Unraveling the effects of a rehabilitation program on the socioeconomic wellbeing of beggars and begging motivation: evidence from an urban area of Bangladesh

Beggars in Bangladesh are socially excluded and face chronic poverty, marginalization, and adverse health and safety risks, challenges that intensified during COVID-19. Despite policy attention through safety nets and sporadic rehabilitation efforts, robust evidence on effectiveness remains limited. This study examines a government-led rehabilitation program in Khulna City Corporation (KCC) designed to improve socioeconomic wellbeing and provide alternative livelihoods to beggars. The research addresses two questions: (1) Does the rehabilitation program enhance socioeconomic wellbeing (income, expenditure, food security, and personal wellbeing) of beggars? (2) To what extent does the program reduce begging motivation? Given the importance of inclusivity under SDGs and the dearth of rigorous evaluation for this vulnerable group, the study’s purpose is to assess whether the intervention yields measurable improvements and deters continued begging.

Prior studies identify multiple drivers of beggary including poverty, lack of jobs, illiteracy, sickness, social norms (e.g., polygamy), migration, and shocks (Khan et al. 2016; Sobhani and Murtaz 2015; Jeffreys and Wang 2012). Children are often involved due to extreme poverty, parental coercion, trafficking, and strategic appeals to public sympathy (Kamruzzaman and Hakim 2015; Owu-Sekyere et al. 2018). In Addis Ababa, begging can function as a household livelihood strategy (Abebe 2008). COVID-19 exacerbated vulnerabilities, pushing more people, including children, into begging and deepening food insecurity (Susan 2021; Sukmawati and Prasatyo 2021; Wiseman et al. 2021). Safety nets and Zakah can alleviate hardship, but beneficiary selection challenges, resource constraints, and food-focused spending often limit sustainable poverty reduction (Haider and Mahumud 2017; Zibran et al. 2014; Billah and Alam 2017). In Indonesia, multi-level institutional coordination is emphasized for effective rehabilitation (Wismayanti et al. 2021), yet a persistent gap exists between program design and execution, including weak monitoring and coverage. Overall, the literature suggests that while interventions may offer temporary relief, structural and implementation deficiencies frequently constrain long-term improvements in beggars’ socioeconomic outcomes.

Design and setting: A cross-sectional study was conducted in mid-2022 in Khulna City Corporation (KCC), Bangladesh, an urban area with a sizable slum and floating population. Sampling and data: Using a semi-structured interview schedule (SSIS) built from literature, data were collected from 385 beggars: 59 beneficiaries of the KCC rehabilitation program (treatment) and 326 non-beneficiaries (control). Beneficiaries were sampled randomly from an official list (120 listed; 81 located; 59 eligible). Non-beneficiaries were sampled purposively across 17 city locations due to the absence of a census list, with eligibility criteria of age ≥18 and begging in at least three areas. Intervention description: The program provided livelihood-supportive items in three cohorts: vans (35), sewing machines (37), and tea stalls (48). Some beneficiaries were unavailable or ineligible due to inability to report required information. Quasi-experimental approach: A non-equivalent post-test group design was adopted given infeasibility of randomized assignment and lack of follow-up. Beneficiaries formed the treatment group; non-beneficiaries formed the control group. Groups differed in characteristics (e.g., age, experience, allowance receipt). Econometric strategy: To estimate program effects while addressing endogeneity, the study used multivariate regressions and instrumental variable (IV) methods. Primary outcomes: economic wellbeing (ln household per capita income; ln household per capita expenditure), food security (household dietary diversity, HDD), and social wellbeing (Personal Wellbeing Index, PWI). Additional outcome: begging motivation (binary preference for begging vs other occupations). Measures: HDD captured prior-day consumption across 12 food groups; scores range 0–12 (higher indicates greater dietary diversity). PWI assessed satisfaction across seven domains (standard of living, health, achievements, relationships, safety, community connectedness, future security) on 0–10 scales, aggregated to 0–100. Modeling: For income, expenditure, and HDD, OLS/Tobit specifications controlled for socio-demographics (D), COVID-19-related factors, and livelihood variables (L). PWI was modeled via Tobit and IV Tobit. The treatment indicator (Rehab) was instrumented with beggar type (street, doorstep, both), justified by arbitrary street-level beneficiary selection reported by implementers and the lack of established direct effects of beggar type on the outcomes. First-stage results are provided in Appendix A1; the Montiel-Pflueger robust weak instrument test indicated a weak instrument (Effective F = 3.410 < 29.71). Begging motivation was modeled via OLS and logistic regression. Income and expenditure were log-transformed for interpretation. Covariates included age, gender, schooling, religion, marital history, household size, earning members, family help, migration status, begging experience, debt, asset loss during COVID-19, special COVID allowances, and workless days during COVID-19.

Descriptive profile (N=385): mean age 58.7 years; 51% male; 77% with no formal education (mean schooling 0.88 years); mean household size 3.9; earning members 1.39. Average monthly household income BDT 7,833 and expenditure BDT 7,519; mean debt exceeded both income and expenditure. HDD averaged 4.52/12, indicating poor dietary diversity; PWI averaged 50.69/100, reflecting middling subjective wellbeing. About 70% had migrated; 5% reported another occupation besides begging; 33.8% received family financial support; 9% had other family members begging. During COVID-19, participants experienced on average 96 workless days; 81% received some in-kind aid; 21% reported being forced back into begging post-lockdown. Program coverage among sample: 15.3% (59/385). Earnings patterns by hours spent begging did not differ meaningfully between beneficiaries and non-beneficiaries. Multivariate regressions (Table 2): No significant associations between beneficiary status and ln per-capita income, HDD, or PWI. A weakly significant positive association with ln per-capita expenditure suggested beneficiaries might spend 15.7% more than non-beneficiaries (p<0.10), potentially reflecting sale or exchange of provided items rather than sustained livelihood gains. IV regressions (Table 3): Instrumenting beneficiary status with beggar type, the program showed no significant effects on ln per-capita income, ln per-capita expenditure, HDD, or PWI. The instrument was weak (Effective F=3.410). Predictors of wellbeing: Having an occupation besides begging was associated with ~42.8% higher per-capita income (p<0.05). Special COVID-19 allowance correlated positively with income (0.319, p<0.01) and expenditure (0.003–0.049, model-dependent). Additional marriages were associated with higher income (0.116, p<0.10), expenditure (0.189, p<0.05), and HDD (0.029, p<0.10). Debt was negatively associated with PWI (p<0.05). Male gender was positively associated with income/expenditure and HDD in several models. More workless days during COVID-19 was positively associated with expenditure, likely reflecting compensating transfers. Begging motivation (Table 4): Beneficiary status was not significantly associated with preferring begging over other occupations. Males were less inclined to beg than females (p<0.10). Greater asset loss during COVID-19 was negatively associated with preferring begging (p<0.05). Overall, the program did not reduce motivation to continue begging.

The study set out to test whether a city-level rehabilitation program could improve beggars’ socioeconomic wellbeing and reduce their motivation to beg. Across OLS/Tobit and IV specifications, the intervention showed no robust effects on income, expenditure, food security (HDD), or subjective wellbeing (PWI), nor did it reduce the preference for begging. These findings suggest a disconnect between the program’s intended aims and on-the-ground outcomes. Potential explanations include arbitrary selection of beneficiaries, misalignment between provided assets (e.g., vans, sewing machines, tea stalls) and recipients’ skills or constraints, lack of systematic monitoring and follow-up, insufficient coverage and resources, and limited oversight. The weak instrument cautions interpretation but does not contradict the overall null findings. The results underscore that marginal improvements (e.g., weakly higher expenditure among beneficiaries) likely reflect short-term coping (e.g., selling items) rather than sustainable livelihood shifts. Importantly, the analysis identifies factors associated with better outcomes, such as having an additional occupation, special allowances during crises, and household composition, while highlighting the detrimental effect of debt on wellbeing. These insights reinforce that effective rehabilitation requires tailored, sustained support, integrated with debt relief/management, skills matching, and continuous accompaniment rather than one-off asset transfers.

In Khulna City Corporation, the evaluated rehabilitation program did not generate significant improvements in beggars’ income, expenditure, food security, or personal wellbeing, nor did it reduce the motivation to continue begging. The evidence points to critical implementation gaps, including selection, targeting, alignment of support with needs and capacities, and monitoring. The study contributes a rare quasi-experimental assessment focused on beggars’ objective and subjective wellbeing. Policy recommendations include: conducting needs assessments and categorizing beneficiaries by constraints and skills; providing appropriate, viable livelihood options with training; instituting regular follow-up, monitoring, and evaluation; integrating motivational counseling to support transitions from begging; and considering complementary measures (e.g., debt relief, access to credit, social protection linkages). Future research should employ stronger identification strategies (e.g., randomized pilots or phased rollouts), longitudinal tracking to capture dynamics, and mixed-methods inquiry to understand behavioral and institutional mechanisms.

Key limitations include: (1) non-randomized, cross-sectional design limiting internal validity and causal inference; (2) weak instrument in the IV strategy; (3) lack of longitudinal follow-up to assess sustained effects and dynamics; (4) potential selection bias in non-beneficiary sampling due to purposive methods; and (5) limited external validity, with generalizability most applicable to similar urban contexts in developing countries rather than high-income settings.