Presentation to Macrogroup Seminar, Friday 21 November, 2025; there is a PowerPoint that goes with this presentation. Available on request.

Social Investment Analysis is another example of project evaluation where economists use Cost-Benefit Analysis. By the 1960s New Zealand was using CBA in the evaluation of irrigation and transport projects when, for various reasons, market prices did not reflect social values. They include distorted market prices, externalities and the significant intangible costs and benefits. The CBA paradigm has been to convert the analysis to the neoclassical market model which has properties economists are well aware of, and which they use fluidly. CBAs came into major use evaluating the Major Projects of the 1980s (‘Think Big’); it became increasingly broadened to projects which have a strong social dimension with outcomes that cannot be valued in the market – ‘intangibles’. Social Investment Analysis is a recent extension.

Most of those in this room will have done CBAs and will be aware of sloppily done CBAs by others – sadly there have been too many. Today I thought I could usefully set out some of the trickier issues which I have come across in my own CBA work, especially from the health sector. Some of this will be familiar – allow the revision.

I shall particularly refer to the learnings from the International Working Party on the Evaluation of the Economic and Social Costs of Substance Abuse, which met over eight years and whose report, International Guidelines for Estimating the Costs of Substance Abuse, WHO published.

Social-cost studies are project evaluation studies. They are the first leg of a CBA. We used that approach so that someone who wanted to do a CBA on a particular policy could use a relevant social-cost study.

Throughout this paper, I use ‘social cost’ to refer to ‘net social cost’, that is offsetting any extra resources by any benefits. There were many social-cost studies on drug abuse before our work, but they were often done with little connection to economic analysis, with sloppy conceptualisation perhaps generating large numbers to meet the client’s desire to gain a media headline. The working party’s task was to provide a rigorous economic conceptual foundation. It consisted of a number of economists from various countries, but it also included epidemiologists and experts in substance abuse. I learned a lot from them; as one almost invariably does in complex issues outside economists’ expertise.

The Counterfactual

Because we are using an economic framework, the first principle we insisted upon was that ‘cost’ for an economist means ‘opportunity cost’ – the value of a resource in its best alternative use.  Opportunity cost underpins the efficiency notion in economics, but it is frequently overlooked by others – and even by the odd economist.

A social cost study – all project evaluations – involve a comparison between scenarios, typically the status quo and what is known as the ‘counterfactual scenario’. The opportunity cost is the difference between them.

Sometimes the choice of comparing the two scenarios is complicated. One of the substance-use experts pointed out that banning alcohol consumption in Pakistan had generated a cocaine problem.

When I did my social cost of alcohol misuse study, I chose to compare the status quo with a scenario where there was no alcohol misuse, since limited alcohol consumption is pleasurable without generating the externalities which cause the social costs. (Additionally, at that time low levels of alcohol consumption were thought to beneficial the imbiber’s health; this ‘J-curve’ affect is largely discounted nowadays.)

But when I did the tobacco use study the comparison had to be made with the counterfactual that there had never been tobacco consumption in New Zealand; terminating all tobacco consumption today would still leave an overhang of healthcare and early death from past smoking. (Dealing with such overhangs involved calcualtions of ‘avoidable costs’.)

Specifying the scenarios is critical. A recent social cost of alcohol study did not. It was criticised by another economist which resulted in a very confused discussion. As far as I could work out, the two were discussing different counterfactual scenarios but each was so vague about what they were talking about one could not be sure.

The Distribution of the Impact

It is also important to be clear about the frame of reference – whose costs? For instance, when we were evaluating the 1980s Major Projects we ignored their impact on those outside New Zealand.

Consider an evaluation of a new healthcare treatment. Do we do it from the perspective of the healthcare system, or the government – the fiscal approach – or the community as a whole? A more expensive treatment in the healthcare system may result in the patient getting back to work earlier or using less of government services provided by other government agencies, reducing the cost to the government even though it was more costly to the healthcare system.

The fiscal approach ignores the impact upon the patient and their associates, major concerns when the perspective is the whole community. Ignoring the community impact biases decisions towards cost shifting onto private individuals and households.

The working party recommended a whole-of-community perspective but struggled with distributional issues. There is no simple resolution without a welfare function. As is typical in distributional economics, there is no single statistic which summarises the equity outcome without imposing a value judgement.

It matters. During the Major Project debate there was advocacy from localities which benefited from a project being built there even if the rest of the country suffered – in effect the locality was being subsidised by everyone else.

Or consider the political challenge that Pharmac continually faces. There is always strong pressure from patients who may benefit from a pharmaceutical, irrespective of how expensive the drug is; it may not be cost-effective when the community faces so many other demands.

The working party concluded that the distributional impact of the counterfactual should best be dealt with by tabulating the social cost to the various distributional groups separately.

The Value of Life

The early studies on irrigation schemes and the Major Projects were narrow, assessed only by their impact on output. But even at that time, the transport system was concerned about road design which reduced accidents and saved lives, which involves imputing a value for the lives saved.

Lives saved is too simple a concept for healthcare. It may not make sense to prioritise a cheap treatment which extends life for a short period compared to an expensive one which eliminates the condition so that the treated life a normal length of life.

In alcohol misuse and tobacco use studies, tobacco kills far more people than alcohol. But tobacco deaths tend to be of smokers towards the end of their normal lives, whereas alcohol kills younger people by accidents. When I switched from measuring lives-saved to life-years-saved, I found that the disparity between the two effects was substantially reduced.

Even that is not the whole story, especially in healthcare. Today, many treatments are not about saving or prolonging life. They are also about improving the quality of life – a much fuzzier concept.

The working party chose to use QALYs, quality adjusted life years, as its measure for life improvement. QALYs warrant a seminar in their own right. Briefly, health states are scaled between unity, when there is a full quality of life, and zero, when there is death. Surveys of individuals’ subjective assessments are then used to score states between. In my experience the scale can be very clumsy despite the enormous amount of international work to refine it. [One of its world experts is New Zealander, Nancy Devlin.] However, the QALY measure is the best we have.

Putting a value – a price – on a QALY proves conceptually treacherous. When the economists on the working party were discussing how to measure it, an epidemiologist said, ‘We Moslems do not accept there is such a “value of life”.’ Another added ‘neither do we Catholics’.

That set the economists back. What are we trying to measure? Economists need the value to get useful answers. Set the value of life at zero, we would adopt policies/actions which would kill people; set it at infinity, the policies would try to eliminate death irrespective of the resources available and their alternative uses.

The discussion in the WHO report is tortuous but what it amounts to is this: recall that the framework being used is CBA, which has to put a dollar value on life in order to understand the tradeoffs necessary to make sensible resource allocation decisions. Since social-cost studies are intended to be foundations for CBAs, they have to use the same conceptual frame for the value of life.

We settled on the willingness-to-pay measure to estimate the value of life. These involve surveys in which individuals are asked how much they are willing to pay (in taxes) for actions (such as public healthcare) which may extend their lives. They are being asked to trade off the certainty of income now (thereby reducing the current quality of their lives) to increase to the quality life years in other circumstances. Valuing life in a CBA is about a trade-off, not an absolute value.

(Life can be valued in other ways. An earlier approach was to value it as the discounted future earnings of an individual. There are a variety of objections, including that it would mean that the value of a woman’s life is less than a man’s because she earns less over her lifetime.)

In social-cost studies which use willingness-to-pay estimates of the value of life the magnitude of the intangibles of the value of life usually far exceeds the value of the market resources, often exceeding annual GDP. What this suggests is that life is very much more important than material consumption, perhaps by a factor of ten. It is a humbling thought for economists; we are working in only a small corner of the human garden. (I think we garden a lot better than some of our colleague professions who think their bit of the garden is just as important.)

Miscellaneous

Before developing this as my final point, I mention briefly that there were other significant issues with which the working party struggled. For example, as in the case of the Major Projects debate, there was the choice of the discount rate for comparisons involving time. It greatly affects the magnitude of the social-cost estimate measured as a stock – the gee-whiz figure we like to discuss. Discount that figure to an annual flow and there is not too much difference. (Nowadays a sensitivity analysis is used in a CBA to investigates the impact of different discount rates.) 

Moreover, over time economic understanding progresses and the data base improves. (One of the benefits of social cost studies is to draw attention to deficiencies in the data base.) The working party did not, for instance, deal with the challenge behavioural economics poses of time inconsistency, which may be particularly important when there is addiction. Paul Krugman describes behavioural economics as the most revolutionary thing which has happened to economics for ages. Our paradigm was heavily anchored in neoclassical economics before the revolution.*

Conclusion

This paper reviews some lessons I learned over the years for CBAs particularly relevant to Social Investment Analysis. In another venue I would trace how the experience has played a critical role in my arguing that economists focus too narrowly on material output provided by the market and that we should be looking at a wider – if fuzzier – welfare notion, currently called ‘wellbeing’.

* The subsequent discussion raised how to deal with the impact of a personal event on others – a treatment delaying a death for a short time may enable family and friends to gather, grieve and adjust. The neoclassical paradigm pays little attention to interpersonal utility (although by treating a household as a person it allows interpersonal utilties with in it – but not in market transactions.)