Appraising infrastructure on what it prevents, not what it earns
The situation
Some public infrastructure earns little or no cash. Its value is in what it prevents: damage from rare events, emissions, and costs that land on someone else.
On cash alone, it is never worth building. The question is whether what it prevents is worth more than it costs.
Budget holders see the whole cost up front. The benefits arrive over decades, and nobody receives them as cash.
Why it is hard
The benefits are losses that do not happen. They have to be estimated, they arrive over decades, and some of them have no market price.
- Avoided damage depends on how often a damaging event happens, which is a probability, not a forecast.
- Carbon has no single price. The price used is a policy choice, and the answer can turn on it.
- Benefits and costs arrive at different times. Money spent now is worth more than benefits in twenty years.
- One assumption often carries most of the case, and a single-number answer hides which one.
The approach
We built a cost-benefit analysis that counts avoided damage and avoided emissions alongside cash.
- Costs are the cost to build and the cost to run, year by year, for the life of the asset.
- Avoided damage is the damage from one event, times its chance in a year. That is multiplied by the share the project prevents.
- Avoided emissions are the tonnes saved each year, times a carbon price.
- Everything is discounted at a social rate. Benefits minus costs, in today’s money, is the net present value.
- The benefit-cost ratio is benefits over costs. Above one, the project pays its way.
- The economic rate of return is the discount rate at which benefits exactly equal costs.
Then the analysis turns the question around. Instead of assuming a carbon price, it finds the price at which the project pays its way.
A grid tests the answer against every input, and a scoring of options on several criteria sits beside it.
A worked example
This is a synthetic project, not the sponsor’s. It costs 10 million to build and 0.2 million a year to run, for 25 years.
| Line | How | Millions |
|---|---|---|
| Cost to build | paid up front | 10.00 |
| Cost to run | 0.2 a year for 25 years, discounted | 2.82 |
| All costs | build and run | 12.82 |
| Damage avoided | 40 an event, 1 percent a year, 80 percent prevented | 4.51 |
| Left for carbon to cover | costs less damage avoided | 8.31 |
| Worth of one unit of carbon price | 5,000 tonnes a year, discounted | 0.070 |
| Carbon price that covers the rest | what is left, divided by the line above | 117.9 |
Discounted at 5 percent, avoided damage covers part of the cost. Carbon has to cover the rest, 8.31 million in today’s money.
Each unit of carbon price is worth 0.070 million over the project’s life. So the project pays its way at 118 a tonne.
Left undiscounted, the same project seems to pay its way at only 56 a tonne. The gap between the two lines is discounting alone.
Undiscounted, a tonne saved in year 25 counts as much as a tonne saved today. Discounted at 5 percent, it counts for less than a third as much.
At a carbon price of 85, the undiscounted ratio reads 1.24 and looks like a pass. Discounted, it is 0.82, a fail.
Where it breaks
- It uses one size of event and one yearly chance. A real range of events, from small to severe, would give a different answer.
- It counts full benefits from the first year, with nothing left at the end of the asset’s life.
- The carbon price is a policy choice, not a market fact, and the answer rests on it.
- Scores that compare options on several criteria depend on judgment, and whoever sets them shapes the result.
- Benefits that overlap, such as damage and the disruption it causes, can be counted twice unless someone checks.
If a project's case rests on what it prevents, discount the benefits the same way as the costs.