Working Papers
How well does EITC help low-income households in the age of automation? (Job Market Paper)
Abstract: Recent studies suggest that technological progress such as robots and Artificial Intelligence (AI) is replacing human labor, and not just routine-intensive jobs. At the same time, welfare programs for low-income families in the U.S. have increasingly emphasized in-work aid over out-of-work aid since the 1990s. Consequently, the U.S. safety net provides only modest benefits to non-workers. In this paper, I empirically examine to whether low-income families are still supported by Earned Income Tax Credit (EITC) under automation by exploiting variation in exposure to industrial robot adoption across commuting zones. The results show that there is no significant difference in the EITC recipiency rates across commuting zones and a negative effect on the average benefit for the married filers with children. Also, the estimates present some suggestive effects. For the married with children, the EITC recipiency rates appear to decrease in commuting zones with low routine share due to an increase in the share of families with higher earnings. For the single filers with children, automation could increase their recipiency rate in commuting zones with high routine share. Overall, there is no evidence that current EITC is ineffective in local labor markets experiencing growth in automation.
Abstract: Recent studies suggest that technological progress such as robots and Artificial Intelligence (AI) is replacing human labor, and not just routine-intensive jobs. At the same time, welfare programs for low-income families in the U.S. have increasingly emphasized in-work aid over out-of-work aid since the 1990s. Consequently, the U.S. safety net provides only modest benefits to non-workers. In this paper, I empirically examine to whether low-income families are still supported by Earned Income Tax Credit (EITC) under automation by exploiting variation in exposure to industrial robot adoption across commuting zones. The results show that there is no significant difference in the EITC recipiency rates across commuting zones and a negative effect on the average benefit for the married filers with children. Also, the estimates present some suggestive effects. For the married with children, the EITC recipiency rates appear to decrease in commuting zones with low routine share due to an increase in the share of families with higher earnings. For the single filers with children, automation could increase their recipiency rate in commuting zones with high routine share. Overall, there is no evidence that current EITC is ineffective in local labor markets experiencing growth in automation.
Persistent poverty counties and the distribution of economic development funding
Abstract: There are 425 counties in persistent poverty, their poverty rates continuously have been 20 percent or more since 1970. This paper tries to examine determinants of the distribution of federal economic development funding at the county level by focusing on whether there are differences between persistent poverty counties and non-persistent poverty counties. Regression results with county fixed effects and time effects show that (i) in the case of non-persistent poverty counties, more distressed areas receive relatively less per capita funding, whereas funding distribution on persistent poverty counties is not affected by economic need or political concerns, (ii) while project grants (or funding of USDA) are likely to be influenced by economic condition, formula grants (or CDBG funding of HUD) are affected more by voting margin, not economic need, (iii) there seems little structural differences in funding distribution between persistently poor counties and non-poor counties.
Abstract: There are 425 counties in persistent poverty, their poverty rates continuously have been 20 percent or more since 1970. This paper tries to examine determinants of the distribution of federal economic development funding at the county level by focusing on whether there are differences between persistent poverty counties and non-persistent poverty counties. Regression results with county fixed effects and time effects show that (i) in the case of non-persistent poverty counties, more distressed areas receive relatively less per capita funding, whereas funding distribution on persistent poverty counties is not affected by economic need or political concerns, (ii) while project grants (or funding of USDA) are likely to be influenced by economic condition, formula grants (or CDBG funding of HUD) are affected more by voting margin, not economic need, (iii) there seems little structural differences in funding distribution between persistently poor counties and non-poor counties.
Publications (written in Korean)
Estimation and Forecast of Korean Government Bond Yields Using a Macro-Finance Model of Term Structures (with Byungin Lim and Yongil Jeon), Journal of Korea Economy Study, 31(2), pp.187-210, 2013.
A study on the Effective Inspection of Workplace Conditions (with Yongil Jeon), Quarterly Journal of Labor Policy, 12(2), pp.165-187, 2012.
Vulnerability Index for Workplace Spot Inspection (with Sungtai Kim and Yongil Jeon), Journal of the Korean Labor Economics, 34(3), pp.29-58, 2011.
Reason for the Slow Adoption of Retirement Pension Plans by Korean Firms (with Doobin Yim and Yongil Jeon), Pension Study, 1(2), 2011.
A study on the Effective Inspection of Workplace Conditions (with Yongil Jeon), Quarterly Journal of Labor Policy, 12(2), pp.165-187, 2012.
Vulnerability Index for Workplace Spot Inspection (with Sungtai Kim and Yongil Jeon), Journal of the Korean Labor Economics, 34(3), pp.29-58, 2011.
Reason for the Slow Adoption of Retirement Pension Plans by Korean Firms (with Doobin Yim and Yongil Jeon), Pension Study, 1(2), 2011.
Working Papers
"The effect of automation on earning distribution: does it increase inequality?"