Skip to main content

The growth of US science and technology parks: does proximity to a university matter?

Science and technology parks are seen increasingly as a means to create dynamic clusters that accelerate economic growth and international competitiveness

—National Research Council


In this paper, we present a generalized model of US university science and technology parks, and we identify covariates that might serve as target variables not only to perpetuate the growth of existing parks but also to provide information for those nations, regions, and universities starting new parks. Relevant covariates are the distance between the park and the university and if the park was founded during the information and communications technology (ICT) revolution (post-2000).

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2


  1. See, See also, UNESCO-WTA (2015).

  2. Additional important target variables are provided in Link and Scott (2006); unfortunately, several of the controls available for that paper are not available in the present study.

  3. See,

  4. See,

  5. See,

  6. See,

  7. As explained in Link and Scott (2003, p. 1337), the estimated parameter a provides an estimate of the efficient start-up scale for a park. For all of the models estimated below, the natural logarithm of the parameter is very well estimated to be about 5, and so the start-up scale is estimated, apart from random error, to be about 150 employees. Observe that, although we do not think it is necessary to do so in the present paper, the parameter a can be made a function of any variables that the modeler believes to be an important determinant of the start-up scale for a park.

  8. Some of the variables among the explanatory variables are powers or interactions of other variables, so the function will be linear in the parameters estimated but nonlinear in the variables.

  9. Link and Scott (2003, 2006) estimated a model similar to that in Eq. (5) using a limited sample of US science and technology parks. Among other things, the purpose of those papers was to estimate the average annual growth rate of US parks in an effort to approximate their overall productivity. Below we use Eq. (5) on a larger and up-to-date sample of US parks, and we estimate the life cycle of park growth; however, several variables that proved to be important in Link and Scott (2003, 2006) are not available for the present study; subsequently, we discuss some of the implications for our conclusions.

  10. We thank an anonymous reviewer for suggesting other performance variables besides employment. He/she is correct that park performance is heterogeneous and not homogeneous at a point in time (Dimaggio and Powell 1983), and no one metric is applicable to all parks.

  11. It is important to point out that our analysis emphasizes employment growth in a spatial context. Our analysis is not an evaluation of either the private benefits to a firm from locating in a park or the social benefits to the region for having a park. But, there is literature related to the evaluation of parks that underscores the importance of the mission statement of the park when considering its performance. We appreciate an anonymous reviewer emphasizing this point. However, we have no information on park mission statements; Bigliardi et al. (2006) provide an excellent and important template, based on Italian parks, for future case studies of US parks.

  12. To the best of our knowledge, there is no national compilation of data on STPs in Europe or Asia. However, the Organisation for Economic Co-operation and Development (OECD 2013) does chronicle the activities of university and non-university STPs through case studies and commissioned research. The closest counterpart to such OECD efforts in the USA is funded research by the National Science Foundation, although the National Science Board (biannual) does not regularly consider STP activities in its biannual publication, Science and Engineering Indicators.

  13. Figure 1 is not identical to the comparable figures in Link and Scott (2003, 2006). Figure 1 is constructed on the basis of the most recent park information available. It is not uncommon for historians of science and technology parks to have different factual information on the date a park was founded emphasizing that archival data are not always available on the date a park was chartered, and institutional memories vary over time. For example, we found two founding dates for Cornell Business and Technology Park in New York—1951 and 1952. The source for the 1951 date is used in Fig. 1 because it came from a more recent source.

  14. This paper, although in some dimensions a sequel to Link and Scott (2003, 2006), is, to the best of our knowledge, the only other paper to investigate empirically covariates of STP growth at a national level (Hobbs et al. 2017).

  15. See,

  16. Our dating of the start of the so-called ICT revolution as being after 2000 is not at odds with OECD data on the growth of total communication access paths in OECD countries as a group and in the USA. See, See also Table 2.6 at:

  17. See, We have not included variables capturing the characteristics of the regional economy (such as orientation toward particular sectors of the national economy, regional GDP, and regional access to venture capital). We did include the regional dummies, and as shown in Sect. 4, they are not significant. If they were significant, we could replace them with variables defined at the regional level of aggregation and explore the differences across regions that would have caused the significance of the regional dummies. However, the regional dummies are not significant; thus, any variance across the regions in variables such as the regional GDP does not appear to be important in the context of the questions that we ask. However, it is certainly possible that variance in certain variables is important when the variables are defined across subsets of regions, such as the individual states. However, the organizations that occupy the parks typically come from many different states, supporting the view of the census region as the relevant “local” area. We have, however, considered, in addition to the regional effects (and hence the effects of any variables defined at the regional level of aggregation), the possibility that variables defined at the national level have different effects within each region (see footnotes 26 and 27).

  18. Our sample of parks (\(n=106\)) differs from what might ostensibly be referred to as the population of US parks in Fig. 1 (\(n=146\)) in the fact that the parks in our sample are slightly younger when park age is defined as (2016—year the park was founded). For a comparison, consider the following descriptive statistics:

  19. The range of employees in 2016 in Table 1 does not define outliers in our sample of parks because parks were founded in different years. For example, Research Triangle Park in NC had 46,000 employees in 2016 and was founded in 1959; Texas Research Park, with 12 employees in 2016, was founded in 2004. The t variables are included in our model to account for nonlinear growth.

  20. A number of empirical studies have also shown that the survival rate and the productivity of firms in science and technology parks in the UK is greater than that of comparable off-park firms (Westhead 1995; Westhead and Cowling 1995; Westhead and Storey 1994, 1997; Westhead et al. 1995; Siegel et al. 2003; Link 2016).

  21. Feldman and Kogler (2010) provide an excellent overview of the geography of innovation literature.

  22. We know from the economically and statistically significant t variables (see footnote 25) that parks do not on average stay the same size. However, observe that if one interjected the counterfactual of no growth for parks, one would expect, for the reasons that we have stated, the positive estimates for the coefficient on the variable ICT. Put differently, even if one did not need to control for time because parks have a life cycle, it would be important to control for ICT and capture its effect mitigating the negative effect of distance of the park from a university.

  23. We explored the possibility that distance might enter the regression model nonlinearly, but in such an alternative specification the coefficient on (\(distance^{2}\,x\,t)\) was completely insignificant statistically. For example, adding the additional variable to the specification (2) in Table 2, the estimated coefficient on (distance x t) was still negative and essentially the same size (\(-0.00087\) rather than \(-0.00088\)). The insignificant coefficient on the squared distance term was negative and very small, implying that the negative impact of distance increased with distance, although the result is wholly insignificant statistically. The results for the other variables in the model are qualitatively the same. These results are available from the authors on request.

  24. UNESCO (2013) also discusses the role of STPs in the context of being a permanent growth policy in a number of developing and less developed countries.

  25. As a group, the estimated coefficients on the three t variables are highly statistically significant. For example, for specification (2) in Table 2, against the null hypothesis that the three coefficients are zero, (F(3, 94) = 6.90, \(\hbox {Prob} > F = 0.0003\)). The econometrics log producing these results is available from the authors on request.

  26. We estimated the model in specification (2) with interaction terms for ICT and the regional dummies to see whether there is regional variation in the effect of ICT. The effect is more positive in the northeast region, although the result is significant at only the 15% level. The performance of the other variables is essentially the same; the model with the regional interaction terms is available from the authors on request.

  27. We also estimated the model in specification (2) with interaction terms for propGR and the regional dummies to see whether there is regional variation in the effect of propGR. The effect is less negative in the west region, although the result is significant at only the 11% level. The performance of the other variables is essentially the same, and this model is also available on request.

  28. Regarding the availability of the additional control variables, as compared with the earlier papers, we had little success this time getting park directors to respond to inquiries.

  29. These results are available from the authors on request.

  30. The four parks with the longest distance from their universities do not appear, apart from the distance variable, to be special cases within their regions. One of the parks is the Clemson University International Center for Automotive Research in Greenville, South Carolina, only 9 miles from the Greenville/Spartanburg Airport. Clemson University is 48.5 miles from the airport. Another is Russ Research Center in Beavercreek, Ohio, while Ohio University is 94.4 miles away in Athens, Ohio. The land for the research center was a gift from the Russ family and is on the location of their family business. Reese Technology Center is on the grounds of the former Reese Airforce Base in Lubbock County, Texas, and the airport is important to the tenants. The Reese Technology Center supports Texas Tech University Health Science Center that is 126 miles away in Amarillo, Texas. There are three such supported Texas Tech centers, but the one in Amarillo is closer than the ones in El Paso and Odessa (we used the closest university node). Treasure Coast Research Park is on the east coast of Florida about (depending on the route taken) 215 miles from the University of Florida in Gainsville and was built there to take advantage of the US Department of Agriculture being an anchor tenant. The research done on the park is agricultural. The interpretation of the finding about distance certainly remains open because something else—other than distance from the university centers that they support—may explain their underperformance. See footnote 17 for discussion of what such factors might be; future research may discover such additional explanatory factors.

  31. One could, rather than use the model in this paper, certainly support a model-free examination of the available, limited information about the parks. Yet we believe that the model adds a great deal. The life-cycle model is significant; the specifications using t, \(t^{2}\), and \(t^{3}\) are needed to show what is clearly a very highly significant life cycle for employment growth for the parks. Despite not always being significant individually, the t, \(t^{2}\), and \(t^{3}\) terms are significant economically and statistically; as reported above, together their statistical significance is very high. The variable numuniv is significant, when the life-cycle model is controlled, at about the 12% level for the conservative two-tailed test. With the complete life-cycle model, the variables ICT and propGR are significant. The model lets us see the significant negative effect of distance and also lets us discover that in the present sample that effect depends on the four observations that are farthest from their universities. In all, despite a lack, as discussed above, of some of the variables used in earlier papers, we believe the model performs well and is necessary to make the points that we make in the paper.


  • Acs ZJ, Audretsch DB, Feldman MP (1994) R&D spillovers and innovative activity. Manag Decis Econ 15:131–138

    Article  Google Scholar 

  • Adams JD, Jaffe AB (1996) Bounding the effects of R&D: an investigation using matched establishment-firm data. RAND J Econ 27:700–721

    Article  Google Scholar 

  • Audretsch DB, Feldman MP (1996) R&D spillovers and the geography of innovation and production. Am Econ Rev 86:630–640

    Google Scholar 

  • Battelle Technology Partnership Practice (2013) Driving regional innovation and growth. Battelle, Columbus

  • Bigliardi B, Dormio AI, Nosella A, Petroni G (2006) Assessing science parks’ performances: directions from selected Italian case studies. Technovation 26:489–505

    Article  Google Scholar 

  • Dimaggio PJ, Powell WW (1983) The iron cage revisited: institutional isomorphism and collective rationality in organizational fields. Am Sociol Rev 48:147–160

    Article  Google Scholar 

  • European Commission (2014) The role of science parks in smart specialisation strategies, S3 Policy Brief Series, No. 08/2014, Seville

  • Feldman MP, Kogler DF (2010) Stylized facts in the geography of innovation. In: Hall BH, Rosenberg N (eds) Economics of innovation. Elsevier, Amsterdam, pp 381–410

    Google Scholar 

  • Gallaher MP, Link AN, Petrusa J (2006) Innovation in the U.S. service sector. Routledge, London

  • Gertler MS (2003) Tacit knowledge and the economic geography of context, or the undefinable tacitness of being (there). J Econ Geogr 3:75–99

    Article  Google Scholar 

  • Hardin J (2013) “North Carolina’s innovation ecosystem,” presentation to the committee on science, engineering, and public policy. National Academy of Sciences, Washington

    Google Scholar 

  • Hobbs KG, Link AN, Scott JT (2017). Science and technology parks: an annotated and analytical literature review. J Technol Transf 42(4):957–976

  • Jaffe AB (1989) Real effects of academic research. Am Econ Rev 79:957–970

    Google Scholar 

  • Leyden DP, Link AN (2015) Public sector entrepreneurship: U.S. technology and innovation policy. Oxford University Press, New York

    Book  Google Scholar 

  • Link AN (2016) Competitive advantages from university research parks. In: Audretsch DB, Link AN, Walshok ML (eds) The oxford handbook of local competitiveness. Oxford University Press, New York, pp 337–344

    Google Scholar 

  • Link AN, Link KR (2003) On the growth of U.S. science parks. J Technol Transf 28:81–85

    Article  Google Scholar 

  • Link AN, Scott JT (2003) U.S. science parks: the diffusion of an innovation and its effects on the academic missions of universities. Int J Ind Organ 21:1323–1356

    Article  Google Scholar 

  • Link AN, Scott JT (2006) U.S. university research parks. J Product Anal 25:43–55

    Article  Google Scholar 

  • Link AN, Scott JT (2007) The economics of university research parks. Oxf Rev Econ Policy 23:661–674

    Article  Google Scholar 

  • Maskell P, Malmberg A (1999) Localized learning and industrial competitiveness. Camb J Econ 23:167–185

    Article  Google Scholar 

  • National Research Council (2009) Understanding research, science and technology parks: global best practices. National Academy Press, Washington

  • National Science Board (biannual) Science and engineering indicators. National Science Foundation, Arlington

  • Organisation for Economic Co-operation and Development (OECD) (2013) Regions and innovation: collaborating across borders. OECD Publishing, Paris

  • Polanyi M (1958) Personal knowledge: towards a post-critical philosophy. University of Chicago Press, Chicago

    Google Scholar 

  • Polanyi M (1966) The tacit dimension. Doubleday, New York

    Google Scholar 

  • Siegel DS, Westhead P, Wright M (2003) Assessing the impact of science parks on research productivity: exploratory firm-level evidence from the United Kingdom. Int J Ind Org 21:1357–1369

    Article  Google Scholar 

  • United Nations Educational, Scientific and Cultural Organization (UNESCO) (2013) UNESCO science report: Toward 2030. UNESCO, Paris

  • United Nations Educational, Scientific and Cultural Organization and World Technopolis Association (UNESCO-WTA) (2015) Science park and technology business incubator UNESCO-WTA initiatives (2006–2010). UNESCO-WTA, Paris

  • Westhead P (1995) New owner-managed businesses in rural and urban areas in Great Britain: a matched pairs comparison. Reg Stud 29:367–380

    Article  Google Scholar 

  • Westhead P, Cowling M (1995) Employment change in independent owner-managed high-technology firms in Great Britain. Small Bus Econ 7:111–140

    Article  Google Scholar 

  • Westhead P, Storey DJ (1994) An assessment of firms located on and off science parks in the United Kingdom. HMSO, London

    Google Scholar 

  • Westhead P, Storey DJ (1997) Financial constraints on the growth of high-technology small firms in the U. K. Appl Financ Econ 7:197–201

    Article  Google Scholar 

  • Westhead P, Storey DJ, Cowling M (1995) An exploratory analysis of the factors associated with the survival of independent high-technology firms in Great Britain. In: Chittenden F, Robertson M, Marshall I (eds) Small firms: partnerships for growth. Paul Chapman, London

    Google Scholar 

  • Wolf M, Terrell D (2016) The high-tech industry, what is it and why it matters to our economic future. Beyond the Numbers 5:1–6

    Google Scholar 

Download references


We thank an anonymous reviewer for many helpful comments and suggestions that have greatly improved the paper.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Albert N. Link.

Rights and permissions

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hobbs, K.G., Link, A.N. & Scott, J.T. The growth of US science and technology parks: does proximity to a university matter?. Ann Reg Sci 59, 495–511 (2017).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:

JEL Classification