Investigating the Impact of Socioeconomic Factors on Poverty Entry in the US

Background/Purpose: A struggle against poverty has been still critical in U.S. politics and public policy, regardless of the controversial legacy of President Johnson’s War on Poverty and its impacts. A myriad of previous research on poverty has revolved around two main strands: cross-sectional measures of poverty (i.e. poverty rates) and time-series trajectories of poverty with related covariates. Given the precarious nature of work and resulting income volatility, policy demands of the study on poverty dynamics have continued to grow because the static poverty research cannot provide a comprehensive and dynamic perspective of poverty. Thus, the purpose of this study is to examine poverty entry over time and its relations with socioeconomic factors in the U.S. 

Methods: Using the Panel Study of Income Dynamics (1999-2011) data, this study conducted a piecewise complementary log-log model to examine the relationship between poverty entry and socioeconomic factors such as age, gender, race, education, having healthcare coverage, childcare burden as measured by the number of children, single parenthood, presence of family composition change, and region. The advantage of using the complimentary log-log model was that it yields the odds ratio coefficients as the effects of covariates on relative risk or hazard of poverty. For this analysis, the 1999-2011 cross-year individual data merged with corresponding year’s family files was used. The analytic sample was 11,152 individuals aged 18-64 between 1999 and 2011. In terms of poverty status, individuals at each time were determined as poor if their annual equivalized income was below the relative poverty threshold, as measured by 60% of the median income per capita. 

Results: Descriptive results showed that 32.4% out of the sample were transient poor (experiencing poverty only once) whereas 7.9% were chronic poor (experiencing poverty consecutively) in the 1999-2011 study period. For model selection, AIC value was used for the goodness-of-fit test. The lowest AIC of our unrestricted model (13199.89) suggested that it was more improved compared to other five restricted models. Hazard analysis of poverty entry indicated the predicted probabilities of poverty by each predictor. The effect of single parenthood on hazard of poverty was the strongest (OR = 2.91; p <.01). Childcare burden was positively related to risk of poverty (OR = 1.25; p <.01). Individuals living in South region were higher in hazard of poverty by about 8% compared to other regions (OR = 1.08; p <.01). However, higher education (OR = 0.83; p <.01) and having health insurance (OR = 0.47; p <.01) was negatively related to hazard of poverty. 

Conclusions/Implications: Findings provide insight and detailed information about the effects of socioeconomic factors on hazard of poverty. Findings suggest that single parenthood is strongly related to the risk of poverty controlling for other factors. In addition, nonwhites tend to have higher risk of poverty compared to whites by about 66%. This study has implications for policymakers to gain insight on identifying which combination of factors would result in the heightened risk of entering into poverty for low-income families in the long term.