Carow Hall, Conference Room
July 17, 2017, 02:00 PM to 04:00 PM
Chapter 1 of my dissertation is a study of a federal infrastructure program. I examine cost forecasts for 49 fiber-optic and wireless networks funded by the Recovery Act. I find that grantees did not systematically underbid for projects. They did, however, underestimate and overestimate costs with nearly equal frequency, and escalate costs on average by 202 percent in cost per institution and 37 percent in cost per fiber mile. For four outputs, I ask whether cost forecasts predicted actual construction. I find that grantees overestimated fiber miles and indirectly connected institutions, while meeting targets for directly connected institutions and points of interconnection. I do not find budget, technology, or institution effects to explain the quality of cost forecasts in these outputs. My findings suggest that low-price bids can be cost-effective across project types. With data from the applicant pool, I compare offers from 116 projects with 657 rejected proposals.
Chapter 2 of my dissertation focuses on a universal service program for broadband subsidies to schools and libraries. Broadband in schools has been financially supported by the E-rate program for over fifteen years in the United States. This study focuses on distribution effects of priority 2 internal connections funds from 1998 to 2012. Regressions estimate the effects of the discount matrix and effects of the National School Lunch Program (NSLP) student population on fund distribution by state. Regressions also provide per-student and per-school estimates of funds from the number of city, suburb, town, and rural students and schools per state. Treatment groups provide more detailed comparisons across the 50 states and the District of Columbia, a smaller sample of 48 states excluding New York, California, and Texas, and a quasi-experiment from FY2010. The effect of the discount rate is statistically significant, along with the number of NSLP students in the treatments. Results show that NSLP students in city locales account for a large proportion of funds directed toward New York, California, and Texas. Given these results, E-rate reformers may consider student population demographics implicit in the discount matrix to evaluate distribution of priority 2 funds.
Chapter 3 of my dissertation studies technology adoption of converter boxes subsidized by the Digital Television Transition of 2005. I find that broadband adoption explains participation in the coupon program at the zip code level. Broadband adoption may have facilitated coupon requests, but the opposite may be true as well. Coupon seekers might have adopted broadband for access to savings. In regression analysis, I also control for over-the-air television households, coupon redemption rates, broadband availability, location affordability, income, population density, age, and dependents per capita. My study does not solve the endogeneity problem, but suggests a preference explanation for broadband adoption. Broadband adopters may use the internet to seek coupons, deals, and opportunities. By finding a relationship between coupon participation and broadband adoption, I add to the digital divide literature and suggest a question for further study. I provide empirical analysis of a campaign that spent 160 million dollars to distribute 1.34 billion dollars in coupons redeemable in electronics stores nationwide.