This dissertation is the first study to provide evidence of the effects state-level research and development tax credits have on firm research choice behavior. It is also the first on the federal or state level to provide evidence of the effectiveness of the alternate form of R&D tax credit. Based on a unique data set developed for this study, econometric modeling of firm R&D investment is tested with data from two sources on a sample of pharmaceutical and software firms headquartered in California or Massachusetts. As of 1997, California offered increased R&D standard computation tax credit rates for basic and in-house R&D and instituted the alternate form of credit. Massachusetts has made no change in its tax R&D credit rates, nor does it offer the alternate form of credit. The before (1994–1996) and after (1997–1999) R&D choice of firms in the two states is compared. There are two principle findings from the analysis. First, firms in the same industry but different located in different states face statistically different research tax prices. The R&D expenditure choice of the pharmaceutical firms in California applying the standard form of credit computation is consistent with the increased state-level R&D tax credit rates affecting firm R&D choice. The pharmaceutical results are robust to changes in computational assumptions and firms tested. Within the same state, firms in the two industries faced different tax prices and exhibited different R&D investment behavior. The R&D choice behavior of the software firms is consistent with the availability of the alternate credit leading to increased R&D expenditure as well. The outcomes highlight the need for research policy analysis to be performed at the industry level. Second, the model's results are robust differences in data source. The estimates obtained from the 10-K and Compustat data were similar but the Compustat results were biased upward. This suggests that studies based on Compustat data may overstate firms' sensitivity to research tax credit policy.
展开▼
