Evidence is still missing

Over the past two weeks, numerous politicians (mostly federal Conservatives) have vilified our study, Did Government Stimulus Fuel Economic Growth in Canada? After all, our conclusion that government stimulus in Canada had virtually no impact on last year's economic turnaround runs counter to the current wisdom in Ottawa. While politically motivated attacks on sound, rigourous research is somewhat understandable, we expected more from Serge Coulombe, University of Ottawa economics professor.

Professor Coulombe's criticism starts by calling our methodology for analyzing Statistics Canada data a tricky data mining exercise and odd. However, the methodology we use is actually standard. For example, Stanford University professors John Cogan and John Taylor and Goethe University professor Volker Wieland, all internationally renowned macro-economists, use the same methodology to assess the U.S. government's stimulus package.

To gauge whether government stimulus had an impact on the economic turnaround in 2009, we examined Statistics Canada data on the contributions of government consumption (i. e., spending), government investment (i. e., infrastructure) and private sector activity to the improvement in economic growth.

For example, between the second and third quarter of 2009, GDP growth improved from -0.9% to 0.2%. Of this 1.1 percentage point improvement in GDP growth, government consumption and government investment each contributed only 0.1 percentage points. The same analysis was done for the improvement in GDP growth from the third to fourth quarter, during which government consumption and government investment contributed nothing.

Professor Coulombe dismisses this analysis by stating that by our account government expenditures would have contributed to the recovery if and only if the expenditures accelerated between the second and the fourth quarters of 2009. But that is exactly what stimulus spending is supposed to do. By definition, stimulus spending means speeding up the increases in government spending.

Not so, according to Professor Coulombe, who thinks that a steady increase in government expenditures would be enough to show that stimulus works. The problem with this logic is that government expenditures almost always increase. And they certainly did under the Conservative government and well before the recession. Consider that from 2005/06 to 2008/09 federal government spending increased by an average of 4.6% a year. Steady increases in government spending over time (regardless of economic conditions) is the reason the percentage point contributions to GDP growth of government consumption and government investment fluctuate very little on a quarterly or annual basis.

As an alternative to our methodology, Professor Coulombe proposes a straightforward approach to analyzing the impact of government stimulus spending by examining what Canada's GDP growth would have been in the third and the fourth quarters of 2009 without the contribution of government expenditures. By deducting the contributions of government consumption and government investment to GDP growth in the third and fourth quarter of 2009, he concludes that government stimulus played an important role in the economic turnaround of last year.

This analysis is simply wrong. Since government consumption and government investment are components of GDP that regularly contribute to its growth, simply deducting their contributions in a given quarter does not tell us whether government stimulus contributed to the change in GDP growth last year.

Most surprising however, is that Professor Coulombe fails to understand that subtracting the contribution of government spending to GDP growth simply cannot be done. The reason being, government spending reduces other components of GDP such as consumer spending and private sector investment. Professor Coulombe makes the all-too-common mistake of assuming the resources to finance government spending magically appear.

Professor Coulombe then dismisses the large body of academic research on fiscal stimulus that has found government spending is an ineffective tool for combating recessions. He assures us that because the interest rate charged by the Bank of Canada is near zero, this time will be different.

To support this view, professor Coulombe refers to a single study whose authors even acknowledge that viewed overall, it is hard to argue, based on the literature, that the government spending multiplier is substantially larger than one. In plain English, the authors concede that the evidence suggests that government stimulus spending likely doesn't work.

But the authors proceed to construct a theoretical model of a hypothetical world and use abstract reasoning to show that the government spending multiplier could be larger when the interest rate is near zero. Based on their model, the authors conclude the government spending multiplier can be larger than normal if certain, highly unlikely assumptions hold. They also conclude that [their] analysis abstracts from a host of political economy considerations which might make an increase in government spending less attractive than [their] analysis indicates.

While Professor Coulombe advises not to throw the baby out with the bath water, he is guilty of doing so by brushing aside the findings of several decades of empirical research on government stimulus. Instead he opts for spending billions of taxpayer dollars without any strong empirical evidence to show that stimulus spending works.