[PLUG-TALK] Expansion, growth, the world, and time: PLOS paper

Keith Lofstrom keithl at kl-ic.com
Fri Mar 25 09:29:11 PDT 2011

> >On Tue, Mar 22, 2011 at 10:42:53PM -0700, Michael M. Moore wrote:
> >>"Two weeks ago, the scientific journal PLoS Medicine published a study
> >>that sought to define the relationship between economic growth and the
> >>pervasive problem of undernourished Indian children.  In India, 25-50%
> >>of the deaths of children between 6 months and 5 years old are due to
> >>inadequate food intake.  This slow, sustained starvation causes stunted
> >>growth and puts children at greater risk for infection and disease.
> >>
> >>"The primary policy tool to combat chronic childhood undernutrition has
> >>been economic development, but despite two decades of booming growth,
> >>the average calorie intake in India has actually declined.  Indeed, the
> >>authors found that there was no link between economic growth and
> >>childhood undernutrition; stagnant or thriving, the status of a state’s
> >>economy had no effect on its number of underweight children." [1]
> >
> On 03/23/2011 01:01 PM, Keith Lofstrom wrote:
> >Ah - look carefully at that study, and what it does NOT include -
> >actual calories, birthrate patterns, and mobility.  The Mumbai slums
> >are not full of the grandchildren of previous Mumbai slum dwellers.
> >They are mostly immigrants from poor regions in the countryside.
> >The previous slum dwellers are living in the high rises.  Interesting
> >things happen to compartmentalized data when the compartments are
> >leaky.  That is why an immigrant can move from Guatemala to the US
> >and get a 10x bump in income, while simultaneously lowering the per
> >capita GNP of both countries.  The "capita" changes, so the process
> >is not "adiabatic."

On Wed, Mar 23, 2011 at 09:12:45PM -0700, Michael M. Moore wrote:
> I'm not sure what you're getting at here -- the study wasn't of children 
> just in one region, it was across the country.  The point of the study 
> was to determine the correlation between economic growth and childhood 
> undernutrition.  It found none.  I thought it related to this discussion 
> in the sense that it points to "increasing resources" not, absent any 
> concerted effort to disperse those resources in something approaching an 
> equitable manner, necessarily improving the situation across the board.

Keith replies:

Lets explain that more carefully.  The PLOS study was flawed.
The conclusions it draws are based on a bogus model of human
behavior.  It also did not directly measure calories - it
inferred them from outcomes.  The conclusions were drawn
from differential statistical behaviors of regions of India.  

The authors' claim is that economic growth does not lead to
better outcomes for children, based on regional statistics
for child physical size, and regional statistics for per-capita
economic growth.  But regional statistics are a poor measure
when people can move between regions.

People in India move, from poor regions to rich ones.  They are 
exporting poverty and the results of it from poor regions to rich
ones.  This worsens the statistics by region.  The PLOS study did
not consider this, and it is the dominating effect, like smoking
is on US health statistics.

The most massive migration ever seen is occuring right now,
in India, from rural to city.  Cities mean better jobs, living
standards, education, medical care, support in old age, and
more freedom.  Rousseauian fantasies aside, life in the country
can be brutal.

Why do people migrate?  Certainly not to poorer conditions.  And
they move only if they can afford it - the very poorest don't
have the means to travel and support themselves in the city.

One hundred years ago today, March 25 1911, my grandfather left
Liverpool, England on the steamship The Empress of Ireland.
His ticket cost 311 Kroner, a year's wage in Sweden, a
month's wage in Oregon.  Farther north in Sweden, people
were starving.  Most could not afford to leave.  

What happens when a family moves?  They leave their old life 
behind, but they don't leave the lingering medical effects of
poverty.  They bring their underweight children with them. 
Those children will be small for life.  Though the PLOS study
does not mention it, most stunting causes occur during gestation,
because Mom could not get enough food.  So an underweight 10yo
child in Mumbai in 2005 does not reflect conditions in Mumbai
in 2005, but maternal health during pregnancy in rural Uttar
Pradesh (for example) in 1994.  The PLOS study ignores this.

Statistically, what happens when people move?  The population
of rich regions go up.  If you don't look carefully, it seems
that birthrate is skyrocketing there, and income dropping.
But cities historically have lower birthrates than replacement. 
The cities are filling with higher-than-median-income people
from the countryside.  The per-capita income drops in the
countryside, with the poor babymakers left behind. 

Where the migrants are going, they will start out with lower-
than-median income, though much higher than was available in
the country.  The per-capita income in the city is reduced,
for a generation.  The migrants move from shanties in the
country built with agricultural waste, to shanties in the
city built with construction scraps.

But - now the migrants have access to doctors, and education
for their kids.  The weak children no longer die, as they
would in rural areas, dropping out of the rural statistics.
The dead children are no longer measured for wasting,
improving the apparent aggregate health in rural regions. 
In the city, the weak children get counted.  In the city,
those weak children get educated, and get much better jobs
than their parents, and get nice places to live where the
electricity is always on.  The process takes time, a
generation.  But Indian people think in generations, not
about what is on TV tonight.

Capturing all that is statistically messy.  It is easier to
look for the contact lens under the streetlight, not where
it was lost.  While avoiding the messiness leads to a more
mathematically rigorous result, it does not lead to an
accurate one. 

Assuming people are not complete idiots, they migrate from
worse to better.  Measuring average child size in a
destination region tells you nothing about the desirability
of that region.  If you want to know what is better for
people, observe their migration patterns.  

Consider hospitals.  The "per capita income" of hospitals
is enormous.  Highly paid specialists, patients with enough
money to be there, with the wealthier able to stay longer. 
However, the average health in a hospital is low, the death
rate is high.  Hospitals are cesspools of infection.  So why
does a patient leave their comfy house or apartment in a
healthy residential neighborhood and move into a hospital? 
Obviously, because individuals perceive (justifiably) that
that they will survive accident or acute illness better in a
hospital than if they stayed home.  If the same two populations
(at home and in hospital) were studied by the regional aggregate
methods of the PLOS study, they would show that hospitals are
very wealthy but have poor health outcomes.  The study would
be seriously flawed because it leaves out migration.  Sadly,
such flawed studies are abundant.

Always be skeptical of tinkertoy models.  When they leave out
what is important because it is complicated, they reveal only
methodological and ideological bias, not the objective truth.


Keith Lofstrom          keithl at keithl.com         Voice (503)-520-1993
KLIC --- Keith Lofstrom Integrated Circuits --- "Your Ideas in Silicon"
Design Contracting in Bipolar and CMOS - Analog, Digital, and Scan ICs

More information about the PLUG-talk mailing list