Tuesday, September 7, 2010

An Appropriate Discount

Here is the question:  How much should we discount future problems to create solutions for today?  For example, if we have an economic policy that can make us 10B dollars richer today but we have to pay back 11B dollars in 10 years, is that a good idea?  Economists generally assume that there is a discount rate we apply to future wealth and benefits that is multiplicative year after year.  For example, if we discount 6% per year then 1 dollar's worth of benefits in 5 years is only worth 73 cents of benefits right now.  This is especially critical when evaluating the costs of environmental policy years down the road as in order to talk about the net benefits or penalties of various schemes we need to understand how we weight various outcomes.  If we cost ourselves 1 dollar now but our descendants will save 2 dollars by not having to respond to issues later on, is that a good tradeoff?

It turns out this issue is really contentious.  In both The Rational Optimist and The Skeptical Environmentalist this issue was really brought to the fore because both authors felt that it was critical to assume that our descendants would be much, much richer than we are today.  For them paying 2 dollars to prevent a flood due to climate change is much less of a problem than it is for us to pay 1 dollar to use renewable energy sources today, so in that particular case we should just keep burning oil.  Obviously everyone agrees that it is sensible to make the easiest cuts first and that some cuts should be made, but people strenuously disagree on exactly how deep to cut into our economy today to benefit those will come later.  In the books Eaarth and Earth in the Balance the opposite view was held:  Those authors felt that we should weight our descendants' problems very highly and that discounting benefits and penalties in the future was unconscionable.  In fact this issue is often framed as a moral one with some people even claiming that a discount of zero is the only moral choice, valuing the lives of those many generations from now as equal to those alive today.

I think the idea of a zero discount is bonkers, in particular because the authors themselves don't live that way.  People care about those close to them and care much less about others they do not know; even those who are very generous do not spend the same time and money helping random strangers as they do their family and friends.  If people really behaved this way the rich wouldn't buy expensive cars or clothes - they would set up trust funds to distribute their money equally among their 1000 potential descendants 10 generations hence and live a middle class lifestyle.  People naturally apply their own future discounts and if that behaviour is immoral then I don't know if there exists a moral person.

The other major issue with zero future discounts is uncertainty.  We can be pretty confident that 1 dollar of benefit today is worth about 1 dollar.  We have a sketchy idea of how much it will be worth in 25 years and absolutely no clue about its worth in 50 years.  Think about someone from 1960 trying to work out the economic value of something in the world of the internet, cell phones, facebook and the collapse of communism. They would have no idea, and using valuations of concepts and even raw goods from that long ago is a joke.  Because of this we have to value the future less simply because we don't know whether what we are doing will matter at all.  Even barring the seemingly unstoppable increases in economic power that each new generation brings the lack of predictive power means we must strongly weight benefits today over seemingly equal benefits far down the road.

None of this means we should pour arsenic into the water just for fun but it does mean that we need to approach these issues cautiously and not idealistically.  Future discounts aren't just an economist's construction - they actually model human behaviour if the number is set correctly.  We need to accept that, set reasonable numbers and move on.


  1. There's always market considerations for inflation or deflation to take into account. With inflation expected, money today is worth a good deal more than money in the future, and the opposite is true for deflation.

  2. Two weeks ago I got to go to a two day course at U of T on cost benefit analysis. We talked about discount rates for about half a day, and he specifically used the Stern Review as an example, so the discussion focused on environmental issues and to some extent on the value of human life.

    While obviously a zero discount rate is total madness, there isn't really any way to decide what the discount rate should be. The Ontario government's treasury board insists on 8%. Where this comes from I have no idea. Mean real income in Canada increased less than 1% between the mid 90s and the mid 00s. We currently have a prime lending rate under 1%. What would make anyone think that a dollar next year will be worth 8% less than a dollar this year? In the current climate in Canada, at least, we have every reason to think a discount rate pretty close to zero is about right. Of course, projecting this 50 years in the future doesn't really work, since no one knows what will happen. But what do you do then? Whatever you do is a guess.

    Also, though, I think that it is important to figure out what you are discounting. For a lot of these disasters the major part of the cost isn't clean-up, it's death and displacement. If someone asked me whether I would rather die today or in 10 years I would take my ten years. But as a person in charge of a society if someone asked me whether it was better for a random person to die today or for a random person to die 10 years from now (the person being selected in 10 years, not selected now and then wait to die) then there is no reason for me to have a preference. The issue is that if we really are richer, then the value of human life and the loss due to human displacement should have increased by the same amount. Today people use values for human life in the range of 1.5M to 15M (it's crazy that it varies this much, but that's another issue). What if we get rich at a fantastic and exponential rate and real mean income rises to 1M. We couldn't possibly think that the value of a human life is only 1.5 or even 15 years of mean income.

    Setting the discount rate actually is an ethical issue as much as it is an economic one. Since projects done for the public good are typically paid for in advance and have benefits for many years, higher discount rates tend to discourage them while lower rates encourage them. While I agree that real growth in income and human psychology regarding the future are both real things we could base the discount rate on, what is the argument that either of these will result in the best benefit to society? Making decisions based on the faults of human psychology seems like it might be one of the worst possible ways to set public policy. After all, don't humans naturally discount on a hyperbolic rather than exponential curve?