increase was offset by the pass-through of business tax savings in the form of many small price cuts. Some studies, including the government's independent HST panel chaired by Jim Dinning, assumed that up to 90% of business tax savings is passed-through to consumers. Where is the evidence to support that assumption?
The initial price rise is easy to understand. If a new 7% tax (8% in Ontario) is applied to 20% of what you buy, then you might expect your cost of living to increase by 1.4% (20% times 7%). The evidence shows initial price increases are closer to 1.0% because on average the increase applies to slightly less than 20% of expenditures (although more for some people) and, most importantly, depending on the characteristics of each market, taxes may not ever be passed-through, may not be passed-through in the first month the tax is effective or may not be measured by the consumer price index (CPI).
In his article on the initial evidence on the implementation of the HST in Ontario, Michael Smart has a good categorization of various types of markets and how they deal with the pass-through of taxes (p. 8). Smart's article is also worth reading for its clear description of various features of the CPI. He is an economist who understands the strengths and weakness of the data he uses. For example, in a footnote on input tax credits (ITCs p. 10) he wrote: “In other words, the overall effect of ITCs on prices simply cannot be measured accurately with the CPI data.” He went on to argue that excluding sectors that could benefit from ITCs because they aren't affected by an increase in tax could understate price-reducing effects.
Kesselman looked at changes in the CPI or its components from July (when the tax was implemented) to December and “standardized” that change relative to a control group (an average of Alberta, Saskatchewan and Manitoba). Smart did the same for Ontario for the same period using Quebec as a control group. They then argued that to the extent inflation was lower than the control group from July to December it measured a pass-through of business tax savings and mitigated the June to July price increase. The flaw in their argument is the July to December price differences between a new HST province and the control group could be due to many things. Businesses get their tax break in the form of deductions for the HST they pay, ITCs. To Smart's credit he wrote (p. 13): “... it is difficult to distinguish the effect of ITCs from other, unobserved factors that may have affected Ontario's general rate of inflation over the last five months of 2010.” That is the point I made when I applied Kesselman's method for analyzing the BC CPI to 30 years of data and found almost two out of three times picking a month at random from any month over the last three decades would yield the same result Kesselman creatively interpreted. Surprisingly, while aware of the limitations of the CPI and the statistical method, Smart also wrote: “The observed diminution in tax effects over time is consistent with a gradual pass-through of input tax credits to lower consumers prices.”
HST proponents need to prove that differences in inflation between BC or Ontario and control groups subsequent to the implementation of the tax is due to the pass-through of cost savings rather than due to random fluctuations. I repeated the analysis I did on Kesselman’s study and found that, as was the case with Kesselman, Smart’s conclusion is based on a difference that frequently occurs at random. At least Smart included a footnote warning of that possibility. Smart simply looked at the difference in the percentage change in the CPI between Ontario and Quebec. He reported: “… the month-over-month increase in Ontario’s CPI was 0.9 percent in July, compared to a CPI decrease of 0.3 percent in Quebec. Under the assumption that Ontario’s prices would have evolved in the same way as Quebec’s in the absence of reform, we therefore estimate that tax harmonization caused a 1.2 percent increase in Ontario’s prices in July.” He went on to say: “The cumulative difference in Ontario's CPI between June and December was 1.6 percent compared to 0.9 percent for Quebec, suggesting that the impact of the tax harmonization after six months was to increase Ontario's overall price level by just 0.7 percent.” Actually the difference was 0.792 percent when the calculation is done to three decimal points.
The problem of interpreting the difference in the six month percentage change in the CPI between Ontario and Quebec as evidence of businesses passing-through tax savings in the form of lower prices is, as Smart acknowledged, that there are many other factors that could account for the difference. The chart shown here plots the difference in six month percentage CPI changes between Ontario and Quebec for both all items and services. Over the past 30 years, there were 351 times when the six month CPI difference between Ontario and Quebec was less than 0.792 percent, and only 51 times when it was greater. Looking at services where the June to December 2010 difference was 0.792 percent, there were 300 times when the six month difference was less than 0.792 percent, and only 59 times when it was greater. It looks like there is a chance that random events could have produced the same figures that Smart used to suggest evidence “consistent with a gradual pass-through of input tax credits to lower consumers’ prices.”
It is possible to apply a much more difficult test. It is interesting to look not just at the difference between Ontario and Quebec in six month percentage changes in the CPI but at the number of occasions when the second month of the six shows a difference in the percentage increase of more than one percent only to end the period with a difference of less than one percent. In other words, circumstances similar to what Smart observed for the all item CPI between June and December. There are only five occasions in the past thirty years when the month-to-month increase in the all item CPI was one percent or more greater for Ontario than for Quebec. In three of those occasions, including December 2010, the period ended with a smaller difference over six months than the difference at the beginning of the period. It is not sound to draw any conclusions from so few observations, but to the extent one can, the data are consistent with the phenomenon that Smart observed occured by chance. The five months are: March 1982, June 1983, May 1988, February 1994 and July 2010. The 1982 and 1983 periods ended like the recent example with a lower difference after six months. There were no sales tax changes in Ontario in 1982 or 1983.
Smart attempted to construct a price index that would concentrate the effect of the HST by taking a weighted average of 16 sub-components of the CPI, using the percentage each component represents in the CPI basket of goods and services. From that index, taking the difference between Ontario and Quebec, he concluded: “.. the best estimate is that harmonization caused prices to rise by 0.9 percent in July, falling to 0.6 percent in December.” There are problems with the index Smart constructed. He chose sub-components based on his assessment of whether it included items that were subject to the HST, but not previously subject to the PST. The problem with that method is mistakes are made, both by Smart and by Statistics Canada. For example, Smart excluded food purchased from stores with the note “none” under examples of newly taxed items, but only “basic” groceries are exempt. Many food and beverage items purchased in a grocery store are subject to the HST, e.g. non-dairy milk substitutes such as soy drink. Smart wrote: “Statistics Canada does not sample prices of all commodities in all provinces on a monthly basis, due to its limited resources for data collection and because prices for some commodities (especially services) change infrequently. The agency did sample some service prices extraordinarily in July 2010 to capture price changes around the introduction of the OHST, but some reported CPI changes in July may reflect inferences based on how tax rates changed under OHST, as well as actual prices observations.”
Rather than repeating Smart’s calculations to compose an index based on the 16 sub-components he used, I simply looked at the CPI for services as published by Statistics Canada on CANSIM. The difference between Ontario and Quebec for June to December in Smart’s index was 0.6 percent; the difference using the CPI for services is 0.792 percent. The CPI for services difference is shown in the graph here because it is arguably as close to Smart’s index for tracking items most subject to the effect of the HST.
My conclusion is that the proponents of HST have failed to make a case that 90 percent of business tax savings have been passed-through to consumers in the form of lower prices. I have shown that the price changes they identified can be due to random events or unidentified factors.