Tag: cost-benefit analysis

Mid-December saw a rapid rise in coronavirus cases in London and the South East and parts of eastern and central southern England. This was due to a new strain of Covid, which is more infectious. In response, the UK government introduced a new tier 4 level of restrictions for these areas from 20 December. These amount to a complete lockdown. The devolved administrations also announced lockdowns. In addition, the Christmas relaxation of rules was tightened across the UK. Households (up to three) were only allowed to get together on Christmas day and not the days either side (or one day between 23 and 27 December in the case of Northern Ireland). Tier 4 residents were not allowed to visit other households even on Christmas day.

The lockdowns aimed to slow the spread of the virus and reduce deaths. But this comes at a considerable short-term economic cost, especially to the retail and leisure sectors, which are required to close while the lockdowns remain in force. In taking the decision to introduce these tougher measures, the four administrations had to weigh up the benefits of reduced deaths and illness and pressure on the NHS against the short-term economic damage. As far a long-term economic damage is concerned, this might be even greater if lockdowns were not imposed and the virus spread more rapidly.

In a blog back in September, we examined the use of cost–benefit analysis (CBA) to aid decision-making about such decisions. The following is an updated version of that blog.

The use of cost–benefit analysis

It is commonplace to use cost–benefit analysis (CBA) in assessing public policies, such as whether to build a new hospital, road or rail line. Various attempts in the past few months have been made to use CBA in assessing policies to reduce the spread of the coronavirus. These have involved weighing up the costs and benefits of national or local lockdowns or other containment measures. But, as with other areas where CBA is used, there are serious problems of measuring costs and benefits and assessing risks. This is particularly problematic where human life is involved and where a value has to be attached to a life saved or lost.

The first step in a CBA is to identify the benefits and costs of the policy.

Identifying the benefits and costs of the lockdown

The benefits of the lockdown include lives saved and a reduction in suffering, not only for those who otherwise would have caught the virus but also for their family and friends. It also includes lives saved from other diseases whose treatment would have been put (even more) on hold if the pandemic had been allowed to rage and more people were hospitalised with the virus. In material terms, there is the benefit of saving in healthcare and medicines and the saving of labour resources. Then there are the environmental gains from less traffic and polluting activities.

On the cost side, there is the decline in output from businesses being shut and people being furloughed or not being able to find work. There is also a cost if schools have to close and children’s education is thereby compromised. Then there is the personal cost to people of being confined to home, a cost that could be great for those in cramped living conditions or in abusive relationships. Over the longer term, there is a cost from people becoming deskilled and firms not investing – so-called scarring effects. Here there are the direct effects and the multiplier effects on the rest of the economy.

Estimating uncertain outcomes

It is difficult enough identifying all the costs and benefits, but many occur in the future and here there is the problem of estimating the probability of their occurrence and their likely magnitude. Just how many lives will be saved from the policy and just how much will the economy be affected? Epidemiological and economic models can help, but there is a huge degree of uncertainty over predictions made about the spread of the disease, especially with a new strain of the virus, and the economic effects, especially over the longer term.

One estimate of the number of lives saved was made by Miles et al. in the NIESR paper linked below. A figure of 440 000 was calculated by subtracting the 60 000 actual excess deaths over the period of the first lockdown (March to June 2020) from a figure of 500 000 lives lost which, according to predictions, would have been the consequence of no lockdown. However, the authors acknowledge that this is likely to be a considerable overestimate because:

It does not account for changes in behaviour that would have occurred without the government lockdown; it does not count future higher deaths from side effects of the lockdown (extra cancer deaths for example); and it does not allow for the fact that some of those ‘saved’ deaths may just have been postponed because when restrictions are eased, and in the absence of a vaccine or of widespread immunity, deaths may pick up again.

Some help in estimating likely outcomes from locking down or not locking down the economy can be gained by comparing countries which have taken different approaches. The final article in the first list below compares the approaches in the UK and Sweden. Sweden had much lighter control measures than the UK and did not impose a lockdown. Using comparisons of the two approaches, the authors estimate that some 20 000 lives were saved by the lockdown – considerably less than the 440 000 estimate.

Estimating the value of a human life

To assess whether the saving of 20 000 lives was ‘worth it’, a value would have to be put on a life saved. Although putting a monetary value on a human life may be repugnant to many people, such calculations are made whenever a project is assessed which either saves or costs lives. As we say in the 10th edition of Economics (page 381):

Some people argue ‘You can’t put a price on a human life: life is priceless.’ But just what are they saying here? Are they saying that life has an infinite value? If so, the project must be carried out whatever the costs, and even if other benefits are zero! Clearly, when evaluating lives saved from the project, a value less than infinity must be given.
 
Other people might argue that human life cannot be treated like other costs and benefits and put into mathematical calculations. But what are these people saying? That the question of lives saved should be excluded from the cost–benefit study? If so, the implication is that life has a zero value! Again this is clearly not the case.

In practice, there are two approaches used to measure the value of a human life.

The first uses the value of a statistical life (VSL). This is based on the amount extra the average person would need to be paid to work in a job where there is a known probability of losing their life. So if people on average needed to be paid an extra £10 000 to work in a job with a 1% chance of losing their life, they would be valuing a life at £1 000 000 (£10 000/0.01). To avoid the obvious problem of young people’s lives being valued the same as old people’s ones, even though a 20 year-old on average will live much longer than a 70 year-old, a more common measure is the value of a statistical life year (VSLY).

A problem with VSL or VSLY measures is that they only take into account the quantity of years of life lost or saved, not the quality.

A second measure rectifies this problem. This is the ‘quality of life adjusted year (QALY)’. This involves giving a value to a year of full health and then reducing it according to how much people’s quality of life is reduced by illness, injury or poverty. The problem with this measure is the moral one that a sick or disabled person’s life is being valued less than the life of a healthy person. But it is usual to make such adjustments when considering medical intervention with limited resources.

One adjustment often made to QALYs or VSLYs is to discount years, so that one year gained would be given the full value and each subsequent year would be discounted by a certain percentage from the previous year – say, 3%. This would give a lower weighting to years in the distant future than years in the near future and hence would reduce the gap in predicted gains from a policy between young and old people.

Cost–effectiveness analysis (CEA)

Even using QALYs, there is still the problem of measuring life and health/sickness. A simpler approach is to use cost–effectiveness analysis (CEA). This takes a social goal, such as reducing the virus production rate (R) below 1 (e.g. to 0.9), and then finding the least-cost way of achieving this. As Mark Carney says in his third Reith Lecture:

As advocated by the economists Nick Stern and Tim Besley, the ideal is to define our core purpose first and then determine the most cost-effective interventions to achieve this goal. Such cost–effectiveness analysis explicitly seeks to achieve society’s values.

Cost–effectiveness analysis can take account of various externalities – as many of the costs will be – by giving them a value. For example, the costs of a lockdown to people in the hospitality sector or to the education of the young could be estimated and included in the costs. The analysis can also take into account issues of fairness by identifying the effects on inequality when certain groups suffer particularly badly from Covid or lockdown policies – groups such as the poor, the elderly and children. Achieving the goal of a specific R for the least cost, including external costs and attaching higher weights on the effects on certain groups then becomes the goal. As Carney says:

R brings public health and economics together. Relaxations of restrictions increase R, with economic, health and social consequences. A strategic approach to Covid is the best combination of policies to achieve the desired level of infection control at minimum economic cost with due respect for inequality, mental health and other social consequences, and calculating those costs then provides guidance when considering different containment strategies. That means paying attention to the impact on measures of fairness, the social returns to education, intergenerational equity and economic dynamism.

Conclusion

Given the uncertainties surrounding the measurement of the number of lives saved and the difficulties of assigning a value to them, and given the difficulties of estimating the economic and social effects of lockdowns, it is not surprising that the conclusions of a cost–benefit analysis, or even a cost–effectiveness analysis of a lockdown will be contentious. But, at least such analysis can help to inform discussion and drive future policy decisions. And a cost–effectiveness analysis can be a practical way of helping politicians reach difficult decisions about life and death and the economy.

Articles (original blog)

Articles (additional)

Questions

  1. What are the arguments for and against putting a monetary value on a life saved?
  2. Are QALYs the best way of measuring lives saved from a policy such as a lockdown?
  3. Compare the relative merits of cost–benefit analysis and cost–effectiveness analysis.
  4. If the outcomes of a lockdown are highly uncertain, does this strengthen or weaken the case for a lockdown? Explain.
  5. What specific problems are there in estimating the number of lives saved by a lockdown?
  6. How might the age distribution of people dying from Covid-19 affect the calculation of the cost of these deaths (or the benefits or avoiding them)?
  7. How might you estimate the costs to people who suffer long-term health effects from having had Covid-19?
  8. What are the arguments for and against using discounting in estimating future QALYs?
  9. The Department of Transport currently uses a figure of £1 958 303 (in 2018 prices) for the value of a life saved from a road safety project. Find out how this is figure derived and comment on it. See Box 12.5 in Economics 10th edition and Accident and casualty costs, Tables RAS60001 and RA60003, (Department of Transport, 2019).

It is commonplace to use cost–benefit analysis (CBA) in assessing public policies, such as whether to build a new hospital, road or rail line. Various attempts in the past few months have been made to use CBA in assessing policies to reduce the spread of the coronavirus. These have involved weighing up the costs and benefits of national or local lockdowns or other containment measures. But, as with other areas where CBA is used, there are serious problems of measuring costs and benefits and assessing risks. This is particularly problematic where human life is involved and where a value has to be attached to a life saved or lost.

Take the case of whether the government should have imposed a lockdown: an important question if there were to be a second wave and the government was considering introducing a second lockdown. The first step in a CBA is to identify the benefits and costs of the policy.

Identifying the benefits and costs of the lockdown

The benefits of the lockdown include lives saved and a reduction in suffering, not only for those who otherwise would have caught the virus but also for their family and friends. It also includes lives saved from other diseases whose treatment would have been put (even more) on hold if the pandemic had been allowed to rage and more people were hospitalised with the virus. In material terms, there is the benefit of saving in healthcare and medicines and the saving of labour resources. Then there are the environmental gains from less traffic and polluting activities.

On the cost side, there is the decline in output from businesses being shut and people being furloughed or not being able to find work. There is also a cost from schools being closed and children’s education being compromised. Then there is the personal cost to people of being confined to home, a cost that could be great for those in cramped living conditions or in abusive relationships. Over the longer term, there is a cost from people becoming deskilled and firms not investing – so-called scarring effects. Here there are the direct effects and the multiplier effects on the rest of the economy.

Estimating uncertain outcomes

It is difficult enough identifying all the costs and benefits, but many occur in the future and here there is the problem of estimating the probability of their occurrence and their likely magnitude. Just how many lives will be saved from the policy and just how much will the economy be affected? Epidemiological and economic models can help, but there is a huge degree of uncertainty over predictions made about the spread of the disease and the economic effects, especially over the longer term.

One estimate of the number of lives saved was made by Miles et al. in the NIESR paper linked below. A figure of 440 000 was calculated by subtracting the 60 000 actual excess deaths over the period of the lockdown from a figure of 500 000 lives lost which, according to predictions, would have been the consequence of no lockdown. However, the authors acknowledge that this is likely to be a considerable overestimate because:

It does not account for changes in behaviour that would have occurred without the government lockdown; it does not count future higher deaths from side effects of the lockdown (extra cancer deaths for example); and it does not allow for the fact that some of those ‘saved’ deaths may just have been postponed because when restrictions are eased, and in the absence of a vaccine or of widespread immunity, deaths may pick up again.

Some help in estimating likely outcomes from locking down or not locking down the economy can be gained by comparing countries which have taken different approaches. The final article below compares the approaches in the UK and Sweden. Sweden had much lighter control measures than the UK and did not impose a lockdown. Using comparisons of the two approaches, the authors estimate that some 20 000 lives were saved by the lockdown – considerably less than the 440 000 estimate.

Estimating the value of a human life

To assess whether the saving of 20 000 lives was ‘worth it’, a value would have to be put on a life saved. Although putting a monetary value on a human life may be repugnant to many people, such calculations are made whenever a project is assessed which either saves or costs lives. As we say in the 10th edition of Economics (page 381):

Some people argue ‘You can’t put a price on a human life: life is priceless.’ But just what are they saying here? Are they saying that life has an infinite value? If so, the project must be carried out whatever the costs, and even if other benefits are zero! Clearly, when evaluating lives saved from the project, a value less than infinity must be given.
 
Other people might argue that human life cannot be treated like other costs and benefits and put into mathematical calculations. But what are these people saying? That the question of lives saved should be excluded from the cost–benefit study? If so, the implication is that life has a zero value! Again this is clearly not the case.

In practice there are two approaches used to measuring the value of a human life.

The first uses the value of a statistical life (VSL). This is based on the amount extra the average person would need to be paid to work in a job where there is a known probability of losing their life. So if people on average needed to be paid an extra £10 000 to work in a job with a 1% chance of losing their life, they would be valuing a life at £1 000 000 (£10 000/0.01). To avoid the obvious problem of young people’s lives being valued the same as old people’s ones, even though a 20 year-old on average will live much longer than a 70 year-old, a more common measure is the value of a statistical life year (VSLY).

A problem with VSL or VSLY measures is that they only take into account the quantity of years of life lost or saved, not the quality.

A second measure rectifies this problem. This is the ‘quality of life adjusted year (QALY)’. This involves giving a value to a year of full health and then reducing it according to how much people’s quality of life is reduced by illness, injury or poverty. The problem with this measure is the moral one that a sick or disabled person’s life is being valued less than the life of a healthy person. But it is usual to make such adjustments when considering medical intervention with limited resources.

One adjustment often made to QALYs or VSLYs is to discount years, so that one year gained would be given the full value and each subsequent year would be discounted by a certain percentage from the previous year – say, 3%. This would give a lower weighting to years in the distant future than years in the near future and hence would reduce the gap in predicted gains from a policy between young and old people.

Conclusion

Given the uncertainties surrounding the measurement of the number of lives saved and the difficulties of assigning a value to them, it is not surprising that the conclusions of a cost–benefit analysis of a lockdown will be contentious. And we have yet to see what the long-term effects on the economy will be. But, at least a cost–benefit analysis of the lockdown can help to inform discussion and help to drive future policy decisions about tackling a second wave, whether internationally, nationally or locally.

Articles

Questions

  1. What are the arguments for and against putting a monetary value on a life saved?
  2. Are QALYs the best way of measuring lives saved from a policy such as a lockdown?
  3. If the outcomes of a lockdown are highly uncertain, does this strengthen or weaken the case for a lockdown? Explain.
  4. What specific problems are there in estimating the number of lives saved by a lockdown?
  5. How might the age distribution of people dying from Covid-19 affect the calculation of the cost of these deaths (or the benefits or avoiding them)?
  6. How might you estimate the costs to people who suffer long-term health effects from having had Covid-19?
  7. What are the arguments for and against using discounting in estimating future QALYs?
  8. The Department of Transport currently uses a figure of £1 958 303 (in 2018 prices) for the value of a life saved from a road safety project. Find out how this is figure derived and comment on it. See Box 12.5 in Economics 10th edition and Accident and casualty costs, Tables RAS60001 and RA60003, (Department of Transport, 2019).

The Winter Olympics are full on as athletes from all over the world compete against each other, hoping to set new world records, win medals and be known as Olympians. Pyeongchang, the South Korean county that hosts the 2018 Winter games, enjoys a large influx of tourists – estimated at 80,000 people a day. This is certainly an unusually large number of tourists for a region that has a regular winter-time population of no more than 45,000 people.

Having such a high number of visitors to the Winter Olympics, and even more to the larger Summer Olympics, is not an unusual occurrence, however, and it is often mentioned as one of the benefits of being a host to the Olympic Games.

Baade and Matheson (see link below) distinguish between three key benefits of hosting the Olympic Games: “the short-run benefits of tourist spending during the Games; the long-run benefits or the ‘Olympic legacy’, which might include improvements in infrastructure and increased trade, foreign investment, or tourism after the Games; and intangible benefits such as the ‘feel-good effect’ or civic pride”.

On these grounds, a number of studies have been authored, attempting to analyse some or all of these benefits, distinguishing between short-term and long-term effects. Müller (see link below), uses data from the 2014 Oympic Games in Sochi, Russia, to assess the net economic outcome for the host region. He concludes that any short-term economic benefits caused by the investment influx (before and during the games) could not offset the long-term costs, leading to an estimated net loss of $1.2 billion per year.

Zimbalist (2015) and Szymanski (2011) report similar results when analysing data from the London Games (2012) and past major sporting events (Games and FIFA World Cup). Kasimati (2003) points out the significant economic benefits that host regions tend to enjoy for years after hosting the games, but argues that the overall effect depends on a number of factors (including pre-existing infrastructure and location).

The jury is, therefore, still out on what is the overall economic effect of being host to this ancient institution. But I must now dash as women’s hockey is soon to start. “Let everyone shine”.

Articles

For the sake of the games, South Korea needs to show hosting an Olympics can be economically viable CNBC, Yen Nee Lee (15/2/18)
South Korea’s Olympic bet is unlikely to pay off, economics professor says CNBC, Andrew Wong and Andrew Zimbalist (12/2/18)
Going for the Gold: The Economics of the Olympics Journal of Economic Perspectives, Robert A. Baade and Victor A. Matheson (Spring 2016)
After Sochi 2014: Costs and Impacts of Russia’s Olympic Games Eurasian Geography and Economics, Martin Müller (9/4/15)
Circus Maximus: The Economic Gamble Behind Hosting the Olympics and the World Cup The Brookings Institution, Andrew Zimbalist (14/1/15)
About Winning: The Political Economy of Awarding the World Cup and the Olympic Games SAIS Review of International Affairs, Stefan Szymanski (Winter/Spring 2011)
Economic aspects and the Summer Olympics: a review of related research International Journal of Tourism Research, Evangelia Kasimati (4/11/03)
“Let Everyone Shine”: the song for the PyeongChang 2018 Torch Relay unveiled with 200 days to go Olympic Committee (24/7/17)

Video

The Olympic Winter Games PyeongChang 2018 Torch Relay Official Song PyeongChang 2018

Questions

  1. Using supply and demand diagrams, explain whether you would expect hotel room prices to change during the hosting of a major sports event, such as the Winter Olympics.
  2. List three economic (or economics-related) arguments in favour of and against the hosting of the Olympic games. Relate your answer to the empirical evidence presented in the literature.
  3. Why is it so difficult to estimate with accuracy the net economic effect of the Olympic Games?

With first Houston, then several Caribbean islands and Florida suffering dreadful flooding and destruction from Hurricanes Harvey and Irma, many are questioning whether more should be spent on flood prevention and reducing greenhouse gas emissions. Economists would normally argue that such questions are answered by conducting a cost–benefit analysis.

However, even if the size of the costs and benefits of such policies could be measured, this would not be enough to give the answer. Whether such spending is justified would depend on the social rate of discount. But what the rate should be in cost-benefit analyses is a highly contested issue, especially when the benefits occur a long time in the future.

I you ask the question today, ‘should more have been spent on flood prevention in Houston and Miami?’, the answer would almost certainly be yes, even if the decision had to have been taken many years ago, given the time it takes to plan and construct such defences. But if you asked people, say, 15 years ago whether such expenditure should be undertaken, many would have said no, given that the protection would be provided quite a long time in the future. Also many people back then would doubt that the defences would be necessary and many would not be planning to live there indefinitely.

This is the familiar problem of people valuing costs and benefits in the future less than costs and benefits occurring today. To account for this, costs and benefits in the future are discounted by an annual rate to reduce them to a present value.

But with costs and benefits occurring a long time in the future, especially from measures to reduce carbon emissions, the present value is very sensitive to the rate of discount chosen. But choosing the rate of discount is fraught with difficulties.

Some argue that a social rate of discount should be similar to long-term market rates. But market rates reflect only the current generation’s private preferences. They do not reflect the costs and benefits to future generations. A social rate of discount that did take their interests into account would be much lower and could even be argued to be zero – or negative with a growing population.

Against this, however, has to be set the possibility that future generations will be richer than the current one and will therefore value a dollar (or any other currency) less than today’s generation.

However, it is also likely, if the trend of recent decades is to continue, that economic growth will be largely confined to the rich and that the poor will be little better off, if at all. And it is the poor who often suffer the most from natural disasters. Just look, for example, at the much higher personal devastation suffered from hurricane Irma by the poor on many Caribbean islands compared with those in comparatively wealthy Florida.

A low or zero discount rate would make many environmental projects socially profitable, even though they would not be with a higher rate. The choice of rate is thus crucial to the welfare of future generations who are likely to bear the brunt of climate change.

But just how should the social rate of discount be chosen? The following two articles explore the issue.

Articles

How Much Is the Future Worth? Slate, Will Oremus (1/9/17)
Climate changes the debate: The impact of demographics on long-term discount rates Vox, Eli P Fenichel, Matthew Kotchen and Ethan T Addicott (20/8/17)

Questions

  1. What is meant by the social rate of discount?
  2. Why does the choice of a lower rate of social discount imply a more aggressive climate policy?
  3. How is the distribution of the benefits and costs of measures to reduce carbon emissions between rich and poor relevant in choosing the social rate of discount of such measures?
  4. How is the distribution of the benefits of such measures between current and future generations relevant in choosing the rate?
  5. How is uncertainty about the magnitude of the costs and benefits relevant in choosing the rate?
  6. What is the difference between Stern’s and Nordhaus’ analyses of the choice of social discount rate?
  7. Explain and discuss the ‘mortality-based approach’ to estimating social discount rates.
  8. What are the arguments ‘for economists analysing climate change through the lens of minimising risk, rather than maximizing utility’?

Many politicians throughout the world,
not just on the centre and left, are arguing for increased spending on infrastructure. This was one of the key proposals of Donald Trump during his election campaign. In his election manifesto he pledged to “Transform America’s crumbling infrastructure into a golden opportunity for accelerated economic growth and more rapid productivity gains”.

Increased spending on inffrastructure has both demand- and supply-side effects.

Unless matched by cuts elsewhere, such spending will increase aggregate demand and could have a high multiplier effect if most of the inputs are domestic. Also there could be accelerator effects as the projects may stimulate private investment.

On the supply side, well-targeted infrastructure spending can directly increase productivity and cut costs of logistics and communications.

The combination of the demand- and supply-side effects could increase both potential and actual output and reduce unemployment.

So, if infrastructure projects can have such beneficial effects, why are politicians often so reluctant to give them the go-ahead?

Part of the problem is one of timing. The costs occur in the short run. These include demolition, construction and disruption. The direct benefits occur in the longer term, once the project is complete. And for complex projects this may be many years hence. It is true that demand-side benefits start to occur once construction has begun, but these benefits are widely dispersed and not easy to identify directly with the project.

Then there is the problem of externalities. The external costs of projects may include environmental costs and costs to local residents. This can lead to protests, public hearings and the need for detailed cost–benefit analysis. This can delay or even prevent projects from occurring.

The external benefits are to non-users of the project, such as a new bridge or bypass reducing congestion for users of existing routes. These make the private construction of many projects unprofitable, except with public subsidies or with public–private partnerships. So there does need to be a macroeconomic policy that favours publicly-funded infrastructure projects.

One type of investment that is less disruptive and can have shorter-term benefits is maintenance investment. Maintenance expenditure can avoid much more costly rebuilding expenditure later on. But this is often the first type of expenditure to be cut when public-sector budgets as squeezed, whether at the local or national level.

The problem of lack of infrastructure investment is very much a political problem. The politicians who give the go-ahead to such projects, such as high-speed rail, come in for criticisms from those bearing the short-run costs but they are gone from office once the benefits start to occur. They get the criticism but not the praise.

Articles

Are big infrastructure projects castles in the air or bridges to nowhere? The Economist, Buttonwood’s notebook (16/1/17)
Trump’s plans to rebuild America are misguided and harmful. This is how we should do it. The Washington Post, Lawrence H. Summers (17/1/17)

Questions

  1. Identify the types of externality from (a) a new high-speed rail line, (b) new hospitals.
  2. How is discounting relevant to decisions about public-sector projects?
  3. Why are governments often unwilling to undertake (a) new infrastructure projects, (b) maintenance projects?
  4. Is a programme of infrastructure investment necessarily a Keynesian policy?
  5. What accelerator effects would you expect from infrastructure investment?
  6. Explain the difference between the ‘spill-out’ and ‘pull-in’ effects of different types of public investments in a specific location. Is it possible for a project to have both effects?
  7. What answer would you give to the teacher who asked the following question of US Treasury Secretary, Larry Summers? “The paint is chipping off the walls of this school, not off the walls at McDonald’s or the movie theatre. So why should the kids believe this society thinks their education is the most important thing?”
  8. What is the ‘bridge to nowhere’ problem? Why does it occur and what are the solutions to it?
  9. Why is the ‘castles in the air’ element of private projects during a boom an example of the fallacy of composition?