The 9 Most Important States for Entrepreneurship

• John Vandivier

This article describes some preliminary results from an ongoing <a href="https://en.wikipedia.org/wiki/Exploratory_data_analysis">EDA. Specifically, we look at the 10 states which are significantly associated with the rate of self-employment.

Intro

I've been conducting an EDA as part of an econometrics class in my Ph.D. program and I thought I would blog it. Readers will learn specific techniques in data analysis and also more generally how to interpret results. There will be at least 5 articles in this series and this article is the first.

The data analysis is currently being conducted in STATA using the <a href="https://www.census.gov/programs-surveys/acs/data/pums.html">ACS 2014 1-year PUMS data set from the US Census Bureau. You can replicate the analysis using the do files and instructions found on my <a href="https://github.com/Vandivier/data-science-practice">GitHub project.

The central research question under investigation is, "Do workers in the IT sector have a higher rate of self-employment?" While the paper for class is focused on that question, here on the blog I will deviate and explore anything I find interesting in the course of research. For example? Today's topic.

The 9 Most Important States for Entrepreneurship

There are 9 states which have a coefficient which is significant at the 2% significance level in the prediction of the rate of self-employment at the individual level. 6 states have a negative correlation and 3 states have a positive correlation:

  • Good States: Georgia, Minnesota, and South Dakota
  • Bad States: Alabama, Colorado, Hawaii, New York, North Carolina, Tennessee, and Utah
There is a huge caveat here: The regression run to determine which states are significant is corrected for a variety of factors other than wealth and income. Wealth and income are obviously important and they also interact with the state variable due to local differences in the cost of living, income, and costs associated with running a business like corporate taxes and so on.

Later in the series I will discuss the change to these 9 states after correcting for income and wealth.

The specific regression is in the .do file on GitHub and it looks like this: "reg self isit sciengrlp agep agesquared ismale schl pwgtp _marriage2 _state* _cit2"