Dani Sandler

@dismalscientist86

Economist at the Census Bureau studying women's employment, post-prison outcomes, and low-income housing. Living in Davis, CA.

1,913 Following5,943 Followers479 Posts

Bluesky

Is this your profile?Claim it and dismalscientist86.dirty.social is yours.Claim profile

Dani Sandler@dismalscientist86· 1d

This paper presents statistics on the prevalence of U.S. export controls, the types of firms that export controlled products, and the impact of controls on producers.

National-security restrictions on exports constrain exporters' market access, but empirical evidence on their domestic costs is limited. We leverage data on export transactions that includes the product classifications used to administer U.S. export controls to document a doubling in the share of U.S. exports subject to controls between 2010 and 2024 and a skewed firm-size distribution among firms that export controlled products. We then exploit the 2014 U.S. restriction on exports to Russia as a natural experiment to examine the impact of export controls on firm-level outcomes. These controls not only reduced U.S. firms' probability of exporting targeted goods to Russia but also had indirect effects: they reduced affected firms' exports of other products and to other destinations and lowered affected firms' average payroll. Our findings imply that the domestic costs of export controls extend beyond direct restrictions.

Figure 3. Targeted vs. Non-Targeted Product Exports to Russia and China

Notes: Value of targeted and other products, in millions of 2019 U.S. dollars, exported to (a) Russia and (b) China. Targeted products are defined as one of 52 Schedule B codes or eight ECCN codes listed in FR 45675; non-targeted products refer to all other products not listed in FR 45675. Source: Authors’ calculations using U.S. Census Bureau (2024) and FR 45675.

Dani Sandler@dismalscientist86· 1d

This paper combines Census survey and IRS administrative data to estimate participation in the Child Tax Credit and Additional Child Tax Credit programs for 2021.

The Child Tax Credit (CTC) and Additional Child Tax Credit (ACTC) help ease the financial burden of families with children. This paper provides taxpayer and dollar participation estimates for the CTC and ACTC covering tax year 2021. In this tax year the credit was temporarily expanded under the American Rescue Plan Act. The estimates derive from linking the 2022 Current Population Survey Annual Social and Economic Supplement (CPS ASEC) to IRS administrative data. This approach, called the Exact Match, uses survey data to identify CTC/ACTC eligible taxpayers and IRS administrative data to indicate which eligible taxpayers claimed and received the credit. Overall, in tax year 2021, eligible taxpayers participated in the CTC and ACTC program at a rate of 91 percent while dollar participation was 85 percent.

Dani Sandler@dismalscientist86· 1d

Rural firms’ innovation gap stems mostly from deficits in digital infrastructure, R&D, and human capital - not from rural location itself.

Place-based policy is premised on the idea that productive investments that address endowment deficiencies may release latent comparative advantages, stimulating growth while reducing regional disparities. Extending place-based policy to the domain of innovation is conceptually challenging given conventional wisdom that settlement size is the principal determinant of agglomeration economies that fuels innovation. However, growing concern that more places are being left behind by the innovation economy has prompted greater interest in ways to expand the geography of innovation. Using confidential firm-level data from the 2018 Annual Business Survey, this study examines differences in self-reported innovation activity between firms in rural and urban counties to identify possible endowment deficiencies. We do this by employing the Oaxaca-Blinder decomposition method to quantify the urban-rural innovation gap to better understand the relative contributions of firm characteristics, owner characteristics, and county-level factors.

Figure 1 Estimating Logit Models for Metro and Nonmetro Firms

Notes: Estimated coefficients from separate logit models for metro and nonmetro firms for three innovation outcomes: new-to-market, process, and marketing innovation.

Dani Sandler@dismalscientist86· 1d

More underfunded U.S. public pensions earn lower average PE returns and match with smaller average GPs. Target firms financed see labor productivity declines.

I study investments of U.S. public pensions in private equity (PE), and trace them to ultimate micro assets—target firms which PE funds invest in, using micro-data on private investments combined with confidential U.S. Census data. I show that more severely underfunded public pensions receive lower average PE returns, and match with smaller GPs on average, than less underfunded pensions. Consistent with matching and returns, firms financed by most underfunded public pensions and smallest PE funds face labor productivity decreases. I introduce a novel instrument—public unionization—in support of underfunding positions driving selection into funds. Lastly, I evaluate alternate mechanisms.

Table 3: Correlation Between Public Pension Underfunding Positions and Average Private Equity Returns for 2001-2021

Notes: The table shows the relation between underfunding ratios of public pensions and average PE returns for 2001-2021. The PE performance is measured using the net internal rate of return (IRR). Column (1) controls for LP size, Column (2) controls for LP size and includes current PE allocation, Column (3) controls for LP size and includes current alternative asset allocation, and Column (4) controls for LP size, includes the average past three-year PE, equity, fixed income allocations, and investment consultant dummies. Robust standard errors in parentheses. Significance levels: ∗∗∗ p < 0.01, ∗∗ p < 0.05, ∗ p < 0.10.

Dani Sandler@dismalscientist86· 1d

SNAP work requirements reduce participation by 7% without raising employment and disproportionately screen out lower-income recipients.

We examine the effectiveness of work requirements as a screening device in the Supplemental Nutrition Assistance Program (SNAP). Work requirements for “able-bodied adults without dependents” were suspended after the Great Recession and gradually reinstated across counties and states in the 2010s. Using linked administrative SNAP and employment data from five states and a triple-differences design, we find that work requirements reduce SNAP participation by seven percent without increasing labor supply and disproportionately screen out low-income individuals. We develop a welfare framework to interpret these results and find that the social costs of work requirements exceed budget savings.

Figure 1: Effect of Work Requirements on SNAP Participation
Notes: This graph plots event-study estimates of Equations (1) and (3) where the outcome is monthly SNAP participation. The triple-differences (DDD) estimates are shown with solid dots, and the difference-in-differences (DiD) estimates are shown with square markers. Both the DDD and DiD samples include childless individuals aged 45 to 55 in five states that partially reinstated work requirements: Colorado, Connecticut, Maryland, Massachusetts, and New Jersey. The event-study specifications are estimated separately for each state and then aggregated using a weighted average across states. The weights are based on each state’s share of individuals aged 45-49 residing in treated counties. Bars show 95% confidence intervals.

Dani Sandler@dismalscientist86· 5d

Questionable product at a local store. Is the dog ... dead? Why is it spurting water?

Sprinkler toy that appears to be a dog flattened on the ground with water spurting from its back.

Dani Sandler@dismalscientist86· 8d

This study provides guidance in understanding the Census Bureau and USDA, FNA (formerly FNS) estimates of WIC eligibility, participation, and access rates.

This study estimates eligibility and access rates for the U.S. Department of Agriculture’s (USDA) Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) administrative data linked with American Community Survey (ACS) data. This study is one result of a long-term research collaboration among USDA’s Economic Research Service; the U.S. Department of Commerce, Bureau of the Census; USDA’s Food and Nutrition Service (FNS); and participating state WIC agencies. By analyzing WIC participation at the state and substate levels, the report provides insights into program reach and demographic differences. The findings confirm that the Census Bureau estimates meet high statistical reliability standards, providing valuable data for program officials and managers, and other stakeholders, to enhance program outreach and effectiveness. A key focus of the report is the comparison between Census Bureau and USDA, FNS estimates, which differ in methodology and measurement scope.

Figure 1. WIC coverage rate and WIC access rate in Arizona, by years of age, 2021

Note: USDA, Food and Nutrition Service (FNS) estimates the Special Supplemental Nutrition Program for Women, Infants, and Children program (WIC) coverage rates and the U.S. Department of Commerce, Bureau of the Census estimates the WIC access rates for Arizona for 2021. The figure shows 95-percent confidence intervals based on reported 95-percent margins of error for Census Bureau estimates and deriving the intervals from report coefficients of variation for the FNS estimates. Estimated access rates do not include infants and children who receive benefits from Navajo Nation WIC or Inter Tribal Council of Arizona WIC. Source: USDA, Economic Research

Dani Sandler@dismalscientist86· 9d

This paper details the data linkage and estimation methods used to produce WIC eligibility and access rates at the state, county, and demographic levels.

The Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) provides benefits to low-income, nutritionally at-risk women, infants, and children. To administer WIC, officials and program managers at the federal and state level want to understand who is eligible for the program, who among the eligible population chooses to participate, and who is not accessing the program despite their eligibility. Novel individual-level data linkages between restricted-use WIC Administrative Records and the American Community Survey provide WIC access rates estimated at the state and county levels, as well as estimates disaggregated by the demographic and socioeconomic characteristics of individuals and their households. These estimates are developed by the Census-FNS-ERS Joint Project, a research partnership among the U.S. Census Bureau, the US Department of Agriculture’s Food and Nutrition Service and Economic Research Service, and state WIC agencies that provide the requisite WIC administrative data to the Census Bureau. This paper details and evaluates our current data linkage and estimation methods, reports results, and identifies areas for improvement and further research.

Table 1. Eligibility and Access Rates (%) by Infant and Child Characteristics, Infants and Children, 2019

Note: MEs represent the margin of error at the 95 percent confidence level. NH indicates Non Hispanic; AIAN indicates American Indian or Alaska Native. Sources: ACS (2019); and Arizona and Wisconsin WIC ARs (2018-2019). CBDRB-FY19-167, CBDRB-FY19-575, CBDRB-FY20-154, & CBDRB-FY21-108

Dani Sandler@dismalscientist86· 9d

Entrepreneurs build organizations that are better able to make use of workers from similar backgrounds.

What are the implications of unequal access to entrepreneurial careers for labor markets? Using data from the U.S. Census and LinkedIn profiles, we document that entrepreneurs are significantly more likely to hire workers from similar social backgrounds (gender, race, age, education, etc.). These effects are quantitatively large across several demographic dimensions. For example, female employee share at female-founded startups is 36.4pp higher after controlling for industry-by-metro area-by-cohort fixed effects, with corresponding estimates of 51.2pp for Blacks, 37.3pp for Hispanics, and 11.3pp for non-college individuals. Large effects are present in high-growth startups, across industries and occupations, and remain stable across new firm cohorts. In addition, we find that these differences persist out to at least 20 years. We use wage data and an AKM research design to untangle whether the relative differences are driven by labor demand or labor supply effects. We find that demand drives the differences: groupspecific wage decompositions show that new firms pay higher relative wages to individuals from similar backgrounds to the entrepreneur. Using these estimates, we calibrate a model of entrepreneurship with heterogeneous ability and production functions, and assess the impacts on relative wage from reducing access barriers to entrepreneurship.

Figure 1: Estimates from Balanced Sample of New Firms

Notes: Each panel plots the coefficient βt from Equation 1 using a different demographic variable. The three series correspond to specifications with: (i) no fixed effects; ii) 6-digit NAICS-by-founding year fixed effects; iii) 2-digit industry-by-MSA-by-founding year fixed effects. All standard errors are clustered at the firm level and are negligible relative to the point estimates, so we have not added confidence intervals to the plot. Source: authors’ calculations based on U.S. Census data (FSRDC Project Number 2229, Disclosure Review Board (DRB) approval numbers: CBDRB-FY25-P2229-R11972 and CBDRB-FY26-P2229-R12962).

Dani Sandler@dismalscientist86· 9d

Minimum wage hikes spur automation: a 10% rise in the wage floor increases manufacturing robot adoption by ~8%, new plant-level Census data show.

This paper studies how minimum wage policy affects firms’ adoption of automation technologies. Using both state-level measures of robot exposure and novel plant-level data on industrial robot imports linked to U.S. Census microdata from 1992–2021, we show that increases in minimum wages raise the likelihood of robot adoption in manufacturing. Our preferred identification exploits discontinuities at state borders, comparing otherwise similar firms exposed to different wage floors. Across specifications, a 10 percent increase in the minimum wage increases robot adoption by roughly 8 percent relative to the mean.

Figure 1: State-Level Exposure to Robots and Minimum Wage

Dani Sandler@dismalscientist86· 12d

Another item of interest at the Money Museum at the Chicago Fed is 1 million dollars in singles, twenties, and hundreds.

Dani Sandler@dismalscientist86· 12d

TIL: The US Treasury stamped all the money issued on Hawaii with the word HAWAII in WWII in case it the island were invaded.

Dollar bill stamped with HAWAII

Text: $1 Federal Reserve Note
During World War II, at a time when there was a threat that Hawaii might be invaded, the Treasury replaced all U.S. currency on the island with a special Hawaiian issue. These notes were overprinted with the word "Hawaii."

Federal Reserve Bank of Chicago

Dani Sandler@dismalscientist86· 17d

TIL: U.S. born women married to non-citizen men lost their U.S. citizenship in 1907, until and unless their husbands became citizens. This was reversed in 1922, but then those women who had not become citizens again through their husband's naturalization had to reapply independently for it.

The year 1907 saw the creation of a law that had a strange effect on some married couples. The Expatriation Act decreed that any woman who married a foreigner would lose her American citizenship and take on that of her husband.

Dani Sandler@dismalscientist86· Jul 9

Using business survey data linked to the LBD, this paper explores how affirmative action bans in state contracting affect minority and women-owned business enterprises.

Using restricted Census business records, I explore how banning affirmative action in state contracting affects minority- and women-owned business enterprises (MWBEs). I find that ending affirmative action led MWBE contractors to gradually downsize, with the most pronounced reductions in force experienced by Black-owned businesses and larger MWBEs. Despite these workforce changes, existing MWBEs were no more likely to shut down than other businesses. New MWBEs were relatively less common after a state’s ban, highlighting how bans can shift the demographic composition of new contractors. A calibrated model suggests bans are equivalent to considerable reductions in MWBE productivity and scrap values.

Figure 1: The Effect of Affirmative Action Bans on Employment
Note: Figure plots the three-way interaction coefficients, δs, and 95% confidence intervals attained from estimating equation (1). The dependent variable is the inverse hyperbolic sine of employment. Regressions use SBO weights, with data winsorized from above and below by 1%. Results were approved for release by the U.S. Census Bureau under FSRDC Project Number 2075, with approval number CBDRB-FY26-P2075-R12853.

Dani Sandler@dismalscientist86· Jul 9

This paper uses tax reports of restaurant workers and employers to measure the importance of bunching in reported tip income in response to minimum-wage incentives.

We study the importance of bunching in the context of tip-income reporting by workers at fullservice, single-unit restaurants in the United States. Using tax reports at both the individual and the employer levels, we show that reported tip income varies with minimum-wage laws that provide an incentive for tipped workers to report some, but not necessarily all, of their tips. As a result, reported tips bunch at the minimum required threshold. We quantify missing tips due to bunching at nearly $63 million per year in 2018 dollars, on average over the period 2005-2018. Bunching is stronger for jobs at small employers and in the earlier part of the time series and declined monotonically from 2010 to 2018. Using restaurant-level revenue data, we also estimate the total value of unreported tips assuming an average tip rate of 12%. We find that tips are missing throughout the distribution. All told, missing tips exceed $4 billion per year, implying that bunching explains only 1.5% of all missing tips.

Figure 2. Distribution of Normalized Tips-to-Earnings Ratios by Employer Size, 2005–2018
Notes: The fgure shows the distribution of normalized tips-to-earnings ratios by employer size (small versus large) for a sample of 27.5 million job-years in states with tip credits (representing 9.5 million unique workers). This ratio is 1 when a worker in a job subject to a two-tiered minimum wage and tip credit reports exactly the tips to meet the tip credit. The shaded area is the bunching window: [0.96, 1.05], with a vertical line at the reporting threshold of 1. Authors’ calculations from Form W-2 and state minimum wages. Tails removed to comply with Census Bureau disclosure-avoidance rules.

Dani Sandler@dismalscientist86· Jul 9

When women had more female teachers at young ages, they completed more education, married later and better, had higher household earnings, and died later.

We document an enduring influence of female primary school teachers on their female students in rural America circa 1940. Using historical US Census data linked to IRS records, later censuses, and death records, we find that exposure to female teachers significantly improved girls’ educational attainment—particularly among those with poorly-educated parents and in areas with limited labor market opportunities for women. These educational gains translated into delayed marriage and childbearing, marriage matches to better-educated husbands and higher household income. We also find that exposure to female teachers increased longevity of women relative to men.

Figure 3: Estimated of Local FFT on Marriage and Childbearing, Measured in the 1940 Census, for Females Aged 10–18 in the 1930 Census
Notes: These are estimated efects of FFT on “any child” and “ever married” estimated in nine separate regression, using the ED design, by age for cohorts of females aged 10–18 in 1930 census records matched to 1940 census records. In all cases the dependent variable is a binary variable equal to 100 or 0.

Dani Sandler@dismalscientist86· Jul 9

This paper explores the differences in SNAP receipt between the CPS ASEC, the ACS, and the SIPP.

The Census Bureau asks households about the receipt of the Supplemental Nutrition Assistance Program (SNAP), also known as food stamps, using the Current Population Survey Annual Social and Economic Supplement (CPS ASEC), the American Community Survey (ACS) and the Survey of Income and Program Participation (SIPP). SNAP receipt is important in its own right as a metric, and it is also used as a noncash resource for the Supplemental Poverty Measure. Studies done at the U.S. Census Bureau and at U.S. Department of Agriculture have shown that SNAP is underreported in all three surveys. However, studies have not addressed the difference in SNAP receipt between the CPS ASEC, the ACS, and the SIPP. The SNAP receipt rate for households was approximately 13.3 percent in the SIPP, 12.2 percent in the ACS, and 10.2 percent in the CPS ASEC in 2023. In this paper, SNAP receipt is examined and compared in the three surveys over time and by demographics and labor market characteristics.

Figure 1. Percent of Households that Reported Receipt of SNAP Benefits by Survey

Dani Sandler@dismalscientist86· Jun 24

This report documents some of the research results from the Administrative Records Dual System Estimation study.

This report expands upon the innovation of utilizing administrative records and third-party data implemented in the 2020 Census. The 2020 Census used administrative records and third-party data in address canvassing and nonresponse followup operations. The Census Bureau also has a long history of using administrative records of births, deaths, and other information to produce Demographic Analysis coverage estimates. Since 1980, the Census Bureau has produced capturerecapture coverage estimates by conducting an independent post-enumeration survey and utilizing dual system estimation approaches. This report presents the research results of attempting to see if administrative records and third-party data could be utilized to produce capture-recapture coverage estimates. This work uses an Expectation Maximization Log Linear Modeling approach previously researched by Statistics Netherlands and Statistics New Zealand. This report documents some of the experimental results from an evaluation that was part of the 2020 Census Program for Evaluation, Experiments, and Assessments.

Figure 1: Net Coverage Error for 0-to-4 Population by Hispanic Origin

Dani Sandler@dismalscientist86· Jun 24

This paper uses the Environmental Impacts Frame to highlight new facts about the distribution of flood risk in the US.

Flooding is among the most salient natural hazards facing households in the United States. A large body of evidence has documented a pattern of disproportionate social vulnerability in floodplains. However, little evidence exists on how household-level exposure to flood risk is distributed. We fill this gap by combining parcel-level flood risk with confidential linked survey and administrative data held at the US Census Bureau. Although net migration to Census blocks in floodplains has increased in recent years, there has been essentially no net migration to parcels with flood risk or change in the overall share of households living in floodplains. Income gradients in flood risk are highly non-linear at the household level, with slightly negative income gradients for the bottom 90 percentiles of the income distribution that are dwarfed by disproportionate exposure in the top decile, especially when considering multiple property ownership. This nonlinearity is largely driven by differences in building type and homeownership within narrow income groups. In contrast to the conclusions in the literature using aggregate data, our household-level analysis suggests that households in floodplains are less disadvantaged and increasingly protected from the impacts of flooding, even as a vulnerable subpopulation of low income, uninsured homeowners remains.

Figure 4: Trends in the Share of the Population in Floodplains by Income, 1999-2023

Dani Sandler@dismalscientist86· Jun 18

Zorro is trying out my new notebook

Grey and white cat sitting regally on a red spiral notebook

Dani Sandler@dismalscientist86· Jun 15

BLS reports many more workplaces than Census because it records twice as many multi-establishment firms, and includes more home health care employees.

Since the 1990s, the Bureau of Labor Statistics (BLS) has reported much more rapid growth in
U.S. private sector employer establishments than has the Census Bureau– the gap reached
roughly 1.6 million by 2023. Using linked BLS-Census microdata, we document two main drivers.
First, a large and growing number of employers providing services to the elderly and persons with
disabilities are in scope for the BLS frame but not the Census Bureau’s. Second, many firms
appear with substantially more establishments in the BLS frame. These discrepancies
substantially affect the measured establishment size distribution and quantitative policy analysis.

Notes: The Business Employment Dynamics (BDM) is derived from the Quarterly Census of Employment and Wages
(QCEW) and published by the Bureau of Labor Statistics. The Business Dynamics Statistics (BDS) is derived from
the Longitudinal Business Database (LBD) and published by the Census Bureau.

Dani Sandler@dismalscientist86· Jun 15

Does military service help immigrants assimilate? New research finds the answer is self-selection - not service itself.

Seminal theories in political science argue that military service is a critical driver of minority integration. However, a major obstacle bedeviling the study of military service is self-selection: individuals who are better integrated may be more likely to join the military in the first place. We address the selection problem by examining the effects of military conscription during the Vietnam War using an instrumental variables approach. Conscription during 1970--72 was decided on the basis of national draft lotteries that assigned draft numbers based on an individual's date of birth. Using the draft lottery instrument, we find no evidence of a causal effect of military service on a range of integration outcomes from the 2000 decennial census. At least for the Vietnam era, the link between service and long-term integration is largely driven by selfselection, which points to important scope conditions for the integrationist view.

The table reports results for the effect of DRAFT RISK on integration outcomes. Dependent variables appear in the rows. Observation counts are rounded to comply with Census Bureau policy. Results have been approved for release under FSRDC Project Number 2896 (CBDRB-FY24-P2896-R11579 and CBDRB-FY25-0074).

Dani Sandler@dismalscientist86· Jun 15

Through quasi-random variation in office lease expirations, remote work causes migration for lower income taxes, public schools, space, and homeownership.

How has remote work reshaped residential sorting and housing demand, and what are the implications for state and local governments? To estimate causal effects, I propose a novel instrument for remote work that exploits quasi-random variation in the timing and size of office lease expirations, captured through a Bartik-style exposure measure at the residential block level. Expirations allow tenant firms to reduce office space and switch employees to remote work, generating strong first-stage effects. Remote work causes modest increases in housing and property tax expenditures in exchange for space, homeownership, and public schools, but not other neighborhood characteristics. It significantly increases migration, particularly out of cities and states that levy income taxes. At the neighborhood level, higher 2020 remote work shares cause subsequent residential turnover, demographic clustering, and property tax revenue windfalls. Taken together, the results indicate that remote work induces migration consistent with Tiebout sorting, and accounts for 10% of migration since 2020. Residential choices and tax bases now depend less on employment proximity and more on affordability and tax-benefit linkage.

Notes: This table presents the results of the second stage regression shown in Equation 5, which estimates the causal impact of remote work on selected outcomes. The explanatory variable is 1{remote}, an indicator for self-reported remote work, defned for employed individuals only. The frst outcome indicates changing residences in the past year. The next 2 outcomes are indicators that the worker lives in a state or city that levies income taxes. The fnal outcome is annual property taxes paid by the worker’s household, set to 0 for renters. All columns include controls for local ofce real estate characteristics and year fxed efects. To condition on prior year place of residence, they include linear controls for neighborhood income and housing characteristics, as well as fxed efects for the prior year city and county of residence, as described in Section 3.3. All standard errors are clustered by interactions of three-digit NAICS industry, city, and county. All fgures are rounded to 4 signifcant digits. * p < 0.1, ** p < 0.05, *** p < 0.01.

Dani Sandler@dismalscientist86· May 22

This paper uses matched employee-employer data to show how the employment and earnings of program participants change as they enter and exit housing assistance programs.

Federal housing assistance programs, such as those run by the U.S. Department of Housing and Urban Development (HUD), have been shown to reduce rent burden and improve housing stability for program participants, which may in turn have downstream impacts on their labor market attachment and career trajectories. However, existing studies from individual cities or states provide mixed evidence on the association of housing assistance with labor market outcomes. By linking HUD administrative records to matched employee-employer earnings records from the Longitudinal Employer-Household Dynamics (LEHD) program, we document how the labor market trajectories of program participants change as they enter and exit federal housing assistance programs, examining outcomes over a 14-year window surrounding entry or exit. In our analysis of entry, we find that the employment rates and earnings of first-time HUD program participants begin to increase upon entering a HUD program, which represents a reversal of prior declining trends in these outcomes. Suggestive of a positive association, these increases in employment and earnings trends exceed those of low-income non-participants from the American Community Survey (ACS). In our analysis of exits, we find that program participants who eventually leave a HUD program have increasing pre-exit trends in employment and earnings that then flatten upon exiting. Comparing these negative changes in trend to the relatively stable trajectories of those who remain in HUD programs throughout the analysis suggests that exits are associated with diminished employment and earnings trajectories.

Figure 2: Labor market trajectories of HUD entrants and low-income ACS comparators

Dani Sandler@dismalscientist86· May 22

This paper analyzes intergenerational mobility of wealth and income across U.S. counties, highlighting spatial differences due to house prices and the Great Recession.

We provide new county-level estimates of intergenerational mobility, covering multiple economic concepts: total income, labor income, homeownership, housing wealth, and total wealth. This is possible via small-area estimation techniques and linked survey and administrative data covering millions of U.S. children born between 1978 and 1986. We find that relative mobility in wealth concepts shows less spatial clustering and more spatial variation than relative mobility in income concepts. Many cities and their suburbs exhibit lower relative mobility (i.e. higher intergenerational persistence) in wealth concepts than in income concepts. Next, we show that various local characteristics are associated with some concepts of economic mobility but not with others. For example, we estimate a strong negative association between the local severity of the Great Recession and child income, regardless of parent position in the income distribution. However, the negative association between recession severity and wealth only exists among children from poorer families. We provide a public-use data package on census.gov to facilitate further research.

Notes: These maps show the county-level variation in β for total income and labor income, our measure of relative mobility for children of parents on the intensive margin (parents with the lowest non-zero amount of the resource concept in question). Cool colors correspond to higher relative mobility, warmer colors correspond to lower levels of relative mobility.

Dani Sandler@dismalscientist86· May 22

Should bankruptcy forum shopping be allowed? Census data show Delaware filings save firms, speed cases, boost recoveries, and lift jobs by 24.8%.

Many non-Delaware firms strategically file for bankruptcy in Delaware. Should this "forum shopping" be allowed? This question has motivated nine proposed congressional bills over decades of policy debate. Using a novel natural experiment and Census-Bureau microdata, we inform this debate. Comparing similar firms within a Delaware-adjacent state, we show that proximity to Delaware predicts forum shopping. Instrumenting with proximity, we find that forum shopping causally: (i) prevents closures and liquidations, (ii) shortens bankruptcies, (iii) boosts creditor recovery, and (iv) increases post-bankruptcy employment by 24.8%. Proximity to Delaware is uncorrelated with growth for not-yet-bankrupt or never-bankrupt firms, validating the exclusion restriction.

Figure 1: Forum Shopping Boosts Post-Bankruptcy Employment: Results and Placebo Tests This figure displays 2SLS estimates of the causal effect of forum shopping to Delaware on industry-adjusted employment changes. Each observation is an establishment of a firm filing for bankruptcy in year t. The outcome is Δ Adj. Employt−1 → T , defined as the change in employment from year t − 1 to year T , minus the average employment change for establishments filing in year t in the same 4-digit NAICS industry. If an establishment is inactive in either year t − 1 or T , we assign it zero employees in that year. We instrument for filing in Delaware using the distance from the bankrupt firm’s headquarters to the Delaware bankruptcy court (see Table 2). Each specification mirrors the specification described in Table 3. The black circles plot 2SLS estimates for our baseline sample across horizons T = t − 3, t − 2, t, t +1, t +2, t + 3, with estimates on the y axis and horizons on the x axis. The gray diamonds plot 2SLS estimates for a placebo sample of never-bankrupt establishments, each randomly selected from the same ZIP code as an establishment in our baseline sample and assigned that establishment’s filing characteristics. Standard errors are clustered at the firm level. Error bars display 95% confidence intervals. This graph was estimated at a Federal Statistical Research Data Center under FSRDC Project Number 2910. (CBDRB-FY26-P2910-R12910). Estimates have been rounded to four significant digits according to the disclosure avoidance practices in place at the Census Bureau.

Dani Sandler@dismalscientist86· May 22

This paper combines new evidence from the Annual Capital Expenditure Survey with a model of custom software to examine its macroeconomic implications.

Custom software is distinct from other types of capital in that it is non-rival—once a firm makes an investment in custom software, it can be used simultaneously across its many establishments. Using confidential U.S. Census data, we document that while firms with more establishments are more likely to invest in custom software, they spend less on it as a share of total capital expenditure. We explain these empirical patterns by developing a model that incorporates the non-rivalry of custom software. In the model, firms choose whether to adopt custom software, the intensity of their investment, and their scope, balancing the cost of managing multiple establishments with the increasing returns to scope from the nonrivalrous custom software investment. Using the calibrated model, we assess the extent to which the decline in the rental rate of custom software over the past 40 years can account for a number of macroeconomic trends, including increases in firm scope and concentration.

This figure plots the establishment-size bin fixed effects, γk, from Equation (1)), with single-unit firms as the omitted category. Panel A plots the share of firms investing in custom software in each establishment size category. Panel B plots the software investment intensity, measured by the share of custom software expenditures relative to total capital expenditures, for each establishment size category. The regression controls for industry-year fixed effects at the 6-digit NAICS level. We report the 95% confidence interval. Standard errors are clustered at the industry-year level.

Dani Sandler@dismalscientist86· May 10

I also am not sure where to put my arms when sleeping on my stomach

Dani Sandler@dismalscientist86· May 2

April showers bring May flowers

Dani Sandler@dismalscientist86· Apr 28

Early career hires in the industries most exposed to AI dropped by 9% after ChatGPT's release, explaining a 12% decline in employment through early 2025.