For Good Measure Read online

  But the same device that was used by the colonizers to convince themselves of their mission civilisatrice was turned on them by those struggling for independence. The National Planning Committee of the Indian National Congress, headed by India’s future Prime Minister Jawaharlal Nehru, produced a report in 1936, the findings of which Nehru referred to in his book Discovery of India: “There was lack of food, of clothing, of housing and of every other essential requirement of human existence.” Independence was needed “to ensure an adequate standard of living for the masses, in other words, to get rid of the appalling poverty of the people.” Nehru wrote these words in prison, having been put there by the British authorities for his role in the Quit India movement of 1942. But a generation of Indian analysts had been using official statistics and doing their own surveys to bolster, in effect, the empirical case for independence.

  Given this role of statistics in the Indian independence struggle, and especially the role played by statistics on well-being of the population, it is perhaps not surprising that special attention was paid in India after independence to data on the distribution of consumption and poverty, and on access to public services. The Indian National Sample Survey is the longest-running household survey in developing countries, stretching back to the 1950s. Each release of data is accompanied by lively debate and discussion, with the key statistics providing an assessment of policy outcomes and directions for the future (Deaton and Kozel, 2005).

  The SDG process, and the emphasis given to goals, targets, and indicators in that process, has thrown into sharp relief the generation and use of statistics in developing countries, particularly in Africa. This includes the accessibility and availability of data to researchers and the population at large. In his presentation to the HLEG workshop, Pali Lehohla, former Statistician General of South Africa and head of Statistics South Africa, emphasized that GDP provided a good framework for what it intended to measure, but that it was badly used. In principle, for example, a Social Accounting Matrix (SAM) framework could be used to enrich distributional discussions anchored on GDP. These sentiments were echoed by Rashad Cassim, now Deputy Governor of the South African Reserve Bank and former Head of National Accounts in Statistics South Africa: “Getting GDP measures and its components right is not trivial and there are many challenges that a middle-income country like South Africa, let alone developing countries, face in getting a set of conventional economic indicators right.… Tensions are not only between social and economic data but between high frequency economic data and structural long term economic data. Put differently, should we gear up our statistical infrastructure to track, as accurately as we can, the business cycle or sacrifice this for something else—like putting more resources into estimating the value added of the informal sector, conduct area sampling to better understand small enterprises?”10

  Cassim went on to elaborate upon a number of trade-offs faced in practice by National Statistical Offices, including those involving quality of data, even in the relatively standard area of national accounts, let alone in expanding their remit as seemingly required by the SDG process so as to track and monitor a vast number of indicators.

  These concerns were further underlined by Daniel Masolwa of Tanzania’s National Bureau of Statistics, who emphasized the cost of running regular establishment and household surveys, as well as specialized surveys on informal transactions such as unrecorded cross-border trade.11 Chukwudozie Ezigbalike, Chief of the Data Technology Section of the UN Economic Commission for Africa, estimated that, in 2005, the cost of running a survey of 3,000 households exceeded $500,000.12 However, he also argued for using new technology, the benefits of which included, along with improving and expanding administrative data, the opportunity to initiate an African data revolution in which agricultural and other data could be collected rapidly and at low cost.

  For many low-income countries, these financing needs have driven their statistical offices into the hands of donors who have their own—and often shifting—priorities. The entire statistical system of some low-income countries is geared to the statistics that donors wish to collect. This may be no bad thing if the government is encouraged, for example, to collect gender-disaggregated data on well-being. But, as a general rule, statistics in democracies should be driven by data the government has to collect to satisfy the monitoring and planning needs on behalf of the population.

  The data revolution and the use of new technology emphasized by Ezigbalike is not simply a technical fix to collect relevant data more cheaply. It also highlights the role that civil society and the population at large can play in the statistical discourse, taking it beyond the preserve of technical experts. A key requirement is, of course, the independence of statistical systems from partisan politics. But, beyond this basic governance requirement, we are back again to the question of how many top-level indicators there should be in a national dashboard. It can be argued that too many and too complex a set of indicators would actually be deleterious to an informed debate in society, including the vigorous participation of civil society.

  There can be, and there always will be, specialized interactions on specific sectors, and resources will move back and forth to assess and monitor progress and prospects in these areas to reflect the ebb and flow of political interest. But if a relatively small number of top-level indicators can be agreed upon—for example, the five outlined in the previous section—national discourse can focus on these, and adequate resources can be made available to the National Statistical Office to provide the database for such discussion. The provision of additional resources for data collection is, along with helping develop tools and methodologies, an essential contribution needed from the international community.13

  Measurement at the Global Level

  Although the major significance of SDGs lies at the national level, as laid out in the previous sections, the goals are developed at the global level. The national discourse is, of course, central to the development process, but there are also uniquely global dimensions to key elements of the SDGs, for which we have to take a perspective that goes beyond the national. This triggers the need for establishing internationally agreed statistical standards, for which global and regional organizations such as the ILO, the OECD, or Eurostat (at the European level) have a major role to play.14 We consider three such examples—global poverty, global inequalities, and global climate change.

  SDG 1.1, the first quantitative target of the first SDG is: “By 2030, eradicate extreme poverty for all people everywhere, currently measured as people living on less than $1.25 a day.” This is also the first of the new “twin goals” of the World Bank. The usual operational definition of “eradicate” is to reduce something down to 3%. But note that this is a global goal; in other words, it is a goal for a global measure of poverty. This immediately raises the question of how global poverty is to be measured. Dean Jolliffe’s presentation at the HLEG workshop set out the World Bank’s current thinking and the dilemmas it raises.15 The report of the Atkinson Commission on the Measurement of Global Poverty (Atkinson, 2016) also takes up the issue in more detail.

  Focusing on monetary measures, two questions arise in getting a global count of poverty. First, how are nominal incomes and consumption around the world to be turned into comparable real-income measures? Converting local currency values into a common currency globally by using official exchange rates (say to the US dollar) opens up the question of whether these exchange rates measure true cost-of-living differences between different countries. In general they do not, because market exchange rates reflect only traded commodities and may also reflect financial flows and government interference in market exchange rates. To overcome these problems, the World Bank and others use Purchasing Power Parity (PPP) exchange rates, the use of which is itself steeped in controversy (Deaton, 2010; Ravallion, 2014), one that reignites every time a new set of PPP exchange rates is published. The issue is not whether to use PPPs or not, but the methodology underlying their calculation.
And, of course, PPPs are meant to be conversion factors for some aggregate basket of goods and services, rather than being representatives of what the poor consume.

  The second question arises even if we were to successfully arrive at a true distribution of real income in the world as a whole. Where then do we draw the poverty line? There are various conceptual bases—for example, starting from basic capabilities inspired by Sen and working down from those to a line in the income space (e.g., Reddy and Pogge, 2010). But, as a practical matter, the World Bank has constructed its global poverty line using as inputs various national poverty lines (Ferreira et al., 2015), presuming that these national poverty lines reflect a range of actual normative perspectives. This method led to one poverty line of $1.25 per person per day at 2005 PPP, which is the line stated in SDG 1.1, and another poverty line of $1.90 at 2011 PPPs, as set out in Ferreira et al. (2015). The two lines do not lead to a big difference in the global poverty count (just over 14% of the world’s population).

  Turning now to inequality, SDG 10 is, “Reduce inequality within and among countries,” which actually raises an interesting set of issues that go beyond statistics and measurement, to the conceptual. Take, for example, the case of income inequality. Overall inequality among all individuals in the world can be decomposed into inequality between countries and inequality within countries.16 Inequality between countries is the inequality of the world distribution of income if each person in a country was given that country’s average income—in other words, it is the inequality that would be left if within-country inequality were eliminated in each country. The difference between this inequality and total inequality is then the contribution of within-country inequality to total world inequality.

  What do the numbers look like on this decomposition of global inequality into between-country and within-country components? For the “mean log deviation” measure of inequality (which takes a value of zero when everyone has the same income, rising as incomes become more unequal), Lakner and Milanovic (2015) find that the between-country contribution was 77% in 2008, down from 83% in 1988. The overall global inequality index fell by 10% over this same 20-year period. These trends capture broadly what we know about global inequality trends. Within-country inequalities have been rising in the large countries of Asia (Kanbur and Zhuang, 2012) and, because of their population size, this effect dominated the falling within-country inequality in Latin America. However, low-income countries have grown much faster than high-income countries, with the result that between-country inequality has fallen. The overall combination of these effects has been a fall in global income inequality by this measure.

  These patterns—rising within-country inequality but falling between-country inequality—raise the conceptual question of how, if at all, we weight these components of inequality. The between-country component is numerically much larger—the well-being chances of an individual are predominantly determined by the probability that they are born in this or that country. Thus from this perspective it is important to monitor both between-country inequality and within-country inequality, and SDG 10 recognizes this imperative, although, perhaps surprisingly, no indicators in the “global list” agreed by the UN Statistics Division refer to this between-country element.

  Our third example of global measurement is the most obvious case of why monitoring and assessment at a global level is crucial, and that is climate change and its determinants. Although the short-term consequences of climate change can vary by locality—rising sea levels will devastate small island states, but rising temperatures may be beneficial to some temperate zones—the long-term consequences pose an existential threat to humanity, especially if certain tipping points are reached. These global tipping points are precisely that—global. The extent to which we are approaching them is determined not just by greenhouse gas emission by this or that country, but by global emissions in total. Similarly, the carbon sequestration potential of the planet is determined by total forest cover in the world, and weather systems around the world are linked to each other.

  Thus while action on adaptation and mitigation in response to climate change will necessarily have a national component, the monitoring and assessment is equally necessarily global in nature. Such global monitoring is not as prominent as it should be in the SDG platform. Under SDG 13, it can be glimpsed in the target SDG 13.3, “Improve education, awareness-raising and human and institutional capacity on climate change mitigation, adaptation, impact reduction and early warning,” or perhaps in target 17.19, the last target of the 17th and last SDG, on partnership for sustainable development: “By 2030, build on existing initiatives to develop measurements of progress on sustainable development … However stated, global monitoring of global climate change is surely a key component of the measurement of economic and social progress, and common global measurement instruments and accounting systems such as the System of Environmental-Economic Accounting (SEEA) are crucial in developing common indicators. It is indeed the classic public good, like measuring and monitoring global poverty or global inequality.

  Conclusions

  The Stiglitz, Sen, and Fitoussi (2009) report came after the MDGs, but well before the SDG process got under way. The authors’ insistence on going beyond GDP meshed well with, and greatly contributed to, the broadening of the agenda on the measurement of economic and social progress. But that report did not give as much emphasis as is appropriate to issues that arise in developing countries. The SDG process does indeed have a focus on development, although, of course, it is meant to encompass developed countries as well, and the time is right for us to take stock of where we have come and where we need to go in measuring economic and social progress in developing countries and globally. This chapter attempted such an exercise.

  Three central themes emerge from our discussion, and from the HLEG workshop on which our discussion is based.

  First is the inevitable and enduring tension between the pull to broaden and expand our indicators for assessing and monitoring economic and social progress in development, on the one hand, and on the other, the imperative to keep a relatively small number of indicators at the “top level of the dashboard,” in order to facilitate national discourse and policy-making. The first pull is what explains the expansion of goals from the 8 MDGs to the 17 SDGs and 169 targets. This list is useful as a platform from which to choose and narrow down, but choose we must at the national level.

  Second, National Statistical Offices must be given the governance independence and the financial resources with which to provide the framework for a data-based dialogue on economic and social progress at the national level.

  Third, some aspects of the measurement of progress and development are truly global and beyond the remit of any National Statistical Office. For these exercises, and as a conduit for providing support to National Statistical Offices, the international community needs to commit resources to regional statistical offices and to multilateral agencies for the provision of this global public good.

  Notes

  1. The spectrum of inputs-outputs-outcomes is familiar in the evaluation literature. Of course, any classification of a continuum into three categories is bound to be problematic, but is useful as an analytical device. To use an example from infrastructure, in a road-building project concrete is an input, miles of road built an output, and “travel time saved” an outcome. An example from education would be school expenditure as an input, number of students enrolled as an output, and test scores measuring learning as an outcome.

  2. www.oecd.org/statistics/measuring-economic-social-progress/HLEG%20workshop%20on%20measurement%20of%20well%20being%20and%20development%20in%20Africa%20agenda.pdf.

  3. https://www.slideshare.net/StatsCommunications/hleg-thematic-workshop-on-measurement-of-well-being-and-development-in-africa-lorenzo-fioramonti.

  4. https://www.slideshare.net/StatsCommunications/hleg-thematic-workshop-on-measurement-of-well-being-and-development-in-africa-joseph-stiglitz. At a more technical
level, only if the Lorenz curves of two distributions do not cross can one say that one distribution is unambiguously more or less equal than the other.

  5. https://www.slideshare.net/StatsCommunications/hleg-thematic-workshop-on-measurement-of-well-being-and-development-in-africa-sabina-alkire.

  6. https://www.slideshare.net/StatsCommunications/hleg-thematic-workshop-on-measurement-of-well-being-and-development-in-africa-william-baahboateng.

  7. But that view seems to privilege formal jobs over productive informal work. The problem is that it is hard to distinguish from the available data truly productive informal sector work that increases the size of the national income pie from work that mostly entails getting a large share of some commons rents.

  8. As an intermediate step between the setting of indicators at the global level and at the national level, initiatives have been launched in different regions in the world. A set of more than 100 sustainable development indicators—structured around ten themes—has been defined at the level of the European Union for over a decade. Two-year monitoring reports (http://ec.europa.eu/eurostat/web/sdi/indicators) are compiled and published by the statistical office of the EU (Eurostat). These reports evaluate progress on the long term (since the year 2000) and on the short term (looking at the last 5 years). Eurostat is currently reflecting on how to adapt its monitoring activity on sustainable development to the SDGs. In 2013, the Conference of European Statisticians also agreed on a set of recommendations on measuring sustainable development. Drawing on their experience gained in the European region, the UN Economic Commission for Europe (UNECE), OECD, and Eurostat are now developing a road map on statistics for the SDGs, which will help to structure the statistical reporting in the UNECE region.