Martin Ravallion

Martin Ravallion has written 46 posts for ECONOMICS & POVERTY

Inequality and Social Distancing during the Pandemic

The COVID-19 pandemic in the U.S. saw a remarkable behavioral response through physical mobility. There was a sharp contraction in non-residential mobility in the first half of 2020—over 40-50% declines relative to the pre-pandemic levels—which then dissipated over time as the pre-pandemic patterns started to be restored, though with smaller contractions in each of the two winters. Residential mobility had returned to something close to pre-pandemic levels by mid-2022.

Whose mobility are we talking about here? An important role has been played by individual choices related to physical mobility, especially in the early stages when pharmaceutical interventions were not yet available. The individual choices comprised both personal actions (such as choosing to shop online rather than in person, and compliance with local policies) as well as efforts to influence the actions of others (advocating local “stay-at-home” mandates, for example). Those choices were undoubtedly influenced by the socioeconomic characteristics of people and communities. We can thus ask about the socioeconomic incidence of the mobility response. Depending on that incidence, the pandemic may come to reflect, and possibly reinforce, antecedent socioeconomic inequalities.

In a new paper, “Inequality and Social Distancing during the Pandemic,” Caitlin Brown and I study the socioeconomic incidence of the dramatic behavioral responses through physical mobility over the course of the pandemic. We merge Google Mobility Reports across the 3,000 U.S. counties with socioeconomic characteristics as well as (more standard) covariates suggested by the epidemiological literature. We use these data to try to understand the joint epidemiological and socioeconomic covariates of the mobility responses across counties, and how these changed over the course of the pandemic.

Our paper points to theoretical ambiguities in the socioeconomic incidence of the mobility responses to the threat of infection. Among the socially excluded and largely non-working poor (such as the elderly or disabled poor), social distancing may not be much of a burden. Yet the cost could be high among the working poor, since such families cannot easily maintain their consumption in isolation. The pre-pandemic levels of social and economic interaction are likely to be higher for wealthier people, and they face costs of adjusting quickly to a lower level of physical mobility. The marginal effects on social distancing of income differences may also vary with income across counties, though here too we argue that the direction of this effect of between-county inequality on social distancing could go either way.

For the US we find that counties with a higher median income tended to experience greater reductions in mobility outside the home in the initial phase, though re-bounding substantially by mid-2022. This holds when we control for the poverty rate, suggesting that the effect is coming from the attenuated mobility of the non-poor in the early phase of the pandemic, alongside enhanced mobility in the subsequent re-adjustment. While pre-existing inequalities were reflected in social distancing in the pre-pharmaceutical phase, this was partly reversed later.   

Counties with a higher poverty rate and higher income inequality tended to see larger declines in non-residential mobility. (Since we are controlling for average income, these are relative distributional effects, rather than absolute.) Behavioral responses through physical mobility in the pre-pharmaceutical phase were more protective of those living in more affluent and unequal areas. The near-linearity of the median income effect across counties implies little or no trade-off between reducing geographic inequality and the aggregate mobility response to the pandemic. The distributional effects are mainly through within-county channels, with only a small reduction in non-residential mobility attributable to between-county inequality.

Our results are not consistent with the view that higher inequality undermines social distancing, such as by eroding local prosocial norms relevant to the chances of infection. The results are more suggestive of greater scope in high-inequality areas for enforcing such norms during the pandemic, possibly backed-up by local policies. By interpretation, the same desire among the rich (buttressed by their ability-to-pay) to protect themselves from infection by personal effort helped promote local public efforts for social distancing in high-inequality areas.

We point to two broader implications of our study. First, efforts to understand social distancing, and to respond through policy, cannot ignore the distribution of income. While voluntary social distancing can be a strongly protective response, it is a response that is firmly grounded in antecedent socio-economic inequalities. The behavioral response to the threat of infection can be highly heterogeneous across income strata and over time during the course of the pandemic. However, the specific pattern of such dependence is hard to predict on a priori grounds, and it is not (as we have shown) simply a situation in which inequality impedes collective social distancing—indeed, our results suggest the opposite.

Second, there may be implications for social policies. Our findings suggest that, in the absence of enforced policies to support social distancing, it will be the poorer and yet relatively equal areas that are more vulnerable to the spread of infection. Self-protection is easier for those in relatively well-off and unequal areas. Our interpretation is that poorer families are less able to afford to protect themselves, which leads them to make different social-distancing choices. This suggests that there may be a role for antipoverty policies as a complement to more direct health-policy measures in combating infectious disease, especially in the initial pre-pharmaceutical phase. Such policy implications beg for exploration in greater evaluative depth than we have been able to provide.

Poor people photography: when is it OK and when is it not?

In Part 1 of the Economics of Poverty (and in ECON 156: Poverty and Inequality) we briefly discuss some of the views of the great C18th German philosopher, Immanuel Kant. Kant is not often identified as a key thinker about the economics of poverty (and I can claim little expertise in moral philosophy). But in my view Kant took a key step in thinking relevant to poverty and anti-poverty policies, namely to argue that poor and other disadvantaged people should be given respect in society—yes, they may be poor but that should not mean that they are treated disrespectfully.   

That brings me to consider the two types of “poor people photography” that we see in modern times, especially in development settings.

First, we see serious efforts, like Gapminder’s “Dollar Steet,” aiming to use photography to provide objective information that helps teach people how others in the world live. Efforts like Dollar Street are good for raising public awareness of poverty. Similar efforts go way back in the history of thought on poverty (as also discussed in Part 1 of EOP).

The second type is what has come to be known as “poverty porn.” I refer to stereotypical, over-simplified, often cringe-worthy, photographic depictions of poor people, typically aiming to attract donations or promote some cause. We have seen the pictures of starving children with flies swarming around them. Radi-Aid has often pointed to other examples, including through their Rust-Radiator Awards.

As development folk resume field work and other travel “post-pandemic,” and seek to illustrate their talks, fund-raising efforts, high-profile reports and so on, we might reflect on how to differentiate the first type of poor people photography (which we want more of) from the second (less please). There is more sensitivity to this issue today, but not enough in my view.

A key requirement (that Kant would approve of) is obvious respect for the subject. A minimal list of criteria for that would surely be:

  1. The subject’s approval; don’t take that photo without asking for the subject’s approval, in clear terms.
  2. No obvious damage is done to the interests of the subjects and preferably there is a good case for believing that they will benefit.

OK, but that minimal list seems inadequate. Something is missing. Any thoughts? Please go to my tweet on the topic (May 12 2022) if you have suggestions or comments that might help us come up with some practical ethical guidelines/norms.

New Estimates of Poverty in India since 2011

India has had a long and (mostly) distinguished record in monitoring poverty, based on the National Sample Surveys (NSS), done by the Government of India’s National Sample Survey Organization (NSSO). (The surveys have had many applications, but here the focus is on measuring poverty.) Prior to the early 1990s, researchers had to rely on NSSO’s published tabulations from the surveys. This changed when the micro data became publically available. There have been debates about how poverty should be measured using the NSS, and how the results should be interpreted. But these debates were at least well-informed, given public access to the data and an active research community within and outside India’s borders.

That changed with the 75th NSS round for 2017-18, which the Government chose not to release to the public in any form. (A Press Release from the Ministry of Statistics pointed to concerns about the quality of the data, though with few details.) The prior round of the large sample (five-yearly) surveys that have been most relied upon in measuring poverty was 2011-12. That means that it is now over ten years since we have been able to update the estimates of poverty in the country that has historically held more of the absolute poor than any other (judged by a global poverty line that aims to have constant purchasing power across countries). In the World Bank’s much used interactive PovcalNet website, this became a gaping hole in the global picture of poverty. (You can see the hole here; notice how the South Asia rows for recent years are labelled “Survey coverage is too low”; that is entirely due to the missing NSS data for India.) 

There were some efforts to figure out what the missing survey data would have told us about how India’s poor were doing. Some observers argued that the poverty rate had risen after 2011, others claimed that it had fallen, though by different amounts depending on who you listened too. (See Sandefur, 2022, for further discussion.) Some observers have tried to provide an answer largely avoiding the need for household survey data (as in, for example, Bhalla et al. 2022). However, it has long been agreed that household surveys are necessary for credibly measuring poverty. Here my focus is on the problem of getting an internally consistent survey-based series of poverty measures for India in the context of measuring global poverty.

Until early April 2022 it seems that we were still pretty much in the dark about how India’s poor have been doing over the last ten years. That changed with the release of a World Bank working paper by Sutirtha Sinha Roy and Roy van der Weide (2022).

In a new Note, “Filling a Gaping Hole in the World Bank’s Global Poverty Measures,” posted on the website of the Center for Global Development, I review what this paper has done, and then discuss what their results tell us.  Here I will just summarize the new estimates in the following graph, which combines the PovcalNet estimates of the poverty rate in India with those of Roy and van der Weide. 

The new estimates indicate a continuing decline in the national poverty rate since 2011. This is, of course, good news. To put the Roy and van der Weide results in historical context, India’s progress against poverty was undeniably slow prior to the 1990s, but has picked up its pace since then. (For a longer-term perspective on India’s progress against poverty and how it differed between the “pre-reform” and “post-reform” periods see Datt et al. 2020.) Eyeballing the Roy and van der Weide results in the context of the series for India in PovcalNet, and past work in the literature, I would not interpret their findings as either a marked “acceleration” or “deceleration” in India’s progress against poverty since the early 1990s, but it is indicative of continuing progress. If one does simple end-point comparisons, the annual rate of decline in the poverty rate was slightly higher post 2011 than post 1993 (1.5 pp per year 1993-2019 versus 1.6 in the period 2011-19), but I would not make too much of this difference.

Roy and van der Weide (2022) only provide the headcount index, but they kindly provided their estimates of the poverty gap (PG) and squared poverty gap (SPG) measures (Foster et al. 1984). My CGD post provides these measures. Unlike the headcount index, these measures are responsive to changes in consumption below the poverty line. (The headcount index only changes when households cross the line.) So this is also worth checking. The results generally follow a similar pattern over time to the headcount index, also echoing past findings in the literature related to earlier periods. Over the period as a whole, the correlation coefficients between the headcount index and PG and SPG are 0.99 and 0.98 respectively; taking the first differences, the correlation coefficients are 0.85 and 0.69. (One difference is that the SPG in urban areas exceeded that in rural areas in 2018—the only year that was observed.)

 Another perspective on progress against poverty is provided by estimating the consumption floor, as defined Ravallion (2016). This can be thought of as a Rawlsian measure of poverty, focusing on the level of living of the poorest. Of course, there is likely to be considerable measurement error if one were to use the (literally) lowest observed level of consumption in any sample survey as the measure of the “floor”, so some averaging is clearly called for. Ravallion (2016) proposes that the consumption floor should be measured by taking a weighted mean of all consumptions below the poverty line, with highest weight on the lowest observed consumption, and with weights declining linearly as consumption rises, reaching zero at the poverty line.

 The next graph provides this measure of the consumption floor, again for both PovcalNet and the Roy and van der Weide series. (My CGD post provides a tabulation of the numbers.)

We can see that the floor has generally risen over time, though rather slowly, from $1.09 a day in 1983 to $1.35 in 2019. Progress in lifting the floor has clearly been slow since 2011, when it was $1.32 a day. There has been convergence between the urban and rural consumption floors up to 2011. The Roy and van der Weide series suggest a fall in the urban consumption floor since then, and a reversal in the prior pattern of a higher floor in urban than rural areas. The signs of a drop in the urban floor could well reflect changes in the type of migrants moving from rural to urban areas over the last ten years.

These findings on the consumption floor provide a qualification to the evidence pointing to India’s progress in reducing the proportion of the population living in poverty in the last decade (and prior to that), namely that this progress appears to have come with only limited gains for the poorest, who are probably living at a similar standard of living today to that found in the last 2011-12 NSS round. Fewer people live at or near the floor, but the floor has changed rather little. 

Next steps

We still need to see what India’s poverty measures over the last ten years look like when using India’s national poverty lines, applied to the careful re-shaping of the CPHS consumption data by Roy and van der Weide. The World Bank’s international $1.90 a day line in 2011 prices is anchored to the poverty lines found in low-income countries (updating the $1.25 line for 2005 proposed by Ravallion et al. 2009, based on a survey of national poverty lines). Other lines can also be defended, as PovcalNet readily allows.

Relative poverty measures are also of interest, to better reflect changing perceptions of what “poverty” means in growing developing countries such as India. The “strongly relative” measures that have been popular in Western Europe are hard to defend on theoretical grounds, especially in developing countries, but “weakly relative” measures offer a better compromise (Ravallion and Chen 2011).

Of course, the best “next step” would be for the Government of India to return to its policy of providing public access to the NSS data. Until then, we at least have something credible to work with, to fill that gaping hole in our knowledge about global poverty.


Bhalla, Surjit, Karan Bhasin and Arvind Virmani, 2022, “Pandemic, Poverty, and Inequality: Evidence from India” IMF Working Paper 2022/069.

Datt, Gaurav, Martin Ravallion, and Rinku Murgai, 2020, “Poverty and Growth in India over Six Decades,” American Journal of Agricultural Economics 102 (1): 4-27.

Drèze, Jean, and A. Somanchi, 2021, “View: New Barometer of India’s Economy Fails to Reflect Deprivations of Poor Households,” The Economic Times June 21.

Foster, James, J. Greer, and Erik Thorbecke, 1984, “A Class of Decomposable Poverty Measures,” Econometrica 52: 761-765.

Ravallion, Martin, 2016, “Are the World’s Poorest Being Left Behind?” Journal of Economic Growth 21(2): 139–164.

Ravallion, Martin, and Shaohua Chen, 2011, “Weakly Relative Poverty,” Review of Economics and Statistics 93(4): 1251-1261.

Ravallion, Martin, Shaohua Chen and Prem Sangraula, 2009. “Dollar a Day Revisited,” World Bank Economic Review 23(2):163-184.

Roy, Sutirtha Sinha, and Roy van der Weide, 2022, “Poverty in India Has Declined over the Last Decade But Note As Much as Previously Thought,” Policy Research Working Paper 9994, World Bank, Washington DC.

Sandefur, Justin, 2022, “The Great Indian Poverty Debate 2.0,” Blog Post, Center for Global Development.

Fleshing out the Olive? Income Polarization in China

Beyond the popular goal of “ending poverty,” national leaders rarely articulate a reasonably well-defined goal for the national distribution of income. In an exception, at a prominent and widely-reported meeting in August 2021 of the Chinese Community Party’s Central Committee for Financial and Economic Affairs, President Xi Jinping argued that the goal of “common prosperity” for China required an “olive-shaped distribution structure of large middle and small ends” (as reported by the Xinhua News Agency, August 17). In short, the proposed aim of the Chinese leadership is to “expand the proportion of middle-income groups” (Xinhua News Agency)—what the Economist magazine dubbed “fleshing out the olive” in an August 28 article reporting on President’s Xi’s speech.

President Xi was clearly not saying that this is the only policy goal for China, even within the gamut of goals related to the distribution of income. (Xi has often emphasized the goal of ending poverty.) So, the question naturally arises as to what trade-offs might exist against other goals. That is a difficult question. Trade-offs can be hard to identify ex-post in observable data, which also reflect past policy choices (given the trade-offs faced at the time) and shocks. Nonetheless, it is of interest to see what the historical experience suggests about trade-offs with regard to this new goal. This requires that we can quantify attainments of the multiple distributional goals, including defining and measuring the idea of “fleshing out the olive.”

The well-documented success of China in reducing absolute poverty came (of course) with a rising share of the population living above the absolute poverty line, many of whom joined what can be thought of as China’s “middle-class.” Naturally, what this means depends on the setting. The prevailing definition of a “middle-income group” can be expected to change over time with rising living standards; what was considered a “middle” income in the China of the 1980s is clearly not the same as today. “Fleshing out the olive” can be interpreted as reducing the spread of incomes relative to the current median, which arguably provides a more relevant reference point than a fixed absolute level of real income.

This perspective suggests that the concept of “polarization” found in economics is relevant to monitoring China’s performance in “fleshing out the olive,” and identifying potential trade-offs against other goals, including poverty reduction. And there is a measure available in the literature, namely the Foster-Wolfson (FW) polarization index. This measures the spread of incomes relative to the median.

While much has been written about poverty and inequality in China, rather little has been said about polarization. A new paper, “Fleshing out the Olive,” with Shaohua Chen, provides polarization measures for China spanning the post-reform period after 1980. This allows us to identify some key sub-periods when polarization was stable and even falling. The variance in the time-series allows us to explore the covariates of polarization. For example, we will be able to see whether there are signs in the historical record that less polarizing periods saw lower rates of economic growth.  

Conceptually, polarization is not the same thing as inequality, which suggests the possibility of a trade-off between the two. While it is not something that has attracted much attention in the literature, one might expect that the process of economic development through structural transformation in a country such as China may have a de-polarizing effect, as the poorest move closer to the middle. Nor is this an aspect of the potential distributional changes with development that is likely to be captured well by the standard inequality indices. These potentially de-polarizing gains among the poorer half may, however, come hand-in-hand with polarizing gains among the (primarily urban) upper half, comprising an elite of skilled workers and those who own the capital stock and/or rental properties.

Also relevant in the context of China is the evolution of the large disparities found between mean incomes in urban and rural areas. This reflects long-standing inequalities in social policies (health, education and social protection) as well as impediments to internal migration, notably through the hukou registration system, and administrative land allocation processes. (Our paper provides references to the literature on these points.) Given the large mean income gaps between China’s urban and rural areas, the degree of urban-rural sectoral fractionalization—the extent to which people live in different sectors—may also matter to both income inequality and polarization.

Our new paper points to some potential trade-offs between reducing income polarization and other valued goals. Some policies that are good for fighting poverty and inequality could well be polarizing. Policy makers need to be aware of these potential trade-offs. In addition to arguing that the Foster-Wolfson index is a close match to the spirit of the idea of “fleshing out the olive”—and so provides a valuable tool for monitoring progress in attaining that goal—the paper has looked for signs of such trade-offs in the aggregate time series data for China since 1981.

A focus on polarization begs some new policy questions that have so far been largely ignored. A prominent example in contemporary China is the Central Government’s goal of eliminating the hukou registration system—the internal “passport” system in China that restricts the access of rural migrants to urban services and markets. While ongoing reforms to the hukou system would undoubtedly help reduce poverty, the impact on polarization is unclear, given that the bulk of both the personal benefits and the costs of relaxing hukou restrictions may well fall on the lower side of the median, suggesting that these reforms could be polarizing. The potential for such polarizing effects of relaxing hukou restrictions would need to be balanced against other considerations, including poverty reduction.

However, our paper finds rather little evidence in the time-series data we have assembled of any negative co-movement between polarization (on the one hand) and economic growth or reducing poverty and inequality (on the other). Granted, polarization rose with rising average incomes up to 2009, but this appears to be spurious, reflecting common time trends. Periods of higher poverty reduction or higher economic growth did not typically see more rapid polarization. And there is strong co-movement between the Gini index and the Foster-Wolfson polarization index. Nor do we find that periods of a more rapid rise in the urbanization of the poorer half of the population (who started off almost only in rural areas) tended to be more polarizing.

To the extent that reducing polarization is a new policy goal for China, the historical record does not point to any serious trade-offs with past goals going forward, including with economic growth and poverty reduction. The recent reversal in the generally upward path for polarization in China has been driven almost entirely by attenuated median-normalized incomes among the upper half.

Of special relevance to thinking about the policy options in reducing polarization is our finding that the rise and fall in China’s national polarization index is largely accountable to the evolution of the gap between urban and rural mean incomes. Here too, the historical record provides little support for the idea that reducing urban-rural disparities would be polarizing—indeed, the data suggest the opposite. However, potential trade-offs would need to be considered further in the context of specific policy efforts, such as in expanding social service coverage in rural areas, also taking account of how those efforts are financed.

It might look like a “Kuznets curve,” but appearances can be deceptive: Income inequality in China since 1980

There has been a tendency to declare a “Kuznets curve” whenever we see an “inverted U” in how some social, economic, political or environmental variable evolves with economic development. Yet the turning point in the inverted U may have little or nothing to do with the process of economic growth through modern-sector enlargement that was postulated in the classic development models of Simon Kuznets and Arthur Lewis (as reviewed in EOP, Chapter 8).

The implications (including for policy) may depend crucially on why we see such a turning point.

Inequality has become a prominent concern in China, both in its own right, and as a possible threat to future growth prospects, to the extent that high inequality restricts investment by relatively poor people (including investment in human capital) and possibly undermines growth-promoting policy reforms, which get blocked by the new, economically and politically powerful, elites.

In this context, it is now widely believed that, around 2008-9, China reached its Kuznets-Lewis turning point for income inequality. Since then, inequality has trended downwards. This is good news, but we need to understand better why it happened.

My new paper with Shaohua Chen assesses whether the claimed “Kuznets curve” for China has anything to do with Kuznets (Ravallion and Chen 2021). We confirm that our new measures look like a “Kuznets curve,” at least since the mid-1990s. When we go back further in time, to 1981, it starts to look more like an upward sloping roller coaster ride than the famous “inverted U,” as can be seen from the following graph, plotting our new estimates of two popular inequality indices for China.

A deeper analysis of the data indicates that the country’s turning points for inequality have rather little to do with the Kuznets model of economic growth through urbanizing structural transformation, as formalized in development economics. We show that the Kuznets Hypothesis would not have generated anything like the path for inequality measures seen in China. Key assumptions of that model simply do not hold. The Kuznets-type growth process alone would only have delivered a very flat Kuznets curve, with virtually no long-run increase in inequality, and a trend decline since around 2000 (Ravallion and Chen 2021).

Nor do the data for China support the view that the urbanization process has been driving up urban-rural disparities in mean incomes, or inequalities within either urban or rural areas; if anything it is the opposite.

We need to look elsewhere to understand why we have seen such a large long-term increase in inequality in China, with multiple turning points. There has been a long-term pattern (going back to the early Maoist period) of “urban bias” in China’s development policies. Economic and human development policies have systematically favored urban areas. This has been reinforced by internal impediments to migration and administrative controls of agricultural land assignments, making it harder and more risky to migrate.

The pattern in the inequality data over time is very consistent with the various date-specific reversals in the longer-term pattern of divergence between urban and rural mean incomes. Indeed, once we control for the ratio of the urban mean to the rural mean–keeping this ratio at its initial, 1981, value–we find no sign of a trend increase (or decrease) in income inequality in China since the mid-1990s. The counterfactual Gini index would have actually seen a trend decline!

Causal attribution is difficult of course, but our paper points to specific policy changes that helped the rural sector, and coincided closely with all three periods of declining inequality in China’s economic history since 1980, including the two (known) inverted-U turning points.

Just as China’s first turning point for inequality was short lived (see the graph above), we cannot be complacent about the latest turning point, which has received so much (hopeful) attention in China. These inequality trends and reversals in the direction of change are not the realizations of some more-or-less inevitable, theoretically grounded, process in economic development through urbanizing structural transformation. In large part, for China, the turning points in inequality appear to reflect policies.

Unless there is a continuing commitment to redistributive effort—which (as in most developing countries) includes the prioritization given to agriculture and rural development—China could well return to its upward trajectory for overall inequality. 

Macroeconomic Misery by Levels of Income in America

At the time of writing (mid-2021), there is a public debate in the U.S. stemming from fears that the macro policies in the wake of the 2020-21 pandemic are stimulating inflation. Concerns have been raised about the impacts of macro policy choices on poverty and inequality.

The relative importance of inflation versus growth and employment as macroeconomic indicators has long been debated. Arthur Okun’s famous “Misery Index” (OMI hereafter)—developed when Okun was an advisor to the Johnson Administration in the U.S. during the 1960s—added up the inflation rate with the unemployment rate. Since then, others have argued that the growth rate should be included (weighted negatively).

Economic theory has offered some insights, but what does the evidence suggest on how these macro variables impact levels of real income in America, ranging from the poorest to the richest?

Micro evidence can tell us (for example) whether poorer families have higher rates of unemployment. However, macro evidence can also reveal indirect effects on real incomes; for example, a higher economy-wide unemployment rate may reduce the wage-bargaining power of workers in poor families or reduce (public or private) transfer payments to those families. There is also likely to be heterogeneity within any given income group, such as due to differences in dependence on the labor market, wage setting, discrimination by race or gender, and wealth portfolios. It is thus of interest to see if systematic differences in the mean impacts of these macro variables are evident at different levels of income.

In a new paper I ask how the relative importance of three prominent macro indicators—the rate of unemployment, the inflation rate and the growth rate of GDP per capita—depends on whether one is talking about the real incomes of the poor, middle-income groups or the rich.

The new paper explores these issues using real income distributions assembled from almost 30 years of survey data since the 1980s, spanning a wide range of macro-outcomes, including the Great Recession (GR) of 2009-10.  The literature does not suggest that any single summary measure of “inequality” or “poverty” could adequately capture the nature of the distributional changes induced by these macro variables. So I look at data on incomes from the poorest to the richest Americans. The regression specifications aim to isolate the short-term effects of the macro variables at each income level.

The paper’s results indicate a systematic pattern in how the key macroeconomic indicators influence real incomes in America. The unemployment rate should have higher weight than inflation in a Misery Index calibrated to real incomes across the whole distribution, at least in this time period.

A higher unemployment rate unambiguously increases poverty measures (for all measures and lines). It also reduces the skewness of the distribution—an effect that is not evident in its (ambiguous) implications for inequality. This more complex distributional pattern found in the study is not evident if one only looks at a summary statistic of overall inequality, such as the popular Gini index.

Inflation matters more in the middle of the distribution than in the tails. GDP growth rates matter at all levels of income, and especially for the poorest; however, this effect is largely attributable to the impact of growth on the unemployment rate.

The restrictions implied by the OMI—namely equal weighs on unemployment and inflation and excluding the GDP growth rate—are rejected across the bulk of the distribution, and strongly so for the poorer strata; indeed, the OMI appears to only be defensible for the top income groups.

Missing top income recipients and what to do about them

Low response rates among rich households are thought to be a serious problem in many applications using household surveys, including the measurement of poverty and inequality and in distributional policy evaluation. A new paper, “Missing Top Income Recipients” (link below), discusses the various ways the problem can be dealt with, and makes some recommendations for practice. The context includes developing countries with weak statistical systems, as well as rich countries with “state-of-the-art” systems.

Finding that richer households are less likely to participate in surveys when sampled does not necessarily imply that we under-estimate inequality, by any standard measure. That is an empirical question. (We can actually be more confident on theoretical grounds that standard poverty measures will be over-estimated for this reason.) The empirical studies so far (mainly for the U.S.) do suggest sizeable under-statement of the extent of inequality based on survey data, including with the commonly used methods of trying to correct the problem, either ex-ante or ex-post.  There are also serious concerns about under-estimation of mean income from surveys. On balance, poverty measures tend to be reasonably robust to this problem.

What can be done about the problem? It will be obvious to any user of micro survey data for empirical analysis that it is desirable to correct for the bias in top-income shares due to selective compliance internally to the survey data. Doing so can retain both the statistical integrity of the survey design and the great many applications for micro-data files in distributional analysis. How can that be done?

Non-compliance with survey sampling can take the form of outright refusal or not being home for interviews. Either way, it is an outcome of choices made by those sampled. A theme of the paper is that thinking about behavioral responses helps to clarify the concerns about the non-behavioral methods underlying current practices, and it helps in thinking about better methods and measures.

As has long been recognized in the statistics literature, under certain conditions, income-selective non-compliance with an initially randomized assignment can be corrected by reweighting the survey data. Economists have given too little attention to how survey weights are estimated; indeed, it seems that the weights are almost always taken for granted.

Past methods of calculating these weights do not appear to take proper account of the behavior of those selected for a sample, and thus do not deal fully with the survey compliance problem. Newer methods are reviewed in the paper—methods that can do a better job, and their data requirements should make them feasible in many settings. Modelling compliance requires identifying assumptions, and these are contestable, though certainly no more so than the assumptions made by ad hoc non-behavioral methods.

The (old and new) reweighting methods require that the surveys pick up at least some top incomes. That may be a serious concern—called a “failure of common support”—although there is remarkably little evidence one can point to. Indeed, the literature on top incomes using non-survey data has boomed with rather little evidence on whether common support holds, such that survey data could be adequately re-weighted.

If common support does not hold, then income tax records can help, including in estimating distributional national accounts. There has been much progress in obtaining and carefully using these data, also in combination with survey data. There are also some reasons for caution. Tax data come with their own concerns including tax avoidance/evasion, weak coverage of informal sectors and illicit incomes, including capital flight, and concerns about construct validity, given the limitations of taxable income as a basis for inter-personal comparisons of economic welfare. The paper reviews the methods and issues going forward, but it is clear that more work is needed.

While such measurement problems need not prevent us making the best estimates possible from the data available, the validity of the data and methods can, and should, be challenged continually, and under public scrutiny. That is how we make progress in measurement.

How much all this matters to policy remains to be seen. Just as top incomes are hard to reveal for measurement purposes—and we may never know just how unequal income and wealth are distributed—they can be hard to tax for financing and redistributive purposes. The hope remains that progress in advancing measurement will also help foster better policies.

Poor people can gain from allowing benefits-in-kind to be tradeable, but that is not all that is needed to realize those gains

Governments typically prohibit the resale of the benefits-in-kind often provided by antipoverty programs. Yet the personal gains from those benefits are likely to vary and to be known privately, so there can be gains to poor people from trading their assignments. We know very little about those gains.

To help address this knowledge gap, my new Working Paper, “On the Gains from Tradeable Benefits-in-Kind,” models a competitive market for assignments, and simulates the market using an unusual survey of workers on a rural public-works scheme in Bihar, a poor state of India.

The results indicate large gains from tradeable assignments after first randomizing. A competitive market for tradeable assignments would generate aggregate gains that are around 2 to 3 times the current mean gains to these (mostly very poor) families. It would also have greater impact (by a similar magnitude in terms of mean gains) than an allocation without re-sale options targeted to workers from consumption-poor families. Allowing the assignments to be tradeable in this setting can also make workfare more effective against poverty than (budget-neutral) cash transfers. The simulated allocations with tradeable assignments imply a tendency for somewhat larger gains among poorer households, not the opposite. Gains are similar between male and female workers.

However, the paper also argues that, given the realities of the setting, fully realizing the gains from trade in practice may require complementary policies to help people access the market and to support its administration and regulation. 

On the history of impact evaluation

Good evaluative research has had a huge impact on the lives of people. Careful observational studies have long been important tools for such research. (Chapter 6 of EOP provides a non-technical review of the methods found in practice.)

An example is the use of border discontinuities—an instance of what is often called these days a “regression discontinuity design.” Here the key assumption is that the counterfactual in the absence of a “treatment” of interest is continuous across some geographic borderline; all that differs across that border is the presence of the treatment. Under that assumption, we can infer the impact of the treatment by comparing outcomes either side of the border. The assumption is important, and one should look for any confounding factors in each application—any reason why the counterfactual outcomes might differ across the border.

A famous example of this type of evaluation method is found in the work of David Card and Alan Kruger, in their book, Myth and Measurement. Card and Kruger were interested in the effect of a higher minimum wage rate on employment. They compared employment in the fast-food industry either side of the border between two US states. The minimum wage rate had risen on one side of the border, but not the other. They found that the higher minimum wage increased incomes but did not reduce employment. (There has been a continuing debate on this issue, though the best evidence so far that I know of, from work by Arindrajit Dube and colleagues using comparisons of contiguous counties between US states with different minimum wages, comes to a similar conclusion to Card and Kruger.)

If you had asked me yesterday for the first example of the use of border discontinuity in impact evaluation, I would probably have said Card and Kruger. I just found out that I would have been wrong by 100 years! In an 1893 volume (available here in a 1895 English translation), the German microbiologist and medical doctor Robert Koch used this method to identify the effect of public water filtration on cholera in Germany. In case you don’t know, cholera is a truly awful bacterial disease. If untreated, the dehydration brought on by cholera will kill even a healthy person in just a few hours.

Famously, John Snow had mapped the incidence of deaths in a severe and geographically concentrated outbreak of cholera in 1854 London. The map suggested that the incidence of deaths due to cholera was associated with a specific drinking water supplier (the Broad Street pump in Soho) that had been contaminated by sewerage. Prior to this, the prevailing view was that cholera was an air-borne disease, not water borne. Snow’s map was essentially a descriptive tool, albeit a powerful one. It did not clinch the case for believing that dirty water was the culprit, though it had much influence on the debate.

In the 1892 cholera outbreak in Hamburg, Robert Koch conducted an observational study that was more suggestive of a causal impact of contaminated water. Koch observed that along the border between Hamburg (with no public water filtration) and neighboring Altona (with water filtration), there was cholera on the Hamburg side but not the Altona side. Koch (p.25) argued that (in today’s terminology) the counterfactual was continuous across this border:

On both sides of the frontier, the state of the soil, the buildings, the sewerage, the population, in short all of the conditions that are of importance in this connection, are perfectly similar, and yet the cholera in Hamburg spread only to the frontier of Altona, and stopped there.” (p.25).

Interestingly, Koch goes on to provide a further test. It turns out that there was a group of residences on the Hamburg side that got their water from Altona. Yes, you guessed it: little or no cholera in that group of houses.

This was influential in persuading European cities to filter their water. In his path-breaking history of the world viewed through the lens of infectious diseases, Plagues and Peoples, William McNeill writes that, based on Koch’s observational study, “Doubters were silenced and cholera … never returned to European cities.” (p.278).

There can be no doubt that this empirical demonstration of the health benefits of clean drinking water has saved countless lives, and made life much better for countless others who did not die from diseases such as cholera, but suffered directly from the illness, or through the harm to their loved ones.  

If you know of other early examples of the use of observational studies in impact evaluation I would like to hear about them. Tell me about it at

Martin Ravallion

Poverty in China since 1950: A Counterfactual Perspective

China’s progress against poverty since around 1980 has been much applauded. On Christmas day, 2020, China Xinhua News (the official news agency) released a video claiming that China’s progress against poverty was “the greatest achievement in world history.” One need not go quite that far to agree that reducing the number of poor by something like 800 million (judged by the World Bank’s international poverty line) is a huge accomplishment.

However, to draw useful policy lessons, our applause for China’s success against poverty needs to come hand-in-hand with an acknowledgment of the preceding policy failures. That is not to deny that China has made enormous progress against poverty since Deng Xiaoping unleashed the country’s pro-market reforms in the late 1970s. Rather, it is to remind us of both stages in China’s history post-1949.

Any evaluative interpretation of a measure of social or economic outcomes over time requires consideration of a relevant counterfactual trajectory. My new paper, “Poverty in China since 1950: A Counterfactual Perspective,” argues that the historical record suggests that two relevant counterfactuals for China were geographically and culturally close at hand around 1950, namely South Korea and Taiwan. The armed conflicts, political realignments and economic setbacks within North East Asia during the first half of the C20th left China, Taiwan and South Korea very poor by 1950, though China was clearly poorer. In all three, the rural populations were especially poor. Very few people in the rural sectors would have met the World Bank’s current international poverty line.

During the 10 years after the end of WW2, two relatively new but very different economic models emerged out of the poverty of North East Asia. The Maoist path, under Communist Party Chairman Mao Zedong, did not appear immediately in mainland China, and there were options and debates, especially over the relative importance of rapid industrialization versus agriculture and rural development. So too for South Korea and Taiwan, but they took a different course to China, namely a form of political capitalism, which was (by the late 1970s) to be influential in Deng’s reform path for China.

Like mainland China, both South Korea and Taiwan had historically been supportive of governmental intervention in production, and this did not fundamentally change after 1950. The main point of departure was in whether the state should actually own the means of production. My new paper argues that, based on the historical record, a version of political capitalism was a viable option for China around 1950. The country took another path, in no small measure reflecting the personality and power of Chairman Mao.

When judged against the development paths of South Korea and Taiwan, the “difference-in-difference” estimates provided by my paper indicate that the Maoist path meant that an extra quarter or more of the Chinese population were living in poverty by the time Deng’s reforms began. The historical data are far less than ideal and need to be handled with caution. My paper provides various checks and tests, including placebo tests using the data prior to Mao taking control of mainland China in 1949; the available historical data go back to 1820. There is no sign of anything one could call a significant impact of the Maoist path prior to Mao taking over.

Acknowledging the data uncertainties, it is nonetheless clear enough that a large share of China’s post-reform reduction in the incidence of poverty was the country’s success in correcting the past failures in its economic policies. While much has been written and debated in efforts to explain China’s success since 1980, there is a lot less to explain when we view the country’s record against poverty within this broader historical perspective.

The new trajectory after Deng’s reforms allowed China to catch up over the following 10-20 years. My main estimates suggest that by about 1990, just after Deng had resigned as leader, the post-reform trajectory of poverty reduction had fully made up for the “lost ground” that I attribute to the Maoist regime. There are data uncertainties here too (notably in merging historical data with new survey-based data). An alternative estimation method suggests this catch-up took an extra 10 years or so.

When my new findings are combined with other evidence from the literature, it also becomes clear that Deng’s initial focus on agrarian reforms was a crucial element of this remarkable pro-poor catchup. Here we have an historical echo of the debates among China’s political elite (then including Deng) back in the early 1950s.

China’s progress against poverty revisited

The widely held view that China has greatly reduced income poverty over the last 40 years does not accord with all the evidence. My new paper with Shaohua Chen, “Reconciling the Conflicting Narratives on Poverty in China,” tries to reconcile the conflicting findings.

The fact that strongly-relative measures provided in the graph above show rising poverty is easily understood, since such measures depend solely on relative distribution, and inequality in China has been rising until recently. (Strongly relative measures entail setting the poverty line as a constant proportion of the mean or median.)

More surprising, and revealing, is the story told by the official lines, which were revised twice since the original 1985 line. Our new paper shows that the official lines are neither absolute nor strongly relative. Rather, they are weakly relative, with a positive elasticity to the mean that is less than unity. Along with rising inequality, this feature slowed the pace of measured poverty reduction when compared to absolute measures.

Nonetheless, substantial progress against poverty is indicated, as we confirm using our independent, and higher, weakly-relative lines calibrated to cross-country data as in the graph above. Our alternative schedule of weakly-relative lines shows a marked reduction in poverty—falling from 88% to 23% over 1981-2014—though not as much as the absolute $1.90 lines.

Our implied weakly-relative line for 2014 accords well with independent estimates of the subjective poverty line for China, which is about three time the official poverty line.

Download the paper here

A Short History of the Idea of Ending Poverty

There are times and places when announcing a goal for ending poverty is clearly little more than a symbol of good intentions. It tells poor citizens, and those who care about them, that the government (or international agency) purports to be on their side, even if nothing much is done to ease poverty. This can be called a “symbolic goal.”

At times there have also been more substantive aims. Advocates against poverty have variously seen it as: the most morally objectionable aspect of inequality, stemming mainly from economic and political forces rather than bad choices by poor people; a key material constraint on human freedom and social inclusion; a risk of deprivation, whether currently poor or not; and a cost to other valued goals, including economic efficiency, human development and environmental sustainability. The actions that might be motivated in response range from specific policies to efforts to help poor people organize collectively for things that matter to them. Thus, goal setting is seen as an incentive mechanism for attaining better outcomes. We can call this the “motivating goal.”

My new paper, “On the Origins of the Idea of Ending Poverty,” provides a short history of the idea of ending poverty as a motivational goal, and tries to draw some lessons from that history. (This blog post is little more than a summary to hopefully stimulate reading the paper.)

Ending poverty is a modern idea, little evident in pre-modern times. The balance of factors influencing the motivating goal changes with economic development, and varies from one place to another. Politically, the perceived benefits depend on the weight given to poor people, which depends in turn on their voting power and their capacity to organize. The cost of ending poverty through redistribution depends (in part) on how much poverty there is, relative to the resources thought to be available. It can be no surprise that calls for ending poverty have been heard more often when a society’s total resources make it more feasible to do so.

History confirms the intuition that “ending poverty” has little political traction as a near-term goal when mass chronic poverty is seen to be the norm and poor citizens have little political influence. When those conditions no longer hold, a political goal of “ending poverty” can motivate public action to end poverty. The late 18th century saw the intellectual germ of the idea, but it did not get far in economics or policy making until much more recently. Over the 19th century, poverty rates fell substantially in Western Europe and North America, and we started to see mainstream advocates of ending chronic poverty, and policies for doing so.

While the history of the idea of ending poverty confirms that political constraints matter, it also suggest that they are not deterministic. Social and economic thought, and data, have often played a role. One could not talk seriously about ending poverty until it was agreed that less poverty was a good thing, and here Adam Smith was influential in overturning the prior mercantilist thinking that saw poverty as essential for wealth generation. Descriptions (both qualitative and quantitative) of the lives of poor people have also had much influence, often shaming the non-poor into supporting actions to help poor people. The effort to document poverty, especially those of the late 19th century, also fostered the development of modern empirical social science, including economics.

In the wake of high inequality and the critiques and rising influence of the socialist and labor movements, and the heightened public awareness of poverty, the period around the late 19th century saw the beginning of a concerted effort to reduce poverty and inequality in much of today’s rich world. This was echoed in economic thinking; the most famous economist of the time, Alfred Marshall was asking in 1890, “May we not outgrow the belief that poverty is necessary?” Welfare states started to emerge, alongside progressive income taxation and minimum wage laws in the early part of the 20th century.

The poverty focus gained political momentum in the wake of the Great Depression. Famously, in America, President Franklin D. Roosevelt’s new social programs—bundled under the label “New Deal”—included the Social Security Act, which introduced federal pensions for the elderly, transfers for families with dependent children, and unemployment benefits.  

There was new interest globally in the idea of ending poverty after the Second World War, and an explosion of interest and effort from around 1960, with policy responses in many countries, including America, notable under the Johnson administration’s War on Poverty. The debates about poverty in America in the 1960s and ‘70s both reflected past debates, but also anticipated issues that would be prominent going forward, especially about the relative importance of economic growth versus redistribution.

In the post-Colonial period, the newly independent states—what we came to call the “developing world”—were keen to see an end to poverty. Some of this was clearly little more than symbolic goal setting. Progress was slow for most countries. An acceleration in progress against poverty emerged around the year 2000.

The U.N.’s first Millennium Development Goal (MDG1) of halving the 1990 poverty rate by 2015 was achieved ahead of time. The fact that MDG1 was achieved has been taken by some observers to imply that it was hugely motivational, though some of the claims made for the power of MDG1 have clearly been exaggerated. One might equally well argue that MDG1 was not ambitious enough. More worryingly, however, is that halving the 1990 poverty rate was attained with only modest gains to the poorest.

The U.N.’s Sustainable Development Goals came to include ending extreme poverty by 2030. This is more ambitious than MDG1, and more politically challenging. SDG1 focuses attention on the poorest 10% globally, although it also highlights regional priorities; 40% of the population of Sub-Saharan Africa still live below that line. Importantly, SDG1 cannot be attained if the poorest are left behind, as we saw in the MDG1 period.

Attaining SDG1 will clearly not be the “end of poverty” (as the U.N.’s rousing labelling of the goal suggests). Many of those who are no longer poor by the global $1.90 standard will still be poor by the (defensible) standards of the country they live in. Nonetheless, getting everyone above a global line that 10% do not currently reach, and 40% did not attain 40 years ago, would be an achievement.

The path to attaining SDG1 calls for some combination of economic growth, especially when fueled by pro-poor technical progress, and pro-poor redistribution. The political context clearly matters to the relative importance of growth versus redistribution, but so does the level of economic development. When there is a lot of poverty—such that redistribution is politically and economically challenging, if not impossible—economic growth may be all that we can hope for as a politically feasible response. There have been cases of rising poverty with economic growth, but they are rare over the longer term. The Catch-22, however, is that poverty typically makes it harder to grow an economy.

History suggests that the dynamics of poverty reduction can sometimes work synergistically with the political economy to accelerate progress; the heavy lifting is done by growth, but then redistribution starts to take over. This virtuous cycle has been evident at times in the history, but it can come unstuck, especially when the poorest are harder to reach, and one can point to arguments and evidence as to why that might be so. It is undeniably good news that fewer people live near the floor to living standards, but it is sobering that the floor has not risen more.

SDG1 will probably not be attained with a return to “business as usual” after the COVID-19 pandemic. Restoring economic growth in poor countries will almost certainly be required. There is scope for more effective redistributive policies, and even efficiency-promoting redistributions, though there are continuing challenges in assuring that these policies reach the poorest. There is also a more widespread recognition that the economic growth that has helped so much to reduce aggregate poverty measures has also come with environmental costs, including global warming. Huge challenges lie ahead in how to manage the likely tradeoffs between the “social” and “environmental” SDGs.

Can the World’s Poor Protect Themselves from the New Coronavirus?

In rich countries, the messages have got out reasonably well on how best to protect oneself from COVID-19. Lockdown and social distancing have been important, as have personal hygiene measures. Learning from public announcements has also been key. The World Health Organization (WHO) has played an important role in getting the message out. It is clear that these nonpharmaceutical measures can help contain the spread of illness.

Virtually all of the WHO recommendations require that a household environment supports the capacity to protect from the virus—what can be called the “home environment for protection” (HEP). The HEP is the result of past wealth-constrained choices, and these are unlikely to change quickly. Dwelling attributes such as its size, construction and location (determining access to treatment) cannot be easily adjusted in response to the immediate virus threat, and nor is health all that people care about when allocating their resources.

Importantly, all of the aspects of the HEP required for compliance with the WHO recommendations are likely to have a wealth effect, meaning that poorer households will have less capacity to protect themselves by following WHO recommendations. This is to be expected between countries as well as within them.

A new paper with Caitlin Brown and Dominique van de Walle asks “Can the World’s Poor Protect Themselves from the New Coronavirus?”? In a nutshell, our answer is “no.” We assemble data on some key attributes of the home environment for one million sampled households from the latest Demographic and Health Surveys. We find that prevailing WHO recommendations for protection make unrealistic assumptions about the home environments of the bulk of the world’s poor. Our calculations indicate that 90% of households have inadequate HEP by one or more of the dimensions considered. Strikingly, the recommendation of having a place to wash hands with soap is not satisfied by the majority of households in the developing world overall, and only satisfied by one-in-five households in sub-Saharan Africa. 40% do not have a formal health-care facility within 5km. Only 6% of the poorest 40% have a home environment that supports full compliance with the WHO recommendations, and the proportion is virtually zero in sub-Saharan Africa.

Our analysis leads us to conclude that the developing world, and especially its poorest half, is ill-prepared to protect itself from this virus. For most households, the recommendations that have been implemented on a massive scale in the rich world must be considered near fiction for the world’s poor.

Given the contagion rate of this virus, the likely degree of exposure to be expected among a large segment of the population of the developing world also points to a serious concern for the entire population.

What can be done? The housing stock cannot be changed rapidly. But there are still things that can be done now, as I discuss further in my paper, “Pandemic Policies in Poor Places.” The current infrastructure for information (particularly cell-phone coverage) holds promise for getting the messages out on public health and access to consumption support. Policies such as distributing or subsidizing soap and improved water access could be feasible in the near term, and justified by both the external benefits and the equity impacts. Home-grown innovative adaptations to the realities of life in the developing world will be crucial.



On the virus and poor people in the world

While this is a signed posting, I do that only to take responsibility for the content, rather than to claim originality. These notes draw on enumerable sources from media, emails, conversations, as well as my personal observations and knowledge drawing on the more mainstream literature. The post is provided as a contribution to public debate, not as any attempt at a definitive statement.

Martin Ravallion

The lack of systematic testing for the Covid-19 virus makes nonsense of the counts of incidence, especially in the developing world. Death counts are probably more reliable, but still flawed. Nonetheless, from what we know so far, the developing world is on roughly the same trajectory as for the developed world, just with a lag. Death rates might be somewhat lower, given younger populations. However, less well-nourished and less healthy young populations could well turn out to be just as vulnerable.

On balance, it is a reasonable expectation that this will be a huge health and economic shock to the developing world, and especially to poor people.  

Reliable public data and communication is crucial at this time

Yet in many developing countries, the government is pretending that it is in control, or that the threat is minimal. This delusion does not help. A different approach is needed.

  • Each government needs to establish an authoritative independent communication channel on public health during the pandemic. A trusted health authority should be providing public health warnings and protocols on social distancing, hand washing etc.
  • On the technology side, internet connectivity is weak in many poor places, though cellphone usage is more widespread. Texting is a promising method, but his can also be a source of miss-information. Text messages from the dedicated health authority must be clearly recognized by all. Also make use of TV, radio and the web. Dedicated local leaders should also be responsible for getting the message out on public health from the central unit.
  • A lesson from the literature on famines is the role of democracy in helping to assure credible public information, to help avoid policy mistakes, and help assure a more rapid response. This crisis should not be an excuse for more closed and authoritarian rule.

Sharper trade-offs and harder constraints can emerge in poor places, with bearing on policy responses 

Isolation at home slows the spread of the virus, and reduces the strain on health systems. It also comes with a large cost, especially to poor people.

  • Public health systems are under strain even in normal times. Regular health shocks will go on. If anything, the case for social distancing and hygiene is even greater in poor countries.
  • At the same time, the welfare cost of isolation is also likely to be high. The world’s poor rarely have much in the way of a buffer that they can draw on; they have little savings, low food stocks; a heavy dependence on casual daily labor. The majority cannot work from home.
  • Food delivery systems will also come under stress, as is happening in rich countries. Supply chains will be broken.
  • So this is not solely a trade-off between economic welfare and health. It is also a trade-off between two aspects of health: illness due to the virus, and hunger and poor nutrition due to isolation and the disruption to markets and institutions, including private social protection.

Is the case for isolation weaker in poor countries? No. But the case for lockdown is weaker. Lockdown can pose new threats in some poor places.

  • Globally, poor people often have a crowded home environment. This is not just about crowded urban slums. Yes, overall population density is lower in rural areas, but rural people tend to be crowded into villages, often quite closely in multi-generational and sometimes polygamous families. There are ample opportunities for the virus to spread quickly in poor rural areas.
  • Some work environments may well be safer than being in a crowded home (as some observers have pointed out), but the worker already has the home exposure (at least for those who do not sleep at work). So this is not a valid argument against staying home while the virus threatens.
  • On balance, the case for isolation at home in this period for controlling the spread of the virus remains strong even in poor places.
  • However, we can make a distinction between social isolation practices in response to the virus and a strict lockdown, which could turn the virus response into a famine in some poor places. I do not say this casually; I believe it is a looming threat to be aware of now. The literature on famines has many examples of how the kind of institutional and market breakdown implied by a strict lockdown could cause local famines.
  • Lockdown also produces large short-term migration flows, as we have seen in India and Kenya. This threatens to spread the virus even faster, and especially into poor and vulnerable rural populations. We are likely to see outbreaks throughout the rural areas of countries that have imposed strict lockdowns.
  • The key reason for caution is that one is also locking down the markets and institutions that help protect poor people in normal times, and avoid poverty traps that may always exist, but are normally shielded from view. The risk is that one plants the seeds of longer term destitution, that may be hard to get out of once the threat of Covid-19 has largely passed.

Administrative and fiscal constraints loom large in poor countries.

  • Administrative capabilities and information systems tend to be weaker in poorer places, both poorer countries and poorer regions within countries. Many of the options being used or discussed for rich countries are simply not feasible in poor places. Adaptation to the administrative capabilities of poor places is essential.
  • Similarly, for fiscal resources; the constraints are severe. The rich world can help. Granted this is not the best time for that, but the moral and economic case is strong. Inaction by the rich world now could come at a huge cost later.
  • Of course, the rich world should not be making matters worse for the poor world, such as by banning the export of essential medical supplies.

A lot of what we routinely recommend for social protection in poor countries may need to be re-thought in this pandemic

Some of the classic policy prescriptions for dealing with other shocks don’t make as much sense during a  pandemic, and this is especially so in poor places.

  • This is no time to be worrying about “targeting the poor.” Fine targeting in poor places is over-rated in normal times. Now it makes little sense.
  • Nor should adverse incentive effects of welfare programs loom large. There have been exaggerated concerns here too in normal times. Do not worry about adverse incentive effects now. However, incentivizing better behavior, also recognizing the externalities involved, is worth thinking about.
  • Workfare versus transfers: Workfare is the classic policy response to recessions and famines. But workfare sites in poor places tend to be even more crowded that typical homes, even in poor places. The risk of catching the virus is likely to be greater on workfare sites.
  • On the other hand, one of the (often hidden) costs of workfare is the foregone earnings from other sources. These will tend to be lower in the pandemic, as economic activity is curtailed. The net gains to workfare participants may be greater.
  • On balance, however, the pandemic context is likely to tilt the balance against workfare in favor of transfers.
  • Countries such as India that have large rural workfare programs can use these to target transfers to poor families. All workfare participants in the last year (say) could be given a transfer payment, without (of course) any work requirement. This is starting to happen in India.
  • Transfers in cash versus food: Cash transfers are not be of much use if food markets and shops are closed. Direct food distribution may be needed. But food distribution will be a challenge. Some governments are making promises, but can they deliver? Include soap with the food package.
  • Conditions on transfers (cash or kind)? The common ones of keeping your kids in school do not apply. But a “stay at home most of the time” condition makes sense if it can be implemented in a reasonable way. Also, direct support to firms could be made conditional on not firing workers.

As in rich countries, a rapid, bold and forward-looking response is needed

A large, albeit time-bound, scaling up is needed of existing programs that are considered to work adequately. Anything less than an immediate fiscal transfer of something like 2% of GDP would probably be judged inadequate.

  • A rapid response calls for working with existing institutions and policies, though in a differentiated way, recognizing that some work better, and are more relevant, than others. Don’t wait for confirmed cases, or even symptoms; anticipate ahead, such as by closing schools and banning large events, as has happened already in some developing countries
  • Some modifications may be needed (such as removing work requirements, as noted above). But new programs take a long time to put into place. (This also illustrates the often heralded, but too little acted upon, advantages of having a good social protection system in place in normal times.)
  • Size of the economic shock? Even aside from the health threat, absent action, a month ago we were seeing estimates of an aggregate economic shock of about 2% of global GDP this year. I expect that would now be considered on the low side.
  • Scale of fiscal response? In a normal macroeconomic shock, we would expect sizable short-term multipliers given that so many economic resources are underemployed due to the virus. But those resources can’t be safely re-employed until the virus threat subsides. This is not about Keynesian multipliers.
  • At this point, the argument for a fiscal response in poor countries must be largely about current relief and avoiding even worse future outcomes, with the threat of famine coming on top of the virus. This is about getting money and food to poor and vulnerable people. As the virus threat subsides, the high propensities to spend of poor people will help the recovery. But the short term need is for helping to restore the food purchasing power of the world’s poor, without exposing them to greater risk from the virus.
  • As a rough rule of thumb—OK, more like a wild guess—I think that a near-term fiscal injection of transfers less than 2% of GDP should be judged as inadequate.

There is an important role for the private sector

The immediate health risk naturally take center stage, but getting people back to work is a high priority, and certainly no less for poor workers.

  • Rapid recovery for the private sector should anticipate that the threat from the virus will subside. Keeping now largely unemployed resources, including labor, in an “idling mode” makes a lot of sense. Instead of firing workers, try to keep them, maybe on reduced wages. Think of that as both insurance and investment.
  • The financial sector can play a positive role. Financial inclusion programs have helped expand banking coverage for poor people (though still a long way to go). This can facilitate rapid transfers, in partnership with public agencies.
  • There may also be some scope for adapting work to a more isolated, home-based, environments. Some encouragement/support from government might help to nudge the private sector.


After this paper, I am never going to look at a household Engel curve the same way

In the 1850s, Ernst Engel famously studied household budgets for 200 working class Belgian families, and found that the share devoted to food tends to decline with total household spending—a property that came to be known as Engel’s Law. Since then, “Engel curves” for budget shares have been widely used and much studied across the world, with near universal confirmation for Engel’s Law. Engel curves have found a wide range of applications, including in the assessment of policies related to agriculture, taxation, trade, industrial organization, housing, and in the measurement of poverty and inequality.

While there have been methodological and computational advances in the specification and estimation of Engel curves over the last 100 years or so, a common and persistent feature has been the reliance on household aggregates that Engel pioneered, along with a degree of imposed homogeneity in the Engel curves, allowing only limited variability in the parameters across and within households.

In a new paper, “Unpacking Household Engel Curves,” Philippe De Vreyer, Sylvie Lambert and  myself have studied some neglected but potentially confounding sources of heterogeneity in standard household Engel curves. Three sources are postulated.

First, there can be latent household effects on individual demand behavior. Members of a given household are not autonomous individuals who happen to be living together, but rather they come together selectively, and then interact and influence each other’s behavior through the process of consuming (and often working) together. While we may reject the unitary model, it can be expected that there are aspects of the household, and shared local environment, that can have a powerful influence on individual choices.  This can happen via individual preferences, which are to some extent formed within a household. Or it can stem from household- or location-specific aspects of the constraints on exercising personal preferences.

Second, there are differences in individual demand parameters within households. Engel’s Law may cease to hold at the household level when income gains are assigned to people with different consumption patterns and different preferences over how the extra money should be spent.

Third, there is heterogeneity in the extent of inequality within households. The existence of intra-household inequality is known to be a source of bias in the measurement of poverty and inequality. It is less well known that intra-household inequality can also bias estimates of empirical consumer demand functions, as invariably estimated from household aggregate data. Yet for many goods, and (hence) expenditures, there is a typically an unobserved individual assignment within the household, that may be a source of intra-household inequality, reflecting different reservation utilities outside the household. Furthermore, intra-household inequality can interact with individual parameter heterogeneity in influencing household demands, whereby greater intra-household inequality magnifies the effect of differences in preferences.

In our new paper, Philippe, Sylvie and I use an unusual survey for Senegal (which Philippe and Sylvie developed, in collaboration with others) that gives us a window on consumption distribution within the household. The families are typically multigenerational. Polygamous unions are common, with 25% of married men and 39% of married women engaged in such unions, which mostly comprise a husband and two wives.

Using these data, and with suitable modelling, we can unpack the traditional household Engel curve. In essence, what we do is estimate individual Engel curves (strictly, they are for sub-household units called “cells”) and then aggregate these up to the household level. We then compare the results we get this way with the traditional household Engel curve, pretending that we do not have the sub-household cell data.

We find that the traditional household Engel curve hides quite a lot about distribution within the household and preference heterogeneity, and these hidden factors are quite confounding about the true Engel curve. For example, intra-household inequality (not observed in standard data sets) surfaces in the error term of the traditional Engel curve. (To be more concrete, for those familiar with the literature on Engel curves, a form of the Theil index of intra-household inequality is found in the error term of the traditional Working-Leser household Engel curve.)

Two key lessons emerge. First, the (often-assumed) two-stage structure in bargaining-collective models of the household carries a testable implication with our data, namely that household spending should only matter to individual choices via the intra-household allocation of total spending. This exclusion restriction is generally consistent with our results. The exception is education spending, for which cell-specific budget shares are independently, and significantly, affected by the household’s overall standard of living. We suspect that this may be a “social effect” on cell Engel curves whereby the father exercises influence over the spouse(s) to spend more on his children’s schooling (including making conditional monetary transfers to), though some role may also be played by competition among the wives.

Second, our data reveal large biases in the standard household-level Engel curves. The sources of bias do not all go in the same direction; in particular, the bias associated with intra-household inequality tends to offset that due to latent heterogeneity in preferences. However, large net biases are indicated. For example, for the food share Engel curve, the coefficient on log total household spending is -0.11 using only household data but -0.28 when one estimates the Engel curve from the sub-household data and aggregates up to the household data. This is enough to reduce the income elasticity of demand for food (evaluated at mean food share) by one third, from 0.82 to 0.55.

In these data, we find that the bulk of the bias in standard household-level Engel curves is accountable to the influence of household fixed effects on sub-household Engel curves. The fact that the channel of bias via intra-household inequality partially offsets that due to the latent household effects in standard Engel curves implies that only adding controls to reflect intra-household inequality will tend to increase the bias in household-level Engel curves.

Given that we find that the bulk of the bias in standard household Engel curves is due to household effects on sub-household Engel curves, it may be expected that the most promising means of removing (or at least attenuating) the bias is to use longitudinal data, assuming that the confounding household effect in individual consumption behavior is time invariant. That conclusion is to be investigated further in future work.

What might a US-Mexico market for work permits look like?

In a new NBER Working Paper, “A Market for Work Permits,” Michael Lokshin and I have explored further the case for introducing a market in work permits (building on a shorter World Bank paper we wrote a few months ago). The policy idea we study is that working-age citizens in high-wage economies should be given the option of renting out their work permit—viewed as an asset of citizenship, though an asset that is not currently marketable. On the other side of the market, foreigners can buy (taxable, time-bound) work permits. The price of a work permit would balance supply and demand.

We argue that creating such a market would help capture the economic gains from freer migration, while keeping the host-country government in control of the migration flows, and aggregate labor supply. A minimum income can be assured for workers in host countries, financed by tapping into the currently unexploited gains from international migration. Thus, this market would offer a new instrument for social protection, as well as an efficient, growth-promoting, means of managing immigration, which would now be seen as an asset to workers in host countries, rather than a threat.

In the new paper, Lokshin and I elaborate further on the idea and provide illustrative calculations for the US and Mexico. We simulate the economic returns to sampled Mexican workers from migrating, and compare this to the likely costs, including the equilibrium price of the work permit, also allowing this to be taxed by the host country.

The results suggest that the missing market for work permits is large, with 18-36 million participants (depending on the chosen tax rate on work permits and other parameters for the costs of migration). For example, with a 10% host-country tax on the work permits and a 10-20% “remittance tax” on the US wage earnings of the Mexican migrants, the equilibrium price of the WPs would be about $20,000 per year, and around 30 million workers would participate. The US tax revenue would be around $300 billion, and the gain in earnings would represent about 6% of US GDP. The poverty rate in the US would fall by two percentage points, reflecting the pro-poor feature of the market’s implicit targeting mechanism.

Our simulations for the US and Mexico are only intended to be broadly indicative of likely orders of magnitude. More research is needed, including on the costs of migration and the spillover effects on the labor markets for workers who do not directly participate. Lokshin and I argue that further exploration of these and other issues discussed in our new paper is warranted, given the huge potential benefits of a market for work permits.


Lifting the floor to living standards by exploiting the gains from international migration

A recent paper with Michael Lokshin, “Market for Work Permits,” (with many updates to the earlier draft) proposed the creation of a market for work permits. This would allow citizens to rent out their right-to-work (RTW) for a period of their choice. On the other side of the market, foreigners can purchase time-bound work permits (WPs). This would help tap into the (potentially huge) unexploited gains from restrictions on international migration. Yet host countries would retain control over the flow of migrants and total employment. There are many gains to the host country, as discussed in the paper.

There is one aspect of the policy proposal that is worth considering. Under certain conditions, this policy will create a new binding floor to labor earnings in the host country—a new lower bound, above the current floor. The only estimate of the level of the income floor in America (averaged over reported incomes of the poor, with higher weight on poorer people) puts the floor at about $5 per person per day (Jolliffe et al., 2019). Allowing for (say) one dependent, this implies an income of $10 a day. It would be reasonable to assume that this is lower than the equilibrium price of a WP in our proposal. Indeed, $10 a day is lower than the minimum wage rate in the US for an eight hour day.

Workers in the host country will sell their RTW if they earn less than the going price in this new market (and some earning more than it will also do so if they experience a disutility of work). Similarly, foreign workers will only take up migration under this scheme if they earn something more than the going price  of WPs (sufficiently higher to cover costs of moving and any tax levied). This holds for all contracted time periods of the WPs. Thus, creating a market in WPs along the lines Lokshin and I suggest can be thought of as a new way of providing a guaranteed minimum income for each time period. And it is self-financing.

To better understand this argument, we can posit a first-best distribution in the host country that maximizes some weighted aggregate of utilities, with the weights reflecting the government’s social preferences. The first-best distribution of income is bounded below by some value, lets call it ymin. However, in the absence of this policy, the first-best is not implementable given other constraints (notably on information and administrative capabilities). Thus, the observed distribution has incomes below ymin due to uninsured shocks or longer-term disadvantage. With the policy in place, the host government can now solve for the tax rate on WPs required to assure ymin (as explained in the new version of the paper). Thus, the market for WPs now makes it feasible to implement the host country’s socially optimal minimum income.

There is another control available to the host country, namely its power over eligibility to purchase WPs, or sell the RTW. For example, the US might (initially at least) choose to make the market only available to citizens of (say) Mexico. Restricting migrant eligibility, or expanding eligibility to sell the RTW among citizens of the host country, will reduce the equilibrium price.

The big difference between these two policy instruments—the tax on WPs and eligibility conditions—is that the tax instrument can raise revenue, albeit at the expense of both citizens selling their RTW and foreigners buying WPs. It is reasonable to assume that the (positive) partial equilibrium effect of a higher tax rate on revenue dominates the (negative) effect stemming from the deterrent effect of a higher tax on migration. Then the host government faces a trade-off between the level of the income floor and the extra revenue generated by a higher tax on WPs.

Under certain conditions (explained in the paper) one can solve for the host government’s optimal tax on the new WPs, obtained by balancing its desire for revenue against its desire to implement its first-best level of the floor to living standards.

Jolliffe, Dean, Juan Margitic, and Martin Ravallion, 2019, “Food Stamps and America’s Poorest,” NBER WP 26025.

Social protection in rich countries: How to do better than a UBI (while also reaping the gains from international migration)

A Universal Basic Income (UBI) gives everyone the same transfer amount. Of course, the net benefit may not be uniform after the extra taxes, or spending cuts, used to finance the UBI. However, the question here is whether it is feasible to do better than a UBI—to assure that more goes to poor people who clearly need it more. There are many ways in practice of doing that—or at least trying to do so. The solutions proposed, or found in practice, vary greatly in their efficacy. Information and incentive constraints are known to loom large. (Incentive effects may well be less of a concern than information.)

In a new paper, “The Missing Market for Work Permits,” Michael Lokshin and I have argued that creating  a two-sided market in work permits would provide both pro-poor social protection in high-wage economies and new options for migration from low-wage economies. (A revised version is found here, with a fuller treatment of the literature.)

The idea is to create a market that helps capture the gains from international economic migration, while keeping the host government in control of domestic employment. An anonymous market exchange would allow workers to rent out their right-to-work (RTW). There is clearly much they could then do, including financing education or training, homecare of loved ones, or taking a long vacation. Simultaneously, someone else pays for a work permit (WP) and is then free to take up any job offer in that country.

A competitive market mechanism can be implemented (such as through a computerized double auction) to determine the market prices of these new WPs, conditional on the stipulated length of time and start date. Once that period ends, the seller gets back her RTW. The marketable WP is fully disembodied from the person selling it, and also independent of who is buying it. The market is anonymous.

Transaction costs would probably be low—almost certainly lower than for immigrant sponsorship schemes. The WPs could be taxed to finance the scheme’s costs, and (if desired) support other objectives. Development agencies and financial institutions could help applicants from developing countries, including in financing the costs of the WPs.

The currently missing market would no longer be missing. This can be seen as a social protection policy as well as an efficient policy for managing immigration, while capturing at least some of the (seemingly huge) economic gains from freer international migration. And freer migration would become a more popular idea—relieving public concerns by helping to internalize the externalities in host countries generated by migrants (or at least perceived to be). If the option of selling your RTW is confined to those in the workforce then aggregate labor supply would stay the same. A broader base of eligibility would allow rising employment. That is a policy choice.

Going back to the question I posed at the outset, our proposal will undoubtedly have a more pro-poor incidence than a UBI; specifically, it will bring both direct (first-order) gains to relatively low-wage workers who take up the option of renting out their RTW—a “self-targeting” mechanism—and indirect gains to others via the likely tightening in the low-wage labor market.

We probably can do better than a UBI, which can be a rather blunt instrument. For example, a UBI has been advocated as a means of addressing job-loss due to automation. But why would one give the transfer to everyone, including those who stay working? Our scheme would directly help those who lose their job due to automation.

Also, unlike a UBI, it is self-financing. This overcomes a widespread concern about UBI proposals that require higher domestic taxes or are only available as an option to existing welfare programs, thus reducing the net gains to poor people from the UBI.

Time to rethink global poverty measures

Today we find two main approaches to measuring poverty and monitoring progress in reducing it. The first focuses on “absolute” measures that strive to use poverty lines with constant real value. For example, this is essentially what the official poverty measures for the US strive to do. It is also how the World Bank measures global poverty, aiming to apply a “rigidly unchanged” real line across countries as well as over time.

The second approach uses “relative” measures for which the poverty line varies in real terms, being set at a constant proportion of the current mean or median—an approach that emerged in the 1960s and became popular in Western Europe in the late C20th. There has been much debate on the choice between absolute versus relative measures.

In a new paper, “On Measuring Global Poverty,” I argue that neither approach makes economic sense. A new approach is needed for measuring and monitoring global poverty going forward.

The nub of the problem is that existing poverty measures tend to opt for one of two very different assumptions, neither of which can be seen as acceptable any longer:

  • that relative income does not matter to economic welfare, or
  • that relative income is all that matters.

When applied globally, the fixed real line cannot capture relative economic deprivation at country level or the need for higher outlays for economic well-being in richer countries—a higher cost-of-living not reflected in the existing Purchasing Power Parity (PPP) rates. However, it is no less obvious that the absolute standard of living, at given relative income, also matters, thereby ruling out measures in which the poverty line is set at a constant proportion of the mean or median.

The new paper proposes a welfarist interpretation of global poverty lines, which is augmented by the idea of normative functionings, the cost of which varies across countries. In this light, current absolute measures are seen to ignore important social effects on welfare, while popular strongly-relative measures ignore absolute levels of living. It is argued that a new hybrid measure is called for, combining absolute and weakly-relative measures consistent with how national lines vary across countries.

Illustrative calculations indicate that we are seeing a falling incidence of poverty globally over the last 30 years. This is mainly due to lower absolute poverty counts in the developing world. While fewer people are poor by the global absolute standard, more are poor by the country-specific relative standard. The incidence of purely relative poverty has been rising in the developing world, as can be seen from this graph. The vast bulk of poverty, both absolute and relative, is now found in the developing world.

Global poverty count

Review of Tony Atkinson’s last book

It is a sad task to be reviewing the last book by Anthony (Tony) Atkinson, who passed away in January 2017, before the book could be completed. Shortly before his death, he asked two of his past collaborators, Andrea Brandolini and John Micklewright, to bring the book to publication under the title Measuring Poverty around the World .

Throughout his career, Tony Atkinson bridged his considerable technical skill as an economist with a commitment to rigorous thinking about distributional measurement and policies. He combined deep scholarship with social concern. It is a combination that has long made him a role model for all those who seriously study, and care about, poverty and inequality, and social issues more broadly.

Here is my review: Review of Atkinson’s Measuring Poverty Around the World.

Enter your email address to follow this blog and receive notifications of new posts by email.