Consenting to inequality: The distributional consequences of Copenhagen Consensus approach


Bjorn Lomborg is, undoubtedly, seriously concerned with poverty and inequality. Both in the work of the Copenhagen Consensus Center (CCC) and in his popular writings, this is a common theme. In this context, he has championed some very progressive ideas, including eradicating barriers to international migration.[1] Unfortunately, he has also used rather distorted arguments about inequality to attack some of his favourite bugbears, such as subsidies for renewable energy.

The problem I want to address in this post is that the central methodology of Lomborg and the CCC is at best blind to inequality and, in its application, could tend towards policy prescriptions that increase inequality. Moreover, as we shall see, there are good arguments to suggest that if we take a broader view of inequality to include intergenerational equality, the CCC methodology is not even equality-blind; it is equality averse.

Inequality Blind Methodology

The basic idea of Cost Benefit Analysis (CBA) is straightforward and, indeed, literally high school-level economics. You work out the economic cost of a particular investment (or policy) and estimate its economic benefits (including monetarized estimates of less directly economic costs and benefits such as health), and generate a ratio of benefits to costs (BCR, Benefit-Cost Ratio). The idea of the CCC is that this allows us to rank policy interventions in climate change, international development, or other global challenges.

The methodology is, in itself, blind to its distributional consequences. This is because, in its simple application, it is based on a Benthamite assumption that the objective is utility maximization irrespective of inequality. Put simply, an investment of $100 that returns $1000 accruing to an already rich person is, in CBA terms, better than an investment with the same cost that generates $800 return that accrues to a poor person.

As Duke University’s Matthew Adler has consistently argued, the CBA methodology can be adjusted relatively easily to incorporate ‘aversion’ to inequality by simply weighting the cost and benefit calculation according to whom it benefits. In a simplistic scenario in which the world is divided into ‘poor’ and ‘rich’, we might for instance weight benefits accruing to ‘poor’ people twice as highly as benefits accruing to rich people. Applying this weighting to our previous example would reverse the ranking of the investments; the $800 return to the poor person would now be weighted as worth $1,600, making it more attractive than the $1000 return to the rich person.

I should stress that this is technically easy in the sense that it is quite a simple calculation, even in more realistic situations where you have gradations of wealth and poverty rather than two straightforward categories. But it is ethically more difficult. How much inequality aversion should we build in? This will necessarily be a somewhat arbitrary decision and subject to contested views.

Indeed, there are a number of measures of inequality, notably the Atkinson Index, which are derived from a Social Welfare Function[2] and require an exogenous (independently-determined) inequality aversion parameter, and agreeing an appropriate value for this parameter has proved impossible.

This is not surprising, and we should not expect it to be possible. Precisely the point of the inequality aversion parameter is that it allows different individual, groups, or societies to make different value judgements about tolerable levels of inequality, or for a single individual (or group, or society) to compare outcomes at a range of levels of aversion.

The same could be applied to incorporating inequality within CBA analysis. One would not need to decide on a single inequality aversion parameter; one could compute a range of values (including zero aversion, that is the raw CBR), and present the relative rankings of different investments (or policies) accordingly.

Distributional Impacts of CBA Methods in International Development

Lomborg might assert that this is all very pretty, but because his centre is set up to look primarily at problems affecting the poor, it doesn’t really matter because policies that benefit the rich are ruled out by fiat. But this is insufficient defence because, my explanatory example above notwithstanding, the world is not just divided into ‘rich’ and ‘poor’. There are gradations of poverty, and while many individuals and families move in and out of poverty throughout their lives (in a process termed churning), there are many others who live in situations of ‘chronic poverty’, and it is these who are often missed or under-serviced by international development assistance.

It is my contention that a CBA approach to international development would simply exacerbate this problem, contributing to a widening divide between middle and low income groups and countries on the one hand, and those countries and groups trapped in chronic poverty on the other hand.

It is easiest to see why with a simple, but realistic, example. Let us take the case of public health interventions, say immunization against infectious diseases. CCC analyses of public health often return very sizeable Benefit-Cost Ratios for such policies, and not surprisingly so: few international development experts would dispute that immunization is, in principle, a very cheap and effective way of improving livelihoods.

But the CCC papers on this topic are rather general in their scope; in contrast to the sophisticated modeling for some of the other policies they examine (including climate change), the BCRs on public health are often described by the analysts as ‘indicative’, based on an estimated economic benefit expressed in terms of Disability-Adjusted Life Years (DALYs).

DALYs are a common tool in health economics. They essentially place a monetary value on one year of healthy living for one individual. This often raises the ethical heckles of non-economists[3], but within a economic perspective it is entirely defensible as, in effect, a means of comparison for the greater good. In cross-country comparisons, DALYs are often approximated in relation to GDP. The CCC papers typically take a value of $5000 for a DALY.

This may be fine at the abstract level, but let us take the analysis a stage further. Suppose we have agreed with the 2012 CCC outcome that of a hypothetical budget of $75 billion over four years, we would invest $1 billion per year in child immunization. Where would we invest it? Inevitably, a CBA analysis would lead us to invest in immunization in the relatively wealthy (or less poor) countries and communities in our scope.

This can easily be demonstrated without the need for complex calculations.

Let us initially assume that the costs of administering the immunization is the same wherever we go. The benefits, however, in terms of DALYs will necessarily be higher in the relatively wealthy countries precisely because DALYs are typically derived from the overall economic standing of the country in question. The same logic would apply within countries: the economic benefit in DALYs from immunizing children in wealthy regions of a country would be much greater than immunizing children in poorer regions.

What about the cost side of the ratio? It seems evident that the costs for administering immunizations would be much higher in poorer countries, for a range of reasons. Firstly, poorer countries typically have worse physical infrastructure, so the simple cost of getting to the beneficiaries would be higher. Secondly, poorer countries often lack a sizeable pool of trained health professionals, so there would be increased personnel costs in terms of administering the immunization because of the greater numbers of international health workers required. Thirdly, many of the poorest countries in the world are bedeviled by insecurity and conflict. This is both a major vector for the transmission of infectious diseases and, clearly, would add additional risks and costs to immunization programmes, as has been tragically demonstrated in Pakistan over recent years. Again, the same kind of logic applies within countries.

If costs are higher in poorer contexts and benefits lower, the BCR would clearly push us to invest in immunization in relatively richer areas.

Lomborg and his defenders might argue that their approach was never intended to be applied at this level of implementation, but I have used this as a clear example of a more generic point: CBA analysis of development intervention without inequality aversion will, almost inevitably, be targeting marginal, rather than chronic, poverty and the distributional consequences are clear: we would end up exacerbating inequality in the developing world.

The Intergenerational Sting in the Tail

Thus far, we have seen that because the straightforward CBA methodology is inequality-blind, its application can tend towards exacerbating inequality; the poorest of the poor get left out.

There is one caveat to this analysis, however, which is the issue of the discount rate. I will discuss the discount rate further in a subsequent posting. Suffice it here to say that discount rates are often included in CBA calculations to place a lower value on returns the further they are in the future. They are, in effect, a kind of negative interest rate: $100 next year is valued at less than $100 today.

There are good reasons for including discount rates in CBA analysis, but they are hugely contentious, particularly in environmental economics.

Psychologically, discount rates simply provide a good empirical description of the way we act. Typically, humans are not good at ‘deferred gratification’; we would much rather have $100 today than next year.

But ethically, the situation is much less clear, particularly when we are considering investments or policies that will affect future generations, rather than ourselves. A simple illustration. When I turned 18, my father was employed by the British civil service and, inasmuch, I was entitled to enroll in a very cheap health insurance scheme. I declined: I was young and immortal, and four pints of beer a month seemed a much better investment than insurance in my future health. Now, rather older and with blood pressure somewhere around 40lbs per square inch when untreated, this was clearly a stupid decision. I discounted the future far too heavily. It is my own health, though: it might have been stupid but it is far harder to argue that my decision was unethical.

But imagine my father had been offered a job outside the civil service when I was 17, paying him slightly more than his civil service job, and that I was inclined to take the insurance. For him, it may have been economically advantageous to take the job, but it would have denied me the opportunity to obtain cheap life insurance that would have had long run benefits for me. Here, an ethical argument can be made. Did my father have the ethical right to discount my future in order to secure a small advantage in his own present circumstances?

This, in fact, is at the heart of debates in environmental economics over the appropriate discount rate for calculating the costs of climate change. The UK Stern Review used a very low discount rate (0.1% per year) precisely because it argued that there were no ethical grounds to discount the welfare of future generations against our own; its critics argued that this was both economically and psychologically unrealistic.

I will develop this point further in my next post, with specific reference to the climate issue. My point here is the more generic one that any discount rate can be interpreted as a preference for intergenerational inequality: it systematically values the welfare of future generations at a lower level than our generation.

Conclusion: Is CBA Appropriate for International Development Policy?

I have argued that because simple CBA is, by design, inequality blind (and may even be predisposed towards intergenerational inequality), its application to international development policy would likely have perverse distributional consequences, tending to favour policies that target the marginally poor, rather than chronic poverty.

I want to conclude by making two generic points about the applicability of CBA to international development policy. I want to begin by noting that CBA was developed largely in the context of ranking different investments or policies that affect the same outcome group, whether an individual’s decision on financial investments (where the benefit always accrues to that individual), a firm’s decision on how much to invest in R&D (where the benefit always accrues to the same shareholders), or a government’s decisions on infrastructural investment (where the government is making a decision on behalf of the whole society, and the benefit accrues to the same society as a whole). This is important because it minimizes the distributional consequences. At the extreme, for an individual’s financial decision-making, there are no distributional consequences at all.

But the more people who are affected by the investment decisions of a CBA, the more important it becomes to consider the distributional consequences. Consider, for instance, an area where CBA is often used in public policy: infrastructural investment. If a development agency working with a single remote community had to decide whether to invest in a road linking the community to the nearest economic centre or a railway along the same route, CBA might be a reasonable tool to use, on the basis that the village as a whole is likely to benefit more-or-less equally from either option (although even this is disputable if one looks at gendered impact of infrastructural projects).

But if the same decision was taken in relation to a major investment in national infrastructure in a developing country – road or rail – it would certainly seem remiss to ignore the distributional consequences. Roads might generate a better overall economic return, but also disadvantage the poorer sections of society who are unable to afford cars of their own.

Put simply, the larger scale the level on which the overall calculation is made, the more important it seems to consider the distributional impacts between groups within that population, whether neighbourhoods, provinces, or countries. The global aspirations of the CCC project are its distributional Achilles’ heel. By calculating benefit-cost ratios at the global level, it at best risks massive inadvertent distributional consequences and, I have argued, in practice may indeed favour policies with such negative consequences.

[1] I have often thought it a tragic irony that those politicians busy championing the liberalization of international trade and investment for the sake of free markets are often the same politicians busy opposing the free movement of labour internationally, which is just as fundamental to efficient market allocation in neo-classical economics as the free movement of goods and capital.

[2] Social Welfare Functions are used in economic analysis to rank different social states and are, in effect, a generalization of the basic logic underlying CBA.

[3] As a high schooler attending an economic study class at Oxford where I was first introduced to the concept, I was so appalled I had a standing-up row with an Economic Professor there, whose name I have long forgotten. On the off chance he is reading this and remembers me: my apologies. I’m sold on DALYs now.

5 thoughts on “Consenting to inequality: The distributional consequences of Copenhagen Consensus approach

  1. Prof. Brown – as I understand it, the term discount rate is employed by economists in at least 2 senses. One is the discounting of future benefits you describe which is due to preference for immediate rewards (a bird in the hand is worth two in the bush). The other is the type of discounting used by individuals like Nordhaus, which appears somewhat different and is essentially a bet on economic growth rates. You seem to refer to the former, which is not, as I understand it, the discounting procedure used in most models.

    1. Thanks for your comment, Roger. You’re right that there are these two senses to the discount rate, but most environmental economics combine the two in what is known as the Ramsey equation, which includes an element of ‘pure time preference’. Nordhaus’ model includes such a pure time preference, although this is indeed to some extent determined endogenously. The details were a bit too geeky for a blog post. See this graph which shows the regional variation in cumulative discounting in Nordhaus’ RICE model. The details of discounting in the DICE/RICE model are here.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s