Professional Documents
Culture Documents
Macro part
Week 1
Measurement
• How to measure development?
• Most common standard: levels and growth of income/production per capita
• Development ≠ income per capita growth
• Health improvements (nutrition, clean water, sanitation, health services), education (access
schooling, literacy), material well-being, political rights and freedoms, environmental improvements
• Last few decades focus on income criticized too narrow
• Per capita growth not automatically leads to development sometimes no connection
• Still per capita GDP fairly good proxy
• Ultimately rising income levels will lead to development (growth is good for the poor)
x-axis: average annual growth in per capita income
y-axis: average annual growth in per capita income, poorest quantile
positive slope, poor is also very vulnerable for declines in income.
Important development index constructed by UNDP for all countries: Human Development Index
(HDI)
• 3 elements HDI:
– life expectancy at birth (reflecting i.a. infant and child mortality)
– educational attainment (weighted index of adult literacy 2/3 and enrollment rates 1/3)
– per capita income
• HDI takes simple average, weights arbitrary Nevertheless HDI often used
• High correlation GDP per capita and HDI (logical because GDP per capita income is a component of
HDI)
• High correlation individual indicators and HDI (like life expectancy, infant mortality, literacy)
• Income and life satisfaction high (positive) correlation
So, GDP per capita not perfect, but good proxy for development
• additional measurement issues:
• GDP measures both total output and income
• Total value of final goods and services produced in a country during a given year
• With trends over time, real GDP, income adjusted for domestic price inflation
• To compare per capita income across countries, we may use exchange rate method
• Convert income in local currencies into common currency (US$?) and then divide by population
(per capita)
• But cross-country comparisons of per capita income using exchange rate method can be biased for
three reasons:
1. Underreporting of income in developing countries (tax evasion, subsistence production for
own consumption)Example India 18-21% of income unrecorded. Cambodia 1/3 of labor force
unpaid family worker
May overestimate growth
Correct through imputations.
2. More serious bias, non-tradable goods not reflected in exchange rates, reflect prices tradable
goods (tradable goods prices equalized through arbitrage). Non-tradable goods lower prices
in poor countries, therefore conversion to US$ using exchange rates underestimates real
incomes in poor countries.
Correct for by using PPP dollars (expresses all non-tradable goods (use average price of
haircuts in all countries instead of4$ in India and 45$ in US)
Summers-Heston dataset PPP-based country incomes since 1950, also called Penn World
Tables
3. Other subtle measurement problems:
a. Distorted pricing: GNP measurement uses market prices, assumed to reflect people’s
preferences. But, not all markets perfect competition, neither fully flexible prices, price
information sometimes absent or not nationally representative.
b. Externalities: Prices may not reflect true cost, e.g environmental damage, resource
depletion
Correct by using shadow prices
• Growth rates accelerated since 1820, especially after 1880, and again since 1950
• Striking, richest regions recorded highest growth rates, poorest regions slowest, at least until 1950
• 1950-2001 patterns of economic growth changed:
– Asia, then poorest region recorded fastest growth, narrowing gap with richest
– Widened gap W. Europe-W. offshoots narrowed
– LA and Africa growth stagnated in 80s and 90s, resumed mid-1990s
– Eastern Europe collapsed after fall Berlin Wall 1989, resumed mid1990s
– Since 1950: end of WWII and end of colonial era
• Although we see clear regional differences, within regions different country experiences
negative growth, slow growth, moderate growth, rapid growth (see table)
Harrod-Domar
• Harrod- Domar model emphasizes the role of saving for economic growth
• Firms produce goods which creates household income
• both consumption and capital goods produced
• Part household income used to buy consumption goods, part of it is saved (for future consumption)
• Firms use these savings to buy capital goods, investment
• Cycle of production, consumption, saving, and investment
• Macroeconomic balance if saving = investment
• The modeI:
θ∗Y t+1 P t +1 Y t +1 P t +1
* = (1-) θ*yt + s*yt = yt+1 and = 1+n
Pt +1 Pt P t +1 Pt
θ∗ y t+1
* (1+n) = (1-) θ + s
yt
y t +1
= (1+g*)
yt
θ(1+g*)(1+n) = (1-) θ + s
Devide by θ
s
(1+g*)(1+n) = (1-) +
θ
s
(1+ n + g* + g*n) = (1-) +
θ
Ignore g*n
s
1+ n + g* -1 + =
θ
s
n + g* + =
θ
• Per capita Harrod-Domar equation (10) gives per capita growth rate for given values of saving rate,
capital-output ratio, population growth and depreciation rate
• Shortcoming: many of these parameters are endogenous, determined by growth as well
• For instance savings can be endogenous, influenced by the level of per capita income in a country
• As incomes change, s that enters into the Harrod-Domar equation will change
• This feedback mechanism is missing
• Important, because can cause development trap: low s <-> low y
• Also population growth may be endogenous, vary with per capita income
• Demographic transition
• May cause development trap For low-income countries:
Increasing per cap. income -> higher population growth
Higher population growth -> lowers per cap. income (for given g) divided by more n
• Harrod-Domar model emphasizes the role of saving (and investment and capital accumulation) and
implies saving is sufficient for sustained growth s/θ = g * +n+ (higher s -> higher g and g *)
• The model doesn’t address endogeneity of s and n
• The model doesn’t address changes in technology/productivity (are important for g)
• Also model assumes fixed capital-output ratio θ while marginal product of capital declines as more
capital is accumulated -> capital-output ratio endogenous
Solow model
• Solow model makes the capital-output ratio θ endogenous.
• Endogenous capital-output ratio θ :
• This ratio is low for poor countries (labor abundant and capital scarce), because MPC is high if little
capital is used.
• As countries accumulate more capital and income per capita grows, MPC declines (each additional
unit of K produces less additional Y, so Y/K ↓), so the capital-output ratio θ = K/Y ↑.
• => K/Y (= θ) determines growth, but is in turn also affected by growth
• Endogeneity of capital-output ratio is incorporated in Solow model through production function
with diminishing returns to capital (diminishing marginal product of K).
• We will see this results in long-run equilibrium (steady state) where growth in per capita capital
and per capita output/income is zero
Y K Y
Slope / =
L L K
Θ = K
Y
• So, as capital per capita increases output-capital ratio Y/K ↓ (or θ = K/Y↑) because relative
shortage labor
• As noted above, production function with diminishing returns to capital is combined with second
key equation of the Solow model, the capital accumulation equation
• We then write (4) Kt+1=(1- )Kt + sYt in per capita terms first by dividing through population Pt and
then assume that population grows at rate n: Pt+1= (1+n)Pt next, substituting Pt+1/ Pt by (1+n) gives per
capita form of (4):
K t +1 P t +1 Kt Yt
= (1 – 𝛿) +s <=>
P t +1 Pt Pt Pt
y = Akα
y* is constant Y/L
k* is constant in the steady state L is growing at
rate n, and K is also growing at rate n
• In the graph, starting from any value for kt , you can find kt+1 on the concave curve. Then moving
horizontally to the 45-degree line, you can find the position of kt+1on the horizontal axis, which you
can then use to find the level of kt+2 on the concave curve. Keep repeating to find the dynamics of k
and thus of y (remember k and y move together).
• Graph shows if per capita capital is low, per capita capital grows fast, because output-capital ratio is
high (high marginal product of capital). Relatively high additional output per capita means relatively
high additional saving and investment per capita (large enough to more than compensate for n and
), thus increase in k is large.
• If k grows fast, y grows fast
• Starting on the right side of k* , the marginal product of capital has become so small that not
enough to compensate for n and , so per capita capital stock k declines until you end up in k * .
• Economy thus converges to the equilibrium, the steady state, where k is constant at k*
• Also income per capita y is then thus constant! No long-run per capita income growth due to
saving
• Implies that K and Y grow at population growth rate n in steady state (so total K and Y do grow)
• So, to the left of k* , capital is accumulated faster than labor, which is why you get diminishing
returns to capital and the growth process slows down.
• Income per capita y is constant in the long run in Solow model savings cannot sustain per capita
income growth. Contrast with Harrod-Domar model, where savings affected the per capita income
growth rate.
• Difference is exactly due to diminishing returns to capital (and thus increasing capital-output ratio)
in Solow, and constant capital-output ratio in Harrod-Domar.
• Savings rate in Solow doesn’t influence long-run per capita growth rate (which is zero in this
model), but it does affect the long-run level of per capita income.
• Illustrate increase in s, level-effect in steady state (and growth effect only during transition to new
steady state).
• Also illustrate increase in n or , increase in either implies you need higher s to keep k* constant.
• Distinguish:
– Per capita growth of capital and income in steady state
– Per capita growth of capital and income in transition towards steady state
• Solving for the steady state:
• Use per capita capital accumulation equation
(5) (1+n)kt+1=(1- )kt + syt
• Plug in k* for kt and kt+1 to obtain equation describing steady state:
• k * /y* = s/(n+ )
• s↑, n ↓, and ↓ increase capital-output ratio and thus also k* must be higher (and y* )
• Increase in A shifts production function upwards and increases per capita income in steady state,
so continuous increase in A (=continuous technological growth) can sustain per capita income growth
curve shifts upwards (see notes)
• Introducing units of effective labor
L’(t)=E(t)*P(t) and expressing the model in these L’(t) instead of P(t), you find that in steady state
growth of capital per unit of effective labor L’(t) is also zero, but because efficiency E(t) grows with
rate 𝜋, also capital per capita has to grow with rate 𝜋. Story is rather similar to basic model. Use
K
^k = instead of capital
E∗P
K
per person k=
L
L’= E*P
𝜋 growth rate of the
efficiency term
Et+1 = (1+ 𝜋) Et
• Illustration of Solow model with China and US. What does the model predict about their steady
states and thus about their long-run per capita income?
• Look at difference in savings rate and level of technology
• So, theories of economic growth take us quite far in understanding the development process, at
least at some aggregate level
• Both the Solow and the Harrod-Domar model put saving (and investment and capital
accumulation) at core of growth process.
• In Harrod-Domar this is sufficient to sustain growth in per capita income, in Solow not
• Extended Solow model emphasizes technological growth for sustained growth of per capita
income
• Both models include exogenous variables (s, n) that may be endogenous and cause development
traps
Week 2
• Overall global population will increase strongly. UN forecasts of future pop growth. All scenarios
similar assumptions declining mortality, different assumptions fertility declines (due to economic
improvements, better education, family planning programs).
• Higher population growth in developing countries, low in developed countries also implications for
age distributions and again for population growth.
• Relatively high population growth in developing countries is partly result of very sudden and
widespread decline in death rates
• Medical innovations (antibiotics, use of pesticides to reduce malaria), elementary methods of
sanitation and hygiene etc. did not have to be reinvented, widespread application
• Decline in death rates took about 300 years Europe and North America, while in developing
countries only several decades
how quickly will birth rate follow death rate in downward course?
• Adjustment birth rates: various factors keep birth rate high:
• At macro-level, large share population in reproductive age keeps birth rates high even if fertility
rates are reduced
• At household level, children are substitutes for missing markets (mostly social security in old age,
but also insurance: medical care, life insurance, loss of employment, disability)
• Number of children is chosen with intention of receiving support at old age, which may not be
forthcoming because (1) child may die, (2) child may not earn enough income, (3) child may
deliberately not support its parents. More children increases probability of support.
• Falling death rate reduces (1), but not (2) and (3), which may well increase
• Gender bias may further increase fertility (when a daughter is born, they want a son get more
children)
1-(1-p)n > q
p = prob. that a given child will support you
1 – p = prob. That a given child will not support you
(1-p)n = none of you children will support you
1- (1-p)n = prob. that at least one of your children will support you
q = threshold prob. of receiving support of at least one child that is found acceptable by a couple
f.e. if (1) reduces p↑ n↓
Ad 1. Demographic dividend:
• The temporary increased population growth produces a boom of children that will move through
the age distribution.
• First, boom concerns children, then 15 to 20 years later they become the economically active
population (age 15- 65) and finally they are the elderly.
• So first increasing youth-dependency, then declining youth-dependency and then increasing
elderly-dependency.
• First, the effect on economic growth is negative (more children), then it becomes positive (more of
working-age), and finally it may become negative again (more elderly).
• Middle phase is called the demographic dividend.
demographic dividend is not automatic
Depends on the implementation of effective policies (improving health and education of working-
age population to improve their productivity, employment creation).
Historical evidence:
• Industrial countries as a group received a dividend of 1 percentage points.
• Cross country growth regression by Bloom and Williamson (1998) finds that 1/3 of the ‘economic
miracle’ in East Asia attributed to demographic dividend
• Relates to growth accounting exercise by Young (1995) for Asian tigers (1965-1990), also included
correction for increasing participation rates.
• Recent IMF study suggest developing countries in total as result of rising share working age pop will
enjoy “demographic dividend” resulting in stronger economic growth next 20 to 30 years
• Especially SSA, potentially largest demographic dividend
• In Africa, since 1985 dependency ratio falling, mirrors Asia, 20 years earlier dependency ratio falling
(in Asian tiger countries)
• Dependency ratio = ratio of young plus elderly to those in the labor force
Aid can also help to push the economy beyond the threshold to prevent poverty trap
• Although aid not always worked well, overall record positive, critical for poverty reduction and
growth many countries, helped prevent worse performance in others
• Argue many weaknesses aid more to do with donors than recipients, substantial aid for political
reasons, so not surprising not always effective fostering growth
• Point to successful recipients Botswana, Indonesia, Korea, Taiwan and more recently Ghana,
Mozambique, Uganda
• Several successful aid-financed initiatives: green revolution, campaign against river blindness
Main proponents: Jeffrey Sachs, Joseph Stiglitz
• Arguments opponents: Aid no effect on growth, may undermine:
– Aid invites corruption (ends bank accounts, cars)
– Entrenches corrupt dictators (supports political regimes that impoverish the poor)
– Aid is wasted, large bureaucracies, TA reports, new projects, no maintenance
– Undermines incentives for private sector activity, e.g. Dutch Disease, real appreciation exchange
rate undermining export, drawing workers and investment from productive activities
– Aid may reduce government investment, aid is fungible, effectively finances activities not intended,
e.g. aid used for investments government would have made without aid, resources freed up for other
purposes
• Cite widespread poverty Africa and South Asia despite decades aid
• Call for aid programs reformed, curtailed, or eliminated altogether
• Point to countries high amounts aid, disastrous growth: DRC, Haiti, Sudan
Main opponents: Dambisa Moyo, William Eaterly
Aid recipients:
View 1
Although not always successful, on average aid has had a positive impact on economic growth
• Majority of studies find positive relationship aid and growth, after controlling for the impact of
other factors on growth
• Since mid 1990s, aid-growth studies began investigate whether aid spurs growth with diminishing
returns. Small amounts of aid, large impact, each additional dollar less effect. Standard assumption in
growth theory, but earlier research only tested linear relation
• Aid can also have a positive effect on other development objectives that affect growth only
indirectly, such as better health and education. Large share of aid is allocated to health and
education . Impact on growth can take long to materialize.
• Micro studies find strong impact, especially on health & education
View 1
View 2
View 3
View 2
aid has had little effect on growth
• Especially early studies, not controlling endogeneity, finding simple negative correlations
• Also early studies testing only linear relation
• Health and education effects not yet apparent in old studies
• But also some more recent studies no effect
View 3
Aid has a conditional relation with growth, stimulating growth only under certain circumstances
• Tries to explain why aid effective in some cases, not in others
• Three conditional explanations emerged:
– Characteristics of recipient country (quality of institutions)
– Types of aid (emergency and humanitarian aid, negatively associated with growth, aid
affecting growth after very long time (health, education), aid directly aimed at affecting
growth, strong positive aid-growth relation)
– Donor practices (e.g. bilateral vs multilateral aid, tied vs. untied aid, donors that coordinate
with others vs acting on their own, donors with effective monitoring/ evaluation)
• Nowadays: increased use of RCTs (Randomized Controlled Trials) to evaluate direct impact aid
intervention, rather than impact on economic growth.
• Impact of micro-interventions (especially health and education) tested with treatment and control
group
Week 3
Convergence
• Prediction one could make from Solow model under certain assumptions about poor and rich
countries’ predicted growth rates
• Two types:
– Unconditional convergence
– Conditional convergence
Unconditional convergence
• Implies per capita incomes of countries move ever closer, converging to same steady-state • Initial
conditions k0 and y0 don’t matter in the long run given steady-state k* /y* = s/(n+ + 𝜋)
• kt k* from any k0
• due to diminishing returns to capital…
• … and the equality of s, n, 𝜋 across countries
• Assertion is poor countries will grow faster
• Assertion poor countries will grow faster implies negative relation initial per capita income and per
capita income growth
• Does this assertion fit the data?
▪ Option 1. Small number of countries, long horizon
▪ Option 2. large number of countries, short horizon
Conditional convergence
• If countries have own steady state, then not necessarily that poorer countries grow faster
• Suppose e.g. similar technological progress 𝜋 across countries, but different s, n, and
• So, growth rate of k and y in steady state is similar (𝜋), but
levels different, due to s, n, and
• Then initially wealthier countries may grow faster, see
graph
Convergence condition on and conditional on openness, savings, and
population growth population growth
• How do we control for effect of s, n, A, 𝜋, 𝛿 on position steady state? need relation between per
capita income and these parameters
• Based on Solow model and using a Cobb-Douglas production function, following relationship
emerges between per capita income and parameters (3.15)
α α
𝑙𝑛𝑦(𝑡) ≈ 𝐴 + ln 𝑠 − 𝛼 ln(𝑛 + 𝜋 + 𝛿)
1−α 1−α
• This relationship can be tested with data
• 𝛼 ≈1/3 such that 𝛼/(1-𝛼)≈ 0.5 𝛼 = share of capital income in total national income
• Mankiw, Romer and Weil (QJE 1992) did this using 𝜋 + 𝛿 ≈ 0.05
• More than half of world variation in 1985 per capita GDP can be explained by variation in s and n
strong result
• They do find positive coefficient for ln(𝑠) and negative coefficient for ln(𝑛 + 𝜋 + 𝛿), but too large,
not close to 0.5
• Observed variations in per capita income are thus too large relative to those predicted by theory,
too little convergence (so maybe less diminishing returns to capital? But constant returns to capital
simply not true)
• So, once we drop assumption that all parameters are same across countries, Solow model does
make some sense of the data, some evidence of conditional convergence, but still too little
convergence
Growth Accounting
• In growth models, output/income can grow for three reasons:
– Growth in capital input (by saving/investment)
– Growth in labor input
– Technological progress
• So far, we have identified these as core determinants of growth
• How important are each of these in explaining growth?
• Solow pioneered early efforts to quantify contribution of each of these proximate causes
• This approach is more of an accounting framework based on actual data than economic model
growth accounting
• Seeks to find proportions of economic growth we can attribute to growth capital stock, growth
labor force and TFP growth
• 𝑌 = 𝐴𝐾𝛼𝐿1−𝛼 𝑔𝑦 = 𝑔𝐴 + 𝛼𝑔𝐾 + (1 − 𝛼)𝑔𝐿
• Economic growth, growth capital stock and growth labor force can be measured, TFP growth
measured as residual (Solow residual)
• Limits to this analysis: TFP is measured as residual, will include combination of influences that
analysis cannot disentangle
• TFP growth due to faster computers, new seed varieties agricultural crops, other technology?
• Or efficiency gains from improved trade policies, reduced corruption, less red tape?
• Growth accounting cannot answer these questions, part of growth that is not explained by the data
on accumulation K and L measure of ignorance
• Residual also includes measurement error
• Sources of growth analyses have been carried out for many countries
• Solow’s initial study on US attributed 88% of growth to TFP growth, only 12% to increased capital
per worker
• Subsequent studies included better measures of inputs (including skill categories, quality of capital)
• For industrialized countries, particularly rapid growers, increases in capital per worker frequently
account for less than half increase in output, while TFP growth for bit more than half.
• Was surprise, as most basic models put capital formation at heart of growth
• Similar studies for many developing countries; generally these attribute a larger role to capital
formation (main source of growth)
• This is consistent with Solow model, as developing countries have lower levels of capital per worker
and can increase incomes (converge) through investment
• TFP growth tends to become more important as income rises
• Example: economic growth East Asia 1965-90 higher than any other region in world history
• East Asian Tigers: Japan, Hong Kong, Korea, Taiwan, Singapore, Indonesia, Thailand, Malaysia
• Clues to explain their success? Policy implications
• Countries had high accumulation of physical (s↑) and human capital (especially secondary
enrollment ↑)
• TFP growth according to World Bank publication very high (explaining 1/3 growth, high relative to
other developing countries), thus productivity-based catching up due to openness
• But, Young (1995, QJE) did more proper growth accounting exercise, accounting for all changes in
inputs
• Since 1960s, several developments have contributed to high growth rates
– Participation rates have risen
– Labor moved from agriculture to manufacturing
– Education levels improved
– Capital has been accumulated faster (I/GDP investment ratio)
• Is there a role left for TFP-growth, residual?
• For Singapore, all growth can basically be attributed to accumulation of capital and (to lesser
extent) labor
• For other East Asian Tigers, similar story holds
• Conclusion: although there has been productivity growth in East Asia, this has not been
exceptionally high. Factor accumulation has played most important role.
• Broad clues from this empirical research about why some countries grow faster
• Most rapidly growing developing countries tend to share 6 characteristics (e.g. Sala-i-Martin, AER
1997):
1. Macroeconomic and political stability
2. Investment in health and education
3. Effective governance and institutions
4. Favorable environment for private enterprise
5. Trade, openness, and growth
6. Favorable geography
6. Favorable geography
• Poorest countries are almost all in tropics, richest countries in more temperate zones
• Even within temperate zones, regions closer to tropics tend to be less well off
• Strong relation location and growth
• Tropical countries greater burden diseases, erratic climate, some areas poor quality soil
• Landlocked countries and small-island states isolated from major markets (overland transport costs
can be three times higher)
• Being landlocked doesn’t mean growth is impossible, but adds to production costs (-> improve
infrastructure)
In sum
• Growth accounting indicates capital accumulation contributes substantially to growth in low-
income countries, while increase in size and quality of labor force also important contribution
• Productivity growth tends to account for a larger share of growth in high-income countries
• Our understanding of precisely how these and other factors affect the growth process is far from
complete.
Graph:
Linear axis: emphasizing high global income
inequality (distribution stretches out far to the
right, much further than cutoff at 14,000
intern.$)
• But inequality declining, Gini coefficient from
68.7 (2003) to 64.9 (2013)
• Also rise of global median income from
1,090 int.$ (2003) to 2,010 int.$ (2013), almost
doubling
• Changes in total global income inequality driven by changes in inequality between countries and in
inequality within countries.
• Inequality between countries declining since 1980s
• Inequality within countries increasing since 1970s ‘
• Trend in global inequality now driven mostly by changes inequality between countries, because this
inequality higher than within-country inequality.
• Within-country inequality (Gini) over last three decades increased in many countries, especially in
advanced countries and Asia and Pacific.
• In Latin America and Sub-Saharan Africa decline, but high levels
• Also, rising share of top income earners in several countries.
Poverty
• Share of population in poverty declined since 1820.
• World population living in extreme poverty fallen from around 90% in the 19th century, to 9.6% in
2015.
• Extreme poverty: < 1.90 $ per day (in 2011 intern. $)
• This continuous decreasing share of people in extreme poverty over last two centuries one of most
remarkable achievements of humankind.
• Especially the last 4 decades very steep declining share.
• Number of extreme poor until 1970 did increase (despite declining population share), but since
then also number of extremely poor decreased (despite strong
population growth).
• Also using higher poverty lines (more dollars a day), the
number of poor has been declining in recent decades. (see
graph distribution of popultaion between different poverty
thresholds)
• Decline, yet still large share of world population below 10 int.
$ per day (from 75% to 66%).
• Striking that perception especially in rich countries (graph UK) is share of poor increased.
only 12% decreased
55% increased
• Concentration of world’s poorest shifted from East Asia in 1990s to South Asia in the 2000s to
SubSaharan Africa now.
• Steep decline poverty during last generation because the poorest people lived in countries with
strong economic growth, large part in Asia, but also in countries like Ethiopia and Ghana.
• This is now different. Many of world’s poorest now live in countries with low economic growth in
past. Half a billion face the prospect to remain stuck in extreme poverty.
• Goal number one of Sustainable Development Goals (SDGs), agreed on by all nations, is
“eradication of extreme poverty for all people everywhere” in 2030.
• Although projections show a further decline in extreme poverty, world not on track to end extreme
poverty by 2030.
• Number of people in extreme poverty will likely stagnate at almost 500 million in 2030, mostly in
SSA.
• Big lesson from history of extreme poverty is that it is ec. growth of an entire economy that lifts
individuals out of poverty. Micro-level interventions (short-term support households with productive
asset grant, consumption support, skills training, health education etc.) also help.
• In 2013 (2015) still 746 (705) million people (10.7%) in extreme poverty.
• Africa is now region with largest number people in extreme poverty (in 1990 this was Asia).
• India (in 2013) the country with the largest number of people living in extreme poverty (218 million
people). Nigeria and the Congo (DRC) follow with 86 and 55 million people.
• Breakdown by (region &) country:
• There are very large differences in income per capita across countries today
• Growth China/India: millions moved above $1/day
• Other countries have seen their poverty rate increase over time
• Why do some countries fail to improve their technologies, invest more in physical capital, and
accumulate more human capital?
• Capital, human capital and technology are only proximate causes of economic growth and
economic success they are endogenous variables in the model (and not exogenous variables)
Institutions:
• Economic institutions: economic rules of the game (property rights, contracting institutions)
• Political institutions: political rules of the game (democracy vs dictatorship, electoral laws,
constraints)
• Socials institutions: religion, social norms, culture
Exogenous in the short run, endogenous in the long run
• Do Institutions change slow enough so that they can be considered exogenous, and thus an
explanation of growth differences?
• Possible mechanisms:
Property rights land Less worry about losing land More investments in land higher
productivity
Democracy Rulers worry about re-election less corruption allocation of capital
towards production
• Cross country regressions: Positive relation between institutions and GDP but is it causal?
Causal evidence
• “Causal effect” refers to the answer to the following counterfactual thought experiment: if, all else
being equal, a particular characteristic of the country were changed (“exogenously”), what would be
the effect on its growth rate?
• Answering such causal questions is extremely difficult because:
I it is difficult to observe changes in country characteristics that are not driven by something
that will also affect growth directly
so difficult to isolate the impact of a given characteristic
• There is no definitive answer context matters!
• Any statement we make must have theoretical foundations and is grounded in rigorous evidence
Rigorous evidence = no anecdotes, because that is not representative we are all subject to
“confirmation bias” (we tend to notice/remember things that match our prior and discount
stories hat does not agree with us)
• Not just look at correlations because of reverse causality, omitted variable
• Empirical setting in which one variable changes, ceteris paribus
Poverty: Measurement
Definition of poverty: people experiencing well-being below some minimal acceptable level
Vulnerability to poverty
• Many households are not poor, but vulnerable to falling into poverty
• vulnerability: risk that household will fall into poverty at least once in the next few years
• Indonesia: 20 percent of people below poverty line
• close to 60% of total Indonesian population have a ≥ 50% chance of falling into the bottom 20%
• The percentage of being sometimes poor is much higher than the percentage of being always poor
Main point: no matter how you measure is, world poverty is massive and a very serious challenge
• And lack of income is a very good proxy for lack of development
Poverty traps
• Poverty trap = the poor are poor because ... they are are poor.
• Initial conditions play a crucial role in determining where an individual or country ends up
a self-perpetuating condition where a household (economy), caught in a vicious cycle, suffers
from persistent poverty (underdevelopment).
Many possible traps:
• Low schooling trap: Poor children can’t afford school once adults, they are stuck in low-wage
occupations such as farming they remain poor and cannot send their kids to school
• Low asset trap (with fixed cost in production): Suppose you need 100 dollars to start a profitable
business, otherwise stuck in subsistence farming: if there is no credit market, only those with enough
wealth to start with can invest and grow richer
• Infant industry trap due to Learning-By-Doing Externalities: Young firms are inexperienced and
less productive, so they cannot offer wages high enough to attract/keep experienced workers
Remain inexperienced and unproductive.
• Demographic Trap: the economy may be caught in the vicious cycle of high fertility/low human
capital and child labor traps
• Scarcity trap: Extreme scarcity may cause the poor to be worse decision-makers than the non-poor
the non-poor. I Financial stress results in lower mental resources (cognition, attention span, self-
control, etc.). Because mental resources are important for economic outcomes, this creates a
poverty trap.
I Each model of poverty trap is designed to highlight one particular feedback mechanism behind
the vicious cycle In reality, of course, many sources of poverty trap are likely to co-exist
Theory
The capacity curve: relates income and work capacity (productivity)
• Higher income better nutrition higher work capacity / productivity
But the relationship is not quite linear Calories you ingest: first 1700 of them used by the
body for basic (resting) metabolism, only then does it translate into higher capacity, up to a
certain point
Capacity curve
• Use the capacity curve to introduce a relationship between
income from work today and income you can generate tomorrow
• Inter-generational version: can also think of it as relationship
between income inherited from parents and income once adults.
Inherited income = income today: determines health status
Income from work = income tomorrow: determined by
health status
Everyone gets poor: no trap, unique equilibrium When does a poverty trap emerge?
If the capacity curve is about the 45 degree line, everybody ends up poor
• Note that a double feedback loop (e.g. poor nutrition lowers your income, and low income means
you cannot buy a lot of food) does not always mean there is a poverty trap: the relationship has to be
sufficiently strong
In mathematical terms: the product of the two elasticities has to be greater than 1
The math
• Income is a function of health:
yt+1 = f (ht), f’ > 0 investment in health today raises income tomorrow
• But at the same time health is a function of income:
ht+1 = g(yt+1), g’ > 0 higher income today means higher health investment
• Combining the two functions defined above gives us a dynamic map:
yt+1 = f (g(yt)) = ϕ(yt)
• Necessary (but not sufficient) condition for there to be a trap: The capacity curve needs to intersect
the 45◦ line from below. This means that there must be a range where for a dollar in extra income,
capacity increases by more than 1. That will be the case only if there exists a fixed point where ϕ’ (.)
> 1.
• In terms of elasticities, this would mean that the elasticity of capacity wrt income must be > 1,
which means that the product of the elasticity of capacity wrt nutrition and the elasticity of nutrition
wrt income must be > 1 (at least somewhere)
f ’ (h) g’( y)
ϕ' (y ) = f’ (g(y )) × g’ (y ) = h× y = εfh × εgy
f (h) g( y)
εfh is the elasticity of capacity w.r.t health (nutrition)
εgy is the elasticity of health (nutrition) w.r.t income
Empirical evidence
• To test whether the conditions are such that there could be a trap, need to look for evidence on:
1. Effect of income on nutrition
2. Effect of nutrition on work capacity
• And the product of the two elasticities has to be at least 1 somewhere
Conclusion
• Empirical evidence does not really suggest there is a nutrition based poverty trap
• But the poverty trap model is not a literal description of the reality. The model is there to force us
to think about how the nexus between human capital and income can lead to a vicious circle
• This circle may be found across generations, rather than within
• Effect of parental income on children nutritional status:
In rural Cote d’Ivoire, children are more likely to suffer from malnutrition when rainfall in
abnormally low (Jensen, 2000)
In South Africa, old-age pensions have a large positive impact on the anthropometric status
(weight for height and height for age) of grandchildren (Duflo, 2003)
• Effect of children nutritional status on future adult income:
Barker hypothesis (1992): poor nutrition in utero leads to irreversible losses (IQ, physical
fitness)
Evidence: iodine deficiency in first trimester of pregnancy leads to permanent irreversible
brain damage. Study estimating impact of Iodine Supplementation campaign among
pregnant women in Tanzania find very large impacts on educational attainment (Field, Robles
and Torero, AEJ Applied 2007)
Policy implications
• Poverty traps can exist outside the health context (school, business etc)
• Sachs “Big Push” idea: Give money for basic health, education, productive investment get
people out of trap people can start getting richer and richer on their own
• If there is no trap, basic income may reach humanitarian goals but not promote growth
• Tropical regions tend to be poor most poor countries are between the two tropics
• tropical diseases
malaria, yellow fever, dengue, helminths, sleeping sickness, lymphatic filariasis,
schistosomiasis, …
• Is there a causal relationship from a high disease burden to poor development?
Theoretical mechanisms
Direct effect of disease on productivity
• Disease burden Poor nutrition Lower productivity
• Disease burden Many illness spells Fewer days worked (because sick oneself or caring for sick
relative)
• Disease burden Exposure in utero or childhood Reduced endowments (physical, IQ) Lower
productivity
• Figure B-1: Ratio of malaria mortality to total mortality by state circa 1890. Darker colors indicate
more malaria.
• Figure B – 2: Malaria Intensity by State in Brazil Notes: Displays a map of an index of malaria
ecology as constructed by Mellinger et al. (2004). Darker colors indicate climatic and geographic
conditions more conducive to the transmission of malaria..
Results: US
• For those born before 1900, more pre-eradication malaria in one’s state of birth predicts lower
adult income on average (coefficient is just below -.1)
that’s because they were exposed in utero/childhood
• For those born after 1920, more pre-eradication malaria in one’s state of birth has no effect on
adult income on average (coefficient is around 0)
Results: Latin-America
Robustness
• Can perform “placebo” tests:
Old versus very Old
• Test if the assumption on parallel trends is likely to hold:
OK (no diverging pre-trend) not OK – diverging pre-trend across two sets of
• If diverging pre-trends: then diff-in-diff doesn’t pick up effect of malaria eradication campaign but
some other thing
• Bleakley performs such placebo tests, results pass the test!
Competing risks
• e.g. consider a population in which 50% of children are healthy, 20% have TB, 20% have malaria
and 10% have both diseases.
• If having Malaria and TB makes you more sick than having Malaria or TB added up, adding up the
disease specific effects overestimates the true impact of eliminating both diseases
• So ideally we would want to look at macro-estimate (that takes into account effects across
generations and diseases)
• But what’s hard there is to find plausible source of variation (identification)
if countries that eradicate malaria are those with better institutions, then comparing them
with countries not eradicating malaria would not be very helpful in estimating impacts of
malaria eradication
Supply of health
Failures in service delivery
• Services fail poor people often
(a) in access
(b) in quality
Lack in access
• The nearest school or health center can be quite far
• More public spending for the rich than for the poor
• This does not mean that poor people visit health providers less often
In rural Rajasthan, people visit a doctor about six times a year.
In the US, people visit doctor about 3.4 times a year.
• But they pay much less, and receive lower quality
Many visits to private professionals without formal license or pharmacies.
Surveys indicate informal sector accounts for 65−77% of care seeking in Bangladesh, 36−49%
in Nigeria, 33% in Kenya, and 55−77% in Thailand.
• Implications:
Households get discouraged
Don’t bother to bring child to center for immunization since with more than 50% chance
center will be closed when they show up
• Fascinating work by Das and Hammer:
“Which doctor? Combining vignettes and item response to measure clinical competence,”
“Money for nothing: The dire straits of medical practice in Delhi, India,” J
• Setting: Delhi. Both public and private physicians interviewed
• Majority of the private physicians don’t have any medical training
• Vignettes to measure health providers’ depth of knowledge:
Providers know the correct treatment only about 40 percent of the time
• Fake patients sent to measure health providers’ practice:
In the average interaction, physician sees the patient for 3.8 minutes, asks 3.2 questions and
performs less than one examination procedure
So not only they don’t know much, they don’t even use all of their knowledge! (there is a
“know-do” gap)
Median provider is ranked as “harmful”
• It is not just India
• “A recent study in Benin found that one in four sick children received unnecessary or dangerous
drugs from health workers”
• The World Bank started measuring the competence of providers in several countries in Sub-
Saharan Africa through its Service Delivery Indicators initiative:
diagnostic accuracy was 34 percent in Senegal, 57 percent in Tanzania and 72.2 percent in
Kenya
So high provider absenteism may be a blessing in this context
Recap
• Contracting out
Works if and only if well-regulated
For services which are well defined
The private sector in itself is not the solution to government failures
• Bottom-up monitoring
Relying on communities to ensure that they receive health services is tricky; on their own,
they often do not have the resources / information necessary
• Top-down monitoring
Ensuring better selection or providing incentive schemes (publicly provided or publicly
regulated) are essential to increase performance
• Overall, good quality health care must be a politically salient issue to guarantee the quality of the
public sector
Otherwise government doesn’t implement own rules / doesn’t monitor / doesn’t audit
Conclusion
• A lot of “low-hanging fruits” as Banerjee and Duflo call them in Poor Economics
We know bed nets prevent malaria, water filters or water purification prevent diarrheal
diseases, etc
• But poor governance prevents these low-hanging fruits from being picked
• Ultimately, having to constantly use a bed net, filter one’s water, etc. is not the solution
None of us have to do that
We have a malaria-free environment, clean water
In developed countries, most improvements in health came through big pushes, e.g. clean
water infrastructure in cities
• Issue of incentives / accountability of public providers not specific to health
• Crucial topic on the development agenda!