Skip to content

“Empirically sufficient and empirically necessary” – Lant Pritchett on economic growth as the (only) key to poverty reduction

Source: Lant Pritchett – Poverty Reduction and Economic Growth (



We show that any cross-national measure of human material wellbeing that is

(i) basics (not luxuries),
(ii) general (uses indicators from multiple domains), and
(iii) plausible (uses any defensible choices for weights)

will have a statistical relationship with country GDP per capita with four features.

First, the relationship will be strong, with nearly all cross-national variation in basics associated with variation in GDPPC. Second, the relationship will be non-linear, with a stronger elasticity of basics to at lower than higher levels of GDPPC. Third, GDPPC is empirically sufficient: no country has high levels of GDPPC and low levels of basics. Fourth, GDPPC is empirically necessary: no country has high levels of the basics at (very) low levels of GDPPC.

These findings extend the existing literature as, while there are large literatures showing a strong connection of GDPPC to money-metric measures of wellbeing (e.g. headcount income/consumption poverty (Pritchett 2020), inequality adjusted incomes (Dollar,Kleineberg, and Kraay 2015)), or, to direct physical measures of specific dimensions of wellbeing like health, nutrition, education, access to sanitation, or indices like multidimensional poverty, our argument is that all plausible, general, measures of the basics of human material wellbeing will have a strong, non-linear, empirically sufficient and empirically necessary relationship to GDPPC.


 In 1988 Robert Lucas said “the consequences for human welfare” of differences in economic growth “are simply staggering,” so much so that, once one starts, “it is hard to think about anything else.” But something strange has happened in the field of development: the importance of sustained economic growth for human wellbeing is being actively downplayed and development economics seems taken up with “anything but” economic growth.

In a February 2021 blog the executive director and the communication director of J-PAL (Abdul Lateef Jameel Poverty Action Lab) made the (outlandishly and obviously false) claim that: “for millions of people living in poverty, growth is not enough. Specific, targeted social programs based on rigorous empirical evidence are equally important to prevent people from being left behind.” (emphasis added).

When Bill Gates argued that anti-poverty programs providing chickens would be an important avenue for reducing poverty in Africa, Professor Chris Blattman responded not with the corrective that without sustained broad based growth such programs would do little to reduce poverty, but rather that he thought: “the best investment we could make to fight world poverty” (emphasis added) would be randomized studies comparing the efficacy of programs that transferred livestock assets like chickens versus those that distributed cash.

Even development think tanks whose name includes “growth” are skeptical: Rohini Pande the director of the Economic Growth Center at Yale University in October 2021 has a blog published by the International Growth Centre at LSE titled: “not by growth alone” (emphasis added, twice).

Claims that economic growth is not central to poverty reduction are just wrong (Dollar and Kraay 2002, Sala i Martin 2006, Dollar, Kleineberg, and Kraay 2016, Roser 2021). Pritchett (2020) shows that growth in median incomes is strongly empirically sufficient for poverty reduction: growth is enough as enough growth alone does (roughly) eliminate extreme poverty. The relative contributions of levels of income and targeted social programs to levels of poverty is not “equal” as J-PAL, with zero evidence, asserts (50:50), it is more like 99:1 in favor of growth. Moreover, it is also well-known that important elements of wellbeing, like child mortality or life expectancy, are strongly associated with GDPPC (Filmer and Pritchett 1999, Pritchett and Viarengo 2010). And, it is also easy to show that some cross-national general indicators of the basics of human wellbeing, like the Social Progress Imperative’s Basic Human Needs measure, are strongly related to GDPPC (Pritchett 2022).We up the ante and demonstrate that the skeptics about the benefits from economic growth are not just wrong poverty or about this or that indicator, they are wrong about everything, or at least wrong about everything basic to human material wellbeing (hereafter rather than use the extended phrase or invent an acronym BHMWB we just say ‘basics’).

At least since Amartya Sen’s (1985, 1999) arguments for a “capabilities” approach and the UN’s Human Development 1991, many have agreed that normative measures of country development as human wellbeing should go beyond “money metric” measures and include outcomes in health, education, nutrition, access to water and sanitation, the natural environment (and others). The Sustainable Development Goals (UN 2015) include a very wide array of goals and targets.

We argue the relationship between any general, plausible, measure of the basics of human material wellbeing and GDPPC will have four characteristics:

(i) is very tightly correlated with GDPPC. The correlation between a very wide array of measures of basics and their “predicted” value based on GDPPC alone is typically around .9 (equivalently, an R2 around .81). The p-levels for zero on all terms in GDPPC are smaller than 10-37. Even “data undermining” which does specification searchers over the space of possible measures of basics to minimize the correlation of basics with GDPPC still produce strong correlations,

(ii) the relationship of basics and GDPPC is strongly non-linear, more non-linear than logarithmic and not constant elasticity. The elasticity of basics wrt to GDPPC is between twice and eight times larger for countries in the second quintile income than for those in fourth quintile,

(iii) the statement “growth is not enough” is demonstrably false for basics. GDPPC is empirically sufficient: while at any given level of GDPPC there are countries with higher or lower values, there are no countries with high GDPPC that have low achievement on basics.

(iv) GDPPC is also empirically necessary: there are no countries with low levels of GDPPC with high levels of basics.

After a first section that describes the underlying data sources the next four sections describe various approaches to creating a measure of the basics of human material wellbeing and the relationship of each of those approaches to GDPPC…

Lant Pritchett – Economic growth is enough and only economic growth is enough
Country score on Legatum Prosperity Index elements / indicators mapped against GDP per capita. Source:

1 thought on ““Empirically sufficient and empirically necessary” – Lant Pritchett on economic growth as the (only) key to poverty reduction”

  1. An interesting return to classical economic development thought: it’s the rising water lifts all boats hypothesis championed domestically (USA) in the 1980’s. Hard to dispute, but also really hard often to incentivize via local government action.

I'd love to hear your thoughts and recommended resources...