The story that American children learn of their nation is one of adaptation to adversity, expansion into the unknown, and overcoming obstacles—in a word: innovation. Since the United States declared its independence 239 years ago, the world has changed significantly, but innovation and invention remain at the heart of the American cultural mythos. Intellectual property protections that encourage and give substance to innovation have been with Americans since at least the 1789 Constitution, and the patent system has since shown itself to be an increasingly significant driver of the American economy.
The states of the Union encompass a remarkable breadth of demographic, geographic, social, and cultural variety, and inventiveness is just as variable. Inventiveness is a broad concept to be sure, but one tractable measure of inventiveness is the number of patents applied for and the rate at which those applications are allowed. Patents come in three broad varieties: utility patents, plant patents, and design patents. While design patents are innovative, we will focus on plant and utility patents because they protect new things and new ways of making and doing things, while design patents only protect ornamental designs.
To find our measures of innovation and inventiveness, we first found the state from which an inventor hailed. We then aggregated the total number of applications, the number of disposed applications (the number allowed or abandoned), and the number of allowed applications coming from each state (what we will call “allowances” or “allowed applications”). We were interested in discovering the difference in inventiveness between the states, as well as finding out whether (and what) major demographic and educational benchmarks correlated with a state’s inventiveness. We selected a few social, educational, and demographic measures from the National Science Foundation’s Science and Engineering state profiles, the US Census Bureau, the National Center for Education Statistics, Education Week, and Juristat’s own database of patent prosecution data., 
Looking at raw numbers, California was by far the most inventive state, with over 329,000 non-design patents filed, over 254,000 disposed, and over 180,000 allowed between 2000 and 2010. Texas, New York, Massachusetts, and Washington rounded out the top five, but together had fewer applications, disposed applications, and allowed applications than California alone. The states with the highest allowance rates were Vermont (84%), Idaho (77%), Kansas (74%), Washington (74%) and Iowa (73%).
The least inventive state by raw numbers was Alaska, with only 329 non-design applications, 265 disposals, and 174 allowances between 2000 and 2010. Wyoming, South Dakota, North Dakota, and the District of Columbia joined Alaska at the bottom of the list. Together these five states had just under 3,800 disposed utility patent applications, and just under 2,500 allowances. These seemingly low numbers should not be entirely surprising, as these five states, along with Vermont, were the smallest in terms of population. However, the five states with the lowest allowance rates were the District of Columbia (57%), Montana (59%), Maryland (59%), Nevada (60%), and Delaware (62%).
Given the disparity in population sizes, in order to get a better look at how innovative the states have been, we found the number of applications, disposed applications, and allowed applications per 1,000 residents (APK for short). Idaho, at over 10 APK, had the largest number of applications and disposed applications per 1,000 residents, though Vermont was the most successful with over six APK allowed. Along with Vermont, Idaho (5.74 APK), Massachusetts (4.87 APK), California (4.85 APK), and Washington (4.68 APK) had the most allowances per 1,000 residents. Arkansas (0.34 APK), Mississippi (0.34 APK), Alaska (0.34 APK), West Virginia (0.43 APK), and Hawai’i (0.49 APK) had the fewest allowances per 1,000 residents.
The significant variability between the inventiveness of the different states begs the question of what drives those differences, or at least what factors might contribute to those differences. We looked at a number of demographic, educational, and economic indicators and performed simple correlational calculations to see whether any one factor was predictive of inventiveness. Our correlations (simple Pearson’s r) are reported on a scale from 1 (perfect direct correlation) to -1 (perfect inverse correlation), where 0 means there is no meaningful relationship.
The raw number of applications (.88), dispositions (.88), and allowances (.88), strongly correlated with a state’s population; however, the allowance rate had no meaningful correlation with average population (.08). A state’s average GDP was strongly correlated with the numbers of applications, dispositions, and allowances (all near .91), though there was virtually no correlation between GDP and allowance rates (.07). Interestingly, a state’s population density showed a -0.39 correlation with allowance rates, so it would seem that the more densely populated a state is, the less likely it is to have a high allowance rate. Surprisingly, average individual income also showed an inverse correlation with allowance rate, though the correlation was much weaker (-0.24). Average individual income was, however, positively correlated with the number of allowances (.29) and dispositions (.35) per 1,000 residents, and more modestly correlated with the raw numbers of applications (.20), dispositions (.20), and allowances (.19).
Tracking with population size, the number of science, engineering, and healthcare PhDs granted in the states correlated strongly with the numbers of applications, disposals, and allowances (all near .90), but was virtually uncorrelated with the allowance rate (.06). The percentage of a state’s population with an advanced degree of some kind (Master’s or above) was inversely correlated with allowance rates at (-0.27), as was the correlation between allowance rate and percentage of a state’s population with a Bachelor’s degree or higher (-0.13) (based on the 2009 American Community Survey from the US Census Bureau). However, the District of Columbia is an extreme outlier, with the lowest allowance rate (57%) but the largest percentage of the population with a Bachelor’s degree or higher (48.5%) and Master’s degree or higher (28%). When we removed DC from the calculation the correlation between Master’s degree or higher and the average allowance rate was only -.06, and the correlation between allowance rate and the percentage of Bachelor’s degree holders rose to .06. Application, disposition, and allowance rate correlations with the percentage of university degree holders also rose to about .25 and .24.
High school graduation rates more strongly correlated with allowance rates (.36) than PhD conferrals or college degrees, though, like PhD conferrals, there was no meaningful correlation with the numbers of applications, dispositions, and allowances. Average National Assessment of Educational Progress mathematics tests were almost equally correlated with allowance rates (.36) as were the high school graduation rates, and, like the high school graduation rates, were nearly uncorrelated with application, disposal, and allowance numbers (around -0.06). NAEP reading test scores also positively correlated with allowance rates, though to a lesser extent (.26).
Overall, our data makes clear that education is not the main driving force behind innovation in the United States, although more education likely increases the chance that a patent will be granted. Population density likewise is not a driver of innovation, in contrast to some theories that suggest that denser populations lead to more interactions of people and ideas and, thus, to more innovation. It is clear that the more people a state has, the more patents will be filed, disposed, and allowed, but just as clear that there is no relationship between a state’s population and its success rate at the USPTO. Our data is, admittedly, incomplete, as some other factors may explain the variability in innovation between the states, but, as far as we can tell, no single factor drives innovation other than an inventor’s desire to solve a problem or to create something.
Invention and innovation are at the core of economic growth and productivity, and encouraging these ideals is essential to any nation’s economic future. We were unable to find any single educational or demographic factor that reliably and greatly encourages innovation, but we can see that primary and secondary educational achievement have some measureable effect on the success of patent applications. Future research should more deeply investigate the effect of education on innovation by looking to teacher quality, spending per pupil, and other factors to determine what – if any – factors encourage innovation. Future research should also be done to find the age of patent applicants so that cohort analyses can more confidently be conducted.
It is (patently) clear that the United States is an inventive country. We should strive to discover what leads to that innovative spirit and try to foster innovation as much as possible, as efficiently as possible.
 We chose to look at a number of benchmarks including:
The number of science, engineering, math, and healthcare PhDs conferred in each state;
The state’s 2010 US Census population;
The state’s average personal income over 2000-2010;
The state’s average GDP over 2000-2010;
The state’s average high school graduation rate from 1996-2006;
The percentage of the states’ populations with Bachelor’s degrees or higher as of 2009;
The reading and mathematics test score averages from the National Assessment of Educational Progress from 1992-2000.
 We staggered the time series in our benchmarks in order to as much as possible see an influence from those benchmark factors in the broader patent data. We assume that few patents are filed by anyone younger than 22 and restricted our patent application period to the years 2000-2010. These limits mean that we had to find primary and secondary school data for years in which a 22-year-old would have been in the tested cohort. We therefore used 1996 as the first year in which we looked at high school graduation rates and 1992 as the earliest NAEP year because 4th, 8th, or 12th graders tested would have been 22 years old sometime between 2000 and 2010, and most individuals who turned 22 between 2000-2010 would have graduated from high school between 1996 and 2006.
 The NAEP tests 4th, 8th, and 12th grade students, though we excluded 4th graders from our data as only the 1992 NAEP 4th grade cohort would meet our age window.