THE IMPACT OF PROGRAM KELUARGA HARAPAN ON HOUSEHOLDS’ POVERTY LEVEL AMIDST COVID-19 PANDEMIC IN BALI PROVINCE OF INDONESIA Текст научной статьи по специальности «Экономика и бизнес»

DOI 10.18551/rjoas.2021-07.06

THE IMPACT OF PROGRAM KELUARGA HARAPAN ON HOUSEHOLDS' POVERTY LEVEL AMIDST COVID-19 PANDEMIC IN BALI PROVINCE OF INDONESIA

Prayasta I Gede Heprin1,2*, Budhi Made Kembar Sri1

1Faculty of Economics and Business, University of Udayana, Bali, Indonesia 2BPS, Statistics Bangli, Bali, Indonesia *E-mail: gedeheprinprayasta@gmail.com

ABSTRACT

The outbreak of the unprecedented Covid-19 pandemic resulted in an economic downturn that declined household income. Bali experienced the most severe impact because of the deepest economic contraction during the last two decades. The (GoI) provided so many schemes for social assistance. However, it was the Program Keluarga Harapan (PKH) which is designed as the center of excellence and the largest comprehensive assistance for reducing the burden of poor households even before pandemics. This study objective is to investigate the impact of PKH on the household poverty status during the Covid-19 pandemic. The data was analyzed from the result of Susenas 2020. Further, Ordinal Logistic Regression Model (OLRM) is estimated to evaluate the impact of PKH. The findings showed a positive and significant impact of PKH on the household poverty levels. A household that received PKH also has a significant positive odds ratio to escape the poor class compared to others who are not. Thereby, this research proposed a recommendation to maintain the sustainability of PKH as one of the effective poverty alleviation programs in Indonesia.

KEY WORDS

Poverty, PKH, ordered logistic, Covid-19, Pandemic.

The outbreak of the Covid-19 pandemic has resulted in an economic downturn since it was announced as a global pandemic on 11 March 2020. Indonesia's economic contracted by 2.07 % (c-to-c) during 2020 (BPS,2021). The Socio-Demographic Impact Survey conducted by BPS, Statistics Indonesia found the most vulnerable groups affected by the pandemic are the lowest income groups (income less than IDR 1.8 million). In the lightest scenario, the impact of Covid-19 was predicted to result in an increase of poverty up to 9.7%, while in the worst scenario poverty will reach 12.4% (Suryahadi et.al., 2020). In other words, 8.5 million people will likely fall into poor. Consequently, it is urgent to expand the social security policies to reach vulnerable groups and rise the coverage immensely.

Bali is a province in Indonesia that has been affected the worst by the Covid-19 pandemic. Its economic growth experienced the deepest contraction within the past two decades by 9.31% (BPS Provinsi Bali, 2021). It was estimated that 24.69% or around 98.18 thousand people stopped working and became unemployed during February-August 2020 due to the Covid-19 pandemic. The Unemployment Rate (TPT) was also reported to increase sharply to 5.63% in August 2020, up 4.06% compared to August 2019 which was recorded at 1.57% {Berita Resmi Statistik No. 69/11/51 / Th. XIV, 5 November 2020). Before the pandemic, Bali Province always first placed for the lowest unemployment rate but now it has to settle for being ranked 18th out of all provinces in Indonesia1.

The slowing down economy will automatically impact the level of household welfare. The declined sources of household income which in turn threatens the fulfillment of basic needs, namely household consumption for food and non-food ingredients. As the result, the Covid-19 pandemic might pull the households deeper into living under poverty. The number of poor in Bali Province by September 2020 was estimated to reach 196.92 thousand people, an increase of 31.73 thousand as compared to the number of poor people in March 2020 which was reported at around 165.19 thousand people. An increase of 0.67% occurred which

1https://www.balipost.com/news/2020/11/06/156211/Jika-Kasus-COVID-19-Masih-Terus...html

was reported at 4.45% in September compared to March 2020 at a 3.78% level (Berita Resmi Statistik, No.15/02/51 /Th.XV, 15 February 2021).

Poverty is a multidimensional issue. Oxford Poverty and Human Development Initiative (OPHI) defined multidimensional poverty as a pool of deprivation experienced by the poor to meet their minimum daily basic needs. The World Bank measured poverty with monetary approaches in which those living less than US$1.99 (PPP) a day with lack access to services and essential goods classified as poor. Meanwhile, BPS, Statistics Indonesia measuring poverty utilizes the basic-needs consumption approach. The poor is a household with a monthly average expenditure per capita lower than the poverty line. The concept considers the household capacity to meet basic needed both for food and nonfood. This research utilized and referred to the poverty concept and data from BPS, Statistics Indonesia.

The Government of Indonesia (GoI) considers poverty seriously through implementing social protection program even before the pandemic. It is expected to minimize the poor and near-poor burden avoiding the pass to their future generation (Bappenas, 2014). The International Labor Organization (ILO) (1984) as cited in the Social Counseling Center of the Ministry of Social Affairs of the Republic of Indonesia defines social protection as a policy design to minimize the impact caused by economic and social shocks that lead to reduced or even loss of income. Social protection, fundamentally at least should cover four important aspects titled the Social Floor Initiative (SPF-I) consisting: health service insurance, education, and other social services such as basic income security for children, productive age, and the elderly (United Nations (2009) in Bappenas (2014)).

Indonesia's social protection scheme is classified into social assistance (non-contribution) and social security (with contribution). Social assistance is funded by the state through certain procedures sources entirely from the APBN. There are some programs included as social assistance such as Program Keluarga Harapan (PKH), Bantuan Pangan Non-Tunai (BPNT) or Bantuan Sosial Beras Sejahtera (Bansos Rastra), ASPD, ASLU/BANTU-LU. Meanwhile, social security is regulated in Law Number 24 of 2011 which includes health insurance programs, work accident insurance, elderly benefits, pension benefits, and death benefits which are implemented through the Social Security Administering Institution (BPJS) for Health and Employment.

The Family Hope Program or PKH is Indonesia's social assistance well-known as Conditional Cash Transfer (CCT) to very poor households/very poor families (RTSM/KSM) that have been designated as PKH beneficiaries. They are required to meet requirements and commitments related to efforts to improve the quality of human resources both for education and health. In the short term, this program goal is to decrease the burden on RTSM consumption while it is expected to break the inter-generational poverty chain in the long term so the next generation can escape from the poverty trap (Kemensos, 2018).

PKH implementation is a comprehensive effort to achieve sustainable development goals (SDGs). Five components of SDGs that PKH is expected to address: reducing poverty and hunger; basic education; gender equality; reduction in infant and under-five mortality rates; reduction in maternal mortality. PKH was carried out in a sustainable design, starting with trials in 7 provinces in 2007. This trial was intended to test various instruments that were related and needed in the implementation of PKH, including targeting methods, data validation, verification of requirements, payment mechanisms, public complaints, processes mentoring, and others. The program then has been implemented in 34 provinces, including 512 districts and cities and 7,214 sub-districts in 2018 (Kemensos, 2018).

PKH is directed to become the epicenter and center of excellence for poverty reduction that synergizes various national social protection and empowerment programs among other social assistance. PKH distributed with the most accurate data and continuous assistance which is a pioneer to synergize social protection and empowerment programs compared to various other social protection programs. PKH complementarity with various other programs is expected to accelerate the increase in the welfare level of KPM and thus contribute to reducing national poverty (Habibulah, 2017).

There is are some CCT programs in several countries which successfully implemented with a significant impact on poverty. Conditional Cash Transfer (CCT) in Brazil known as

Bolza Familia has succeeded in increasing the income of the poor by up to 47% for food and groceries (Kamakura and Mazzon, 2015). Fernandez and Olfindo (2011) found that the CCT program in the Philippines, entitled Pantawid Pamilya, was able to reduce inequality by 5.3 points and the depth of poverty by up to 4.3%. Meanwhile, other CCT programs such as Progresa in Mexico succeeded in reducing poverty by 17%, and Familiasen Accion reducing poverty in Colombia by more than 6% (Brauw, et.al, 2015). Familiasen Accion's initial evaluation showed a positive effect on short-term outcomes such as household consumption. The household consumption increased by 13-15% (Attanasio et al., 2005, 2006, and 2009; Attanasio and Mesnard, 2005 in Javier and Adriana 2011) after two years of the program implemented. There is a positive effect of Juntos, the CCT program in Peru, on reducing the chance of inequality as measured by the Unsatisfied Basic Needed (NBI) method Secanella (2017). Cruces et.al (2011) evaluated the significant impact of the Juntos CCT in Peru on consumption and found a moderate impact on poverty reduction.

The success of conditional cash transfer programs in poverty alleviation has been demonstrated in several countries around the world. The program succeeded in providing the poor with adequate access to meet basic needs in the short term. PKH's is accounted for a flagship and important program to accelerate poverty alleviation in Indonesia even before the economic shock hit the country. These priorities can be seen implicitly from the projection of the social protection beneficiaries projected coverage in 2020-2024 published by the National Team for the Acceleration of Poverty Reduction (TNP2K) in 2018. PKH integration is targeted to be the main priority for social protection in Indonesia with the largest beneficiary family target among other social protection programs. The problem of poverty alleviation does not only depend on the effectiveness of PKH impacts. Thus, the study objectives are examining the impact on household poverty status during the hard time in the unavoidably economy shock lead by Covid-19 pandemic in Bali.

METHODS OF RESEARCH

This research analysis section began with descriptive analysis with Cross-Tabulation. This statistical method describes two or more variables simultaneously and the results are displayed in a table that reflects the joint distribution of two or more variables with a limited number of categories (Agresti, 2002). In order to estimate the impact of PKH on household poverty level, this study utilized the Ordered Logistics Regression Model (OLRM). OLRM methods applied to investigate whether the PKH can boost the probability of the poor household entering the higher income class or even escaping of being poor by identifying its odds ratio. The model for ordinal logistic regression is a cumulative logit model with the characteristics of the response variable expressed in cumulative odds (Agresti, 2002). In the model with the s response category with the predictor variable in vector xi with the number of predictor variables q, the cumulative probability P (Y<s|xi) is defined as follows:

P(Y<s\Xi) = n(x) (1)

n{x) = exp ^^ (2)

1+ exp (fios+Xr=1 PrXir)

Where xt = (xi1, xi2,..., xir) is the value of the ith observation (i = 1, 2, 3,..., n) of each q predictor variable. The logit transformation is used to estimate ordinal logistic regression parameters so that the equation is obtained:

Logit P(Y<s\Xi) = In (¿£<<2L) (3)

After being substituted for the initial equation, the empirical model of ordinal logistic regression in Agresti (2002) is formulated as:

Logit P(Y<s\Xi) = pos + S=i Pr*ir (4)

The ordinal logistic regression model will be formulated through some stages as following:

Parallel Lines Test. The parallel lines assumption means that the odds ratio is not affected by where the variable is not freely dichotomized (Kleinbaum and Klein, 2010). This assumption can be tested by using the log-likelihood ratio test (Azen and Walker, 2011).

The Goodness of Fit Test. Model fit testing can be done by comparing the observed value for a subject with the predicted value for that subject (Kleinbaum and Klein, 2010). A model is perfectly fit when there is no difference between observed and predicted for all observations. A Goodness of fit test is using the Pearson test and Deviance test (Fagerland and Hosmer, 2012).

Determination Coefficient Analysis. This analysis resulted in how the variation of dependent variables can be explained by covariates included within the regression model. The higher coefficient means the larger covariates contribution to explaining the variation of dependent variables.

Simultaneous Test. Simultaneous testing aims to examine the role of the proposed independent variables on the dependent variable together using the Likelihood Ratio Test (Hosmer, Lemeshow, and Sturdivant, 2013). If there is a significant result, the overall test can proceed to evaluate every single particular independent variable.

Partial Test. This test sequentially examines whether single testing the meaning of the model parameters with one predictor variable is carried out to determine whether or not there is a relationship between the predictor variable and the response variable (Le, 1998).

Model interpretation through Odds Ratio. The common form of regression model interpretation is defining the dependent variable's unit of change which is caused by the independent variable and determines the functional relationship between the dependent and independent variables. In order to make it easier to interpret the ordinal logistic regression model, odds ratio values are used (Hosmer and Lemeshow, 2000). It is the value that shows the comparison of the level of the tendency of two or more categories in one of the independent variables with one of the categories is used as a comparison. Therefore, the relationship will be more meaningful and very much likely easier to understand regarding the objective of the study.

This study analyzed the data result of the National Survey on Social Economics (Susenas) in March 2020. Susenas is a cross-section survey conducted by BPS in order to estimate the poverty level until the regional level. This research analyzed a 6.231 household sample where 351 of them are received PKH. Household samples at Susenas March 2020 were selected using a two-stage one-phase sampling technique with implicit stratification according to the education of the head of the household, under-five years (balita) household members, and nine months pregnant women. In order to complete the figure before the pandemic, some part of the analysis were also conducted by using the 2018 and 2019 data.

There were three variables defined to support the analysis, such as treatment, outcome, and control variables. The treatment variables included in the analysis described according to the household PKH status (0: not received PKH and 1: received PKH). Meanwhile, the others variables were outcome and control variables which are described in Table 1 and Table 2.

Table 1 - Outcome Variable

Variable_Scale_Details_

Flag poor Ordinal 0: Extremely poor1: Poor2: Nearly poor3: Other nearly poor4: Not poor_

Source: Susenas, 2020 (processed).

The flag poor is classified according to the monthly expenditure per capita (MEPC) compared to the poverty line (PL) as follows: Extremely poor: (MEPC< 0.8 PL); Poor: (0.8 PL < MEPC); Nearly poor: (PL < MEPC< 1.2 PL); Other nearly poor: (1.2 PL < MEPC < 1.6 PL); Not poor: (MEPC > 1.6 PL). The poverty line is referred to March 2020, where it is published by BPS, Statistics Bali Rp 443 070,- (Urban) and Rp 401 291,-(Rural).

Table 2 - Control Variables

Variable Scale Details

HH marital status Nominal 0 Unmarried

1 Married

HH productive age Nominal 0 No &1: Yes

HH education Ordinal 0 No school

1 Primary

2 Junior

3 Senior

HH industry sector Ordinal 0 Not work

1 Primary

2 Secondary

3 Tertiary

HH occupation Ordinal 0 Not work

1 Employment

2 Entepreneur

Pregnant women Nominal 0 No &1: Yes

Toddler Ratio Number of household members age (0-5 years old)

Elderly Ratio Number of household member age > 60 years old

Type of fuel Nominal 0 Traditional & 1: Modern

Type of wall Nominal 0 Others &1:Concrete/plastering

Type of roof Nominal 0 Others & 1: Concrete

Clean water Nominal 0 No & 1 Yes

Lighting Nominal 0 : Not Electricity & 1: Electricity

Toilet facilities Nominal 0 No facility

1 Public

2 Private

House ownership Nominal 0 No & 1 Yes

Type of floor Nominal 0 No & 1 Yes

Credit Nominal 0 No & 1 Yes

Land ownership Nominal 0 No & 1 Yes

Refigerator ownership Nominal 0 No & 1 Yes

Jewelery> 10 gr Nominal 0 No & 1 : Yes

Car ownership Nominal 0 No & 1 : Yes

BPNT Nominal 0 No & 1 : Yes

Regional Gov. assistance Nominal 0 No & 1 : Yes

JKN / Jamkesda Nominal 0 No & 1 : Yes

Diff. visual Nominal 0 No & 1 : Yes

Diff. to walk Nominal 0 No & 1 : Yes

Diff. to move hand Nominal 0 No & 1 : Yes

Diff.concentration Nominal 0 No & 1 : Yes

Diff.emotional Nominal 0 No & 1 : Yes

Diff.communication Nominal 0 No & 1 : Yes

Diff.selfcace Nominal 0 No & 1 Yes

Source: Susenas, 2020 (processed).

RESULTS OF STUDY

The different characteristics were revealed between household that received PKH (KPM) and the other that did not receive the program (Non-KPM). On average, the Non-KPM spent higher in both urban and rural areas. The wider gap to those living in an urban area was predicted due to the higher inequality that existed.

Table 4 - The Household's Monthly Average Expenditure in Bali 2020

Type of Expenditure Category KPM Non-KPM

Urban Rural 3,874,442.04 3,689,566.13 6,295,424.22 4,190,243.46

Source: Susenas, 2020 (processed).

The diverse source of income and job availability in most of the cities resulted in a higher difference in monthly household expenditure between KPM and Non-KPM in the

urban area. Further, the pattern can be also concluded that the Non-KPM is wealthier as compared to the KPM because they can afford higher basket commodities monthly. It is also shown that PKH already benefited those in the lower-income group the most.

The dynamic poverty rate in Bali before and amidst the Covid-19 pandemic is depicted in Figure 1. It is revealed that the number of poor rose significantly in both urban and rural areas. Long before the pandemic, the slowing down was achieved until September 2019 but the shock expanded the poverty rates during the following period. According to the figure, the poverty rate in Bali was started to increase by 3.78% in March 2020 and 4.45% in September 2020. The rural area was reported to host a higher number of poor as compared to those living in urban. However, it was tended to increase by March 2020 to September 2020 while from March 2018 to September 2019 felt smooth. By contrast, it was expanded started from September 2019 in an urban region. The drawback pattern of the poverty dynamic is predicted due to the different economic structures between urban and rural regions. There is a higher number of people working in the formal sector living in the urban. Given the high population density and economic activities in cities lead to those living in the areas are more likely to impact higher than those in rural (Sharifi and Garsmir, 2021). In order to seek a clear picture of how the dynamics of the poor before and amidst pandemic this study also analyzed other poverty indicators such as the Gini ratio (Table 2), Poverty Gap Index (Table 3), and Poverty Severity Index (Table 4).

5.38 ---- 4.88 4.86 5.40

401 3.91 ,7Q 3.79 3.78 4.45

3.32 3.36 3.04

Mar-18 Sep-18 Mar-19 Sep-19 Mar-20 Sep-20

Urban Rural Urban+Rural

Figure 1 - Poverty Headcount ratio (P0) before and amidst Covid-19 Pandemic in Bali Province.

Source: Susenas, 2020 (processed)

Table 5 - Gini Ratio before and amidst Covid-19 pandemic in Bali Province

Type of Place Period

Mar-18 Sep-18 Mar-19 Sep-19 Mar-20 Sep-20

Urban 0.38 0.36 0.37 0.37 0.37 0.38

Rural 0.32 0.31 0.31 0.31 0.30 0.30

Urban+Rural 0.38 0.36 0.37 0.37 0.37 0.37

Source: Susenas, 2020 (processed).

The impact of the pandemic on inequality can be examined through the trend of the Gini ratio. The tools measured the degree of population expenditure which is used approaches to measure the household income. Poverty alleviation program is necessarily considered poverty not only from the percentage of the poor but also how the program eradicates the inequality. The outbreak of the Covid-19 pandemic reported worsened the disparity across provinces in Indonesia. It tends to increase the inequality in urban areas but

declines in rural areas. It was predicted that the pandemic increasing the spatial inequality at the urban level which probably also related to the distribution of the government subsidies (Brata et.al.,2021). The serious implication of the current pandemic is also predicted to the poverty in Indonesia (Suryahadi, Izzati, and Suryadarma (2020)). Those two previous findings support the pattern in Bali. It was predicted that the inequality in urban areas increased in September 2020 to become 0.37 while previously reach 0.38. By contrast, the coefficient Gini in rural areas decreased from 0.31 by September 2019 to 0.30 during March 2020 and September 2020 (Table 6). Considering the total analysis of the Gini coefficient not much likely changed rapidly from March 2018 to September 2020. The rapidly developing program on rural development is predicted to cause the decrease in inequality during the respective periods while in urban fluctuate around 0.36 to 0.38. This signal provokes the urgency of addressing the poverty from the countryside and proved it is the best option moreover during difficult times because of the Covid-19 pandemic.

Table 6 - Poverty Gap Index (P1) before and amidst Covid-19 pandemic in Bali Province

Type of Place Period

Mar-18 Sep-18 Mar-19 Sep-19 Mar-20 Sep-20

Urban 0.50 0.44 0.49 0.52 0.47 0.55

Rural 1.04 0.67 0.64 0.47 0.65 0.75

Urban+Rural 0.69 0.52 0.54 0.50 0.52 0.61

Source: BPS, Statistics Bali (2020).

Moving on to the Poverty Gap Index (P1), it is explained the gap of the poor from the poverty line. It also considers how far the poor household expenditure on average to the poverty line during the respective period. Both of the urban and rural areas poverty gap indexes rose rapidly from March 2020 to September 2020. Before the pandemic, the trend slowed smoothly. In contrast, it tended to climb after September 2019. The gap in rural areas is higher as compared to those in urban which mean they tend to lives deeper under the poverty line. Consequently, the governments will require to design different policies scheme to reduce poverty in urban and rural areas. In other words, the poor in the rural area will cost the poverty alleviation program higher as compared to those in urban escaping from living under the poverty line.

Table 7 - Poverty Severity Index (P2) before and amidst Covid-19 pandemic in Bali Province

Type of Place Period

Mar-18 Sep-18 Mar-19 Sep-19 Mar-20 Sep-20

Urban 0.12 0.10 0.11 0.11 0.09 0.11

Rural 0.28 0.15 0.13 0.08 0.12 0.14

Urban+Rural 0.18 0.12 0.11 0.10 0.10 0.12

Source: BPS, Statistics Bali (2020).

Taking into account the variety among the poor, the Poverty Severity Index is also known as the squared poverty gap. The higher index means the more various characteristics of the poor household within the region or the higher inequality among the poor. Before the pandemic backward from September 2019, similarly to other poverty indicators, the trend is predicted to decline both in urban and rural areas. The different pattern is reported during September 2019 to March 2020 decreased in urban while in rural increased. It is probably due to the early impact of the Covid-19 pandemic which directly affecting the poor household expenditure. The sudden huge shock burdened the poor household as they lost jobs or laysoff soon after the pandemic reduced economic productivity. Consequently, the household income becomes more unequal in the urban region before distributing to those living in rural regions. The higher the Poverty Severity Index (P2) the more complex scheme needs to be managed to address the poverty because of their diverse characteristics.

Further, the ordinal logistic regression is performed to study the effect once the household benefited from PKH on their level of poverty. This model is begun with a parallel

line test. The test is used to test the assumption that every category has the same parameters or relationship between the independent variable and the logit is the same for all logit formulas.

Table 8 - Parallel Line Test

Model -2 Log Likelihood Chi-Square df Sig.

Null Hypothesis 7,463.23

General 7,304.72 158.51 132 0.06

Source: Susenas, 2020 (processed).

According to Table 8 the significance value (p-value) is 0.06 (> 0.05) with Chi-Square 158,51 so decided to accept the null hypothesis that the resulting model has the same parameters or the link function selection is appropriate. Thereby, the model estimation can proceed to examine the next stages. The following prerequisite test is the Goodness of Fit Test. A perfect fit model is when the value observed perfectly matches it's for all observations. Meanwhile, if there are only a few in the model categorical independent variables, then a contingency table can be compiled with the number of rows is as much as the covariate pattern and the number of columns is as much as the variable category dependent. Then, the observed and estimated values of each cell can be calculated and model fit testing can be done using the Pearson test and test Deviance (Fagerland and Hosmer, 2012).

Table 9 - Goodness of Fit Test

Test Statistics Chi-Square df Sig.

Pearson 18,837.44 19,660 1.000

Deviance 7,214.45 19,660 1.000

Source: Susenas, 2020 (processed).

Table 9 shows the Pearson chi-square value was obtained 18.837,44 with a p-value 1.00. Meanwhile, the chi-square Deviance value was 7.214,45 with a p-value of 1,00. These two tests can be concluded that all lead to the conclusion that the model the resulting fit (fit) with the data. The strong relationship between the dependent variable with independent variables is figured by the determination coefficient. It was 23.9% of the dependent variable can be explained by the independent variables included to the model which is shown by Nagelkerke Pseudo-R Square (see Table 10 below).

Table 10 - Coefficient Determination

Pseudo R-Square Value

Cox and Snell Nagelkerke McFadden 0.183 0.239 0.139

Source: Susenas, 2020 (processed).

Table 11 - Model Fitting Information

Model -2 Log Likelihood Chi-Square df Sig.

Intercept Only Final 8,724.02 7,463.23 1,260.79 44 0.00

Source: Susenas, 2020 (processed).

After analyzing the determination coefficient, the process then checking the fitting of the estimated model. Table 8 revealed that the value of -2 Log-Likelihood in the intercept-only model is 8.724,02 meanwhile when the independent variable was entered into the model the -2 Log Likelihood value fell to 7.463,23 and this decrease was significant at 0.00. This shows that the model with the inclusion of independent variables is better than the

intercept-only model and it can be concluded that the model fit and (Ghozali, 2011). The result also considers a simultaneous test which means that the overall variables show a significant impact on dependent variables or household poverty status.

After completing the prerequisites test then the ordinal logistic regression model is estimated. The partial test examines the significance of the model for each independent variable. This research found some variables that statistically significantly increase the probability of the poor household entering a higher level of income class. Table 12 provides information on some particular variables that significantly affect the household poverty status. The interpretation of the model is presented by the odds ratio. PKH's is significantly to increase the poor household to escape their group to a higher level of income by 2.07 as compared to those who don't receive the program. This finding answered the research question straightforwardly.

Table 12 - The Odds Ratio of Some Significant Variables

Variables Standard Error Sig. Odds Ratio

PKH 0.32 0.02 2.07

HH industry sector 0.09 0.03 1.23

Type of fuel 0.09 0.00 1.63

Toilet 0.13 0.00 2.45

Credit 0.11 0.03 1.28

Refrigerator 0.08 0.00 1.63

Jewelry> 10 gr 0.10 0.00 1.72

Car 0.15 0.00 3.66

Land 0.08 0.00 1.29

Toddler 0.06 0.00 2.46

Elderly 0.05 0.00 1.49

Source: Susenas 2020 (processed).

The OLRM analysis provides an insight that once the household benefited from PKH during the pandemic their welfare may improve through basic needed household consumption. Other variables may interpret that they are possibly escaping being underprivileged as they have sufficient house amenities, intangible assets, and living longer with a sustainable income when getting older. It is clear that the higher income group determine by their asset and decent housing amenities.

Despite the findings, the aforementioned limitation of this research is the impact evaluation conducted in one year only. The cumulative generated impact before the pandemic is minimized under the context of the study to the short-term goals addressed. In order to expand the research, the longer term with education and health outcome need to be concerned as the impact probably will may not feasible soon after the program implemented. However, due to the lack updated of longitudinal data available, the experimental design is one of the possible options to overcome the limitation. It surely should consider the funding, time length of the study, and other related sources. Without declining the impact generated and the urgency of the current situation this research then utilized the available sources to provide an initial finding on PKH's performance during the pandemic.

CONCLUSION

PKH significantly addressed poverty amidst the Covid-19 pandemic. Once the household received PKH it means that their opportunity is predicted around 2.07 times higher to reach the higher income level as compared to those who did not receive the program. During the pandemic, there is an increasing number of poor and inequality. The impact of the shock led by the pandemic during the early period (March 2020) began at the urban level before expanding to the rural level in the following periods. It is probably due to the different variety of economic structures and development between urban and rural.

The poverty alleviation programs should be designed differently to increase the sensitivity of the program due to a higher Poverty Gap Index and Poverty Severity Index.

That indicator means that the poor characteristics are heterogeny, hereby to accelerate the effort targeted-specifics policies need to be considered. Conditional Cash Assistance (CCT) in Indonesia namely Program Keluarga Harapan (PKH) is one of the Indonesian social security specifically targeting the child, maternal mother, student, elderly, and people with disabilities. This specifically targeted program is successfully helping the burden of the poor household amidst the Covid-19 pandemic in Bali.

REFERENCES

1. Agresti, Alan.(2002). Categorical Data Analysis. New York: Inc. John Wiley and Sons.

2. Avila, Pedro. (2012). On the Evaluation of Conditional Cash Transfer Program An analysis of CCT programs and poverty. Quantitative Methods Thesis.Boston: Columbia University.

3. Azen, Razia & Walker, Cindy. (2011). Categorical Data Analysis for the Behavioral and Social Sciences. 10.4324/9780203843611.

4. Badan Pusat Statistik.(2019). Penghitungan dan Analisis Kemiskinan Makro Indonesia.Jakarta: Subdit Statistik Kerawanan Sosial. BPS.

5. Badan Pusat Statistik Provinsi Bali.(2020). Keadaan Ketenagakerjaan Provinsi Bali Agustus 2020. Berita Resmi Statistik Keadaan Ketenagakerjaan Provinsi Bali No. 69/11/51/Th. XIV, 5 November 2020. https://bali.bps.go.id/pressrelease/2020/11/05/717415/keadaan-ketenagakerjaan-provinsi-bali-agustus-2020.html.

6. Badan Pusat Statistik Provinsi Bali. (2021).Pertumbuhan Ekonomi Triwulan IV Tahun 2020. Berita Resmi Statistik No.14/02/51/TH.XV, 5 Februari 2021. https://bali.bps.go.id/backend/materi_ind/materiBrsInd-20210205083 36.pdf

7. Badan Pusat Statistik Provinsi Bali.(2021). Profil Kemiskinan di Bali September 2020. Berita Resmi Statistik Kemiskinan Provinsi Bali No. 15/02/51/Th.XV, 15 Februari 2021. https://bali.bps.go.id/backend/materi_ind/materiBrsInd-20210215092852.pdf.

8. Baez, Javier & Camacho, Adriana. (2011). Assessing the Long-Term Effects of Conditional Cash Transfers on Human Capital: Evidence from Colombia. https://www.researchgate.net/publication/228135655_Assessing_the_Long-Term_Effects_of_Conditional_Cash_Transfers_on_Human_Capital_Evidence_from_Colo mbia/citation/download

9. Bappenas.(2014). Perlindungan Sosial di Indonesia: Tantangan dan Arah ke Depaan. Direktorat Perlindungan dan Kesejahteraan Masyarakat. Jakarta: Bappenas. https://www.bappenas.go.id/files/5114/2889/4558/Perlindungan_Sosial_di_Indonesia-Tantangan_dan_Arah_ke_Depan.pdf.

10. Brata,Aloysius Gunadi; Pramudya,Eusebius Pantja; Astuti, Esther Sri ;Rahayu,Heffi Christya; Heron, Heronimus. (2021). COVID-19 and Socio-Economic Inequalities in Indonesia: A Subnational-level Analysis. ERIA Discussion Paper Series No. 371. https://www.eria.org/uploads/media/discussion-papers/ERIA-Research-on-COVID-19/COVID-19-and-Socio-Economic-Inequalities-in-Indonesia_A-Subnational-level-Analysis.pdf.

11. Brauw, Alan De., Gilligan, Daniel O., Hoddinot, John., Roy, Shalini. (2015).Bolsa Familia and Household Labor Supply.Article in Economic Development and Cultural Change The University of Chicago DOI 0013-0079/2015/6303-0005$10.00.

12. Cruces, G., Perez-Truglia, R., Tetaz, M.,(2011). Biased perceptions of income distribution and preferences for redistribution: evidence from a survey experiment. IZA Working Paper 5699.

13. Department of Social Welfare and DevelopmentPhilippines.2017. Philippines' Conditional Cash Transfer (CCT) Program. https://pantawid.dswd.gov.ph.

14. European Union, INFID, Koalisi Perempuan Indonesia, dan OXFAM.Tujuan 01 Mengakhiri Kemiskinan dimana pun dan dalam segala bentuk. https://www.sdg2030indonesia.org/page/9-tujuan-satu.

15. Fagerland, Morten & Hosmer, David. (2012). A Generalized Hosmer-Lemeshow Goodness-of-Fit Test for Multinomial Logistic Regression Models. Stata Journal. 12. 447453. 10.1177/1536867X1201200307.

16. Festa Secanela, Patricia. (2017). Conditional Cash Transfer Programs and Inequality of Opportunity A case study Juntos in Peru. Master Thesis Department of Economic History: Lund University. http://lup.lub.lu.se/student-papers/record/8915919.

17. Ghozali, Imam. (2011). Aplikasi Analisis Multivariate Dengan Program SPSS. Semarang: Badan Penerbit Universitas Diponegoro.

18. Habibullah.(2017). Perlindungan Sosial Komprehensif di Indonesia Social Protection Comprehensive in Indonesia. Sosio Informa Vol.3 No.01, Januari-April, Tahun 2017. Kesejahteraan Sosial. https://ejournal.kemsos.go.id/index.php/Sosioinforma/article/download/492/458.

19. Hosmer, D.W. dan S. Lemeshow, (2000): Applied Logistic Regression.Second Edition, John Willey & Sons, New York.

20. Hosmer, D.W., Lemeshow, S. and Sturdivant, R.X. (2013) Applied Logistic Regression. 3rd Edition, John Wiley & Sons, Hoboken, NJ.https://doi.org/10.1002/9781118548387

21. Kamakura, Wagner A., Mazzon, Jose Afonso.(2015). Measuring the Impact of a Conditional Cash Transfer on Consumption Behavior with Propensity Scoring, Research Article 2:302-316 24 Februari 2015. Springer DOI 10.1007/s40547-015-0037-0.

22. Kementerian Sosial RI. (2018). Pedoman Umum Keluarga Harapan (PKH) Tahun

2019.Jakarta: Kementerian Sosial RI.

23. Kleinbaum, D.G. and Klein, M. (2010) Ordinal Logistic Regression. In: Kleinbaum, D.G. and Klein, M., Eds., Logistic Regression, Springer, New York, 463-488. https://doi.org/10.1007/978-1 -4419-1742-3_13.

24. Le,C.T.(1998). Applied Categorical Data Analysis. John Wiley and Sons, Inc. USA.

25. Maya, Citra. Jika Kasus Covid-19 Bertambah, Penutupan "Pintu Wisman Berpotensi Diperpanjang" Bali Post 6 November

2020,https://www.balipost.com/news/2020/11/06/156211/Jika-Kasus-COVID-19-Masih-Terus...html.

26. Olfindo, R. and Fernandez, L.(2011). Overview of the Philippines' Conditional Cash Transfer Program: The Pantawid Pamilyang Pilipino Program (Pantawid Pamilya), The World

Bank.http://documents.worldbank.org/curated/en/313851468092968987/pdf/628790BRI0 Phil0me0abstract0as0no010.pdf .

27. Sharifi; Ayyoob, Khavarian-Garmsir; Amir Reza. (2021). The COVID-19 pandemic: Impacts on cities and major lessons for urban planning, design, and management. Science of The Total Environment Volume 749, 2020 142391 ISSN 0048-9697,https://doi.org/10.1016/j.scitotenv.2020.142391.

28. Suryahadi, Asep., Al Izzati, Ridho., Suryadarma, Daniel. Estimating the Impact of Covid-19 on Poverty in Indonesia. Bulletin of Indonesian Economic Studies Volume 56, 2020-Issue 2 halaman 175-192. https://doi.org/10.1080/00074918.2020.1779390.