Small Business Programs, Innovation, and Growth: Estimating Policy Effects Using Comprehensive Firm-Level Panel Data

Submitted by NSFAdmin on

Award Number: 

This project investigates the growth of start-ups and young firms, which are generally believed to be the most important job-creators, and evaluates the probability of becoming a high-growth productivity leader in an industry. The project provides direct measures of innovation and research and development (R&D), firm-level labor and total factor productivity, the dispersion of firm productivity and performance, the growth of high- vs. low-productivity entrants and incumbents, and exit rates of under-performing firms. With a specific focus on the Small Business Administration (SBA) 7(a) and 504 loan facilities, the analysis provides evidence on the importance of financial constraints, and the degree to which they may be attenuated by major government programs. The results will shed light not only on direct program impacts but also consider the indirect effects on productivity-enhancing reallocation of resources. In particular the central issue as to whether the policy intervention promotes growth of productive but financially constrained businesses, or instead sustains low-productivity firms that should release their resources to superior uses. The results have policy implications for the growth, innovation, and productivity of the U.S. economy.

The project relies on new concepts, methods, and data for estimating the effects of financial access and small business programs. Conceptually, the project draws on models of firm dynamics to analyze innovation as the result of experimentation by entrants and incumbents, thus as part of the creative destruction process. A fundamental insight is that the full distribution of outcomes is useful for revealing innovation, as experimentation raises dispersion and selection raises skewness. The empirical methods include a credible identification strategy based on panel-matching and instrumental variables capturing geographic variation in SBA loan access. Estimation includes not only average treatment effects for productivity and innovation measures but also other moments of the outcome distribution: overall dispersion and skewness, performance at top percentiles, and allocation and selection effects that work through growth and exit at different points in the distribution. The methods provide an original way to assess the impact of a government intervention on productivity-enhancing reallocation. The data work links SBA administrative data with Census Bureau universal data on all U.S. employers. Variables from the Business Register, Census of Manufacturers, and Annual Survey of Manufacturers are added to measure productivity. Linking to the new Annual Survey of Entrepreneurs permits direct measurement of innovation and R&D effort. The conceptual approach, identification strategy, methods of analysis, and data development of this project should have positive externalities for other research including the evaluation of other programs in the U.S. and around the world.

Principal Investigator: 
John Earle
George Mason University
PI Email Address:
NSF Organization: 
NSF Directorate: 
Start Date: 
Jul, 01 2017
Last Amendment Date: 
Mar, 20 2017
Expiration Date: 
Jun, 30 2020
Award Instrument: 
Continuing grant
Program Manager: 
Maryann Feldman
Awarded Amount To Date: 
Organization Street: 
Organization City: 
Organization State: 
Organization Zip: 
Organization Phone: 
Program Element Code(s): 
Program Reference Code(s): 
ARRA Amount: