Petroleumworld`s
Opinion Forum:
viewpoints on issues in energy, international
politics & civilization.
Sunday's
Feature
How
Not to Defend the Revolution: Mark Weisbrot
and the Misinterpretation
of Venezuelan Evidence
By Francisco
Rodríguez
Abstract: Mark Weisbrot (2008) has claimed that under the
Chávez administration in Venezuela the share of pro-poor
spending has increased, inequality has declined, poverty
has fallen rapidly, and there has been a massive reduction
in illiteracy. All of these conclusions are based on the
use of
heavily slanted data and on the misinterpretation of the
existing empirical evidence. Weisbrot uses estimates of social
spending that are upward biased by the inclusion of
large infrastructure projects, debt refinancing, and even
military spending; his inequality data is distorted by the
inexplicable exclusion
of households that received no income; his econometric estimates
on illiteracy actually show the exact opposite of what he is
arguing for. Weisbrot confuses basic economic concepts and
offers a bizarre interpretation of events leading up to the
2002 currency crisis. Once one corrects
for Weisbrot’s biases, the evidence paints a consistent
image of an administration that has not effectively prioritized
the well-being of the Venezuelan poor.
In a paper published in the
March/April issue of Foreign Affairs (Rodríguez,
2008), I argued that there is little evidence that
the government of Hugo Chávez has given priority to
the well-being of Venezuela’s poor. In recent
days Mark
Weisbrot (2008) published a rebuttal on the
website of the Center for Economic Policy Research - a
Washington
thinktank
that he co-directs - arguing that some of my conclusions
were “altogether
wrong, and others grossly exaggerated and/or misleading.” In
particular, Weisbrot argued that I am mistaken in asserting
that the share of pro-poor spending has not increased under
Chávez,
that inequality had risen, that the government had not taught
1.5 million persons how to read and write, that the rate
of poverty reduction has been slower than normal given
Venezuela’s
economic growth, that other health and human development indicators
show a deterioration in the living standards of the poor,
and
that the 2002 recession was not caused by the country’s
political crisis. On each of these, Weisbrot argues the
exact for the exact opposite conclusions to those that
I have drawn.
I welcome the opportunity to have an in-depth discussion of
the evidence regarding the well-being of the Venezuelan
poor under Chávez.
Indeed, many of the points raised in Weisbrot’s paper
as well as in this response have been discussed previously
in academic fora. The broad dissemination of both papers
thus offers an extraordinary opportunity to involve a broader
group of policymakers and academics in the discussion and analysis
of Venezuela’s social and economic policies.
As I will show, Weisbrot’s critiques are generally
invalid, relying on erroneous reading of the evidence or
use of severely biased indicators that do not accurately reflect
the evolution of the Venezuelan economy or the well-being of
the poor. For example, I will show
that Weisbrot’s estimates of social spending are upward
biased by the inclusion of large infrastructure projects, debt
refinancing, and even military spending in what
he contends is pro-poor spending, that his inequality data
is distorted by the inexplicable exclusion of households that
received no income, and that his econometric estimates on the
effect of the Robinson program on illiteracy actually
show the exact opposite of what he is arguing for. Weisbrot’s
other criticisms are based on a misinterpretation of the concept
of elasticity and on the questionable interpretation
of existing health indicators and of the evidence leading
to the 2002 recession.
Before delving into these differences, I would like to emphasize
one basic point of agreement with Weisbrot. Official Venezuelan
statistics are far from the ideal of what we would need in
order to properly evaluate the performance of the Chávez
administration. Well-designed impact assessments of the government’s
social programs are either inexistent or have not been
made public by the administration. The raw data and methodological
descriptions necessary to replicate official
calculations are only made available with severe lags,
and often not at all. Many series that are vital to the
analysis
of the government’s policies are not public, and
it is not uncommon for different entities to produce contradictory
numbers. These weaknesses cause an inherent
ambiguity in the interpretation of the evidence regarding
the Chávez
administration, a fact that helps to underline the usefulness
of a serious academic debate on how to read the data.
In the rest of this note, I will take each one of Weisbrot’s
criticisms and show why they are invalid. In most cases,
I will show that he has misinterpreted the evidence or used
severely biased indicators,
and that when we correct for these biases we come to conclusions
which are opposite to what he contends. In a number of issues,
our disagreements reflect alternative possible interpretations
of ambiguous data, and it is useful
to lay out the sources of these differences in interpretation
for readers to make up their own minds. All in all, I will
argue, the image that emerges from a close reading of the evidence
is still one in which there is little evidence that the Chávez
administration has prioritized or produced favorable effects
on the well-being of the poor above and beyond what we could
have expected any other government to do.
1. Has the share of pro-poor spending gone up?
In my article, I argued that government spending figures show
no evidence that the Chávez administration is giving
greater priority to the categories of spending that benefit
the poor. As an example, I cited the fact that the average
share of the central government’s budget allocated
to health, education, and housing during Chávez’s
first years in office was 25.12 percent, essentially identical
to the share in the previous eightyear period, 25.08 percent.
Weisbrot has countered with three pieces of evidence: that
the share of social spending – a broader category
- in total spending has increased markedly since 1998,
that the absolute amount of resources received by the poor
has also increased significantly, and that my calculations
exclude the contributions
by PDVSA to social projects, which he claims summed to $13.3
billion, or 7.3 percent of GDP, in 2006.
Before
looking at the data in detail, it is relevant to think
a bit about what we should be looking for. Let us start
from the following fact: the Venezuelan state is undeniably
much richer today than it was nine years ago, to a great
extent (if not completely) due to the ten-fold increase
in oil prices that has occurred since 1999.
As a result, the Venezuelan government has substantially
increased its spending levels, and therefore is indeed spending
more in real terms on just about any type of expenditures.
This means
that all categories of spending can be expected to have increased
in real terms since Chávez reached office, be
they social programs, infrastructure projects, military spending,
or growth of the public bureaucracy.
But the absolute level of pro-poor spending is not
what should concern us if we are interested in evaluating
a government’s priorities.
Precisely because the government has experienced such
a huge windfall, we want to study how it has allocated it among
different possible objectives. To use an intuitive metaphor,
if you want
to know how much your rich uncle cared about you, you’d
like to compare how much of his inheritance he left you with
what he gave everyone else. If all of your siblings got a million
dollars in his will, while you received the old man’s
poodle to take care of, itwould be hard to argue that
you were his favorite nephew. Thus, of all the pieces of evidence
thrown about by Weisbrot, the ones that we should study closely
are those that reflect the relative distribution of government
spending among different categories of spending.
What do we see when we look at these figures? The upper
panel of Table 1 shows the percentage share of different
categories of “social” spending
in total government spending.2 I
compare two periods of equal length: 1999-2006, corresponding
to the Chávez
administration, and 1991-1998, corresponding to the previous
eight years, in order not to contaminate the comparison by
the behavior of the data in one particular year. I enclose
the word “social” in quotation
marks precisely because, as should become clear by looking
at the categories included in it, only some of these types
of expenditure are pro-poor. Financing of the education and
health system, as well as public housing programs are clearly
programs that are primarily oriented towards the poor. This
is also arguably the case with social development expenditures,
but definitely not the case with investment in science
and technology or culture and communications.
In any case, what strikes one as evident when one looks
at Table 1 is that the only significant change in government
priorities that appears to have occurred in this period is
an increase in social security spending. Indeed, if one takes
out social security, the share of social spending in total
spending actually goes down
from 32.0 to 29.8% between the pre-Chávez and
pro-Chávez
periods. Therefore, the increase in the share of “social” spending
that Weisbrot presents as evidence of the government’s
pro-poor priorities is completely driven by a more than
doubling in the share of social security spending.
Table
1: Decomposition of Social Spending
as a Percent of Total
Spending, 1991-2006
| |
1991-1998 |
1999-2006 |
Education
Health
Housing |
14.4%
5.7%
5.1% |
16.0%
5.4%
3.7% |
Education, Health and Housing |
25.1% |
25.1% |
Social Development
Culture and Communications
Science and Technology
Social Security |
5.6%
0.8%
0.5%
4.5% |
3.3%
0.7%
0.7%
9.9% |
All Categories |
36.5% |
39.7% |
Venezuelan social security spending is essentially
spending on pensions. Roughly half of the budgetary
assignment to the social security system goes to the financing
of the Venezuelan Social
Security Institute (Instituto Venezolano de los Seguros
Sociales)
while the other half goes to direct payment of public sector
pensions, including the very generous military pension
system, which by itself accounts for 8.9% of total government
social security spending.3 There
is a vast array of evidence
showing
that pension expenditures in Latin America are highly regressive
because poor individuals tend to belong to the informal
sector, and thus do not enjoy coverage by the formal social
security system.4 This
is certainly the case in Venezuela, where only
13% of individuals in the lowest quintile participate in the
social security system.5
In sum, the share of the central government budget
devoted to pro-poor spending has not increased. The
only category of “social” spending
that has increased markedly is spending on pensions,
which are regressive. In a particularly befuddling instance
of trickle-down thought, Weisbrot actually argues that
even if the beneficiaries of the pension system are
relatively better-off formal sector workers “their
pensions are in many cases shared with families and
extended families who are in the informal sector, thus contributing to
poverty reduction among others besides the recipients.” Using
this reasoning, Weisbrot could just as well go ahead and
throw in the rest of the budget into social expenditures.
Alas, this is not that far from what he goes on to
do. Weisbrot goes on to criticize me for excluding
PDVSA “social” expenditures,
which according to him sum to $13.3
billion, or 7.3% of GDP, in pro-poor spending. The
simple reason why one need not
include them is that including them makes little difference
to the aggregate calculations, as most of these expenditures
are not, under any conceivable definition of the category, classifiable
as social or pro-poor expenditures.
Table 2 presents the decomposition of what PDVSA calls “expenditures
for social development,” calculated using PDVSA’s
financial statements, both for 2006 and for the whole
2001-06 period.6 We have classified PDVSA expenditures according
to one of three categories, corresponding to the categories
of spending that we have used for the central government budget:
education, health and housing; other social spending; and non-social
spending. In 2006, only 26.8 % of PDVSA’s “expenditures
for socialn development” corresponded to spending
on educational, health, or housing programs (including
the educational and health misiones and low-income
housing projects) while an additional 10.5 corresponded
to other social expenditures (Mercal, Misión
Ciencia, Community Development). Thus the total share
of social spending was 37.4% of total PDVSA expenditures.
Note that these percentages (26.8% and 37.4%) are
very similar to those found for the central government
budget and shown in Table 1 (25.1% and 39.7%). Adjusting
government expenditures for PDVSA’s
expenditures thus leaves essentially unchanged the
conclusion that the priorities of the current Venezuelan
administration – at least as far as its spending
figures reveal – are
very similar to those of its predecessors.6
Table
2: Decomposition of PDVSA's
" Expenditures
for Social Development"
|
Education,
Health
and Housing |
Other Social |
Total Social |
Non-social |
Total (Social
and Nonsocial) |
Year
2006 |
|
|
|
|
|
Direct
Expenditures
|
2130
|
1265
|
3396
|
676
|
4072
|
Investment
Funds
|
207
|
0
|
207
|
859
|
1066
|
FONDEN
|
882
|
0
|
882
|
5974
|
6856
|
Total
|
3219
|
1265
|
4484
|
7510
|
11994
|
Percent
|
26.8%
|
10.5%
|
37.4%
|
62.6%
|
100%
|
Years
2001-2006 |
|
|
|
|
|
Direct
Expenditures
|
4957
|
2081
|
7037
|
1110
|
8147
|
Investment
Funds
|
857
|
0
|
857
|
6709
|
7566
|
FONDEN
|
1078
|
0
|
1078
|
7303
|
8381
|
Total
|
6869
|
2081
|
8972
|
15122
|
24094
|
Percent
|
26%
|
8.6%
|
37.2%
|
62.8%
|
100%
|
Source: See Appendix
What, then, makes up the bulk of PDVSA’s “social” expenditures?
Table 3 lists the eight largest non-social
items included in PDVSA’s “social” expenditures.
These include $3.2 billion for debt refinancing,
$1.1 billion for Ministry of Defense projects, $1.3 billion
for four large infrastructure projects, and $613 million
for capitalization of two sectoral finance entities. These
projects are representative of the broader pattern of PDVSA’s
non-oil business related expenses. Whatever the relative
merit of these investments, it is very hard to categorize
them as expenditures whose primary aim is to benefit
the poor.
Table 3: Largest Non-Social Expenditures included in PDVSA's
" Expenditures for Social Development"
Project |
Expenditures
(Million US$) |
Intermediary
Financing Entity |
Ministry
of Finance - Debt Restructuring |
3221 |
FONDEN |
| Ministry of Defense - Various Projects |
1093 |
FONDEN |
| Highway - Ezequiel Zamora |
476 |
FONDEN |
| Agricultural, Fishing and Forestry Fund |
327 |
FONDEN |
| Venezuelan Argicultural Bank |
286 |
FONDEN |
| Orinoco River Bridge |
273 |
FONDEN |
| Caracas-Tuy Railroad |
272 |
FONDESPA |
| Barquisimeto Mass Transit System |
251 |
FONDESPA |
Source:
PDVSA (2006), pp.71-76
In sum, regardless of whether one includes
or excludes PDVSA expenditures in the
aggregate budget figures, there is little
evidence of a change in the priority given
to propoor spending in the Venezuelan budget.
The relative shares of pro-poor spending
appear to be very similar to those of
previous periods, although of course the absolute
levels are higher. The only way to reach
a different conclusion is to include either spending
in pensions or large infrastructure projects,
debt refinancing, and military spending
in the definition of pro-poor spending.
This is clearly not reasonable.
2. Has inequality increased or decreased?
In my article, I argued that Venezuelan
income inequality had gone up since Chávez
came into office, and in particular since
the second semester of 2000, the moment
in which Chávez gained control
of all branches of government and was
first granted enabling law legislation
allowing him to approve laws by Executive
Decree.
In particular, I cited the findings of
a study by the Venezuelan Central Bank
that found this deterioration in the
2000-2005 period.7 Weisbrot takes issue
with the
comparability
of the indicators used by the Central
Bank. In response, he cites a series
of income inequality calculated by the
National Institute of Statistics that
shows a
strong improvement in inequality starting
in 2006 which offsets the increase occurring
up to 2005.
The series cited by Weisbrot is highly
problematic because it excludes from
the
calculation all households with reported
income equals to zero, thus omitting
the poorest households from the construction
of an inequality index. This fact is illustrated
in Figure 1, where I show the effect
of excluding zero-income households from the calculation
of income inequality.8 If they are excluded,
the Gini index declines from .461 in
the first semester of 1999 to .448 in
the first semester of 2006, a decline of .012 points.
When all households are counted, however,
the relative comparison changes, and the
index increases from .475 to .488 for
the same period. Thus it appears that the decline
in inequality in the series cited by
Weisbrot up to 2006 is an artifact created by
the exclusion of the poorest households
from the sample. Figure 2 shows that the deterioration of
inequality is also present for a variety of alternative inequality
indicators as long as families with zero reported
income are included in the sample.
The National Institute of Statistics
has not yet released its raw data
for the second semester of 2006 or the first semester
of 2007; in the last of these periods
it estimated a further decline in
inequality. Note, however, that there are periods during
which the Gini excluding zero-income
households and that including all households move
in different directions, so that it is not a foregone
conclusion that the properly calculated
series will also decline.9 More
importantly, inspection of Figures
1 and 2 appear to reveal three distinct
periods: a period of decline in inequality
that starts in 1998 (the
last year of the Caldera administration)
and continues up to 2000, a sharp
increase in inequality
that occurs in 2001 (before the political
crisis start), and a stabilization
at high levels of inequality after
2002, with large roughly offsetting
up and down swings in the
final three periods. Whether this
will be followed by a reversion to
the levels characteristic
of the start of the decade remains
to be seen, but what is true is that,
at least
until very recently, inequality climbed and remained high
during the Chávez
administration.
3. Did the Chávez government
teach 1.5 million persons how to
read and write?
Weisbrot takes issue with the claim,
made in my Foreign Affairs article
as well as in a more extensive and technical research
paper jointly written with Daniel
Ortega (Ortega and Rodríguez, 2008a), that
the data does not support the hypothesis
that
the Venezuelan government enrolled
1.5 million persons in the Misión
Robinson literacy program. He claims
that the simplicity of the Households Survey question – which
asks the respondent whether each
family member knows how to read and write - makes it impossible
to use it to assess the effects of
a literacy program.
Furthermore, he also claims to have
found, together with his coauthor David Rosnick
- also of the Center for Economic
Policy Research – that our results
were not robust and an artifact of
the specification
that we chose.
We have already provided an extensive
rebuttal of Weisbrot and Rosnick’s
objections to our paper (Ortega and
Rodríguez, 2008b). As we show
there, Weisbrot and Rosnick’s
arguments provide no support for
the hypothesis
that Venezuela carried out
a large scale literacy program – indeed,
their estimates show the exact opposite.
It is only by systematically ignoring
the results of his own empirical
analysis that Weisbrot
is able to keep on claiming that
the Venezuelan government taught
upwards of one million
persons how to read and write.
Let us consider for a moment the leap
of faith one must take to believe
that the Venezuelan government carried out a large
scale literacy program but that nevertheless
that program fails to register in our data. Imagine that
the Venezuelan
government
did indeed give reading and writing
classes to 1.5 million Venezuelans. Surely, if
these people already felt that they
could read and write before the program, then they
would have answered “yes” to
the literacy question both before
and after
the end of
the program, and thus would not be
picked up by the survey. But the
fact is that there were
1.1 million people who were claiming
not to know how
to read and write in early 2003.
So in order to believe that the program
took place one would either have
to believe that
(i) the program did not reach these
people, or (ii) these people still
claimed not to know how to read and
write even after finishing the program.
The first possibility (a massive
program not reaching a substantial
fraction of those who claimed to
be illiterate) is hard to believe and in itself
would be
an indictment of the program’s
effectiveness. We would have to
buy the story
that the Venezuelan government
devoted the massive resources necessary
to put upwards
of a million
persons in classrooms and yet ended
up
putting exactly the wrong million
persons in
the program - those who already
felt that they knew how to read
and write – while
systematically excluding those
who claimed to need the program
the
most. In order
to get an
idea of the dimension of the necessary
assumptions, imagine that the program
was composed
in 90% of people who claimed to
know how to read and write, with
only the
other 10%
coming from the group that believed
to be illiterate. Even in that
case,
taking
the government’s claims at
face value, 150 thousand people
who claimed to be illiterate
before the
start of the program would have
enrolled in it. But this number
is more
than three times
as high
as the largest point estimates
found in our study (as well as
in Weisbrot
and Rosnick’s analysis).
The second possibility is almost
as farfetched. It would imply that
upwards of a million persons who
claimed not to know how to read and write were
enrolled in
a sevenweek program, showed their
skills by composing a letter to President
Chávez
at the end of the program, received
a certificate that indicated that
they had passed
the Robinson program,
yet would still assert that they
did
not know how to read
and write
when asked by an interviewer. It
would appear that in order for
this to be true
the program participants would
have to be extremely skeptical
that
anything that they
did while in the program even remotely
resembled a literacy course. Again,
if this
were true, it
would in itself constitute a striking
demonstration of the program’s
failure.
How about Weisbrot’s claim
that our results were not robust
and an
artifact
of our
choice of specification? As we
show in our detailed rebuttal,
what Weisbrot
and Rosnick found was that there
was one of the dozens of specifications that
we
evaluated – an aggregate
time-series regression with a lagged
program
indicator – in
which you could find a program
effect. This is of course a common
phenomenon – whenever
you run hundreds of specifications,
some of them will be significant.
Weisbrot
and Rosnick
have selected the one specification
that gives the results which are
most favorable
to the Venezuelan government. But
indeed, even Rosnick and Wesibrot’s
preferred specification imply that
only between 34 and 42 thousands
persons benefited
from the program. This is quite
far from 1.5 million persons. In
other words,
Rosnick and Weisbrot’s results
show that the Venezuelan government
did not
carry out
a massive literacy program – indeed,
their estimates point to a program
of approximate
size less than 1/30th of that claimed
by the Venezuelan government. Remarkably,
this is a result that Weisbrot
has chosen to ignore, despite the
fact
that it arises
from his own
empirical work.
4. Has poverty reduction been faster
than normal given Venezuela’s
economic growth?
In my article, I argued that the performance of the Chávez
administration in reducing poverty compared unfavorably with
that of most other countries. As an illustrative calculation,
I showed that the decline in poverty between the first semester
of 2003 and the first semester of 2007 corresponded to a
decline of roughly one percent in the poverty rate for every
percentage point increase in per capita income,
a ratio that compares unfavorably to that of most cross-national
estimates, which put the elasticity at around minus two.10 Weisbrot argued that if this was true, Venezuela should have
eliminated poverty during this period, a conclusion which
appears absurd.
It is a basic mathematical fact that with a finite elasticity
of poverty reduction,
poverty will never reach zero for any positive level of income,
a truth that seems to escape Weisbrot. I realize that I may
have inadvertently contributed to Weisbrot’s confusion
in this case by using an approximation to the income elasticity
of poverty reduction which was reflective of the patterns
found in more detailed calculations which I did not discuss
for reasons of technical complexity. The source of confusion
in this case is that the percent change definition that I
used is an approximation to the point elasticity (the relative
change in the dependent variable for an infinitesimally small
relative change in the independent variable), which will
only be accurate if the poverty-income relationship is linear.
A more precise estimate of the income elasticity of poverty
reduction will require at least two adjustments. One of them
is the use of a functional form that allows us to estimate
the point elasticity adequately. The other is to properly
account for the effect of scarcities of consumption items
in the poverty basket in the first semester of 2007. These
adjustments operate in opposite directions, and for some
reasonable parameter
values their effects cancel out. However, even if one accepts
only one of these adjustments, Venezuela’s elasticity
of poverty reduction appears to be at best near the lower
bound of –
and at worst substantially lower than – cross-national
estimates.
The first adjustment reflects the need to properly take into
account the nonlinearity of the poverty-income relationship.
Properly estimating the point elasticity requires the assumption
of a functional form that allows us to traduce the observations
of discrete changes in poverty (what we observe) into point
estimates corresponding to infinitesimally small changes.
The simplest and most common way to do this is by assuming
an isoelastic logarithmic specification, such as:
log pov it = ai + ßlog
y it (1)
Note
that using the log elasticity definition will raise the
income elasticity of poverty reduction,
a reflection of the fact that for an isoelastic function
like (1) the elasticity will always be underestimated by
the percentage change definition. If one uses definition
(1) for Venezuela for the period between the first semester
of
2003 and the first semester of 2007, one gets an elasticity
estimate of -1.67 ln(27.5/54)/ln(942,308/629,589). This
10 12 estimate is still lower in absolute value than – though
closer to the lower bound of - the values found in most
of the cross-national literature.
A second adjustment necessary to accurately estimate the
elasticity of poverty reduction in Venezuela is to take
into account the effect of the chronic scarcities of basicfoodstuffs
subject to price
controls that occurred in Venezuela since early 2007. According
to the Venezuelan Central Bank, the scarcity index (percentage
of times for which surveyors were not
able to
find the item in question) was 14.3% in the first quarter
of 2007, with extremely high levels for basic staples of
poor
families’ diet like sardines (86%), black beans (85%),
and milk (65%). Since the regulated prices of these goods
are used to construct the poverty line, this line will
cease to be an adequate indicator of the resources necessary
to
escape poverty when scarcity is prevalent as it will be
measuring the price of a basket of goods that poor individuals
cannot
buy.
As individuals either buy these goods in black markets
which are not covered in the price index surveys, or substitute
towards more expensive sources of caloric intake, the real
income necessary to attain a standard of living above poverty
will increase.
Quantifying
the effect of scarcities on the
poverty line is an extremely difficult task, because
it would require information on the prices that individuals
pay on the black market as well as the substitutability
between
different sources of calories. As we lack that information,
we can try to put some reasonable bounds on its magnitude,
which will ultimately depend on the prevalence of scarcity
in the consumption basket of the poor and the magnitude
of the black market/substitution premium. Given that
scarcities
are concentrated in the goods that are under price controls,
and that the strictest controls affect precisely the
prices of the consumption baskets of the poor, it is intuitive
to
assume that the scarcity level affecting the poor is
larger
than that of the population as a whole.
Let us suppose that it is between 2 and 3 times as large.
On the other hand, let us assume that the markup on the
prices of goods in the black market is between 20 and
40%. In that
case we can create a lower bound scenario in which accounting
for scarcity would raise the poverty line by 5.72% (2*.143*.2)
and an upper bound scenario where it would raise it by
17.16% (3*.143*.4) in comparison to the poverty line
used by the
National Institute of Statistics.
How many additional people would this place under the
poverty line? Since the
National Institute has not yet made publicly available
the raw survey data for the first semester of 2007, we
use the
latest available survey (first semester of 2006) to make
this calculation. In order to do this, we find a cutoff
level of income in this distribution such that the number
of people
below it will be similar to those in poverty according
to official statistics in the first semester of 2006.11 We
then raise the poverty line by our upper and lower bound
estimates and calculate the new number of households
in poverty. The resulting estimates of the number of
households under poverty in the lower and upper bound
scenarios are
respectively 29.3% and 35.8%. We call these the scarcity-adjusted
poverty rates. Using the scarcity-adjusted poverty
rates, the log-elasticity
of poverty reduction decreases to -1.51 in the lower
bound scenario and -1.02 in the upper bound scenario,
numbers
that are substantially smaller than cross-national estimates
and
close to the number presented in my article.
Relying on this same methodology, we can back out what
the poverty rate should have been in Venezuela in the
first semester
of 2007 if the country had has an income elasticity of
poverty reduction
equal to -2, closer to the average of other countries.
In that case, we can derive that the scarcity-adjusted
poverty
rate should have fallen to 24 points. Remember,
however, that since the rate of poverty that is reported
by the National Institute is not scarcity-adjusted, we
would have to use our upper and lower bound effects again
to calculate the poverty rate that we would have observed
in this scenario. Once we do that, we derive a range of
values between 18.5% and 22.5%. This is the range to which
the poverty rate should have fallen if the Venezuelan economy
had evolvedaccording to the patterns observed in other
developing economies.
5. Other objections
Among the other elements of my analysis that Weisbrot has
objected to are the use of indicators of low birthweight
babies, quality of dwellings, and drinking water access to
evaluate the well-being of the Venezuelan poor, as well as
my interpretation
of the events preceding the 2002 economic crisis. I will answer
each of these briefly.
Weisbrot claims that the increase in the percent of low
birthweight babies is due to the decline in the percentage
of children monitored for low birthweight. Weisbrot does
not present an argument why the change in the population
sample
would lead to an increase in the number of low birthweight
babies. Indeed, one might expect just the opposite – that
as the prevalence of monitoring falls, low income areas are
left out of the sample first, thus leading to a decline in
measured number of low birthweight babies. In any case, the
claim that the reason for the increase in measured birthweight
is that the Venezuelan state is no longer monitoring the
weight of infants does not paint the most encouraging picture
of the state of the country’s
public health system.
Regarding the rest of the series,
I agree with the idea that there could be multiple
factors that explain the deterioration in other indicators
of the well-being of the poor, among them measurement
error in the reported series. But this is precisely the point:
if Venezuela had seen a sea change in government policies,
we would expect to see consistent improvements in all
of these indicators. But we instead find that the infant
mortality rate has continued declining at the historical
rate and that
many indicators of the health and well-being of the poor have
deteriorated.
Perhaps all of these indicators are so poorly measured that
they can’t tell us anything
about whether the poor are better off or not. But they
are definitely not providing evidence that the lot of the
Venezuelan poor has improved under Chávez.
Regarding the 2002 recession, a close analysis of Venezuela’s
fiscal and external
accounts at the close of 2001 easily shows that the country
was headed for a balance of payments and fiscal crisis
even in the absence of a political crisis. Despite a decline
in the price of oil in 2002, the government had maintained
a highly appreciated exchange rate, leading to a significant
decline in external demand. The appreciated exchange rate
hurt Venezuelan competitiveness – the Bolívar
appreciated in real terms by 55% and 137% with respect
to the Colombian and Brazilian currencies – and
also hurt the Venezuelan fiscal accounts, leading the
country to run a fiscal deficit of 4.0% in 2004 (and of 5.9%
after discounting Central Bank distributions of forex gains).12 The
attack on the currency lasted through 2001, and by November
the Central Bank had lost $5.4 billion dollars in reserves
trying to defend the currency, well before any political
crisis had started.13 In order
to believe that this currency crisis was politically motivated,
one would have to buy into an interpretation where a full
year before there was even talk of a national strike the
private sector started withdrawing its deposits from domestic
banks
in order to cause economic and political instability. It
is difficult to see why one would want to appeal to this
fantastic conspiracy theory when
there is clear evidence of deteriorating internal and
external fundamentals leading up to the crisis.
6. Concluding Comments
It is useful to recap some of our key arguments. First,
there is no evidence that the Chávez administration
is devoting a higher share of resources to pro-poor
spending. Second, inequality increased between 1999
and 2006, unless by inequality one means inequality
among everyone except those who earn no income. Third,
the Venezuelan government did not teach 1.5 million
persons how to read and write – at most the magnitude
of the program was 1/30th of what was claimed. Fourth,
however one calculates it, Venezuela’s income
elasticity of poverty reduction is below typical values
for developing countries. Fifth, the majority of human
development indicators do not show striking improvements
under Chávez, and
some show deteriorations.
Weisbrot
has not produced a convincing counterargument to any of
these claims.
He has argued that social spending has increased
by using series that are distorted by the inclusion of
regressive pensions, large infrastructure projects,
and even military spending. He has argued that inequality
has declined on the basis of a series that excludes
the poorest families from the sample. He has argued
that the Venezuelan government put more than a million
persons in literacy courses while presenting regression
estimates that indicate that at most forty thousand
persons were enrolled in these courses. He has misinterpreted
the concept of elasticity, and furthermore argued
that the reason why government statistics do not show an
improvement in the health of newborns is that the
monitoring
system has collapsed. To top this all off, he has
presented an incredible conspiracy theory of the 2001 Venezuelan
balance of payments crisis according to which the
private
sector withdrew funds from the domestic system during
more than a year in order to provoke a political
crisis.
In closing, I would like to stress my conviction that
an in-depth academic discussion is indispensable in order
to evaluate and understand the nature of recent changes
in Venezuela’s
economy and society. In any serious academic debate, there
must be mechanisms that subject research to high standards
of quality. One such mechanism is the peer review system,
which ensures that research papers that do not pass basic
standards of quality will not be published. I would like
to invite Mark Weisbrot to make use of that mechanism in the
future, and wish him the best of luck.
Footnotes:
1. Assistant
Professor of Economics and Latin American Studies, Wesleyan
University,
238 Church Street, Middletown CT 06459. E-mail:
frrodriguez@wesleyan.edu.
2. These
calculations use the Escuela de Gerencia Social’s
spending indicators, available
(http://www.gerenciasocial.org.ve/bases_datos/gerenciasocial/cuadros/Gasto/PresupuestoSocial.zip.
3. See Ministerio del Poder Popular
para las Finanzas (2007), p. 184.
4. See, e.g., World Bank (2004).
5. Own calculations from the Households
Surveys available at http://frrodriguez.web.wesleyan.edu/space/FP_debate.zip.
6.
PDVSA(2006). “Expenditures for Social Development” actually
sum to $12.0 billion. Once one includes
subsidiaries, the figure goes up to $13.7 billion. We use the
more detailed accounts for PDVSA
expenditures as these are
available since 2001; the differences once one includes subsidiaries
are minor
and do not alter our main point.See the Appendix for
the full decomposition
and classification of PDVSA
expenditures.
7.
This trend is actually present for many different income
inequality measures for the 2000-05 period,
a fact that Weisbrot misses in part because he only reports
first semester figures. Alternative series differ
in the magnitude of this increase and in whether the improvement
after 2006 is large enough to offset it.
8. All of these calculations are done using the raw data from
the Venezuelan Households Surveys, available
at http://frrodriguez.web.wesleyan.edu/data.htm.
9.
An additional problem in evaluating income inequality figures
in the first semester of 2007 would come
from the fact that during this period there were growing scarcities
for consumption items of the poor, a fact
that we discuss in more detail in the next section. Given that
scarcity increases the income level necessary
to achieve certain levels of consumption, it is also likely
that even a decline in inequality of nominal
incomes would not represent a decline in inequality of real
incomes.
10.
By contrast, Ravaillon and Chen (1997) estimate the log-elasticity
of poverty reduction at between -2.59
and -3.12 on a sample of 64 developing countries. World Bank
(2000) estimates -2, a value that is similar
to that estimated by Bourgoignon (2003) and routinely used
in World Bank simulations.
11.
That number is 27.2% of the population. Since there is a
discrete mass of individuals at this income level,
it is not possible to find a level that will exactly replicate
the rate of 27.5% corresponding to the first
semester of 2007.
12.
Moreno and Rodríguez, 2007, Figure 3, p.6.
13.
Indeed, I discussed many of the causes of the looming recession
in
the article “¿Por qué se desacelera
la
enonomía venezolana?” (“Why is the Venezuelan
economy decelerating?”) published in the Venezuelan
newspaper El Nacional on January 7, 2002.
References
Bourgoignon, Francois (2004) “The Growth Elasticity
of Poverty Reduction: Explaining Heterogeneity across Countries
and Time Periods.” In
Eichner, Theo S. and Stephen J. Turnovsky (2004) Inequality
and Growth: Tehory and Policy Implications. Cambridge:
MIT Press.
Moreno, María Antonia and Francisco Rodríguez
(2008) “Plenty of Room? Fiscal Space in a Resource-Abundant
Economy” (with Maria A. Moreno) forthcoming
in Roy, Rathin and Heuty, Antoine, eds. Fiscal Space
for Human Development. New York: United Nations Development
Programme.
Ministerio del Poder Popular para las Finanzas (2007) “Exposición
de Motivos del
Proyecto de Ley de Presupsueto para el año 2008,” Caracas:
Ministerio del Poder Popular para las Finanzas
Ortega, Daniel and Francisco Rodriguez (2008) “Freed
from Illiteracy? A Closer Look at Venezuela's Robinson Literacy
Campaign.” Economic Development
and Cultural
Change. Forthcoming, October, 2008.
Ortega, Daniel and Francisco Rodriguez (2008) “A Response
to Rosnick and Weisbrot,” reproduced Wesleyan Univeristy.
http://frrodriguez.web.wesleyan.edu.
PDVSA (2006) “Información Financiera y Operacional
al 31 de diciembre de 2006”
Caracas: Ministerio del Poder Popular para la Energía
y Petróleo.
Ravallion, M. and S. Chen. 1997. What Can New Survey Data Tell
Us about Recent
Changes in Distribution and Poverty? World Bank Economic Review,
11(2), 357-382.
Rodríguez, Francisco (2008) “An Empty Revolution:
The Unfulfilled Promises of Hugo
Chávez.” Foreign Affairs, March/April 2008.
Weisbrot, Mark (2008) “An Empty Research Agenda:
The Creation of Myths About
Contemporary Venezuela,” Washington, DC: Center
for Economic Policy Research.
Weisbrot, Mark and David Rosnick. “Illiteracy Revisited:
What Ortega and Rodriguez
Read in the Household Survey” Forthcoming. Washington,
D.C.: Center for Economic and Policy Research.
World Bank (2001) World Development Report 2000/2001. Washington,
DC: The World Bank.
World Bank (2004) Keeping the Promise of Social Security in
Latin America.
Washington, DC: The World Bank.
Francisco
Rodríguez,
Assistant Professor of Economics and Latin American Studies
at Wesleyan University, was Chief Economist
of the Venezuelan National Assembly from 2000 to 2004 (frrodriguez@wesleyan.edu)
.Petroleumworld does not necessarily share these views
Editor's
Note: All
comments posted and published on Petroleumworld, do not
reflect either for or against the
opinion expressed
in the
comment as an endorsement of Petroleumworld. All comments
expressed are private comments and do not necessary reflect
the view of this website. All comments are posted and published
without liability to Petroleumworld.
Fair
use Notice: All works published by Petroleumworld are in
accordance with Title 17 U.S.C. Section 107, this material
is distributed
without profit to those who have expressed a prior interest
in receiving the included information for research and educational
purposes. Petroleumworld
has no affiliation whatsoever with the originator of this article
nor is Petroleumworld endorsed or sponsored by the originator.
Petroleumworld encourages persons to reproduce, reprint, or
broadcast Petroleumworld articles provided that any such reproduction
identify the original source, http://www.petroleumworld.com
or else and it is done within the fair use as provided for
in section 107 of the US Copyright Law. If you wish to use
copyrighted material from this site for purposes of your own
that go beyond 'fair use', you must obtain permission from
the copyright owner.
Internet web links to http://www.petroleumworld.com are appreciated.
Petroleumworld
04/06/08
Copyright© 2008
respective author or news agency. All rights reserved.
We welcome the use of Petroleumworld™ stories by anyone
provided it mentions Petroleumworld.com as the source. Other
stories you have to get authorization by its authors.
