No time to tax! Can a gradually increasing carbon tax really provide a cost-efficient green transition?

It is widely assumed in climate economics that a uniform, gradually increasing carbon tax mirroring the “social cost of carbon” (SCC) leads to cost-optimal emission reduction. The underlying idea is that emitters switch to carbon-saving technologies as soon as the tax becomes so high that this switch saves money. If we let the tax equal the SCC, i.e. the extra damage caused by emitting one extra ton of CO2, then everybody who can save carbon at a lower price than the damage caused by this emission will do so, whereas those for whom the emission-saving costs more than the associated damage will not. Models like Nordhaus’ famous DICE model find a roughly exponentially increasing SCC, corresponding to an exponentially increasing carbon tax.

We implemented such an exponentially increasing carbon tax in a simple agent-based model, the Dystopian Schumpeter-Keynes (DSK) model. Agent-based models dispense with restrictive perfect rationality and market equilibrium assumptions and are able to describe non-equilibrium dynamics and tipping as emergent properties of collective behaviour. They are not yet widely used in climate economics.

The DSK model contains two types of firms which manufacture machines or a consumption good, respectively, using labour and energy (electricity and/or fuel). Electricity is provided by a monopolist using green (carbon neutral) or brown (fuel-based) plants.

In DSK, the DICE-based carbon tax is far from satisfying as climate policy. In the first ≈30 years, the tax is too low to trigger a green transition in the electricity sector. When green plants finally do become competitive, it still takes decades until the transition is completed, because power plants have a long lifetime and are replaced only gradually. Higher taxes can speed up the process somewhat, but even modest increases in the transition rate require big increases in the tax (and hence considerable side effects on the economy, including unemployment). The exponentially increasing carbon tax is thus low at the beginning of the transition, where higher taxes would be most needed, but becomes high (with associated side effects) at later stages, when the transition is already gaining momentum by positive feedbacks, most notably innovation reducing the price of green plants. It would be preferable to implement a constant (or even slightly decreasing) tax which is sufficiently high from the beginning.

Apart from green electricity, decarbonisation also requires fuel-using firms to switch to electricity. However, the carbon tax does not incentivise this switch initially, as the tax increases not only fuel price, but also electricity price. Again, the switch takes time, while rapid decarbonisation requires a swift start of electrification. One way around it is to levy a higher carbon tax on manufacturing firms than on the electricity sector (to make electricity use more attractive); alternatively, one could simply impose regulations.

Our results suggest that a carbon tax should not be gradually increasing and uniform, but high from the beginning and sector-dependent. We also find that tax-free policies, such as green subsidies or regulations, can bring about a green transition with possibly less side effects.