Technology companies and investors that are heavily ingrained in R&D often look to inventive university partners to head-up their research before bringing candidate products in-house to be further developed. This saves time and money on infrastructure and talent, among other benefits. However, finding a university that has the expertise and capabilities to deliver can be risky and time-consuming, potentially leaving you on shaky ground going into commercialization.
When you consider patent output alone, it's not surprising that the usual suspects that rise to the top ranks are heavily concentrated near academic research centers like MIT (MA) and Stanford (CA). But patent output is only a single feature of the intellectual property (IP) of a university research system; publication output must also be considered. Together, patent and publication output capture overall research activity and are common proxies for Inventiveness. Pairing these outputs with known inputs allows you to assess how well a potential research partner performs at converting ideas into viable, comercializable products.
R&D Spend is the Driver of Inventiveness?
If patents and publications are considered as outputs, what defines the input to this system? Broadly speaking, universities rely on external dollars to fund the cost-intensive academic endeavors required to rigorously discover, explore, and pilot-test the feasibility of potential market solutions. The products of such inventive programs are the precursors to innovation: solutions waiting to solve industry problems and market needs. True innovations are made when these solutions are coupled to one of many applications and then successfully brought to market.
We initially expected that inputs and outputs would move in relation to each other, showing a crisp connection between a univeristy's spend proporational to their Inventiveness output. Our research described below suggests otherwise - it's actually the relationship between these two that gives a higher resolution assessment metric.
To test our expectations, we compiled an initial sample of US universities with data sourced from the NSF HERD Survey1 (Higher Education Research and Development). We then ranked the Top 30 of all 640 US universities receiving funding, based on total federal R&D expenditure for the most recent 10-year period. From here, we were able to illustrate how R&D spend is allocated across the US.
The top spender over the 10-year period is California with $56.5B, or 24.3%, of the total spend of the top 30. It is possible that this is an unbalanced result because of the tier structure used, however, California uses 9.7% of all R&D sped for all 640 universities included. The next big spenders are MD and TX with just over $19B each (8.4% and 8.2% of top 30, respectively); NC and PA finish the top 5. Keep in mind, this does not accound for the number of universities per state. Here, we just show the total R&D spend for the top 30 universities.
Hot Spots of Inventiveness are Diffused Across the US
For the sake of being thorough, look at publication and patent putput, individually. Try to see if you can discern which of these best indicates how well a university research partner would deliver. (Answer: there's no clear pattern, but there is a shift.)
For publication output: CA, MA, NY, TX, and IL dominate as the top 5 states2.
For patent output: MA, CA, IL, NY, and PA take the lead - a slight rearrangement in the top 5 states3, when compared to publication output.
Each top-tier state has unique strengths and empasizes different types of activities. Some publish their work while others patent. Others do both very well.
Ultimately, we are interested in a university's inventiveness and their role in bringing transformative technologies to market. To do this, we need to understand the relationship of R&D spend to IP output for each using a conversion ratio. The conversion of R&D to IP is an analogue to a metric established in systems engineering circa 1824 (back when there were only 24 states in the Union). Here, this metric works effectively by normalizing an entity's conversion of cash assets to IP assets4.
Based on this conversion ratio, new players emerge as top performers, where the states that spend the fewest dollars per IP asset are MA, CT, WA, MI, and IL. Based on the maps described above, we would expect states like California and Massachussetes to dominate the top of this list as well. In ranking states by the conversion ratio, MA maintains dominance, while California drops to the middle of the pack. Of particular interest, and an illustration of the use of this method, Washington ranks in the top 5 as an efficient spender. Based on R&D input or patent/publication output alone, we may not have considered WA, however, given this new information, it may be a state to look toward as someone is considering potential research partners.
Getting the Biggest BANG out of Your Investment
Patent and publication output, alone, cannot sufficiently capture the inventiveness of a research system. Instead, to best qualify a potential research partner, it is useful to also know how well they convert investment dollars into IP assets. A clearer picture of the inventive landscape will reveal previously unforseen technology development opportunities and strategic research partners.
This broad, foundational view gives a clearer juxtaposition of universities as systems that measurably covert investment dollars into IP assets. Moreover, by applying a conversion metric, new institutions begin to show promise as strategic R&D partners who can efficiently utilize resources to maximize IP coversion. Our method can be further customized to evaluate an institution on a case-specific basis.
If you have reached this point, you're serious about making the best-infomed decision before launching into a research venture. This is how you approach it and end up ahead of the game.
Using an IP Conversion metric is a powerful decision-making tool for evaluating investments. Here, we apply it as an illustrative case study where university research organizations are the converting systems. This can further be applied to evaluate any strategic research partner, whether they are in academia, government, or industry (from start-ups to established companies).
Data Sources & Notes. Because Scientific Research.
3We crafted a custom patent search procedure to include patents by assignee (i.e., the university of interest) during the period between 2006-2015. The resulting dataset was further filtered to remove any anomalies and errors in assignee designation. A note about the University of California System: an assignee that included the University of California and University of California plus "location" (i.e., UC Berkeley) was included if at least one owner address was located in reasonable proximity to the university under investigation.
4Method for conversion: we normalized R&D expenditure with respect to the cumulative number of patents and publications. Defining R&D expenditure as the driver/input and patent and publication number as the outputs, we can compute the output-input ratio, an established metric of efficiency. To get to the targeted conversion metric, we can simply take the reciprocal, to determine the number of dollars spent per IP asset.