Poverty in Pakistan 1984-1985.
Havinga, I.C. ; Haanappel, F.W. ; Louter, A.S. 等
I. INTRODUCTION (1)
Poverty analysis on Pakistan is an area of research which has
generated papers from various authors such as Naseem (1973, 1977);
Alauddin (1975); Mujahid (1978); Wasay (1977) and de Kruijk and van
Leeuwen (1985). However, due to the different units of measurement and
counting units, the results of these studies are not comparable.
Moreover, the poverty profiles developed in the past were
restricted to the geographical areas only because the authors had to
work with the published household data. This paper will extend the
analysis to most of the discriminating socioeconomic attributes of
households which are available by using primary survey data on
households and heads of households.
In short, this paper will serve a dual purpose. Firstly, it will
illustrate by way of comparison to what extent various methodologies,
using different counting units and units of measurement affect the
degree of poverty. Secondly, it will determine poverty profiles by
socio-economic attributes of the households and the heads of the
households in Pakistan.
2. THE DATA BASE
The Household Income and Expenditure Survey (HIES) provides the
only data base in Pakistan which contains a comprehensive range of
household sector variables. HIES surveys are conducted at regular
intervals and provide statistically representative data for the rural
and urban segments at provincial level. In the HIES, the household is
the statistical unit for collecting details about household composition,
patterns and levels of consumption expenditure and sources and levels of
household income.
In this study we will use the household expenditure data as a proxy
for household income. The reason is, that the expenditure data are more
reliable than the income data.
3. THE STRUCTURE OF THE PAPER
The paper is organized in four sections. First, we deal with the
concept and definition of poverty. Then, in Section 5 the measures of
poverty are discussed and in Section 6 the poverty profiles. Conclusions
are drawn in the last Section.
4. THE CONCEPT AND DEFINITION OF POVERTY
When the poverty line is defined as a limit under which persons,
families and groups of persons are excluded from the minimum standard of
living, it is important to define the minimum standard of living
expressed by this poverty line.
In this study, calorie intake has been used as the discriminating
indicator to measure poverty. It has been selected because its use has
become an accepted methodology in Pakistan; threshold calories intakes
for Pakistan have been determined; and the data are available.
The methodology used in this study is to work out the calorie
intake of households (standardized to household equivalents) on the
basis of the actual consumption of food. Households below the threshold
have been selected for further analysis. The more commonly used
methodology to determine the threshold in money terms on the basis of
the cost of a predetermined minimal bundle of goods and an applied Engel
coefficient to include non-food expenditure was rejected for two
reasons. Firstly, the application of a minimal bundle of goods assumes a
uniform consumption behaviour across households. Secondly, as the Engel
coefficient depends on total household income, the choice of the
"right" coefficient is subjective, but affects the poverty
threshold.
The selection of the households for the determination of the
poverty line based on total household food and non-food expenditure is
not correct. Measuring the calorie intake on the basis of total
expenditure per household is also inaccurate. It ignores the reality
that households differ in size and composition. An often used remedy to
adjust for household size is to deflate household expenditure by
household size, so that the total expenditure is expressed in per capita terms. However, this approach is equally unacceptable because an earner
generally needs a higher calorie intake than an adult dependent, while
an adult dependent needs more than a child. In this respect it is
mentioned that adults are defined here to include all persons of ten
years of age and above, and children as those below the age of ten.
To overcome the deficiencies outlined above, the household
equivalent scale has been used in this study to correct for size and
composition. The household equivalent scale is used for the
transformation of the values of household variables to a reference
household (i.e., a single earner household). This transformation is done
on the basis of the following conversion:
HE = [a.sub.1][X.sub.1] + [a.sub.2][X.sub.2] + [a.sub.3][X.sub.3] +
[a.sub.4][X.sub.4]
[a.sub.1] = Parameters;
[X.sub.1] = Number of earners in the household;
[X.sub.2] = Number of other adults in the household;
[X.sub.3] = Number of children which are younger than 10 years;
[X.sub.4] = Economies of scale; and
HE = Household equivalent.
The values of the parameters are obtained from the paper of Wasay
(1977) from which it can be derived that, after normalization against an
adult earner, [a.sub.1] = 1, [a.sub.2] = 0.8 and [a.sub.3] = 0.7. The
parameter applied for children (0.7) used by Wasay (1977) seems to be
rather high, taking into consideration that the average age of the
children-group is less than five years. However, for reasons of
consistency and lack of a better estimate, the value has been used here
as well. In some studies an economies of scale factor is introduced, but
we discard this factor.
For the calculation of the calorie intake per household, a detailed
conversion table [Government of Pakistan (1985)] has been used to
convert the quantities of the actual food consumed into calories. Where
necessary, these quantities have been imputed on the basis of urban and
rural provincial prices. Those prices have been obtained from the HIES
data base. The household calorie intake thus obtained is then converted
into calorie intake per household equivalent according to the above
described formula.
For the assessment of the poverty line, the calorie intake forms
the basis: Per adult earner, a minimum daily intake of 2000 kilo calorie
(kcal) is recommended by FAO/WHO (1973) and a standard of 2550 kcal is
used in Pakistan. In this study, the poverty line in rupee terms is
based on the expenditure patterns of all household equivalents with
calorie intakes between these two standards, that is, 2000 to 2550 kcal
per day per adult earner. This poverty line will be referred to as the
"high" poverty line. To test the sensitivity of the results
for the level of the poverty line, all calculations have also been
carried out for a poverty line based on a calorie intake of 1500 to 2000
kcal per adult earner per day, which will be referred to as the
"low" poverty line. Moreover, two rupee values have been
determined for poverty lines, namely the food only, "austere"
type, and the overall expenditure, "broad" type. In the
presentation of the results of the analysis, only the poverty lines on
the basis of the "broad" type have been used. For comparison,
the poverty lines has also been calculated in terms of monthly
consumption expenditure per capita. In this method, no correction for
household composition has been made.
The approach taken does not restrict the assessment of poverty to
the low income households. Those households which fall below the
stipulated minimum calorie intake level due to their socio-economic and
cultural context are covered in this study. As such, it differs from the
analysis of other authors, which arbitrarily exclude middle and high
income households.
In addition to the correction for household size and composition,
the food expenditure in rupee terms should also be corrected for
purchasing power differences between the geographical areas. The
"purchasing power parity" (ppp) has been calculated in the
following steps. First, each of the provincial food expenditure patterns
has been expressed in terms of the prices of rural Punjab, being the
area with the largest proportion of the population. Price indexes were
then determined by dividing the actual expenditure by the expenditure in
prices of rural Punjab. Subsequently, the inverse of the individual food
price indexes have been applied to derive purchasing power parity
between the rural and urban areas, and provinces. As a result,
"ppp" poverty lines can be generated for Pakistan, for urban
and rural areas, and provinces separately. The introduction of
ppp's in the regional tabulations did result only in marginal
changes. Results of the calculations using the two units of measurement
which include food price parity (ppp-adjusted household equivalents and
individuals, respectively) have therefore not been presented in this
article.
5. THE MEASURES OF POVERTY
For the assessment of poverty, various measures will be used to
develop the poverty profiles. It should be noted that the poverty line
will be expressed in either per adult earner equivalent or per capita.
In addition, three different statistical counting units will be applied
to measure the extent of poverty: households (H), individuals (I) and
adult earner equivalent (A). This latter distinction is rather crucial
to the measurement of the extent of poverty and, therefore, remarkable
that most poverty studies do not make this distinction. The measures of
poverty will be discussed briefly below.
The poverty incidence (H) expresses the proportion of the
population whose consumption is below the poverty line:
H = m/N;
m = Number of individuals, households and adult equivalents below
poverty line; and
N = Total number of individuals, households and adult equivalents
in the population.
The poverty intensity poor (I) indicates the average consumption
gap between the actual expenditure of the poor and the poverty line:
I = [m.summation over (i=1)] ([Y.sub.p] - [Y.sub.i])/m[Y.sub.p];
m = Number of individuals, households and adult equivalents below
poverty line;
[Y.sub.i] = Consumption expenditure of individual i; and
[Y.sub.p] = Poverty line.
The poverty intensity population (P) indicates the relative (over
all population) average consumption gap between the actual expenditure
of the poor and the poverty line:
P = I x H.
The Sen index ([P.sub.s]) measures the intensity of poverty:
[P.sub.s] = H[I + (1 - I)G]
G = Gini poor.
Contrary to the previous indexes, it takes into account the
distribution of the expenditure among the poor using the Gini
coefficient. The index reduces to the poverty intensity population index
in case of complete equality (G = 0). In case of complete inequality (G
= 1), the index equals the poverty incidence index.
The redistribution index (R) indicates whether the poor population
can be compensated through a shift in consumption from the rich
population to the poor. With the index smaller than 1, the rich
population is able to provide the compensation:
R = [m.summation over (i=1)] ([Y.sub.p] - [Y.sub.i])/[N-m.summation
over (i=j)] ([Y.sub.i] - [Y.sub.p])
6. THE POVERTY PROFILES
The poverty profiles developed in this section will be based on the
poverty measures presented above. The description of the profiles will
be related to the attributes of the household and the head of the
household. However, firstly the poverty lines are assessed and compared
using different methodologies.
Poverty Line Assessment with Different Methodology and the Effect
on Poverty Measures
An introductory comparison of the calorie intake measured in per
capita and per adult earner equivalent shows how sensitive the asessment
of poverty is to the unit of measurement. A simple distribution table of
calorie intake presented in Table 1 for total Pakistan indicates large
differences in percentages. With the unit of measurement being
individuals, 59 percent of the households and to 66 percent of
individuals are deficient in the daily calorie intake of 2550 kcal per
individual. When correcting for household composition and hence
measuring in adult earner equivalent, only 35 percent of the households
and 40 percent of the adult equivalents falls under the poverty line.
Already we have reflected on the theoretical need to use the equivalence
scale as a unit of measurement. As expected, the results indicate a
considerable decline in the poverty incidence when the household
composition is taken into account.
The definition of the "high" and "low",
"broad" and "austere" type poverty lines used in
this study is given in Section 4. For comparison with other studies, the
poverty lines have also been calculated on per capita basis. In Table 2,
the "broad" type poverty lines in terms of monthly expenditure
are shown in columns 3 and 7. After adjustment for price differentials
of food items, the normalized poverty lines appear in columns 4 and 8.
The poverty line expressed in monthly household expenditure is obtained
by multiplying with the average number of adult earner equivalent or
individuals per household, which is given in column 5.
The "broad" type "high" poverty lines for adult
earner equivalent and per capita, indicate a monthly expenditure of 249
rupees and 244 rupees respectively. This is equivalent to 1415 and 1610
rupees per household. For the "austere" type poverty lines
household expenditure would be 690 rupees and 784 rupees, respectively.
The latter poverty lines determine the absolute minimum to enable an
average household to meet the basic calorie intake.
Conventionally, the poverty line should include the non-food
expenditure. This will also be done in the proceeding analysis. Although
the poverty lines might seem on the high side, it should be borne in
mind that the poverty lines are based on all households whose average
daily calorie intake per individual or adult earner equivalent lies
below 2550 kcal and 2000 kcal daffy intake respectively.
Even some so-called middle income households are included in the
poor category. This is due to some extent to the fact that income
classes are based on total household disposable income. Therefore large
households with multiple low income earners have been classified in the
middle income group. Furthermore, high house rents, private health and
education charges in urban areas along with socio-economic and cultural
pressures result in much higher non-food expenditures, in absolute terms as well as in share of total expenditure, for some middle income groups
in urban area.
The concentration and the extent of poverty in the urban and rural
areas is assessed in Table 3 along the various poverty measures. At the
same time, the table indicates the sensitivity of the magnitudes of the
poverty measures for the two units of measurement, that is, per adult
equivalent and per capita.
Column 3 indicates to what extent the urban and rural areas suffer
from a particular incidence of poverty while column 4 shows the
deviation of the area's poverty from the all Pakistan average.
Comparing the "high" poverty lines, measured in adult
equivalents, 39 percent of the households, 45 percent of the individuals
and 44 percent of the adult earner equivalents can be considered poor.
The extent of poverty is larger in the urban than in the rural areas
where respectively, 42 percent and 30 percent of the households, 49
percent and 36 percent of the individuals and 48 percent and 36 percent
of the adult earner equivalents fall below the poverty line.
Similar patterns are observed for the estimation of the poverty
measures based on the "low" poverty lines. Since this
observation holds also for the other poverty measures, the description
of the results will be limited to the "high" poverty line
estimation.
The "Gini poor" indicates the expenditure inequality
between the poor household members. Comparison between the units of
measurement shows that the magnitude of the inequality increases when
measured in per capita as compared to adult equivalents. This feature is
explained by the fact that poor households have a larger number of
household members and higher dependency ratios. On the other hand, the
differences in magnitudes are marginal with respect to the different
counting units. The factual magnitudes of the Gini inequality measures
in urban areas are slightly larger than in the rural areas. Still, the
inequality is small with magnitudes of, for instance, 0.135, 0.108 and
0.120 for urban, rural and overall Pakistan measured per adult
equivalent and counting in adult equivalents.
The "intensity poor" measure shows that on average the
poor segment of the population requires an increase of around 24 percent
in expenditure to meet the shortfall from the poverty line. This measure
is reduced to 11 percent if the total population is taken into account.
Both values refer to total Pakistan measured per adult equivalent. The
urban expenditure gap was found to be larger than the rural gap.
After ranking the expenditure gaps of the poor population with the
Gini index, the Sen index is determined. The Gini index being small, the
values of the Sen index tend to reflect values similar to the
"poverty intensity population" measure. The redistribution
index indicates that 11 percent of the surplus household expenditure of
the non-poor segment of the population (above poverty line) is
sufficient to compensate the poor segment of the population measured in
adult equivalents. This index rises sharply when measured per capita.
This difference is due to the much larger proportion of the population
living in destitute conditions when measured in per capita terms.
Poverty Profiles (2)
Where is poverty concentrated, who is affected and to what extent,
are questions relevant in the poverty analysis. To answer those
questions and at the same time meet the policy requirement,
socio-economic profiles have been drawn-up. The profiles are developed
using attributes from households in general and heads of the households.
The measures of poverty will be calculated using the adult earner
equivalent unit of measurement and adult earner equivalent as counting
unit.
Although the absolute values of the poverty measures are affected
by the methodology selected, the focus of this subsection will be on the
differences in values between the classifications of attributes.
In the description below, the major focus will be on the comparison
between the urban and rural profile. The salient features of the
profiles are the following.
The size distribution indicates that in urban areas household sizes
above 7 persons (67 percent) are by 10 percentage points (57 percent) a
more observed phenomena. As the urban and rural earner distributions are
similar, the urban areas have higher dependency ratios.
Turning to the attributes of the heads of the households, the age
and sex distributions do not reveal significant discriminating features.
The comparison of distributions by educational attainment, however, does
indicate marked differences. The number of heads with educational
background above primary (87 percent) is 24 percentage points higher in
urban areas. Similarly, those with an employee working status (38
percent) are 21 percentage points higher in urban areas. This is
balanced by the decline in those with a self-employment status.
Employment of the heads of households in urban areas is 31 percent
points higher in the sales and transport occupations, but 48 percent
points lower in the agricultural occupations than in rural areas. This
characterization is reflected in the employment in the manufacturing,
wholesale and services activities, which is 38 percentage points higher
than in urban areas.
By Provinces
The extent of poverty is larger by 13 percentage points in the
urban areas (49 percent) versus the rural areas. In relative terms, the
incidence of poverty in rural Baluchistan is 34 percent higher than the
average of the rural areas, while in the rural areas of the other
provinces the extent of poverty is more or less equal to the overall
average of rural areas. The detail by provinces in the urban areas is
that poverty in Sindh is more severe by 15 percent points, while in
urban Punjab the same is 11 percent points lower than the average
incidence for urban areas. With respect to the other measures of
poverty, no striking differences between provinces can be observed.
By Size and Number of Earners
While the number of earners does not seem to influence the extent
of poverty, the size of the household seems to affect the same. This
increase in poverty with the increasing size of the households is
observed both in the urban and rural areas. This confirms our
expectations, since it is known that lower income (traditional)
households tend to be larger.
Monthly Income per Adult Equivalent
It is interesting to note that, given the prevailing poverty line
of 313 rupees of monthly expenditure per adult equivalent, still 43
percent of urban adult equivalents with a monthly income between 250 and
500 rupees fall below the poverty line. This represents 20 percentage
points of the 49 percent incidence of poverty observed in the urban
areas. Less striking is that the urban incidence of poverty is almost
100 percent for the income brackets below 250 rupees. However the
opposite phenomena is observed in the rural areas. For instance, only 63
percent poverty incidence is observed for the expenditure bracket between 150 and 250 rupees.
By Sex
Although only 4 percent of households reports a female head, the
extent of poverty is 13 percentage points less in urban areas (36.4
percent)and 11 percentage points less in rural areas (25.7 percent) than
that of the households with a male head. In relative terms, the
incidence of poverty is even more than 25 percent less.
By Age
Similar inverse U-shape patterns are observed in the rural and
urban areas with regards to the incidence of poverty, both in absolute
and relative terms. The lowest extent of poverty is observed in the age
group below 30 years of age, after which it steadily increases with a
peak for heads of households between 40 and 50 years of age.
By Education
Educational attainment seems to influence the extent of poverty
considerably given the fact that the highest absolute and relative
incidence of poverty is observed for those heads of the households
having only primary education or less after which it steadily declines
with higher levels of educational attainment. This trend., however, is
more pronounced in urban than in rural areas. Moreover, the severity of
poverty is much larger in urban than rural areas. For instance, in urban
areas, the absolute incidence of poverty of household members with a
head of the household with primary education is 60 percent, versus 38
percent in rural areas. For household members where the head of the
household has an educational level above intermediate, the incidence is
24 and 15 percent, respectively.
By Working Status
The comparison of the poverty pattern between urban and rural areas
based on the working status attribute of the heads of the households
shows interesting differences. In absolute terms, the incidence of
poverty of employers is higher in the rural areas but much lower for the
self-employed. For those with employee status, the incidence of poverty
is about the same. In relative terms, however, those in the rural areas
with an employee working status have a relative high incidence of
poverty. It may be observed that those households which heads are
unemployed are not adversely affected by additional poverty as compared
to other working status. This phenomena is of course a direct
consequence of the extended family system.
By Occupation
Members of households of which the occupation of the head is in the
professional, administrative and clerical categories are less prone to
poverty than in other categories. This phenomena is observed in absolute
and relative terms both in urban and rural areas. In contrast, those
characterized by occupations in the service, production and transport
categories are confronted with higher relative and absolute incidence of
poverty, both in rural and urban areas.
By Activity
In line with the incidence of poverty according to the
classification by occupation, relatively higher incidences of poverty
are observed in mining, manufacturing, construction, sales and transport
activities. Those household members characterized by heads employed in
the remaining activities like electricity, finance and government
services are relatively less affected by poverty. This contrast is, of
course, a reflection of the dual nature of the labour market.
Finally some remarks about the comparison between the poverty
profiles based on the poverty "high" and "low"
lines. As was observed earlier, generally, both poverty lines generate
similar patterns in the poverty profiles although the absolute values of
the poverty measures have declined except for the household attribute
'region'.
By region, a clear change in the pattern is observed. For total
Pakistan, a significant drop in the incidence of poverty for Baluchistan
and to a lesser extent for Punjab is shown. Those declines are explained
by the drop in incidence of poverty both in the rural and urban areas.
In contrast, the incidence of poverty hardly declines in the urban areas
of NWFP and rural areas of Sindh. As a result, it may be concluded that
the degree of deprivation found in the urban areas of NWFP and the rural
areas of Sindh is more severe (and contain a structural element) than
that observed in other geographical areas.
7. SUMMARY AND CONCLUSIONS
The analysis of poverty presented in this report uses the original
full HIES data base. It is the first time that this type of analysis has
been carried out in Pakistan.
A second important contribution of this study is the development of
poverty indicators using different methodologies, units of measurement
and counting units. It was found that substantial differences exist in
the magnitude of poverty and inequality when measured in various ways.
This implies that extra care should be taken in international
comparisons and inter-temporal analysis to use compatible methodologies
and specifications.
A further contribution of this analysis lies in the determination
of poverty lines for the four provinces of Pakistan, in addition to the
overall average.
The methodology used differs in some aspects from that found
elsewhere in the literature, in particular with respect to the
determination of the Engel coefficient, which is not needed in this
study.
It was found that the poverty line for 1984-85 for overall Pakistan
per household in terms of household equivalents (average household size)
was 690 rupees per month for food expenditures (i.e. the absolute
minimum, "austere" type of poverty line) and 1415 rupees per
month when including all expenditure. The poverty line varies across
provinces and between rural and urban areas.
Another contribution of this report is the presentation of detailed
poverty profiles for Pakistan based on attributes of the household and
the head of the household. The poverty profiles are drawn up using
conventional poverty measures Future poverty analysis should extent to
the identification of specific contribuants "to poverty in a joint
analysis of the various attributes.
Comments on "Poverty in Pakistan 1984-85"
The subject of income distribution is of great interest to the
economists and policy-makers of this country. With the availability of
primary data it has now become possible to carry out more meaningful
analysis of poverty and income inequality. Poverty profiles can be
generated on the basis of different attributes of households like
education, age, occupation, geographic location, number of earners in a
household, etc. All this has been accomplished in the present paper,
hence it represents an improvement over those earlier works based on
secondary or grouped data and were limited in scope.
Most of my comments are of a specific nature. Household size and
composition have been taken into account in estimating the poverty
level. For this, adult earner equivalents of each household have been
worked out. The following three groups have been defined to compute adult earner equivalents: adult earners, adults and children. All the
individuals of age 10 and above are treated as adults. This
classification is quite restrictive. Adult equivalence scales have been
worked out for more detailed age groups of females and males separately,
(1) whose use can refine the work.
Poverty lines have been estimated by using the caloric requirement
approach. Two caloric intake ranges of 1500-2000 and 2000-2550 have been
defined. All the households having these levels of calories per adult
earner equivalent have been separated. By adding food and non-food
expenditure of those households, two poverty lines have been
estimated--one corresponding to the lower and the other to the higher
range of caloric intake. The problem with this approach is that actually
rich households can enter in determining the poverty line. There might
be many rich households whose caloric intake per household equivalent
may be within the range of 2000-2550 calories. It is possible that they
are diet conscious or their consumption patterns are such that they fall
into this particular range of caloric intake. Rich households generally
consume better quality food products whose prices are relatively higher
and consequently their food expenditure would be higher. The inclusion
of food expenditure of such households to determine poverty lines would
appear less problematic when we consider non-food expenditure. Non-food
expenditure of rich households would certainly be higher than needed to
define a poverty line. In a nutshell, with the approach followed in the
paper, the expenditure of rich households can enfluence the
determination of poverty lines, making them biased upward.
I see a problem, at least theoretically, in using a range of
different caloric intake levels instead of one specific value in
defining a poverty line. Poverty lines have been estimated for provinces
and their rural and urban areas. If households in one region are highly
concentrated towards the upper range of caloric intake and the converse is true for another region, the estimated poverty lines ill such cases
would not be comparable. It is possible that this problem may not have
arisen in the paper.
In one of the exercises poverty lines computed for different
regions have been normalized for price differentials by applying the
price level prevailing in the rural Punjab to all other regions of the
country. Another normalization is possible which removes differentials
in consumption patterns. In this case, the same consumption pattern for
all regions of the country would have been used with regional price
differentials. I consider this normalization more interesting and useful
because the same consumption basket is being allowed to everyone in the
country and considering those people who cannot afford to have that
basket.
In the second part of the paper, inequality has been estimated in
consumption expenditure only. It would have been interesting to see
inequality estimates on the basis of income also, because inequality on
the basis of expenditure is depressed.
Muhammad Hussain Malik
Resident Mission of the World Bank, Islamabad.
REFERENCES
Alauddin, Talat (1975) Mass Poverty in Pakistan: A Further Study.
The Pakistan Development Review 14 : 4.
F.A.O. (1973) Energy and Protein Requirements. Report of a Joint
FAO/WHO ad hoc Expert Committee. Rome: Food and Agriculture Organization
of the United Nations.
Havinga, I. C., F. W. Haanappel, A. S. Louter and W. A. van den
Andel (1989) Poverty and Inequality in Pakistan 1984-85. Project on
Improvement of National Account Statistics. Islamabad: Federal Bureau of
Statistics. The Hague: Institute of Social Studies.
Kruijk, H. de, and M. van Leeuwen (1985) Changes in Poverty and
Income Inequality in Pakistan during the 1970s. The Pakistan Development
Review 24 : 3 & 4.
Mujahid, G. B. S. (1978) A Note on Measurement of Poverty and
Income Inequalities in Pakistan: Some Observations on Methodology. The
Pakistan Development Review 17 : 3.
Naseem, S. M. (1973) Mass Poverty in Pakistan: Some Preliminary
Findings. The Pakistan Development Review 12 ." 4.
Naseem, S. M. (1977) Rural Poverty and Landlessness in Pakistan.
Poverty and Landlessness in Rural Asia. Geneva: ILO.
Pakistan, Government of (1984) Household Income and Expenditure
Surveys Islamabad: Ministry of Finance, Planning and Development,
Statistical Division.
Pakistan, Government of (1985) Food Composition Table for Pakistan.
Islamabad: Planning and Development Division.
Wasay, Abdul (1977) An Urban Poverty Line Estimate. The Pakistan
Development Review 16 : 1.
(1) This article is an abbreviated version of the study
"Poverty and Inequality in Pakistan, 1984-85", carried out by
the Pakistan/Dutch Project on Improvement of National Accounts
Statistics implemented in collaboration with the Federal Bureau of
Statistics
(2) The description of the profile will be restricted to the
estimates based on the "high" poverty line. Besides, it is
mentioned that the profile reported coincides with the profile based on
the conventional relative poverty line of 75 percent of the average
expenditure per adult equivalent. For that matter, the profile developed
for the "low" poverty line coincides with a relative poverty
line of 65 percent of the average expenditure per adult equivalent.
Complete sets of tables are available on request.
I. C. HAVINGA, F. W. HAANAPPEL, A. S. LOUTER and W. A. van den
ANDEL *
* The first three authors are associated with the Institute of
Social Studies, The Hague, The Netherlands. The fourth author is
associated with the Erasmus University, Rotterdam, The Netherlands.
(1) See M. Irfan and Rashid Amjad (1983) Poverty in Rural Pakistan.
In A. R. Khan and E. Lee (eds) Poverty in Rural Asia. Bangkok:
ILO/ARTEP.
Table 1
Percentage Distribution of Calorie Intake for Pakistan 1984-85
Per Adult
Calorie Intake Per Capital Equivalent Average Calorie
Intake per
HH Ind HH Ind Adult Equivalent
< 1500 8.5 10.2 2.7 2.9 1052
1500 - 2000 22.2 25.8 10.0 11.8 1799
2000 - 2550 28.2 30.0 22.4 25.1 2291
2550 - 3000 16.2 15.5 21.0 21.9 2769
> 3000 24.8 18.5 43.9 38.3 4063
Total 100.0 100.0 100.0 100.0 3088
Source: [Government of Pakistan (1984, 1985)].
Table 2
Poverty Line Assessments, 1984-85
(Rupees per Month)
Per Adult Equivalent
Normal- Number of
ized Individuals
Region Food Total Total per HH
1 2 3 4 5
Pakistan Urban 141.1 *) 312.6 299.0 6.011
116.4 *) 260.4 248.7 6.486
Rural 110.6 214.4 213.5 5.525
89.5 179.8 178.9 5.716
Total 121.3 248.8 243.4 5.686
101.0 214.3 208.8 6.021
Per Household
Normal-
ized
Region Food Total Total
1 6 7 8
Pakistan Urban 848.2 1879.0 1797.3
755.0 1689.0 1613.1
Rural 611.1 1184.6 1179.6
511.6 1027.7 1022.6
Total 689.7 1414.7 1384.0
608.1 1290.3 1257.2
Per Individual
Normal- Number of
ized Individuals
Region Food Total Total per HH
1 2 3 4 5
Pakistan Urban 144.1 323.0 309.2 6.9
108.9 240.5 229.9 7.7
Rural 108.9 212.8 211.7 6.5
85.6 167.7 167.0 7.0
Total 118.8 243.8 239.1 6.6
94.0 193.9 189.6 7.2
Per Household
Normal-
ized
Region Food Total Total
1 6 7 8
Pakistan Urban 988.2 2215.1 2120.5
835.9 1846.1 1764.7
Rural 709.2 1385.8 1378.6
597.1 1169.9 1165.0
Total 784.8 1610.5 1579.5
678.0 1398.6 1367.6
Notes: (1.) The two values in each cell reflect the estimates based
on the "high" poverty line (top) and "low" poverty line (bottom),
respectively.
(2.) Similar poverty lines by provinces are available in the
comprehensive report, [see 1. C. Havinga, et al. (1989)].
Table 3
Poverty Measures and the Effect-of Methodologies, 1984-85
Poverty Poverty
Count Incidence Incidence Gini
Region Unit (Absolute) (Relative) Poor
1 2 3 4 5
Based on Adult Equivalents
Urban H 0.420 *) 1.088 0.130
0.283 *) 1.076 0.113
I 0.491 1.098 0.135
0.340 1.083 0.118
A 0.485 1.095 0.135
0.334 1.077 0.117
Rural H 0.302 0.782 0.105
0.163 0.620 0.095
I 0.362 0.810 0.108
0.203 0.646 0.097
A 0.358 0.808 0.108
0.201 0.648 0.096
Total H 0.386 1.000 0.116
0.263 1.000 0.104
I 0.447 1.000 0.120
0.314 1.000 0.107
A 0.443 1.000 0.120
0.310 1.000 0.107
Based on per Capita
Urban H 0.585 1.102 0.160
0.377 1.102 0.129
I 0.666 0.103 0.166
0.448 1.106 0.134
A 0.658 1.102 0.165
0.440 1.103 0.133
Rural H 0.473 0.891 0.126
0.266 0.778 0.104
1 0.551 0.912 0.130
0.323 0.798 0.107
A 0.545 0.913 0.129
0.318 0.797 0.107
Total H 0.531 1.000 0.139
0.342 1.000 0.115
1 0.604 1.000 0.143
0.405 1.000 0.118
A 0.597 1.000 0.142
0.399 1.000 0.118
Intensity Intensity Sen Redist
Region Poor Population Index Index
1 6 7 8 9
Based on Adult Equivalents
Urban 0.269 0.113 0.153 0.142
0.228 0.064 0.089 0.031
0.280 0.138 0.186 0.323
0.239 0.081 0.112 0.068
0.279 0.135 0.182 0.302
0.238 0.079 0.109 0.064
Rural 0.207 0.063 0.088 0.050
0.185 0.030 0.043 0.007
0.215 0.078 0.109 0.101
0.189 0.038 0.054 0.015
0.214 0.077 0.107 0.097
0.189 0.038 0.054 0.014
Total 0.237 0.091 0.126 0.110
0.205 0.054 0.076 0.027
0.247 0.110 0.151 0.219
0.212 0.067 0.093 0.053
0.246 0.109 0.149 0.208
0.211 0.066 0.092 0.051
Based on per Capita
Urban 0.338 0.198 0.260 0.702
0.260 0.098 0.134 0.084
0.354 0.235 0.307 1.838
0.271 0.122 0.165 0.201
0.352 0.231 0.302 1.680
0.270 0.119 0.162 0.184
Rural 0.259 0.122 0.166 0.297
0.208 0.055 0.077 0.031
0.269 0.148 0.200 0.666
0.215 0.069 0.097 0.067
0.267 0.146 0.197 0.623
0.214 0.068 0.095 0.062
Total 0.290 0.154 0.206 0.458
0.235 0.080 0.110 0.067
0.303 0.183 0.243 1.015
0.243 0.098 0.135 0.143
0.301 0.180 0.239 0.945
0.242 0.096 0.132 0.133
Source: [Government of Pakistan (1984, 1985)].
Note: 1. The two values in each cell reflect the estimates based
on the "high" poverty line (top) and "low" poverty line (bottom),
respectively.
