The winning entries for the Indigo Prize highlight both the importance of improving economic statistics and some of the challenges of doing so.

The launch of the Indigo Prize by LetterOne is a welcome initiative which recognises both the limitations of conventional economic statistics and the need for a little razzmatazz to bring the issue to the attention of a wider public (encouragingly there have been reports in several UK newspapers).  Entrants were asked to submit a 5,000 word article presenting a design for a new economic measure for global economies and explaining how it should be used to improve the measurement of GDP.  The new measure was to take full account of social and economic factors and the impact of creativity, entrepreneurship and digital skills.

The remit made no mention of environmental or natural resource issues.  A clue as to why may perhaps be found in the contribution on the Indigo website by Mikhail Fridman, a co-founder of LetterOne, in which he rejects the view that there is an imminent problem of depleting resources.  Fortunately, both entrants and judges appear to have taken a broad view of their task.

Given the current state of knowledge, no entry could have been expected to offer a definitive solution to the problem of economic measurement.  Nor is it the sort of problem that lends itself to such a solution: what is best will depend on economic circumstances and (so far as high level presentation is concerned) on the capacity of policy-makers and the public to interpret economic indicators.  Nevertheless, both of the joint winning entries are insightful and thought-provoking (and contain much more than I shall mention here).  That by Haskel et al begins by identifying what it regards as two key strengths of GDP, namely, that it avoids double-counting of intermediate outputs, and that it adds outputs of different goods in an objective way using market prices as weights. Recognising however the limitations of GDP, it proceeds to identify various ways in which GDP could be “fixed” and “extended” to address those limitations while preserving its strengths. It outlines how it might be possible to define and estimate a single measure of well-being that would be a substantial improvement on conventional GDP.

The entry by Coyle and Mitra-Kahn proposes, for the long term, an approach which differs from that of Haskel et al and of conventional GDP in two key ways.  One is that it focuses on assets rather than annual flows.  If such an approach were adopted, people would no doubt consider annual increases or reductions in assets as well as their absolute magnitude. But that itself is quite different from focusing on annual GDP, much of which (eg most consumption) does not contribute to the maintenance and growth of assets.  The second difference is that it rejects the idea that economic debate should focus on a single key number, proposing instead a dashboard presenting separate numbers for broad asset groups.

Of these differences it is the first that is the more radical.  It has never been the case that public debate on economic matters focused solely on GDP.  A fairly conventional ‘dashboard’, though perhaps not presented as such, would include measures of output, employment, inflation, poverty and perhaps a few more variables.  The interesting question here is how far it is sensible or meaningful to aggregate items using weights, with the implication that categories that cannot sensibly be aggregated should be treated as separate indicators within a dashboard.  But this is really two questions, one presentational and one technical.  For presentation to the general public, and as a high-level framework for public debate, a dashboard of six indicators (the number proposed by Coyle and Mitra-Kahn) seems about right, in the sense that it would inform but not overload with detail the lay person with no special interest in economic policy.  The technical question is how many indicators we need to include relevant items that cannot meaningfully be aggregated using weights, and the answer to that question is probably many more than six (within natural capital, for example, consider whether it is meaningful to aggregate petroleum, fresh water, forests and fisheries).

So what do the winning entries say about environmental and resource issues?  As part of ‘fixing’ GDP, Haskel et al propose treating enhancement of the natural environment (eg planting a forest) as investment (an addition to GDP) and degradation (eg destruction of a coral reef) as disinvestment (a deduction).  Presumably they would treat depletion of minerals in a similar manner.  This approach requires valuation of the enhancement and degradation, which is challenging both in principle and in practice.  Taking the coral reef example, an issue of principle is whether the loss of value attributed to destruction of a given quantity of reef should be constant, or depend on how much coral remains elsewhere in a country’s territorial waters, or elsewhere in the world (on the basis that scarcity of a good  increases the marginal value per unit of what is left).  A practical issue is the collection and assessment of data from below the ocean surface.

A more fundamental problem is how the measure obtained by adjusting GDP is to be interpreted. Conventional GDP is, at least, a fairly good measure of economic activity (though it could be improved even in that respect).  The measure that would be obtained after the above adjustments would be such that a year-on-year reduction could reflect either a fall in economic activity or a reduction in net enhancement of the natural environment.  Most people, I imagine, would want to know which, rather than focusing on a measure which combines both.

As part of ‘extending’ GDP, Haskel et al propose that it should count future as well as current consumption (a line of thought which may suggest a concern with resource depletion, although it is introduced in the context of the balance between consumption and investment).  Their reference to Weitzman (1) indicates that what they have in mind here is that suitable adjustments can convert GDP to a concept of net national product (NNP) which measures what Weitzman calls “the stationary equivalent of future consumption”.  In my view this is not a workable or especially useful proposal.  It isn’t workable because the adjustments needed to obtain Weitzman’s NNP depend upon shadow prices of capital goods, and these shadow prices are obtained by solving a long term dynamic optimisation problem requiring assumptions about production technology far into the future (2).  It isn’t very useful because, although a future-oriented NNP concept may sound as if it has something to do with sustainability, “the stationary equivalent of future consumption” is not the same as sustainable consumption.  The relation between the two concepts is quite complex, as explained in this post, and if (implausibly) we had the data needed to calculate the former, we would be able to calculate sustainable consumption directly.

Within the dashboard approach of Coyle and Mitra-Kahn, natural capital is one of the six groups of assets.  Their definition of natural capital as “the renewable resources provided by nature” (p 6) should presumably have read “renewable and non-renewable”, since their chart showing natural capital declining in England Wales (p 12) includes minerals, oil and gas.  They state that measurement of natural capital is currently very incomplete and based on market prices.  The former is correct, but the latter seems to refer to official statistics, and overlooks the considerable academic literature on non-market valuation techniques including the hedonic pricing method (for local environmental quality) and the travel cost method (for recreational sites).

I would agree however that, in the context of a high-level dashboard of six items, it is appropriate that one of the six should relate to the environment including natural resources, and should be a capital measure.  A slight disadvantage of such a measure is that it will tend to ignore the short-term (though often recurrent) effects of non-cumulative pollutants.  Greatly outweighing that is the fact that most of the more important environmental and natural resource issues (eg depletion of minerals, deforestation, over-fishing, greenhouse gas emissions) involve long-term effects that a suitable capital measure should reflect.

Coyle and Mitra-Kahn note that valuing their asset groups, including natural capital, is a major challenge.  I wonder however if it is necessary or appropriate to value each group.  If each item in a dashboard is quantified in monetary terms, that could be taken as an invitation to add the values together and focus on the total rather than the separate items.  Having distinct indicators that cannot be aggregated would seem more consistent with a dashboard approach (think of a car dashboard including speed, fuel level and oil pressure, each in separate units that could not be aggregated, even if bizarrely someone wished to do so).  In the case of natural capital, there is also what might be termed the ‘baseline problem’.  For example, should the UK place a positive value on the fact that it does not suffer from serious earthquakes?  Or should the valuation baseline be defined to include ‘absence of serious earthquakes’ so that the value of natural capital for Japan, say, would include a negative element for earthquake risk?   There is a case therefore for including natural capital in terms of some non-monetary measure based on an assessment of its adequacy to sustain current living standards into the future, although so far as minerals are concerned that leads back to the issue of assumptions about future production technology.  So although the idea may have merit, working out the detail would be just as challenging as trying to value the whole of natural capital.

Notes and References

1. Weitzman M L (1976) On the Welfare Significance of National Product in a Dynamic Economy  The Quarterly Journal of Economics  90(1) pp 156-162
2. Weitzman, as above: the “investment prices” he introduces on p 158, which vary over time, would have to be obtained as shadow prices in the solution of the dynamic optimisation problem set out on p 159. The relevance of future production technology is implicit in Equation (6) on p 158 which features the marginal product of each capital good.