Bucking the time-honored tradition of “leaving the nest,” an astounding 68% of young Gen Z adults are choosing familial confines as opposed to independent living, thus weaving a complex narrative of economic, social and familial ties in modern America. What’s more, Millennials aren’t far behind with 20% struggling with leaving the nest and still residing in their parental homes. But, what does this mean for our society, economy and the individuals involved? Digging into the data reveals a tale of generational adaptation, financial pragmatism, and evolving familial structures that are reshaping the American Dream.

HIV/AIDS has been one of the largest public health crises in recent history, and the US federal government has spent hundreds of billions of dollars fighting the disease. This study examines the impact of the large amounts of federal funding allocated to US cities to combat HIV/AIDS through the Ryan White CARE Act's first title. The findings indicate that the cost to avoid an HIV/AIDS death through the program is roughly $334,000, that the program has saved approximately 57,000 lives through 2018, and that funding disparities are responsible for the uneven progress in combating HIV/AIDS across the United States

This case study shows how GIS and Expert Judgment can be used to develop small area population forecasts in the United States. It starts by organizing 2000 and 2010 block group population data into the 2020 block group geography and then examines 2020 indicators used to evaluate the effect of Differential Privacy on the 2020 population data. Preliminary population projections to 2050 are then generated by averaging the results of three standard small area projection methods. Using local expert judgment, GIS overlay maps and satellite imagery in a virtual environment, the 301 block groups of Greenville County, South Carolina were classified into seven categories of future population change. These categories were then applied to the preliminary projections to generate informed forecasts. Following this step, the sums of the BG results were then compared, respectively, to independently generated county population forecasts for 2030, 2040, and 2050. At this point, 25 BGs were selected for additional review, which resulted in a final set of forecasts. We find that the increase of 152,840 people in the year 2050 spread over all of the 301 census block groups in going from the preliminary projections (675,626) to the final informed forecasts (828,467) is largely generated by these same 25 BGs, which expert judgment determined were currently poised to “take off” in terms of population growth. Having this much change generated by such a small number of BGs is consistent with findings elsewhere.

This paper examines the impact of gentrification on criminal activity in urban neighborhoods to determine whether this process has a detrimental effect on communities. The study utilizes a newly-built unique data set of geo-referenced crime reports from 14 major American cities matched with Census data, to identify gentrified areas in the 2010s. To causally evaulate the impact of gentrification on crime I adopt state of the art event study models to causally evaluate the effects of gentrification, taking into account variations in the timing of this process across different cities and neighborhoods. The analysis reveals that gentrified areas experienced a statistically significant increase in crime ranging from 11 to 17%. The findings suggest that gentrification has a negative impact on neighborhoods, with property crimes showing the most significant increases. Overall, the study suggests that gentrification may have a criminogenic effect on neighborhoods, highlighting the need for further research and policy attention to this issue.

Estimating pedestrian exposure for all the intersections in a jurisdiction is crucial for developing strategies with a focus on pe- destrians. Some engineering applications require the annual average daily pedestrian traffic (AADPT) to be disaggregated per crosswalk. When continuous counts are available at the intersection, this indicator can be calculated directly. However, when only short-term counts (STCs) or no information on pedestrian volume is available, the AADPT per crosswalk cannot be calculated and must be estimated using other means. This work (1) evaluated the degree of confidence for estimating the pedestrian volume in each crosswalk based on point es- timates of percentage shares per crosswalk obtained from STCs; and (2) developed models to estimate the percentage share of pedestrian volume per crosswalk as a function of attributes of the intersection that commonly are available for jurisdictions, referred to as the land-use (LU) model. The two methods were evaluated using continuous count data from three different jurisdictions, and a naive estimate assuming equal shares per crosswalk was used as a benchmark. The performance of each method was measured as the fraction of the intersection AADPT that was allocated wrongly to each crosswalk. The use of the LU model generated an average wrong allocation of 0.301, a statisti- cally significant improvement of 11.4% compared with the naive estimate. The use of a STC from a single day produced an average wrong allocation of 0.153, an improvement of 54.9% from the naive estimate. Increasing the number of days of STCs to two or three resulted in average performance indicators of 0.117 and 0.106, respectively. The benefits of using STCs for more than three days are minimal. The STC method was developed using STCs from the same 1-year period in which the observed share was averaged. In practice, STCs are likely to be between 1 and 5 years old. Analysis using STCs from previous years showed that estimation error in practice may be as much as twice as large as the aforementioned errors

Veterans exit military service with varying degrees of service-connected disabilities (SCD). The GI Bill provides educational benefits, which increased substantially in 2009 (‘Post-9/11 GI Bill’). Exploiting the exogeneity of SCD and using a difference-in-difference approach, we find that SCD veterans are 16.2 percent more likely to attend college than non-SCD veterans due to the Post-9/11 GI Bill, an effect driven by lower-level SCD veterans attending public colleges. After the benefit increase many lowerlevel SCD veterans switch from being employed to attending college. We provide insights into the disabled veterans’ college-employment tradeoffs and find that the benefit likely helps disabled veterans improve their labor market outcomes. Future changes in the distribution of SCD levels among veterans will translate into changes in the demand for higher education. Also, the high responsiveness to a more generous financial aid for higher education among disabled veterans may provide useful insights into the effective design of similar subsidies for civilian disabled populations.

Among the U.S. health care workforce, the COVID-19 pandemic appeared to greatly impact employment levels in 2020. However, no research has examined how the pandemic’s impact on employment varied by racial/ethnic group or beyond the initial emergency year. Our study aimed to quantitatively evaluate workforce trends by race/ethnicity before, during, and after the COVID-19 pandemic. This study analyzed each March supplement of the Current Population Survey over a 5-year span (2018– 2022). We restricted the sample to nurses, physician assistants, and other non-physician health care workers (HCW), per specific census occupation codes, and constructed an event-history study to test for differential effects from each year, as compared to 2019, on the proportion of employment between non-Hispanic White, non-Hispanic Black, Hispanic, non-Hispanic Native (American Indian, Alaska Native, Hawaiian Islander), and non-Hispanic Asian HCW. Results suggest that the pandemic’s negative impact on the health care workforce disproportionately reduced employment for HCW selfidentifying as Black or Indigenous. Rates for other groups increased 2–3 percentage points in 2020 but returned to prepandemic levels by 2022. However, for Black and Native HCW, the change was twice as large in 2021 and remained significantly higher in 2022 for Black HCW, providing more evidence that the burden of the COVID-19 pandemic disproportionately fell on people of color. Future research investigating how employment disruptions impacted the health care workforce and, potentially, health equity remains warranted.

The southeastern United States has experienced rapid economic and population growth relative to the stagnation and decline of established manufacturing regions. The Rust Belt–Sun Belt literature has relied on a range of factors to explain southern growth (e.g., low wages, regional amenities, technological advances, market proximity, and the Civil Rights Movement). More recent work has cast growth in terms of bottom–up convergence and providing a conceptual framework consistent with the region’s experience as the recipient of northern capital seeking a low-cost region. Within the region, uneven development remains. While the neoclassical framework can explain regional capital inflows, questions remain regarding how local and spatial contexts influence the trajectories of individual southern economies. To address these issues, this paper utilizes a conditional convergence model to explain southeastern county-level growth from 1990 to 2018 to understand how the balance of neoclassical pro-cesses and localized factors have affected the region’s growth. Results highlight a transition away from low-cost manufacturing toward an increased importance of new, knowledge economy sectors. While new economy sectors are identified as significant drivers, results also suggest future challenges for sustained growth due to a suburbanized growth pattern, weak spillover effects, and a lack of urbanization economies.

Borrowing money is a core instrument of governments to fund goods with high front costs and long-term benefits. Scholars have, however, primarily associated public debt with shortsighted policies by office-seeking politicians. The three essays in this dissertation investigate the determi- nants and outcomes of popular preferences for investment-oriented public debt using novel voting, survey, and budgetary data. The first essay asks: Is a community more amenable to borrowing when its time horizon short- ens? Existing theories argue that individuals with shorter time horizons, like seniors, have a higher inclination towards borrowing because they overvalue current consumption and discount future costs. I verify this assumption by studying how population aging affects support for debt-funded investments. Using novel data sets on U.S. state and local bond referendums over six decades, I show that, conversely, aging decreases support for debt-funded investments. Contrary to main- stream predictions, an original conjoint survey experiment further demonstrates that seniors do not have a greater preference for policies with longer repayment maturities and shorter benefit peri- ods. Rather, aging lowers support for investments by increasing fiscal conservatism and shifting consumption away from capital-intensive goods. The effect of aging varies depending on which age groups cohabit with seniors. In particular, aging communities experiencing an influx of non- relative children show greater opposition to new investments, while increased contact with relative children has no detectable effect on their support for investments. These findings suggest that pop- ulation aging can complicate the construction of political coalitions over investments, particularly in communities with diverse age distributions. The second essay inquires: Do popular preferences affect how governments fund policies? Pol- icy funding is often presented as technical and hardly influenced by voters. I study this assumption by investigating the effect of population aging on U.S. municipal budgets between 1970 and 2017 with the use of data on municipal finances and mayors’ characteristics. In contrast, I find that aging increases appetite for consumption-oriented policies, leading to more short-term budgeting. When a municipality’s population ages, it substitutes current expenditures for capital spending, shortens its debt maturity, and favors liquid revenues over long-term borrowing. Ultimately, this translates into lower indebtedness and higher property tax revenues. In contrast, its expenditure levels and distribution between policies remain stable as seniors’ fiscal conservatism constrains surges in spending, and seniors’ interest in property values limits cuts in municipal amenities. The effects of aging are not uniform, as municipalities ruled by elderly mayors implement debt policies more aligned with seniors’ preferences. These results contradict the dire budgetary predictions associated with aging and show that seniors’ ideological and economic motives can counterbal- ance their distributional demands. They also illustrate that the preferences of minorities are better represented when they elect politicians who resemble them. The last essay questions: Do voters care about how a policy is funded? Even if citizens can grasp the technicalities of public finances, policy funding may still not matter to them. Indeed, the Ricardian equivalence argues that people are indifferent about whether a policy is funded by debt or taxation because they internalize the future costs of debt repayment in their bequests. Using a novel dataset of 22,000 local referendums and two original conjoint survey experiments, I demonstrate that, conversely, voters prefer financing policies by small tax increases rather than borrowing. My surveys also reveal that respondents discriminate against policies with longer repayment periods. This result contradicts both the Ricardian equivalence’s assumption that people are indifferent to how policies allocate costs over time and the premise that people overlook future borrowing costs. Time preferences are important to explain opinions regarding debt and taxation, as each funding method distributes costs and benefits differently over time. Specifically, people’s resistance to long repayment periods lowers support for debt-funded projects. Variations in preferences between debt and tax remain after accounting for their temporal differences. My analyses indicate that preferences do not vary by policy content or relative to personal financial investments, although conservative individuals display greater support for borrowing than liberals

Nitrogen dioxide (NO2) is a regulated pollutant that is associated with numerous health impacts. Recent advances in epidemiology indicate high confidence linking NO2 exposure with increased mortality, an association that recent studies suggest persists even at concentrations below regulatory thresholds. While large disparities in NO2 exposure among population subgroups have been reported, U.S. NO2-attributable mortality rates and their disparities remain unquantified. Here we provide the first estimate of NO2-attributable all-cause mortality across the contiguous U.S. (CONUS) at the census tract-level. We leverage fine-scale, satellite-informed, land use regression model NO2 concentrations and census tract-level baseline mortality data to characterize the associated disparities among different racial/ethnic subgroups. Across CONUS, we estimate that the NO2-attributable all-cause mortality is ∼170,850 (95% confidence interval: 43,970, 251,330) premature deaths yr–1 with large variability across census tracts and within individual cities. Additionally, we find that higher NO2 concentrations and underlying susceptibilities for predominately Black communities lead to NO2-attributable mortality rates that are ∼47% higher compared to CONUS-wide average rates. Our results highlight the substantial U.S. NO2 mortality burden, particularly in marginalized communities, and motivate adoption of more stringent standards to protect public health.

The United States’ Affordable Care Act (ACA) of 2010 improved and expanded availability of non-group health insurance. Previous studies have shown that women in the US workforce value health insurance more highly than men do. Because prior to the ACA self-employed individuals did not have guaranteed access to affordable health insurance coverage, women’s relatively lower rate of self-employment may partly have reflected their greater “job lock” due to employer-based health insurance. This article employs nationally representative survey data for 2009–18 and a quasi-experimental difference-in-difference modeling approach and finds that unmarried women’s probability of self- employment increased by 1.2 percentage points in 2015–18, after the ACA’s expansion of non-group health insurance came into effect. Among women who have never married, overall probability of self-employment increased by 1.2–1.5 percentage points versus trend, and the probability of transitioning into full-time self-employment increased by 0.9 percentage points.

Tap water lead testing programs in the U.S. need improved methods for identifying high-risk facilities to optimize limited resources. In this study, machine-learned Bayesian network (BN) models were used to predict building-wide water lead risk in over 4,000 child care facilities in North Carolina according to maximum and 90th percentile lead levels from water lead concentrations at 22,943 taps. The performance of the BN models was compared to common alternative risk factors, or heuristics, used to inform water lead testing programs among child care facilities including building age, water source, and Head Start program status. The BN models identified a range of variables associated with building-wide water lead, with facilities that serve low-income families, rely on groundwater, and have more taps exhibiting greater risk. Models predicting the probability of a single tap exceeding each target concentration performed better than models predicting facilities with clustered high-risk taps. The BN models’ Fβ-scores outperformed each of the alternative heuristics by 118–213%. This represents up to a 60% increase in the number of high-risk facilities that could be identified and up to a 49% decrease in the number of samples that would need to be collected by using BN model-informed sampling compared to using simple heuristics. Overall, this study demonstrates the value of machine-learning approaches for identifying high water lead risk that could improve lead testing programs nationwide.

Environmental migration studies have gained popularity in the social sciences as environmental change has drastically changed migration patterns around the globe. While previous studies tended to focus on the environmental impacts on migration in the developing world, less attention has been paid to examining the relationship between slow-onset environmental variability and migration in developed settings. To address this knowledge gap, we linked individual migration data with previous climate exposure and other individual- and aggregate-level socioeconomic and demographic factors to explore the effects of slow-onset environmental variability on migration in the U.S. using multi-level regressions. The results demonstrated that slow-onset environmental variability influences migration; their impacts are heterogeneous across age groups with the younger generation’s migratory response to environmental variability being connected to economic opportunities and the elder generation’s being amenity-driven. This study contributes to the environmental literature by extending and validating slow-onset environmental impacts on migration in an economically developed setting.

Research on the division of housework among same-sex partners is limited. This is because gender-cultural theories – which emphasize the significance of gender identity and motivate many studies on the topic – are implicitly assumed to be less relevant in this case. Attending to admonitions that the division of housework in same-sex households is not free from gendering processes and practices, in this study we use the high-quality data of the American Time Use Survey (ATUS; 2003-2019), to compare the association between housework and relative earnings across partnership types. Since gender-cultural theories are based on the performance of gender identity by different-sex partners, we utilize the differences between same- and different-sex partners to better understand the effect of gender-cultural determinants on the division of housework. Our comparison of the relation across partnership-types validates the power of gender-cultural mechanisms in different-sex partners, provides a better assessment of the differences in housework patterns between different types of households, and serves as a novel quantitative test of gender-cultural mechanisms in same-sex partners.

Climate change is an urgent crisis. Currently more than one-quarter of carbon emissions in Massachusetts come from the building sector, primarily through on-site combustion of oil and methane for the heating of residential and commercial buildings. Achieving net zero carbon emissions in Massachusetts will require achieving net zero carbon emissions from buildings statewide, both by creating more energyefficient building envelopes and by decarbonizing heating, cooling, and other appliances. At the same time, Massachusetts residents are experiencing a housing cost crisis that disproportionately and negatively affects low- and moderate-income households. Roughly one out of every two renter households (49 percent) and nearly one out of every three (30 percent) homeowner households with a mortgage in Massachusetts were housing cost burdened in 2021, meaning that they were paying 30 percent or more of their income towards housing costs. Today, it is more expensive to build most energy-efficient homes than it is to build conventional homes. Increased construction costs lead to increases in home prices and decreases in housing production. This report identifies a range of effective policy tools that policymakers can use to meet critical goals for climate emissions, housing production, and housing affordability. This report analyzes and answers three questions: 1. How does the cost of building homes that are compliant with Massachusetts’ new municipal optin specialized stretch energy code compare to the cost of building similar homes that are compliant with the current stretch code? 2. Based on these cost estimates, how is the opt-in specialized stretch energy code likely to affect housing affordability in Massachusetts? 3. How can public policies simultaneously advance the transition to net-zero housing while also addressing the housing affordability crisis? Models of single-family home and townhouse construction and surveys of builders of single and multi-family housing indicate that the specialized stretch energy code is likely to increase the cost of construction of single family homes and townhouses by roughly 1.8 to 3.8 percent (approximately $10,000 to $23,000 for the median single family home), depending on the pathway to compliance selected, and increase the cost of construction of large multi-family buildings by roughly 2.4 percent, at least initially. It is important to interpret these numbers in the context of an evolving and diverse marketplace. Labor and material costs have been in flux; low-carbon technologies continue to advance; and builders themselves are still working out the least expensive routes to compliance for diverse projects on diverse sites. Several large non-profit builders of multi-family affordable housing noted that as developers and contractors have become more accustomed to Passive House construction, incremental costs have declined and will continue to decline. The economic modeling here shows that these increases in construction costs could push the median single-family home in Massachusetts out of reach for between 15,000 and 33,000 households, before taking into account any public financial incentives for green building. The increased energy efficiency of homes will potentially offset some of the increase in construction costs, but not all, as explored below. Some of the increased construction costs will be passed on to homebuyers and renters, while some will be absorbed by builders in the form of reduced profit margins. Increases in construction costs, however, are likely to reduce construction starts overall, as fewer projects pencil out and fewer households can afford new homes, putting further pressure on housing supply and affordability. We find that the increased costs will be felt most by households with low and moderate incomes. Policymakers have multiple tools to increase housing affordability while advancing the Commonwealth’s climate agenda. Since local land use regulations are currently a major driver of the high cost of housing, reform of local land use regulations to facilitate the construction of more housing units should accompany the strengthening of green building standards. Zoning reform can also improve the carbon efficiency of the state’s settlement patterns. Other tools to simultaneously improve the energy efficiency of buildings and housing affordability include efforts to streamline the permit process, to restructure financial incentives to keep them in place for longer periods of time, to increase technical assistance for green building, to facilitate use by realtors and appraisers of the Residential Green and Energy Efficient Addendum, to improve underwriting of energy-efficient homes, to expand financing sources through the new Massachusetts Community Climate Bank, to increase support for low-income renters, and to consider new tax classifications and exemptions for highly energy-efficient buildings. One of the most powerful tools to simultaneously increase housing affordability and decrease carbon emissions is land use policy. Current land use regulations prevalent in municipalities across the Commonwealth require carbon intensive living patterns and drive up the cost of housing. Some of the municipalities adopting the specialized stretch energy code have extensive large-lot, large-unit, singlefamily land use regulations that mandate energy intensive, environmentally destructive, and unaffordable living patterns. Legislative action tying the adoption of the opt-in specialized stretch code to land use regulations enabling dense and pedestrian-oriented land use patterns, such as smaller minimum lot and unit sizes, larger height limits, and more by-right multi-family zoning could increase the reduction in carbon emissions and simultaneously increase affordability. State legislative action is urgently needed to address local zoning barriers to energy-efficient and affordable housing production. State and local permitting requirements add substantial time, cost, and uncertainty to current home construction. State efforts to streamline these permit requirements could enable the construction of more affordable energy-efficient homes. Financial incentives for more energy-efficient construction are an important tool. An obstacle to the effectiveness of some existing incentive programs is lack of certainty about their duration. Explicit commitments by the state as to the durability of these incentives would enable home builders to rely on them in their financing models. The Massachusetts Commission on Clean Heat recently proposed the creation of a Building Decarbonization Clearinghouse to be a single point of contact for all decarbonization programs. The creation of such a clearinghouse as a single point of contact, with a commitment to longterm incentive availability, streamlined application processes, and adjustment of incentives for inflation, would be a beneficial step to advancing housing production, affordability, and decarbonization. Increased technical assistance and training to small home-builders and subcontractors regarding how to comply with the new specialized stretch energy code would be beneficial. The Massachusetts Clean Energy Center’s Passive House Design Challenge and its support for large multi-family builders beginning to build to Passive House standards could serve as a model for smaller builders beginning to comply with or exceed the specialized stretch energy code. Realtors not only connect buyers and sellers but can play a role in educating buyers and sellers about the value of energy-efficient homes, while appraisers are essential to the underwriting process. Increased use of the Residential Green and Energy Efficient Addendum is necessary to increase home purchasers’ awareness of the increased value and decreased utility costs of energy-efficient homes and also to accurately underwrite these homes. Federal action to ensure that energy-efficient improvements and associated utility cost savings are capitalized into property appraisals and that those appraisals will be accepted by the government-sponsored enterprises (GSEs) would simplify the financing of more energy-efficient homes for home buyers. In the interim, it would be helpful for the state to encourage Massachusetts financial community development financial institutions to offer energy-efficient mortgages that accurately underwrite energy-efficient homes and their associated cost savings. The new Massachusetts Community Climate Bank can facilitate investment in low-carbon, climateresilient housing. A mission-driven financial institution designed to use public funding to attract private investment to clean energy projects through loans or investment funds, Green Banks can be an important new tool in reducing building carbon emissions and simultaneously increasing affordability. To the extent that some energy-efficient multi-family buildings may shift some utility costs from landlords to tenants, support for low-income tenants is important to ensure that vulnerable households in the Commonwealth do not disproportionately bear the burden of decarbonization. The combination of the current climate change and housing affordability crises requires bold action. The Commonwealth’s Global Warming Solutions Act of 2008 and the Act Creating a Next Generation Roadmap for Massachusetts Climate Policy of 2021 are fundamental steps to reduce carbon emissions in Massachusetts but will not, alone, reduce carbon emissions from new residential buildings to zero. Housing Choice legislation and the MBTA Communities Act are helpful steps towards reducing some of the obstacles that local land use regulations create to building more energy-efficient and affordable housing, but it falls far short of the changes to land use needed to meet current housing needs and far behind more ambitious legislation other states, such as Oregon, California, and Maine, have recently passed. In sum, action by the state leg…

While scale is an essential factor in discussions about sustainable cities, there is no common understanding of what scale is or how it should be measured. This paper sheds light on the issue of scale by suggesting how it might be measured and evaluated. My purpose is to offer both a methodological and empirical contribution to the understanding of scale, using Chicago as a case study. Using historical Sanborn maps, I first investigate scale change over time, focusing on a selection of 31 sites that are now “mega-developments” but were originally composed of small-scale buildings and blocks. I quantify that the historical urban fabric had five times as many buildings, and a much higher percentage of buildings with mixed use. I then look at the city as a whole and compare urban scale to pedestrian quality to assess whether there is a quantifiable difference between large- and small-scale urbanism. I find that, at least for Chicago, small scale urbanism is associated with higher pedestrian quality. For the third part of the analysis, I correlate scale and socio-economic characteristics at the census tract level. The results illuminate a mixed set of differences between scale and socio-economic characteristics like income and housing value.

Despite infrastructure’s importance to the US economy, evidence on its cost trajectory over time is sparse. We document real spending per new mile over the history of the Interstate Highway System. We find that spending per mile increased more than threefold from the 1960s to the 1980s. This increase persists even conditional on pre-existing observable geographic cost determinants. We then provide suggestive evidence on why. Input prices explain little of the increase. Statistically, changes in income and housing prices explain about half of the increase. We find suggestive evidence that the rise of “citizen voice” in government decision-making increased spending per mile.

U.S. HIV diagnoses disproportionately affect Non-Hispanic Black (NHB) and Hispanic gay and bisexual men. Using data from the National Health Interview Survey (2013-2018), we examined race and ethnicity and primary care access, an HIV prevention resource, among gay and bisexual men. The explanatory variable was NHB, Hispanic or Non-Hispanic White (NHW). Outcomes were primary care-specific usual place of care (potential access) and saw general doctor <12 months (realized access). We used multivariable logistic regression, adjusting for individual sociodemographic characteristics, health status, and care barriers. In sensitivity analysis, we examined general access (any place/doctor) and subgroups (1) NHB (2) has usual place of care. The sample included 1,858 adult, gay and bisexual men (unweighted). Nearly one- third self-identified as NHB or Hispanic. Compared to NHW men, NHB and Hispanic men were younger, with lower household income, and more care barriers (p < 0.05). NHB and Hispanic men had lower realized access (aOR 0.7058, p = 0.030) than NHW men. Potential access was lower for NHB only (versus NHW) and, among those with any usual place of care, NHB and Hispanic men versus NHW men. Lower primary care access for NHB or Hispanic, rather than NHW, gay and bisexual men, may reduce HIV prevention access.

The abundance of data, often containing private and sensitive information, coupled with an ever-growing interest in using the data for research and driving business value at scale has raised concerns about privacy protections. Formal policies have made access to such data heavily regulated, often resulting in users waiting months or years before they can even start analyzing the data to determine fit for their tasks. In the recent past, generation of synthetic data has emerged as an alternative. A synthetic dataset is a collection of records that is designed to capture structural and statistical properties in the sensitive dataset while ensuring that private properties of individuals are not revealed. Synthetic data generation, especially under Differential Privacy, is gaining popularity for making copies of sensitive data for testing, benchmarking, data analysis, training models and privacy preservation.

The skyline is an optimization operator widely used for multi-criteria decision making. It allows minimizing an n-dimensional dataset into its smallest subset. In this work we present SkyFlow, the first heterogeneous CPU+GPU graph-based engine for skyline computation on a stream of data queries. Two data flow approaches, Coarse-grained and Fine-grained, have been proposed for different streaming scenarios. Coarse-grained aims to keep in parallel the computation of two queries using a hybrid solution with two state-of-the-art skyline algorithms: one optimized for CPU and another for GPU. We also propose a model to estimate at runtime the computation time of any arriving data query. This estimation is used by a heuristic to schedule the data query on the device queue in which it will finish earlier. On the other hand, Fine-grained splits one query computation between CPU and GPU. An experimental evaluation using as target architecture a heterogeneous system comprised of a multicore CPU and an integrated GPU for different streaming scenarios and datasets, reveals that our heterogeneous CPU+GPU approaches always outperform previous only-CPU and only-GPU state-of-the-art implementations up to 6.86×and 5.19×, respectively, and they fall below 6% of ideal peak performance at most. We also evaluate Coarse-grained vs Fine-Grained finding that each approach is better suited to different streaming scenarios.