A debate over discount rates

Transport evaluation, and urban policy in general, invariably requires us to make trade-offs between the present and the future. When we invest in the built environment – roads, rails, buildings, etc – we are expending today’s resources on projects that will primarily benefit people in the future. Conversely, decisions we make about resource use today – e.g. how much greenhouse gas to emit – will impose increasing costs on future generations.

Individuals also make decisions that involve trade-offs between the present and the future. For example, most home-buyers take out a mortgage, which allows them to spread the payments over a long period of time in exchange for paying a bit more in interest charges. For the borrower and for the bank, the interest rate paid on the debt reflects the trade-off between repaying the debt today or repaying next year.

Of course, there are also other ways that individuals make trade-offs between the present and the future. They may, for example, spend money on their children’s education, even though the “returns” from doing so are long-term and indirect. Or they may vote for policies that cost them money now but return long-term benefits, such as environmental protection or pre-funding of superannuation.

The trade-offs that governments make between present and future are codified as discount rates, which describe how much of a “discount” we place on future outcomes relative to present outcomes. For example, a discount rate of 10% means that the government would value a benefit of $100 in a year’s time as being equal to a benefit of $90 today.

Transportblog’s previously taken a look at the discount rate issue here, here, and here. But others are also engaged with the issue. Back in August, University of Michigan economics professor Miles Kimball, who had been visiting the Treasury, strongly criticised its approach to discount rates. The key point in his critique is that the Treasury’s discount rates don’t accurately reflect the financial market returns currently available to the government:

There is an extremely strong argument against using an 8% real discount rate in evaluating government projects. I think the argument below can be sharpened to become institutionally relevant.

Basically, an 8% real discount rate makes no sense to use unless the New Zealand government is actually getting an 8% real return on funds that it saves. It is not enough for someone to claim that the New Zealand government theoretically could get an 8% real return on funds it saves when that is not true or is only theoretical because the New Zealand government would never actually do that with funds saved by not doing a project.


Just to be clear, my view is that (a) all projects that are better than putting the money in the Superfund should be done, and (b) if someone claims that a project is worse than putting money in the Superfund, then money should be put in the Superfund instead, and (c) if a project looks better than paying off some of the debt by buying bonds–or, almost equivalently, good enough that borrowing at the bond rate to do it looks like a positive present value–it should also be undertaken UNLESS the government is willing to issue additional bonds to put more money in the Superfund invested in risky assets.

I’ve skimmed over a lot of the substance of Kimball’s argument, which I encourage you to read. (Warning: it’s wonky.) After Kimball’s blog post, former RBNZ economist Michael Reddell wrote a blog post defending the Treasury’s policy. Reddell’s counter-argument is that financial returns to government is the wrong measure of the time value of money:

I was struck reading Kimball’s material that the cost of the government’s equity did not get a mention. There was a strong tendency to treat the government as an autonomous agent (like a household) managing its own wealth, whose low borrowing costs depends only on the innate qualities of the government and its decision-makers. But that is simply wrong. A government’s financial strength – and ability to borrow at or near a conceptual “risk-free” interest rate – rests on the ability and willingness of the government to raise taxes (or cut spending) as required to meet the debt commitments. That ability to tax is implicit equity, and it has a cost (an opportunity cost) that is considerably higher, in most cases, than 2.5 per cent real.  So long as the government will raise taxes as required, the bondholder bears none of the downside if a project goes wrong. But shareholders – citizens – do.  Bearing that risk has a cost, and that cost needs to be taken into account by government decision-makers.

There is a related argument sometimes heard that governments should do infrastructure projects rather than private firms simply because the government’s borrowing costs are typically lower than those of a private firm. But, again, that rests on the power to tax, and the ability to force citizens/residents to pay additional taxes has a cost from their perspective (even if the government never chooses to exercise the option).   As citizens, the possibility that the government will raise taxes (or cut other spending programmes – eg NZS) impinges on our own ability and willingness to take risks, and hence to consume or invest in other areas.   That often won’t be a small cost. The opportunity cost of the government not undertaking a project is not what, say. the NZSF might be able to earn on the funds, but what citizens themselves might prefer to do if that risk-bearing capacity was freed up.

Again, I’ve skimmed over Reddell’s argument, so I’d encourage you to read it in full if you’re interested. He throws in a brief jab at road projects with low BCRs:

We have too little disciplined analysis of the costs and benefits of most government projects, and too little willingness to allow decisions to be guided by the results of the analysis when it is undertaken (did I hear the words “Transmission Gully”?).

I can see some truth in both sides of the debate. Kimball’s got a good point, which is that current government borrowing costs (and financial market returns in general) are at historic lows, which should lead to lower government discount rates. But Reddell’s also correct that it’s appropriate to set discount rates based on wider social decisions about consumption and investment.

One issue that neither of the two grappled with in detail was the question of how risky government investments actually are. Kimball touched on that briefly, arguing – essentially – that the benefits of projects will inevitably rise in the future due to people’s greater willingness to pay for public goods in the future:

A good method of risk adjustment for projects is to think seriously of the real dollar value they will have dependent on the level of real consumption in the economy. One virtue of thinking about the adjustment this way is also that it provides a reminder that the dollar value of the flow of benefits from many projects will tend to increase in the future simply because trend increases in per capita income will raise the willingness to pay for those benefits.

Frankly, I don’t think this is correct. Changes in technology, changes in prices, and changes in preferences mean that infrastructure that’s useful today can easily become a stranded asset tomorrow. Look at Los Angeles: in the 1950s it was rushing to rip up its rail lines, and now it’s rushing to build a new rapid transit network. Or look at San Francisco: several of the elevated expressways it built in the 1960s have been torn down.

In short, things change. Sometimes they change quite radically. When the Ministry of Transport looked at future transport demand last year, they found that they couldn’t settle on a single forecast for future transport demand. Instead, they arrived at four scenarios, three of which entailed a decline in vehicle travel:


In other words, there can be considerable uncertainty in returns from transport investments. While it might not necessarily be appropriate to try to account for this in discount rates, it’s surely an important consideration for project evaluation more generally.

What do you think about discount rates?

29 comments to A debate over discount rates

  • Brendon Harre

    I think there are all sorts of hidden assumptions to this sort of CBA/discount rate process and if the top-down decision makers want a particular outcome then they probably can find an intellectually agile economist to adjust the assumptions so the ‘right’ outcome is produced. The ‘right’ outcome invariably being decided by those at the ‘top’ while those in the ‘down’ part -get what they are given.

    • Peter Nunns

      I think it’s natural to assume that bad decisions / bad policy outcomes are the product of rigged analysis. My observation is that that’s seldom the case. When bad decisions get made, it’s usually because one of two things happened:
      (1) Nobody did any cost-benefit analysis in the first case, or
      (2) Cost-benefit analysis was done, but decision-makers ignored it in favour of politically convenient mumbo-jumbo.

      Also, I think it’s inaccurate to describe CBA as a top-down process. CBA asks bottom-up questions, such as: “Is this a good way to spend the next dollar of public funds?” or “Is there a case for this individual regulation?” In other words, it can be used to justify (or critique) individual policy choices, but it’s less relevant to top-down planning frameworks.

  • Warren S

    Another high quality post Peter.
    I am not an economist so cannot add anything meaningful about discount rates. What I do know is that the tearing down of the Embarcadero Freeway in San Francisco absolutely improved that city for its residents and tourists alike.
    The completion of the demolition of the Alaska Highway in Seattle will do the same for that city.
    These examples indicate how important “what we do not build” is and I am very wary of any motorway extension and particularly any elevated motorway extension within our cities. NB Waterview interchange?

    • Peter Nunns

      Thanks Warren! Definitely agree about the freeway teardowns in the US cities. Today, it’s very difficult to imagine what the SF waterfront would look and feel like if the Embarcadero Freeway was still in place.

      What this illustrates, I think, is that transport infrastructure can have long-term (although not entirely irreversible) effects on cities. But there can be significant uncertainty about whether those effects are beneficial or not! When it comes to roads, it might be the case that people come to value good urban places more than they value a slightly faster drive.

    • Waterview at least finally completes the city bypass that always should have been done instead of forcing all traffic through the centre of the city. What we all should be much more concerned about are NZTA’s plans for massive interchanges on either side of the proposed road harbour crossing tunnels. On the Shore side alone there will be triple stacked flyovers like at Waterview, massive new reclamation, and 35m exhaust stacks in the Harbour belching out fumes.

      So last century, so wasteful. The approach works alone will cost more than a rail only tunnels, and way way more an elegant Light Rail walk and cycling bridge across from Wynyard.

      Check out this ‘people’s bridge’ that’s just opened in Portland $135m USD; LRT, walking and cycling [considerably shorter than the Waitemata crossing of course but still gives a sense of the cost difference].

    • Dave B (Wellington)

      What concerns me is our lack of retrospective economic-performance monitoring of major projects once they are implemented. The results from this ought to be a major input to the analysis of future projects but it seems that this doesn’t happen. What, for instance, would have been the economic-performance profile of the Embarcadero Freeway over its lifespan and eventual removal? Did it ever achieve a good economic performance or was it a disaster from inception onwards? Were any other such projects influenced (or deterred) by the emerging- and ultimate understanding of how this performed (or failed to)? I would doubt it, since the performance of much-vaunted projects tends to be talked-up to ‘show the project in a good light’ until such time as it becomes impossible to conceal a fiasco. In this respect, the dismal performance of certain Australian PPP toll roads provides a hard-to-gloss-over reality-check of these things, but have lessons been learned? Lessons not simply about PPP funding mechanisms but about the over-stated value of such projects in the first place? Has the NZ government taken any of this on board in deciding its own project-priorities (did I hear the words “Transmission Gully”?).

      And likewise, when a project performs well, do the free lessons in effective project-selection get lapped up and applied to future decisions by policy-makers? Seems not, because there are numerous examples of successful public transport projects that achieve spectacular and expectation-busting performances, and yet still, such projects have to jump through far greater approval-hoops than many lower-performing highway schemes.

      The real problem is that when people of a certain arrogant persuasion find themselves in positions of being able to control things, they don’t want rational analysis to get in the way of their favoured ideas and so they ignore it where it doesn’t suit. Therefore it is vital that credible critics trumpet the actual outcomes of their decisions every bit as loudly as they trumpet their own justifications for them. Informed public-opinion is the best tool for bringing these people to heel and the true effectiveness of their decisions out into the open.

  • Early Commuter

    It’s simple
    The government should nationalise all major industry/infrastructure
    The government should build what is necessary to maximise social utility without empowering the capitalist class

  • John

    Small quibble: If I recall correctly, a discount rate of 10 per cent means that $110 falling due in a year’s time is regarded as having the same value as $100 today (not 100/90 as you said).

  • Graeme Easte,

    I did not see any discussion of the relationship of the discount rate to the payback period for the project (i.e. cost versus savings per year). To me the issue is the failure to differentiate between public transport projects where the full cost is included (inclusive of the rolling stock or buses) and roading projects which exclude the cost of the vehicles (supplied by the users). They also ignore the fact that public transport vehicles last much longer (e.g. locomotives last several decades versus cars that are typically replaced within a decade). For that reason the payback period for PT projects can be longer and the discount rate correspondingly reduced.

    • Peter Nunns

      Hi Graeme – you raise a good point about how we account for the public and private costs of transport. Under current evaluation procedures, we effectively assume that vehicle ownership is fixed – i.e. that people will bear the significant costs of owning cars and parking regardless of what transport options are available to them.

      That seems to ignore an important reality, which is that better transport choice _can_ allow households to downsize from two cars to one, or from one to zero. And vice versa – we probably underestimate the costs of monomodality.

    • Stu Donovan

      Couple of interesting economic issues here:
      1. Private vehicles: I disagree with you here because private vehicles are a private cost. Hence the economic benefits that users receive from switching to PT (and hence having to use their private vehicles less or not at all) should be captured by the demand curve for PT. Put another way, the benefits of less vehicle use are included in benefit cost analysis via the demand curve for PT: If cars were cheap, then we would expect to see less demand for PT, and vice versa. I should mention that not all bloggers and not all economists agree with me, but that’s my line and I’m sticking to it, at least until I see compelling evidence as to why we should include private costs incurred in a secondary market where there is no evidence of externalities (i.e. people owning cars hurt no-one, it’s when cars are used that the problems arise); and
      2. Public versus private discount rates: I’m more sympathetic to the view that we should reduce discount rates for transport projects. I don’t think it’s necessarily clear-cut and can see arguments either way, but as you say it seems to me that people attached relatively low risk to investment in transport infrastructure (i.e. they like it), even the pay-offs are a long way off. This is actually one reason why people may like the RONS: Because they attach a lower discount rate to infrastructure investment. I’m sympathetic to this view.

      • Peter Nunns

        I thought we’d done this debate to death via email!

        I agree with you that all costs of private vehicle ownership *should* be factored into modelled demand curves for PT and driving. However, I can’t see any compelling evidence that they *are* in practice.

        I think you could draw an analogy with agglomeration economies. Some agglomeration economies arise internally to firms (i.e. through economies of scale in production) and thus could in principle be modelled with a complete land use-transport interaction model. But instead of doing that, NZTA said, that’s too hard, so we’ll implement a post-model fudge to account for it.

        • stu donovan

          Yes but i was replying to graemes comment. But your comment has also confused me further: when undertaking economic analysis you usually assume benefit/costs to consumers are factored into decisions about what they consume. So the evidence needs to originate with people like yourself and graeme :).

  • Warren S

    Yes Graeme. With the right amount/proportion of spending on public transport particularly in Auckland to make PT a realistic and viable option, there are opportunities for huge savings in overseas funds for NZ, for monies now spent in the categories of vehicle imports, petroleum products and roading asphalt etc,

  • Matthew W

    Regarding risk and uncertainty, I think this is fundamental to what a discount rate enumerates, together with the time value of money. There is risk with transport projects just like any other investment. If demand is not as high or new technologies come along and reduce the value of the investment this risk will be realised.

    • Peter Nunns

      Yep, that’s why I raised the issue in this post!

      It seems (to me at any rate) that different government investments should come with very different risk premia. For example, building a convention centre is intrinsically more risky than building a hospital – there’s a lot less demand risk for a hospital. Yet we apply the same discount rate in both cases…

  • I think the problem I have with economic analysis of anything – and this came up in the discussion of golf courses a few weeks back – is that economists tend to resist any discussion of what we’re actually trying to achieve.

    In theory, economics is about telling us about what the trade-offs are with any policy that gets proposed, but in practice economics is largely split into ideological camps, mostly right-wing, trying to justify their own policy preferences.

    Discount rates are a fairly reasonable technical way of analysing the benefits of projects over time, but those benefits themselves use largely arbitrary made-up numbers. There’s endless debate about the various merits of, say, Puhoi-Wellsford versus the CRL. Or about the value of houses versus parks. But both of those have totally different goals, and measuring them against each other to the 3rd decimal place is largely a fool’s errand.

    There’s been a lot of complaint on this blog about the Roads of National Significance, and endless jabs about their low BCRs. But the problem isn’t the BCR, which after all is only useful for comparing two similar highway projects. The problem is with what’s being proposed at all – highways to serve more unsustainable urban sprawl, or an urban rail system to give a shot in the arm to urban development.

    Likewise with parks – the issue isn’t whether certain land is “more valuable” as parkland or housing land. By the time you’ve asked that question, you’ve already made up your mind to judge the value in dollars and cents, rather than human values.

    • Peter Nunns

      I think what you’re saying is “sometimes economists come to conclusions that I disagree with, so therefore they must be ideological hacks and idiots”. That seems to miss the point.

      You say we need to consider the realities of the situation – e.g. “highways to serve more unsustainable urban sprawl”. Well, how do we know what reality looks like? How do we know that urban sprawl really is unsustainable? To me, that implies a need to calculate the cost of things – and yes, sometimes to put a monetary value on them so we can easily compare them to other things.

      In fact, if you wanted to do that, you could start by taking a look at some of my research on the topic – see here and here.

      “dollars and cents, rather than human values”

      I’m confused. I thought humans invented money to use as a unit of account. Are you telling me that money is just something that we found lying around, like basalt or the Planck constant?

      Sarcasm aside, if you object to monetary values, I can easily use a different unit of account instead. I could analyse everything in terms of minutes of one’s life, laughter, or helium balloons. It genuinely does not make a difference to the end results.

    • I think my comment was way over the top, and perhaps needlessly personal. I apologise and withdraw, and I’ll start again.

      Economic analysis of government spending involves a lot of assumptions and decisions about what to measure and what to value. The NZTA Economic Evaluation Manual is, what, 500 pages or so? This doesn’t matter hugely comparing different projects of the same type – which of these passing lane projects should we construct? But the more different the competing projects, the more the assumptions matter.

      The Roads of National Significance are a pretty good example, much as I disagree with what’s trying to be achieved. They have low traditional BCRs, but that’s exactly because what’s being measured is not the point of the project. Traditional BCRs measure time savings, and mostly ignore induced demand. Whereas induced demand is the point of the RoNS – they are “lead infrastructure”, and the whole point is to create traffic. The CRL does exactly the same – performs poorly in its BCR, because it’s trying to build patronage, not shorten travel times.

      Dollars are a perfectly good unit, but just because two quantities are both in dollars does not make them interchangeable, without considering how the values were arrived at. The dollars we come up with when we measure time savings on a highway project can be directly compared to, say, the dollar value of a quality-adjusted life-year in a health project. But only by fiat – in both cases, they’re relative scores that reflect the underlying assumptions and decisions that went into the analysis. Not least, how much we value a life, or a minute or someone’s time.

      This doesn’t have much to do with discount rates. But I think economists are too quick to dismiss strategic concerns – dealing with what is, and not what could be. When Kimball says

      > “all projects that are better than putting the money in the Superfund should be done, and (b) if someone claims that a project is worse than putting money in the Superfund, then money should be put in the Superfund instead”

      it’s confusing the dollar scores from economic analysis with real money. It’s entirely possible to compare them, and decide what to spend money on – but that’s a subjective decision by policy makers, not a simple matter of checking which number is larger. Suggesting that projects should be funded automatically removes that subjective, strategic role – or rather, incorporates it into the original decisions about how to evaluate the projects. Hence power moves from policy makers to the writers of the economic evaluation manual.

      • Agree! Opponents to the whole conception of RoNS programme, like me, latched onto the BCRs as an additional way to criticise them but that is not the core of my objection. But rather that the whole thing is conceptually flawed and the wrong fit for our age. And perhaps you have summed that up perfectly in your comparison of them to the CRL:

        ‘Traditional BCRs measure time savings, and mostly ignore induced demand. Whereas induced demand is the point of the RoNS – they are “lead infrastructure”, and the whole point is to create traffic. The CRL does exactly the same – performs poorly in its BCR, because it’s trying to build patronage, not shorten travel times.’

        The RoNS are about expensively and destructively generating more vehicle traffic; something we should not wish for, whereas the CRL is about increasing access without increased vehicle traffic, or FF use, things we should most certainly wish for.

        NB: the CRL also delivers time savings, and ones that will not deteriorate over time unlike with induced traffic, as increased Transit demand should in fact lead to higher frequencies and further increased time savings over time.

        • > But rather that the whole thing is conceptually flawed and the wrong fit for our age.

          Well put.

          > the CRL also delivers time savings, and ones that will not deteriorate over time unlike with induced traffic

          Oh, it certainly does that too – the traditional BCR is not zero. But it would probably not be worth spending north of $2 billion merely to shorten the commutes of the current 50,000 or so daily train riders – the point is to add the next 50,000 train riders, then the next 50,000, and beyond. The same is true of almost all public transport and cycling improvements – they stack up because we want to drive mode change, not (mainly) to make things better for the tiny few who are currently Quaxing.

      • Peter Nunns

        Hi Stephen – thanks for coming back and commenting further – it’s definitely an interesting/contentious topic! However, I think you’re too quick to assume that I haven’t thought through these issues.

        You raise a good point about strategic assessments / value judgments versus quantitative analysis / cost-benefit analysis. In my experience they can be complementary, rather than intrinsically opposed. After you’ve set a strategic direction, you need some way of determining whether any given project is an efficient way of achieving that end. Cost-benefit analysis is a good way of doing that. (And most alternative approaches start to resemble CBA as they get more complete/rigorous.)

        On the RoNS example, I would observe that their low BCRs indicate that they will probably not be successful in achieving their stated strategic objectives. There’s little compelling evidence that they’re a good way to reduce congestion or raise economic productivity – or, at least, they’re likely to be less effective than competing projects. See MoT’s anlysis here:

        Once you realise that, it starts to look less like a diligent pursuit of strategic objectives and more like a mistake.

        That being said, I agree with you that it’s also bad for the technicians to be making undisclosed value judgments. I wrote about that here:

        That’s basically why I write posts on issues like discount rates – to try and make the value judgments embedded in them transparent to the public.