Cost-Benefit Analysis

Government departments like to do cost-benefit analysis before making policy decisions. They evaluate the benefits and assess the costs against them to ascertain whether a policy is justified. Of course, political factors will often overrule the outcomes of the cost-benefit analysis.

Governments make cost-benefit decision all the time, even where human lives are involved. For example, if the government was to build median barriers on every single road in the county, it would reduce the road toll, but the horrific cost means that no government would ever consider doing it. Because some country roads have very few vehicles on them, the benefit does not justify the cost.

Unfortunately, as I explain in an article called cost-benefit analysis the information needed to support good cost-benefit analysis often does not exist. Although there is not much evidence of serious cost-benefit analysis, this is particularly true in the case of the policy decision to impose a major lockdown of the economy to reduce the spread of the coronavirus.


The people making the decision simply did not have the information that they needed to make a really confident decision. On the benefit side, it is almost impossible to assess how effective a four-week shutdown will be for stopping the spread of the virus. It is well understood that quarantines have an effect in reducing disease, but it is less certain about how effective they will be in this case.

The problem is that the key information about the virus is still sketchy. Scientists cannot conduct controlled experiments in which they infect people and measure the outcomes. That would be immoral. Instead, they undertake statistical analysis of the existing cases and identify patterns and measure outcomes from past actions and events.

Strong statistics need robust definitions and classifications of data to be meaningful. And to calculate rates affected in a particular way, they need an accurate measure of the total number of people infected (the denominator). In the case of Covid19, the definitions used have changed significantly over time, and from place to place. "Cause of death" is very difficult to measure at the best of times, because in many cases, a death will have many causes, and it will not be clear which one was the final cause. For some cases included in the statistics, Covid19 may not have been the actual cause of death, even if it was present.

And because not everyone has serious symptoms, and not everyone is tested, it is impossible to know the total number that is actually affected. The uncertain denominator undermines the reliability of any measure of death rates, spread rates and infection rates for different nations and particular groups of the population.

During a pandemic, clinicians are working flat tack to care for the people who are sick. They do not have time to sit around filling in forms and counting numbers of patients, or hold committee discussions to ensure that the cause of a death is recorded accurately.

This was particularly true in Wuhan when the disease first broke out. At first, they did not even realise that they were dealing with a new virus. Doctors look for sicknesses that they know about, so by the time they realised it was a new virus, they were unable to go back and count up the statistics accurately. They did well to complete a DNA analysis of the virus and make it available to scientist around the world in early January. Not surprisingly, their statistics are frequently questioned.

The statistics that the medical bureaucrats publish are going to be full of errors that can make them misleading. For example, the media quote each day the total number of cases in the world, as if it was the correct number. Actually, those counting are just giving an estimate, with a huge amount of error attached to it. The statisticians will hope the double-counts in the data will be balanced out by the missing records, but there is no way of knowing.

This uncertainty about numbers is one reason why there is so much debate about the death rate for Covid19. Both the numbers for the numerator and the denominator are weak estimates, so the calculated rates need to be taken with a grain of salt.

Since they cannot do experiments to measure the outcome of various policy actions, health experts have to develop models to estimate the number of deaths if particular policy actions are taken, or not taken. The validity of these models depends on the robustness of the data that is fed into the model. Unfortunately, much of the data needed for these models, such as death rates and spread rates, is of uncertain quality.

The other problem with models is that researchers need to make assumptions about human behaviour and the way people will respond in different situations. A model is only as strong as the assumptions underlying it. In a way, the results that the model produces can be easily tweaked to get the results that are wanted, and researchers are often pressured to do that. The experience and judgment of the people developing the models will be critical for getting reliable results.

I suspect that part of the problem is the skills and experience of the people in the Ministry of Health who are making the policy recommendations to the government. I imagine that "public health" is seen as a bit of backwater by medical professionals, so it will not attract the best brains. I presume that most trained doctors can make better money in other fields, so those who end up the Ministry of Health are not likely to be the top people.

This seems to be confirmed by the lack of planning that is becoming evident. On the 27th January, a press release from the Ministry of Health said that "There is still a risk that the illness will arrive in New Zealand. However, the likelihood of further transmission - if we do get a case - to another person, is low to moderate, and the likelihood of a sustained outbreak in New Zealand remains low." This was surprising given that so many Chinese tourists visit New Zealand and students arrive to start their study at this time of year.

This lack of concern probably explains why the Ministry of Health has been so slow to order more ventilators and personal protection equipment like masks. It also seems to explain why they have been captured by the health academics who shouted the loudest in support of policy changes.

All this means that the benefit side of the cost-benefit analysis is very uncertain. The number of deaths the shutdown will actually prevent is not very clear. The government has quoted the worst-case outcomes of the modelling, but that has been mainly to scare people into self-isolating as the policy requires.


The costs of the economic shut down will be extensive. The media are asking for the release of Treasury estimates, but they probably don't have any that are defensible. Private-sector economists can make estimates without much accountability, but those in the Treasury need to be able to defend their estimates and be careful they don't cause unwise behaviour by scaring people unnecessarily.

Measuring economic activity is never easy, even in ordinary times. Estimates of GDP are built up from numerous statistical series which have different degrees of reliability. The model used involves many assumptions and weights for aggregating series together. These assumptions and weights are justified in times of gradual economic change, but when a calamitous shock occurs, they probably breakdown.

The Department of Statistics will have difficulty measuring the increase in unemployment and the decline in production that results from the shutdown. The standard method for measuring unemployment relies on a household survey conducted by personal visit and telephone. They will be more difficult to conduct during a shutdown when homes cannot be visited.

Output is measured using a mix of surveys and administrative data, particularly tax data. Survey data might be more difficult to collect, and administrative data will be difficult to analyse and interpret, because businesses that have closed might still be making tax payments and paying final wage payments to employees whom they have already let go. The statisticians will attempt to put some mitigations in place, but they will be less confident in their results at a time when they are more important.

If measuring the economic outcome of the shutdown will be difficult, estimating the impact in advance is almost impossible. The Treasury has economic models, but the assumptions and economic relationships that undergird them will probably break down in a time of such calamitous change. The judgement of the skilled analyst interpreting the results will be really important.

Insiders have recently suggested that there has been a significant decline in the skills and expertise of Treasury officers in the last few years. If this is true, they will be struggling to come up with robust estimates of the impact of the shutdown and asses their reliability.

At least, they are good people who want to do the right thing, even if they do not know what to do. This is different from the situation in the United States, where the same group of people work in the Federal Reserve, the US Treasury and the big investment banks like Goldman Sachs. They rotate from one to the other to get ahead, so they all think the same way, and they all believe that what they are doing is important and worth big money. They do not know what to do either, so they just keep doing more of the same, hoping that it will work, and looking out for their mates.

A big problem for these experts is that they simply do not know how long that the shutdown will last. They probably don't have much information about the cash reserves of businesses, as they go into the shutdown. And they have no understanding of how the owners and managers of businesses are thinking, and how many of them are already feeling that it will too hard to keep going.

All this means that when it comes to assessing the costs of the shutdown (to balance against the benefits) Treasury economists are driving in the dark.

Government Decisions

When deciding if a shut down of the economy is the correct response to a pandemic, no one has the information that is needed to undertake a cost-benefit analysis that would support good decision making. I don't have the information to know if they have done the right thing. Politicians are conservative. They have all learnt the lesson from George W Bush during Hurricane Katrina that it is better to do something, even if it is the wrong thing, than being seen to do nothing. Politicians now have a bias for action, so I presume that they would have introduced a lockdown, even if the decision was not supported by cost-benefit analysis.

The lack of information for cost-benefit and marginal analysis has another serious consequence. No one really knows what effect social-distancing will have on the spread of infection or how effective quarantining has been. Governments will have to decide when and how to reverse the shutdown. That decision will have to be made with very little supporting evidence. If they reverse too soon, a second outbreak of infection might occur. If they act too late, they will exacerbate the damage to the economy.

So again, the politicians will be conservative and wait till they are rarely certain before winding back the shutdown. This is an election year, so they don't want to be embarrassed by a second outbreak of infections, when the election is closer. They can always justify the economic costs by claiming that there was no other alternative, even though there always is.


One week into the lockdown, it seems to be slowing the spread of the virus. Most hospitals only have one Covid19 patient, so the health system is not being overwhelmed. However, in a months' time when the real economic costs are becoming clearer, I expect that people might be saying that the lockdown was unnecessarily severe.

Already, the biggest mistake seems to have been leaving the border open to cruise ships too long, and not restricting visitors from places where the virus had spread. They shut out visitors from China quite early, but then were too slow to restrict visitors from Australia, USA and Europe when it was spreading there.

Another mistake was telling citizens who were travelling overseas to come home quickly, without having a plan in place to screen the huge numbers arriving at airports for symptoms and risk of infection, and enforcing a strict quarantine on those deemed to be a risk. These people were allowed to trek back to their homes on places while mixing with people who had not travelled. Much of the spread of the virus has come from them.

I realise that these people are glad to be home, and that the parents of travelling children are glad to have them back with them, but they should have had health insurance to pay for health care where they were, and they should realise that their quick return has had a significant cost for the rest of the nation by making more a more severe and prolonged shut down necessary. Costs and benefits usually fall unevenly.

This was privileging of people who had travelled overseas, as friends and families of politicians and senior bureaucrats could afford to do, at the expense of ordinary people working to keep the country going who cannot afford overseas travel.

The nation has paid generals and admiral to make plans for dealing with imaginary military invasions and spent a massive amount on military toys for the boys and girls of the military, but it seems that there were not many people being paid to plan for the next pandemic, even though health experts were certain that one would come soon.

Looking back with the benefit of hindsight, I can see that many mistakes were made, but if there were people planning and brainstorming various scenarios, they would have been able to predict many of these problems and have been better prepared for dealing with them. Here are a few that seem obvious:

NZ has 12,000 active military personnel. Having a few thousand people trained and equipped for pandemic protection would have been better value for money, even if they had nothing to do for a few years. Successive governments have grossly underfunded public health preventative activities.
Do not put your trust in bureaucrats,
nor in a son of man, in whom there is no help (Psalm 146:3).
It is better to take refuge in the Lord
than to trust in bureaurcats (Psalm 118:9).