Bunching Together: The Role of Demographic Similarities in the Spread of Bunching in the EITC
Studying how behaviors spread can help target interventions aimed at changing behavior and help predict how future behaviors will spread. I examine how bunching behavior at the first kink in the Earned Income Tax Credit (EITC) spreads locally among self-employed filers along demographic similarities. I use the universe of 1040 tax returns from 2004 to 2018 linked to the 2010 Decennial Census and the Social Security Administration's Numident file. I take both a descriptive and a causal instrumental variables approach to study how similarities between self-employed filers and others who bunched the year before in their area influence their probability of bunching in the current year. Finally, I use the January responses to the American Community Survey (ACS) to study how tax filers who bunch report their income in surveys. I find that the spread of bunching is consistent with it spreading through local networks. I also find descriptive and causal evidence of homophily in the spread of bunching. I find strong evidence of bunching in the tax data, but no evidence of bunching in the ACS. I also find that self-employed filers who bunch in the tax data are more likely to report earnings on their taxes that are above those they report on the ACS compared to other self-employed filers.
Studying how behaviors spread can help target interventions aimed at changing behavior and help predict how future behaviors will spread. I examine how bunching behavior at the first kink in the Earned Income Tax Credit (EITC) spreads locally among self-employed filers along demographic similarities. I use the universe of 1040 tax returns from 2004 to 2018 linked to the 2010 Decennial Census and the Social Security Administration's Numident file. I take both a descriptive and a causal instrumental variables approach to study how similarities between self-employed filers and others who bunched the year before in their area influence their probability of bunching in the current year. Finally, I use the January responses to the American Community Survey (ACS) to study how tax filers who bunch report their income in surveys. I find that the spread of bunching is consistent with it spreading through local networks. I also find descriptive and causal evidence of homophily in the spread of bunching. I find strong evidence of bunching in the tax data, but no evidence of bunching in the ACS. I also find that self-employed filers who bunch in the tax data are more likely to report earnings on their taxes that are above those they report on the ACS compared to other self-employed filers.
Earnings Through the Stages: Using Tax Data to Test for Sources of Error in CPS ASEC Earnings and Inequality Measures
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 also 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.
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 also 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.
Regression Discontinuity Estimates of the Effects of Winning School Bond Elections on School Spending, Test Scores, and Teacher Retention
California school districts spent over $6 billion in capital spending in 2014 and voted on 1,966 school bond measures between 1995 and 2016 meant to finance capital spending, yet it is unclear from the literature what effect capital spending has on student outcomes or whether there are other effects caused by passing school bond measures. I expand this literature by looking at the effect of passing school bond measures on standardized test scores in California using a dynamic regression discontinuity design. I find a large positive and immediate divergence in test scores and in proficiency rates between the districts close to the vote-share threshold that pass and that fail to pass school bond measures. This divergence starts the year of the elections and is difficult to explain based on the change in capital spending. Furthermore, this effect lasts several years. I further explore this effect by looking at the effects on teacher and staff turnover.
California school districts spent over $6 billion in capital spending in 2014 and voted on 1,966 school bond measures between 1995 and 2016 meant to finance capital spending, yet it is unclear from the literature what effect capital spending has on student outcomes or whether there are other effects caused by passing school bond measures. I expand this literature by looking at the effect of passing school bond measures on standardized test scores in California using a dynamic regression discontinuity design. I find a large positive and immediate divergence in test scores and in proficiency rates between the districts close to the vote-share threshold that pass and that fail to pass school bond measures. This divergence starts the year of the elections and is difficult to explain based on the change in capital spending. Furthermore, this effect lasts several years. I further explore this effect by looking at the effects on teacher and staff turnover.
JMP.pdf | |
File Size: | 1082 kb |
File Type: |
Superstitious Parents and Healthier Babies: Birth Outcomes and the Chinese Zodiac
In Chinese culture, children born during the year of the dragon are thought to be luckier and more successful. This leads to increases in fertility for Chinese parents in that year as well possible changes in how those children are raised. Several studies have argued either that differences between children born during the dragon years are driven by differences in parental involvement or that dragon years can be used as a source for exogenous change in the size of the Chinese populations. I show that both of these interpretations are incorrect. Using data from birth records in the United States, I find that children born to Chinese mothers during dragon years are healthier at birth than those born to Chinese mothers in other zodiac years. I also find that there is substantial selection into childbearing by Chinese mothers during dragon years, with those dragon year mothers tending to be younger, more educated, and more likely to have had previous children.
In Chinese culture, children born during the year of the dragon are thought to be luckier and more successful. This leads to increases in fertility for Chinese parents in that year as well possible changes in how those children are raised. Several studies have argued either that differences between children born during the dragon years are driven by differences in parental involvement or that dragon years can be used as a source for exogenous change in the size of the Chinese populations. I show that both of these interpretations are incorrect. Using data from birth records in the United States, I find that children born to Chinese mothers during dragon years are healthier at birth than those born to Chinese mothers in other zodiac years. I also find that there is substantial selection into childbearing by Chinese mothers during dragon years, with those dragon year mothers tending to be younger, more educated, and more likely to have had previous children.
The Peer Effects of High Ability Peers: Variation from the Chinese Zodiac
Parents care a great deal about school quality and about the quality of the students that will be their children's peers. In this project, I investigate the effects of changing peer composition from an increase in classroom diversity using variation induced by the Chinese zodiac. In Chinese culture, children born during the year of the dragon are thought to be luckier and more successful. In other work, I show that there are differences between dragon cohorts and other Chinese children at birth, Here I try to isolate how the differences in Chinese students during dragon years affects their peers by controlling for the percentage of Chinese students. Using standardized test scores in California, I find that exposure to Chinese students born during the year of the dragon lowers the standardized test scores of other students in their schools.
Parents care a great deal about school quality and about the quality of the students that will be their children's peers. In this project, I investigate the effects of changing peer composition from an increase in classroom diversity using variation induced by the Chinese zodiac. In Chinese culture, children born during the year of the dragon are thought to be luckier and more successful. In other work, I show that there are differences between dragon cohorts and other Chinese children at birth, Here I try to isolate how the differences in Chinese students during dragon years affects their peers by controlling for the percentage of Chinese students. Using standardized test scores in California, I find that exposure to Chinese students born during the year of the dragon lowers the standardized test scores of other students in their schools.