growth is good for the poor---Dollar & Kraay
  - WHAT IS A SYSTEMATIC EFFECT???
  - study average income of bottom quintile (poor) in country w.r.t. 
    average income of entire country
  - study developed and developing nations
  - find strong correlation between growth of poor and growth of country, as
    well as average income in both.  That is, they change together in the same
    direction.
  - growth-enhancing policies affect the poor and everyone else 
    equiproportionally
  - greater economic integration between countries benefits the poor equally
  - controlled for changes over time in last four decades, as well as regions
    of the world---found no significant effect
  - negative growth is also equally negative for the poor, but
    interesting quote on p. 209: "It could still be the case that the same
    proportional decline in income has a greater impact on the poor if social
    safety nets are weak, and so crises may well be harder on the poor."
    
economic lives of the poor---Banerjee & Duflo
  - we keep using under $1 or $1.08 as extremely poor, but how do they actually
    live?
  - based on data from 13 countries in asia, africa, latin america
  - udaipur and pakistan are majority poor!  80-ish% poor (under $2.16/day)!
  - base on consumption in PPP.  Note that this adjustment isn't perfect,
    as it's costlier in urban areas, etc.
  - aren't counting the poor---instead are describing their conditions, so if
    the poverty line is a bit off, they have still contributed something.  note
    that the poor and the extremely poor live in similar conditions.
  - large households: 6-12, median 7-8.  adults 2.5-5, median 3.  large households
    spread the fixed costs of living.
  - many children in household.  several moms per household.
    young(<18):old(>51) is median 6:1.  in the US it's 1:1
    old(>51):midage(21-50) is .2-.3:1, whereas in the US it's .6:1
  - food is 50-80% of budget, and 5-10% on tobbaco/alcohol in some countries,
    but less in latin america (where other intoxicants may be popular)
  - festivals are huge.  udaipur---99% of extremely poor spend money on
    wedding/funeral/religious festival.  Less than 1% in 13 countries spent on
    entertainment like movies, etc.
  - radio/tv ownership in extremely poor varies drastically by country
  - extremely poor spend same fraction as poor on food, even though they could
    cut alcohol/tobbacco/festivals by upt o 30%.  don't seem eager to buy more calories.
  - increase in consumption doesn't go to best bang per buck foods.  50% of increase
    goes to extra calories, adn 50% goes to tastier, more expensive calories.
  - poor are spending less money on food, and getting less calories over time,
    but this may be because they do less physical work.
  - tv ownership jumps from extremely poor to poor, but varies by country.
  - land ownership by extremely poor goes from 1.4% in south africa to 99% in
    udaipur.  most of the land is scrubland, not easily cultivated.
  - few extremely poor have durable assets like bikes, clocks, chairs, beds.
  - poorest decile get 1400 calories, whhich is half of what they should
  - other udaipur health statistics: 65% men/40% women are underweight,
    55% aenemic (low red blood cells), 72% report symptoms of disease, 46% were
    bedridden in the past month
  - the poor feel poor, but not unusually unhappy!  stress in udaipur/south 
    africa lower than in usa.
  - extremely poor spend ~2% of budget on education.  that said, in 12/13 countries,
    50% of kids aged 7-12 were in school.  many countries have free school, but
    public schools disfunctional, so some parents send kids to for-pay schools.
  - multiple jobs---sell dosas in the morning, saris in the afternoon.  sometimes
    they are entrepreneurs and day laborers.  having multiple jobs not true in
    all palces---in SA and Panama, almost no one has >1 job.
  - even if they own farmland in udaipur, it's not their primary source of income.
  - many have nonagricultural jobs, especially in urban areas.
  - udaipur survey asked about migrant work.  60% of families had one member
    (58% said head of household) migrate for work.  median length was 1 month,
    and they usually didn't leave udaipur or rajasthan.
  - permanent migration for work is rare---less than 10% even in urban areas
    which attract more immigrants.  indonesia, where migration is subsidized,
    has 41%.
  - because of migration and short stay in jobs, specialization is rare,
    which trades away opportunity of higher income.
  - businesses typically don't achieve scale.  usually the poor who run
    businesses have no paid staff, and usually run with <3 family members.
    Mostly operate out of the same room as living space, and few have a vehicle.
    Dosa-makers don't work in pairs, which would be more efficient, and spend
    more time waiting for clients.
  - having outstanding debt: 11% in rural east timor to 93% in pakistan.  Few
    loans are from formal lending source---usually family or shopkeeper, even
    in Hydrabad where banks are available.  Indonesia has special effort,
    and 1/3 of loans for poor are through formal bank.
  - extremely poor pay 3.84% per month interest, whereas poor pay 3.13%.
    That's high not because of risk of default but because of cost of
    contract enforcement
  - cote d'ivoire 79% of extremely poor have savings account,
    average is 14% in other countries.
  - 10% of udaipur/hydrabad is in self-help groups which pool savings and loan
    to members.
  - health insurance usually nonexistent or informal with neighbor assitance.
  - historically, many poor own land, but it usually doesn't have a title, etc.
  - infrastructure like water, electricity, latrines varies drastically between
    studied countries.  poor have better access to tap water than extremely poor,
    and urban is better off than rural.
  - most indian villages have school within kilometer, and health subcenter for
    each 10K people.

The potato's contribution to population and urbanization---Nunn & Qian
Introduction + Section 5A (p.21)
  - Look at potato cultivation suitability in different regions in the old world
    to conclude that in 18th and 19th centuries, potato played significant part
    in population growth
  - From 1000-1900, world population quadrupled, but only in the last 2-300
    of those years.  Population increase==increase in fertility or decrease
    in mortality.
  - In 1820 fertility began to decline, so decreased mortality has to be it.
  - Previous study (Fogel) shows that caloric intake, height, and life expectancy
    are correlated in those periods.  Infers link between nutrition and
    population growth, but doesn't do it directly.
  - Traditional view is that modern medicine/sanitation increased life expectancy
    in 19th century, but in reality is was also nutrition, which completes the
    trifecta of reasons for population growth.
  - This study finds causal relationship between improved nutrition and
    population growth.
  - They don't try to control for medicine/sanitation.  Instead they study a
    positive shock of the introduction of the potato (better calories/nutrition)
    2 centuries before medicine/sanitation caused the growth.
  - Also show that agricultural productivity leads people out of the countryside
    and into cities due to neednig less farmers.
  - Two potential problems with showing causality:
     1) reverse causality---population expansion led to adoption of potato
     2) omitted/unobserved variables that caused growth rather than the potatos
  - To avoid these, they study potato coming from new world (america) to old
    world.  Combined with different areas that were more or less suitable
    to growing potatos in old world, they can control for a lot of internal
    and time-variant characteristics
  - Controlling for a bunch of other stuff, they show that old world regions
    suitable for potato growth had way more population growth
  - Effect size of potato: 12% increase in population, 22% increase in pop.
    growth, 47% increase in urbanization, and 50% increase in urb. growth
  
The great Bengal famine---Amartya Sen
  - 1943, 1.5-3M people died in famine
  - 3 rice crops: winter autumn spring, winter is most important.  In 1942 cyclone
    killed crops and japan occupied burma, so no imports.
  - cost of rice almost tripled, and it's hard to track since in sept. 1943, govt.
    capped the official price of rice, so its market price is hard to know.
    Unofficial reports were up to 7.5x.
  - starvation death came first, then in next phase came epidemics
    (cholera malaria small pox)
  - famine first revealed itself in early 1943 in districts away from Calcutta
  - Calcutta offered rice to over 1M of its citizens starting 1942, first at
    mandated prices, then at (higher) market prices.  Famine didn't really hit
    city, except that streets filled with famine-affected travelers 
    from rest of bengal
  - Number of sick/destitutes reached 100K in october '43, and calcutta made
    a controversial decision to move them to camps outside the city, where
    they were still given some food.
  - Stable crop and relief came to other provinces after october, but
    famine-induced epidemics hit next
  - Food availability decline (FAD) is cited as reason for death in famines.
    If you look at numbers, though, 1943's yield was only 5% lower than last 4
    years' average, and 13% higher than 1941 (which had no famine).  To explain
    this, they refute a few ideas
     - Even after correcting the slightly incorrect reported yields, the
       1943 yields aren't disastrous
     - Aside from rice yield, wheat imports are big in Bengal.  Turns out
       1943 wheat imports are still higher than 1941 by 11%.
     - Adjusting for population growth in Bengal by 1943, per-capita availability
       index is still higher than 1941 by about 9%.
     - Giving benefit of the doubt that some food was delivered late, there
       was still more per-capita availability in 1943 than 1941.
     - Some confusing argument showing that enough old rice was carried over from
       1942 to 1943.
  - Combine these to get that food shortage was not reason for famine.
  - Keep in mind that the famine was mostly rural, and mainly due to lack of
    calories from rice.
  - Rural areas were agricultural.  While food prices rose in '42-'43, 
    argiculture labor wages dropped because of crop destruction.
    This made it hard to be a farm laborer
    and afford food.
  - Prices for rice vs. haircuts and fish rose (less for fish because large boats
    were denied due to fear of japanese attack), so for some reason rice got
    expensive even though it wasn't that scarce and rural laborers didn't
    have money.
  - Surveys of job titles of the destitute seemed to lean more toward wage laborers,
    fisherman, transport workers, etc., and affected peasant cultivators and share-
    croppers less since they had direct access to rice.
  - So it's starting to seem that for the destitute groups, they had
    less 'entitlement' to rice, which led them to get less calories.
  - Some ideas on why all of this happened
    - Wages dropped for affected groups while inflation was high due to war 
      construction (Phase 1)
    - Due to message that crop yields were low, there was panic hoarding, which
      was done by producers of the product and drove prices up for everyone else
      (phase 2)
    - Adminsitrative observations increased the chaos in phase 3 after observing
      first two phases
    - Goverment policies against trade of rice across indian states slowed recovery
    - Wage laborers took a hit in income while prices rose.  Urban dwellers and
      those involved with the military didn't feel this pressure, as their
      wages rose with inflation.
  - What could have helped?  A large public rice supply to ration out, as was
    mandated by law, but not implemented because an official famine was not 
    decalared.

Making Famine History (CORMAC Ó GRÁDA)
  - Summarizes recent economics and economic history contributions to the
    study of famine.  In last century, more famine caused by war and totalitarianism
    than by overpopulation or economic backwardness
  - Defines famine as excessive death due to starvation or hunger-induced illness
  - In the last century: famine left most of globe, but---china in '59-'61 had
    most deaths to famine in history, world saw 70 million famine deaths (less
    than epidemic diseases, but more than WWI+WWII), and in addition unborn babies.
  - Countries are seeing smaller famines than they did earlier in the century,
    and famines are now mostly only appearing in war-torn/poor areas.  Famines
    now kill thousands, not millions.  In the past, famine killed a larger
    relative proportion of the population.
  = Accounting for Famines = 
  - In the past, poor economics led to huge pain from famines as countries
    could not react, had no insurance, and had no backup crops.  Large
    crop shortfall = many famine deaths.
  - UN output and welfare indices strongly correlate famine and underdevelopment.
    Excess mortality is correlated with population density, and how rural a place is.
      - true of most of famine-ridden africa today, ireland in 1840s, bengal in
        1942-43, china in 59-61.
  - In todays' sub-Saharan Africa (only famine-prone region left today), infectious
    diseases/diarrhea/parasites still kill many people during famine, since
    the famine just compounds the already hard task of getting aid to those who need it.
  - Many famines in poor countries stem from extreme weather that destroys crops,
    though it's not a necessary or sufficient condition for famine.
  - Single bad harvest is usually not enough---need a 2+ crop cycle disruption
    to get even the weakest of economies.  Two year back-to-back disruptions are rare.
  - War and political regimes can lead to famine, even in countries like the
    USSR in the 30s which had 3x the GDP/capita of today's poor African nations,
    but still had a famine.  Totalitarianism and war can lead to such things.
  - Amartya Sen shows that often it's entitlement differences (differences in how
    much each person gets) than the decline in output that leads to famine.
    Sometimes in famines there is enough food to feed everyone, but the indicator
    of the famine is that some fear, hoarding, or other social problem is leading
    some subset of the population to not get a fair share of food
  - FAD = food avail. decline ==> not an entitlement decline, but one where the
    supply actually drops below what's needed.  Low evidence for a pure FAD
    famine in recent time.  Little evidence for a purely entitlement-based
    decline---there's usually some FAD/supply component to famines.
  - Transportation/communication/technology costs are low enough that famines
    are "easily" avoidable.  One reason for famines in 20th century is thus
    war and totalitarianism, where there are inequalities and communication limits
    stop information on famine from spreading.
  - Flipside: democracy makes famines harder, since it reduces inequality,
    leads to more criticism of poor management, and doesn't limit communication.
  - While you'd often be able to hedge poor crop cycle in different parts of the
    country, market speculation often creates famines as people hoard.
  - In general, markets help prices change uniformly across a country that has
    a crop disruption in one area, but the effects of speculation and hoarding
    often reverse this uniform movement, and you see disconnected prices in different regions.
  - Government can help reduce famine.  Usually govt. tries to show solidarity,
    and focuses on how it split the relief, rather than the amount of relief, so
    that it can show it was fair.  Example of Quing dynasty in china---distributed
    relief in size proportional to disaster, but corrupt beaurocrats would give
    more relief to the rich, so the poor got double-screwed.
  - To combat govt. corruption, several schemes: Giving food below market prices
    leads to corruption or hoarding, so this only works with perishables.  Giving
    money by wages for public works projects can bias against the poor who can't
    perform as well, and can lead to spread of infectious diseases or more mortality
    if they are paid below-market wages (this happened in Ireland).  In Ireland,
    soup kitchens worked better than public works projects, whereas in Bengal the
    soup kitchens were too disorderly and didn't provide much nutrition.
  - NGOs have gotten good at linking poverty and famine risk, and have improved
    speed of fundraising for highly publicized crises.  This contrasts with
    their desire to get more funding for bureaucracy and to claim more credit
    so they get more money later.  Monitoring their activities is thus important.
  - NGOs lag the media coverage, since they need emergency funding to react
    to the events.  Because of reliance on emergency funding, they often jump into
    areas where they aren't experts or are too small to help.
  = Famine Characteristics =
  - Soaring food prices/poor harvests are a sign of famine, but not necessary
    or sufficient.  Abnormal jump in mortality is a better sign, but measuring
    abnormal mortality is hard, and is both underestimated and overestimated to
    the tune of millions of deaths.
  - Deaths from famine above the average are smaller in number than in the past.
    (from 15-43M in china in 1959-61 to less than a million for the largest
     african crises in the past century).
  - Relative deaths are more interesting in some ways, but the denominator
    (what region should you consider) is controversial.
  - Famine deaths in relative and absolute terms has dropped in the 20th century
    compared to other centuries.
  - Disease is often cause of excess famine death rather than literal starvation.
    One class of diseases is nutrition-based, including diarrhea and starvation
    itself, while the other class comes from migrating poor people and a breakdown
    in sanitation during the famine.
  - Trend seen in Leningrad and other developed societies---first famines result
    in more disease deaths, but as basic sanitary and medical habits rise, future
    famines see larger share of literal starvation deaths, as disease is moderated.
  - That trend doesn't exist in sub-Saharan africa today, since disease is
    not moderated even in times of no famine.
  - Women have higher survival rate form starvation, since they store more
    fat than muscle.
  - Excess deaths aren't the only thing in a famine---age demographics matter:
    - Lower birth rate (.4M in Ireland, 43M in China) due to lower libido,
      migration of men (to work) from women (to beg), lower nutrition for
      carrying pregnancy, and lower marriage rate.
    - Lost births are actually postponed---often, post-famine, the births rebound
      and the birthrate surpasses previous highs for a while.
    - Ill/elderly often die younger than they otherwise would in a famine, so when
      famine ends, those groups see reduction in number of deaths (premature death).
  - Long-term health---links between nutrition in utero and health/mental development
    later on. Low nutrition later in life means (in China) schitzophrenia,
    breast cancer, heart disease, antisocial personality disorder, defective teeth).
    Children end up to several cm (8 in Leningrad) shorter.
  - Malthusianism: limited resources lead to stable population eventually.  In
    most cases, this doesn't happen, since crop growth is outpacing population growth.
  - sub-Saharan Africa, while seeing drops in famine mortality, still hasn't increased
    crop fertility, due to arid land, poor development, and some effects of global warming.

"Raiders of the Lost Arc Elasticity," or "The Indiana Jones of Economics" by Steven Levitt
  - About Robert Jensen at Brown, with Nolan Miller.
  - Study Giffen Goods, which have positive price elasticity (more people want the
    good as its price rises).
  - Example he gives: you are poor.  you buy bread for cheap to get calories.  with
    small amount of money leftover, you buy meat, which is a luxury.  bread prices
    go up, and you can either buy less bread and same ammt. of meat, or more bread
    and less meat.  only one option to survive: more bread, less meat.  That makes
    bread a Giffen Good (price went up, demand went up).
  - People claim this is true of potatoes in Irish famine, but that's been disproved.
  - Found data showing that in south china, high rice prices increase consumption, and
    in north china, high wheat (noodle) prices increase consumption.
  - Some cities have high price and high consumption, while
    others have lower price, lower consumption.  Identification problem: Did high
    demand lead to high prices (standard economics) or did high prices lead to
    high demand (Giffen behavior)?
  - Needed to have some external factor affect price without affecting demand
    (except through price).  Rainfall affects crops which affects supply but
    not demand, but there wasn't enough data.  Instead, went to Hunan province
    (south---rice) and Gansu province (north---wheat).  These households
    typically bought a lot of the staple item and a little bit of meat for luxury.
  - Studied 1300 households to get basleine, randomly gave some of those households
    vouchers to subsidize wheat/rice purchases during a six-month period.
    They collected data before, during, and after the subsidy period.
  - Ooops...got the exact opposite of what they were looking for.  But it was
    a regression mistake---coded the price increase as a price reduction.
    In truth, they had found a Giffen Good!  When subsidy ended, price effectively
    went up, and consumption of the product went up as well.  
  - In the end, most populous country with the most popular food staples are the
    Giffen goods (rice more than wheat).  Turns out that Giffen Goods are found
    in what was theoretically described as a Giffen household: poor households
    with simple diets, which are spread all over developing countries.
  - Implications for public policy:
    - Already, food subsidies in low-income countries (e.g. India, Egypt) take
      about 1% of GDP make up the largest form of social assistance, and they
      are criticized.  Several reasons: distort market signals---hard to tell
      if demand is meeting supply, since the subsidy changes behavior.  Also,
      they lead to shortages, smuggling, black market activity due to below-market
      prices, which disproportionately benefit the poor the least.
    - Subsidies are supported because people believe they at least guarantee
      a basic level of nutrition, unlike giving out money, which may not be
      spent on nutritional or basic food.  But the Giffen Good research shows
      that subsidizing the poor decreases their consumption of the good!
  - Follow-up paper shows that the Hunan subsidies _hurt_ nutrition, and that
    Gansu subsidies had no effect on nutrition at all!  And the sample they studied
    only included the extremely poor (< $1/person/day), which are the ones
    suffering from malnutrition.
  - So in the end they don't suggest a policy better than subsidies, and don't
    say they are bad when there are huge price swings like the recent food price
    crisis, but they say we should abandon the idea that they are the best form
    of aid since they improve nutrition.

Child Labor---Christopher Udry
  - trade future welfare for additional current income
  - majority of cases are not extreme (not sex work, dangerous conditions, or
    forced labor).  usually it's for family on a farm or something less rough.
  - poverty leads to child labor---parents need more money.  it's also true the
    opposite way, though---child labor removes education, which leads to future
    poverty---vicious cycle
  - 210M children between 5 and 14 worked in 2000, about half full-time (1 in 10
    children work full time)
  - 1 in 5 primary school-aged children not in school.
  - largest absolute # in asia, but largest % in africa.
  - overwhelmingly rural---70% in agriculture
  - more boys than girls, and older children more than younger siblings.
  - benefits: wages + reduced costs of schooling
  - harm: lower future wages
  - babies born to well-educated parents, especially mothers, are healthier
  - family income is strongly negatively related to incidence of child labor,
    with a weaker relationship in more affluent countries
  - as household income increases, child labor decreases and childhood education
    increases.  that's called an "income effect."
  - however, as wages increase, they tend to do so for adults and children,
    and so wage increases increase the cost of sending the child to school instead
    of having them make money, called a "substitution effect."  so increasing
    wages isn't as good as government subsidies for sending kids to school.
  - analyze investment in human capital of children using market-based approach
  - first, assume costs (less education) and benefits (current wages) only affect
    family (assume education doesn't matter for greater society).  you have to
    take undiscounted current wage and compare to interest-adjusted future
    benefits of education.  best value when cost = benefit, so not more school
    than will lead to better pages in the future.  this assumes that the
    financial markets are perfect, and the parent can take a loan now that
    can be paid off when the child is working at higher wages due to
    education later (unrealistic)
  - even worse, the benefits of education are not all internalized to the family,
    as society benefits from the education as well.  this means that parents
    undervalue education, and send their kids to work earlier when cost = benefit.
  - so an imperfect financial market and a lack of complete internalization to
    the family of the benefits of education lead to increased child labor past
    the optimal level.
  - they argue in several ways that there is causality from low income to
    increased child labor and decreased education.  evidence comes from
    temporary price shocks (floods, etc.).  Several studies show that these
    adversely affect the incidence of child labor and decreased education in
    the poor more, since they can't take a loan or count on savings to fund
    the education while they get past the slump.
  - probability of a 16-year-old girl in sao paulo to enter workforce in the next year
    goes from 22% to 35% if her dad becomes unemployed while her likelihood to stay
    in school goes from 70% to 60%.
  - keep repeating that it's poverty and inadequate access to capital markets
    that causes this child labor poverty trap
  - agency---parents making decision on behalf of children
  - if agency has no societal effect, and parents want to leave a positive
    benefit when they die, they will give optimal balance of cost = benefit.  in
    a poor family, this is not possible, since they will have nothing to leave
    behind (money to borrow against to pay for child's education), so they
    have to put child into labor for now or else child won't even live.
  - common problem when considering agency is considering a single household,
    where income is pooled, and parents make decisions as a groupthink.  this is
    inaccurate---as Duflo showed in 2003, the nutrition of girls in south
    africa increases if grandmothers receive old-age pensions, but no effect
    if grandfathers do.
  - poverty and child labor are a vicious cycle.  eliminating poverty will
    eventually eliminate child labor, but can we do it without waiting?
  - governments could outright ban child labor.  it's unclear whether developing
    countries' governments have the ability to enforce that, and it's possible
    that this would devastate families resorting to child labor out of desparation.
  - instead, one focus should be on case where parents have no idea what the
    cost to their child is: hazardous conditions or bonded labor.
  - sanctions from the west to the developing world are also dangerous:
    reducing wages just leads to more need for child labor, and less family
    benefit in general.
  - credit markets aren't likely to improve in the developing world any time soon,
    so that outlet for financing education is unlikely.  so what to do?
  - most promising tool yet is to give subsidies to families that
    send kids to school.  this addresses the agency and cost-benefit issues head-on
  - in mexico this was implemented by paying slightly less than wages to children,
    and was done in a randomized way to study the effects of such subsidies.  the
    program was measured to be quite successful (increased schooling by .666 of a year
    from 6.8 years).  similar results in nicaragua.  in bangladesh they only pay
    15-25% of child wages, but still effective.  all programs drop child labor while
    increasing schooling.
  question: this writeup treats families as calculating machines that weigh
  long-term benefit against short-term needs, and understand the benefit of
  education.  is this realistic?

Policy Dilemmas for Controlling Child Labor---Kaushik Basu
  - policies backfire readily, so making sound set of decisions is very important
  - economically active---above the age of 12, not doing hazardous work is
    economically active but not a child laborer
  - avoid fallacy of single-mindedness->there are worse things to happen
    to a child than child labor---hunger, malnutrition, illness, abandonment, prostitution
  - resisting direct intervention (banning child labor) can be consistent with
    wanting to improve conditions, but doing so without making child labor illegal.
  - Things with a negative effect on child labor (these are good): quality of schooling,
    incentives such as subsidies or meals, improving adult labor market.
  - will show argument that fining firms which employ children can 
    lead to increased child labor.  there's no hard evidence in either direction,
    but wants to provide a counterargument to what lots of people suggest.
  - Britain started with child labor laws in eatly 1800's.  Then Mass. in 1837.
    However, drop in US child labor only came in 1880-1910's.  Evidence shows
    that labor laws weren't the reason, and may have been a detriment.
  - Any time child labor is used as a means for meeting subsistence consumption,
    laws that fine firms that do it make child wages drop (children are
    less attractive workers now).  This drop in wages means children need to
    work harder to meet subsistence.  This holds true until it costs more than
    the wages of a child in terms of fines.  Then child labor will drop.
  - the relationship between the size of the penality against a firm and the
    number of children working (or output of children working) is an inverted U,
    increasing while childhood wages are positive but dropping, and going back
    once child wages near 0.

J-PAL issue 1---Meeting the Millenium Development Goals
  - if child labor brings in a lot of money to family, it will be hard to get them
    in school.  however, if they don't bring lots of money, then lowering cost of
    school will help.
  - random evaluations of attempts in sub-Saharan Africa, latin america, and south
    asia show that making school cheaper and subsidizing attendance, as well as
    improving child health helps increase attendance.
  - randomized evaluations are key, because they help control for many undesired
    effects, like volunteers that are more motivated, or leaders who bring on change
    being more considerate anyway.
  - randomized testing works by (for example) taking 100 representative 
    schools, making change in 50 chosen randomly, phasing in the change
    in the next 50 over time, and observing what happens.
  - attempts were targetted equally at boys/girls, but biggest improvements were
    to girls.
  - improving quality of education did not lead to higher attendance, and higher
    attendance didn't lead to improvements in learning
  - PROGRESA program in Mexico introduced payments for vaccinating children and
    sending them to school.  Program was phased-in randomly, which allowed
    growth to be detected.
  - costs include uniforms, materials, and school fees.  subsidizing these in
    7/14 randomly poorly performing schools in Kenya increased attendance by 15%.
  - giving free meals increased school participation by 30% in 25 schools that
    got free breakfast over 25 that didn't in india.
  - Deworming in 25 Kenyan schools, phased in randomly, decreased absenteeism by
    25%, and helped nearby schools which had reduced sickness as an externality.
  - chart on page 3 of costs is very interesting, but keep in mind that mexico
    is richer, and that many programs don't have keeping kids in school as their
    primary goal---health or putting money in the hands of the poor is often the
    goal, with increased education as a side effect.
  - note: increasing participation doesn't always show evidence of children
    learning more
  - note 2: improving the quality of education, or increasing the amount of
    catch-up help for students left behind, did show an increase in participation
    in one study.
  - girls have seen increased participation even when changes didn't target girls.
  - increasing women teachers where possible led to 50% increase in girl participation
    with no affect on boy participation.

Randomized Evaluations of Interventions in Social Service Delivery
  - Hard to measure causes of changes.  If new headmaster installs a new
    parents' committee, and people learn he cares a lot and more students
    enter the school, was it the headmaster or the committee that led to more
    enrollment?
  - To tell if it works, can pick 100 schools, put parent's committee in 50,
    and phase the other 50 in over time while comparing the results of the
    schools before and after.
  - Same examples as the J-PAL article (notes are above this one) about increased
    school participation.
  - nonrandomized testing can lead to false conclusions.  free textbooks + flipbooks
    for education can improve learning + test scores in nonrandomized test.  randomized
    test showed that textbooks only help students who are in top 40% in pretest, and
    flipbooks only help because they are more popular in richer areas.
  - providing inputs to students isn't useless, though.  in randomized evaluation
    in india, young women provided remedial education to students identified as
    lagging, and raised overall scores by .39 std. dev. and lagging students by
    .6 std. dev.  That program was 6x more cost-effective than computer-assisted
    learning in the same schools.
  - Since mid 1990's, India has decentralized social good delivery to local councils,
    with stipulation that > 1/3rd of positions go to women, and that disadvantaged
    minority tribes and castes get proportional amount of people on the councils
    to their population.  To avoid manipulation, positions were randomly allocated.
    In villages reserved for women or minorities, those groups got goods chosen
    to better benefit them.
  - explanation of phasing in second half
      question: why is randomize testing "half, then phase in the other half."  Why
            is the phasing in important rather than just changing half and
            leaving the other half the way it is until you have something
            conclusive?
      answer: you get funding for X schoools, and people will get mad if they don't
              receive help.  you will annoy people less if you say "everyone is getting
              it, just in a phased rollout due to lack of resources."  It's not a
              requirement of the procedure, but it helps with practical implementation.

Educated For What? William Easterly
  - human capital == education?
  - Everyone thinks education is the silver bullet.  Easterly says that just
    providing people with education without having technology for them to use
    once they graduate, or with still allowing corruption to run daily life,
    will not lead to additional growth.
  - In 1960, 28% of world countries had 100% enrollment.  By 1990, the median
    country had 99% enrollment.  In same time period, the median college
    enrollment per country went from 1% to 7.5%.
  - In africa, there was a blowup of education (human capital),
    but no noticable increase in GDP (sometimes effect was negative!).  In 
    asian countries, like japan, singapore, etc., growth did correlate with
    education, but at a lower rate than the disasters of lack of growth in africa.
    (See chart on page 6 of pdf). Eastern Europe has more years of education 
    than western or america, but smaller gdp.
  - Median output per worker of poor countries has fallen from 3% in 1960s
    to 0% in 1990s.
  - Other studies found no relationship between increase in years of education
    and increase in GDP growth, even when controlling for factors that lead to
    growth.
  - Also, no relationship between increase in educational growth and gdp growth
    even if you cut out Africa.  Also get a noneffect if you use average years
    of education.
  - Initial level of schooling does relate positively to growth, but economists
    say that this is temporary: higher physical capital relative to
    human capital will lead to increased growth until the two are relatively equal.
  - Another study found that 1% higher in gdp growth is explained only
    .06% (less than 1%) by the growth in human capital.
  - Another study finds problem with education growth->gdp growth model.  If
    human capital growth correlates with gdp growth, then you need to keep increasing
    human capital to grow.  So young come out with more growth, and thus produce more,
    leading to higher wages for the young, which you don't see anywhere.
  - Another idea: if initial education is correlated to gdp growth, then maybe it's
    just people valuing education more in a country where there is more value
    to go around, so this leads to higher initial education, not the other way around!
  - Mankiw showed that income disparity is explained about 25%-33% by physical
    capital, but when you combine physical + human capital, you explain 80% of
    income disparity in Solow model (which predicts growth as coming from
    outside capital increases).
  - One problem was that poor countries grew slower than other countries.  Mankiw
    controlled for savings rates (in physical and human capital) and found that
    saving a lot made you move toward being more rich, and not saving a lot led
    you toward being more poor.  Being poor and moving toward being rich made you
    move faster, actually.
  - Mankiw assumed physical capital can cross borders, whereas human capital has
    a harder time of doing so.  So if skilled labor is not available in those
    countries already, adding machinery and technology by investment will result
    in poor returns relative to giving them to skilled countries.  So that might
    explain why capital flows to rich countries more than poor ones.
  - Criticism by Easterly: Mankiw used secondary education as the determinant
    of change in income.  Easterly said that if you use primary education, which
    has less variance (more countries have it at a basic level), you can explain
    _less_ of the variation in income than with the secondary education.  I guess
    this can mean secondary education is the basic need and primary education
    is just a necessary step, or that Mankiw was artificially adding variance.
  - Compares india + US to show that while Mankiw predicted that skilled would
    move to poor countries, where skilled labor is less abundant.  The reverse
    is true, due to wages being higher in US.  So Easterly doesn't believe
    that savings (human capital=>education) leads to higher income, which would
    blame the poor for their poverty.
  - His idea: In economies where government intervenes a lot (controlling exchange
    rate, inhibiting growth, etc.), the best job for a silled educator is to lobby
    government to change its course.  But this job doesn't create growth, it just
    redistributed current income!
  - Similarly, pressuring people to study with no investment in the future won't
    give people a reason to learn.  Instead, you should invest in the future of
    growth, which will show people the light that education can bring them,
    and bring parents/teachers/students in line on what education can bring, thus
    causing everyone to work hard toward an educated populace.
  - Corruption and low teacher salaries can't be fixed by mandating education.  If
    you don't fight corruption and incite teachers to teach, you won't get necessary
    materials to students or have people to deliver content to the students.
  - Teaching (in Pakistan and elsewhere) is often used as political patronage.
    This means there are too many teachers, and each is underpaid.  That makes
    mandated schooling ineffective.
  - High skills go with high technology.  Don't create a supply of knowledge for which
    there is no demand!
  question: Shouldn't you check increase in education today with increase in
    growth 50 years from now, after a person has a chance to use it?
  question: Has anyone said that education _growth_ will lead to gdp growth?  I
    would have thought they would say _education_ leads to gdp growth, and that
    growing education is just a necessary component.  Why does he focus on the
    growth part of things?

The Primacy of Education, Anne Case
  - We know little about impact of education and educational initiatives, even
    though economists will turn to it as a motivator for growth.
  - Educated workers earn more than uneducated ones all over the world.  The causality
    is not fully understood---it might go in either direction.  Additionally, education
    quality is different in different locations, so six years of schooling doesn't
    mean anything across countries, etc.
  - Even if you compare siblings w/ different educational levels to check impact on
    income, you don't know how the factors for one to get more education might affect income.
  - Duflo studied indonesia 1973-78 to find that 61K extra schools resulted in more education,
    and that 1 year of extra school resulted in roughly 10% economic return.
  - Psacharopoulos did the same thing earlier, and also found that girls(12.4%)
    benefit more than boys (11.1%)
  - Benefits of education rise with a demand for the skills once you leave.  Rozenweig
    showed this for farmers w/ primary education in India during green revolution.
  - Duflo followed up her own work by showing that education had larger effect in
    areas of the country that saw more building.
  - Even after education is increased, economic development can lead to increases in
    growth for those that got educated.  China initially tought more students more content,
    but had same growth as India until market reforms in 1979.
  - Education and increased health (1 year leads to 8% reduction in mortality in men)
    are correlated, but it could also be that being healthy convinces you to get more
    education.  Recently, a causal effect was found in study of compulsory schooling
    in US leading to higher health.
  - Education and reduced fertility are also noted.  Some say that kids in school are
    either too busy or too expensive to raise if they aren't working and getting
    schooling, so parents have less of them.  Other say women being educated
    helps decrease fertility as being more educated increases opportunity cost
    of not being on the job market.
  - Causal relationship from education to reduced fertility not seen yet.  There
    may be other reasons, like women in school may just be highly motivated.
  - Two studies, one about europe and one about india, show some causality, but it's
    not a totally strong argument.  In india, 10% increase in education correlates
    with .2 less children.
  - So what do you do with marginal money to put into education? Hard to study, since
    most education is funded around responding to children's needs rather than just
    inputting some factor blindly into the system.
  - Some argument about class size relating to optimal results and bad students
    disrupting class, but it's very confusing.
  - Experiments are important.  Just taking correlations leads us astray sometimes.
    Flip charts in Kenya were correlated w/ increase in test scores by 20% of std. dev.
    When they did a randomized experiment, there was no effect.  So presence of flip
    charts is caused by something else which causes performance, most likely.
  - Natural experiments may help in 1) avoiding unwanted differences between schools
    and 2) preventing "better" parents from selecting their kid sinto some program.
  - 1999 study on Maimonides' Rule in 12th century dictating class size.  Class
    can grow to size 40, and at 41, it gets split into two classes (21 and 20).
    Mandated smaller classroom lets you study how different-sized classrooms
    change performance controlling for other factors.  Reducing class size by 10%
    correlates w/ a .25 std. dev. increase in 5th graders' test scores.
  - South Africa because of apartheid blacks got worse schooling (larger classrooms)
    than whites.  Results in districts controlling for household background variables,
    which also had powerful effect on outcomes, saw correlation between
    teacher-pupil ratio and enrollment, educational achievement, and test scores
    for numeracy.
  - Believes all of the variables are too entangled, and that only reasonably random
    data and experiments can measure anything, and even then causality will be hard.
      - DPEP in india was to measure improving education for females, etc., but
        ended up selecting sites that would show change in reasonable time, so
        ruined data.
      - PROGRESA in Mexico did randomized rollouts due to low budget initially,
        which helped get lots of interesting dependable correlations.

Remedying Education (Randomized Experiments)---Banerjee, Cole, Duflo, Linden
  - randomized experiments conducted in india
  - remedial education hired young women to teach lagging students basic literacy
    and numeracy: +.28 std. dev.'s for all students in treatment schools
  - computer-assisted learning program focusing on math increased math scores by
    +.47 std. dev.
  - after programs ended, gains stuck with them, fading to .1 std. devs after
    a year compared to control group.
  - universal primary education important, but quality is dismal in developing
    countries for the poor.  most students in india enrolled in school, but 44%
    of children 7-12 can't read basic paragraph, 50% can't do simple subtraction.
  - it's not always enrollment---even when enrollment works, you still get poor results.
  - given poor conditions, enrollment != learning.
  - know how to get students to attend school, but how to teach is harder.  randomized
    evaluations show that putting money into additional teachers, textbooks, and
    flipcharts does not raise avg. test scores.  Need to focus inputs on unmet needs.
  - the inputs (teachers, textbooks, flipcharts) may help students who do well on pretests,
    suggesting that they just don't help students who are too far behind and don't
    have educated parents to help them with hard topics.  need to fill this niche.
  - remedial program raised everyone .14 std dev in first year, .28 in second.  raised
    bottom third by .4 std devs in second year.  did this by giving them personal attention
    for two of the four school hours each day.  no discernible impact on classroom peers,
    who ended up in smaller classrooms.
  - computer-aided learning raised scores by .35 std dev in first year, .47 in second,
    equally for all students.
  - remedial education takes young woman (balsakhi---child's friend), who gets 15-20
    students to teach core literacy/numeracy skills to, and only needs 2 weeks
    of training to get the job.  balsakhis are paid little $10-15/month, and stay on avg.
    1 year.  this means program scales cheaply and doesn't depend on dedicated
    balsakhis to succeed
  - computer-aided learning (CAL) didn't get a lot of correlation w/ success before.
    teachers hired got 5 days of training.  each child shares computer with another
    2 hours each week. children played educational computer games. instructors didn't
    provide general instruction in math, just pushed students to play challenging games
    and provided help in getting through tough parts.
  - balsakhi design: 122 schools.  98 participated in first year.  half got 3rd grade,
    half got 4th grade.  second year, got a balsakhi in the grade they didn't have it,
    and remaining 24 schools got randomly assigned to group A or group B.  no redistribution
    of students to classes w/ balsakhis was noticed or likely given indian school organization.
    also tried to repeat results in mumbai for 2-,3-rd graders.
  - CAL was in 55 schools with 56 acting as control for a year, and then assignments
    between these schools swapped (CAL vs. control).
  - to test results, students given a pretest and posttest on basic competencies
    for years 1-4.  attrition was low, but if it happened, researchers tried
    to get student to take the missed test at home.  pretests showed that randomization
    was successful between control and treatment groups.
  - children do horrible on baseline pretests, but better in mumbai than in vadodara.  post tests
    show the results already noted for balsakhis, and mumbai supports vadodara data,
    except for verbal knowledge, which was way higher in mumbai in the pretest, so
    the result is harder to find.
  - CAL had discernible effect on math, and no effect compared to control group on
    verbal, which makes sense, since the content supplemented was only math.
  - CAL has larger effect on math scores but Balsakhi has greater overall effect.
  - both balsakhi and CAL help lower end of ability distribution more than the top end.
    balsakhi, which addressed bottom end more, had a larger effect on bottom end.
    the top end benefitted from both, but less so---this is good, positive effect
    on everyone, but brings up the lower kids more.
  - longer-term effect still exists a year after study ends (+.1 std devs), but that's
    a big decrease.  plan on tracking down some students in a few years, but you can
    either be encouraged (they learned something significant that stuck) or discouraged
    (in a year, they might forget everything!).  Maybe results were just a Hawthorne
    effect, where students were grateful for getting a Balsakhi or CAL.  Longer-term
    studies are required.
  - balsakhi cost $2.25/student/year, and $0.67/std dev.  CAL cost $15.18/student/year,
    but balsakhi is 5-7x more cost effective than CAL.  keep in mind that both are cheap!

Health care in Rajasthan---Banerjee + Duflo
  - 2004 world bank development report: "Social services fail the poor."  This is the
    case in India.  But what to do/improve?  Study rural Rajasthan for answers.
  - Seva Mandir is an NGO, runs several services in Udaipur district.  In health, they
    have helped various villages and trained people in basic health care practices.  Didn't
    feel productive, so got Banerjee to help analyze what to improve on.
    Study employed randomized evaluation.
  - Studied 100 hamlets under Seva Mandir's watch in Udaipur.  Seva Mandir tends to
    help only the poorest villages.  Stratified villages by distance to road,
    and 50 of them were more than 500 meters from road.  Selected groups within each
    stratum randomly weighted by population.
  - Hamlet==group of houses + community center.  1-15 hamlets==village, mean is 5.6.
  - Four components to study.  Village survey, including census, infrastructure, and health
   facilities.  Facility survey, activities, types/cost of treatment, medical treatments,
   physical infrastructure of facilitues, and studies of Bhopas, or traditional healers.  Weekly
   visit to all facilities in village.  Individual family surveys of 5700 individuals, 1024 families.
  - Weekly survey made sure facilities that should be open are, and that nurses/doctors
    that should be there were.
  - Average household expenditures in study are 470 Rupees per capita, 40% live 
    below official poverty line vs. 13% in rajasthan.  Few durable goods, 21% households
    w/ electricity, 46% men self-report literate, and 11% women.
  - 51% men, 56% women are aenemic, and 93% men/88% women are below BMI for low nutrition in US.
    Average person's breathing power indicates difficulty breathing.
  - Lots of self-reports of various sicknesses in last 30 days.  30% or more would find it
    hard to walk 5 km or drawing water from a well.
  - Bottom third of income scale have lower bmi/lung capacity/worse aenemia than top third,
    but top third self-reports more sickness.
  - Life-satisfaction measures don't measure their self-reports as especially unhappy.  Know
    they are poor (70% report being 3/10 or worse), but few are unhappy (9%) with life.
  - Should be one nurse-staffed subcenter per 3000 people, but they measure 3600.
    Primary health center per 48000 people w/ 5.8 medical personnel, 1.5 doctors.
  - Private options include medically trained people w/ grad degrees, traditional
    birth attendants, and pharmacists w/ no formal training.
  - Traditional options include exorcism people, herbalists or a mix of both.
  - Nurses in subcenters have highschool + 1.5 years of training.  Doctors in
    health centers have at least GP training, but many have specialist training.
  - 27% of doctors described as main provider have specialist degree above medical school.
    28% self-report degrees, though many in traditional stuff.  The rest don't claim
    formal degree, though many are pharmacists (compounders) or have some medical training.
    In local parlance, they are called Bengali doctors.
  - Often untrained or minorly trained staff see patients, 67.2% of which have no formal training.
  - Private doctors---36% have no college degree, and avg. 11 of 12 highschool years of school.
    Traditional healers have no formal training, and 4-5 years of school total.
  - Median distance to closest public facility: 1.53km.  Closest hospital is avg. 6.7km.
    Median traditional is .62km, and probably closer since underreported.
  - Govt. doctors are free, but if above poverty, must pay for medicines/tests.
    Subcenter costs ~33Rs. Bengali doctor 105Rs. Hospital 138Rs.  Qualified private
    doctor 179Rs. Traditional healer ~131Rs, as you bring a goat/chicken.
  - Each facility has needles/syringes.  20% of aidposts and 33% of subcenters
    have no stethoscope or blood pressure gauge or thermometer or weighing scale.
    Quarter of sub-centers have a sterilizer.
  - No subcenter has water, <10% have toilet or electricity or fans.  45% of rooms leak in rain.
  - Subcenters/Primary health centers/Community health centers should be open 
    6 days/week, 6 hours/day. Weekly survey indicated 45% of personnel absent in 
    subcenters/aid posts, 36% in PHC/CHCs, and likely not due to staff outreach in community.
  - Subcenters open 44% of the time during regular open hours. Being farther from road
    meant being open less frequently.  Being close to Udaipur/water/electricity don't have
    an impact. For the most part, there was no predictable day or time of day for
    the closures, so closure was -frequent- and -unpredictable-.
  - Avg. visits/month/person to health facility was ~.5.  A bit less for poor
    and a bit more for top third of income dist., but not a crazy difference.
    Most visits were for minor things like Diarrhea rather than chest pains/bloody sweat.
  - Children fully immunized age 1-5 are about 2.5%.
  - 25% of visits to a public health provider.  Majority are to private facilities,
    and small amount to traditional healers.
  - Visits to facilities that have less absenteeism are more than visits to facilities
    that are more often unpredictably closed.
  - Avg. household health budget is 7%/month.  Poor spend same relative ammt.
  - Most money spent at private facilities, then traditional, then public.

Disease and Development---Daron Acemoglu and Simon Johnson
  - Many argue that improved health will lead to increased growth.  Give examples
    from sub-saharan africa getting rid of malaria, etc.
  - Measures of health like life expectancy are correlated w/ economic development/growth,
    but the causality hasn't been studied.
  - Health leads to individual productivity, but unclear of effects on larger society.
  - Study effect of general health conditions, proxied by life expectancy at birth,
    on economic growth, using values of life expectancy that have been increasing since
    1940 due to health/chemical/medical advances.
  - a 1 percent increase in life expectancy is related to an
    approximately 1.7–2 percent increase in population over a 40–60-year horizon
  - No significant effect of increase in life expectancy on GDP.  Actually,
    GDP/capita statistically significantly drops as life expectancy goes up, as
    more people are alive.
  - Interpretation: Increased life expectancy initially drops output as population grows.
    However, once jobs arise, as more people enter the workforce and capital
    accumulates, you see growth due to more output from more people in the workplace.
  - 3 major factors improved life expectancy starting 1940s.
  - The first factor was ajor drug/chemical
    innovations, like mass-produced penicillin.  In 1950s, it became widely available.
    Then other antibiotics helped combat pneumonia, dysentery, cholera, and
    venereal diseases, and helped prevent viruses like influenza from taking over
    weakened immune systems.  Vaccines helped prevent disease, and DDT started to
    wipe out malaria.
  - Second factor was establishment of WHO, which along w/ UNICEF (UN International
    Children's Emergency Fund).  Helped run public health/vaccination campaigns.
  - Third, international values changed to move major discoveries between countries,
    thus improving health faster.
  - Took 15 most important diseases and collected cross-country mortality rates for each
    before the 1940s.  Then collected major intervention dates for each of these diseases.
  - TB was largest single cause of death in 1940s.  Streptomycin in 1944 was the drug
    that really helped fight it well.
  - Pneumonia was also a big one, but vaccines and antibiotics fought it hard since late 1940s.
  - Malaria is the third biggest, and DDT was used as insecticide against it starting
    1940's and intensifying in 1955-57 by WHO.
  - Sampled life expectancy at birth and other ages, and total births from UN 
    and league of nations, mostly for after WWII.  For that info + GDP before WWII,
    used other people's studies.  Collected these statistics for 75 countries in Eastern
    Europe, Oceania, the Americas, and Asia.  Exclude Russia/Eastern Europe (bad GDP data)
    and Africa (bad life expectancy data) to get 47 countries.
  - To study effects of life expectancy on growth without getting reverse causality,
    use predicted mortality as instrument for life expectancy.
  - Study change in the statistics between 1940 and 1980, and also uses a few dates before that
  - See notes for complete picture
  - Fig. 3,4---change in predicted mortality 1940-1980 is negatively correlated with change in
    log life expectancy 1940-1980.  (predict you die soon=>low life expectancy).  Holds
    for base sample and low/middle income places
  - Fig. 5---Change in predicted mortality 1940-1980 is hardly correlated w/ change in
    log life expectancy 1900-1940
  - Fig. 6---Change in predicted mortality 1940-1980 becomes negatively correlated w/ change in
    log life expectancy 1930-1940
  - Fig. 7---Change in predicted mortality 1940-1980 is positively correlated w/ change in
    log population 1940-1980  
  - Fig. 8---Change in predicted mortality 1940-1980 is hardly correlated w/ change in
    log GDP 1940-1980, so this implies gdp/capita is negative

Jeff Sachs Vindicated---Dani Rodrik's blog
  - Sachs says distribute insecticide-treated bednets (ITNs) for free, other 
    say to do it at a nominal but positive price 
  - cohen and dupas do a randomized experiment, and sachs was right.
  - decreased number of takers from higher prices is not replaced by enough increased usage
    by owners.  also, free nets save more infants from dying from malaria.
  - charing for ITNs leads to more leakage of product to unintended audience because
    worker steal them to sell.
  - alternative idea: goal is to encourage people to sell bednets even after donors leave,
    which a nominal price would result in, so longetivity of the ITN industry is more
    important than just distribution and use now.

Free Distribution or Cost-Sharing: Randomized Trial, Malaria Bednets---Jessica Cohen, Pascaline Dupas
  - ITNs reduce childhood mortality by 20% in African countries where malaria is top killer
    of children under five.
  - Cost is $5-7 in USD PPP, which is too much, so countries heavily subsidize it.
  - Untreated households w/in 300 meters of treated villages get same benefits
  - Estimated that need 50% coverage to achieve strong community effects.  No affected
    region has reached this yet.
  - Largest data is from Kenya survey 2003. 19.8% of households had at least 1 bednet,
    6.7% had a treated net, 12.4% of children under 5 slept under net, 4.8% under treated net.
    6% pregnant women slept under net night before, and 3% under treated net.  Ownership
    has increased since 2003, and may be at 30% for long-lasting ITNs.
  - one argument is that freely disitrubing nets results in resources wasted to those that
    will not use or need product.
  - no evidence in randomized price trial in Kenya that people will misuse them based on positive price.
  - positive-priced nets don't go to people that need them more (sicker people)
  - 75% drop in demand when price goes from $0 to $0.75
  - items w/ positive externality should be subsidized.  this is true for things whose benefit is
    independent of usage.  bednets require use, and so this rule doesn't necessarily hold.  charging
    even a minor cost could have a selection effect for those that need it, psychological effect
    about sunk costs, and more usage due to interpreted higher value.
  - alternatively, it reduces demand (bad) by increasing price, especially when it was near 
    zero.  also, might bias against the poorest, which may be most susceptible.
  - 20 clinics, 4 as a control, and then four groups of randomized prices between 0
    and $.60 (90% subsidy, still!).  To see if women who needed it more bought it,
    they checked for aenemia in pregnant women, and tested usage by visiting subsample
    of homes later in study.  Study population was women going to clinics for prenatal care.
  - Clinics for $75 or equipment of their choice per month if no evidence of leakage
    of nets to black market was seen.  Leakage happened at 4/11 clinics that charged
    some money, and at none of the clinics that gave it for free.
  - To separate selection effects and sunk cost, they also offered a lottery to some women
    to get a cheaper net.  This would measure the effect of sunk cost.
  - No large drop in demand from 0 to 15 cents.  But from 0 to 60 cents, demand drops
    60%.  Estimate that at market prices (75 cents), demand would drop 75%.
  - No clear relationship on who would use the net more (high payers or free nets).
    Usage is highest among women with free nets in their first prenatal visit.
    Free nets not more likely to be sold than subsidized nets.
  - No evidence that sunk costs had an effect on usage.
  - Women who pay more for bednets are no sicker than those at the control clinics.
    Women who get the free nets are healthier than baseline population, perhaps
    because of the incentive effect of free nets on followup visits.
  - While results need more data for full significance, the effective coverage of
    the bednets is higher when they are free, and the people lost to decreased demand
    hurt the social benefit of the nets.  Free distribution is at least as cost-effective
    as heavily subsidized nets, and number of children's lives saved is highest for free nets.
  - They analyzed the demand side, but there's an argument that free distribution is
    subject to corruption.  They didn't study this, but didn't see significant evidence
    of diversion of the ITNs to other beneficiaries, and for free and buyers, the retention
    rate of ITNs was >90%.
  - Results not consistent w/ water-treatment maybe because the other project didn't
    study free distribution as an option, and because ITNs have 3-year longetivity,
    whereas treatment only lasts 1 month, so selection effect maybe higher.
  - Results only apply for health products that are heavily advertised to the public,
    such as ITNs by the health ministry of Kenya.
  - Figure 1a (page 31) shows decreasing demand for ITNs as price increases.
  - Figure 2 (though not a significant result) shows that middle prices saw
    least usage, then free saw more usage, then 60 cents subsidized saw the most
    usage, but it was all within small margin and not statisitically significant
    because of low purchase rate for 60-cent group.

Ending famine by ignoring the experts (nytimes)
  - in 2005, 5 of 13 million citizens of malawi needed food aid
  - subsidized fertilizer at the chagrin of western countries who say subsidies
    should stop, but those countries do it was well, so malawi kept doing it, and
    now (2007) exports food to other countries
  - farmers couldn't let land lie fallow since their families couldn't eat, so crop yields
    dropped
  - in 80s and 90s, donors said grow cash crops, sell them, and buy food by importing it.
    trust private markets, not govt. intervention.
  - good fertilizer + good seeds + good rains led to increase in production in 2006-2007.
    basically they used techniques the rest of the world other than africa had been using
    forever, and the only way to make it accessible in malawi was subsidies.
  - us+britain did study and found that subsidy program helped growth, not just rain
  - subsidy was 1/3rd price for fertilizer for an acre of land, and coupons for seeds for
    half an acre for half of farming families.
  - extra corn from $74M subsidy was worth $120M-$140M
  - perhaps now the govt. should wean people off of the subsidies, but for now they are
    giving people more say in who gets the finite subsidies.
  
Fertilizer Use in African Agriculture (introduction only)
  - rest of world uses more fertilizer than africa.  no question that they need it
    to keep up w/ population growth goals, but the question is how to promote it.
  - fertilizer promotion schemes should be sustainable and suggest appropriate levels
    of application of fertilizer, and try to find private-sector led solutions for sustainability
  - inherent lack of fertility + widespread soil nutrient mining through various
    sources has led to using less favorable soils for cultivation
  - organic would be preferable, but there's just not enough.  need to get govts. and
    development organizations thinking about how to get more inorganic production/use.
  - in 70s and 80s, several techniques were used: direct subsidies, govt.-financed input
    credit programs, centralized control of fertilizer procurement/distribution, and
    centralized control of output markets (to stabilize production/prices).
  - these resulted in high fiscal/administrative costs, govts. couldn't support them,
    and one-size-fits-all didn't apply to everyone.  led to 80s/90s fiscal deficits
    in govt., which led to liberalization of fertilizer sector, free market, etc.
    helped deficits, but reduced fertilizer use.
  - policymakers have started to think about how to at least help the rural poor,
    who suffer the most from food insecurity. give them fertilizer==give them more stability.
  - private sector often doesn't fill public sector void in fertilizer.  on demand side,
    this comes from lack of crop yield stability, lack of market knowledge of fertilizer
    prices, and high relative cost of fertilizer.
  - on supply side, private sector hampered by excessive regulation, abundance of taxes/fees,
    and rent seeking.
  - thinking long-term, don't want to influence w/ subsidies which only drop price
    now.  on supply side, interventions can lead to help fertilizer importers, manufacturers,
    and others that will help sustain market.
  - to strengthen demand, you can: strengthen research + trials + demonstrations of
    crop management, improving access to credit or grants, increase tools to manage risk,
    improving market information, protecting farmers against low and volatile prices,
    rural education/training, inveesting in agricultural resource base (soil/water)
    to increase effectiveness of fertilizer.
  - improve supply side by: lowering trade barriers, giving access to national/regional markets,
    reducing distribution costs by improving road/rails, improve business finance/risk management
    instruments, and improving supply chain coordination mechanisms/information systems.
  - fertilizer subsidies died after 60s and 70s when they didn't live up to expectations.
    they are making a comeback now, backed by Jeff Sachs, to get Africa out of poverty
    trap, kickstart fertilizer use, and provide food security.
  - political appeal is understandable: good way to distribute growth to the poor.
  - flipside---drains costs from training/development/seeds/irrigation/responsible
    crop management practices.
  - while long-term view is to promote private sector solutions, the report says 
    subsidies are not a bad idea when you combine them with other techniques.
  - suggests "market-smart" subsidies: demonstration packs, vouchers,
    matching grants, and loan guarantees
  - longer-term, need policy reforms to stimulate private investing, institutional
    reforms to allow commercial exchanges, infrastructure investments, research investments,
    capacity building, and improvements of agricultural resource base to allow farming
    to continue on land/water.
  - 10 principles: promote fertilizer in wider strategy, not alone. favor market-based
    solution for sustainability.  promote competition to drive down prices, increase quality.
    pay atention to demand to determine proper supply strategy.  insist on economic efficiency
    to get most bang for buck.  empower farmers to run their own farms instead of telling
    them what to do.  devise an exit strategy for government eventually.  regional integration
    to benefit from market size/scale.  ensure sustainability both institutionally and
    environmentally. consider pro-poor growth, sometimes even over other choices, to
    ensure poverty reduction and food security happen.

Nonmarket Institutions for credit and risk-sharing in low-income countries---Timothy Besley
  - Poor countries have it harder to deal w/ risk.  Infrastructure creates more
    risk of infection.  Agriculture plays larger part (more uncertainty).  Market
    tools for dealing w/ risk aren't there.  Legal system isn't there for enforcing
    contracts
  - Instead, nonmarket institutions pop up w/ little contractual obligation/legal
    system, but well-defined rules exist nonetheless
  - Most traditional risk-sharing studies have focused on agriculture, such as
    sharecropping (share risk between landlord/tenant).
  - Risk-sharing and credit often linked: credit can be replacement for working
    insurance system, credit/insurance becomes blurred once lenders let
    borrower have slack when payments are delayed by unexpected event, and
    since not all information is known, want to mix credit/insurance.
  - To avoid risk of fluctuating income, you can have savings or diversify jobs.
    It turns out low-income folks do respond to windfalls by saving and draw
    on savings when times are low, as one would expect.
  - Saving isn't perfect in low-income though.  savings instruments are scarce,
    inflation is a concern, and food costs fluctuate.  Familial obligations
    pull people toward not saving as well.  This means other sources of
    risk-sharing would be helpful.
  - Peer monitoring is often better-informed than market!  People in nonmarket
    contexts may know each other well, and thus moral hazards and information
    gaps that larger market players have are less frequent.  Harder to
    screw your neighbor out of money.
  - Sometimes, regional/neighborhood schemes have different enforcement/sanction
    techniques than banks/the legal system does.  Interesting to study these.
  - Example of Grameen bank---Group lending is not popular in USA, but there are
    various degrees of it in developing world.  Groups take on risk together,
    get more freedom in deciding how to allocate it.
  - Group lending can improve monitoring, since group members know each other.
    Analyzed as a game where you can default because someone more successful will
    pay off your debt, or else they will get dinged for it.
  - Credit cooperative is a different model, where a co-op borrows together
    and distributed money amongst members.  Exist even in developed world
    in the form of employee credit unions.  Relative importance of such an
    institution declines as economy becomes more developed.
  - Credit cooperatives don't always succeed.  May be because of covariant risk
    resulting in large failures, or because members can collude to hurt lender.
  - Rotating savings and credit institutions have members enter lottery to win
    the pot periodically.  Once you win, you are removed from the pot.
    Eventually everyone wins, and it serves as a savings vehicle that each
    individual would otherwise not have been able to achieve.
  - Little quantitative evidence exists, but perhaps 80% of Tiwanese adults
    participate, from cost of household appliance to several thousand USD.
    Exist even in devoped markets among immigrants or other groups that may
    be disadvantaged from credit pools.
  - Due to economies of scale, you do eventually see developed market players
    crowding out informal groups.

boston.com microlending article
  - Microlending---pioneered by Grameen bank, $8B in loans, usually under $100,
    98% repayment rate
  - About 150 million worldwide
  - J-PAL study shows that microcredit does not offer way out of poverty.  Helps
    entrepreneurs at the margins, but doesn't result in difference in
    income, spending, health, or education for borrowers.
  - Most loans spent on big purchases, weddings, or paying off other loans.  While
    most loans target women, there seems to be no gender difference.
  - Weak effect or no effect on growth, and interest rates are higher than we're used to.
  - Counterargument: they do work, since interest is cheaper than village lenders,
    and effect time will take longer than 1,2-year studies that were conducted.
  - Easy to paint image of entrepreneurs just needing $30 to start business.
    In truth, starting small business is often only way to make money, and
    what you might need for growth is steady jobs, reliable incomes in larger industries.
  - Grameen bank in Bangladesh started when Yunus realized that artisans were
    being controlled because they didn't have 22 cents to buy materials, and so
    merchants who gave them loans controlled rate so that they wouldn't make enough
    to become self-sufficient.  22 cents broke first bamboo seat builder out of
    the vicious cycle.  inspiring story, but little hard evidence for effects.
  - Speed w/ which microcredit spread made it hard to conduct randomized studies,
    as few villages lacked it.
  - Randomized study by changing algorithm to cause some borderline folks
    to randomly be denied in Phillipines.  Then they studied difference in
    both populations
  - Neither household income or spending rose.  People who put the money
    into business and not personal spending saw businesses shrink (perhaps
    because they fired weak employees that they owed financial favors to).  Profit
    rose a bit.
  - Gains accrued mostly for males, not the females usually targetted w/ microcredit.
  - Another study in India (Banerjee + Duflo) saw small, selective effect in Hydrabad.
    Bank was expanding there, so they could measure effect of expansion.
  - In Hydrabad, businesses w/ microloans saw profits rise a bit, and households
    who were judged to be predisposed to start businesses increased savings,
    presumably to save for some capital investment in starting business.
  - Less money spent on temptation goods (alcohol, tobacco, gambling).
  - No rise in household spending, children's health/education, and no effect
    compared to control group of women's role in household decisionmaking.
  - Again, short time horizon of studies (12 months) may not be enough.  People
    seem to be coming bakc for loans, maybe if only to repay higher-interest loans.
  - Either way, microcredit is not necessarily a way out of poverty, but just
    a tool to help you get by in it.
  - Perhaps instead need an infrastructure that supports _saving_ rather than _lending_

Miracle of Microfinance, evidence from randomized eval. (Banerjee, Duflo, Glennerster, Kinnan)
  - Pick half of 104 slums in Hydrabad to open new microcredit service, and other half
    don't get it.  Studied them 15-18 months after program.
  - Effects: households w/ existing businesses invest in durable goods, and profits increase.
    Households w/ propensity to become businesses reduce nondurable expenditures,
    presumably to invest in durable goods.  Households w/ low propensity
    see increase in nondurable spending.  No evidence for increase in health,
    education, or women's decision-making.
  - Microfinance institutions have seen 150M clients, w/ ~100M women.
  - Difference in differences hard to measure, since microfinance clients are
    self-selected, MFIs only pick some of them, and the client would likely be
    on a different trajectory absent microfinance.
  - Can randomize among slums/villages in a new market as they did.
    Harder to randomize amongst individuals, since there may be spillover effect.
  - MFI (Spandana) opened offices in 52/104 villages, each w/ avg. 68 households.  Studied
    differences after 15-18 months.  Other MFIs opened in many villages,
    but the 52 selected ones had 8.3% higher chance of getting a loan.
  - Spandana does not insist that the loan be used for business purpose. 
    In the case of 30% of Spandana loans the reported purpose was starting a 
    new business; 22% were supposed to be used to buy stock for existing 
    business, 30% to repay an existing loan, 15% to buy a durable for household 
    use, and 15% tosmooth household consumption (respondents can say more than one thing)
  - Model: 6-10 women in group, 25-45 groups in a center.  Women responsible for payments
    in group and in center.  First loan is 10K RS ($200, or if PPP adjusted, $1000).
    Loan takes 50 weeks to repay w/ interest, which is equivalent to 20% APR.  Repay
    and are eligible for 10-12K loan, increasing up to 20K.
  - Unlike Grameen, Spandana isn't as concerned with raising financial literacy,
    purpose of loans, etc.
  - Study started w/ 120 slums.  These were surveyed to exclude heavy construction
    worker/migrant population, since they aren't good candidates for these loans.
    Dropped 16.  Remaining 104 saw randomization of MFI introduction (introduced or not)

The Daily Grind---Collins, Morduch, Rutherford, Ruthven
  - follows families living on $1/day.
  - first family in bangladesh, house was flooded in ganges river flood.  moved
    to govt.-owned land, payed no rent in fire slum (always burned down), built
    house of bamboo/corrugated tin, had three children, dad is 37, mom is 29, oldest
    son 14.  toilet is local one set up by NGO, and electricity is few pennies/month
    for one lamp.
  - money is not just low but also unsteady---rickshaw driving is only possible 4 days/week,
    and good days bring in $2.50, while others bring less.  sons rag-pick and work in factories
    mom became made again.  a boarder paid them $7/month.  altogether, 7 of them make $1.90-3.15/day.
  - eat 1-3 times/day, but always at least once.  usually rice/lentils, but sometimes fish,
    and rarely beef
  - didn't take microloan in year of observation.  in general, savings/microloans are
    for short-term goals, not long-term thinking.
  - monetary instruments were used to ensure constant cash-flow, not saving
    assets to grow their balance sheet.  the idea was not to save money long-term,
    but to have food every day, even when income didn't come.  use financial
    instruments because they are poor, not despite being poor.
  - triple-whammy: income is irregular, small, and financial instruments don't always
    help household's cash-flow patterns
  - no extremely poor families studied live hand-to-mouth.  avg. savings:
    bangladesh $68, india $115, south africa $472
  - in all of these places, net financial turnover by pushing/pulling on financial
    instruments in the above three regions results in cash flow order of magnitude
    larger than year-end assets.
  - most people earn income from multiple jobs, some at factories, but others from
    self-employment, casual employment, and petty businesses.
  - in south africa, 27% of households had govt. grants as first source of income,
    and grants made up 48% of studied income.  income goes to children, elderly,
    disabled, and pays for 1/4 people in family.  regularity of payments helps
    use more sophisticated instruments, and when grant doesn't come on time, hurts family.
  - seasonal labor makes earnings unstable.  farmers/traders make more money during
    harvest/celebration seasons.  Can make 60% of yearly income in 4 months.
  - farming---large farmers tend to have stable income, but small farmers affected
    more by weather, etc., sometimes only recovering 25-30% of what they expected.
    smaller farmers get worse land, which affects output variance.
  - small business owners sometimes buy products based on yesterday's earnings.
    earnings aren't always good for selling meat, for example, so when they
    want to sell tomorrow, they can either go to a moneylender to get temporary
    cash or sell yesterday's old stock, which results in competition from others.
    interest can get to 30%/month
  - some join savings clubs, rotating pot of money w/ regular contributions, but
    these fall apart in unpredictable ways sometimes.
  - even regular income, if too low, can suck.  small unpredictable blips like
    a child's illness can break the flow.  also, predictable income that's too short
    or too long apart (govt. grants) between payments can suck, so people enter
    savings clubs to lengthen or shorten the periods of payment.
  - even formal jobs (factory work) aren't stable.  hired as 'emergency contractors'
    in delhi, meaning you can be out of job for up to 4-5 months at a time.  makes
    remitances back to village at home sporadic and often 0.  get robbed while living
    in squatter housing, and get by only on goodwill of grocers/landlords.  that's
    in south asia.
  - south africa has more rigorously enforced laws, so formal labor and grants come
    in regularly.  this regularity in income makes it easier to get debt/loans, and
    so even the poor have similar debt-equity ratio to the less poor.
  - while places like south africa have transparent lending laws, people often don't
    borrow from formal banks/insurance companies.  most borrow from friends/family with
    small no-interest loans, and try to pay them back immediately, just to guarantee
    food every day.
  - every household surveyed made some attempt at savings, sometimes multiple ways
    by multiple family members.  they would often take loans even when they had
    small cash amounts at home---signals that they choose between financial instruments
  - monetary tools were often close at hand (in hut or with neighbors) and flexible
    (no payback date, etc.).  often didn't join microlenders because just wanted help
    saving, not borrowing.
  - microlending is often billed for small businesses, but one family didn't buy rickshaw
    because it was too risky---nowhere safe to park it.  instead they kept renting
    rickshaw and took loans to bulk up on rice, build a cupboard, make loans to others, etc.
  - while everyone studied in bangladesh had some savings, loans to finance temporary
    blips outnumbered savings withdrawls 4-1.
  - loans are usually informal w/ friends/family, from 88% in bangladesh (poor: 92%)
    to 94% in india (poor: 97%).  most of these loans are interest-free.
  - richer people manage blips w/ credit cards---poor use personal loans
  - some loans are reciprocal---you will help me one day.  others are obligatory---
    rich feel obligation to help poor.
  - moneyguarding---holding on to someone else's savings temporarily.  families
    often do it for local young workers who need to keep it in safe place.
  - money-sharing---get grant at different times of month from neighbor.  so you
    and neighbor pay each other a fraction of your check whenever you get the check.
    this makes money come more frequently.
  - opposite of money-sharing---ROSCAs in which you pool money together and get
    larger lump-sum less frequently.  some people view this as more useful than
    the more frequent small sums.  studying such practices might help designing
    formal instruments
  - formal financial instruments, as you can see, are often not used.  sometimes
    it's because using closer-to-home instruments is easier (less travel).  other times,
    too much paperwork isn't good for savers/borrowers/lenders (much of india/bangladesh
    is illiterate)
  - that said, informal system has issues that can make formal system better
    - unreliability---others may not have cash on hand
    - lack of privacy---neighbors know your problems, and if you mess up, you lose
      social capital
    - lack of transparency---while friends/family/neighbors are often trustworthy,
      they can also run off with your money or give you fake stuff in exchange.
  - takeaways for formal systems
    - take small payments, allow small, frequent repayments with somewhat loose 
      schedule (e.g. grameen bank)
    - flexible credit (sometimes short-term, sometimes long-term) for people
      like farmers w/ unpredictable income will be used as well.

Microinsurance: The next revolution?---Jonathan Morduch
  - even though natural disasters are frequent and public/govt. response is
    limited, it is hard to buy insurance plans in the developing world.
  - survey of nicoragua in 1998 after hurricane mitch: 21% of people drasitically
    reduced consumption as first response, and 18% as second most important response.
    89% received no help at all, and most of rest got NGO help, not govt. or private
    insurance.
  - poverty and lack of insurance co-occur.  in US, 44 million poor have no health
    insurance, and US farmers have insurance subsidized $5 for every $1 of theirs
    put into insurance.
  - avg rural income in china in 1990's saw 8% growth, but 20% in any year
    suffered from drought, flood, hail, pest infection, livestock disease
  - in many countries, the cost of revenue for insurance is higher than 1, so it's
    not wise to sell insurance.
  - several reasons for no insurance.  moral hazard of getting insurance makes
    farmers less likely to get best fertilizer, so likelihood of danger goes up.
    adverse selection makes it harder for insurers to find right premium: price too
    high for good farmers, but don't have enough information to charge different prices
    to different farmers. economies of scale hard to reach because on individual
    basis, most insured farms are small.
  - transaction costs are high, private insurers don't want to enter, and many
    farmers are illiterate/inumerate.
  - the grim situation sounds a lot like microfinance in the early days.  perhaps
    microinsurance is possible
  - to understand risks/generate good cost models, need to repeatedly survey 
    families over the long-term, with a large time dimension to the study.
  - hard to run these studies, as poor are resourceful to prevent risk, and so
    they invent their own methods/technologies to weather the storm.  this undercuts
    studies that aim to compare the cost of these approaches to the true 
    cost of market-based insurance.
  - if it was true that villages pool risk, then governments could focus on aggregate
    risks that affect entire villages, like floods and other natural disasters, but
    some evidence shows that even unofficial insurance isn't that developed/effective
  - in fact, Townsend found that a lot of insurance in developing world is not
    through risk-sharing, but instead through personal savings of grains, etc.
  - so there are many unknowns, but should we wait for the data before proceeding?
    morduch says no---we learned a lot about microfinance by comparing different
    techniques in different villages by the grameen bank and control villages
    _because_ the bank was established.  similarly, maybe we should run microinsurance
    experiments in the wild and learn by doing.
  - life insurance is easier than health or crop insurance, as the moral hazard
    to die is lower, and it's easy to measure event (death) happening over time.
  - yet there's no life insurance institution in developing countries.  most are informal
    like burial societies (cochin, india) where you pay 2 Rs/week, and when you die
    during year of being in program, you get 500Rs for each Rs you put in per week.
    sustainable as long as no more than 5% of population dies in a year.  yearly
    surpluses redistributed, and sometimes extra collections are necessary.
  - getting 1000 Rs for 100 Rs a year means you expect chance of family member dying
    is 10%/year.  If it's lower, then you may as well put money in the bank (if one exists)
  - but probabilities are hard for people to understand (invest more in operation if
    you are told you have 80% chance to live than 20% chance to die).  also, even
    if you understood probability, it's not that easy to measure survival rate over 10 years.
  - another problem: burial societies collect 4 cents/week for each member at meetings
    that happen weekly anyway.  insurance companies would have to pay someone to go
    door-to-door to collect 4 cents/week---transaction cost too high.
  - bank in china got around this by giving people savings accounts that resulted
    in insurance on death instead of interest.  but that means need a deposit of
    2000 Rs if bank earns 5% interest---a lot to put upfront.
  - can have middle ground---collect 16 cents/month, 50 cents/quarter, etc.  but
    that requires individual savings over time, which is hard in developing world.
  - finally, can attach insurance collection to collector already on the ground,
    like microfinance agents.  some companies do this already---insure your debt
    repayments in event of death, etc., but it's hard to do unless you're already
    in the microfinance industry.
  - health insurance is harder---moral hazard and adverse selection apply.  health
    microinsurers cover their ass by limiting payout---cover usual inpatient and
    outpatient care, but not AIDs, for example.  this reduces risk/variance.
  - while copayments and deductables decrease moral hazard and cover the really
    scary/unexpected events, patients want "first dollar" coverage---payments
    starting w/ the first dollar spent.  this costs more in transaction costs for
    lots of little spending, but some argue this is good because it avoids
    participant waiting until small problem gets larger.  avoid moral hazard by
    paying people back if they incurred no/little cost at the end of the year.
  - some issues aren't insurable, like price drops in crop sales (?).
    people prefer to just be able to save so their money isn't tied
    up in predicting every kind of unpredictable/uninsurable risk, so
    maybe savings is better than insurance.  
  - problems tend to come in waves (lose money selling crops, and then health
    event comes).  so maybe establish emergency loan fund for families that have
    paid premiums in good standing and are having a rough crop yield so that they
    can keep paying rates from loan so that health problem doesn't double up the
    issues due to temporary slip in paying premiums.
  - insuring agriculture may get to the core of issues more than life/health insurance.
    rather than insuring crop earnings, can insure against too much/little rainfall.
    this eliminates moral hazard/adverse selection, since it's out of hands of farmer.
  - benefit of helping more than just farmers---anyone who is seasonally affected
    can sign up.  also, it's easily measurable, so disbursements can come
    faster, which aids recovery.
  - drawback---because rainfall events will happen to lots of people in one area,
    you either need to have sophisticated international operation or reinsure
    with one to spread risk.  this means that unlike microfinance, you can't start
    small and locally and then scale up---you need to start big to spread risk.
  - in 2002, world bank partnered for $80M global weather risk facility.  this is
    good news, but it's only a pilot, and mostly concerned with whether it's
    practical to do for an insurer, not necessary its benefit on the insured.
  - rainfall insurance may help farmers, but famines and other issues often hit
    the landless poor the worst.  if landless can also buy rainfall insurance,
    then it's a good thing, as they have more protection in a natural disaster.
    however, if they can't (and they often can't since they don't have collateral,
    and insurers often work through established clients), then they are worse off,
    since farmers will be more stable than landless in times of crisis, and will
    drive prices even higher to profit from dearth of food.
  - summary of big issues to solve for microinsurance to have a chance:
    - collecting data on risks
    - collecting data on workable premiums
    - solving need for reinsurance to spread risk
    - need to cut cost of small transactions
    - educating people on benefits of insurance & nature of risks
    - subsidizing insurance for the poor

Land Tenure and Urban Poverty Alleviation---Mooya + Cloete
  - Some say owning property gives the urban poor security and provides them
    with collateral that they can take loans against
  - Others say they already have de facto permanence on their land, and that
    they don't use loans for anything other than consumption anyway
  - property->collateral->credit->income is nice, but there have been few
    studies to see whether land titling helps.
  - Three rights you can associate w/ land: alienate/sell to others, sole use of
    land, and right to bequeath land to others.  This is a social construct + 
    set of regulations enforcing it.
  - Theoretical benefits: long-term planning encourages investments. credit markets
    can use land as collateral.  makes commerce between strangers easier, expanding opportunities.
  - land titles have five benefits: open market to larger set of buyers. allow market
    to form around demand rather than defensive practices. make mortgage finance eaiser
    to manage.  make information assymetry go down as more is publicly known.
    formalizes the tax base of the municipality.
  - now it's time for empirical evidence to back these ideas.
  - Besley showed in 1999 that in Ghana, land tenure: encourages investment, 
    encourages formal access to credit, lowers cost of exchange.
  - Alston et al. in Brazilian Amazon show that land titles increase land values
    and wealth, and create incentives for long-term planning.
  - Evidence from Peru shows that titles make households spend less resources on
    informally protecting land.
  - Ecuador: land titles increase land value by 24%
  - Vietnam: land title transfers result in 3-10% more cash than ones w/ incomplete titles
  - Nicaragua: titles increase value by 30% and increase propensity to invest.
  - Indonesia, philipines, cambodia, brazil similarly 25-73%
  - Evidence from Jordan, Tanzania, rural south africa suggest that tenure increases
    inequality and landlessness.  Evidence from Texas shows that regularizing titles
    don't correlate with increasing land vaues.  These are the minority---most evidence
    shows positive results.
  - w.r.t. land ownership->collateral for credit, the evidence is more clear and less
    positive. botswana, trinidad, zambia: people don't hold their land as collateral often.
    receiving full title in texas doesn't result in much use of home as collateral.
    bogota: titles don't change availability of formal finance. mexico: reduces flexibility.
    peru: largest such study shows titles don't increase private bank approval rates,
    and does little to public funding approval rates.
  - why the low effect on collateral for credit? three implicit assumptions
    must hold for there to be a measurable effect:
    investment decisions are hitting credit constraints, households agree to
    use land/house as collateral for financing business activities, and
    financial institutions must accept the collateral for financing.
  - this might not be the case in informal rural areas.  titling confers benefit
    when informal land transactions are already common and there's a credit
    market ready to take the collateral.  part of that is believing they
    can enforce foreclosure, which is not the case for many poor, so it's not
    worth lending to them.
  - last assumption to be tested---formal titles increase market activity.
    there's some evidence for this: formal titles make people more comfortable,
    so number of exchanges goes up. on the flipside, this is based on assumption
    that informal titles are inefficient.  in ghana and vietnam, studies show
    that beaurocracy and cost associated with formal route is inefficient, and
    informal trades are more efficient.
  - so policy ambiguity, procedural complexity can sometimes lead people away
    from formal titles even when the option exists.
  - in some places, land is not viewed as marketable.  it's a source of family
    and community stability, and is something you bequeath to your relatives later.
  - summary so far: titles probably increase land values, increase investment in
    land.  little or no effect on credit markets.  sometimes increases market
    activity, but others has no effect.
  - big question: does land ownership aid poverty alleviation?
  - capital assets framework lists various kinds of capital 
    (human, social, financial, etc.). because it's such a large part of the
    capital of many poor households, tenure security would give them 
    reason to invest more their house.
  - what's needed to achieve this: well-defined, secure, and enforced property rights.
    liquidity (frequent impersonal transactions).  low levels of uncertainity in
    transactions.  low levels of transaction-specific investment.

Geography, Institutions, and the Poverty of Nations---Acemoglu, Johnson, Robinson
  - Two contending reasons for poor countries being poor: geography and institutions
  - geography/climate/ecology usually has three flavors: climate determines
    work/productivity, geography dictates technology (especially agriculture),
    and disease burden in tropics higher than in temperate zones.
  - good institutions need three things: enforcement of property rights, constraints
    on powerful groups/elite, and equal opportunity to invest in human capital and
    productive economy.
  - their findings: institutions affect both the slice of the income pie and 
    its distribution among citizens.  study this through lens of colonialism.
  - geography helped decide where people settled vs. extracted resources, but
    institutions seem to have mattered the most in the outcome today.
  - geography matters: most places w/ less than 1/20th of USA income are close
    to equator.  tropical agriculture is less productive than in temparate regions.
    diseases seem to be worse in tropics (thus called tropical diseases).
  - but it's more correlation than causation, just like swamps don't cause malaria,
    mosquitos do
  - similarly, property rights and income are correlated, but which causes which,
    or are there external factors?
  - conduct "natural experiment" by coding different historical events in colonialism
    to find out which of the two hypotheses is right
  - colonization by europeans started in 15th century.  since it doesn't affect
    geography much, if it was the key factor, we'll see places w/ institutions that
    set up property rights, etc., to be prosperous.  otherwise if geography is 
    important, we'll see places that were rich before colonialism stay rich.
  - congo, slave trade in carribean, forced labor in mines of latin america
    were extractive institutions.  none of the three tenets of health institutions
    are met.
  - australia, NZ, canada, usa were more protective of basic rights and restrictive
    on elites.
  - no evidence or geographic hypothesis: mughals/aztecs/inca were richest before
    1500, and are poorest now, so geography didn't give them persistence.
    less-developed countries such as usa and australia are some of richest economies.
  - one good proxy of prosperity today is urbanization.  % urbanized in 1995
    correlated well w/ log GDP/capita in 1995.
  - to combat the geography hypothesis, % urbanized in 1500 (before colonialism)
    is negatively correlated w/ log GDP/capita in 1995, meaning that people who
    were well-urbanized then are poor today, namely those in the tropics.  so
    while they were the richest, they are now the poorest.
 question: doesn't looking at % urbanized in 1500 vs. log GDP/capita 1995 not tell
   you anything about prosperity in 1500? need urbanized vs log gdp/capita, both
   in 1500 to see if urbanization means prosperity then.
  - but what if some time-varying factor of climate/ecology/disease led the change?
    one example is "latitude-specific technology hypothesis," which says that
    agricultural technology such as plows, crop rotation, domesticated animals,
    and high-yield crops are more favorable to the temperate areas than tropics.
    this isn't backed by evidence though: reversal in income seems to have happened
    before europeans arrived, so it was probably not geographic.  also, the technology
    would have caused a change of fortune when they showed up, but at least in
    america, it didn't, and instead happened in 1900s post-industrialization.
  - european colonialism shows evidence for institutions hypothesis.  places
    where they extracted resources got totally messed up, and places where they
    settled became prosperous
  - in places that were already densely populated and prosperous (mughals, incas,
    aztecs), they set up extractive colonies to destroy that.  in places that were
    dense and unpopulated, they settled and set up societies with more rights
    (for their own people).  so that's a precise reverse of fortune, all designed
    by institutions
  - further evidence: no major changes pre-colonization 1000-1500.  then 1500-onward,
    changes only in places affected by colonization
  - similar change in urbanization and industrialization: before 1800, urbanization
    was opposite where it happened post-1800.  industrialization/capita moved from being
    higher in India in 1750 to being higher in usa, australia, NZ.
  - industrialization seems to have reversed roles significantly.  industrialization is
    just the kind of thing that needs good institutions in place to happen:
    market transactions supported by law/order, people investing in skills/human
    capital protected by institutions of private property.  countries that stayed
    under control of elite plantation-owners didn't manage to switch over: they
    didn't dopt industrialization as much.
  - another factor in settling: settler mortality.  in temperate regions,
    malaria/yellow fever not as prevalent (10 deaths/1000 settlers).  50 in india,
    150 in jamaica, and 500 in west africa.  so a lot less likely to settle those
    areas, and instead set up extractive societies.
  - settler mortality in colonialism seems to be negatively correlated w/
    protection against expropriation 1985-1995 (protection against land being taken),
    meaning less dead settlers->more property protection.
 question: i bet settler mortality rates are also correlated w/ native mortality
   rates until 1950's at least!
  - they argue against the above question, saying europeans were more susceptible
    to disease than natives, and both had similar mortality rates in their home
    countries, so it's not that natives have higher mortality than europeans.
    what about morbidity, though---malaria only kills 1% or so if its inflicted.
  - so is geography unimportant in development?  yes as a direct effect.  but it's
    also important for four reasons.  first, geography and disease _do_ matter for economic
    outcomes (e.g., no agriculture at the poles, and disease hurts you in schooling, etc.),
    but they can't seem to explain large cross-country outcomes.  second,
    just because it doesn't affect differences today doesn't mean it doesn't
    affect them 500 years ago, where geography and disease probably played a larger
    role in technology.  third, geography had an affect on the institutions,
    so it matters in setting up the "correct" institutions.  finally, even if
    it had no effect on income, it still have effects on social welfare
    by way of health, etc.
 note: this is the first time they mention morbidity, which is such a larger part
  of malaria than mortality.

Whither the World Bank and the IMF?---Kreuger
  - World Bank and IMF, called International Financial Institutions (IFIs),
    are in a different world from 50 years ago at bretton woods where they were established.
  - sumarize founding ideals, successes, and failures since
  - had a good 25 years, and then as more developing countries became industrialized,
    the two IFIs started to merge in visions/actions.
  - argues that developing nations are doing well enough, and that WB should focus
    on poor subsaharan, south/central american countries
  - IMF has to refocus on how to provide crisis management in a world w/ so much
    private capital.
  - The IFIs were started postwar to handle issues inspired by great depression.
  - IMF was funded to avoid beggar-thy-neighbor (bad trade tarrifs that made depression worse)
    policies and work in a fixed but adjustable exchange system.
  - needed a bank to handle loss of private capital to promote investment
    in areas w/ high rates of return and low savings (reconstruction countries
    and developing nations).  The international bank for reconstruction and development
    (IBRD) was started for this purpose.
  - The IMF, which was IRDBs sister organization, was to just provide short-term
    trade/finance related help, and to put currency controls on certain countries.
  - initial purpose of IMF: member countries could come if they had currency problems.
    imf would loan to them if they were stable, or if it was really bad, allow
    them to change their exchange rate.  funding was at near-market rates for
    3-5 year loans.
  - imf could advise in interest rate changes/capital controls.  its documents were
    largely private (necessary to prevent gaming the system), but lack of transparency
    also annoyed people.
  - fund is usually dealing w/ countries w/ high expenditure/low income, the decisions
    to fix economies are usually correlated w/ slow growth/recession.  this annoys people
  - up until 1960s, IMF's purpose was unquestioned.  then people started to question
    idea of a fixed exchange rate system.  by 1973, rates were no longer fixes, and imf's
    role was relegated to developing countries.  then in 1973-74, oil crisis gave it new purpose
  - irdb was founded to make long-term loans to good development projects.  it can
    either borrow from private investors and lend to countries, or guarantee loans
    they take out.  in practice, it lends but doesn't guarantee
  - irbd was eclipsed by marshall plan of USA ($1B vs. $6.2B), so it focused more
    on development projects.  it made loans on commercial terms, which were often
    too high for developing countries, so international development association (ida)
    was formed to give "soft loans" to those countries.  together, irdb+ida+others
    made up the WB.
  - WB became powerful not only as credit source, but also as advisor to inexperienced countries
  - in 70s, US put more money into WB and less into its own development efforts.  at the
    same time, regional development banks (asian, inter-american, african) started.
  - wb is typically seen as less negative than imf.  wb technical staff usually work
    with agriculture minister to implement projects, and extra money is appreciated
  - whereas imf comes in to distressed country for short-term fix, wb is seen even
    by stable developing countries as source of growth
  - after fixed-rate exchanges ended, oil price increases, and 1980s recession,
    developing countries went to the IMF for advice.  WB kept lending in 70's,
    but for success, it needed macroeconomic stability.
  - by the 1980's recessions, world bank started offering stability programs
    to countries, and IMF started lending beyond its normal term length,
    since both realized they needed the other's features for stability.
  - started to lend in contradiction (argentina---imf didn't, wb did),
    and decided to share some information on places where they worked side-by-side
  - 1997, wb started to focus on soft issues like women's roles, environment, etc.
  - hard to evaluate imf/wb, because they loan so little compared to aggregate
    investment, and they do it to the hardest-hit countries.
  - wb reports rate of return of 10% on projects, which is pretty good (for the projects)
    but its lending is a small piece of the pie.  1970: $.7B wb vs. $6B private investors.
    in 1996, it's 7.3 vs 244
  - a few fund evaluations show initial negative growth but ultimate reduction in inflation,
    and more positive long-term prospects in growth than negative ones.
  - imf has tied academic research + policy decisions for many years, and produces data,
    statistics, and original research regulardly.  wb started this in 70's,
    and now puts out yearly world development report w/ academics and policymakers.
  - imf-provided data on financial indicators, and wb data on development indicators
    and household surveys are cited in ~100 academic publications/year
  - imf and wb train employees which then become leaders of policy teams in home countries.
    both also train civil servants in member countries.  finally, the wb has paid
    for thousands of students in developing countries to study abroad
  - hard to measure effect, but imf and wb also provide policy advice to countries
    to modify economies.
  - people try to separate the advising/researching parts from the funding parts,
    but that's harder than it sounds: countries won't open their books up to anyone
    unless they get money out of it, and giving a loan also keeps the imf and wb
    honest, since their money is at stake if they misuse the information.
  - many criticisms of the IFIs, including that they use cookie-cutter approaches
    which are dogmatic, free-market strict approaches.  it seems that they haven't
    hindered growth in the aggregate since war.  they also could not have made a huge
    dent in growth, given that other than in subsaharan africa, they never gave more
    than 2% of GDP.  often learned from their mistakes, so past mistakes aren't future ones
  - so what now? wb was started when there were less private capital options, and imf
    was started under fixed but adjustable exchange rate system.  world has changed,
    what to do with the IFIs?
  - analyze in three ways: is international cooperation better than international
    laissez faire?  is there public good from international transactions?  do these
    international organizations make more rational choices?  no good analysis on
    any of these, since they are hard frameworks to develop.
  - there are 350 international organizations employing 100k people---lots of variables
    move together/against each other, so hard to measure effect of any one.
  - future considerations: wb has three options: focus only on poor countries, focus on soft issues like
    environment or womens rights, or close down.  imf has issue of so much
    private funding that it can only really provide advice, so what is best way to do this?
  - is wb better than bilateral agreements?  maybe---international organization
    gets a lot of flak for disagreeing w/ governments, but it is also more resistant
    to political desires of individual group in charge at the time.  thus, private
    lenders often wait until wb or imf put their stamp of approval by putting
    their own money into a deal before private money goes into a deal.
  - given more private capital options, wb should spend more on poor countries, less
    on developing countries w/ other options.  only invest in developing ones if
    they have horrible political situation or if the staff of the bank can bring
    experience that otherwise wouldn't exist.
  - imf's future: G7 meets when large industrialized countries have issues w/
    exchange rates, often w/o IMF.  so role exists for imf, but probably only for
    poorer countries and for crisis management.  so it would be a more advisory
    role, and would develop an early warning system for exchange rate issues
    that larger countries could finance fixes for.
  - sach argues that in international financial crises, there is no international
    bankruptcy process.  so especially need someone to be lender of last resort,
    perhaps the imf.
  - as to combining wb and imf, one deals long-term and the other short-term, so
    their goals are different, and competition is good when they do overlap,
    since they sometimes swap in making the right decision.

Globalization and all that---Banerjee
  - globalization is hard to define.  it would be nice to call it unrestricted
    movement of people, goods, ideas, but it's not like our laws are converging
    and people can live anywhere
  - has done bad things (e.g., cotton farmers in india having competition's
    prices so low that they kill themselves) and good things (e.g., maids cleaning
    homes of the newly rich instead of being owned by their husbands)
  - globalization seems to help by allowing specialization to have a larger effect.
    but the argument must be that it helps more people than lose from it.  what
    if the poor are the ones that lose?  it's important to keep track of the distributional
    effects of trade, etc.  People did this in 1930-1960, but we live in a different world.
  - samuelson-stolper theorem: labor-abundant country opens trade w/ labor-scarce country,
    that benefits laborers in the first/employers in the second, and hurts employers
    in the first/laborers in the second.  That's because labor-rich countries
    tend to be poorer, so poor laborers get more labor-intensive jobs, and employers
    who had poor labor must pay more to become competitive.
  - until mid-1800s, big wage gaps between different countries.  removed trade barriers,
    increased locomotive and ship technology improved trade, and wages equalized
    between labor-abundant, land-scarce (western/central europe) and 
    labor-scarce, land-abundant countries (americas, other places in the new world).
  - while wages and land prices didn't quite converge, their ratios did, meaning that
    although political issues still led to differences, the pattern of land and wages
    worked out somewhat.
  - still, it could be that trade wasn't the big game-changer: pro-trade years
    were also pro-immigration years.  immigration alone in early 1900s can explain
    all changes in the income distribution (in magnitude, though it seems that it
    wasn't the only reason for the distribution changing)
  - after WWII, trade opened up, and the poor in poor countries got lower wages.
    then in 1980's-late 1990s, trade barriers were removed, and the rich got richer,
    while the poor got poorer (they couldn't find jobs as easily).  ratio of wages
    for college-educated folks to wages in unskilled workers went up by 2-3x
    in mexico, argentina, uruguay, chile, etc.  freer trade hurt unskilled workers.
  - same happened in china and india: open trade => poorer poor, richer rich.
  - note: this is talking about the income distribution, not growth, which went up
  - other factors: with increased trade came increased foreign investment.  theory:
    foreign firm builds factory in mexico, its design will not take advantage of
    as many low-skilled workers since it was designed by high-tech country.  can explain
    half of increase in skilled labor wages with foreign investment.  that still
    leaves another half, and it still doesn't explain the widening income gap
  - but this doesn't answer everything: 90% of latin america's trade is w/
    rich labor-scarce countries, so they should have benefitted from trade.  also,
    there's little correlation between tariffs and and changes in employment.
    so trade seems to play a lesser part?  trade theory has to explain some holes.
  - banerjee claims that reputation is the missing element.  people who are rich
    and buy a lot build brand-name reputation, and thus brands will only
    trust skillfull manufacturers
  - quality being important means price is less important: indian firm could
    double price of line of code every few months, and clients wouldn't leave them still.
    that's because cost of bad code at new consultancy would be discovered
    months later, so it's not worth it.  similarly, macy's would discover a year
    late that the clothes it ordered a while back are not up to standards.
  - a few correlaries: reputation is like a fixed cost, so once you have it, you
    expand around the brand (gucci and amazon.com sell lots of products now).  firms
    are willing to handle higher costs for more predictability, and might even
    buy skilled labor in a third world country.  service industry is especially
    picky of brand, and is also where skilled work is more necessary, so the skilled
    become richer.  similarly, manufacturing is less reputation-intensive, so
    unskilled workers see more wage pressure to undercut.
  - what comes of this?  it's the opposite of the samuelson-stolper theorem:
    the poor in poor countries stand to suffer the most.  most important for countries
    is thus to gain reputation.
  - gaining reputation: improve court system, to make it easy to litigate.  developing
    countries should open trade w/ other developing countries, as poorer customers will require
    less reputation to switch brands.  when free trade opened up among latin american
    countries, poor in argentina benefited.
  - important to pull in multinational, since they have brand and reputation in rich
    country, and can pull trade strings in rich country to get poor country to
    produce and trade a lot.
  - improve capital markets to allow firms that do well to scale quickly
  - traditional economists' outlook on trade is that it's good as long as pie gets bigger:
    even if poor lose out more, you can always redistribute.  in practice, that's
    not so easy---finding and taxing the winners is hard, and they have ways to
    avoid paying redistribution.  so in practice, poor end up poorer.
  - final thoughts: he would go with openness in trade.  given that we're already
    hard-wired to think that growth is good, we can't put the genie back in the bottle.
    so let's go for growth and guarantee that the poor have a guaranteed 
    minimum standard of living, so that the losers don't lose as much.
  - guaranteed standard would be delivered by health and education, but that's not easy
    to deliver.  suggests targetting income support at elderly and women w/ young
    children, and having negative tax for the poor (modeled after earned income tax
    credit in USA).  others that will suffer are owners of small firms, but he doesn't
    know of much research to show how to help the nondestitute.
  - issues with open trade: rich countries already have welfare in place, but it's harder 
    for the hurt poor countries to install support _now_.  also, globalization 
    makes it easier for brain drain to happen, since rich countries can more
    easily pull in the best-in-class from poor countries.
  - so poor countries should charge rich countries explicitly for staying open.
    maybe do this by setting up quotas for worker migration (not just the engineers
    and doctors), so that 20-30 year-olds of all skillsets can migrate.  they can
    stay for 5 years, and are paid to come and return by the rich country, so
    that the earned wealth can go back to the poor country.

Latin America's Growth and Equity Frustrations During Structural Reform---Ocampo
  - 1980s latin america: lost decade, -.9% GDP/capita growth.  1950-80 saw 2.7% 
    GDP/capita growth. High state intervention and high protection of domestic
    markets was reversed in early 1990s.
  - reforms included reduced tarrifs, reduced nontarrif barriers, removing foreign
    investment blocks, removing foreign exchange regulation, increasing autonomy
    of central banks, and privatizing some state banks.
  - improved tax code, strengthened social security, added VAT, limited privatization
    of market, but not in oil/mining.
  - around 2% GDP growth 1990-1997, but then until 2003, lost half-decade of
    decreased growth---re-evaluated structural reform.
  - while reforms led to growth in some fields, it was mostly in skilled labor,
    and there was an decrease in informal/unskilled labor utilization, leading to
    overall per-capita growth decline.
  - result of structural + fiscal + monetary reform were low inflation, high capital
    flows, high investment rates, and strong productivity performance/economic growth.
  - lowering inflation came from reducing budget deficits to 2% of GDP while increasing
    spending by 20% (from actual underlying growth) and spending on social programs.
    this worked until 2003, when inflation crept back up.
  - even still->1990-97 growth never hit 1950-1980s levels, and dropped 1998-2003.
    that's consistent w/ poor labor productivity during that same period.
  - growth was dependent on foreign capital.  after Brady plan converted debt
    to securities in 1989, lots of capital entered system.  after foreign money dried
    up after asian financial crisis, things went downhill.
  - biggest contributor was cyclic motion of low interest/high growth/high foreign
    investment followed by high interest rates/contracted growth/reduced investment.
  - exports helped a lot (correlated w/ gdp growth).  surged to 9% in 1990s, 
    then went to 1.5% in 2001-2002 to get to 4.4% in 2003.  Mexico accounted 
    for most of this, but many countries saw it - two agreements to export 
    within latin america---andean countries and southern common market 
    (mercosur) increased intraregional trade.
  - countries closer to US had manufactured exports w/ imported inputs.  farther
    south they exported commodities/oil.  in panama/mexico, they exported tourism/finance
  - Productivity performance picked up for manufacturing due to alliance in mexico
    w/ US.  Other industries, such as natural resource-export declined in productivity
    as competition rose when globalization happened.  Experts thought productivity
    growth would spread to all sectors, but it didn't.  This led to booms/busts
    as specific sectors rose/fell.
  - summary: macroeconomic sluggistness came from lack of foreign investment or
    decline in trade.  mesoeconomic (sectoral/intersectoral) sluggishness
    came from lack of spread of productivity in some competitive sectors to other
    sectors that didn't get better.  microeconomic sluggishness came from defensive
    policies within companies that avoided investments in new equipment/technology/structure.
  - best growth was 1950-1980, which featured excessive state controls.  1980 sluggishness
    led to free-market reforms that resulted in better growth, but not as good at 1950-1980.
    this counters traditional theory that free-market reform leads to growth.
  - while you need strong macroeconomic framework to see growth, there is less
    evidence to the effect of strong structural reform to liberalize 
    economy resulting in growth.
  - research often characterizes structural characteristics (properties of how
    things are now) with structural reform (changing how things work).  certain
    characteristics might lead to growth, but changing the state of things has
    less certain effect.
  - in 1990s to early 2000s, social spending as % of GDP rose "as a dividend of 
    democracy."  this led to increased government spending, but benefitted from
    better identification of those in need, and improved spending on
    health, education, other social standards, leading to higher human 
    development index (HDI) even in the lost decade.
  - still, social security coverage remains low in most countries.  many of these
    services were privatized, which led to increased efficiency, but also led
    to a concentration on higher-income, lower-risk people, which is a poor
    feature of social security systems.
  - labor market has seen worst results.  while it expanded, this was mostly
    into low-skill, low-productivity jobs.  7 of 10 new jobs in 1990-1999 came
    from informal-sector jobs.  result: deterioration in job quality, relative
    increase in temporary work, reduction in social security coverage, and
    increased work w/o labor contract.
  - sector that benefitted: international specializations such as manufacturing
    in mexico.  less benefit: southern countries which export resources (oil/mining)
    that require less skilled work/specialization.
  - increased income gap between college-trained workers and everyone else.  one
    reason is that specialization leads to increased wages, whereas unskilled
    work has to compete on global market w/ low-income countries like china.
    women entering workforce coincides with decreased (but still large) gender income gap.
  - poverty rate went from 40.5% in 1980 to 48.3% in 1990.  people left poverty
    in 1990s, and stagnated at 200M post 1997, but during lost half decade,
    poverty went up again by 20M.  As GDP has increased by 6% since 1980, poverty
    rate in 2002 was 3% higher.
  - unlike poverty rates, income distribution across region is uneven and appears
    to be getting worse.  in most places, it's getting worse due to specialization.
  - re: income distribution---education coverage seems to be improving.  reduction in births
    means more women entering workforce.  this is offset by rising unemployment
    and low-skill employment, and income gap in education.  that said, it's not all
    trends: disparity in income has been passed from generation to generation.
  - given: not all of the liberalization and reform has resulted in stable internatinal
    support and growth.  three takes on the way forward:  you haven't done enough,
    keep up the reform.  macroeconomic reform was good, but need to complement with
    social reform and strong domestic institutions.  finally, that the original reforms
    were bad, and we should remove them.
  - his take (as UN undersecretary for economic/social affairs and previously
    executive secretary for UN economic commission in latin america + carribean)
    is closest to the second approach, but also to fix some issues from the
    first reform period.
  - problems to fix: look not only at inflation/budget deficits, but also employment
    and capital inflows.  fix stability issues that arise in trade, technology
    adoption, exchange rates, and issue temporary subsidies and through 
    public-private partnerships to fix roadbumps.  don't go back to state-run
    everything, but include saftey net in newly liberalized economies.

Making Aid Work---Banerjee and He
  - Why don't we just give people money?  We instead give it to NGO or govt. and
    say "build a school."  It's better to save/invest than consume, but why
    don't people do that instead?
  - Used to be taboo to think that people can't prioritize, or don't think of kids
    future when given money today.  Now most economists feel this way, even though
    there have been no longitudinal studies to see what happens when you shower
    people w/ cash.
  - Another argument is that giving services rather than money prevents rich
    from pretending to be needy.  Even still, they value their time---would
    they stand in line for a small payment?  The point: we know too little.
  - but giving things away is harder, since you have to produce, oversee, standardize,
    and prevent corruption.  Also---different people believe that different goods/services
    are most productive, so each donor does what their intuition says.  There
    are many ways to promote education---which is best?  And is promoting education
    best, or creating jobs to justify education better?
  - so donors oversee projects, research them, and some contract them out to NGOs.
    how effective are they?  World bank lending is down almost a third because of
    administrative overhead.
  - Most foundations don't report data to help evaluate effectiveness.  World bank
    is only org. to publish rate of return, but it's incomplete.  Hard to calculate:
    what's the counterfactual for rate of return had you not helped people?
  - WB and Asian Development Bank (ADB) are only ones to publicly report rates of return
    and open books to outside evaluation.  Other orgs. report test scores
    went up or number schools built, but it's from some internal calculation.
    WB can serve as an example in reporting, but people aren't taking its cues.
  - WB stats: gives more money to projects that showed improvement in last three
    years.  For two projects w/ equal improvement performance, gives more to one that's worse off.
  - The asian bank is different: past performance and improvement in performance
    are negatively correlated to future money into a sector.
  - So either WB sees itself as a leader: shifting money to fields that are improving
    more rather than sticking to what worked in the past, or it's just easily affected
    by shifting fads.
  - Then they look to see how funding affects future performance in a sector.  WB
    changes in funding don't affect performance of projects, neither does ADB.
    This is consistent w/ view that WB is faddish/ineffective at managing projects.
    They are also consistent w/ view that WB is leader and invests in projects
    that other's wouldn't to get immediate return.
  - WB, like all organizations, is subject to inefficiencies.  Must spend money,
    so employees might be biased to spend money on potential recipients they meet
    even if it's not the most effective way to spend the money.  Ideological
    conflicts also arise.
  - Smaller organizations might be subject to pressure, e.g., from USA or people
    that hate USA.  On flipside, large organizations have to be open to conflicting
    ideas, and may want to placate many sides.  Most importantly, decision-making
    isn't always scientific.
  - Rarely do they use randomized evaluation to test ideas.  randomizing isn't
    always effective either---different geographies, cultures, or just infeasible
    to run an evaluation of a certain technique in a certain place.
  - Still, WB sourcebook of different instruments for implementing aid is a long
    list with no indications of what has been evaluated or how it fared.  Only
    school vouchers were randomly evaluated in this list.
  - Why do people resist evidence-based policymaking?  mixed motives of those
    giving/receiving aid.  it's nice to prove something works, but what if you
    end up proving it doesn't?  But for every initiative that's running,
    there are two others that aren't, so you'd figure the champions of those
    would be interested in supporting evidence.  so maybe everyone at WB just
    views evidence-based policymaking as impractical?
  - senior WB officials give two reasons for impracticality: will bias aid
    to what is easily measurable/evaluatable, and at this point so little
    is backed by randomized trials that they'd have to sit on their hands for
    a while until the evidence came in.
  - authors disagree w/ both points.  sure there are things that haven't/can't yet
    be evaluated.  but in the meantime, there are so many projects (bad exchange
    rates, pension plans on way to being bankrupt) that they could obviously
    help while waiting on data for other initiatives.
  - there are also macro intervensions that have possible micro evaluation.  maybe
    decentralizing political power is macro, but haivng small randomized evaluations
    at local level can teach you a lot about whether it's effective or not.
  - as for things not actually being measurable: be creative!  empowerment is harder
    to measure than consumption, but example from one duflo study: measure
    number of questions asked by females at village meetings.  that's not too much
    worse than using income as a measure of well-being, and economists do that all the time.
  - as for lack of evidence to get things going: did a survey of techniques
    that are sustainable, had randomized evaluation, and have statistical significance.
    when there was disagreement, picked result that had better experimental conditions.
    list spans 20 or so pages, if you leave out programs that just give away money.
  - sure, many interventions are uncovered or left out.  that gives a starting point
    for aid.  estimate cost of scaling up from per capita to per country.  results
    show that a recurring annual expense of $11B is possible just based on
    well-evaluated projects that have reported success.  More evaluations since
    2004, so the number grows, even though the amount is larger than WB
    currently gives out in such loans.
  - They say: give these projects money.  until you find more proof of other
    working projects, use the rest for humanitarian aid to just give money to
    people who have no food/water/basic supplies.  can enforce w/ harsh punishment
    on those caught cheating through random checkins w/ recipients.  can evaluate
    richness based on simple tests: having a car, etc.  Adding programs such as
    PROGRESA and other voucher programs can scale aid up to $34B/year.
  - MCA (Millenium Challenge Account) from USA seeks to only give aid to countries
    that show aid money leading to performance.  sure, it's a problem that money
    goes into pockets of corrupt beaurocrats, but it's also a problem that the
    money goes into unproven programs!  MCA is basically penalizing poor for
    issue in their country that they have nothing to do with.  Instead, look for
    innovative programs that target them and are proven effective.