Tales from the Mouth of the Fraser: Tax Cuts and Economic Growth

The latest from the Fraser Institute concludes that BC’s 2001 tax cuts spurred economic growth, investment and all other good things. To their credit, this is a far more sophisticated analysis than one usually sees from the Fraser. Rather than relying on the in-house talent, they farmed the work out to University of Alberta economist Bev Dahlby, known for advocating that Canada was on the wrong end of the Laffer curve back in the 1990s. Given that prior, it is not surprising that Dahlby comes to the view that tax cuts have been beneficial.

It is certainly plausible that tax cuts have had some benefit in the short run. Personal tax cuts amounted to $1.5 billion per year and corporate tax cuts $800 million. Fully phased in, they were worth almost 2% of BC’s GDP, and were fully deficit financed. The large deficits in the early years of the Campbell government (after two years of surplus in the final years of the “fiscally irresponsible” NDP) were masked in later years by a massive increase in both federal transfers and resource royalties. It really was not until 2005 that BC’s economy really got moving (the study goes to 2006), however, and even then the straight growth numbers are not hugely impressive and productivity numbers continue to be dismal (though falling unemployment much more prominent in this boom).

But to make a convincing supply-side case will take many more years of data. For other countries, over longer periods of time, this has been studied extensively, and there is nothing even close to consensus in the growth literature that tax cuts (smaller government) lead to sustained long-term growth rates. I’ve written about this here. To a large extent, a researcher in this area can find what they are looking for if they are willing to try out different specifications, functional forms and variables. There are studies that can be trotted out as supporting the Fraser view but more do not, and some come to the opposite conclusion. A dispassionate observer would likely conclude that the debate is, at best, inconclusive.

The press release plays up the benefit was from corporate income tax cuts, with personal income tax cuts as best supporting actor. The paper itself is more nuanced, in empirical content if not rhetoric. There are a number of regressions published in the paper and if you look closely, not all of their regressions have CIT and PIT as statistically significant at the 5% level, and the ones that are significant are just barely so. The single strongest explanatory variable in their regression is not taxes at all but US growth, and significant at a 1% level – a fact left out of the Fraser’s press release entirely. Even the finding on corporate tax cuts (which are “responsible” for about 0.3 percentage points of additional GDP growth in subsequent years, it is concluded based on their regressions) would seem to be in contradiction to their Figure 3 which shows essentially no relationship between growth rates and corporate tax levels (there is no basic correlation, so how they move on to conclude causation is striking).

But two other factors that are clearly behind BC’s boom are missing entirely from the regressions. The first is interest rates, which starting in 2001 dropped to lows not seen since the 1960s, boosting the housing market, and well, you know the rest. This is a much more persuasive explanation than tax cuts, but interest rates are not in any of the regressions, and the term appears exactly zero times in the study.

The second factor is commodity prices which surged over the period they are claiming credit for the tax cuts. Dahlby pretends to cover this off by including the “growth rate of export prices”. This is odd: if back in the 1970s, oil went from $4 a barrel to $6, this would show up as a 50% increase in Dahlby’s regressions. But a sustained price of $80 over several years would show up as zero. Clearly, they are measuring the wrong thing.

In econometrics they call this an omitted variables problem. You cannot make reliable claims on the numbers if you are missing key explanatory variables. Which is probably why their published r-squared numbers are so weak. I could point to other questionable methodological choices that may or may not affect the conclusions on taxes (use of five-year averages, the choice of which five years get grouped together, the absence of a time trend variable, the fact that their dummy variables for BC were not significant and subsequently left out of the final specifications). But those are just finer points when the basic model is wrong.

[Thanks to new CCPA economist Iglika Ivanova for talking through this with me.]