Working Paper in Economics and Business Volume IV No. 03/2014

Analysis of Poverty Determinant in West Province

Sartika Djamaluddin

September 2014

Institute of Economics and Social Research Faculty of Economics, Universitas Working Paper in Economics and Business Chief Editor: Hera Susanti Editors: Djoni Hartono, Beta Y. Gitaharie, Femmy Roeslan, Riatu M. Qibthiyyah Setting: Rus’an Nasrudin

Copyright c 2014, Department of Economics Department of Economics Building 2nd Floor , Indonesia 16424 Telp. +62-21-78886252 Email: [email protected] Web: http://econ.fe.ui.ac.id/workingpage Contents

Contents 3

List of Tables 4

List of Figures 5

1 Introduction 1

2 Research Framework3

3 Illustration of Impoverished Households in West Java4

4 Data & Research Methods8 4.1 Data...... 8 4.2 Model...... 8

5 Outcome of Research 10 5.1 Demographic Variables...... 10 5.2 Economic Variables...... 10 5.3 Social Variables...... 12 5.4 Regional/Area Variables...... 12

6 Characteristics of Inter-Area Poverty 13

7 Conclusion 14

8 References 14 List of Tables

1 Gender of Head of Impoverished Households in West Java...... 5 2 Age of Head of Impoverished Households in West Java...... 5 3 Size of Impoverished Households...... 6 4 Number of Impoverished Households by Home Ownership...... 6 5 Types of Assets Owned by Impoverished Households...... 7 6 Educational Participation of Impoverished Households...... 7 7 Symptoms/Medical Conditions Reported in the Past Month...... 8 8 Employed Impoverished Households...... 8 9 Insurance Financing/Healthcare Insurance of Impoverished Households...... 9 10 Result of Estimation of Logistic Regression Model at Province Level...... 11 11 Result of Estimation of Logistic Regression Model at Level...... 15 List of Figures

1 Poverty Level in Regencies and Cities in West Java...... 3 Analysis of Poverty Determinant in West Java Province

Sartika Djamaluddina,∗

aDepartment of Economics

Abstract Comprehensive profiling of impoverished households plays a fundamental role in enabling the government to compose quintessential and antipoverty policies that effectively lower poverty on a significant level. This study analyzes household assets and poverty reduction policies as one of the determinant factors of poverty. This research is based on data cumulated from a national socio-economic household survey (susenas) in 2010 as well as logistic regression model to identify factors proximately associated with poverty level in Province, regency and City in West Java. The number of observations as much as 20,541 households. Upon comprehensively evaluating samples, the outcome of the research shows that West Java is facing complex issues related to poverty. All determinant factors including demographic, economy, social and government policies are identified as significantly impact on poverty rate in the region. At all province, city and regency level, size of household member and assets variables are found to be the factors that consistently and significantly determining poverty rate. At the province level, the high probability of poverty is triggered among other by the large size of households’ members, family head is married and/or employed in agriculture sector or work as laborers and having low education level as well as living at house with the ground floor/bamboo. Results of regression analyses conducted in respective sample cities/regencies nevertheless illustrates that the level of influence on poverty level vary accordingly. While city and regency are facing more complex poverty issues to address, cause of poverty in and regency are identified as exactly the same.

JEL Classifications: I32, D1, C25 Keywords: Poverty Determinant, Household, Logistic Regression Model, West Java

1. Introduction each year. Moreover, up to 4.7 million poverty- stricken people were reportedly living in West West Java is one among several provinces in Java in 2010, the third-highest in the nation af- Java Island with a relatively high poverty rate. ter East Java and where around Although the province’s per capita-poverty 5.5 million and 5.3 million individuals live in rate saw a decline from 2001 to 2011, around poverty respectively. 12.02% of residents continue to live in poverty Data cumulated from the 2010 national so- cio economic survey also indicates that the dis- ∗ Selasar Building, 2nd floor, Lecture Room. tribution of impoverished households varies by Faculty of Economic and Business, Universitas In- city or regency. Figure 1. shows that among donesia, UI Campus, Depok, West Java 16424. Ph.:+6281388377484, +622129504211. Email: cities/regencies with an impoverished house- [email protected]. hold rate of above 10% are Tasikmalaya, Indra- 1 Djamaluddin, S./Analysis of Poverty Determinant... 2 mayu, Kuningan, Cianjur, and Ban- lowering wages. Meanwhile, government poli- dung city. Tasikmalaya has the highest poverty cies also play a significant role in countering rate. Simultaneously, among cities/regencies in poverty – apart from assistance programs, so- West Java with the lowest impoverished house- cial protection programs for families and com- hold rates are Depok, city, city munity empowerment programs, the govern- and at 1.8%, 3.1%, 4.1% and ment also extends assistance to empower mi- 4.4% respectively. cro, small and medium enterprises (UMKM) which often encounters difficulty in accessing fi- The characteristics of households and indi- nancial services due to failure in meeting bank- viduals residing in any one household are also ing terms and conditions. The government is an determinants of poverty level, as backed by the important source in providing financial assis- fact that low education and healthcare house- tance and opening financial services to fortify holds are presented with limited job oppor- the economy and expedite creation of new jobs tunities. Because such jobs are typically low- in various fields as ultimate means of counter- paying, individuals in the household will en- ing poverty. counter challenges in achieving an adequate standard of living. Meanwhile, a large house- Facilities such as electrical infrastructures, hold population will also exacerbate burdens roads, bridges, water pipes, land distribution, borne by the head of household. In such house- access to public facilities (e.g. distance to holds, higher income is required to uphold or schools and hospitals), social structure and so- sustain an adequate standard of living. More- cial capital are also determinants of poverty over, a household with no or limited-valued as- level. Limited access to electrical networks will sets do not have sustenance to depend on as consequently interfere with activities and im- supplementary income to attain an adequate pact education, healthcare and social aspects standard of living. An individual’s incapability in an impoverished household. Poor road condi- to meet the adequate standard of living distin- tions will also impact distribution of goods and guishes that particular individual or household economic activities, both of which will in turn as impoverished. Poverty is also gender-based – affect productivity in an impoverished house- research conducted by the International Labor hold. Organization (ILO) 2004, denotes that women An array of studies has also been con- are more likely to fall into poverty as com- ducted to expose determinants of poverty. pared to their male counterparts. Among rea- Geda Mwabu and Kimenyi (2001) investigated sons for this is that women are often employed determinants of poverty in Kenya and found as farmers in low-output agricultural fields, re- poverty to be substantially inclined to educa- ceive lower wages compared to men, do not tion, household population and agricultural ac- have guaranteed rights over land that they tivity. The study was carried out by assess- own, are lowly educated, have limited access ing household data in 1994, binomial regres- to healthcare and bear the burden of overseeing sion model and logit polychotomous. Similarly, the daily affairs of their respective households. Mok, Gan and Sanyal (2007) investigated de- terminants of poverty in Malaysia and iden- Conditions surrounding an area are also de- tified human capital as a significant factor in terminants of poverty. Isolated areas are gen- reducing the probability of poverty. Simultane- erally more prone to poverty as households in ously, the study also found migrant laborers to such areas do not possess the required infras- be likelier to fall into poverty and that house- tructures to effectively propel economic activ- hold size, race and religion are also important ity, consequently limiting job opportunities and determinants of poverty in the country. The Djamaluddin, S./Analysis of Poverty Determinant... 3 Figure 1: Poverty Level in Regencies and Cities in West Java

study involves 2,403 households in the city and the number of dependents in a household as was carried out from 2004 to 2005. the primary poverty determinant in Bonang, Demak. Haughton & Khandker (2009) attributes The researches above are based on data, poverty to four primary aspects – character- specifically information on the characteristics istics of an area, society, household and indi- of households, individuals, area and govern- vidual. Area-level characteristics such as isola- ment policies, cumulated from the 2010 na- tion, natural resources, weather, natural disas- tional socio economic survey involving residen- ters, local governance and income are determi- tial households in West Java province. Unlike nants of poverty. Society-level characteristics previous studies, the study takes into account that determine poverty are broken down into government programs such business loan pro- infrastructure (roads, water and electricity), gram as one among several factors which are services (healthcare and education), relations significantly correlated to poverty level. Logis- (capital) and social aspects. Meanwhile, resi- tic regression model to analyze factors that are dential and individual-level characteristics such strongly correlated to and are strong factors of as household size, household dependence ratio, poverty is also utilized. gender of the households head, assets, type and status of employment are also determinants of poverty. Moreover, the household size and 2. Research Framework characteristics of family members also differ Analyses conducted to identify poverty de- between impoverished and non-impoverished terminants neither serve to analyze nor arbi- households. Gibson (1999) in Haughton & trate the cause(s) of poverty. Variables dis- Khandker (2009) explained that the 1934– tinguishged based on economic theory are ac- 1994 Cambodian Socio-Economic Survey indi- knowledged as factors that affect poverty level. cates that impoverished households are gener- Such factors, or factors identified as poverty ally made up of large families with an aver- determinants, are determined based on eco- age of 6–7 family members in a household. On nomic theory. Empirical modelling consecu- the opposite end of the spectrum, the richest tively serves to test out theory. In this re- households consist of around 4–5 members in a search, household, individual and area-level household. characteristics are evaluated as determinants Afandi (2011) from Pariaman, Padang found of poverty level, the two latter of which covers households with more than 4 members, with democraphic, economic as well as social factors. less than 8 meters2 of land per capita, with Area-level characteristics comprise of local fac- a head of household below 35 years of age, tors and public business loaning policies. which have taken up a business loan and/or Data is divided into two categories which are with a head of household employed in sec- data cumulated as a whole from West Java and tors other than the agricultural and industrial data of respective cities and regencies in the sector more likely to fall into poverty. Hay- province. Regression analysis as well as data ati (2012) exposed geographical location and from West Java serves the purpose of identi- an addition of household member as high-risk fying the prevalent cause(s) of poverty in the poverty factors in cities and regencies in Ban- province. An analysis using both sets of data ten. Sari (2014) found assets, employment and will also be conducted to exclusively identify Djamaluddin, S./Analysis of Poverty Determinant... 4 the cause(s) of poverty in a specific city or re- poverty level in other areas span between 5% gency in West Java. A second analysis will also to below 11%. be carried out as the poverty level in respec- Contratry to beliefs that poverty is ex- tive cities and regencies in West Java will likely clusively confined to villages, the 2010 na- vary. While the first analysis will more likely tional socio-economic survey reveals that im- accomodate the interests of the central govern- poverished households are almost evenly dis- nment, the second analysis will more likely ben- tributed among villages and cities in West Java efit the local government. province, where up to 505,438 households in Poverty criteria refers to the poverty line es- villages and 439,735 households in cities are tablished by the local Central Statistics Bu- impoverished. The majority of such households reau (BPS) in any one city or regency. House- are overseen by men. holds with a per capita monthly income be- low the poverty line are classified as impov- Based on the classification of household lo- erished. The poverty line itself is determined cation, the majority of heads of impoverished based on minimum household expense which households live in villages. Moreover, impov- esrimates the approximate earnings of a house- erished households in villages and cities also hold. The hypothesis of this research is that the do not differ on a significant scale. Meanwhile, probabibillity of poverty are relatively high in various facilities are available in cities from households retaining the following characteris- kindergartens, junior high schools, public high tics; households with a female head, with more schools, markets, shops, movie theaters, hos- than 3 married family members, with mem- pitals, hotels, billiard clubs, nightclubs, mas- bers working in the agricultural sector, with sage parlors, salons as well as electricity and no assets and/or with members working as an telephones. While the majority of impover- employee and not a laborer. Households with ished households in regencies reside in villages social-level characteristics such as low educa- equipped with minimal public facilities, impov- tion, illness spanning to over 6 months and erished households in cities are easily able to households with earthen or bamboo floors are access public facilities as compared to impov- also at high-risk of falling into poverty. Syn- erished village households. The sole exception chronously, regional/area-level characteristics is where the majority of im- such as living in a city and receiving business poverished households are located in the city. loans are also low-risk poverty factors. Over 72% of household heads are in the pro- 3. Illustration of Impoverished House- ductive age (between 20–55 years old). The ma- holds in West Java jority of such heads reside at villages situated in cities as well as regencies. However, the ma- The 2010 national socio-economic survey jority of heads of Impoverished households in states that up to 945,172 or 9% of house- high-poverty level areas such as Tasikmalaya holds in 17 provinces and 9 cities in West Java city, regency, live under the poverty line. With poverty lev- and are over 55 years of els of up to 12%, Tasikmalaya, Indramayu, age. Simultaneously, the majority of household Majalengka and Kuningan possess the high- heads in low-poverty areas such as Bekasi re- est impoverished household percentage in West gency, Bekasi city, Bandung, and De- Java. Meanwhile, among cities/regencies with pok are in the productive age. Up to 259,887 the lowest impoverished household rates are impoverished households with heads whom are Depok, Bandung city, Bekasi city and Bekasi not in the productive age are located in cities regency at below 5% respectively. The average and villages throughout West Java, 54% and Djamaluddin, S./Analysis of Poverty Determinant... 5 Table 1: Gender of Head of Impoverished Households in West Java

Gender Village City Total Female 50,37 41,485 91,855 Male 455,068 398,25 853,318 Total 505,438 439,735 945,173 Source: 2010 national socio-economic survey

Table 2: Age of Head of Impoverished Households in West Java

Age Village City Total < 20 715 715 20–< 35 79,699 79,271 158,97 35–< 45 153,365 132,351 285,715 45–< 55 131,358 108,654 240,013 > 55 141,015 118,743 259,759 Total 505,438 439,734 945,172

46% of which are located in cillages and cities At 62.4%, a large percentage of impoverished respectively. households do not own assets. Merely 355,142 Impoverished households disbursed through- households own assets that are mainly com- out cities and villages tend to consist of many prised of refrigerators, gas cylinders amount- members. Around 60% of impoverished house- ing to 12 kilograms or more and/or boats. holds comprise of 3–5 members. Simultane- Impoverished households frequently rely on ously, an approximate 29% and 4.2% of im- assistance from others to purchase food as poverished households comprise of 6–8 and 9 well as non-food items. While 15% of impov- or more members respectively. Household size erished households borrow money from rela- reflects the extent of economic burden borne tives, 22.67%, 12.7% and 4.4% borrow from by the head of households. The majority of im- neighbors, use up their savings and borrow poverished households in clusters 1, 2 and 3 from creditors respectively. The remaining per- comprise of 3–5 members, with the exception centage acquires loans from banks, coopera- of Bekasi and Depok where the majority of im- tive banks or pawn loans. 382 households also poverished households comprise of 6–8 mem- rely on other sources to make ends meet. Im- bers. poverished households which depend on other Data from the 2010 national socio-economic sources (outside jobs) are mainly located in Bo- survey also reveals that 82.6% of impover- gor, Sukabuni, Cianjur, Bandung and Tasik- ished households own their own homes and that malaya. the majority of homeowners reside in villages. The educational participation rates of im- Meanwhile, 72.6% of the remaining 163,809 im- poverished households is relatively low. Data poverished households that do not own their from the 2010 national socio-economic survey own homes reside in cities, a fact which can be indicates that 9% of household heads have no clearly observed in many cities and regencies schooling experience whatsoever. The remain- in West Java. In Bandung, a large percentage ing 91% of household heads are no longer in of impoverished households are not homeown- school. 9.8% of impoverished households are ers. Meanwhile in Bekasi, there are roughly the unable to read or write the latin alphabet. same number of impoverished households that Moreover, impoverished households seldom in- own their own homes and do not own their own teract with technology. The majority of impov- homes respectively. erished household members are also unable to Djamaluddin, S./Analysis of Poverty Determinant... 6 Table 3: Size of Impoverished Households

Household Members Village City Total ≤ 2 46,23 20,532 66,762 3–5 324,819 238,152 562,971 6–8 124,022 151,539 275,561 9–12 9,921 25,764 35,685 >12 445 3,748 4,193 Total 505,438 439,734 945,172

Table 4: Number of Impoverished Households by Home Ownership

Home Ownership Status Village City Total Privately Owned 460,538 320,825 781,363 Others 44,900.2 118,909 163,809 Total 505,438 439,734 945,172 access the intenet. Merely 0.48% of impover- Medical conditions and sickness are found to ished household members have accessed the in- impact the daily activities in a household. The ternet in the past three months. work and school of 55.3% of household heads 93% of heads of impoverished households were affected by medical conditions and sick- who have schooling experience are also lowly ness. Despite this, not all households choose educated. The highest education level that was to go to the doctor and merely 88.5% of pursued or is currently being pursued by is households sought medical attention to recover. basic elementary or junior high school educa- Meanwhile, 44.67% of households claiming that tion or equivalent. A substantial percentage of illness do not impact their activities choose to such individuals deliberately chose to end their disregard medical treatment. This is likely due schooling. Around 26% of impoverished house- to economic demands obliging households to holds also do not hold elementary school certi- continue working. fication. 833,706 impoverished households are cur- Lowly educated households are disbursed rently employed in four fields which are agricul- evenly throughout cities and regencies. In De- ture, mining, trade and services. Agriculture- pok, the percentage of lowly and highly edu- related jobs comprise of and vegetable cated households are essentially equivalent. A farming as well as jobs in the horticulture, relatively large number of impoverished house- plantation, fishing, cattle farming, forestry and holds in Depok therefore holds more than a ba- other plantation-associated sectors. sic educational certificate. Meanwhile, mining-related jobs revolve As many as 314,192 or 33.2% of heads of around mining and excavation, the processing impoverished households were sick in the past industry as well as the electricity and gas sec- month. The majority of symptoms and med- tor. In the trade sector, jobs are correlated to ical conditions reported, including accidents, construction/development, trade and the hos- measles, ear discharge and jaundice, are consid- pitality and food industry. Services-oriented ered severe. Among other frequently reported jobs include transportation and warehousing, ailments are cough, cold or flu (influenza) and information and construction, insurance and chronic headaches. Despite this, merely 43.7% financing as well as healthcare and educational of impoverished households seek medical atten- services. The trade and agricultural sector tion to recover. are the primary recruiters of members of Djamaluddin, S./Analysis of Poverty Determinant... 7 Table 5: Types of Assets Owned by Impoverished Households

Type of Asset Count Bicycle 154,046 Motorcycle 122,049 Boat 2,242.37 Motor Boat 462 Refrigerator 58,122.4 Gas Cylinder (≥ 12kg) 18.220,20 Others 590,03 Total 945,172

Table 6: Educational Participation of Impoverished Households

Educational Participation Village City Total No/Have Yet to Gain Schooling Experience 54,117 32.749 86,865 Active in School - - - No Longer in School 451,321 406,986 858,307 Total 505,438 439,734 945,172 impoverished households. erished households are registered members of Up to 11.8% of impoverished households are the healthcare insurance for the poor (JPK also unemployed. The daily activities of mem- MM), healthcare card (kartu sehat), health- bers of such households include managing daily care insurance for poor families (JPK Gakin), affairs in the household, exercising, attending poverty card (kartu miskin) and social health courses and picnics and social activities such insurance program. 1.8%, 0.65% and 0.58% as participating in organizations and doing so- of impoverished households receive health- cial work. The percentage of unemployed and care insurance from jamsostek (social secu- impoverished households is also higher in high- rity program), other healthcare insurance pro- poverty level cities such as Majalengka regency grams and healthcare insurance for civil ser- as well as Sukabumi and Cimahi. Meanwhile, vants/veterans/pensioners respectively. the majority of impoverished households in re- gencies work for the agricultural sector. On the other end of the spectrum, impoverished households in cities tend to work for the ser- vices, construction and trade sector as con- Compared to the allocation of business loans cluded based on observations made at , to non-poor households, loan provided to the Sukabumi, Cirebon and Bekasi. The employ- impoverished household is still very small. Of ment status of employed impoverished house- a total 996,877 households receiving impover- holds is broken down into freelancers (34.5%), ished business credit, 96.3% of them are classi- self-employed (22.5%), workers/employee/staff fied as non-poor households and the remaining (20.5%) and employers with unpaid workers 3.67% is considered as improverished house- (20%). holds. , Tasikmalaya, Sukabumi and With regard to paying off medical bills, only Garut are identified as regencies that distribute a small percentage of impoverished households more loans for impoverished households. At the have access to insurance. Over 50% of impov- city level, most business loans recipients are erished households have yet to receive insur- identified living in Tasikmalaya, Bandung and ance coverage. Furthermore, 35.7% of impov- Banjar. Djamaluddin, S./Analysis of Poverty Determinant... 8 Table 7: Symptoms/Medical Conditions Reported in the Past Month

Symptom/Medical Condition Village City Total Fever 3,562 4,68 8,242 Cough 15,762 14,999 30,761 Cold/Flu 32,969 36,006 68,975 Asthma/Rapid Breathing/Breathing Difficulty 7,623 7,549 15,172 Diarrhea 5,244 3,228 8,472 Chronic Headache 21,952 17,78 39,732 Toothache 6,936 9,364 16,3 Others* 68,912 57,627 126,539 Total 162,958 151,233 314,192

Table 8: Employed Impoverished Households

Employment Field Count Agriculture 381,509 Mining 93,111 Trade 203,878 Services 155,208 Total 833,706

4. Data & Research Methods financial security of the highest earner in the household. Social aspects comprise of educa- 4.1. Data tion level, healthcare and area of residence. This study takes into account data from the Area-level characteristics are observable by the 2010 national socio-economic survey involving conditions of any one specific area (village or residential households in West Java. The sur- city) and government policies, the latter of vey published by the Central Statistics Agency which pinpoints to anti-poverty policies in the (BPS) comprises of 20,541 households span- form of disbursing business loans. The business ning across 10 cities and 16 regencies. The loan program comprises of the national inde- survey also comprises of individual-level and pendent society empowerment program as well household-level data. Household-level data sets as other government sponsored programs, peo- in the survey were constructed by compiling ple’s business loan program and banking pro- individual-level data sets. This was done so grams outside cooperation banking programs as a number of household-associated informa- and the people’s business loan program (KUR). tion is encompassed in individual data. New data sets were subsequently created by compil- ing information on the characteristics of 20,541 4.2. Model households in West Java. Based on accumu- lated data, 945 households were found to be The type of empirical model used in this re- impoverished. 172 observations were conducted search is the logistic regression model where the in 64% of all households. dependent variable is a binary variable either Household characteristics comprise of demo- valued ’1’ if the household is impoverished or graphics as well as economic and social as- ’0’ for other categories. The independent vari- pects. Demographic aspects comprise of gen- able consists of a binary and continuous vari- th der type, household size and marital status. able, where Pi is the i household probability Economic aspects are represented by the em- existing below the poverty line. Pi is a bernouli ployment field, employment status, assets and variable and its distribution is correlated to in- Djamaluddin, S./Analysis of Poverty Determinant... 9 Table 9: Insurance Financing/Healthcare Insurance of Impoverished Households

Insurance Financing/Healthcare Insurance Count a. Healthcare Insurance for Civil Servants/Veterans/Pensioners 5,521.56 b. Social Security Program - Healthcare Insurance (Jamsostek) 17,031.9 c. Private Health Insurance 2,970.05 d. Company Reimbursement 3,844.9 e. Healthcare Insurance for the Poor (JPK MM) / Healthcare Card (Kartu Sehat) / 337,316 Healthcare Insurance for Poor Families (JPK Gakin) / Poverty Card (Kartu Miskin / Social Healthcare Insurance (Jamkesmas) f. Healthcare Funding 3,722.47 g. Other Healthcare Insurance Programs 6,175.37 h. Others 568,59 Total 945,172 dependent vector X. Therefore: Economic variables Agriculture = ’1’ if the household works for eα+βX the horticulture, plantation, fishing, cattle Pi(X) = α+βX (1) 1 + e farming, forestry or plantation sector, ’0’ for Similarity (1) is a cumulative logistic distri- others bution function. The function above (1) is non- Laborer/worker = ’1’ if the household mem- linear but can be lineated by re-writing the ber(s) is a laborer/worker, ’0’ for others function: Asset = ’1’ if the household owns a bicycle, P i = eα+βX (2) motorcycle, boat, motorboat, refrigerator or a (1 − Pi) gas cylinder with a capacity of 12 kilograms or The odd ratio, or Pi , is obtained from impov- above, ’0’ for others 1−Pi erished households and explains the probability ratio of impoverished versus non-impoverished Social variables households. The similarity (2) can be re- Basic education = ’1’ if the household either written in its natural logarithm form as: holds elementary school certification or do not Pi hold an educational certificate, ’0’ for others ln = α + ΣkβkXki (3) 1 − Pi Healthcare = ’1’ if the household has sought medical attention in the past six months, ’0’ This form (3) is also referred to as the logit for others model. ln Pi is a natural logarithm acquired (1−Pi) Home flooring = ’1’ if the household has either from the odd ratio of impoverished households earthen or bamboo flooring, ’0’ for others below the poverty line. The similarity (3) is estimated using the maximum likelihood method. Regional/Area variables Area of residence = ’1’ for city, ’0’ for others Demographic variables Loan programs = ’1’ if the household accepts Male = ’1’ if the household head is business loans from the people’s business a man, ’0’ for others (KUR) program and/or the national indepen- Household size = Number of members in a dent society empowerment program and/or household other banking programs exclusive of the Marital status = ’1’ if the household head is KUR program and other programs held by married, ’0’ for others cooperation banks, ’0’ for others Djamaluddin, S./Analysis of Poverty Determinant... 10

5. Outcome of Research In conclusion, the probability of poverty in- creases as household size expands. This is be- The research’s outcome is that the poverty cause households are required to meet higher level in West Java is significantly influenced basic living requirements with every addition by demographic, economic and social factors to the household. Furthermore, the probability as well as area-level classification and govern- of poverty in an average household (compris- ment policies. The government’s ’rice for the ing of 5 members) is relatively high at 62.2%. poor’ program is also presented as the most ef- Observations conducted in two regencies also fective approach in lowering poverty as the pro- came up with the same findings on the rela- gram’s benefits are directly enjoyed by impov- tionship between household size and probabil- erished households. Other factors which signifi- ity of poverty. In cities and regencies, house- cantly influence poverty level is household size, hold size significantly impacts poverty with a marital status, employment in the agricultural probability of poverty of over 60%. Meanwhile, sector, low education and household-level char- marital status is not a significant determinant acteristics. of poverty in all regencies with the exception of several regency and cities (Regency of Suk- 5.1. Demographic Variables abumi and Ciamis and City of Cimahi, Bogor, Provincial-level data identified men as the Tasikmalaya and Cirebon). primary breadwinners in impoverished house- The conclusion above is in line with a dis- holds. Data from the national socio-economic covery exposing the relatively high probability survey indicates that 90% of impoverished of poverty in married households (63.4%) as households are headed by men with a median marriage is associated with the addition of a salary of Rp196,874. Meanwhile, results of the family member(s). Facts cumulated from cities regression analysis indicate that a household and regencies nevertheless indicate that mar- headed by a man is only 36.5% at risk of ital status does not influence probability of falling into poverty. The research also identified poverty. On a regency level, marriage is a sig- women as relatively significant contributors to- nificant influence of poverty in Sukabumi and ward household finances. Working female heads Ciamis where probability of poverty of married of households earn an average of Rp190,133 per households is well over 80%. Simultaneously in capita. cities, marriage significantly influences poverty Meanwhile, facts prove otherwise in the ma- only in Cirebon, Cimahi dan Tasikmalya where jority (76.5%) of regencies and cities. In a re- probability of poverty in married households is gency and city level, having male household above 70%. heads do not impact household poverty (with the exception of regencies Sukabumi, Ciamis, 5.2. Economic Variables and Indramayu). Facts cumulated also shows that both men and women impact In line with the hypothesis, economic fac- the household in a relatively equal sense. Mean- tors significantly influence the probability of while, male household heads in cities signifi- poverty of households. Merely 1,572 or 7.65% cantly impact the poverty level of households in out of a total 20,451 impoverished households different ways. In Bandung, households headed receive a salary of Rp196,433 per capita. Em- by males are highly vulnerable and are 89.8% ployment by the agriculture industry is also likely to fall into poverty. In Cirebon, Cimahi, exposed as the most influential determinant Tasikmalaya and Banjar, the probability of of poverty level with a 61.7% probability of poverty of households headed by males is rela- poverty percentage. This is so because house- tively low, or below 30%. holds employed by the agriculture industry Djamaluddin, S./Analysis of Poverty Determinant... 11 Table 10: Result of Estimation of Logistic Regression Model at Province Level

Variable Odd Ratio Standard Error P > Z Demographic Variable Male 0.57725410 0,114161 0.0050 Household Size 167.744.400 0.035795 0.0000 Marital Status 173.206.500 0.33791 0.0050 Economic Variable Agriculture 161.382.200 0,113317 0.0000 Laborer/worker 117.360.700 0.073137 0.0100 Asset 0.33071560 0.022405 0.0000 Social Variable Basic Education 207.719.800 0.187528 0.0000 Healthcare 0.85995510 0.062326 0.0370 Home Flooring 199.574.700 0.168248 0.0000 Area Variable Area of Residence 0.58391270 0.039888 0.0000 Loan Program 0.44401490 0.061937 0.0000 Constant 0.01127660 0.001784 0.0000 receive lower wages compared to other em- ers/laborers is Rp210,030 or lower compared to ploment fields. At Rp189,521.30, the average the average income of individuals with other per-capita salary of households employed by employment statuses (Rp222,469.6). In Suk- the agriculture sector is lower than the aver- abumi, Garut and West Bandung, the prob- age per-capita salary of inpoverished house- ability of poverty of households with labor- holds (Rp201,128.6) employed by other sec- ers/workers is relatively high at above 60%. tors and industries. Meanwhile in cities, jobs Meanwhile in Karawang, the probability of in the agriculture industry is relatively limited poverty of the same type of household is con- and is therefore not a significant determinant sidered low at 36%. of poverty in cities other than Bandung and As a financial reserve, assets play an impor- Tasikmalaya where the average probability of tant role in lowering the probability of poverty poverty is over 60%. of households. Results of the research indicate Employment as a laborer is also a determi- that households with assets are less likely to nant of poverty in households. In a provin- fall into poverty (24.8%) and that the more cial level, the probability of poverty in house- assets owned by a particular household, the holds headed by a laborer is also relatively high less likely that household is likely to fall into at 54%, a phenomenon likely due to the fact poverty. Furthermore, data from the national that laborers and non-laborers are paid roughly socio-economic survey indicates that 62.4% or the same as shown by the average per-capita the majority of impoverished households do monthly income of laborers and non-laborers not own assets and that the remaining 355,142 at Rp194,388.6 and Rp199,148.5 respectively. impoverished households own relatively low- Meanwhile, employment status is not a sig- valued assets. The majority of assets owned by nificant determinant of poverty level in cities impoverished households are bicycles. and regencies with the exception of regen- Inter-area observations also identified house- cies Sukabumi, Garut, Karawang and West holds with low-valued assets to be likelier to Bandung. Many workers/laborers are hired fall into poverty. Similar conclusions were made in Karawang, an industrial zone. The aver- in observations conducted on a provincial, city age per-capita income of impoverished work- and regency level (with the exception of regen- Djamaluddin, S./Analysis of Poverty Determinant... 12 cies Cirebon and Indramayu). Simultaneously, year (Rp195,632), an interesting fact indicat- the average probability of poverty for house- ing that impoverished households with differ- holds with assets is low (20%). Results are in ent health conditions receive roughly the same line with the proposed hypothesis. income. This particular fact also reflects that the productivity of impoverished households 5.3. Social Variables remain relatively stagnant. The impact of healthcare on poverty also Education level also impacts the probabil- varies on a provincial-level and city and ity of poverty of households on a substantial regency-level. Bandung, Depok and Tasik- scale. Outcome of the research indicates that malaya are the only three cities where health- lowly educated households are 67.5% likely to care is a significant determinant of poverty. In fall into poverty, seeing that education plays Bandung, households that sought medical at- an important role in raising employee value. tention at least once in the past six months is A lowly educated household is not only lim- only 20% likely to fall in poverty, unlike Depok ited to certain jobs but will also be likely to and Tasikmalaya where probability of poverty earn a relatively low salary. Data from the 2010 in households with the same circumstances are national socio-economic survey shows that at significantly high at over 60%. Results are in Rp194,634, the average per-capita monthly in- line with the proposed hypothesis. Meanwhile come of lowly educated impoverished house- on a regency level, the probability of poverty holds is lower than the per-capita income of of households that sought medical attention at households holding an educational certifica- least once in the past six months in Sukabumi, tion above the primary/elementary level at Tasikmalaya and Karawang is relatively low. Rp208,747 per month. Simultaneously, the probability of poverty of Furthermore, the impact of education level households that sought medical attention at on poverty varies among different cities and re- least once in the past six months in West Ban- gencies. Data cumulated from the majority of dung regency is relatively high at 66.6%. areas on a regency level indicates that low edu- Data from the 2010 national socio-economic cation does not influence the poverty level and survey also indicates that 173,193 households that merely 30% of areas reported of a high have earthen or bamboo flooring and that the probability poverty percentage (above 65%) in number of such households varies significantly lowly educated households. Once again, results among different cities and regencies. A no- are in line with the proposed hypothesis. Mean- ticeable relationship between area of residence while, education level plays a significant role and poverty level was also observed at regen- in over 60% of cities. Lowly educated house- cies Bandung, Garut, Indramayu, Subang, Pur- holds are found to be over 70% likely to fall wakarta, Karawang, Bekasi and West Bandung into poverty. where the average probability of poverty is rel- A healthy household head capable of work- atively high at 75% (with exception of West ing optimally is also likelier to receive a higher Bandung at below 1%). The probability of income. Meanwhile, the probability of poverty poverty for the same variable in cities Suk- of households which sought medical attention abumi and Tasikmalaya is also high at almost at least once in the past six months is 46%. 70%. The average per-capita monthly income of a healthy impoverished household is Rp199,693 5.4. Regional/Area Variables or more or less the same as the average per- Cities generally offer many alternative jobs capita monthly income of households that seek and job opportunities. It is hence an unsur- medical attention at least once each month or prising fact that the probability of poverty of Djamaluddin, S./Analysis of Poverty Determinant... 13 households in cities is relatively low at 36.8%. The factors are identified as gender of house- Impoverished households working and living hold heads, household size, marital status, em- in cities earn an average monthly per-capita ployment in the agricultural industry, employ- salary of Rp205,260, or higher than the aver- ment status as worker/laborer, asset, low edu- age monthly per-capita salary of impoverished cation, healthcare, area classification and busi- households in villages at Rp189,473. ness loan. Among other regencies/cities with a The relationship between area classifica- high number of poverty determinants are re- tion and poverty is significant in all but gencies Bandung, Garut and West Bandung five regencies which are Sumedang, Subang, with 7, 8 and 7 poverty determinants respec- Karawang, Bekasi and West Bandung. Com- tively. Poverty determinants in Bandung are pared to households living in regencies areas, identified as household size, employment in those living in city hve lower poverty proba- the agricultural industry, employment status as bilty rate of around 46%. Meanwhile, admin- workers/laborers, asset, low education, home istrative cities (kota administrasi) generally do flooring, area classification and loan programs. not contain an area classified as a village. Con- Meanwhile, poverty determinants in Bandung sequently, the effect of classification of areas is and Garut are the same with the exception of not reflected in this study. employment status as workers/laborers which Government policies play a relatively sig- is a poverty determinant in Garut but not in nificant role in countering poverty. However, Bandung. the level of influence proves to be different Meanwhile, the characteristics of poverty in from the proposed hypothesis. The probability Purwakarta and Karawang vary from other of poverty of households that receive business areas. The six poverty determinants in Pur- loans is relatively low at 30.7%, a fact which is wakarta are identified as gender of household in line with the proposed hypothesis. Research head, household size, assets, household-level also indicates that 5.5% of impoverished house- characteristics, area classification and loan pro- holds and 10.6% of non-impoverished house- gram. Simultaneously in Karawang, poverty holds are recipients of business loans. In cities, determinants are identified as household size, however, business loans do not significantly in- employment in the agriculture industry, mar- fluence poverty and that business loans only ital status, asset, healthcare and household- significantly impact poverty in regencies Bo- level characteristics. gor, Bandung, Indramayu, Subang and Pur- wakarta. The probability of poverty in house- The problem of poverty is considered simpler holds that receive business loans is also rela- in regencies Kuningan, Cirebon, Majalengka tively low at below 26%. and Sumedang. There are only three determi- nants of poverty were identified in each of the four regencies, or household size, employment 6. Characteristics of Inter-Area Poverty field and area characteristics in Cirebon. The determinants of poverty in Kuningan and Ma- The inter-area poverty determinant analy- jalengka are the same – household size, assets sis indicates that the characteristics of poverty and area classification. Meanwhile, the deter- vary among diferent areas. The dimenson and minants of poverty in regencies Subang and complexity of poverty in a regency level espe- Bekasi are roughly the same, with the excep- cially vary. Compared to other regencies and tion of business loans as a significant deter- cities, is facing a more com- miner of poverty in Subang but not Bekasi. plex poverty issues with at least 10 factors sub- Similarly, the determinants of poverty in re- stantially influence poverty in need to address. gencies Majalengka and Subang are also more Djamaluddin, S./Analysis of Poverty Determinant... 14 or less the same, with the difference being area the difference being household-level character- classification which is a significant determinant istics as a significant determinant of poverty in in the former but not the latter. Sukabumi but not in Bekasi. The characteristics of poverty in Cianjur is roughly the same as that in Kuningan and Cire- 7. Conclusion bon, with the primary difference being employ- ment by the agriculture industry which is a sig- The determinants of poverty vary among nificant poverty determinant in Cianjur but not different cities and regencies in West Java. Cirebon. Based on observations to measure the signif- Meanwhile on a city-level, the question of icance of each determinant of poverty con- poverty is the most complex in Tasikmalaya ducted through the regression model, house- where 9 determinants of poverty were iden- hold size and assets are the primary deter- tified. The determinants of poverty in Tasik- minants of poverty in cities and regencies in malaya are gender of household head, house- West Java. At province, city and regency level, hold size, marital status, employment by the the size of households members are identified agriculture industry, employment status as a as among the major factors which can cause worker/laborer, asset, low education, health- higher probabilty of poverty. Meanwhile, own- care and home flooring. ership of assets is the factor that causes low 5 and 6 poverty determinants were identified probability of poverty in regencies and cities. respectively in Bandung and Cirebon and that The problems of poverty is most complex in both cities more or less have the same poverty Sukabumi regency and Tasikmalaya city where determinants with the difference being business 10 and 9 significant determinants of poverty field and healthcare as significant poverty de- were identified respectively. Several regencies terminants in Bandung but not Cirebon. Si- and cities also possed the same or similar types multaneously, marital status is also a signif- of poverty determinants. icant determinant of poverty in Cirebon but not Bandung. The four poverty determinants 8. References in both cities are gender of household heads, size, asset and low education. [1] Afandi, W. N. (2001). Identification of Poverty Unlike the cities mentioned above, the ques- Characteristics of Household in Padang Pariaman tion of poverty is relatively simpler in remain- Regency (Case study: Nagari Malai V Suku). Prodi Perencanaan Pembangunan, Program Pas- ing cities. Three determinants of poverty are casarjana, Universitas Andalas. http://pasca. identified in cities Bekasi, Depok and Banjar, unand.ac.id/id/wp-content/uploads/2011/09/ with similar poverty determinants being house- IDENTIFIKASI-KARAKTERISTIK-RUMAH-TANGGA- hold size and asset. The difference is that low MISKIN-DI-KABUPATEN-PADANG-PARIAMAN-STUDI- KASUS-NAGARI-MALAI-V-SUKU.pdf eduction, healthcare and asset is a significant [2] Geda, A., de Jong, N., Mwabu, G., & Kimenyi, determinant in Bekasi, Depok and Banjar re- M. (2001). Determinants of Poverty in Kenya: A spectively. Household Level Analysis. ISS Working Paper Se- ries/General Series, 347, 1–20. Netherlands: Insti- Meanwhile, the determinants of poverty in tute of Social Studies. http://repub.eur.nl/pub/ Banjar is also more or less the same as that 19095/wp347.pdf in Cimahi, with the difference being marital [3] Haughton, J. H., & Khandker, S. R. (2009). Hand- status as a significant determinant of poverty book on Poverty and Inequality. World Bank Pub- in Cimahi but not in Banjar. Simultaneously, lications. [4] ILO. (2004). Jender dan Kemiskinan di Indone- the determinants of poverty in Sukabumi is also sia, Seri Rekomendasi Kebijakan: Kerja Layak more or less the same as that in Bekasi, with dan Penanggulangan Kemiskinan di Indonesia, Djamaluddin, S./Analysis of Poverty Determinant... 15 Constant gram Region Loan Pro- flooring Healthcare Home Level Asset Education Employment Status Agriculture Sector Marital Status Number of house- hold member Table 11: Result of Estimation of Logistic Regression Model at Regency Level lamin RegenciesBogor RegencySukabumi RegencyCianjur RegencyBandung Regency JenisGarut Regency Ke- 0.236577* 0.563695Kuningan Regency 2.262156* 0.382315Cirebon 0.765621 0.7170032 Regency 1.848991*Majalengka 4.132641* RegencySumedang 1.834426* 2.221533* Regency 1.722832* 218.499 1.583.155Indramayu 2.051967* Regency 1.001.637 2.454.236 107.747 0.065124*Subang 1.053.802 Regency 1.166.252 0.676197 1.714211**Purwakarta 1.823873* Regency 0.653728 1.851758* 1.815672* 1.695911*Karawang 0.261836* 1.526897** Regency 1.953.832 1.271.615 1.539.088Bekasi 0.686613 0.309453* 1.674924* Regency 1.207.362 0.882666 6.922637** 1.435.741 2.148064*Bandung 3.309972** Barat 2.634441* Regency 0.378078* 0.091241* 1.624.813 1.302.704 1.987360*Bogor 1.270.194 City 0.396354** 2.075793* 1.557.173 1.170.748 0.526128 0.325822* 0.264505*Sukabumi City 0.527951 0.438959 0.779635 1.669200* 1.853800** 1.081.267 1.430.848 0.148756*Bandung 1.488.489 0.991496 2.995907* City 1.814.018 0.950534 1.961903* 2.202379** 0.957202Cirebon 1.613139* City 1.209.947 104.698 0.198407* 2.144834* 1.393.779 1.735525* 1.163.897Bekasi 0.186546* 1.107.587 0.327173* 1.474.724 City 12.338 0.784996 1.586.127 1.160.049Depok 0.226914* 145.549 City 1.753.559 1.514.111 1.064.433 1.137.565 0.314041* 0.709425 1.289.612 2.885.114Cimahi 2.895125* 1.306749* City 1.346.827 1.193.602 1.203.738 0.474713Tasikmalaya 1.875725* City 2.769.699 1.104.465 0.001091 2.185627* 1.645.738 4.053968* 0.190228* 0.59923 0.498794*Banjar 1.021.137 1.701.804 0.531674* 0.435954 0.607737 City 0.149374* 0.505493* 8.890726** 0.878696 13.694 1.665.574 2.229616* 2.375589* 0.388132* 0.568573* 1.118.024 0.351413* 0.190220* 0.157086* 3.050276* 3.845610* 2.041732* 0.167052* 4.711060* 1.854577* 115.209 0.330146 128.219 0.347881* 0.311354 0.169776* 0.758737 1.333.527 0.528329* 0.305863 0.013333 0.741254 0.01192 0.701257 0.444598* 1 0.513706 0.434071* 2.215539* 1.274.958 1.168.388 4.136970* 0.007284 1 1.413.697 2.127297** 1.482.134 0.53582 0.430925* 0.368729** 0.329998 0.951471 1.176.667 0.361766 0.00745 1.571666* 7.318382* 1.551.553 1.209.834 0.439070** 1.116154* 0.679188 0.231337* 2.071138* 1.886801* 0.820916 0.409991* 0.199975 154.628 2.563564* 3.109277* 1 0.350727* 0.838379 1.275851* 0.001284 1.486.892 0.762609* 0.942979 0.814439 0.093948* 6.432886* 0.617471 6.274683* 2.607953* 1.822830** 2.213019* 0.656437** 1.632284* 0.073657 2.199190* 0.533986 0.150612 1.095.583 0.397950* 0.114262* 0.183124* 0.724205 0.017403 0.285741 0.090244* 0.848445 2.053641* 0.338652* 0.924471 0.772579 2.168002* 1.554.637 1 0.117441** 1.256318* 5.512849* 1.512.033 0.006226 0.068062 2.715469* 0.170693 0.022293 6.481388* 1.071.158 0.347463** 0.423182* 0.028577 0.321936 0.001449 1.345.857 0.637123 0.273479* 1.966.845 0.003014 1.590.986 0.039976 0.910884 0.061061* 1 5.593050* 1.247.329 0.009837 0.748154 0.155046* 1.192470* 2.727.437 1.696.025 0.392822* 1.076.809 1 0.262313* 3.728366* - - - 5.117238* 0.538717 0.005843 2.080.121 1.060.275 0.306442 0.618488 3.318.088 1.104.855 1.544.605 0.262728 0.108909* 0.472957 - 1.915.135 - 0.900683 0.647814 0.922448 1.673.457 0.029995 0.150880* 0.702142 0.001116 0.000162 0.003983 3.361596** 2.638931* 0.612592 210.639 1 0.822622 1.511.021 0.029426 - 0.005324 0.004022 268.266 - - 0.01148 - 0.005816 *Significant at 5 percent, ** Significant at 10 percent. Djamaluddin, S./Analysis of Poverty Determinant... 16

2003. : Kantor Perburuhan Interna- sional – International Labour Organization. http://www.ilo.org/wcmsp5/groups/public/ ---asia/---ro-bangkok/---ilo-jakarta/ documents/publication/wcms_125243.pdf [5] Mok, T. Y., Gan, C., & Sanyal, A. (2007). The Determinants of Urban Household Poverty in Malaysia. Journal of Social Sciences, 3 (4), 190– 196. [6] Mukherjee, S., & Benson, T. (2003). The Determi- nants of Poverty in Malawi, 1998. World Develop- ment, 31 (2), 339–358.