Politicians around the world like to argue that climate policy will create jobs and stimulate innovation. Such a message is largely unsupported but more palatable than the typical result of academic research that shows that climate policy would increase the costs of energy and slow down economic growth (Clarke et al 2009).
Most studies of the impact of climate policy are ex ante, for the simple reason that greenhouse-gas emission reduction is still something that is planned for the future rather than achieved in the past. Models are calibrated or estimated to data, of course, but by necessity have a stylised representation of reality.
In a recent paper (Commins et al 2011), we estimate the impact of energy taxes on corporate behaviour. We use the AMADEUS database of firm-level panel data for publicly listed European firms across various sectors. The data cover financial and economic information for around 11 million firms for the years 1996–2007.
Energy taxes are similar to a cost-effective climate policy. The main difference is that carbon pricing changes the relative prices of alternative energy sources in a specific way, whereas energy taxes affect relative prices in an arbitrary manner. Unlike price shocks on world energy markets, energy taxes and carbon pricing raise revenue in the domestic economy.
We test four hypotheses:
- Do energy taxes create employment?
- Do energy taxes stimulate innovation?
- Do energy taxes increase profits?
- Do energy taxes raise investment?
We estimate four models. The model of employment tests whether energy and capital are highly complementary (Koetse et al 2008), with firms switching to more labour-intensive activities as energy prices rise. On the other hand, firms may move activities to countries with lower prices (Eskeland and Harrison 2003).
The second model is of changes in total factor productivity (TFP), a measure of innovation. According to the Porter Hypothesis (Porter and van der Linde 1995), regulation forces firms to innovate themselves out of a tight spot – that is, overcome an initial setback. More conventionally, regulation creates a wedge between firm performance and the production frontier.
For our third model, return on capital employed is included as a profitability indicator. Energy taxes would be expected to decrease profitability under the assumption that taxes act as an additional cost on doing business.
For our final hypothesis, we model that energy taxes would have a negative effect on investment as firms substitute capital for labour and locate elsewhere. However, the Porter Hypothesis would suggest investment in new technology in order to improve productivity.
We use panel data to control for unobserved time- and company-specific heterogeneity. We include sector-tax interaction terms for energy tax levels and lagged tax levels. We use first differences to control for autocorrelation and non-stationarity, and for unobserved firm-level heterogeneity. We include a range of other explanatory variables.
Figure 1 shows the impact of energy taxes on employment by sector. There are substantial sectoral differences. Most effects are statistically significant. Some sectors show a positive effect; notably wearing apparel, textiles, and primary-resource sectors. Labour, typically less skilled in these sectors, may be used as a substitute for energy. Air transport shows a strongly negative association, which may be spurious – the sample period coincides with the rise of the discount airlines. Other sectors exhibit weaker negative effects, notably machinery and construction. For these sectors, higher energy costs imply an overall contraction. For employment, the average effect is negative and significant. Overall, increased energy taxes reduce employment.
Figure 2 shows the impact on TFP growth rate by sector. Sectoral variation is large. Primary resources, manufacturing, power generation, pulp and paper and the media sector all show a positive effect. Most of these sectors are energy-intensive. This suggests that higher energy costs forced these firms to innovate. However, energy taxes reduce input price variability and raise barriers to market entry; both could increase TFP as measured here. TFP in chemicals, textiles, wearing apparel, leather and quarrying shows a negative association with energy taxes. These sectors are in decline in Europe, and higher energy taxes seem to spur this trend. The average effect of a tax change on TFP growth is positive. Overall, higher energy taxes accelerate TFP growth in Europe.
Figure 3 shows the impact on the return on capital employed by sector. This is positive in most cases, particularly for tobacco and air transport. Air transport is exempt from energy taxes. Higher energy taxes, however, imply a lower demand for energy and hence a lower price of kerosene. Higher energy taxes also reduce the price of air travel relative to road and rail. However, the effect may be partly spurious because of the restructuring of the airline industry. Water transport, wood products, quarrying, and refining have significant negative coefficients. Faced with higher energy costs, firms in these sectors take a hit on the bottom line. However, other firms see positive effects as they are able to pass on the higher costs to their customers and take advantage of improved productivity. The average effect is positive. Overall, higher energy taxes increase profitability in Europe.
Figure 4 shows the impact on corporate investment. The tobacco sector again stands out, but this sector was in decline too for reasons other than energy taxes. It may be that energy taxes were raised faster in countries that reduced smoking most substantially. Textiles and plastics show a negative impact – these sectors substitute energy and capital for labour. In contrast, metal mining, gas extraction, basic metals, refining, and water transport have large positive coefficients. High energy taxes spur investment, we presume in primarily in energy-saving equipment. The average effect is positive. Overall, higher energy taxes spur investment in Europe.
We consider whether the sectoral pattern of energy tax effects shown above might be driven by broader sectoral characteristics such as energy intensity or technology intensity. However, grouping sectors by these classifications did not reveal any obvious association with the tax effects. The impact of energy taxes on employment, TFP growth, profitability, and investment vary by sector even amongst industries which have similar energy and technology use.
In sum, the following results emerge.
- First, results vary dramatically between sectors, both in size and in sign.
- Second, energy taxes reduce employment.
- Third, TFP accelerates with higher energy taxes.
- Fourth, energy taxes increase the returns to capital.
- Fifth, energy taxes increase investment.
Overall, these results provide only mixed support for the political ‘green growth’ agenda. Energy taxes appear to favour capital over labour. Innovation accelerates, but it may not be the type of innovation that spurs economic growth.
Figure 1. Impact of a 1% rise in energy taxes on employment by sector
Figure 2. Impact of a 1% rise in energy taxes on TFP growth by sector
Figure 3. Impact of a 1% rise in energy taxes on return on capital by sector
Figure 4. Impact of a 1% rise in energy taxes on investment by sector
Clarke, L, J Edmonds, V Krey, R Richels, S Rose, and M Tavoni (2009), "International climate policy architectures: Overview of the EMF 22 international scenarios", Energy Economics, 31(S2):S64-S81.
Commins, N, S Lyons, M Schiffbauer, and RSJ Tol (2011), "Climate policy and corporate behaviour", Energy Journal, 32(4):51-68.
Eskeland, GS and AE Harrison (2003), "Moving to greener pastures? Multinationals and the pollution haven hypothesis", Journal of Development Economics, 70(1):1-23.
Koetse, MJ, HLF de Groot, and RJGM Florax (2008), "Capital-energy substitution and shifts in factor demand: A meta-analysis", Energy Economics, 30(5): 2236-2251.
Porter, ME and C van der Linde (1995), "Towards a New Conception of the Environment-Competitiveness Relationship", Journal of Economic Perspectives, 9(4):97-118.