Effect of Non-Cash Food Assistance on Work Disincentives Among Beneficiary Households in Dukuhturi, Tegal

1. Introduction

Poverty in Central Java remains relatively high. In Tegal Regency alone, there are 123,520 poor residents and more than 65,000 pre-prosperous families. This condition indicates that poverty continues to be a serious issue, despite a relatively large budget allocated for poverty alleviation. Dukuhturi Subdistrict was chosen as the study site because all of its villages are categorized as developing, with an economic structure dominated by informal sectors such as trade, industry, services, and agriculture.

The Non-Cash Food Assistance Program (BPNT) is a reform of the subsidized rice program (Rastra), gradually implemented starting in 2017. Through BPNT, assistance is distributed non-cash via electronic accounts, which can only be used to purchase specific food commodities at designated e-warongs. In addition to enhancing food security, the program is designed to expand financial inclusion and improve the effectiveness of aid distribution. However, various studies indicate that social assistance-whether cash or non-cash-has the potential to influence beneficiaries’ work decisions.

Pre-survey results from 30 KPM households in Dukuhturi Subdistrict showed initial indications of reduced working hours after receiving BPNT, particularly among households with low proportions of assistance and income. Educational attainment and age also appeared to influence variations in work behavior. This situation underscores the urgency of empirically examining how BPNT affects work decisions through recipients’ economic and demographic characteristics. The importance of this study is further supported by previous research showing inconsistent findings: some studies found reduced working hours (Ningtiyas, 2018; Salsabila & Purwanti, 2020; Hoynes & Schanzenbach, 2012; Prifti et al., 2019), while others reported that assistance does not necessarily reduce adult working hours and may instead affect child labor or sectoral shifts (Sulaiman, 2010; Skoufias et al., 2008). These differing results highlight a research gap regarding how food assistance—which differs in nature from cash transfers—affects poor households’ labor behavior. Although several studies have explored the effect of social assistance on beneficiaries’ work decisions, research specifically addressing work disincentives in food assistance programs, particularly BPNT, remains limited. Furthermore, few studies have examined how the proportion of assistance received, share of household income, education level, and age simultaneously influence the likelihood of reduced working hours among BPNT recipients.

The theoretical framework of this study is grounded in labor supply theory, which explains that non-wage income can shift individual decisions from work to leisure, as leisure is considered a normal good. This perspective aligns with Borjas (2012) and Kaufman and Hotchkiss (1999), who demonstrate that increased non-labor income can lead to reduced working hours, particularly in low-income households. In the context of welfare programs, Stiglitz (2000) and Hyman (2011) emphasize that social assistance, whether cash or non-cash, can alter budget constraints and consumption choices, thereby influencing beneficiaries’ work decisions. Although BPNT is non-cash, routine receipts of significant value can still elicit responses similar to cash transfers.

Based on this theoretical foundation and empirical evidence, this study develops hypotheses regarding the influence of BPNT recipient household characteristics on reduced working hours. The proportion of BPNT assistance is expected to be positively associated with decreased working hours, consistent with Prifti et al. (2019), who found that higher proportions of assistance received increase the likelihood of reduced working hours. Conversely, the proportion of household income is predicted to be negatively associated with reduced working hours, as low-income households are more likely to reduce labor participation when receiving aid (Salsabila & Purwanti, 2020), in line with labor supply theory.

Moreover, education level and age are also considered influential factors in reducing working hours. Higher education increases employment opportunities and incentives to remain in the workforce, while individuals with lower education levels are more responsive to assistance by reducing working hours (Kaufman & Hotchkiss, 1999; Vere, 2011). Older age is expected to increase the likelihood of reduced working hours due to declining physical productivity, as Rasyid (2012) found that older household heads are more likely to reduce labor participation.

Based on this background, this study aims to analyze the effect of the Non-Cash Food Assistance Program on work disincentives among beneficiary households in Dukuhturi Sub-district, Tegal Regency.

2. Method

This study employs a quantitative approach to analyze the effect of the Non-Cash Food Assistance Program (BPNT) on the decrease in working hours among beneficiary households. The study population includes all BPNT-recipient households in Tegal Regency in 2022, totaling 112,550 beneficiary households. The sample was selected using a purposive proportional random sampling technique, which involves choosing subjects based on specific objectives or considerations relevant to the research context. The sample was drawn from six villages in Dukuhturi Sub-district: Sidakaton Village, representing areas with trade-based livelihoods; Lawatan, Karanganyar, and Pengabean Villages, representing areas with industrial or service-based livelihoods; and Dukuhturi and Sidapurna Villages, representing areas with agriculture-based livelihoods. The sample size was determined using Slovin’s formula with a population (N) of 112,550 and a 10 percent margin of error, resulting in a minimum sample size of 100 respondents. Data were collected through surveys (interviews and questionnaires) and supplemented with a literature review.

The dependent variable in this study is the decrease in working hours after receiving BPNT, measured as a binary variable. It is coded 1 for households experiencing a decrease in working hours and 0 for households not experiencing a decrease.

The decrease is measured for both the household head and the wife, resulting in the following equation:

The independent variables include the proportion of BPNT and the income proportion for the household-level model. For the sub-models at the household head and wife levels, the dependent variables used are the proportion of BPNT, income, education, and age.

The data were analyzed using a binary logistic regression model because the dependent variables are dichotomous.

The logistic regression model for the household-level analysis is formulated as follows:

The logistic regression model for head of household is as follows:

The logistic regression model for the wife is as follows:

With:

Subsequently, statistical tests were conducted, as reflected in the coefficient of determination (R²), the F-statistic, and the Z-statistic values.

3. Results and Discussion

3.1 Results

3.1.1 Respondent Household Characteristics

The sample used in this study consists of 100 households. In this study, households are categorized as pre-prosperous or poor families. The proportion of the Non-Cash Food Assistance Program (BPNT) reflects the role of BPNT in household food expenditures. The BPNT assistance is fixed for each Beneficiary Family (KPM) at IDR 200,000 per month. This assistance is intended to cover part of the household’s staple food needs to improve food security and the welfare of poor families.

Based on respondent characteristics by BPNT proportion in Dukuhturi Sub-district, the distribution shows considerable variation. The majority fall within the 10.01 - 20.00 percent BPNT proportion group, while only 3 percent of households are in the 0 - 10 percent BPNT proportion group. Of the 100 BPNT-recipient households, 74 households experienced a decrease in working hours, whereas 26 households did not experience a decrease. This indicates that most households have a moderate to high dependency on BPNT assistance to meet their monthly food needs. The higher the proportion of BPNT relative to household food expenditure, the greater the tendency for a decrease in working hours. This suggests that BPNT has the potential to decrease household motivation to work.

Table 1: Household Characteristics Based on BPNT Proportion and Household Income Proportion

Source: Primary Data, 2025, processed.

The household income proportion reflects the percentage of a household’s total income relative to its total expenditures. This proportion indicates the household’s capacity to meet monthly needs. Income refers to all earnings obtained by the household, whether from work or non-work sources. The households in question are Beneficiary Families (KPM) that receive BPNT. Data in Table 1 show that households with a medium income proportion (40.01 - 80.00 percent) have a higher tendency to experience a decrease in working hours, at 80.77 percent. Conversely, households with a high-income proportion (120.01 - 160.00 percent) tend not to experience a decrease in working hours.

In this study, the analysis of the decrease in working hours was conducted by examining changes in the working hours of household heads and wives. This approach was applied to identify whether the decrease in working hours occurred for both or was concentrated on one of them.

Based on Table 2, the decrease in working hours among household heads is more dominant in groups with a higher BPNT proportion. In the low BPNT proportion group (0 - 20 percent), most household heads maintained their working hours. The decrease then rose from 36.84 percent in the 20.01 - 30 percent group to 80 percent in the 30.01 - 40 percent group and 100 percent in the 40.01 - 45 percent group. Among wives, the decrease in working hours was more pronounced compared to household heads. It increased from 51.92 percent in the 10.01 - 20 percent group to 73.68 percent in the 20.01 - 30 percent group, reaching 100 percent in the higher BPNT proportion group (30.01 - 45 percent). Overall, the decrease in working hours in BPNT-recipient households mostly occurs among wives. This indicates that with the provision of assistance, wives tend to reduce work intensity and focus more on domestic activities or household management, while husbands’ productive roles remain relatively stable.

Table 2: Household Characteristics Based on BPNT Proportion (Head of Household and Wife)

Source: Primary Data, 2025, processed.

The analysis of respondent characteristics by income level is presented in Table 3, which classifies respondents into specific income intervals. This grouping aims to provide a more structured overview of income distribution between household heads and wives, as well as its relation to the tendency of changes in working hours in beneficiary households.

Table 3: Household Characteristics Based on Income (Head of Household and Wife)

Source: Primary Data, 2025, processed.

Income-based analysis shows that the decrease in working hours occurs more frequently in lower-income households. For household heads, the decrease is dominated by the IDR 0 - 500,000 income group (84.61 percent), decreasing to 28.85 percent in the IDR 500,001 - 1,000,000 group, and not occurring in households earning above IDR 1,000,000. A similar pattern is observed among wives, with the highest decrease in the 0 - 500,000 income group (72 percent), decreasing to 36.36 percent in the 500,001 - 1,000,000 group, and no decrease above 1,000,000. Overall, the decrease in working hours is more dominant in low-income households, whereas middle- to high-income households tend to maintain their working hours.

Education reflects an individual’s ability to make decisions and take appropriate actions. It also plays a key role in shaping thought processes and behavioral patterns. Based on Table 4, the education level of household heads and wives in BPNT-recipient households in Dukuhturi Sub-district is generally low. Most respondents have only completed primary education (≤6 years), with a small proportion reaching lower or upper secondary education.

Table 4: Household Characteristics Based on Education (Head of Household and Wife)

Source: Primary Data, 2025, processed.

The decrease in working hours mainly occurs among household heads with low education levels. Household heads with ≤6 years of schooling experienced a 33.33 percent decrease, while those with 7 - 9 years of schooling had a 9.09 percent decrease, and no decrease was observed in the 9 - 12 years education group. This indicates that the higher the household head’s education, the less likely a decrease in working hours occurs. Among wives, the decrease is also more pronounced at lower education levels, with a higher proportion compared to household heads. Wives with ≤6 years of schooling experienced a 63.53 percent decrease, and those with 7 - 9 years of schooling experienced a 57.14 percent decrease, while no decrease was observed in the 9 - 12 years group. These findings indicate that wives’ education also influences labor participation patterns within the household.

Age is an important demographic factor closely related to physical capacity in determining whether to work or not. In general, the productive age group tends to have higher work capacity and greater economic participation compared to the non-productive age group.

Table 5: Household Characteristics Based on Age (Head of Household and Wife)

Source: Primary data, 2025, processed

Based on Table 5, the pattern of decrease in working hours differs between household heads and wives. Household heads in the productive age range (37 - 56 years) mostly maintained their working hours (82.93 percent for ages 37 - 46 and 80.77 percent for ages 47 - 56). The decrease increases in older age groups, reaching 58.33 percent for ages 57 - 66 and 100 percent for ages 67 - 75, influenced by declining physical capacity and productivity. Meanwhile, among wives, the decrease in working hours occurs more in the productive age range, especially ages 25 - 44. Wives aged 25 - 34 experienced a 78.57 percent decrease, while those aged 35 - 44 experienced a 59.57 percent decrease, mainly due to childcare responsibilities. In older age groups (45 - 73 years), the decrease remains high, ranging between 56 - 66.67 percent.

Overall, it can be concluded that respondents in this study, both household heads (husbands) and wives, generally fall within the productive age range. However, there are differences in the tendency of decrease in working hours between them. Among household heads, the decrease occurs more in older age groups (over 57 years). In contrast, among wives, the decrease is more dominant in the productive age group, particularly ages 25 - 44.

3.1.2 Results of Logistic Regression

Household working hour decreases include both head of household and wife. Household-level analysis uses BPNT proportion and household income proportion as observable variables. Individual-level analysis includes BPNT proportion, income of husband and wife, education, and age.

Table 6: Estimation Results Using the Binary Logit Regression Model (Households)

Source: Eviews Output Results (processed)

Table 7: Estimation Results Using the Binary Logit Regression Model (Head of Household)

Source: Eviews Output Results (processed)

Table 8: Estimation Results Using the Binary Logit Regression Model (Wife)

Source: Eviews Output Results (processed)

Based on the logistic regression results for households, heads of households, and wives, the results are as follows:

1.     DHWi(Y) = 5.130606 + 0.310625 (BPNT₁)* - 0.116960 (IncomeProp₂)*

2.     DHW_KK = 1.370508 – 0.013738 (BPNT₁)* - 5.12E-06 (IncomeKK₂)* - 0.095378 (EduKK_3D1)* + 0.053681 (AgeKK_4U1)*

3.     DHW_Istri = 2.856051 + 0.219618 (BPNT₁)* - 4.04E-06 (IncomeIstri₂)* - 0.002238 (EduIstri_3D2)* - 0.110613 (AgeIstri_4U2)*

Note: *: significant at the 5 percent level of the Z-test

Table 6: Summary of Results of the Effect of BPNT on the Decrease in Working Hours by Household, Head of Family, and Wife

Source: Eviews Output Results (processed)

3. 2 Discussion

3.2.1 BPNT Proportion

The BPNT proportion variable at the household level shows a positive and significant effect on the decrease in household working hours. This means that the higher the proportion of BPNT received, the greater the likelihood that households experience a decrease in working hours. This finding is consistent with Prifti et al. (2019), who found that the higher the proportion of assistance, the greater the percentage of households that decrease their working hours.

Furthermore, since the decrease in household working hours consists of the decrease in the working hours of the household head and the wife, an analysis of the household head shows that the BPNT proportion variable is not significant for the decrease in the household head’s working hours. In contrast, this variable has a positive and significant effect on the decrease in the wife’s working hours. These findings indicate that BPNT does not significantly influence the household head’s decision to reduce working hours. As the primary earner, the household head tends to maintain work intensity due to economic orientation and family financial responsibilities. Additionally, the BPNT amount received is relatively small and insufficient to replace part of the income from labor. Thus, even with social assistance, the household head continues to work at the same intensity to maintain family economic stability.

Overall, at the household level, the decrease in working hours primarily occurs among wives. This finding aligns with Alzua, Cruces, & Ripani (2013), who analyzed the PROGRESA program in Mexico and found that social assistance led to a significant decrease in female labor participation. Teixeira (2010) also reported similar findings, indicating that the decrease in working hours is more dominant among female beneficiaries.

3.2.2 Household Income Proportion, Head of Household Income, and Wife’s Income

The household income proportion variable shows a negative and significant effect on the decrease in working hours. Household head income and wife’s income also exhibit a negative and significant effect. This indicates that both household heads and wives with lower income have a higher likelihood of decreasing working hours compared to households with higher income.

In this context, although the opportunity changes are relatively small, beneficiary households tend to adjust their working hours when household economic burdens are slightly eased through social assistance. However, since the BPNT amount is not substantial, the reduction in working hours is not extreme but rather represents flexible adjustments. These results are consistent with Salsabila and Purwanti (2020), who found that individuals earning lower wages reduce labor participation when receiving government assistance.

3.2.3 Education

The household head’s education variable does not significantly affect the decrease in working hours. Most household heads in Dukuhturi Sub-district work in the informal sector, such as farm laborers, construction workers, or daily wage jobs, where income depends on the number of working days or output. In this context, education is not a primary factor in decisions regarding working hours. This finding aligns with Rasyid (2012), who stated that education level does not significantly affect the labor supply decisions of assistance recipients.

On the other hand, the wife’s education shows a negative and significant effect on the decrease in working hours. This supports Vere (2011), who found that individuals with lower education levels are more responsive to government social protection programs. In other words, the lower a person’s education level, the higher the likelihood that they adjust their economic behavior in response to received assistance.

3.2.4 Age

The household head’s age variable does not significantly affect the decrease in working hours. Most household heads in Dukuhturi Sub-district are in the productive age range (30–55 years) with high economic responsibilities, so they maintain their working hours even when receiving social assistance. In contrast, the wife’s age shows a negative and significant effect. This means that as the wife’s age increases, the likelihood of reducing working hours decreases. This finding differs from Rasyid (2012), who stated that older individuals are more likely to reduce working hours.

The decrease in working hours among younger wives can be explained by the age characteristics of BPNT recipients in Dukuhturi Sub-district. Field observations indicate that most recipients in this area are young women with young children, which leads them to reduce their working hours. Meanwhile, older women generally have adult children and lighter caregiving responsibilities, so they maintain their usual working hours without being significantly influenced by the assistance received.

4. Conclusion

Based on the findings from households receiving the Non-Cash Food Assistance Program (BPNT) in Dukuhturi Sub-district, several conclusions can be drawn regarding the influence of socio-economic characteristics on the decrease in working hours at the household, household head, and wife levels. At the overall household level, the BPNT proportion has a positive and significant effect on the decrease in household working hours, whereas the household income proportion has a negative and significant effect on the decrease in working hours.

For household heads, income has a negative and significant effect on the decrease in the household head’s working hours. Meanwhile, the variables of BPNT proportion, education, and age of the household head do not show significant effects. This is due to the household head’s obligation to continue working, as the BPNT amount received is relatively small and insufficient to meet the household’s total needs without employment. Additionally, the generally low economic conditions of beneficiary households lead household heads to maintain their work activities regardless of education level or age.

For wives, the BPNT proportion has a positive and significant effect on the decrease in working hours, while income and age have negative and significant effects. Wife’s education, however, does not have a significant effect because most wives have relatively low education levels, which limits their influence on work decisions.

Overall, the BPNT program has been shown to decrease household working hours. This reduction primarily occurs in households where the BPNT assistance proportion is relatively large compared to total household expenditures. The assistance encourages wives to reduce working hours and allocate more time to other non-work activities. Meanwhile, for household heads, BPNT does not have a significant effect on working hours, as they continue to feel responsible for meeting the household’s overall needs.

For future research, it is recommended to broaden the household characteristics studied. This should not only focus on households consisting of a household head (husband) and wife, but also include other household types such as single-person households, households without a wife, or households with other family members involved in productive activities. In addition, future studies should consider adding other independent variables, such as the number of dependents, as these factors may influence household work decisions.

Author Contributions: All authors contributed to this research.

Funding: Not applicable.

Conflicts of Interest: The authors declare no conflict of interest.

Informed Consent Statement/Ethics approval: All respondents provided informed consent before data collection. This study adhered to Diponegoro University's research ethics guidelines.

Declaration of Generative AI and AI-assisted Technologies: This study has not used any generative AI tools or technologies in the preparation of this manuscript.