-
Earnings Through the Stages: Using Tax Data to Test for Sources of Error in CPS ASEC Earnings and Inequality Measures
September 2024
Working Paper Number:
CES-24-52
In this paper, I explore the impact of generalized coverage error, item non-response bias, and measurement error on measures of earnings and earnings inequality in the CPS ASEC. I match addresses selected for the CPS ASEC to administrative data from 1040 tax returns. I then compare earnings statistics in the tax data for wage and salary earnings in samples corresponding to seven stages of the CPS ASEC survey production process. I also compare the statistics using the actual survey responses. The statistics I examine include mean earnings, the Gini coefficient, percentile earnings shares, and shares of the survey weight for a range of percentiles. I examine how the accuracy of the statistics calculated using the survey data is affected by including imputed responses for both those who did not respond to the full CPS ASEC and those who did not respond to the earnings question. I find that generalized coverage error and item nonresponse bias are dominated by measurement error, and that an important aspect of measurement error is households reporting no wage and salary earnings in the CPS ASEC when there are such earnings in the tax data. I find that the CPS ASEC sample misses earnings at the high end of the distribution from the initial selection stage and that the final survey weights exacerbate this.
View Full
Paper PDF
-
Citizenship Question Effects on Household Survey Response
June 2024
Working Paper Number:
CES-24-31
Several small-sample studies have predicted that a citizenship question in the 2020 Census would cause a large drop in self-response rates. In contrast, minimal effects were found in Poehler et al.'s (2020) analysis of the 2019 Census Test randomized controlled trial (RCT). We reconcile these findings by analyzing associations between characteristics about the addresses in the 2019 Census Test and their response behavior by linking to independently constructed administrative data. We find significant heterogeneity in sensitivity to the citizenship question among households containing Hispanics, naturalized citizens, and noncitizens. Response drops the most for households containing noncitizens ineligible for a Social Security number (SSN). It falls more for households with Latin American-born immigrants than those with immigrants from other countries. Response drops less for households with U.S.-born Hispanics than households with noncitizens from Latin America. Reductions in responsiveness occur not only through lower unit self-response rates, but also by increased household roster omissions and internet break-offs. The inclusion of a citizenship question increases the undercount of households with noncitizens. Households with noncitizens also have much higher citizenship question item nonresponse rates than those only containing citizens. The use of tract-level characteristics and significant heterogeneity among Hispanics, the foreign-born, and noncitizens help explain why the effects found by Poehler et al. were so small. Linking administrative microdata with the RCT data expands what we can learn from the RCT.
View Full
Paper PDF
-
The Icing on the Cake: The Effects of Monetary Incentives on Income Data Quality in the SIPP
January 2024
Working Paper Number:
CES-24-03
Accurate measurement of key income variables plays a crucial role in economic research and policy decision-making. However, the presence of item nonresponse and measurement error in survey data can cause biased estimates. These biases can subsequently lead to sub-optimal policy decisions and inefficient allocation of resources. While there have been various studies documenting item nonresponse and measurement error in economic data, there have not been many studies investigating interventions that could reduce item nonresponse and measurement error. In our research, we investigate the impact of monetary incentives on reducing item nonresponse and measurement error for labor and investment income in the Survey of Income and Program Participation (SIPP). Our study utilizes a randomized incentive experiment in Waves 1 and 2 of the 2014 SIPP, which allows us to assess the effectiveness of incentives in reducing item nonresponse and measurement error. We find that households receiving incentives had item nonresponse rates that are 1.3 percentage points lower for earnings and 1.5 percentage points lower for Social Security income. Measurement error was 6.31 percentage points lower at the intensive margin for interest income, and 16.48 percentage points lower for dividend income compared to non-incentive recipient households. These findings provide valuable insights for data producers and users and highlight the importance of implementing strategies to improve data quality in economic research.
View Full
Paper PDF
-
Incorporating Administrative Data in Survey Weights for the Basic Monthly Current Population Survey
January 2024
Working Paper Number:
CES-24-02
Response rates to the Current Population Survey (CPS) have declined over time, raising the potential for nonresponse bias in key population statistics. A potential solution is to leverage administrative data from government agencies and third-party data providers when constructing survey weights. In this paper, we take two approaches. First, we use administrative data to build a non-parametric nonresponse adjustment step while leaving the calibration to population estimates unchanged. Second, we use administratively linked data in the calibration process, matching income data from the Internal Return Service and state agencies, demographic data from the Social Security Administration and the decennial census, and industry data from the Census Bureau's Business Register to both responding and nonresponding households. We use the matched data in the household nonresponse adjustment of the CPS weighting algorithm, which changes the weights of respondents to account for differential nonresponse rates among subpopulations.
After running the experimental weighting algorithm, we compare estimates of the unemployment rate and labor force participation rate between the experimental weights and the production weights. Before March 2020, estimates of the labor force participation rates using the experimental weights are 0.2 percentage points higher than the original estimates, with minimal effect on unemployment rate. After March 2020, the new labor force participation rates are similar, but the unemployment rate is about 0.2 percentage points higher in some months during the height of COVID-related interviewing restrictions. These results are suggestive that if there is any nonresponse bias present in the CPS, the magnitude is comparable to the typical margin of error of the unemployment rate estimate. Additionally, the results are overall similar across demographic groups and states, as well as using alternative weighting methodology. Finally, we discuss how our estimates compare to those from earlier papers that calculate estimates of bias in key CPS labor force statistics.
This paper is for research purposes only. No changes to production are being implemented at this time.
View Full
Paper PDF
-
Connected and Uncooperative: The Effects of Homogenous and Exclusive Social Networks on Survey Response Rates and Nonresponse Bias
January 2024
Working Paper Number:
CES-24-01
Social capital, the strength of people's friendship networks and community ties, has been hypothesized as an important determinant of survey participation. Investigating this hypothesis has been difficult given data constraints. In this paper, we provide insights by investigating how response rates and nonresponse bias in the American Community Survey are correlated with county-level social network data from Facebook. We find that areas of the United States where people have more exclusive and homogenous social networks have higher nonresponse bias and lower response rates. These results provide further evidence that the effects of social capital may not be simply a matter of whether people are socially isolated or not, but also what types of social connections people have and the sociodemographic heterogeneity of their social networks.
View Full
Paper PDF
-
A Simulated Reconstruction and Reidentification Attack on the 2010 U.S. Census: Full Technical Report
December 2023
Authors:
Lars Vilhuber,
John M. Abowd,
Ethan Lewis,
Nathan Goldschlag,
Robert Ashmead,
Daniel Kifer,
Philip Leclerc,
Rolando A. Rodríguez,
Tamara Adams,
David Darais,
Sourya Dey,
Simson L. Garfinkel,
Scott Moore,
Ramy N. Tadros
Working Paper Number:
CES-23-63R
For the last half-century, it has been a common and accepted practice for statistical agencies, including the United States Census Bureau, to adopt different strategies to protect the confidentiality of aggregate tabular data products from those used to protect the individual records contained in publicly released microdata products. This strategy was premised on the assumption that the aggregation used to generate tabular data products made the resulting statistics inherently less disclosive than the microdata from which they were tabulated. Consistent with this common assumption, the 2010 Census of Population and Housing in the U.S. used different disclosure limitation rules for its tabular and microdata publications. This paper demonstrates that, in the context of disclosure limitation for the 2010 Census, the assumption that tabular data are inherently less disclosive than their underlying microdata is fundamentally flawed. The 2010 Census published more than 150 billion aggregate statistics in 180 table sets. Most of these tables were published at the most detailed geographic level'individual census blocks, which can have populations as small as one person. Using only 34 of the published table sets, we reconstructed microdata records including five variables (census block, sex, age, race, and ethnicity) from the confidential 2010 Census person records. Using only published data, an attacker using our methods can verify that all records in 70% of all census blocks (97 million people) are perfectly reconstructed. We further confirm, through reidentification studies, that an attacker can, within census blocks with perfect reconstruction accuracy, correctly infer the actual census response on race and ethnicity for 3.4 million vulnerable population uniques (persons with race and ethnicity different from the modal person on the census block) with 95% accuracy. Having shown the vulnerabilities inherent to the disclosure limitation methods used for the 2010 Census, we proceed to demonstrate that the more robust disclosure limitation framework used for the 2020 Census publications defends against attacks that are based on reconstruction. Finally, we show that available alternatives to the 2020 Census Disclosure Avoidance System would either fail to protect confidentiality, or would overly degrade the statistics' utility for the primary statutory use case: redrawing the boundaries of all of the nation's legislative and voting districts in compliance with the 1965 Voting Rights Act. You are reading the full technical report. For the summary paper see https://doi.org/10.1162/99608f92.4a1ebf70.
View Full
Paper PDF
-
The 2010 Census Confidentiality Protections Failed, Here's How and Why
December 2023
Authors:
Lars Vilhuber,
John M. Abowd,
Ethan Lewis,
Nathan Goldschlag,
Robert Ashmead,
Daniel Kifer,
Philip Leclerc,
Rolando A. Rodríguez,
Tamara Adams,
David Darais,
Sourya Dey,
Simson L. Garfinkel,
Scott Moore,
Ramy N. Tadros
Working Paper Number:
CES-23-63
Using only 34 published tables, we reconstruct five variables (census block, sex, age, race, and ethnicity) in the confidential 2010 Census person records. Using the 38-bin age variable tabulated at the census block level, at most 20.1% of reconstructed records can differ from their confidential source on even a single value for these five variables. Using only published data, an attacker can verify that all records in 70% of all census blocks (97 million people) are perfectly reconstructed. The tabular publications in Summary File 1 thus have prohibited disclosure risk similar to the unreleased confidential microdata. Reidentification studies confirm that an attacker can, within blocks with perfect reconstruction accuracy, correctly infer the actual census response on race and ethnicity for 3.4 million vulnerable population uniques (persons with nonmodal characteristics) with 95% accuracy, the same precision as the confidential data achieve and far greater than statistical baselines. The flaw in the 2010 Census framework was the assumption that aggregation prevented accurate microdata reconstruction, justifying weaker disclosure limitation methods than were applied to 2010 Census public microdata. The framework used for 2020 Census publications defends against attacks that are based on reconstruction, as we also demonstrate here. Finally, we show that alternatives to the 2020 Census Disclosure Avoidance System with similar accuracy (enhanced swapping) also fail to protect confidentiality, and those that partially defend against reconstruction attacks (incomplete suppression implementations) destroy the primary statutory use case: data for redistricting all legislatures in the country in compliance with the 1965 Voting Rights Act.
View Full
Paper PDF
-
Producing U.S. Population Statistics Using Multiple Administrative Sources
November 2023
Working Paper Number:
CES-23-58
We identify several challenges encountered when constructing U.S. administrative record-based (AR-based) population estimates for 2020. Though the AR estimates are higher than the 2020 Census at the national level, they are over 15 percent lower in 5 percent of counties, suggesting that locational accuracy can be improved. Other challenges include how to achieve comprehensive coverage, maintain consistent coverage across time, filter out nonresidents and people not alive on the reference date, uncover missing links across person and address records, and predict demographic characteristics when multiple ones are reported or when they are missing. We discuss several ways of addressing these issues, e.g., building in redundancy with more sources, linking children to their parents' addresses, and conducting additional record linkage for people without Social Security Numbers and for addresses not initially linked to the Census Bureau's Master Address File. We discuss modeling to predict lower levels of geography for people lacking those geocodes, the probability that a person is a U.S. resident on the reference date, the probability that an address is the person's residence on the reference date, and the probability a person is in each demographic characteristic category. Regression results illustrate how many of these challenges and solutions affect the AR county population estimates.
View Full
Paper PDF
-
An In-Depth Examination of Requirements for Disclosure Risk Assessment
October 2023
Authors:
Ron Jarmin,
John M. Abowd,
Ian M. Schmutte,
Jerome P. Reiter,
Nathan Goldschlag,
Victoria A. Velkoff,
Michael B. Hawes,
Robert Ashmead,
Ryan Cumings-Menon,
Sallie Ann Keller,
Daniel Kifer,
Philip Leclerc,
Rolando A. Rodríguez,
Pavel Zhuravlev
Working Paper Number:
CES-23-49
The use of formal privacy to protect the confidentiality of responses in the 2020 Decennial Census of Population and Housing has triggered renewed interest and debate over how to measure the disclosure risks and societal benefits of the published data products. Following long-established precedent in economics and statistics, we argue that any proposal for quantifying disclosure risk should be based on pre-specified, objective criteria. Such criteria should be used to compare methodologies to identify those with the most desirable properties. We illustrate this approach, using simple desiderata, to evaluate the absolute disclosure risk framework, the counterfactual framework underlying differential privacy, and prior-to-posterior comparisons. We conclude that satisfying all the desiderata is impossible, but counterfactual comparisons satisfy the most while absolute disclosure risk satisfies the fewest. Furthermore, we explain that many of the criticisms levied against differential privacy would be levied against any technology that is not equivalent to direct, unrestricted access to confidential data. Thus, more research is needed, but in the near-term, the counterfactual approach appears best-suited for privacy-utility analysis.
View Full
Paper PDF
-
Coverage of Children in the American Community Survey Based on California Birth Records
September 2023
Working Paper Number:
CES-23-46
The U.S. Census Bureau's American Community Survey (ACS) collects information on individuals and households. The ACS provides survey-based estimates of children drawn from a sample of the U.S. population. However, survey responses may not match administrative records, such as birth records. Birth records should provide a complete account of all births, along with child-parent relationships and demographic characteristics. California is a state that has both a large population of children and a high undercount for young children. This paper uses California as a case study to examine differences between reported versus unreported children in the ACS based on state birth records. Child reporting rates were lower for more recent data years, younger children, for Black and Hispanic mothers, and for more complex households. Child reporting rates were higher for more educated mothers and for households above the poverty line. Using mother's race and Hispanic ethnicity from the birth records combined with poverty indices from the ACS, this analysis also finds that child reporting does not uniformly vary with poverty status across all race and ethnicity groups. This research builds support for the utility of state birth records in analyzing the undercount of children.
View Full
Paper PDF
