Climate Change 2001:
Working Group III: Mitigation
Other reports in this collection

8.2.2.3 The Distributional Effects of Mitigation

A policy that leads to an efficiency gain may not improve overall welfare if some people are in a worse position than before, and vice versa. Notably, if there is a wish to reduce the income differences in a society, the effect on the income distribution should be taken into account in the assessment.

An evaluation of the distributional incidence of higher energy prices is significantly conditional upon the indicator used. Distributional impacts appear to be higher when additional costs are measured in terms of percentage of total household expenditures rather than income, and higher if current income is considered instead of lifetime income. Lifetime income is relevant in the sense that households can borrow or save, and also move between different income classes. According to Poterba (1991), a person had only a 41% chance of being in the same quintile of income distribution in 1971 and in 1978. This percentage rises to 54% if the person initially belongs to the poorest quintile. However, current income is relevant to studies on the short-term intergeneration impact of a new tax. For example, an elderly person is more adversely affected by new taxes on expenditures than are those on an income, even if subsequent generations pay the same lifetime tax bill under each factor influencing the macroeconomy.

International competition limits the ability of firms to pass the tax onto prices, thus reducing the size of the indirect distributional effect. In the same way, the degree of production factor substitution determines the extent to which the tax changes prices. Moreover, as the substitution is generally supposed to be limited in the short term, but increasing as existing plants are replaced, the distributional effect of an environmental tax changes over time. Last, but not least, the distributional effect depends basically on the utilization of the tax revenue.

Two British studies looked at distributional effects of climate policies. Barker and Johnstone (1993) investigated the distributional effects of a carbon–energy tax. Revenues are recycled through an energy efficiency programme and compared to lump-sum transfers. The results show that the burden of a carbon–energy tax falls most heavily on low-income groups. At the same time, for these low-income groups the potential gains to be realized by increasing energy efficiency are higher to offset this regressive outcome. Symons et al. (1994) investigated other various assumptions of revenue recycling for the UK, and found that to introduce a carbon tax without recycling or with recycling through VAT or petrol excise-duty reductions is significantly regressive. Conversely, recycling the carbon tax by a combination of VAT rate reductions and benefits reforms directed towards poorer households results in favourable distributive effects.

The conclusion is similar for other countries. For Ireland, O’Donoghue (1997) found that carbon taxation is generally regressive, but that recycling the carbon tax through a fixed basic income for all individuals allows the distributional effects to become almost neutral. For Norway, Brendemoen and Vennemo (1994) concluded that a global carbon tax of US$325/tC in 2000 and US$700/tC in 2025 (1987 prices) has no significant impact on the regional distribution of welfare. For Australia, Cornwell and Creedy (1996) found that a carbon tax only affecting households (the input–output matrix is constant in their model) is clearly regressive, but can become neutral if adequate recycling is implemented. In addition, the distributional differences across income are not affected much. On the other hand, Aasness et al. (1996) conclude for the same country that poor households are less favourably affected than rich households, because of smaller budget shares on consumer goods (which imply relatively more CO2 emissions) in the rich households. Harrison and Kriström (1999) studied the general equilibrium effects of a scenario in Sweden in which the existing carbon taxes increase by 100% and labour taxes are reduced to maintain constant governmental revenue, but without removing the existing exemptions from carbon taxes. All households lose from this carbon tax (with tax exemptions) increase. They point out that the distributional effects are very dependent on the size of the household (the more affected being those with children). In a study for 11 EU member states, Barker and Kohler (1998) examined emission reductions of 10% below baseline by 2010. They concluded that the changes would be weakly regressive for nearly all the member states if revenues are used to reduce employers’ taxes, and strongly progressive if they are returned lump-sum to households.

In summary, most studies show that the distributional effects of a carbon tax are regressive unless the tax revenues are used either directly or indirectly in favour of the low-income groups (see also Poterba, 1991; Barker, 1993; Hamilton and Cameron, 1994; OECD, 1995; Cornwell and Creedy, 1996; Oliveira Martins and Sturn, 1998; Smith, 1998; Fortin, 1999). This undesirable effect can be totally or partially compensated by a revenue-recycling policy if the climate policy is implemented through carbon taxes or auctioned permits.

Three other issues of distributional effects, not dealt with here, are industry sector impacts, regional effects, and how people are affected by environmental damage. For instance, a tax on CO2 emissions obviously leads to very different effects in energy-intensive industries than in sectors producing labour-intensive services (see Chapter 9). In addition, the poor household generally lives in the most polluted area and then benefits first from the amelioration of air quality induced by GHG reduction policy (see Section 8.2.4).



Other reports in this collection