Malaysia’s colonial legacy left both high inequality—with a marked ethnic/racial dimension—and high poverty—with an especially high incidence of poverty among the ethnic majority, the Malay people and other (non-Malay) “Bumiputera.” And ethnic inequality was somewhat neglected as a policy issue in the 12 years or so after the county’s Independence in 1957.
Ethnic riots broke out in Malaysia in 1969, prompting a national effort at affirmative action favoring the Bumiputera. The main policy instrument was the New Economic Policy (NEP). Its twin aims of Malaysia’s were ethnic redistribution and poverty reduction. The Bumiputera were to receive favorable treatment in access to education, housing, public-sector hiring, and corporate share ownership. The NEP lasted 20 years. To varying degrees, successive governments have intervened against ethnic inequality. The policy has been controversial, with both ardent supporters and critics (seemingly correlated, though imperfectly, with ethnicity).
Since 1969, Malaysia’s official poverty measures indicate one of the fastest long-term rates of poverty reduction in the world, due to both economic growth and falling inequality. Indeed the official poverty rate has gone from about 50% to virtually zero. Judged by Malaysia’s poverty line, the only country I know that has had as seen a faster rate of poverty reduction over the longer-term is neighboring Thailand. (Yes, they both beat China in terms of the pace of decline in the poverty rate, though not, of course, in terms of the count of numbers of poor.)
As an aside, I am not convinced that Malaysia has virtually eliminated poverty. The current official poverty line is almost certainly too low by prevailing standards of what “poverty” means in a country such as Malaysia. I take up this issue in another blog post, which can be found here.
Malaysia can also claim more success than most countries in managing inequality. The Gini index of household incomes fell from 0.51 in 1970 to 0.40 in 2016. This was due to progress in reducing ethnic inequality. For developing countries as a whole, average inequality has been roughly unchanged over a similar period.
Did ethnic inequality fall since 1969 and was that a key factor in the country’s success in reducing poverty and in managing inequality? Thankfully, we have data from 18 reasonably comparable, nationally representative, household surveys to draw on in addressing that question, though the historical tabulations are rather limited, and the micro data are not publically available.
A natural question to ask of the available data is how much ethnic inequality fell, and how much that contributed to poverty reduction. My new NBER working paper, “Ethnic Inequality and Poverty in Malaysia since 1969″, tries to answer these questions. (My research for the paper was done while visiting the Ungku Aziz Centre in the Faculty of Economics, University of Malaya in January 2019. Many thanks to the Faculty for their hospitality.)
The paper provides new measures of ethnic inequality spanning 50 years. These show a large reduction in relative ethnic inequality. And the decline in the national Gini index is fully accountable to the country’s progress in reducing the Gini index of between-group inequality (see Figure below). Nonetheless, the ethnic differential in growth rates has not been enough to attenuate the large absolute gaps in mean incomes by ethnicity, given the extent of the initial ethnic inequality. Relative ethnic inequality has fallen, but absolute ethnic inequality has risen.
Importantly, I find no sign in the data of a robust effect of the growth process on inequality within ethnic groups. It is not the case, for example, that the higher growth rate for the Bumiputera in the wake of the NEP was shared unevenly among the Bumiputera when one defines “uneven” in terms of proportionate changes. However, that growth did come with rising absolute inequality. Nor do I find any sign that the policy efforts to reduce ethnic inequality came at a cost to the overall rate of growth, though this is an issue that merits further research.
The paper also finds that Malaysia’s success in reducing income inequality over the last 50 years played a non-negligible role in the country’s success at reducing poverty, in combination with economic growth. Using the official poverty measures, about 10% of the overall rate of poverty reduction is accountable to reduced inequality in average incomes between the main ethnic groups. However, overall economic growth has been the more important driver quantitatively. Changes in the ethnic composition of the population tended to be poverty increasing, though this effect turns out to be small.
While the reduction in ethnic inequality has not been as quantitatively important to poverty reduction as overall growth that does not imply that ethnic redistribution is a blunt tool against poverty in this setting. Nor do the facts that both ethnic inequality and poverty have been reduced substantially imply that ethnic inequality no longer matters to poverty. Indeed, the paper finds quite sizable elasticities of national poverty to inequality-reducing ethnic redistribution. And the elasticities have stayed high—indeed, they have increased—through this period of ethnic redistribution and poverty reduction spanning 50 years.
In short: Malaysia’s long-term effort to reduce ethnic/racial disparities in living standards has played an important role in its ability to manage overall relative inequality. That effort has also helped reduce absolute poverty, although on that score overall economic growth has been more important. However, the potential gains to poor Malaysians from progress toward ethnic equality do not appear to have been exhausted. Going forward, even small reductions, or increases, in ethnic inequality can still matter.
Officially there is almost no poverty left in Malaysia. Yet since coming to visit the country this month, I have read and heard a few people question this claim. So I thought I should look at the issue more closely. (I have been based at the University of Malaya, as the visiting Royal Ungku Aziz Professor; many thanks to this great University for their hospitality. This blog post is based on a presentation I gave at the University.)
The data available from Malaysia’s Department of Statistics (DOSM) suggest that the country has made huge progress against poverty. The official poverty rate has fallen from nearly 50% in 1970 to just 0.4% in 2016 (the latest year available).
However, when one probes further there are reasons to question the claim that Malaysia has virtually eliminated poverty. The official poverty counts are based on a line that has had a fixed real value over time. Currently the line is MYR 920 per household per month, or a little over MYR 7 per person per day, at average household size.
How does that compare with other countries? For that purpose we need to use Purchasing Power Parity (PPP) exchange rates from the International Comparison Program (ICP). The last ICP was for 2011, when the official line was MYR 800. At average household size this is equivalent to MYR 6.41 per person per day. When converted using the PPP for consumption from the 2011 ICP that comes out at almost exactly $4 per day. (Note that the PPP rate is well below the exchange rate given that many goods and services are cheaper in Malaysia than in the US. Unlike market exchange rates, PPP rates are based on the actual prices paid by people.)
Malaysia’s official line is well above the World Bank’s international poverty line of $1.90 a day. But it should be well above that line! The $1.90 line is deliberately low, being anchored to poverty lines found in the poorest countries. If one looks instead at countries with a roughly similar average income as Malaysia, one would expect the poverty line to be about $12 a day—three times the current line in Malaysia. That would indicate a national poverty rate of about 20%, not 0.4%! The following graph shows how Malaysia’s poverty line compares to the poverty lines found in other (non-OECD) countries. Malaysia’s current official line is clearly well below what one would expect for a country with its current average standard of living.
I am not the only one to ask whether the current official line is defensible. For example, in one comment in the Star, in 2018, a Malaysian politician (the Deputy Minister of International Trade and Investment), Dr Ong Kian Ming, wrote that “As wages climbed and we became a middle-income nation, we didn’t increase the standard for what is considered decent living above the poverty line.” Dr Ong has a good point in my view. $4 a day may well have been a sensible line for Malaysia in 1970. But that is no longer clear; real income per household in Malaysia has increased over five fold since 1970! In terms of the World Bank’s income classifications for countries, Malaysia has gone from “low-income” to “upper middle income.”
To see how much all this might matter, I have calculated illustrative alternative lines that rise with average income over time in a seemingly sensible way. The slope as mean income rises is 1:3; so for each MYR 10 increase in the mean, my line rises by MYR 3.33. But I also gave the line a lower bound; I set this at the official “hard-core” poverty line, which I reckon to be about $2.50.
This gives what I call “weakly relative lines.” The “weakly” refers to the fact that the poverty line is not directly proportional to the mean (so that is has an elasticity less than unity). Making the line directly proportional to the mean (“strongly relative”) has the very odd property that if all incomes (including that of the poor) rise by the same percentage then poverty does not change. (They also have the strange property that the poverty rate can fall in a recession. For example, if one uses strongly relative lines for Malaysia I have found that the poverty rate fell during the Global Financial Crisis of 2008-10. That is clearly wrong.) Weakly relative lines are much more sensible.
The following graph compares my new series of poverty measures for Malaysia with those for $4 a day. We see a very similar pattern over time, and a marked longer-term decline in poverty incidence. But there is still a lot of poverty left. Progress, yes, but there is no room for complacency about poverty in Malaysia!
By the way, when I said above, “from the data available” I was hinting at another fact about Malaysia. Unlike many countries today, Malaysia does not have full open access to its household survey data on incomes and expenditures—the data used to measure poverty and inequality. I have had to work from (less than ideal) published tabulations, helped by the World Bank’s PovcalNet data site. I understand that researchers can apply to DOSM for selected variables for sub-samples of the survey data. But the full household-level data are not available. And only the full data can reveal the many dimensions of living standards and their correlates–the joint distribution of the household data.
Access to public data also needs to improve if Malaysia is to maintain high standards of research on poverty, inequality and socio-economic policy going forward.
Academics often try to educate the public at large, by explaining important stuff in an accessible way. This can be socially valuable. But that is not so clear if they oversimplify, or even get it wrong, in their expository zeal. Then bad ideas can get perpetuated, and have excessive influence.
For example, all of the following “expository” claims one hears about randomized control trials (RCTs) are deceptive, if not wrong:
My paper, “Should the randomistas (continue to) rule” gives other examples, and tries to explain, in reasonably transparent terms, why all this matters to sound development policy making.
Please tell me if you think I have oversimplified anything!
The evidence we are seeing of falling relative inequality between countries has left some observers optimistic about the prospects of falling absolute inequality. (Relative inequality is about the ratios of incomes—typically normalized by the mean—while absolute inequality is about the absolute gaps, such as the $ gap in consumption between “rich” and “poor”.) Based on evidence from economic research, some observers anticipate the prospect of a far more equal world ahead. In this vein, Zanny Minton Beddoes wrote a few years ago that: “The gap between the world’s rich and poor will be far narrower in 2050.”
Is that right? We are not talking here about countries (as in the recent paper by Johnson and Papageorgiou). We are talking about people. The answer then depends on what absolute gaps we focus on within the distribution of global incomes and whether relative inequality is declining. If we are talking about the world’s richest 1% (say) and the poorest 1% then it is plain from what we know that we will not see a declining absolute gap in the foreseeable future if recent trends continue, since we are not even seeing falling relative inequality between the two groups of people.
However, there is a range of “middle” incomes for which recent trends do suggest declining absolute inequality over the next few decades. Compare the world’s middle incomes—the 50-60th percentiles, say (just above the global median)—with the income of the 80-90th percentiles, i.e., the group that Branko Milanovic identifies as the rich world’s “middle class.” Branko’s famous “elephant graph” shows that the former group has seen its incomes growing strongly at 3.6% per annum over 1988-2008, while the latter group has seen little growth (0.23% per annum). The ratio of mean incomes in 2008 was 5.7 (based on Lakner and Milanovic, Table 3). Then it can be readily verified that absolute inequality between the two will decline, though it will take 53 years for the two income levels to converge if recent trends continue.
One might question whether a near zero growth rate of the rich world’s middle class is sustainable. Suppose instead that this income group sees a 1% per annum growth rate, with all else unchanged. Then absolute divergence between the rich world’s middle class and the world’s middle will rise for the next 20 years, and only then start to fall, vanishing after about 70 years.
Of course, such calculations should be taken with a grain of salt. Here they only serve to illustrate that absolute inequality is very likely to persist for some time even with falling relative inequality. Indeed, with current trends, the gap will rise between the world’s richest and poorest, and may well also do so between the rich world’s middle classes and the new middle class of the developing world.
Bourguignon, Francois, 2016, The Globalization of Inequality, Princeton: Princeton University Press.
Lakner, Christoph, and Branko Milanovic, 2016, “Global Income Distribution: From the Fall of the Berlin Wall to the Great Recession,” World Bank Economic Review 30(2): 203-232.
Milanovic, Branko, 2016, Global Inequality: A New Approach for the Age of Globalization. Cambridge, Mass: Harvard University Press.
Johnson, Paul, and Chris Papageorgiou, 2018, “What Remains of Cross-County Convergence?,” Journal of Economic Literature, forthcoming.
Ravallion, Martin, 2010, “The Developing World’s Bulging (but Vulnerable) Middle Class,” World Development 38(4): 445-454.
______________,, 2018, “Inequality and Globalization: A Review Essay,” Journal of Economic Literature 56(2): 1-23.
The prevailing approach to measuring global inequality pools all household incomes—as measured in sample surveys—across the world. One then measures inequality in this global distribution the same way one measures inequality within one country. A good example of this approach is found in Lakner and Milanovic (2016).
These measures of global inequality attribute no economic advantage to living in a richer country beyond what is already reflected in the household incomes measured from household surveys. This restriction is hard to defend on either theoretical or empirical grounds.
One way that national income may matter stems from the longstanding idea of relative deprivation. This postulates negative economic gains to co-residents. Then we can rationalize a nationalistic view that “global inequality” is just the average national inequality across countries. This emerges as the limiting case in which it is relative income within the country of residence that matters.
However, that is hardly plausible. One can point to reasonable arguments for positive external effects of living in a richer country at given own income. Examples of how this can happen include the likely positive correlation between national income and factors conducive to a higher long-run personal income, better public services, and greater security. None of these gains are likely to be properly reflected in current incomes as measured in surveys.
The implication is clear: the (large) differences in average incomes found between rich and poor countries create an extra (horizontal) inequality between their residents, not reflected in their current incomes, as observed from surveys. This is a source of downward bias in prevailing measures of global inequality. Yet the likelihood that living in a richer country delivers gains to economic welfare that are not reflected in survey-based incomes has been entirely ignored by past measures.
I explore this issue in a new paper, “Global inequality when national income matters,” which proposes a new way of measuring global inequality that allows national income to matter at given “own-income” as measured in surveys (Ravallion, 2018).
I find that this issue is highly salient to the quantitative measures obtained for global inequality. If one defines economic welfare in terms of relative income alone then one sees far less inequality in the world than if one puts a sizable value on the external benefits of living in a richer country. Using what can be considered the ideal inequality measure for this purpose, namely the Mean Log Deviation (MLD), the paper finds that relative deprivation theory implies that global interpersonal inequality is far lower than prevailing measures suggest since it is then entirely within countries.
This changes dramatically when one allows a positive value of national income (at given own-income), such as when living in a richer country brings benefits in terms of access to non-market goods and services, and better opportunities for private support in times of need. From what we know based on past global studies using micro data, the national income effect could well be 50% or more of the own-income effect on subjective wellbeing. Then global inequality is far higher than prevailing measures suggest, and far higher than found in even the most unequal country.
Indeed, the differences in levels of inequality due to even rather modest differences in how one values national mean income tend to swamp the differences seen over time in standard measures, or the differences we see between countries, and are also large relative to the impact of even a substantial underestimation of the incomes of the rich.
For example, suppose that incomes of all the richest 1% in the world are actually double the numbers in Lakner and Milanovic (2016) for 2008. This would add about 0.1 to the global MLD. I find that this is about the same as adding a modest 10% of log national-mean income to log own income to allow for the gains from living in an economically-better off country.
So the level of global inequality could well be very much higher than current measures suggest, once one allows for the likely external gains from living in a richer country. Nonetheless, I still find that the stylized fact that global inequality has been falling since around 1990 is robust to all except a seemingly high negative weight on national income, such as due to relative deprivation. The finding of falling between-country inequality is robust whatever value one attaches to national income in assessing individual economic welfare.
There is almost certainly more inequality in the world than we think, but it is likely to be falling.
Lakner, Christoph, and Branko Milanovic, 2016, “Global Income Distribution: From the Fall of the Berlin Wall to the Great Recession,” World Bank Economic Review 30(2): 203-232.
Ravallion, Martin, 2018, “Global Inequality when Unequal Countries Create Unequal People,” European Economic Review, forthcoming.
With today’s release of an updated version of the World Bank’s PovcalNet data site, it is a good time to take stock of progress toward Sustainable Development Goal 1.1 (SDG1.1). SDG1.1 aims to “eradicate extreme poverty” by 2030, as defined by the World Bank’s international poverty line of $1.90 a day line. ($1.90 is an update to 2011 prices of the $1.25 a day line in 2005 prices, which is explained in this paper, “Dollar-a-Day Revisited”.
What does the new data suggest about progress toward SDG1.1?
The following Figure plots the series from PovcalNet for the overall poverty rate for the developing world over the period 1981-2015. This looks very encouraging. Of course, there will be less progress at higher lines than the (deliberately frugal) $1.90 line. But the rate of decline we are seeing for extreme poverty is clearly good news. The poverty rate is falling by a little over 1% point per year, on average, over the period as a whole. (The regression coefficient on time is -0.012, with a standard error of 0.0004, though this only reflects the time series variation.)
A simple linear extrapolation of this series indicates that extreme poverty will be eliminated by 2025. (A crude calculation of the standard error, again based solely on the time series, gives a value of 0.9 years, implying a 95% confidence interval is 2023-2027.)
However, if we dig a bit deeper there is another perspective on the data that is far less encouraging. Let us ask instead how long it will take for the “floor” of the distribution of consumption in the developing world to reach $1.90 a day. This is another way of defining “eradicating extreme poverty,” namely that the poorest person has $1.90 a day or more. The floor is not easily estimated, but I have outlined an approach in another paper, “Are the World’s Poorest Being Left Behind?.” The following Figure gives my estimates of the floor over the same period. This is a weighted mean of the consumption levels of the poor, with highest weight on the poorest, and linearly declining weight as consumption rises, until $1.90 is reached.
This perspective is not nearly as encouraging. The floor in 2015 is almost exactly $1.00 a day, up from $0.87 a day in 1981. There is a (statistically significant) positive slope over time to how the floor has evolved, but the slope is very small. At this rate of progress, extreme poverty will not be eliminated until 2278 (with a 95% confidence interval of 2169 to 2387). From this perspective we are way off target.
The reason is clear: the developing world is not making enough progress in reaching the poorest—well below the $1.90 line. Numbers of poor (by this frugal standard) are falling, which is undeniably good news. But the progress is not being shared enough by the developing world’s poorest. They are not exactly being “left behind” but pretty close to it.
The prospect of a much slower progress in lifting the last few % out of poverty should not be too surprising. In 2012, the President of the World Bank, Jim Yong Kim, asked me to propose a goal for poverty reduction (at a time when I worked for the World Bank, as Director of its research department). An elaboration of the note I wrote for him was subsequently published as, “How Long will it Take to Lift One Billion People out of Poverty?” That paper outlined both pessimistic and optimistic paths to lifting one billion people out of extreme poverty, and the optimistic path evolved into SDG1.1. But an important difference crept in along the way. My “optimistic path” would still leave a few % of the world’s population in poverty by 2025-30. My calculations suggested that the last few % would be a much harder job. Even my optimistic path for economic growth and income inequality would not “eradicate extreme poverty.”
The good news from the latest PovcalNet is that we are making progress in reducing numbers of poor. But a closer look at the new data suggests that “business as usual” will not achieve SDG1.1, which might well take another 200 years!
Greater effort at reaching the poorest will almost certainly be needed.
One sometimes hears claims that rising top incomes are helping to lift up the poorest. This is a version of what is often called “trickle down.” The idea has been much debated.
Is there any evidence that this could be happening in America, where the marked rise in top incomes is (understandably) getting a lot of attention, and raising concerns about the country’s future. These concerns remain even if there are trickle-down benefits to the poorest. But let us focus here on whether there is any sign that the trickle down idea might be right.
Some terminology first. “Top incomes” refers here to the incomes of the richest 1% say. The “floor” refers to the lower bound of the distribution of income—a level of income below which very few people are likely to be found for any reasonable length of time. (There are many methodological issues in measuring these two extremes of the income distribution; while I will mention some, this post is not the place to go into those issues in much detail.)
To help address the question of whether there is any sign that trickle down has been at work, this post draws on my new research paper written with Dean Jolliffe and Juan Margitic, “Social Protection and Economic Development: Are the Poorest Being Lifted-Up or Left-Behind” (coming out on June 4 as NBER WP 24665). The main data source is the microdata from the annual Current Population Surveys done by the US Census Bureau for the Bureau of Labor Statistics, spanning 1988-2016.
Divergence between the richest and the poorest: Figure 1 plots the per capita pre-tax incomes of various quantiles, including the top 1%—the income level above which one finds 1% of America’s population. Call this q(99). The graph also gives the quantiles for 95% and 90% as well as the mean and median.
And Figure 1 gives our estimate of the floor. This is a weighted mean of the incomes of those officially designated as poor in the US, with highest weight on the poorest, and declining weight as incomes rise. We see a marked absolute divergence—roughly speaking, a rising absolute gap between richest and poorest.
Figure 1: Absolute real incomes per capita
Two methodological points are notable here. First, in measuring the floor for the purpose of this graph, incomes are calculated the same way that the US Census Bureau uses for its official poverty measures. This includes money income before taxes from several sources (such as wages, salary, net-income from self-employment, social security payments, pensions, interest, dividends, alimony, other forms of periodic monetary income) but excludes capital gains and non-cash benefits such as fringe benefits or noncash government social assistance programs. Importantly, its excludes a major income source for poor people, namely the Supplemental Nutrition Assistance Program (SNAP), though more often called “food stamps”. This helps recipients cover their food spending and is explicitly targeted to the poor (living below 130% of the official poverty line). SNAP now covers about 14% of the US population. I will return to SNAP shortly.
Second, there is likely to be a downward bias in survey-based incomes of the rich, and correcting for this is likely to make the top few quantiles in the graph higher, and probably steeper. The estimates combining CPS with income tax and national accounts data in the Worldwide Inequality Database (based at the Paris School of Economics) indicate a higher growth rate for incomes of the top 1% than indicated by only using the CPS.
Figure 2: Log real incomes per capita
Since the floor is so low—we estimate that it is around $6 per person per day—relative to the top incomes it is hard to see what is going on in the graph above. So Figure 2 uses logs instead. Now the divergence between the top and the bottom is easy to see. Not only is there no sign that rising top incomes have lifted the floor, the floor is actually falling over time. The growth rate of the floor over the whole period is -1.3% per annum. By contrast the growth rate of q(99) is 1.6% per annum. For a family of four people, the floor fell from about 9% of q(99) in 1988 to 4% in 2016. We see relative divergence between the richest and the poorest.
Food stamps: There is another potential channel for “trickle down” namely via public spending on antipoverty programs. The new paper with Jolliffe and Margitic looks more closely at the role played by SNAP. We find that SNAP has benefited the poorest. Figure 3 gives our estimates of the floor before and after SNAP. (This is expressed as a proportion of the official poverty line, which was $15.15 per person per day in 2010, for a family of two adults and two children.)
Figure 3: The floor before and after food stamps
We find that SNAP helped raise the floor, and to stabilize it after 2000, though with signs of a decline since 2012. SNAP helped prevent the floor falling even further, especially in the wake of the financial crisis starting in 2008. Without the “SNAP stimulus” in the wake of that crisis we would have seen the floor dropping even more. Less encouragingly, we also find that the efficiency of SNAP in reaching the poorest (the gain in the floor per $ spent on the program) has declined over time. However, it is clear from our calculations that SNAP has helped America’s poorest.
So is this a form of trickle down that has worked? That is not so clear. SNAP spending per recipient has grown with rising top incomes, though whether the latter caused the former is an open question, with arguments both ways. More tellingly, SNAP participation rates have fluctuated over time since around 1990, buffeted by politics and reform efforts, and this has led to similarly fluctuating levels of overall spending per capita of the US population. It seems likely that politics has played a more important role than a trickle down process through social spending.
It has long been known that social outcomes in terms of poverty and human development vary across countries at given mean income, such as measured by GDP per capita. An early demonstration of this point is found in Amartya Sen’s 1981 paper “Public Action and the Quality of Life in Developing Countries” (Oxford Bulletin of Economics and Statistics 43(4): 287-319). Sen pointed to countries such as Sri Lanka that had used social policies to attain the social indicators (such as life expectancy) that are more typical of high-income countries. Costa Rica is another example.
There are hazards in inferring that social policies explain why some countries have better social outcomes at given mean income. This relates to a general class of problems in inferring “social efficiency” from regressions of outcomes on mean income, as explained in my paper “On Measuring Aggregate Social Efficiency” (Economic Development and Cultural Change 53:.273-92). However, let me put those problems to one side here, and assume that good social policies help explain cases such as Sri Lanka and Costa Rica.
Does this imply that rich countries can maintain their better social outcomes (in terms of health, education, absolute poverty) at a lower mean income? That is the claim made by the Degrowth Movement, which has argued that the way to improve environmental outcomes (and, in particular, to reach targets for carbon emissions) is to contract aggregate economic activity. Clearly, this would meet with near universal opposition if it was thought that social outcomes would worsen as a result. That is where the experience of countries such as Sri Lanka and Costa Rica becomes so important in the eyes of the Degrowth Movement, which points to those countries to support their claim that degrowth need not have a social cost.
The problem in this argument is that better social outcomes are not only attributable to better social policies. Higher average incomes have also played a role, both directly (through poor people’s greater command over commodities that matter to those outcomes) and indirectly (by creating the resource availability needed to finance better social policies). (Sudhir Anand and I elaborate this argument in our paper, “Human Development in Poor Countries,” Journal of Economic Perspectives 7(Winter 1993): 133-150.)
There lies the Degrowth Fallacy: The fact that some countries have better social outcomes at a given level of mean income does not imply that richer countries can attain the same social outcomes at lower mean income.
In an example of the Degrowth Fallacy, in a radio interview (posted today) Jason Hickel (introduced as an economist and anthropologist at the London School of Economics) points to Costa Rica to support his claim that rich countries can maintain their social outcomes at lower mean income. Yes, Costa Rica has had good social and environmental policies over many decades, and other countries can learn from that experience. However, the country did this in combination with economic growth. Indeed, mean income has tripled in real terms since 1960, and the average growth rate has been above average for Latin America (World Bank). By combining social policies with policies that directly supported economic growth, Costa Rica was able to attain over time good social outcomes for a country at its level of mean income. Costa Rica is definitely not an example of how good social outcomes are possible without economic growth.
The degrowth advocates have also disputed claims that social outcomes tend to improve with growth. For example, Item 3 of the Degrowth Manifesto says that “Global economic growth has not succeeded in reducing poverty.” I will grant that there are cases in which economic growth has by-passed poor people; there is no reason why growth will inevitably come with lower poverty or better outcomes in terms of human development. However, the bulk of country experiences have indicated that better social outcomes (including poverty reduction) generally come side-by-side with economic growth. In marked contrast to the Degrowth Manifesto, economic growth has succeeded in reducing poverty. Also, lower poverty has helped promote economic growth. (On all this see Chapter 8 of the Economics of Poverty.)
Of course, this does not mean that growth on its own, without good social policies, will improve social outcomes. There are also examples of countries that have squandered much of the social welfare gain from economic growth from this perspective—the gains have been largely captured by those who are already well-off, with modest benefit to the broader society. The US today is an example.
As Hickel also argues, many social outcomes were better in the US in the 1970s, when mean income was roughly half what it is today. But one must seriously doubt that halving today’s average income in the US will restore the social outcomes of 50 years ago.
(This is the original English version of my oped in Le Monde, 12/15/2017.)
Probably more than any time in history, we have a broad political consensus today on the desirability of eliminating the worst forms of poverty in the world. We see this consensus in many places, including the UN’s Sustainable Development Goals. There is no such consensus against inequality.
From a global perspective, we are making progress against poverty. Higher rates of economic growth in the developing world during the current period of globalization (though not just due to globalization) have come with success in reducing the numbers of very poor people. In the world as a whole, the number of people living below the frugal poverty lines found in low-income countries has fallen, from around 1,800 million in 1990 to just under 800 million in 2013 living under the World Bank’s international line of $1.90 a day (2011 prices). China has played a major role in this success. But since 2000 we are seeing progress in reducing the numbers of very poor people in all regions of the developing world. We are also seeing progress in human development, with rising literacy rates and falling infant mortality rates.
Growth in the developing world has also helped attenuate global inequality. This is typically measured by pooling all household incomes in the world (ignoring country of residence and) and measuring relative inequality—a summary statistic of the proportionate differences in incomes—in this anonymous global distribution. By this measure, inequality has been falling over the last 30 years or so.
But when we look more closely at the data, there are some troubling features, pointing to serious policy challenges going forward. Relative inequality within countries is trending upwards on average, and markedly so in many countries, now famously so in the US. And absolute inequality—the absolute income and wealth gaps between rich and poor—is probably rising within all growing economies.
We have also learnt that growth processes in all countries have had both winners and losers. Some people have been pushed into poverty, while others escape it. This creates social tensions despite overall progress. We are also seeing rising numbers of relatively poor people, meaning that they are poor by the standards of what “poverty” means in their own countries, even though they live above $1.90 a day.
Nor are we seeing much progress in raising the level of living of the poorest—the world’s consumption floor. Despite higher growth rates in most developing countries since 2000, the consumption floor is still somewhere near the biological minimum in many countries, at around $1 per day. The developing world has been far more successful in reducing the numbers of people living near the floor than in raising the floor.
Before we can expect to see public effort to address these distributional concerns, there needs to be a political consensus that effort is called for. The lack of consensus for inequality is stalling public action. Granted we will never want to eliminate inequality, as that would surely leave too little incentive for innovation and growth. The consensus we need is about reducing high inequality. To some eyes, such inequality is morally objectionable, but not to other eyes.
Broader public support for redistributive effort may well emerge from the body of research that has indicated that high inequality undermines the scope for sustained economic growth, which makes it harder to maintain progress in reducing poverty. Also, the higher the initial level of inequality, the less poor people tend to share in the gains from economic growth. In short, high-inequality countries tend to see less growth and less of it helps those who need it the most. Inequality is thus a policy concern even if we think poverty reduction is the far more important goal.
Not all inequalities matter equally. There is scope for greater consensus about reducing those aspects of high inequality that matter most to growth and poverty reduction. These are the specific inequalities that constrain economic opportunities for poor and middle-class people. Inadequacies in schooling, health care and social protection are socially and economically disabling—creating a loss of freedom to fulfill people’s aspirations in life. So too do the many imperfections in markets for credit and land—creating inequalities in access to production inputs and technologies. All such inequalities limit employment opportunities, impede physical and social mobility, weaken democracy, and render people vulnerable to subordination to the will of others and exploitation in all domains of life. This represents lost economic opportunities, such that poor people have little hope of sharing in the new sources of wealth and in helping to generate that wealth. We then see an inequitable growth process going forward, perpetuating poverty across generations.
This can be the basis for developing a broad consensus against high inequality, as we now have for poverty.
A new paper by Robert Allen, “Absolute Poverty: When Necessity Displaces Desire” (American Economic Review, December 2017), has proposed and implemented a method of measuring global poverty. Allen advocates this as an alternative to the World Bank’s longstanding method using Purchasing Power Parity (PPP) rates across countries in trying to assure that the international poverty line has constant purchasing power globally.
At the core of Allen’s method is the use of Linear Programming (LP) to set a least-cost diet for attaining stipulated nutrients, following George Stigler (“The Cost of Subsistence,” Journal of Farm Economics, 1945). Allen estimates country-specific least-cost diets anchored to globally-constant nutritional requirements with an allowance for spending on his stipulated bundle of non-food goods. Allen does not present his method as a complement to the Bank’s, but as superior. He claims that his method yields substantially high global poverty counts than the World Bank’s $1.90 a day line.
My comment on Allen’s paper, “An interesting step backwards,” argues that his proposed method is the resurrection of a method famously rejected long ago, including by Stigler, because it produces poverty lines of little social relevance. From the point of view of the history of thought on poverty, it is interesting to see what this old method delivers with new data for the developing world.
On a close inspection of his results and on doing some extra calculations, I find that his claim that the global absolute poverty count is much higher than currently thought, based on the Bank’s $1.90 line, is no longer true when one uses nutritional requirements that would seem to accord more closely with practice in low-income countries. This gives a much lower poverty count than the Bank’s $1.90 line. And the trend over time is virtually identical. I also argue that there is an urban bias in the prices Allen uses for valuation, which raises his counts relative to the national lines in low-income countries that have been used to determine the $1.90 line.
The comment argues that Stigler was right, and the method Allen proposes should still be rejected, based on Allen’s own findings. PPP’s remain essential for global poverty measurement.
Partly in response to concerns about chronic undernutrition, there is an expanding effort at social protection in developing countries and this effort is typically focused on transfers targeted to poor families (as discussed in Chapter 10 of The Economics of Poverty). These policies regularly assume that targeting poor households suffices in reaching poor individuals. Is that right?
A new paper, “Are poor individuals mainly found in poor households?“, with Cait Brown and Dominique van de Walle, studies the effectiveness of these household targeting efforts when one is trying to reach poor individuals, as identified by the nutritional status of women and children. Our answers will surprise many readers.
As expected, there is a “wealth gradient” in that poorer families are more likely to include undernourished women and children. But other factors are at work, including intra-household inequality and shared risks from the health environment. In our comprehensive assessment for 30 countries in sub-Saharan Africa we find that undernourished women and children are spread quite widely across the household wealth and consumption distributions.
Strikingly we find that roughly three-quarters of underweight women and undernourished children are not found in the poorest 20% of households, and around half are not found in the poorest 40%. Countries with higher undernutrition tend to have higher shares of undernourished individuals in non-poor households.
It is clear from our results that to have any hope of reaching undernourished women and children in Africa, policy interventions will either require much more individualized intra-household information or they will need to be nearly-universal in coverage.
One often hears that high incomes are simply the reward for greater effort, and poverty reflects laziness. Here, for example, is what Ben Stein has said (In The American Spectator, April 4. 2014): “There is an immense amount of income inequality here [in the US] and everywhere. I am not sure why that is a bad thing. Some people will just be better students, harder working, more clever, more ruthless than other people”. Stein goes on to claim that long-term poverty reflects “poor work habits”.
Is there any truth to the claim that there is less inequality and poverty than we think, once one allows for the differing degrees of work effort that people make? My paper “Inequality and poverty when effort matters,” argues that it is far from obvious that allowing for heterogeneity in effort implies less inequality or poverty.
If one takes seriously the idea that effort comes at a cost to welfare then it is clear that prevailing approaches are not using a valid monetary measure of welfare. While this much is obvious enough, the heterogeneity in effort must also be brought into the picture. Then the distributional outcome is far from obvious. It may be granted that average effort rises with income, but there is also a variance in effort at given income. The implications for measuring inequality and poverty stem from both the vertical differences (in how mean effort varies with income) and the horizontal differences (in how effort varies at given income).
It is unclear on a priori grounds what effect adjusting for effort in a welfare-consistent way will have on standard measures. There are both empirical and conceptual issues. The implications for measurement of taking effort seriously depend crucially on the behavioral responses to unequal opportunities, and not all of those responses are readily observable. Measures with a clearer welfare-economic interpretation call for data on efforts, for which existing surveys are limited to a subset of the dimensions of effort.
The paper provides illustrative calculations for American working adults without disabilities. A positive income gradient in labor supply is evident in the data. This gradient accounts for very little of the income gap between the poorest third (say) and the overall mean. The fact that poorer workers work less appears to contribute rather little to overall inequality in observed incomes. However, the considerable heterogeneity in effort at given incomes imparts a large horizontal element to inequality measures that adjust for effort consistently with behavior.
On calculating distributions of welfare-consistent equivalent incomes to allow for this heterogeneity, I find higher measures of inequality than for observed (unadjusted) incomes. Contrary to the common view, the prevailing practice of ignoring differences in effort understates inequality.
It can be acknowledged, however, that some of the apparent heterogeneity in leisure preferences seen in the data is deceptive given likely rationing and measurement errors. When I use predicted leisure shares based on covariates I find a modest drop in the measured levels of inequality on adjusting for effort. Adjusting for effort does not appear to make much difference in the structure of inequality, as indicated by regressions using a set of circumstances related to gender, age, race and place of birth.
The implications for measures of poverty depend crucially on whether one sets the poverty line consistently with the welfare metric. If one does not do so, then poverty rates are lower using equivalent incomes although this essentially vanishes when one smooths the data. However, these comparisons are arguably deceptive since one is not setting the poverty line consistently with how one is assessing welfare. To correct for this, one needs to include a normative allowance for leisure as a basic need in setting the poverty line.
On introducing even a modest allowance for leisure as a basic need (valued at a low wage rate) I finds higher poverty rates when one adjusts for effort. If half the average amount of leisure taken by American adults is deemed to be a basic need then the poverty rate based on equivalent incomes, adjusted for effort, is nearly twice as high as that based on observed incomes.
In short: allowing for effort in a way that is broadly consistent with behavior does not attenuate the disparities suggested by standard data sources on income inequality or poverty. Indeed, one can reasonably argue the opposite: there is even more inequality and poverty than current measures suggest.
These two graphs are depicting the same information is different ways. Both relate to the distribution of consumption per person in the developing world. Both are from the same data source, using the methods described in my paper with Shaohua Chen, “Developing world is poorer than we thought” (though I have updated the numbers to 2011).
Graph (a) in the left gives the poverty incidence curves (PICs) for 1981 and 2011. Each point on the vertical axis gives the proportion of the population of the developing world living below the point on the horizontal axis. The lower panel in (a) gives the vertical difference between the two (2011-1981). We see a decline in the poverty rate for all possible poverty lines. This also implies that virtually all sensible poverty measures will show a reduction in poverty over this 30 year period. (These claims use well known results from the litertature; see Chapter 5 of EOP).
Graph (b) gives instead the horizontal difference at each percentile. This is the absolute growth incidence curve (GIC). (The concept of the GIC comes from another paper with Shaohua Chen, “Measuring pro-poor growth“). Consistent with graph (a) we see a gain at all levels. But what is also striking in (b) is how much smaller the absolute gain is at the bottom (for the poorest percentile on the left). This is a case of rising absolute inequality.
I have come to realize recently that many people who have learnt that absolute poverty measures are falling in the developing world, and are naturally pleased to know that, are surprised at graph (b), and some are even shocked. But they are just two ways of representing the same information.
Yes, poverty is falling (graph a). But it is coming with rising absolute inequality (graph b).
There is much talk these days about the idea of a basic income. This is an untargeted transfer, set at the same level for all recipients within a domain. It is variously called ‘citizenship income’, ‘universal basic income’ or (my favourite) ‘basic income guarantee’ (BIG). The debate for and against the BIG idea spans the globe. A BIG is often contrasted with a set of targeted transfers that would fill all the poverty gaps.
Without necessarily advocating a BIG, this article points out that some of the arguments we hear against the idea are straw men.
Critics of the BIG idea argue that it costs far too much to be considered seriously. Some BIG proposals have had scary price tags, but that misses the point. Most countries (including many poor countries) already spend public money on poverty reduction. If the same resources would be better spent fighting poverty using a BIG, then that’s what should be done. We can ask this question for any given level of the basic income.
India, for example, has subsidies that benefit the non-poor more than the poor. Cutting these and replacing them with some form of BIG would almost certainly have greater impact on poverty. More remarkably, a BIG might well be more cost-effective against poverty than India’s main programmes targeted to the poor, namely the Public Distribution System (PDS) (for subsidised food) and the Mahatma Gandhi National Rural Employment Guarantee Act (MNREGA), a workfare programme that aims to be self-targeted to poor people. In research with colleagues I have found that a small basic income in Bihar, one of India’s poorest states, would have more impact on local poverty than the labour earnings from MNREGA (Murgai et al. 2016).
A key element of the case for a BIG is how it would be financed. Progressive income taxes are an option in rich countries, and then the BIG is formally equivalent to Friedman’s (1962) negative income tax. When financed by reduced expenditures on other public programmes, the benefits and costs of those programmes should be considered. That could make a BIG proposal look worse if it was introduced at the expense of (say) better quality schooling, or healthcare for poor people. Fine, let’s look at that carefully. But we should not write off a BIG at the outset as too expensive. A BIG should be among the options to be considered by any developing country in a package of anti-poverty policies (Ravallion 2016).
The favourite calculation here is simple. Just measure each poor person’s poverty gap – the distances below the poverty line for all the poor – and make handouts accordingly. Voila, no more poverty! A budget to cover the aggregate poverty gap will almost always be much smaller than the cost of a BIG.
This calculation is deceptive, for three main reasons. The first is about incentives. Perfect targeting like this generates 100% marginal tax rates on poor people because if a poor family’s income increased by US$1, its benefit level would fall by US$1. We would have created a policy-induced poverty trap. If implemented, once everyone realised this was the case, the cost would undoubtedly rise until it was much higher than the initial aggregate poverty gap.
The second reason concerns the information constraints on targeting. These can be severe, especially in poor countries with weak administrative capacity. Policymakers in practice have access to a limited subset of the information needed to reliably establish how poor someone was. The household data that could be reliably used for this purpose are limited, and we know even less about living conditions at the individual level. We are a long way from perfect targeting in practice.
Third, when the budget is free to vary, finer targeting may well undermine public support (notably from the middle class) for anti-poverty policies (Sen 1995). The poor end up with a larger share of a smaller pie, with ambiguous gains.
A simple anti-poverty policy assigns transfers based on observed categories, such as location, household size, or landholding in rural areas. More sophisticated versions use a proxy means test in which a statistical model is calibrated using a limited set of readily observed household characteristics in a sample survey. Data on these characteristics are used to predict who is poor in the population as a whole.
These methods have often disappointed in practice. Even with a fixed budget sufficient to eliminate poverty with perfect information, optimally designed transfers as a function of the information that is available do not come close to eliminating poverty. This has been shown in recent research for countries in Africa (Brown et al. 2016). Even with the aggregate poverty gap as the budget, three-quarters of the poverty would remain.
There are further questions here: what does it mean to target ‘well enough’, and well enough for whom? Experience and research point out the horizontal inequities of targeted social policies in practice. At the community level, these policies may appear opaque or almost random. People can see that equally poor families are being treated differently by these targeting tools. Some get help but others do not. The obvious unfairness of this situation can generate social conflict and undermine well-intentioned social policies. There can be gains from tapping into community-based information, but there are risks there too (Mansuri and Rao 2012).
Whether the information available to policymakers is reliable and sufficient for the task of targeting is an open question. It should be examined carefully in each setting, taking account of the social tensions that can be generated by bad targeting. But we should also not presume that there are large gains for poor people from exploiting the available information in practice.
This one is surprising, because a BIG would probably be the most non-distortionary policy imaginable. Nobody would be able to do anything to change how much they got. Granted, there would probably be a positive income effect on demand for leisure. As with all transfers, however, one must also consider impacts on other relevant constraints facing poor families, including uninsured risk and credit constraints. Transfers can help relax these constraints on employment. There is evidence that South Africa’s ‘older person’s grant’ did this for recipient families (Ardington et al. 2009).
Incentive effects should not be ignored. Some targeted policies create high marginal tax rates for poor people. This is a bad idea. But incentive problems are often exaggerated, and may not be serious in practice (Banerjee et al. 2017).
Undeniably, there are tradeoffs. When thinking about a BIG, however, we could define ‘income’ more broadly than just cash income. A full income concept could include imputed values for services in kind, such as publicly provided health insurance and schooling. A BIG discussion might emphasise these ‘non-cash’ dimensions of welfare.
The composition of the basic (full) income package would then be a policy choice. That choice would be important, and it would never be easy. There are concerns about paternalism that overrides the preferences of poor people. But the key point is that non-cash benefits should be factored into any characterisation of what, exactly, is guaranteed. So this is not a valid case against a BIG.
I doubt if there will ever be a truly universal BIG in which everyone gets a fixed lump-sum. More likely we will see some form of a ‘State-contingent’ basic income, meaning that the transfer would be uniform within some category of households or type of people. It might be defined by where people live, their age, or employment status. This might not achieve the vision of a full citizenship income that advocates have wanted, but it can also help reduce the concerns about fine targeting.
But, wait a minute – a State-contingent basic income is nothing more than a targeted policy based on categorical poverty indicators, and so we have come full circle. In practice, the lines between basic income and targeted assistance are blurred. The core policy issues are how many states or groups we identify for different transfers, with uniform treatment within the group, and how we handle transitions between states, noting the incentive problems this creates.
Whether we see universal BIGs in future or not, the current policy debates will hopefully lead us to be less reliant on finely targeted social policies that focus on avoiding leakage to the ‘non-poor’ yet rarely have the kind of information needed to do this credibly, are often based on an incomplete accounting of the costs incurred (not least by poor people), and end up excluding many who are in real need. In combination with more reliable personal identification systems, a retreat from the fine-targeting fetish we often see today toward more transparent forms of universality promises more socially inclusive, politically effective and more cost-effective anti-poverty policies, not least in poor countries (Ravallion 2017).
A universal basic income at some decent level is not yet feasible in many countries. But more universality in service provision and social protection – and less fine targeting – would create better social policies.
This article first appeared on VoxDev: http://voxdev.org/topic/institutions-political-economy/straw-men-debate-basic-income-versus-targeting.
Ardington, Cally, Anne Case and Victoria Hosegood (2009), “Labor Supply Responses to Large Social Transfers: Longitudinal Evidence from South Africa”, American Economic Journal: Applied Economics, 1(1):22-48. Available here.
Banerjee, A, R Hanna, G Kreindler and B Olken (2017), ‘Debunking the Stereotype of the Lazy Welfare Recipient: Evidence from Cash Transfer Programs Worldwide’, mimeo, Massachusetts Institute of Technology.
Brown, C, M Ravallion and D van de Walle (2016), ‘A Poor Means Test? Econometric Targeting in Africa’, NBER (national Bureau of Economic Research) Working Paper 22919.
Friedman, M (1962), Capital and Freedom, University of Chicago Press, Chicago.
Mansuri, G and V Rao (2012), Localizing Development: Does Participation Work?, World Bank, Washington DC.
Murgai, Rinku, Martin Ravallion and Dominique van de Walle (2016), “Is Workfare Cost Effective against Poverty in a Poor Labor-Surplus Economy?”, World Bank Economic Review, 30(3):413-445.
Ravallion, M (2016), The Economics of Poverty: History, Measurement and Policy, Oxford University Press, Oxford and New York.
Ravallion, M (2017), Interventions against Poverty in Poor Places, World Institute of Development Economics (in press), Helsinki.
Sen, A (1995), “The Political Economy of Targeting”, in D van de Walle and K Nead (eds.), Public Spending and the Poor, Johns Hopkins University Press.
There seems to be much enthusiasm today for efforts to improve access to information about poor people’s rights and entitlements. In a much debated recent example, Facebook’s “Free Basics” platform provides free access to a selected slice of the internet (including, of course, Facebook). In arguing for Free Basics, Mark Zuckerberg says that “everyone … deserves access to the tools and information that can help them to achieve all those other public services, and all their fundamental social and economic rights.” I think we would all agree; less obvious is whether Free Basics will help do that. Critics argue that it is a “walled garden” approach—indeed, a threat to net neutrality. There have been proposals for other options using subsidized internet data packs, as in the proposal for India made recently by Nandan Nilekani and Viral Shah.
Neither the Facebook proposal nor that of Nilekani and Shah includes explicit pro-poor targeting. Is that needed? It might be argued that it is likely to be the poor who are least connected now, so the gains will automatically be greater for them. Against this, those who have the hardware and are currently connected are less likely to be poor and will probably be in the best position to benefit from these initiatives, including enjoying any new subsidies.
Before we decide on Free Basics versus subsidized data packs, or some other option, we should see how well information spreads at present. There is already lots of “public information” out there relevant to poor people in India, and there are various dissemination channels. While there may well be frictions for knowledge diffusion, associated with illiteracy and caste-based social exclusion, how important are they? Are the poor still sufficiently well connected socially to tap into the flow of knowledge, or does poverty come with social exclusion, including exclusion from information about programs designed to help poor people? Is a more explicitly targeted approach called for? An understanding of the sources of current inequality in information access is a pre-condition for thinking seriously about policies.
Using edutainment to learn about knowledge diffusion
The use of entertaining media—“edutainment” as Eliana La Ferrara dubs it in her paper “Mass Media and Social Change”—is attracting attention as a means of both directly informing poor people of their rights and entitlements and changing preferences and how existing communities operate. Such interventions can also provide a lens on existing processes of knowledge diffusion.
In a new paper, “Social Frictions to Knowledge Diffusion,” with Arthur Alik-Lagrange, I have used an edutainment intervention to identify key aspects of how knowledge is shared within villages in rural Bihar (a relatively poor state of about 100 million people in the Northeast of India). We show how an information campaign can throw light on the extent to which information is shared within villages. The campaign we studied used an entertaining fictional movie to teach people their rights under India’s National Rural Employment Guarantee Act (NREGA) (a motivating example used by Nilekani and Shah). NREGA created a justiciable “right-to-work” for all rural households in India. The most direct and obvious way NREGA tries to reduce poverty is by providing extra employment in rural areas on demand. This requires an explicit effort to empower poor people, who must take deliberate unilateral actions to demand work on the scheme from local officials.
In a book I wrote with Puja Dutta, Rinku Murgai and Dominique van de Walle, “Right to Work?” it was found that most men and three-quarters of women had heard about NREGA, but most were unaware of their rights and entitlements under the scheme. Given that about half the adults in rural Bihar are illiterate, a movie made sense as an information intervention. The setting and movie are described in Right to Work? and you can see the movie (audio in Hindi) on my website, economicsandpoverty.com.
The movie was tailored to Bihar’s specific context. Professional actors performed in an entertaining and emotionally engaging story-based plot whose purpose was to provide information on how the scheme works, who can participate and how to go about participating. The story line was centered on a temporary migrant worker returning to his village from the city to see his wife and baby daughter. He learns that there is BREGS work available in the village, even though it is the lean season, so he can stay there with his family and friends rather than return to the city to find work. It was intended that the audience would identify strongly with the central characters.
With the aim of promoting better knowledge about NREGA in this setting, the movie was randomly assigned to sampled villages, with a control group not receiving the movie. Knowledge about NREGA was assessed in both treatment and control villages. Residents were encouraged to watch the movie, but not (of course) compelled to do so. Some watched it and some did not. The new paper studies the impacts on knowledge, and the channel of that impact—notably whether it was purely through the direct effect of watching the movie or whether it was through knowledge sharing within villages.
There is a methodological challenge here, namely how to identify the knowledge gains (if any) for those in the assigned villages who did not actually watch the movie. We postulate that there is a latent process of knowledge diffusion among households within the village. An individual’s knowledge reflects both this process and a latent individual effect representing the individual’s “connectedness.” The latter is assumed to be time invariant, as it depends on long-standing networks of association between people, reflecting how each individual fits within the village social structure including caste positions and the ability of that individual to process the new information. Having two observations within each household allows us to obtain an estimate that is robust to latent heterogeneity in household factors. By exploiting the differences over time, our method is also robust to latent individual effects.
Socially differentiated knowledge spillovers
We find robust evidence of spillover effects, which account for about one third of the average impact of the movie on knowledge about NREGA’s key wage and employment provisions. While knowledge sharing is evident, poorer people, by various criteria, appear to be less well connected, and so benefit less from the spillover effect—relying more on direct exposure to the intervention.
Our key finding is that the knowledge diffusion process is far weaker for disadvantaged groups, defined in terms of caste, landholding, literacy, or consumption poverty. For poor people, it appears that the direct effect of watching the movie is all that really matters to learning about NREGA. There is also some indication of negative spillover effects for illiterate and landless households, suggesting the strategic spread of misinformation.
More effective pro-poor knowledge diffusion does not, of course, assure an effective public response on the service supply side. In another paper, “Empowering Poor People through Public Information?,” it was shown that the (direct and indirect) knowledge gains from the movie did rather little to assure a more responsive program. Right to Work? documents a number of specific, fixable, deficiencies in the responsiveness of NREGA in Bihar to the needs of poor people.
These research findings confirm that efforts are needed to improve the access of poor people to knowledge about public services that can help them, and that edutainment can work. The research also suggests that such efforts need to be directly targeted to poor groups, rather than relying on prevailing processes of knowledge diffusion, which may simply reflect, and reinforce, existing inequities.
(First posted on the World Bank’s Development Impact blog; 1/19/2016.)
The Rusty Radiator Awards for 2015 have been announced, in time for Christmas. These awards have been going for a few years. Not heard about them? From their website: “The Rusty Radiator Award goes to the fundraising video with the worst use of stereotypes. This kind of portrayal is not only unfair to the persons portrayed in the campaign, but also hinders long-term development and the fight against poverty.”
The “winner” this year is Band Aid 30. Apparently this was based on votes. The jury’s report: “Band Aid 30 contributed to the spread of misinformation and stereotypes of Africa as a country filled with misery and diseases. The Ebola outbreak occurred in three countries in West Africa. We resent the idea of a bunch of celebrities joining forces together, giving the impression that they are saving Africa from Ebola. Furthermore, they just make it so much more about themselves! Highly offensive and awful in every way possible. Celebrities cannot stop Ebola.”
Really? In principle I agree that stereotypes about poverty should be avoided. While the aid business has been somewhat prone to them, they do not help the cause of better development knowledge, including more and better aid. But is this particular award justified? The scene at the beginning of the Band Aid 30 clip a poor emaciated Ebola victim in her blood-soaked bed being carried away by health workers in their protective suits is shocking, especially when followed by a slew of cameras taking shots of the celebrity singers arriving for the recording of the clip. Not exactly good taste. But this is not one of those videos of well-healed celebrity musicians handing out food to poor African children, which irk me too (as in the clip for another Rust Radiator award winner). Nor did I get the impression that the Band Aid celebrities were claiming that they were saving Africa from Ebola. They might fairly be accused of bad taste, but I don’t see how they are promoting stereotypes that hinder long-term development and the fight against poverty.
There is a counter-argument. The Ebola crisis awakened many in the rich world to the appalling state of the health systems in West Africa (as in much of the world) AND that this longstanding development problem had spillover effects globally. This needs to be better known, and the Band Aid clip will help. The world is poorly equipped for handling pandemics, and some shock treatments like this may well help mobilize collective action to address the root causes. Showing such a scene at the beginning of the celebrity’s song is a little off-putting, but I think the Rusty Radiator Award is exaggerated in this case, and there is a counter-argument on positive benefits. Excess sensitivity does not help either dear Radiator folk.
In marked contrast, the winner of the Golden Radiator Award is Zalissa’s Choice from Burkina Faso. (From the website: “The Golden Radiator Award goes to the fundraising video using creativity and creating engagement. This kind of charity campaign is stepping outside of the common way with using stereotypes.”) Zalissa’s Choice is an uplifting choice for this award! I recommend it.
This article does not try to summarize the book, but rather to focus on the main lessons that emerge on the challenges in thinking and action about poverty going forward.
A transition in thinking: In reviewing the history of thought on poverty I was struck by how much mainstream thinking has changed over the last 200 years. We see a transition in the literature and policy debates between two radically different views of poverty. Early on, there was little reason to think that poor people had the potential to be anything other than poor; poverty would inevitably persist. Prominent thinkers even argued that poverty was necessary for economic advancement, since without it, who would farm the land, work the factories and staff the armies? Avoiding hunger was the necessary incentive for doing work.
This way of thinking still left a role for policy in providing a degree of protection from shocks, which helped assure social stability in the wake of crises. The protection motive for antipoverty policy goes back well over 2,000 years in both Western and Eastern thought. While the need for social protection was well understood in principle among the elites, their support tended to fade in normal times, and often needed to be re-established in new crises. However, mass poverty was largely taken for granted. Beyond short-term palliatives to address shocks, there was little or no perceived scope for public effort to permanently reduce poverty. Promotional antipoverty policies made little sense to those in power.
In the second, modern, view, poverty is not only seen as a social ill that can be avoided through public action, but doing so is seen as perfectly consistent with a robust growing economy. Indeed, the right antipoverty policies are expected to contribute to that growth by removing material constraints on the freedom of individuals to pursue their own economic interests. Poverty was no longer seen as some inevitable, even natural, condition, but as something that could and should be eliminated.
The state came to be given a prominent role in helping to assure that all individuals have access to the essential material conditions for their own personal fulfillment—arguably the most important requirement for equity, but also the key to breaking out of poverty traps. Antipoverty policy came to be seen as a matter of both promotion and protection. Along with rising real wages and (hence) savings by poor people, public education systems, sound health systems and reasonably well functioning financial markets were deemed to be crucial elements for the next generation to escape poverty, for good.
This transition in thinking came with much struggle. Many people came to protest, join community or religious groups, labor and civil-rights movements, or political coalitions of one sort or another to lobby for governmental action to help fight chronic poverty in multiple dimensions. The resistance to their efforts was often strong, and a great many brave people sacrificed their liberty and even their lives in those struggles over centuries.
Successful promotion policies took time to evolve, and were invariably mediated by politics. However, in due course, a self-reinforcing cycle emerged in the successful countries to help assure a sustained and (over time) more rapid escape from absolute poverty. Success in implementing partial antipoverty policies often fostered success in securing broader coverage and implementing new initiatives.
Progress has been slow in some periods and the cycle has been broken at times, with many setbacks. The history of thinking and action on poverty provides ample illustrations of the fragility of the progress that has been seen. Each major step forward was followed by a backlash in thinking and policymaking. We still see this today, with poor people being blamed for their poverty and even criminalized. But, as we strive to support efforts against poverty going forward, it is important to acknowledge the progress that has been made.
Progress against poverty: The new millennium still has roughly one billion people in the world living in poverty by the standards of what “poverty” typically means in the poorest countries. While the data are far from ideal, as best can be determined, there were also about one billion people in the world living in such poverty 200 years ago. The difference is that then they accounted for about four out of five people in the world, while today they account for one in five.
Since 2000 the developing world has been reducing the extreme poverty rate at about one percentage point per year—over three times the long term annual rate for the world as a whole over the last 200 years. If this progress can be maintained then the developing world will eliminate at least the most extreme forms of absolute poverty in a much shorter timespan than did today’s rich world. We should not, however, presume that the developing world’s new pace of progress against poverty will automatically be sustained. That will require good policies, and a measure of good luck.
Challenges to eliminating extreme poverty: There are two distinct paths going forward. The low-case, “pessimistic,” trajectory entails that the developing world outside China regresses back to the relatively slow progress of the 1980s and 1990s. On this path, it would take another 50 years or more to lift one billion people out of poverty. This would surely be judged a poor performance. By contrast, an “optimistic path” would be to maintain the higher growth rate for the developing world as a whole seen since 2000 without a rise in overall inequality. If that could be achieved then we can be reasonably confident of lifting that one billion people out of extreme poverty by sometime around 2030 (Ravallion, 2013).
What are the principle challenges in assuring that the second path is followed? Among the list of threats one can identify to attaining that goal, inequality stands out as a major concern today. Rising inequality can mean that growth largely by-passes poor people. This has been happening in some countries of the rich world, including the U.S.. Experience among developing countries has been varied. Inequality falls about as often as it rises in growing developing countries, although absolute poverty measures tend to fall with growth. High-inequality countries have a harder time reducing poverty in that they typically need higher growth rates than low-inequality countries to attain the same pace of progress against poverty and their high inequality often makes that growth even harder to attain.
In thinking about the implications for policy it is important to un-pack inequality—to identify the specific dimensions most relevant to progress against poverty. Inequalities in access to good quality schooling and health care stand out in many developing countries today. In many rural economies, inequalities in access to land (including insecurity of rights over land) also remain an impediment to pro-poor growth. Gender inequalities stand out everywhere, though not just in terms of command over material goods.
More pro-poor policies call for better quality public institutions and services that are inclusive of the needs of poor people. With the required political will there is much that can be done to improve health and education services in poor places and in making legal systems more inclusive. These are high priorities for antipoverty policy everywhere. While the “heavy-lifting” against poverty will probably continue to come from pro-poor growth processes, there is also an important supportive role for redistribution and insurance using state-contingent transfers, ideally financed primarily by domestic taxation. And that role is unlikely to be temporary; all countries need a permanent safety net. In thinking about the (many) options, policy makers in developing countries should be more open to the idea of only broadly targeted and largely unconditional transfers (as distinct from finely targeted conditional transfers). Improving tax systems in poor countries to expand the revenue for domestic antipoverty policies must also be a high priority.
External development assistance should continue to play a role. This is ethically compelling in its own right but also as compensation for the costs rich countries impose on poor ones (such as through past contributions to the stock of greenhouse gases and the past injustices of colonial exploitation and trade restrictions). Aid has two important roles. First there will be emergency aid—short-term assistance to deal with crises. There will be concerns about moral hazard, which have to be taken seriously, but wealthier countries should be called upon to help poor countries deal with agro-climatic and other shocks. Second, development assistance should help foster the conditions for sustainable poverty reduction in the longer-term, including institutional development and building better public administrations (such as for domestic resource mobilization).
It must be acknowledged, however, that the record of development aid has been uneven, and not always well-considered in the light of what we have learned about the economics of poverty. For example, a common view among aid donors is that they need to incentivize better policies—to use a carrot and stick approach, rewarding good efforts and punishing bad ones. This is a risky strategy and may well push fragile states into a poor-institutions trap (Ravallion, 2016, Chapter 9).
Market failures are an important reason why inequality matters to progress against poverty. Credit market failures have been a prominent concern. The policy responses entail some combination of efforts to make markets and institutions work better for poor people and efforts to compensate for market failures through other means, including redistribution. Inequality can also undermine the potential for making such policies happen. Those who benefit from their ability to capture new opportunities will often resist reforms that try to assure broader access to those opportunities, also given that poor people on their own have little current capacity to compensate the non-poor losers from pro-poor reforms. History is full of examples. English industrialists in the 19th century lobbied against compulsory schooling and against bans on child labor, and helped stall those reforms for a long period. Indian industrialists in the post-Independence decades lobbied for trade protection that diminished the scope for poverty reduction through labor absorbing export-led manufacturing growth. Powerful landholders in both these countries (and elsewhere) effectively undermined the potential for land reforms and other redistributive policies. And in many countries, insiders in urban formal-sector labor markets (on both sides of the market) act to effectively restrict competition from outsiders.
“Inequality” is not a single idea, but takes many forms, and can be seen very differently by different people. This fact creates much debate—though sometimes this seems like a debate between ships passing in the dead of night, not seeing or understanding each other’s perspective. The differences in how inequality is perceived can also stall pro-poor economic policies. A good example is the reaction that some people (understandably) have to rising absolute inequality. The once widely-held “stylized fact” of development that higher relative inequality is the unavoidable “price” for growth and (hence) poverty reduction, has been overturned in the light of new theories and evidence. Poor growing economies can and have avoided rising relative inequality, but they will have a much harder time avoiding rising absolute inequality—a rising absolute gap between the rich and the poor. And many citizens view inequality in absolute rather than relative terms. They are justified in taking that view; the concept of (relative) inequality held by most economists derives from an axiom that need not be accepted, and (indeed) appears to be rejected by many people. Those who view inequality as absolute and value it independently of poverty will see a trade-off between poverty and inequality. Ameliorating concerns about rising absolute inequality will almost certainly entail less progress against poverty.
Urban poverty is another challenge. The urbanization of poverty—whereby poverty rates fall more slowly in urban areas than in rural areas—is to be expected in almost any developing country that is successful in reducing poverty overall. Urban economies create new opportunities that poor people in rural areas have often sought out to improve their lives. Distorted urban labor markets can readily create excessive urbanization, as can the lack of effective public efforts to promote agriculture and rural development; indeed, many developing countries have gone even further in (explicitly or otherwise) taxing the rural economy to support the urban economy. Removing long-standing policy biases in both taxation and public spending remains a high priority for pro-poor growth. No less misguided are restrictions on migration and urban policies that under-supply services to poor urban residents, including rural migrants. Poor people are often trapped as the victims of policies that simultaneously repress agriculture while making life difficult for rural migrants. Development policymaking needs to be more neutral to these two sectors of economic activity. That will probably still entail an urbanization of poverty, but that should not be a cause for alarm as long as poverty is falling overall.
The sustainability of poverty-reduction efforts poses a further set of challenges in assuring that we can reach the optimistic path. We do not want to reach the poverty-reduction target only to fall back in subsequent years. On an encouraging note, research has suggested that lower initial levels of absolute poverty at a given mean consumption foster higher subsequent rates of growth in average living standards in developing countries and help to ensure that economic growth itself is poverty reducing (Ravallion, 2012). Thus, a “virtuous cycle” can be anticipated that would help to ensure the sustainability of the reduction in poverty.
Relative poverty: The other side of the coin to falling absolute poverty is rising relative poverty. Economic growth has generally meant lower absolute poverty rates, but over time relative considerations have become more important. Such relative poverty is still poverty. Welfare concerns about relative deprivation and costs of social inclusion demand higher real poverty lines as average incomes grow (though it makes little sense for this to be a constant proportion of average income in developing countries). But progress against relative poverty will be slower. Even the optimistic path will leave another one billion or so people in the world who live above the frugal poverty lines typical of the poorest countries but are still poor by the standards of the countries they live in. This type of poverty can also be eliminated but it will almost certainly require much stronger redistributive efforts than we have seen to date in most countries.
The policies are available. The bigger challenges ahead are in assuring the political will and administrative capabilities to implement and enforce sound antipoverty policies, and in adapting them to differing circumstances and evolving knowledge about their efficacy.
Ravallion, Martin, 2012, “Why Don’t we See Poverty Convergence?” American Economic Review, 102(1): 504-523.
____________,2013, “How Long Will It Take To Lift One Billion People Out Of Poverty?” World Bank Research Observer 28(2): 139-158.
____________, 2016, The Economics of Poverty: History, Measurement and Policy, New York: Oxford University Press.
The decade or two after WW2 saw many of the world’s poorest countries gain their independence from Colonial rule, and they were hoping to rapidly become less poor. Economics taught policy makers in those countries that a higher investment rate is crucial to assuring faster economic growth. Being a poor country makes it harder to finance the required investments from domestic savings. Yet rich countries should have ample savings available that might be profitably diverted to this task. In an ideal world, global capital markets could be expected to bridge the gap. But 70 years ago those markets were thin and/or not trusted as a source of finance.
In response, the United Nations Monetary and Financial Conference, held at Bretton Woods in 1944, created the International Bank for Reconstruction and Development (IBRD)—a core component of what came to be known as the World Bank. (The International Monetary Fund was created at the same time.) The essential idea was that the IBRD would borrow money on global markets to lend to developing countries. The Bank’s AAA credit rating (stemming from conservative lending policies relative to its capital) allowed it to lend on favorable terms. An aid-facility (with a large grant component), the International Development Association (IDA), was added in 1960.
Much has changed in the 70 years since the famous Bretton Woods conference. World Bank lending (IBRD+IDA) now represents only about 5% of the aggregate private capital flows to developing countries. In the last 10 years or so there have been prominent calls for radically reforming, or even closing, the institution on the grounds that international capital markets have developed greatly over those 70 years. It is also claimed by some that the Bank’s efforts are wasted due to poor governance in developing countries.
Does the Bank still have an important role? If so, does it fulfill that role, and if not, how might it do better? In a new paper I argue that the Bank’s development role today overlaps only partially with its original role, as conceived at the Bretton Woods Conference 70 years ago (Ravallion, 2015). Its role today is complementary to (rather than competing with) the private financial sector, other development banks, and academia. Knowledge-generation is central to that role. Development knowledge has properties of a public good, which the Bank can generate in the process of actually doing development on the ground.
Threats to the Bank’s effectiveness: There is still much appeal to the bundling of knowledge with lending that has been the distinctive feature of the Bank’s operations. But there are a number of threats to the efficacy of this model.
There have been some longstanding concerns that the Bank’s “lending culture” rewards operational staff for the volume of their lending, with only weak incentives for assuring that knowledge is both applied and generated in the lending operations. The pressure to lend influences the Bank’s ability to deliver objective policy advice to client countries, even when it is not welcome politically. Too often the Bank’s “country strategy” essentially mirrors that of the government, which may or may not serve broader long-term development goals.
Another threat is the perception that the Bank’s most powerful shareholders have excessive influence on its operations and policy advice. The U.S. has long been identified in this role, though some other countries have also been keen to have their say. Some critics are concerned (rightly or wrongly) about conflicts of interest when the Bank gives advice to developing countries.
These are threats to the Bank’s effectiveness as a knowledge leader in both the public and private sectors. All parties—both clients in developing countries and private investors—must have confidence that the institution is not pushing lending for its own sake or beholden to a few powerful owners. Only then can the Bank be accepted as the source of the objective policy advice and information that is needed.
Recent organizational changes have made some effort to put knowledge in the driver’s seat by organizing the Bank around a set of sectorally-defined “practices.” In the end the organogram has changed rather little. However, the threats to the Bank’s effectiveness are unlikely to be solved by changing the Bank’s organogram. The incentives of managers and staff also need to change, to assure a better alignment with development goals. (See Ravallion, 20105, for some examples of specific proposals for reform from past Bank staff.)
Knowledge Bank? There has been much rhetoric about the “Knowledge Bank” over the last 15 years, but I am not alone in believing that the reality has fallen short of the rhetoric. There is a chronic and growing underinvestment in the kind of rigorous research that is needed to identify and address pressing development issues—both the constraints on rapid poverty reduction at country level and the global public bads that threaten us collectively (ranging from climate change to pandemics). Research has been under-valued and under-funded.
Granted we still see some high-quality research at the Bank, though not always on high-priority topics. We see more ex-post evaluations today than 20 years ago. However, much does not get evaluated, and what gets evaluated is a non-random subset of all projects, casting doubt on what we learn about the whole. Too often, methodological preferences drive what gets evaluated rather than the knowledge gaps facing policy makers. Alongside this, we see fewer and less rigorous ex ante evaluations, which make explicit a project’s economic rationale—why the project is expected to have a social value justifying its cost.
Three changes are needed: Echoing the observations of others within and outside the Bank, three things need to change:
Further reading: Martin Ravallion, “The World Bank: Why it is Still Needed and Why it Still Disappoints,” Journal of Economic Perspectives, Winter 2016.
(This was first posted on the World Bank’s Let’s Talk Development blog.)
More thought has been given to the validity of the conclusions drawn from development impact evaluations than to the ethical validity of how the evaluations were done. This is not an issue for all evaluations. Sometimes an impact evaluation is built into an existing program such that nothing changes about how the program works. The evaluation takes as given the way the program assigns its benefits. So if the program is deemed to be ethically acceptable then this can be presumed to also hold for the method of evaluation. (I leave aside ethical issues in how evaluations are reported and publication biases.) We can dub these “ethically benign evaluations.”
Another type of evaluation deliberately alters the program’s (known or likely) assignment mechanism—who gets the program and who does not—for the purpose of the evaluation. Then the ethical acceptability of the intervention does not imply that the evaluation is ethically acceptable. Call these “ethically contestable evaluations.” The main examples in practice are randomized control trials (RCTs). Scaled-up programs almost never use randomized assignment, so the RCT has a different assignment mechanism, and this may be contested ethically even when the full program is fine.
A debate has emerged about the ethical validity of RCTs. This has been brewing for some time but there has been a recent flurry of attention to the issue, stimulated by a New York Times post last week by Casey Mulligan and various comments including an extended reply by Jessica Goldberg. Mulligan essentially dismisses RCTs as ethically unacceptable on the grounds that some of those to which a program is assigned for the purpose of evaluation—the “treatment group”—will almost certainly not need it, or benefit little, while some in the control group will. As an example, he endorses Jeff Sachs’s arguments as to why the Millennium Villages project was not set up as an RCT. Goldberg defends the ethical validity of RCTs against Mulligan’s critique. On the one hand she argues that randomization can be defended as ethically fair given limited resources, while (on the other hand) even if one still objects, the gains from new knowledge can outweigh the objections.
I have worried about the ethical validity of some RCTs, and I don’t think development specialists have given the ethical issues enough attention. But nor do I think the issues are straightforward. So this post is my effort to make sense of the debate.
Ethics is a poor excuse for lack of evaluative effort. For one thing, there are ethically benign evaluations. But even focusing on RCTs, I doubt if there are many “deontological purists” out there who would argue that good ends can never justify bad means and so side with Mulligan, Sachs and others in rejecting all RCTs on ethical grounds. That is surely a rather extreme position (and not one often associated with economists). It is ethically defensible to judge processes in part by their outcomes; indeed, there is a long tradition of doing so in moral philosophy, with utilitarianism as the leading example. It is not inherently “unethical” to do a pilot intervention that knowingly withholds a treatment from some people in genuine need, and gives it to some people who are not, as long as this is deemed to be justified by the expected welfare benefits from new knowledge.
Far more problematic is either of the following:
The latter situation is clearly objectionable if it is seen to hold. But it is often hard to verify in development settings. Ethics has been much discussed in medical research. In that context, the principle of equipoise requires that there should be no decisive prior case for believing that the treatment has impact sufficient to justify its cost. (This is David McKenzie’s sensible modification to clinical equipoise to fit the types of programs in discussion here.) By this reasoning, only if we are sufficiently ignorant about the likely gains relative to costs should we evaluate further. Implementation of such an ethical principle may not be easy, however. In the context of antipoverty or other public programs, a priori (theoretical and/or empirical) arguments can often be made both for and against believing ex ante that impact is likely. A clever researcher can often create a convincing straw man to suggest that some form of equipoise holds and that the evaluation is worth doing. While this cannot be prevented, we should at least demand that the case is made, and it stands up to scholarly public scrutiny. That is clearly not the norm at present.
It has often been argued that whenever rationing is required—when there is not enough money to cover everyone—randomized assignment is a fair solution. (Goldberg makes this claim, though I have heard it often. Indeed, I have made this argument a few times with government counterparts in attempting to convince them on the merits of randomization.) In practice, this is clearly not the main reason that randomistas randomize. But should it convince the un-believers? It can be accepted when information is very poor, or allocative processes are skewed against those in need. In some development applications we may know very little ex ante about how best to assign participation to maximize impact. But when alternative allocations are feasible (and if randomization is possible then that condition is evidently met) and one does have information about who is likely to benefit, then surely it is fairer to use that information, and not randomize, at least unconditionally.
Conditional randomization can help relieve ethically concerns. One first selects eligible types of participants based on prior knowledge about likely gains, and only then randomly assigns the intervention, given that not all can be covered. For example, if one is evaluating a training program or a program that requires skills for maximum impact one would reasonably assume (backed up by some evidence) that prior education and/or experience will enhance impact and design the evaluation accordingly. This has ethical advantages over simple randomization when there are priors about likely impacts.
But there is a catch. The set of things observable to the evaluator is typically only a subset of what is observable on the ground (such information asymmetry is, after all, the reason for randomizing in the first place). At local level, there will typically be more information—revealing that the program is being assigned to some who do not need it, and withheld from some who do. The RCT may be ethically unacceptable at (say) village level. But then whose information should decide the matter? It may be seen as quite lame for the evaluator to plead, “I did not know” when others do in fact know very well who is in need and who is not.
Goldberg reminds us of another defense often heard, namely that RCTs can use what are called “encouragement designs.” The idea here is that nobody is prevented accessing the primary service of interest (such as schooling) but the experiment instead randomizes access to some form of incentive or information. This may help relieve ethical concerns for some observers, but it clearly does not remove them—it merely displaces them from the primary service of interest to a secondary space. Ethical validity still looms as a concern when any “encouragement” is being deliberately withheld from some people who would benefit and given to some who would not.
While ethical validity is a legitimate concern in its own right, it also holds implications for other aspects of evaluation validity. There is heterogeneity in the ethical acceptability of RCTs. That will vary from one setting to another. One can get away with an RCT more easily with NGOs than governments, and with small interventions, preferably in out-of-the-way places. (By contrast, imagine a government trying to justify why some of its under-served rural citizens were randomly chosen to not get new roads or grid connections on the grounds that this will allow it to figure out the benefits to those that do get them.) An exclusive reliance on randomization for identifying impacts will likely create a bias in our knowledge in favor of the settings and types of interventions for which randomization is feasible; we will know nothing about a wide range of development interventions for which randomization is not an option. (I discuss this bias for inferences about development impact further in “Should the Randomistas Rule?”.) Given that evaluations are supposed to fill our knowledge gaps, this must be a concern even for those who think that consequences trump concerns about processes.
If evaluators take ethical validity seriously there will be implications for RCTs. Some RCTs may have to be ruled out as simply unacceptable. For example, I surely cannot be the only person who is troubled on ethical grounds by the (innovative) study done in Delhi India by Marianne Bertrand et al. that randomized an encouragement to obtain a driver’s license quickly, on the explicit presumption that this would entail the payment of a bribe to obtain a license without knowing how to drive. (This study was conducted and funded by the World Bank’s International Finance Corporation. And it was published in a prestigious economics journal.) The study confirmed that the process of testing and licensing was not working well even for the control group. But the RCT put even more drivers on Delhi roads who did not know how to drive, adding to the risk of accidents. The gain from doing so was a clean verification of the claim that corruption is possible in India and has real effects, though I was not aware of any prior doubt about the truth of that claim.
There may well be design changes to many RCTs that could assure their ethical validity, such as judged by review boards. One might randomly withhold the option of treatment for some period of time, after which it would become available, but this would need to be known by all in advance, and one might reasonably argue that some form of compensation would be justified by the delay. Adaptive randomizations are getting serious attention in biomedical research; for example, one might adapt the assignment to treatment of new arrivals along the way, in the light of evidence collected on covariates of impact. (The U.S. Food and Drug Administration issued guidelines a few years ago.)
The experiment might not then be as clean as in the classic RCT—the prized internal validity of the RCT in large samples may be compromised. But if that is always judged to be too high a price then the evaluator is probably not taking ethical validity seriously.
(First posted on the World Bank’s Development Impact blog.)
In 2013 the World Bank announced that one of its two goals is to “share prosperity,” which is to be measured by the growth rate in mean consumption (or income) for the poorest 40% of the population. (Its other goal is eliminating absolute poverty.)
The growth rate in the mean for the poorest 40% has the appeal of simplicity. But that comes with a cost. An important concern is that it does not tell us anything at all about how much rising prosperity is being shared amongst the poorest 40%, or how the losses from economic contraction are being spread. For example, the mean of the poorest 40% could rise without any gain to the poorest.
That limitation is important in the light of recent research. While the developing world as a whole has made huge progress in reducing the numbers of people living in poverty, much less progress has been made in raising the developing world’s consumption floor—the level of living of the poorest. (I show this here.) If we are really committed to sharing prosperity then we should surely not be leaving the poorest behind.
There is a remarkably simple fix for the Bank’s measure of success in sharing prosperity: Instead of measuring the growth rate of the mean for the poorest 40% the Bank should measure the mean growth rate of the poorest 40%. This may sound like some nerdy quibble, but it does matter. This subtle difference in the measure makes a big difference in its properties. With this change, the measure of “shared prosperity” reflects how equitably aggregate gains have been shared amongst the poorest 40%. If inequality falls (rises) among the poorest 40% then the mean growth rate will be higher (lower) than the growth rate of the mean. The mean growth rate of the poorest 40% is also quite easy to calculate from any two standard household surveys. (They do not need to be panel data.)
Let’s take a simple example. Suppose that there are four representative people comprising the poorest 40%, with incomes (in $s per day) of $0.75, $0.75, $1 and $1. After some economic shock or policy change their incomes are $0.50, $0.75, $1 and $1.25, i.e., there is a gain of $0.25 for the least poor, at the expense of the poorest. The growth rate in the mean of the poorest 40% is zero, while their mean growth rate is -2% (the average of -33%, 0%, 0%, 25%).
This change may still not be considered enough. There are other measures, although they often lose the advantage of simplicity. In more careful monitoring, a broader dashboard of measures will clearly be needed in assessing how well prosperity is being shared. For example, if one really cares about not leaving the poorest behind then one should also focus directly on that; there are now operational measures for that purpose, which are also easy to implement, as I show in this paper.
However, my point here is that one can improve the Bank’s measure with only a small change in wording. And the alternative measure can be implemented at virtually no extra cost in monitoring.
(In the interests of full disclosure, I participated in the internal discussions at the Bank in 2012 about its goals. I am writing this post two years after leaving the Bank, and in the light of research since then.)