Tidewater: IELTS Academic Reading practice test

A beaver fells trees by gnawing through the trunk, drags the branches into a watercourse and builds a dam across it, behind which a pond forms. The purpose is not the dam but the pond. A beaver on land is slow and vulnerable, and deep water lets it swim to its food, float timber that would be impossible to carry, and enter its lodge through a tunnel whose mouth is below the surface, where no fox or wolf can follow. The trigger for the whole performance is sound. An animal placed near a loudspeaker playing the noise of running water will pile mud and sticks against the speaker, even where there is no water at all.

Both species were very nearly removed. Beaver fur is dense and waterproof and made the best felt hats in Europe for three centuries; the meat was eaten during fasts on the convenient grounds that the animal has a scaly tail; and a pair of glands produce a strongly scented secretion that was used in medicine and later in perfumery. By 1900 the European beaver had been reduced to perhaps twelve hundred animals surviving in eight separate remnants. In North America a population estimated in the tens of millions before the fur trade had fallen to something of the order of a hundred thousand.

What went with them is harder to see, because nobody alive has seen the alternative. The floor of a typical North American valley before trapping was not a single channel cut into its own sediment but a chain of ponds, marshes and wet meadow, connected by braided threads of water and interrupted every few hundred metres by a dam. Sediment cores from valley floors across the continent show thick deposits of pond mud, laid down over thousands of years and then, within a few decades of the trappers, cut through by a stream that had begun to run in one place instead of everywhere.

The hydrology of all this is now reasonably well understood. A dam slows the water above it, and slow water drops the sediment it is carrying. The stored water raises the water table in the ground around the pond, sometimes for tens of metres on either side, which keeps the floodplain wet into the summer. Water released from a pond leaves more slowly than it arrived, so a rainfall peak passing through a series of dams comes out lower and later. The effect has been measured repeatedly in small catchments and is substantial there. It is important to add that it weakens as the catchment grows, and that no serious study claims beavers would have much influence on the flood peak of a large river.

Britain lost its beavers by the sixteenth century and began getting them back in this one, partly through licensed trials and partly through animals of uncertain origin that turned up in a Devon river and were allowed to stay. The Devon population was monitored for several years by a university team, and the results were consistent with the catchment studies: flow peaks leaving the beaver-occupied stretch were reduced, the water leaving it carried less sediment and fewer nutrients than the water entering it, and the number of species using the site rose sharply. In 2022 the animal was given legal protection in England as a native species, which settled the question of whether the population would be removed and opened a different one about who pays for it.

Because beavers are not diplomatic. A dam raises water onto land whose owner did not ask for a pond. Culverts under roads are exactly the sort of constriction a beaver likes to block, and a blocked culvert floods a road. Orchard trees and ornamental willows are felled with no regard for their value, and burrows in a flood bank are a genuine engineering problem. A set of remedies exists and most of them work: a pipe run through a dam at a level that keeps the pond below the height at which it causes damage, wire mesh around valuable trees, fencing along vulnerable banks. What the remedies require is somebody to install and maintain them, and the awkward fact of the politics is that the benefits of a wetter catchment are spread across everybody downstream while the costs land on a small number of identifiable people upstream.

It is worth being clear about what beavers will not do, because their advocates have not always been. They will not stop a city flooding. The effects that have been measured are real, local and modest in absolute terms, and the enthusiasm with which they are sometimes reported has begun to invite the backlash that overstatement always invites. There are also genuine trade-offs. A pond is warmer than the stream that fed it, and warmer water holds less oxygen, which matters for trout and salmon; the literature on whether beaver ponds help or harm those fish is genuinely divided, with the answer appearing to depend on the temperature the river started at.

What makes the animal interesting as a policy question is its independence. A beaver builds where it decides to build, maintains its works without supervision, responds to a breach within a night, and costs nothing. No agency could afford to install and maintain several thousand small leaky dams across a catchment, and no agency would be permitted to. The whole appeal is that the work is free and unmanaged, and the whole difficulty is exactly the same thing seen from the field that has just been flooded.

Suppose a policy costs a billion pounds now and prevents damage worth ten billion in the year 2150. Is it worth doing? The standard method for answering is to convert the future sum into its equivalent today, on the reasoning that money available now can be invested and will grow, so a pound in the future is worth less than a pound at present. That conversion requires a rate, and the choice of rate is where almost everything interesting happens.

Consider what the arithmetic does over long periods. Discounted at three per cent a year, a billion pounds arriving a century from now is worth about fifty-two million today. At five per cent it is worth about seven and a half million. At seven per cent it is worth around a million, which is to say that the benefit has effectively vanished. Two rates that a reasonable person might defend, separated by four percentage points, differ in their verdict by a factor of fifty. A technique presented as neutral accounting turns out to contain a dial, and the position of the dial determines the answer.

Where does the rate come from? The standard formula assembles it from three quantities. The first is pure time preference: the extent to which a future person's welfare simply counts for less because it arrives later. The second is the rate at which the economy is expected to grow, since if our descendants will be richer, a pound means less to them. The third is a measure of how quickly the value of additional consumption falls as one becomes wealthier. Only the second of these is an empirical estimate. The first and the third are ethical judgements presented in the notation of economics, which is a useful thing to notice before anybody claims that the result is a technical matter.

The point was made unavoidable by two prominent economists who disagreed publicly about climate policy and did not disagree about the science or the modelling. One argued that pure time preference should be set at almost zero, on the ground that discounting a person's welfare because they happen to be born later is indefensible, and that the only legitimate reason to discount the future is uncertainty about whether anybody will be there. The other argued that a policy analysis should use the rate people actually reveal in their own behaviour and in capital markets, because an analysis calibrated to an ethical ideal nobody acts on produces recommendations that no government will follow. Their models were similar. Their conclusions about how much to spend differed by an order of magnitude.

One technical development has narrowed the gap. If the correct rate is itself uncertain, the expected present value of a future sum is not obtained by discounting at the average rate, because the low-rate scenarios dominate the calculation over long horizons. The consequence is that the effective rate applied to a distant benefit should decline as the horizon lengthens. Several governments have adopted schedules of exactly this kind, stepping the rate down over successive decades, and they make long-term projects appreciably easier to justify without abandoning discounting altogether.

For Alma Reyes-Court, who writes on the ethics of public decision-making, the entire apparatus is a sleight of hand. Deciding how much a future life is worth relative to a present one is a moral question of the first order, and embedding it in a parameter inside a model does not make it less so; it conceals it from the people who are entitled to argue about it. Outcomes that are catastrophic and irreversible, she argues, should not enter a calculation of net benefit at all, because the whole logic of a trade-off assumes that a loss can be compensated by a gain somewhere else, and some losses cannot.

The first part of that is plainly correct and the second seems to me to be mistaken. A constraint is not an alternative to a trade-off; it is a trade-off with the terms hidden, since a rule that permits no risk of a given outcome is committing to an unlimited expenditure to avoid it, and no society does that consistently for anything. Refusing to discount at all produces absurdities of its own: any permanent benefit, however tiny, then outweighs any finite cost, however large, and the analysis recommends impoverishing the present for an arbitrarily small perpetual gain. What Reyes-Court has identified, correctly, is a class of cases where expected-value reasoning breaks down because the tail of the distribution is fat enough that the answer depends entirely on outcomes we cannot price.

My own conclusion is about how the tool should be used rather than whether. A discount rate is not a measurement and should not be reported as one. The useful output of the exercise is not the present value it produces but the range of rates over which the recommendation changes, because that tells a decision-maker how much of the answer rests on an ethical assumption they are entitled to reject. Where a project is worth doing at every defensible rate, the analysis has settled something. Where it is worth doing at two per cent and not at four, the analysis has not answered the question; it has located the argument, which is a more modest achievement and an honest one.