3.2  Environmental vs. economic risk

3.2.1  Tax versus tradable permits under uncertainty

Theory in standard domestic regulation situation

In a scenario where the government has complete knowledge about the aggregate costs and benefits to the country of reducing greenhouse gas emissions, producers know their own abatement costs, and where there is perfect certainty, taxes and tradable emission permits achieve exactly the same abatement at the same cost. However, the government is likely to know less than producers about the costs of emission abatement, and even producers have imperfect knowledge about these. Furthermore, no one is likely to know the true benefits of abatement with any degree of certainty. Constantly developing technology and an otherwise dynamic, changing world mean neither costs nor benefits are able to be learned precisely, as both are perpetually buffeted by shocks.

In these circumstances, a price instrument such as an emissions tax and a quantity instrument such as emission permits offer different guarantees and different risks. A tax limits the marginal cost of abatement, thus providing more economic certainty while making no guarantee about emission levels. On the other hand, a permit system limits total emissions, thus providing environmental certainty, but does not necessarily limit the marginal cost of achieving the desired abatement level.

If the government chooses quantity regulation, the optimal choice of quantity is that which maximises the expected benefits minus the expected costs. Similarly, the optimal tax under price regulation is that which maximises the expected benefits minus expected costs, given the way producers are expected to respond to the tax. A classic result shown in Weitzman (1974) is that uncertainty in the costs of abatement, but not uncertainty in the benefits, can have significant effects on the choice between emission taxes and permits, provided shocks to benefits and costs are uncorrelated. The intuition behind the idea that marginal benefit uncertainty does not affect the choice of instrument is that with neither a tax nor a permit system do firms change their behaviour in response to a marginal benefit shock. Thus both instruments yield the same social loss when marginal benefits are not equal to their expected values.

In the unlikely event that the costs of abatement were known with certainty, taxes and permits would yield the same environmental and economic results. However, the greater the uncertainty over the position and shape of the cost curve, the greater the difference between the outcomes likely to be achieved with a quantity instrument and with a price instrument.

Whether a tax or a permit system is preferable depends critically on the slopes of the marginal cost and marginal benefit curves of abatement. In the absence of other considerations, a quantity instrument is preferable if and only if the marginal benefit curve is more steeply sloped than the marginal cost curve. This is illustrated in Figure 2 for marginal cost and marginal benefit curves with different relative slopes. In each panel, MC_E represents the expected marginal cost curve, and MB is the marginal benefit curve. T is the value of the emission tax expected to yield the socially optimal reduction in emissions, and Q is the quantity of emission reduction that would be chosen in the case of a quantity instrument.

However, suppose actual marginal costs are unexpectedly high at all quantities of emission reduction, represented by the line MC_R, and thus the quantity of emission reduction that is ex post optimal is lower, at Q*. Because producers choose their levels of emission reduction with regard to the realised marginal costs, under an emissions tax they choose an emission reduction of Q_T. Under quantity controls, producers achieve the targeted emission reduction, but at a higher cost than expected. Thus under an emissions tax the social loss is represented by the triangle ABC (here the loss is in fact a foregone net benefit), whereas under emission permits the social loss is triangle BDE. In the top panel, where marginal costs are steeper than marginal benefits, the loss under a permit system is greater; in the lower panel, where marginal benefits are steeper than marginal costs, the loss under an emission tax system is greater.

Figure 2 – The effect of marginal cost and marginal benefit slopes on the choice between taxes and permits

Case a): Tax preferred

Case b): Permits preferred

Weitzman (1974) notes that, although taxes yield lower expected losses than permits in many circumstances, choosing a price instrument is far more likely to be disastrous than choosing a quantity instrument. If the government is very averse to extremely bad outcomes, a permit system may be more appealing than a tax system.

A price instrument could deliver very bad outcomes if there exists some threshold for emissions beyond which point the pollution in the atmosphere becomes great enough that the probability of environmental catastrophe increases rapidly. If the threshold were accidentally breached as a result of unexpectedly high costs and low abatement, the loss to society could be very large. A quantity instrument could prevent this breach if the threshold value of emissions were known. A quantity instrument would not deliver disastrous outcomes unless marginal costs were flat and several other conditions were also met. Weitzman’s warning is relevant only if there is a reasonable probability of an emissions threshold.

Weitzman’s analysis assumes that it is greenhouse gas emissions that cause the damage. However, it is not primarily emissions of carbon dioxide that cause the environmental damage, but rather the stocks of carbon dioxide that accumulate in the atmosphere. These stocks are slowly removed by natural processes, but are continually being added to by current emissions. Because of the low rate of removal, stocks of greenhouse gases in the atmosphere are large relative to the annual emissions of these gases.

When we account for the fact that only the rate of change of atmospheric carbon dioxide can be affected, and only to a limited extent, the problem of choosing a control instrument is slightly different. Newell and Pizer (2003) modifies the analysis in Weitzman (1974) to examine how the conclusions are affected when stocks rather than flows of the pollutant cause the harm. This paper concludes that if a permit system limits emissions over any short period of time, then a price instrument is preferred because the marginal environmental benefit of abatement curve is flat. However, if permits are bankable, this short run quantity constraint is absent. In New Zealand’s case, the marginal benefit curve is the international permit price (if one exists), and thus is not directly affected by the slope of the international marginal environmental benefit curve. Even if there is no international permit market, environmental benefits are likely to be such a small proportion of New Zealand’s marginal benefits, and already horizontal, that the distinction between flow damage and stock damage is irrelevant.