Lecture 16, Feb. 26: The Social Discount Rate

Introduction

The discount rate formula: r = η⋅g+δ

r: social discount rate

g: Growth rate

• How much better things will be/ growth of the economy

  • A positive value entails that future generations get richer as they are further into the future; the larger the value, the better off we estimate they will be year by year. A negative value entails that future generations would be getting progressively worse. This is an empirical prediction.

η: Elasticity of marginal utility of consumption (equality parameter):

• The larger the value, the more intergenerational equality has weight. If future generations are getting worse off (g), then η means to discount our benefits; if future generations are getting better off, then we discount their benefits.

δ: rate of pure time preference:

• These are reasons for which we discount future goods because they are in the future. (All other reasons)

• Cohen and Parfit are just interested in δ!

  Cohen and Parfit Analogy

• Cohen and Parfit defend a zero rate of pure time preference, i.e., they think we should not discount for future good based on time. They present this colourful analogy: “Imagine finding out that you, having just reached your twenty-first birthday, must soon die of cancer because one evening Cleopatra wanted an extra helping of dessert. How could this be justified?”

  An opposing analogy

• The analogy is accurate but skewed in favour of their position. Here is an example showing the opposing view. Imagine someone telling you that they will either give you $10,000,000 right now or the exact equivalent of the worth of $10,000,001 in 10 years adjusted to inflation and the profit you would have made during this period. If we do not discount for time, a rational person would wait 10

  years; there is an additional $1. However, we think the first deal is better because we intuitively discount future benefits for some reasons we will discuss.

The argument from democracy

P1) Most people prefer to use a discount rate for time.

P2) Governments must represent most people’s preferences.

C) Therefore, governments must use a discount rate for time.

• If a government does not represent most people’s preferences, then the

government is paternalistic or authoritarian.

Cohen and Parfit distinguish two questions:

(1) Is it morally justified to use a Social discount rate?

(2) Should our government override the decision of the majority to question (1)?

Objection 1 - Moral action are not based on democratic preferences: The opinion of the majority has no bearing on this question. If all citizens believed that the genocide of a population was a good thing, it would not make the genocide good.

The argument that our successors will be better off (η⋅g)

“Benefits that are equally great, when received by people who are better off, may be plausibly claimed to have less moral importance.”

• Why should we sacrifice for future people who will be better off?

Objection 1- This is not δ

• Cohen and Parfit's first argument is that this is not about δ but about η⋅g. In other words, it is outside the scope of their argument.

Objection 2- Some future people will be worse off

• “Some of our successors will not be better than we are now. When applied to these people, the arguments just given fail to apply”.

The argument from excessive sacrifice

“By bearing costs now, we can give our successors greater benefits. For example, if we reinvest rather than consume, the resulting benefits may, in the end, be greater because of the cumulative effect of interests. We may seem morally required to choose policies that would impose great sacrifice on the present generation.”

• Analogy with a 20-year-old self vs 40-year-old-self

• A positive discount rate would avoid this too-high burden on the present

  Objection - The positite discount rate is the wrong approach

• “We shall be misstating what we believe. Our belief is not that the importance of future benefits steadily declines. It is rather that no generation can be morally required to make more than certain sacrifices for future generations. And this is part of a more general view, which has nothing to do with time.” Instead, “We should have a second moral aim: that these benefits be fairly shared between different generations. To our principle of utility, we should add a principle about fair distribution.”

Special Relations

• Most people believe we should give more weight to people close to us, like ourselves, our children or our friends, people with whom we have a special relationship.

• In the discussion about the discount rate, the next generation is like our children to whom we have a strong obligation. The following generation is like our grandchildren, to whom we have a weaker obligation.

• Based on this reason, we could have a discount rate in function of time, as the closeness of relations decreases as we look further into the future.

Objection - The case of grave harm:

• Cohen and Parfit then argue that this discount might

apply to the good of future people but not to great harm. They give the example of duties toward our own citizens.

• “Perhaps the United States government ought in general to give priority to the welfare of its own citizens, but this does not apply to the infliction of grave harm. Suppose this government decided to resume atmospheric nuclear tests. If it predicts that the resulting fallout would cause several pathologies, should it discount the death of aliens? Should it, therefore, move the tests from Nevada to the South Pacific, so that those killed would not be Americans? It seems clear that, in such a case, the special relations make no moral difference. We should take the same view about the harms that we may impose on our remote successors.”
  

Economic arguments

• It is sometimes better to receive a benefit earlier, since this benefit can then be used to produce further benefits

• An investment that yields a return next year will be worth more than the same return arriving in ten years if the earlier return can be profitably reinvested over these ten years.

Objection 1 - Some benefits do not decrease with time

• Cohen and Parfit argue that some benefits, like enjoying the beauty of nature, do not lose any value because they are in the future. The pleasure I get now is of equal value to the pleasure people in the future will have. Thus, the social discount

rate does not adequately account for these benefits.

Objection 2 - Investment consumed

• The argument fails to apply to resources that are simply consumed, as the moral importance of those costs should not diminish based on when they occur.

• How do we know that the benefits are not just consumed instead of ‘invested’ for the future? This is what often happens. Many investments that increase the GDP do not create good for the future; they just increase our consumption in the present. Investment in R&D, ameliorating institutions, etc. have a potential positive impact on the future, but not most of the thing we do to make the economy grow.

Objection 3 - Harm

• The argument does not work for harm. If a policy risks causing harm, such as

genetic deformities of children, the argument cannot validly suggest that a deformity occurring next year should count less than one occurring later based solely on opportunity costs.

The argument from probability

• The Argument from Probability suggests that we should discount more remote effects because they are less likely to occur. However, this argument sometimes confuses two separate questions:

1. When a prediction applies to the further future, is it less likely to be correct?

2. If some prediction is correct, should we give it less weight because it applies to the further future?

• The answer to the first question is obviously yes. We make decisions now according to our best estimates, so we should assume our estimate prediction to be correct. In this context, the question is, should we give less weight to the future because they are further in the future given what estimate?

• This is question 2 and the one we are interested in.

Objection 1: Nuclear Waste

• When considering possible accidents with nuclear waste, we must think far into the future since some nuclear wastes remain radioactive for thousands of years.

• In this case, with a discount rate of 5%, one death next year is worth more than a billion deaths in 400 years. According to the discount rate, to avoid one death next year, we would have to put 1 billion people in danger in 400 years

Death that do not occur do not matter

• “We know that if radiation escapes next year, we will have no adequate defense. We may believe that, over the next few centuries, some kind of countering measure will be invented, or some cure. Thus, we may believe that if radiation escapes in four hundred years, it will be much less likely to cause deaths.”

• If the probability of 1 billion people dying is a smaller probability than 1 billionth, then it is reasonable to put 1 billion people at risk to save a person for sure. This is what EVT entails.

• “Deaths that do not occur, whether now or in four hundred years, do not matter”

Objection 2: Misstate our view

• A second “objection” is that this misstates our moral view. It makes us claim not that more remote bad consequences are less likely, but that they are less important. This is not our real view”. This is because ‘δ’ states that death is less important in the future not that it is less probable.

There are two formulas used in the literature.

1) r = η⋅g+ δ (this is the most common one)

2) r= η⋅g+ δ+ p (this is less common; the only difference is an added variable for probability p).

• The probability that people will exist in the future (so they can receive benefits) is not stating that they are less important but that the consequences are less likely to affect them, because non-existent people cannot benefit or suffer from consequences

Objection 3: The discount rate does not always coincide

• “The two discount rates do not always coincide. Predictions about the further future do not decrease in certainty at some constant rate of percent per year.

• Indeed, when applied to the further future, many predictions are more likely to be true. (Consider the predictions that some policy will have changed, or that certain resources will have been exhausted.) If we discount for time rather than probability, we may thus be led to what, even on our own assumptions, are the wrong conclusions.