The Impact of a Revenue-Neutral Carbon Tax on GDP Dynamics: The Case of British Columbia.

AuthorBernard, Jean-Thomas
PositionGross domestic product

    The effect of environmental taxes on GDP is a major policy concern and it continues to generate heated debates in public squares. Arguments in support of positive and negative net effects are presented. On the one hand, there is the worry expressed by some that environmental taxes increase costs and reduce competitiveness, thus hurting the economy. On the other hand, there is the double-dividend economic argument, whereby not only can environmental taxes reduce negative externalities such as pollution and global warming, but they can also increase income. This can notably happen when the tax is designed to be revenue-neutral; the new tax replaces less efficient duties (such as those applied to personal and business income) ensuring no change in the government budget position.

    While theoretical and simulation-based studies have examined the expected effect of revenue-neutral environmental taxation on output, (1) there is little empirical work aimed at quantifying this effect. A possible reason may be the scarcity of real-world such experience, and consequently, of pertinent data. For one thing, the various forms of environmental taxes introduced by governments generally have not been designed to be budget-neutral. In addition, the definition of an environmental tax is not so clear-cut. For example, often governments justify taxes applied to gasoline and to diesel by appealing to environmental considerations. However, when these taxes are collected to also expand and maintain roads, they are really user fees. Alternatively, when the collected monies also serve to fund general government activities, the taxes cannot be qualified as being strictly environmental levies.

    In this paper, we rely on a unique policy enacted by the government of the province of British Columbia (B.C.) of Canada to study the effect over time of revenue-neutral environmental taxes on GDP. In 2008, B.C. implemented a carbon tax designed to be revenue-neutral on a broad range of greenhouse gas (GHG) emissions originating from fossil fuel use. The tax rate was set at $10/ton of C[O.sub.2] in July 2008 and then raised by $5 in July of every year to reach $30/ton in 2012. The price remained at that level until the end of 2017. In order to fulfill the objective of revenue-neutrality, the government lowered income tax rates to individuals and businesses, and delivered subsidies to low income earners. The two lowest personal income tax rates were thus decreased, providing a tax cut on the first $70,000 earnings of 2% in 2008, and 5.0% in 2009, respectively. The general corporate income tax rate was reduced from 12% to 11% on July 2008, and then to 10% on July 2011. Similarly, the small business income tax rate was reduced from 4.5% to 2.5% over the same period. The income threshold between small businesses and general corporations was lifted from $400,000 to $500,000. In addition, annual tax credits of $100 per adult and $30 per child were granted to low income residents. (2) Note that while the aim is to achieve revenue-neutrality, there exists no actual mechanism to ensure that collected revenues from the tax will exactly equal given tax reductions and credits.

    A few studies have examined impacts of this policy initiative on selected aspects of the B.C. economy. Some find declines in GHG emissions or in the demand of fossil fuels in the province. (3) Others focus on tax effects in the labor market and they document changes in employment in specific categories of industries and firms. (4) Welfare effects of the tax have also been examined, with the conclusion that households of varying income levels were impacted differently. (5)

    The above show that certain sectors or households gain from a revenue-neutral tax while others lose. But what is the net overall tax impact? Almost all policy actions end up by benefitting some economic agents while being detrimental to others. What matters most is whether a policy has a net positive impact, and by how much. Furthermore, focusing on average effects, as is the case with the majority of the above-mentioned studies, may not be enough. Economic agents may face constraints that bind at different points in time, and thus they may adapt to the tax changes in different periods, and by different amounts. In other words, time-varying considerations could also be important and thus merit attention. For example, such information can be useful to policymakers for the design of future tax strategies.

    Little attention has been given to the big picture, namely the effect of the carbon tax on the province's GDP. An early exception is Elgie and McClay (2013) who observe that, over 2008 to 2011, per capita GDP shrank by an average of -0.15% in B.C. compared to -0.25% in the rest of Canada (ROC), and that, at the same time, per capita GHG emissions fell by 10% in B.C. compared to 1.1% in the rest of Canada. (6) The authors thus conclude that the B.C. carbon tax must have had no negative impact on the province's GDP. More recently, Metcalf (2019) estimates the average effect of the B.C. carbon tax on the province's GDP in the context of a number of static panel models for Canadian provinces. The significance of the estimated impact depends on the particular specification adopted and on the sample period used. Another somewhat relevant paper is Beck et al. (2015) who focus on the distributional household welfare impacts of the tax. Simulations based on their static computable general equilibrium (CGE) model yield an aggregate household welfare loss of 0.08% and show that the tax is progressive. (7)

    While useful, the above works measure either average or static effects, and thus they ignore possibly important dynamic considerations. Furthermore, Elgie and McClay (2013) and Beck et al. (2015) do not make use of statistical testing to evaluate their results formally. As for Metcalf (2019), conclusions regarding the importance of the tax effects appear not be robust, given that relatively minor changes in specification (such as using different tax measures, or omitting a province from the analysis) yield different outcomes.

    We resort to time series methods to quantify statistically the impact over time of the B.C. carbon tax on the province's GDP. After constructing suitable aggregate energy price and aggregate carbon tax series for B.C., we study the impact of the tax changes on GDP changes in the context of a vector autoregression (VAR) framework. Such models are well suited for our purposes since they can allow for rich interactions amongst variables of interest while remaining parsimonious and easy to implement statistically. Specifically, we build on the canonical VAR framework of Kilian and Vigfusson (2011) by explicitly integrating into our model possible effects of tax saliency, tax pre-announcements, and tax pass-through. To our knowledge, such considerations have not previously been examined in a VAR framework. Finally, our specifications also allow for open economy considerations, which are important for small open economies like British Columbia.

    Based on impulse response analysis, and on counterfactual investigations, we find that: (i) the B.C. carbon tax has had no overall statistically significant dynamic effect on monthly GDP changes of British Columbia, (ii) in the few months where carbon taxes do have a statistical impact, GDP changes are mostly positive, and (iii) there is complete pass-through of carbon taxes into energy prices paid by consumers.

    The paper proceeds as follows. In Section 2 we describe the data, including the details of the construction of the aggregate price and tax series. The analytical framework and model specifications are presented in Section 3. In Section 4 we discuss the results of estimated impulse responses and of a counterfactual exercise. Finally, section 5 concludes.

  2. DATA

    Our sample consists of monthly data for the province of British Columbia over the period extending from January 1987 to December 2016. These series include domestic sales (in litres or [m.sup.3]), energy prices (inclusive of all taxes, in cents/litre or cents/[m.sup.3]), and carbon taxes ($/ ton of C[O.sub.2] equivalent emissions) for regular grade gasoline, diesel, and natural gas (for the residential, commercial and industrial sectors). We also obtain the province's consumer price index (CPI) and population. Other monthly series that we make use of are: The Case-Shiller U.S. National Home Price Index, U.S. housing starts, as well as the global measure of economic activity developed in Kilian (2009). Finally, we have quarterly data on real gross domestic product (GDP; 2007$) for British Columbia, which we linearly interpolate to obtain monthly figures. The appendix describes the sources of all these variables.

    Our raw series are transformed as follows. We first express carbon taxes for each energy source in comparable price units and subtract these from corresponding total prices to obtain net-of-carbon-tax prices for each of our gasoline, diesel, and natural gas series. The province's CPI is then applied to all of our price and tax series to obtain their real counterparts. In the case of the tax series, and for reasons explained in Section 3, we also generate the real counterparts of the 4-quarter lead series by dividing nominal t+4 tax values by time t CPI values, and which we refer to as real lead-4 carbon taxes, (denoted [[~.T].sub.t+4] in the equations). For sales and GDP data, we obtain per capita terms. Using 2008 conversion factors provided by Statistics Canada, we convert our real prices and carbon taxes into the same energy units (cents/MJ; MJ stands for MegaJoules), and our per capita quantities into units compatible with our prices (MJ). (8) Except for real carbon taxes, we deseasonalize all our series (including per capita quantities) using the ARIMA-X12 filter. (9) Total deseasonalized real prices are constructed by adding together deseasonalized real...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT