Multidimensional Poverty

in the Philippines: Trend,

Patterns, and Determinants

Geoffrey Ducanes and Arsenio Balisacan

Multidimensional Poverty - Philippines
There is government awareness that focus
should be on poverty’s many aspects not
just income poverty
This is evident in the Medium-term Philippine
Development Plan of every president since
1992 which refers to human development goals
and not just income poverty targets.
Due mainly to effective lobbying by NGOs like
the Human Development Network
Multidimensional Poverty - Philippines
e.g. KALAHI-CIDSS
acronym for current government’s flagship
poverty project (roughly translatable to Arm-in-
arm Against Poverty)
involves funding support for likes of road,
water, health and day care projects for selected
towns/municipalities
Multidimensional Poverty - Philippines
e.g. KALAHI-CIDSS
 steps in town selection
1. Choosing 20 poorest provinces out of 78 total in
terms of official income poverty
2. Within each of these 20 provinces, choosing
eligible municipalities based on a composite index
of income level, food consumption, clothing
consumption, quality of shelter, disaster
vulnerability, and citizen participation
3. etc.
Multidimensional Poverty - Philippines
Still, the literature in the country on
multidimensional poverty is lagging
compared to income poverty. Two main
reasons
Income poverty, rightly or wrongly, is seen to
be the more pressing problem. Justification for
this may take the following form, for instance.
Income poverty more pressing?
Medium human % difference
Indicator Philippines development
countries

Per capita GDP 4,170 4,269 -2.3

Adult literacy 92.6 80.4 15.2

Combined
81 64 26.6
enrollment ratio

Life expectancy 69.8 67.2 3.9

Multidimensional Poverty - Philippines
Data constraints. Many important non-income
indicators such as literacy rates, mortality rates,
life expectancy, and nutrition status of children,
access to health and education facilities are
obtained either at long intervals of time or
irregularly
Data frequency

Life expectancy every 10 years

Infant mortality every 10 years

survey held twice in last

Literacy 15 years, with definition
changing
held thrice in last 15
Nutrition years by different
agencies
Multidimensional Poverty - Measurement
Multidimensional indices have been constructed at
the level of provinces. Important particularly in
making local leaders and the people more
accountable for their performance.
 HDI – real per capita income, primary and secondary
enrolment rate, high school graduate ratio, and life
expectancy
 HPI – probability at birth of not surviving to age 40,
functional illiteracy rate, % not using improved water
sources, and % of underweight children under 5
Multidimensional Poverty - Measurement

 Quality of Life Index (QLI) – under-5 nutrition rate,

attended births, elementary cohort survival rate,
 Minimum Basic Needs Index (MBN) – # of families
below the official poverty line (n), incidence of official
poverty in the province (%), cohort non-survival rate
(%), population illiteracy rate (%), infant mortality
rate (per 1,000 livebirths), malnutrition rate (%),
households without access to safe water (%),
households with no sanitary toilets (%)
 Multidimensional Poverty - Measurement
Table 1. Spearman's Rank Correlations of Provincial Welfare Measures*
FLOL Official
MBN'
Indicator HDI HPI GRDI QLI poverty poverty
Index
incidence** incidence***
HDI 1 . . . . . .
HPI -0.53 1 . . . . .
GRDI 0.98 -0.57 1 . . . .
MBN' Index 0.62 -0.76 0.65 1 . . .
QLI 0.65 -0.66 0.68 0.78 1 . .
FLOL
poverty -0.84 0.39 -0.83 -0.59 -0.53 1
incidence** .
Official
poverty -0.80 0.55 -0.81 -0.77 -0.65 0.74 1
incidence***
*Using provincial level data as unit of analysis
**Uses fixed-level-of-living poverty lines and per capita expenditure
***Uses government computed poverty lines and per capita income
Multidimensional Poverty - Measurement
Table 2. No. of provinces identified in common among 20 poorest

FLOL Income
MBN'
Indicator HDI HPI QLI poverty poverty
Index
incidence incidence
HDI 20 . . . . .
HPI 12 20 . . . .
MBN' Index 12 13 20 . . .
QLI 10 10 9 20 . .
FLOL
poverty
incidence 13 9 9 6 20 .
Income
poverty
incidence 15 11 10 8 11 20
Multidimensional Poverty - Measurement
Multidimensional Poverty - Patterns
Table 3. Regional Welfare Indicators (2000)*
FLOL Income
MBN'
HDI HPI GRDI QLI Poverty Poverty
Region** Index
(2000) (2000) (2000) (1999) Incidence*** Incidence****
(1994)
(2000) (2000)
CAR 0.620 19.5 0.574 0.57 0.71 20.1 44.2
1 0.639 12.8 0.602 0.72 0.8 20.2 43.7
2 0.567 14.7 0.539 0.72 0.78 29.6 36.2
3 0.634 11.7 0.591 0.73 0.78 16.4 23.0
NCR 0.830 9.6 0.732 . . 5.6 12.1
4A 0.669 12.1 0.621 0.77 0.78 14.7 24.8
4B 0.535 15.3 0.51 0.64 0.59 39.2 60.2
5 0.523 17.8 0.503 0.56 0.59 49.7 62.9
6 0.587 20 0.552 0.59 0.6 28.1 51.5
7 0.563 17.7 0.537 0.67 0.75 39.3 44.0
8 0.519 18.4 0.495 0.61 0.60 46.8 51.6
9 0.530 23.6 0.505 0.47 0.61 49.0 54.9
10 0.606 16.6 0.558 0.59 0.71 31.2 49.3
11 0.594 21.7 0.553 0.58 0.59 23.1 45.0
12 0.569 20.5 0.538 0.51 0.57 32.5 59.2
13 0.520 17.4 0.499 0.54 0.59 33.9 56.7
ARMM 0.395 31.1 0.381 0.37 0.55 58.9 72.6
*Regional figures are population-weighted averages of provincial figures in Appendix Table 1.
**CAR – Cordillera Administrative Region; NCR – National Capital Region; ARMM – Autonomous Region of
Muslim Mindanao
***Based on fixed level of living poverty lines and per capita expenditure.
****Based on per capita income
Multidimensional Poverty - Patterns
The most glaring pattern is that regardless of
which welfare indicator is used
 Provinces (or regions) adjacent to and including Metro
Manila, the country’s capital, have the most favorable
levels, almost without exception
 The provinces in one region, the Autonomous Region
of Muslim Mindanao, performs most poorly in almost
all indicators. This is the region where majority of the
country’s Muslim population is found and the base of a
long standing armed conflict between secessionist
groups and the government.
Multidimensional Poverty - Determinants
We examine multidimensional poverty in relation to
a. geographical/topographical factors,
b. infrastructure, and
c. political economy variables
Geographical/topographical factors

 Climate and topography, for instance, affect livelihood

patterns, food production, and shelter ,
 Climate is also intimately related with disease burdens
(such malaria in tropical areas, meningitis in mountainous
areas) and health
 Difficult terrain, as well as frequent inclement weather
also makes children’s access to school more grueling.

In our regressions, geography is represented by dummies for

climate type, as well as a dummy for whether a province is
predominantly mountainous and a dummy if it is coastal.
Infrastructure

 Infrastructure facilitates trade and travel, raising income

levels
 Infrastructure, say in the form of a good road network also
facilitates the construction of, and transport to, further
infrastructure such as markets, school buildings, and health
centers.

Infrastructure is represented by road density and an indicator variable

for the presence of international ports in the province. In
addition, the population density, which is closely linked to the
level of urbanization in an area, is included as an additional
proxy infrastructure variable.
Political economy variables

 Good governance, for instance, should lead to better welfare for the
constituents
 The presence of armed conflict in an area, insofar as it represents a
direct threat to life and health, impedes access to education and health
facilities, and represents a grave psychological burden, should be
detrimental to well-being.

As measures of good governance, we include a measure for the extent of local

political dynasty and also provincial per capita budget expenditure on
education. To represent conflict, we include a dummy for significant
presence of communist armed insurgence (CPP-NPA) in the area and
also a dummy for the Autonomous Region of Muslim Mindanao, a
historically contentious region and the main base of Muslim insurgents.
Regression Results
Table 4. Regression Results
HDI 2000 HPI 2000
Variable Coeff p-value Coeff p-value
Climate type 2 -0.08 0.00 *** 1.86 0.25
Climate type 3 -0.05 0.01 *** 3.48 0.02 **
Climate type 4 -0.07 0.00 *** 4.18 0.01 ***
Mountainous 0.01 0.80 0.58 0.59
Coastal 0.01 0.56 1.35 0.45

International port 0.01 0.69 0.20 0.86

Road density 1990 0.02 0.54 -4.64 0.02 **
Population density 1990 (000) 0.16 0.01 *** -2.05 0.44

Dynasty -0.06 0.02 ** 1.04 0.65

Educ expend per capita (P000) 0.04 0.17 0.00 0.80
Communist insurgency -0.02 0.16 2.44 0.06 *
ARMM -0.15 0.00 *** 18.57 0.00 ***

Intercept 0.55 0.00 16.32 0.00

No. of observations 72 72
R2 0.673 0.668
*significant at the 10% level; **significant at the 5% level;***significant at the 1% level
****Regressions were done in Stata 8 using the robust method, which uses White’s adjusted standard error
estimates. Diagnostic tests on multicollinearity, omitted variables, and normality of residuals were made and
except in the case of the normality of residuals in the HDI regression, all were passed at the 5% level.
Regression Results
Table 4. Regression Results
MBN 1994 QLI 1999
Variable Coeff p-value Coeff p-value
Climate type 2 -0.09 0.00 *** -0.05 0.08 *
Climate type 3 -0.09 0.00 *** -0.07 0.01 **
Climate type 4 -0.11 0.00 *** -0.06 0.07 *
Mountainous -0.02 0.48 -0.04 0.03 **
Coastal -0.08 0.01 *** 0.04 0.15

International port 0.08 0.03 ** 0.05 0.02 **

Road density 1990 0.05 0.19 0.14 0.00 ***
Population density 1990 (000) 0.17 0.01 ** 0.12 0.02 **

Dynasty -0.09 0.08 * -0.03 0.30

Educ expend per capita (P000) 0.29 0.01 *** 0.29 0.05 *
Communist insurgency -0.04 0.08 * -0.02 0.21
ARMM -0.22 0.00 *** -0.09 0.00 ***

Intercept 0.57 0.00 0.56 0.00

No. of observations 72 72
2
R 0.70 0.79
*significant at the 10% level; **significant at the 5% level;***significant at the 1% level
****Regressions were done in Stata 8 using the robust method, which uses White’s adjusted standard error
estimates. Diagnostic tests on multicollinearity, omitted variables, and normality of residuals were made and
all were passed at the 5% level.
Regression Results

Regression results show in the case of Philippine provinces

 Geography, infrastructure, and political factors are robustly
related to multidimensional welfare levels.
 For policy, geographical features maybe made one basis for
targeting, although a closer study must be made to trace the
exact path/paths through which geographical factors are
transmitted to welfare levels, and then design interventions
appropriately.
 Infrastructure investment, good governance, and a quick and
peaceful resolution to the armed conflicts must all be pursued
to improve multidimensional welfare in the lagging provinces.
End