10 th National Convention on Statistics (NCS) EDSA Shangri-La Hotel October 1-2, 2007

Transcription

1 10 th National Convention on Statistics (NCS) EDSA Shangri-La Hotel October 1-2, 2007 Population and Poverty Nexus: Does Family Size Matter? by Romulo A. Virola and Arturo M. Martinez, Jr. For additional information, please contact: Author s name : Romulo A. Virola Designation : Secretary General Affiliation : National Statistical Coordination Board Address : 403 Sen. Gil Puyat Avenue, Makati City Tel. no. : (0632) / (0632) ra.virola@nscb.gov.ph Co-author s name : Arturo M. Martinez, Jr. Designation : Statistical Coordination Officer II Affiliation : National Statistical Coordination Board Address : 403 Sen. Gil Puyat Avenue, Makati City Tel. no. : (0632) am.martinez@nscb.gov.ph

2 Population and Poverty Nexus: Does Family Size Matter? 1 by Romulo A. Virola & Arturo M. Martinez Jr. 2 ABSTRACT The emerging development paradigm on poverty reduction recognizes, as it should, the multi-dimensional nature of poverty. Poverty reduction strategies, therefore, require an effective assessment and management of the many facets of economic dearth. This should include a clear understanding of vulnerability to poverty and the important determinants of the welfare status of the Filipino family. Researches show that poverty and population are closely linked. While many past studies have identified the various correlates of poverty, only few dealt specifically with the family size variable. This paper therefore, attempts to establish a quantitative link between the issues affecting the two critical concerns of poverty and population in the development efforts of the nation. In particular, it highlights the relationship between family size and the vulnerability of a family to poverty. Through standard statistical techniques and using official poverty statistics and information provided in the triennial Family Income and Expenditures Survey, the paper provides an analysis of the correlates of poverty with specific focus on family size. It presents differentials in the economic status, e.g., income, expenditure, saving, etc. and social status, e.g., education, health, etc. and characterizes Filipino families by family size. Thus, the study exhibits the relationship between population management and poverty reduction and provides guidance towards the formulation of effective poverty reduction programs and strategies. Keyworkds: poverty, inequality, family size, income distribution, Foster Greer Thorbecke measures, poverty incidence, poverty gap, poverty severity, demographic transition, demographic dividend I. Introduction Statistics show that poverty and inequality in the Philippines are more pervasive than in neighboring countries in Asia. Although decreasing, the poverty incidence in the country has been declining at a slower pace than in many Asian countries. Among five countries in Table 1.1 for which data are available, poverty reduction rate is highest in Thailand at 10.4 percent per year followed very closely by Vietnam at 10.2 percent and lowest in the Philippines at 2.4 percent. In the Philippines, the one dollar poverty incidence decreased by less than 4 percentage points in a span of 9 years. Moreover, inequality is most severe in the Philippines posting a Gini coefficient, the highest among the eight countries in Table Paper presented during the 10 th National Convention on Statistics at the EDSA Shangri-la Plaza Hotel, Mandaluyong City, Philippines on 1-2 October Secretary General and Statistical Coordination Officer II, respectively, of the National Statistical Coordination Board. The views expressed in this paper are those of the authors and do not necessarily reflect the views of the NSCB. The authors acknowledge the assistance of Noel S. Nepomuceno in the preparation of this paper.

3 Table 1.1 Poverty and Inequality in selected Asian countries Poverty 1 Inequality 2 (Consumption) Country Year(s) Annual 3 reduction (%) Initial year Final year Gini (final year) Reference Cambodia UN and ADB Indonesia UN and ADB Korea, Rep. of UN and ADB Lao PDR ADB Malaysia UN and ADB Philippines UN and ADB Thailand UN and ADB Viet Nam ADB Notes: (1) The estimates are sourced from the UNStats Millennium Indicators website. Poverty incidence is the proportion of the population below $1 poverty line. These estimates are generally not equal to the official poverty statistics computed by the national statistical offices of countries. For example in the Philippines, the official headcount poverty rate in 2003 stands at 30% as computed by.the National Statistical Coordination Board (NSCB). (2) Estimates of Gini coefficient are sourced from the ADB Special Chapter of 2007 Key Indicators. These are expressed in percentage and are based on expenditures. (3)The annual reduction of poverty rates is computed as an annualized rate. (4) All estimates for Lao PDR and Viet Nam are sourced from the ADB Special Chapter of 2007 Key Indicators. Table 1.2 GDP at market prices, current US$ (WB estimates) (1960s and 2005) (in millions) Country Year Cambodia , Indonesia 87, , Japan 44, , , ,505, Korea, Rep. of 3, , , , Malaysia 2, , , , Nepal , Philippines 6, , , , Thailand 2, , , , Source: United Nations Statistics Division Common Database website And yet, in the sixties, the Philippines had one of the most vibrant economies in Asia, second only to Japan. And while Japan continues to be an economic power, it is quite

4 obvious that the Philippines has lost so much grounds, having been overtaken by Korea, Thailand and Malaysia. The poverty nexus is therefore an important Philippine development issue that has to be carefully explored and systematically examined if we are to hope to be able to address the Filipino diaspora that has cost social damage to the Filipino nation that will have to be borne by future generations. Also if we are to have a chance against emerging economic giants like China and India. Despite many efforts to study the causes and correlates of poverty, the slow progress in reducing poverty worldwide is indication of the complexity and multidimensional nature of poverty. Towards enriching the literature on poverty research, this paper looks at the most basic unit of the Philippine society the family, and directly examines the relationship between family size and poverty, though not necessarily as one of cause and effect. The debate on the role of population growth and family size in poverty is largely unresolved (Orbeta, 2005). What characterizes a poor Filipino family? What makes a family vulnerable to poverty? When properly addressed, the answer to these questions will aid in the formulation of social policies that are efficiently pro -poor. Obviously, larger family sizes means larger expenditures for food, clothing and shelter, education, health and other needs. It is therefore easy to understand why population management has b ecome central to many development efforts being pushed by the international community in poor countries. According to a report (UN, 1999) from the Cairo Consensus at Ten: Population, Reproductive Health and the Global Effort to End Poverty, there is clear evidence that enabling people to have fewer children, if they want to, helps to stimulate development and reduce poverty, both in individual households and at the macro-economic level. But not everybody is convinced that overpopulation is the cause of underdevelopment. The emergence of China and India, where 2 out of 5 earthlings live lends some credibility to the view that population size per se and population growth may not be that bad for a country. In addition, that many countries have reaped the so-called demographic dividend is proof that population growth may be good for a particular country at a certain stage of development. In fact, developed countries like Singapore and Germany are now pushing for couples to bear children, turning around from their development strategy in the past. In the Philippines, the Catholic church has been visibly opposing some population programs particularly those geared towards population control through artificial methods. Of course, it cannot be denied that other factors like bad governance, graft and corruption, lack of education, etc. are related to poverty, but to say that they are the cause rather than the effect, is debatable. In addition, many complain that despite the economic growth experienced by the country in the recent and even distant past,

5 the benefits have not trickled down to the poorer sectors of society that the rich have indeed, become richer, but the poor have also become poorer. By looking at the income distribution, the validity of such claims can be ascertained; and if validated, points to the need for our development planners to be more creative in crafting more effective development programs as well as to prick the conscience and sense of social responsibility of the rich. This also raises the question whether the persistence of poverty in Philippine society is a case of having dysfunctional poor or malfunctional rich or both. This paper presents empirical evidence on the link between family size and the poverty status of a family and dissects the inequality in income distribution in the next section. In the third section, it quantifies the prospects for the Philippines experiencing the demographic dividend presenting hypothetical poverty scenarios for the Philippines if population management had been more successful in arresting population growth. The last section deals with some concluding statements and recommendations on the use of better statistics in evidence-based decision-making and for better development results. II. Poverty and Inequality: Does Family Size Matter? In 1960, the Republic of Korea, Philippines and Thailand had roughly the same population, less than Vietnam. But by 2007, as depicted in Table 2.1, the Philippines had a population close to double that of Korea, close to 1 ½ times that of Thailand and bigger than Viet Nam. In addition, the total fertility rate (TFR) and the population growth rate for the Philippines during the period remain the highest among the seven countries in Table 2.1. At the same time, Table 2.2 shows that in 2005, the per capita GDP of the Philippines only exceeded that of Vietnam, was less than half that of Thailand, less than a fourth that of Malaysia, about 7 percent that of Korea and 4 percent that of Singapore. And in 2004, together with Indonesia, the Philippines had a double digit unemployment rate, much higher than the rate for the other countries (Table 2.3). With fertility rates remaining high, if rapid population growth indeed causes poverty, prospects for the Philippines are ominously dim compared to our neighbors. From this point, the cross country comparisons will focus on the Philippines and six other neighboring countries in Asia i.e., Indonesia, Republic of Korea, Malaysia, Singapore, Thailand and Viet Na m, encompassing different social indicators. Observations and analysis are then placed in the context of the population and poverty nexus, the topic of interest.

6 Table 2.1 Indicators on Population Population (in thousands) Sex ratio of population Total Fertility rate ( ) Annual Population growth rate ( ) Country Total (1960) Total (2007) Men Women men / 100 women % Indonesia Korea, Rep. of Malaysia Philippines Singapore Thailand Viet Nam Source: United Nations Statistics Division Common Database website Notes: Population projections are revised every two years in order to incorporate new data. T he preparation of projections by the Population Division of the United Nations Secretariat, involves two distinct processes: (a) the incorporation of all relevant information regarding past demographic dynamics of the population of each country or area of the world; and (b) the formulation of detailed assumptions about the future paths of fertility, mortality, and international migration. Table 2.2 Indicators on Income and Economic Activity Country Per Capita GDP (US$) 2005 Adult (15+) economic activity rate Year Total Men Women Source Indonesia 1, a HS Korea, Rep. Of 16, a,b HS Malaysia 5, a,d 64 HS Philippines 1, a,c HS Singapore 26, HS Thailand 2, a HS Viet Nam HS Source: United Nations Statistics Division Common Database website. Per capita gross domestic product (GDP) in US dollars is calculated by the Statistics Division of the United Nations Secretariat primarily from official national accounts statistics in national currencies provided by nation al statistical services.. Notes: HS stands for Household Survey. The superscripts correspond to the following: (a) refers to civilian labour force only, (b) excluding contributing family workers working less than 18 hours per week, (c) including members of armed forces living in private households and (d) Age group 15-64

7 Table 2.3 Indicators on Unemployment Country Adult (15+) unemployment rate Year Total Men Women Source Indonesia Household Survey Korea, Rep. of 2004 a Household Survey Malaysia 2004 b Household Survey Philippines Household Survey Singapore Household Survey Thailand Household Survey Viet Nam Household Survey Source: United Nations Statistics Division Common Database website Notes: The adult unemployment rate refers to the proportion of the adult (aged 15 years and older) labour force that is unemployed, unless otherwise specified. The unemployed are persons who are currently without work, who are currently available for work, and who are seeking or have sought work recently. The base for these statistics is the labour force (that is, the economically active portion of the population), not the total population. The superscripts cor respond to the following: (a) refers to civilian labour force only and (b) corresponds to age group POVERTY AND FAMILY SIZE Computations from the 2000 and 2003 Family Income and Expenditures Survey (FIES) of the NSO show a definite pattern of decreasing per capita income, decreasing per capita expenditure and decreasing per capita savings as family size increases, although some blips show between families with size 7 to 10 and families with size greater than 10. This inverse relationship is true fo r the entire country as well as for each of the major island groups. Table 3 shows that average-sized households enjoy nearly close to 1 ½ times as much per capita income and expenditure as families with more than ten household members. Looking at the savings variable, a logical hypothesis is that smaller families are able to save more per capita than bigger families or that additional children exposes families to increased risks of income shortfalls. For detailed discussion on relationship between savings and family size, the readers are referred to Children and Household Savings in the Philippines (Orbeta, 2006). The good news is that FIES estimates show that the population of larger families is on the downtrend while that of smaller families is on the rise, consistently in Luzon, Visayas and Mindanao. Table 3. Per capita Income and Expenditure, by Major Island Group and Family size island group Phils. family size per capita income per capita expenditure per capita savings Annualized growth of number of families to 3 44,271 50,050 35,248 40,765 9,023 9, to 6 29,105 31,989 23,678 27,067 5,427 4,

8 island group Luzon Visayas Mindanao family size per capita income per capita expenditure per capita savings Annualized growth of number of families 7 to 10 18,910 21,738 15,721 18,172 3,189 3, > 10 21,835 18,356 18,693 15,521 3,142 2, to 3 56,502 54,225 44,722 44,034 11,780 11, to 6 35,442 33,414 28,652 28,474 6,791 5, to 10 23,675 23,304 19,367 19,368 4,308 4, > 10 27,595 19,884 24,768 16,514 2,828 3, to 3 29,430 32,510 23,858 26,880 5,573 6, to 6 21,786 21,588 18,148 18,544 3,638 3, to 10 13,910 14,511 12,259 12,375 1,651 2, > 10 14,764 11,419 11,782 9,932 2,982 1, to 3 28,116 28,567 22,624 23,242 5,492 5, to 6 19,478 20,236 15,944 16,374 3,534 4, to 10 13,466 13,883 11,236 11,669 2,230 2, > 10 13,901 11,535 9,828 10,239 4,073 1, Source: Basic data come from the 2000 and 2003 FIES of the NSO. Notes: Less than 200 observations fall for each of the group corresponding to more than 10 family members. The annualized population growth is computed based on the FIES.. The variable in the last column corresponds to annualized growth of number of families falling under the categoryies of famsize depicted in the second column of the table. This is approximated by using the total number of families with n members in the FIES. Figures 1.1. and 1.2 are also very informative. For the richest members of Ph ilippine society, i.e., the highest income decile, family size does not significantly matter when it comes to per capita income as indicated by the smaller gap between the two curves in Figure 1.2 of per capita income for families with size less than 7 and for families with size 7 or more. But alas, for the poorest of the poor, the lowest income decile, there is an obvious divide poor families with larger families have lower per capita income than poor families with smaller families! Note that the income differential is also visible if we look at bottom and top 30 percent of the income distribution. From the intermediate measures of poverty such as per capita income, per capita expenditures and per capita savings, we now look at empirical evidences of the direct link between family size and poverty. As shown in Table 4.1, using the three Foster Greer Thorbecke (FGT) poverty measures of poverty incidence, poverty gap and poverty severity, the pattern is again unmistakably clear in the last 10 years: as family size increases, poverty worsens. The poverty gap is defined as the total income shortfall (expressed in proportion to the poverty line) of families with income below the poverty threshold, divided by the total number of families. The poverty severity is the total of the squared income shortfall (expressed in proportion to the poverty line) of families with income below the poverty threshold, divided by the total number of families. In 2003 for example, less than 1 out of 4

9 families with size no greater than 5 is poor, defined to be families with income below the poverty threshold set by NSCB. Among families with size 11 or greater, more than half are poor. Figure 1.1 Difference between Per capita Income of Small and Big Families (bottom income households) (bottom 30%) (bottom 10%) Per capita income Per capita income Income quantiles (from poorest to richest) Income quantiles (from poorest to richest) Source: Basic data come from the 2003 FIES of the NSO. Legend: Blue line represents income quantiles for families with at most 6 members; red line corresponds to income quantiles of families with at least 7 members. Figure 1.2 Difference between Per capita Income of Small and Big Families (top income households) Per capita income (top 30%) (top 10%) Per capita income Income quantiles (from poorest to richest) Income quantiles (from poorest to richest) Source: Basic data come from the 2003 FIES of the NSO. Legend: Blue line represents income quantiles for families with at most 6 members; red line corresponds to income quantiles of families with at least 7 members. More informative is the observation that the reduction in poverty in the country from 28.1% in 1997 to 27.5% in 2000 comes generally from the reduction in poverty among families with sizes no more than five. In fact, families with 6 or more members generally suffered a. deterioration in their poverty status between 1997 and However, it is gladdening to note that the reduction in poverty from 27.5% in 2000 to 24.4% in 2003 was felt across all family size groups, as all three FGT measures showed general improvement. It

10 is also quite obvious that the improvement in poverty reduction between 2000 to 2003 is more definitive than that between 1997 and This direct link between family size and poverty demonstrated at the national level is replicated at the level of the major island groups. For instance, among families with more than 10 household members, in Luzon, 45 percent are poor; in Visayas, 58 percent and in Mindanao, 66 percent, compared to 53 percent for the entire country (Table 4.2). Table 4.1 Foster-Greer-Thorbecke Measures, by Family size Family size Poverty Incidence Phils > Poverty Gap Phils > Poverty Severity Phils > Source: Basic data come from the 1997, 2000 and 2003 FIES of the NSO. Note: The values were computed using apoverty..ado file of Stata. Except for the weights, the sampling design was not used to generate the values, in an attempt to facilitate comparability of the figures.

11 Table 4.2 Poverty Incidence Among Households, by Size of Household Income groups Poverty rate(%) in 2003 not large large Phils Luzon Visayas Mindanao Source: Authors own computations using 2003 FIES data Notes: For this case, large households corresp ond to families with more than 10 household members. Family size is considered not large if the household has at most 10 members. This direct relationship between family size and poverty indicated by the FGT measures is supported by other poverty measures such as poverty vulnerability. From a study conducted by NSCB under the Strengthening Institutional Mechanisms for the Convergence of Poverty Alleviation Efforts, Phase II (SIMCPAE-2), an umbrella project spearheaded by the National Anti-Poverty Commission (NAPC) with funding support from the United Nations Development Programme (UNDP), Table 4.3 shows that estimates of mean vulnerability level and vulnerability incidence increase as family size increases. Note that in 1997, about three out of five families with size greater than 10 were considered vulnerable to poverty, about double the rate among smaller sized-families, i.e. those with size no more than five. Table 4.3 Estimated mean vulnerability and vulnerability incidence, by level of predictors of vulnerability, 1997 Indicator of vulnerability Mean vulnerability Level Vulnerability incidence (%) Estimate Standard CV Estimate Standard CV Error Error Family size 1 <= famsize <= < famsize <= < famsize <= famsize > Source: NSCB Assessment of Vulnerability to Poverty in the Philippines study under the Strengthening Institutional Mechanisms for the Convergence of Poverty Alleviation Efforts, Phase II (SIMCPAE-2), an umbrella project spearheaded by the National Anti-Poverty Commission (NAPC) with funding support from the United Nations Development Programme (UNDP). Note: Vulnerability is defined as the risk that a household will fall into poverty in the future. Conceptually, the vulnerability level of a household at time t is formally defined as the probability that it would find itself income poor at time t + 1. A household with vulnerability level greater than a pre-specified threshold is considered as vulnerable. Consequently, vulnerability incidence is es timated as the ratio of the total number of vulnerable households over the total number of households. From a different angle, one can look at the family size of the poor versus that of the nonpoor. More recent computations (Table 4.4) support what has been noted in previous studies, Virola (1995) that the average family size of poor families at 5.87 in 2003 exceeds

12 that of nonpoor families by about one member. Moreover, among the poor families in 2000, around 39 percent have at least 7 household members compared to only 16 percent among the nonpoor. In 2003, the figures are 21 percent and 6 percent, respectively supporting the observation made earlier that larger family sizes are on the downtrend (Table 5). Table 4.4: Average Family Size, by Inco me Groups Income groups Average Family Size Poor Households Non-poor Households Source: Basic data come from the 2000 and 2003 FIES of the NSO. Notes: Households are grouped into two: poor and non-poor. Poor households correspond to Families whose income fall below the official poverty line computed by NSCB. Table 5: Proportion of households with at least 7 members, by Income Groups Income groups Proportion (in %) Poor Households Non-poor Households Sulu Source: Basic data come from the 2000 and 2003 FIES of the NSO. Notes: Households are grouped into two: poor and non-poor. Poor households correspond to Families whose income fall below the official poverty line computed by NSCB. Table 6.1 Average Family Size of 20 Poorest Municipalities Province Oriental Mindoro Municipality Poverty Incidence (%) Standard Error (%) Average family size HADJI PANGLIMA TAHIL (MARUNGGAS) BULALACAO (SAN PEDRO) Maguindanao MAMASAPANO Lanao del Sur MAROGONG Maguindanao SOUTH UPI Maguindanao TALITAY Maguindanao MATANOG Lanao del Sur PIAGAPO Abra TINEG Quezon SAN FRANCISCO (AURORA) Sulu PATA Lanao del Sur TUBARAN Sulu TONGKIL Lanao del Sur SULTAN PUMALONDONG Lanao del Sur CALANOGAS Lanao del Sur BINIDAYAN

13 Province Municipality Poverty Incidence (%) Standard Error (%) Average family size Zamboanga del Norte SIAYAN Lanao del Sur KAPATAGAN Sulu PANDAMI Quezon SAN ANDRES simple average 6.2 Extending the analysis further, one can examine if the established link between family size and poverty carries to the poorest versus richest provinces and municipalities 3 (Tables 6.1 and 6.2). It is noted that for the 20 poorest municipalities, (NSCB, 2005) the average family size is at least five with 12 of the 20 municipalities having an average size of at least 6 members. In comparison, the 20 least poor municipalities all have an average family size of less than 5 (6.2 vs. 4.6). In addition, in the 20 poorest provinces in 2003, the poor families have an average family size of 5.89 compared to 4.37 for the non poor families. In the 20 least poor provinces, the poor families have an average family size of 5.79 compared to 4.33 for the nonpoor (Table 6.3). family size. Clearly, all indications point to the strong positive correlation between poverty and Table 6.2 Average Family Size of 20 Least Poor Municipalities Province Municipality Poverty Incidence (%) Standard Error (%) Average family size 1st district of NCR BINONDO nd district of NCR SAN JUAN Cavite IMUS th district of NCR CITY OF MAKATI st district of NCR SAMPALOC Cavite NOVELETA nd district of NCR CITY OF PASIG nd district of NCR CITY OF MARIKINA Laguna CITY OF SANTA ROSA Laguna SAN PEDRO st district of NCR ERMITA rd district of NCR CITY OF VALENZUELA Cavite BACOOR th district of NCR CITY OF PARAÑAQUE Benguet BAGUIO CITY Laguna CABUYAO The poverty incidence for municipalities was generated by the NSCB (2005).

14 Province Municipality Poverty Incidence (%) Standard Error (%) Average family size Rizal ANGONO nd district of NCR QUEZON CITY nd district of NCR MANDALUYONG CITY st district of NCR PACO simple average 4.6 Source (Tables 6.1 and 6.2): The poverty rates were computed using a national model of household per capita income (NSCB, 2005). The average family size computations come from the 2000 FIESof the NSO Table 6.3 Average Family Size, 20 Least Poor and 20 Poorest Income groups Average Family Size 20 Least Poor 20 Poorest Poor Households Non-poor Households Source: Basic data come from the 2003 FIES of the NSO. Notes: Households are grouped into two: poor and non-poor. Poor households correspond to families whose income fall below the official poverty line computed by NSCB. Provincial poverty rates are ranked based on the proportion of families (to total number of families) whose income fall belowthe official poverty line. POVERTY AND INEQUALITY Using the Gini coefficient as measure of inequality, Table 1.1 clearly shows the wide disparity in Philippine society with a Gini of 43.97, much higher than in other countries like Indonesia, Republic of Korea and Lao PDR. Analysing this further, one sees in Figures 2.1 and 2.2 that even among families above the poverty line, the Gini 4 is quite high. Very noticeable too is the fact that for both 2000 and 2003, while the extent of inequality decreases for the entire distribution and for the families above the poverty line as the family size increases, among the poor, there is an increasing trend of inequality as family size increases. As income inequality generally characterizes poor societies, this gives support to the idea that population management and poverty intervention programs go together. 4 The Gini coefficient in Table 1.1 is based on consumption; in Figures 2.1 and 2.2, it is based on income.

15 Figure 2.1 Gini Coefficient by family size (2003) Source: Basic data come from the 2003 FIES of the NSO. Note: blue depicts gini coefficient for all income groups red bar depicts gini coefficient for income groups above poverty line green bar depicts gini coefficient for income groups below poverty line Gini coefficient is computed for the per capita income variable derived from the FIES data. Of equal importance is to investigate which components of measure of well-being are primarily driving the observed level of inequality. In this context, one can look at other wellknown entropy measures like Theil s and Atkinson s. Unlike Gini, these entropy measures allow one to decompose inequality into the part that is due to inequality within a given category of the variable of interest and the part that is due to differences between categories of that variable. On the average, at least 75 percent of inequality in a country is due to within-group inequality and the other 25 percent may be attributed to between-group differences (Poverty Manual, All, JH Revision (2005)). Figure 2.2 Gini Coefficient by family size (2000) Source: Basic data come from the 2000 FIES of the NSO. Note: blue depicts gini coefficient for all income groups red bar depicts gini coefficient for income groups above poverty lin green bar depicts gini coefficient for income groups below poverty line Gini coefficient is computed for the per capita income variable derived from the FIES data.

16 Decomposition of Inequality Sensitivities of inequality indices vary according to income differences in different parts of the distribution. Greater a (positive) value in the expression G(a), reflects sensitivity of the index to income differences at the top of the distribution. On the other hand, smaller a (more negative) value captures sensitivity of the index to income differences at the bottom of the distribution. Further, GE(0) is the mean logarithmic deviation, Theil index corresponds to GE(1), and GE(2) is half the square of the coefficient of variation. On the other hand, greater (positive) value of the inequality aversion parameter e in A(e) captures sensitivity to income differences at the bottom of the distribution; smaller value (positive) captures sensitivity of the index to income differences at the top. The Gini coefficient is most sensitive to income differences about the middle (more precisely, the mode). The main motivation in decomposing inequality is to identify the possible sources of disparity. Questions like what drives income inequality are common inquiries by policy makers who are fueled by the renewed interest in the welfare distribution. However, there are a number things to consider when interpreting results of inequality decompositions. For the purpose of this paper, descriptive statistical measures of inequality are provided. For more in-depth analysis, readers are encouraged to read Rethinking Inequality Decomposition, with Evidence from Rural China, by Morduch and Sicular (1998). The following table presents the decomposition of General Entropy (Theil s) and Atkinson s inequality (of per capita income) indices for the following variables: urbanity, region, and family size. Table 7. Decomposition of Inequality, by Urbanity, Region and Family size groups within group inequality between group inequality total inequality GE( -1) GE(0) GE(1) GE(2) GE( -1) GE(0) GE(1) GE(2) GE( -1) GE(0) GE(1) GE(2) Urbanity region family size A(0.5) A(1) A(2) A(0.5) A(1) A(2) A(0.5) A(1) A(2) Urbanity region family size Source: Basic data come from the 2003 FIES of the NSO; computations done using the ineqdeco5.ado syntax of stata

17 Measures of within group inequality refer to the contribution of the disparity within each group to total inequality. For example, if the observations are divided into two groups: A and B, the measure of within group inequality corresponds to the extent of the variation of elements of A among each other and the variation of elements of B among each other, which contribute to the observed inequality among all elements of A and B. On the other hand, measures of between group inequality give us an idea of how the disparity between groups A and B contribute to total inequality. For our purpose, we may be more interested on between group inequality especially since we are concerned on the amount of contribution of the variability of per capita income of different family sizes to total income inequality. Among the three grouping variables considered in Table 7, it appea rs that inequality can be mostly attributed to differences among geographic regions and urbanity, the between group inequality measures are almost the same for these two variables. Notice that if we grouped the household units according to family size, Theil index (GE(1)) suggests that approximately 10 percent ( = 0.05/0.49) of total inequality is due to differences of per capita income between households of different family sizes. In comparison, roughly 14 percent ( = 0.07/0.49) of total inequality may be attributed to regional differences; which is roughly equal to the 16 percent (=0.08/0.49) attributable to urban -rural disparity. If one thinks more deeply, it may appear that income inequality is indeed, a family issue. Choices regarding work, whether and how much to educate children, as well as intergenerational transfer of resources, take place in the family. These factors are often correlated to the size of the family. Therefore, the situation when lower-income-individuals are likely to be found in la rger households with more children, may worsen the income disparity between the rich and the poor. This is not only because income must be shared between household members. Rather, decisions made inside the family such as the ones mentioned above, affect the income distribution and hence, income inequality. Let us now see if the premise that the problem of poverty is closely linked with family sizes is supported by non-income indicators. One indicator that has been established in many studies to be a strong correlate of poverty is whether the household head has finished secondary education. Depicted in Figure 3 is an indicator of access to education by size of family. Noticeably, members from large

18 families seem to be less likely to reach college than their counterparts who belong to smaller sized families. Figure 3. Proportion of households members 25 years and over who did not finish high school, by family size Source: Basic data come from the 2003 FIES of the NSO. Note(s): This is computed as the ratio of the number of household member s of age 25 years and over who did not finish high school by the number of household members who are of age 25 years and over. Figures 4.1 to 4.3 also reveal that for both the 2000 and 2003 FIES, the per capita expenditure on education, medical needs and recreation generally goes down with increases in family size. In fact, for 2003, the estimated per capita expenditure on education of members from families of size 10 or more is less than half the per capita education expenditure of smaller families. Figure 4.1 Per capita Expenditure on Education Source: Basic data come from the 2000 and 2003 FIES of the NSO. On the other hand, Table 8 shows that from 1997 to 2003, at the national level we experienced a decreasing proportion of family expenditures going to food, which from Engel s theory indicates improving economic situation. But while this observation holds true across all family sizes, Table 8 clearly indicates that in accordance with Engel s law, the economic status of the larger families is lower compared to smaller families. Moreover, with

19 more than half of the expenditures of larger families going to food, little is left for other basic necessities like education and health, putting greater stress on the socio-economic future of the family. Figure 4.2: Per capita Expenditure on Medical Needs Figure 4.3:Per capita Expenditure on Recreation Source: Basic data for Figures 4.2 and 4.3 come from the 2000 and 2003 FIES of the NSO. Table 8:Ratio of Food Expenditure to Total Expenditure, by Family size (in %) Family size

20 Family size > Source: Basic data come from the 1997, 2000 and 2003 FIES of the NSO. In summary, this section examined the relationship between population, family size in particular and poverty. The FIES suggests overwhelming empirical evidence of the strong positive correlation between poverty and family size. III. What lies ahead: Prospects for the Demographic Dividend The legitimacy of the problem on population dynamics has been acknowledged by well-known scholars. Thomas Malthus, a renowned demographer and political economist, is best known for his influential insights on population growth. He argued that it will be inevitable that population outgrow food supply because of the vices of mankind. He favoured moral restraint as a check on population growth; but limiting this restraint only to the working and poor classes Although Malthus views had been criticized as unreasonably catastrophic, his fundamental principle that living standards decrease when there are too many mouths to feed, stands on solid grounds particularly in today s context (Deming, 2004). Now that data suggest a cannot-be-ignored relationship between population size (or family size) and poverty and inequality, a good question to ask is, is population growth downright bad or can we also get some good apples from it? We will attempt to answer this question by looking at what economic theory and other countries experiences say. Let us try to use a popular statistical tool: regression, to speculate on an answer. Consider the following regression model:

21 Table 9 Regression of Income on Family size and Square of Family size # of obs 96 Source SS df MS F stat Model p-value 0 Residual adj R Total Root MSE lnincp Coef. Std. Err. t p-value [ 95% Conf. Interval] famsize famsize_squared intercept Source: Basic data come from the 2003 FIES for NCR of the NSO. Observations are restricted to households whose per capita income is below poverty line and where the household head is at most 40 years of age. The variables famsize and famsize_squared refer to family size and square of family size, respectively. Notice that in modeling the log transformed household per capita income, the sign of the coefficient for family size is positive while that of the square of family size is negative. Note that the data represent young families, at the bottom of the income distribution residing in NCR. Here, a young family is loosely defined as a family whose household head is at most 40 years of age. Results depicted in Table 9 5 may be interpreted such that, holding all other things equal, a family size of 12 or more may yield negative impact on income, particularly for families below poverty line. Many countries continuously explore for approaches to solve the population problem. There are instances during this search when one stumbles on what appears to be a legitimate solution at first glance, but turns out to be a myth when scrutinized carefully. Later we will enumerate some of these more popular myths. As presented earlier, we had speculated that population growth is not bad at all times. There are some considerations that have to be taken into account. Again, consider the regression equation in Table 9. There, the observations were restricted to households in NCR whose household head is relatively young and whose family incomes fall below poverty line (For households whose head is less than 40 years of age, we would assume that most of the family members excluding the parents are children). As can be observed from Table 9, the coefficient for family size is positive while that of square of family size is negative. Arguably, this may be attributed to the possibility that up to a certain extent, an additional child in the family may mean additional source of family income (particularly, once the child 5 Note however the low value of R 2.

22 reaches working age or if we take child labor into account since we are talking of poor families). However, too many dependent children may exceed the benefits of having an additional member and hence, will have a negative impact on per capita income, especially for poor families. More compactly, Table 9 6 seems to dismiss the notion of some poor parents that additional children means additional sources of income for the family. To achieve a better grasp and appreciation of these ideas, we now review the concept of demographic transition and dividend. In economics, demographic dividend refers to the rise of economic growth due to a rising share of working age people in the population (Bloom, et. al, 1998). This growth is usually observed when the fertility and youth dependency rates in a particular country decline. Because of higher share of working population, accumulation of capital becomes more rapid and spending on dependents is reduced. This stage is usually referred as demographic transition. During this period, increase in output per capita can be observed. It is also accompanied by more economic opportunities. However, it should be stressed that the magnitude of demographic dividend is anchored on the economy s ability to absorb and productively employ the large pool of educated workers, and not solely on the observed demographic changes. For countries like Japan, Singapore, Ireland, and other tiger economies in Asia and Europe, economic growth has been sustained, reaping the demographic dividends. However, when this relatively large working age population cohort grows older with no fewer people to replace them, aging sets in. Consequently, dependency rate increases and so the demographic dividend decreases. This is a problem that many developed countries now face. The outcomes of the steep declines in fertility in the past are being felt at present and these countries are now providing incentives to encourage young couples to bear children. This takes these countries towards the latter stage of the demographic cycle where economic gains may be hampered by fiscal challenges. Figure 5 presents the trend in the ratio of working age to dependent population, by region. Looking at the trend, it seems that in general, Northern America and European countries are on the last stage of the demographic cycle, at present. On the other hand, Asia is on the expansion stage. If the trend continues, Europe and Northern America are expected to be on the contraction stage after The ratio of working age to dependent population in Asia is 6 Readers are cautioned in interpreting the figures. The regression exercise was used only to introduce the subject of the discussion. It is interesting to note that when the young family restriction is not used, the signs of the coefficient interchange, i.e., the coefficient of family size becomes negative while that of the square of family size becomes positive.

23 expected to reach its peak between 2010 and Further, it is anticipated that Africa will peak after Figure 5 Ratio of Working Age to Dependent Population Africa Asia Europe Northern America Latin America and the Carribean Source: United Nations Population Division The 2006 Revision Population Database website In the Philippine context, some economists and experts postulate that the country is in the early stage of a demographic transition and one of the biggest potential gainers among Southeast Asian countries (Bloom and Williamson, 1998). After several years of efforts among Filipino parents to educate their children, it is now time to collect the payday dividends, according to some experts. Over the years, the Philippine population has shifted structurally with a growing share of the working age from 15 to 64 population. According to population projections of the National Statistics Office (NSO, 2006), in 2005, the working age population is projected to constitute 60.38% of the population. This is estimated to rise to 62.54% by 2010, 63.40% by 2015, 64.16% by 2020, and 66.87% by 2040, increasing the ratio of working age to dependent population. This demographic structure of the Philippines is quite different from what many developed countries are now experiencing - a decline in the ratio of working population to dependents because of the increasing number of the elderly. To economists and demographers, such a demographic structure as ours can generate opportunities for economic growth. Unfortunately, reaping the demographic dividend is not automatic according to Harvard University Professor David Bloom. According to him (Bloom, et. al, 2000), the prerequisites include good government institutions, efficiently functioning markets, and considerable investment on the children s education to boost their skills. Otherwise, it will be debacles and not dividends waiting to be reaped. For the purpose of this paper, let us delve only on the last requirement as the first two do not directly involve the topic of our paper.

24 The occurrence of demographic dividend permits growth in per capita output because of the following main reasons (Bloom and Williamson, 1998): (a) the existence of 'pure' age-structure impact on total GDP due to a simple 'factor availability' or 'growth accounting' effect in that having a rising share of the total population in the working-age group increases the ratio of producers to consumers; and (b) the presence of 'behavioural effects' which connote the growing body of prime-age workers in the work force which enhances overall productivity due to the well-known prime age worker productivity effect. It must be emphasized that the problem on population growth can not be resolved by simply meeting unmet need of contraceptive services, or focusing on economic growth. While it is true that over the last three decades, increasing access to contraceptive services has helped reduce fertility rates, it has to be acknowledged that there are socio-cultural dimensions relating to decisions on whether to have many children, that can not be addressed by access to contraceptives alone. This however, should not be interpreted as suggesting that the level of effort in providing contraceptive services be reduced. Indeed, high quality, low cost reproductive health services are necessary components of fertility planning. It is accepted in most societies that reproductive decisions should remain to be a conscious choice of a couple. For this to be realized, proper education is necessary. On the other hand, the myth that the economic development will lead automatically to a demographic transition, seems to contradict the experiences of some countries. In a study of Abernethy (1993), "Land redistribution in Turkey promoted a doubling in family size (to six children) among formerly landless peasants. In the United States and much of Western Europe, a baby boom coincided with the broad-based prosperity of the 1950s. More water wells for the pastoralists of the African Sahel promoted larger herd size, earlier marriage and much higher fertility. The introduction of the pota to into Ireland in about 1745 increased agricultural productivity and caused a baby boom." Other examples by Abernethy include the 17 percent drop in Sudan's fertility rate during the late 1980s at a time of extreme deterioration of the economy and a similar correlation of falling fertility rates and declining economic conditions in Brazil in the 1980s. What this argument is driving at is that economic development alone, without other measures that affect family size desires or the ability to achieve such desires, can not solve the population problem. We can further illustrate this argument through the Philippine experience. According to the NSO (2004), The 2003 National Demographic and Health Survey shows that Filipino women still bear more children than they desire For the last decade, the gap between wanted and actual fertility rates has remained stable at one birth. This may mean that there is more to be done address the population problem in the country.