How does knowledge spur economic growth? The
contribution of universities and the importance of
geographical space
T. Agasisti1C. Barra2R. Zotti3
1Department of Management and Industrial Engineering - Politecnico di Milano 2Department of Economics and Statistics-University of Salerno 3Department of Economics and Statistics “Cognetti De Martiis” - University of Turin
XIX International Academic Conference on Economic and Social Development -April 10th- -April 13th 2018 - Moscow
Summary
Motivation, aims and contribution Main findings
The role of universities and some background Data
Empirical models and analysis Results
Motivation, aims and contribution
Motivation
Universities contribute to the social, economic and cultural development of the areas where they operate
Positive role in the interaction with key stakolders with the final aim of transferring knowledge, disseminating culture and foster economic competitiveness
Universities are important engines of regional and national economic development
Outcomes not confined to the boundaries of academia but wider aspects of society and economy are involved
Motivation, aims and contribution
Aims
Examine the role of universities in sustaining local economic development as institutions able to supply education, knowledge and services
Paying attention to:
Knowledge transfer through education and human resources development (Florida et al. 2008; Barra and Zotti, 2016)
Societal impact of research as the ability to transform knowledge into economically relevant product, services and processes (Bornmann, 2013) New set of activities through which HEIs interact with the community (Hayter et al. 2016; Etzhowitz, 2013)
Motivation, aims and contribution
Contribution
Is there a statistical link between the performances of universities and the economic growth of the geographical area where they operate?
We provide and answer to this question showing that when evaluating the role of universities in developing human capital not only the performances levels should be considered but also their level of efficiency
Main findings
Main findings
1 University efficiency turns out to be a positive and statistically significant
determinant of local economic development
2 Knowledge spillovers occur between areas through the geographical proximity to
the efficient universities, suggesting that the geography of production is affected
3 Productivity gains are larger in areas in which efficient institutions are located
Background
University and economic growth
Large evidence on the positive effects that HEIs have on their regional
environment, but the discussion is still open on the transfer mechanism (Drucker and Goldstein, 2007; Benneworth, 2009; Uyarra, 2010)
Background
University and economic development: Teaching mission
Local human capital levels might increase through the production of highly skilled graduates and consequently of highly educated workforce (Florida, 1999) Highly skilled and well educated individuals are considered as the ultimate drive of economic development (Florida et al. 2008)
The most competitive regions are typically those with high levels of human capital and universities play a key role in bringing the human capital into regions (Haapanen and Tervo, 2012)
More skilled and educated workers have a higher chance of being involved in the implementation of new technologies (Bartle and Lichtenberg, 1987; Woznaik, 1987; Goldstein et al. 1995)
The provision of graduates is the main contribution of the universities to innovation (Etzkowitz and Leydesdorff, 2000)
Background
University and economic development: Codified knowledge
The proximity to high output universities may be important for accessing social science research (Audretsch et al. 2016)
The higher is the quality of academic research, the large is the contribution to industrial innovations (Mansfield, 2005)
University research has a positive impact on the regional distribution of innovation (Del Barrio-Castro and Garcia-Quevedo, 2005)
Research activities contribute to the creation of knowledge spillovers within the regional environment leading to an improvement of local economies (Goldstein and Renault, 2004)
HEIs should focus more on research activities in order to respond to regional needs (Chatterton and Goddard, 200)
Background
University and economic development: Third mission
Knowledge transfers from academia has been investigated through licensing (Shane, 2002), academia spin-off activities (Shane, 2002) and citation to academic patents (Henderson et al. 1998)
The establishment of new companies based on technologies derived from university research is a well recognized driver of regional economic development (Haytet et al. 2016)
The university industry relationship has become more important due to the role played by technological progress in the development of that area (Algieri et al. 2013; Muscio, 2010)
Incubators are effective in supporting new entrepreneurial initiative (Auricchio et al. 2014)
Background
Why universities’ efficiency is important for economic
development
1 Different universities, which produce the same amount of output, can have a
heterogenous impact on local economic development depending on the intensity of their inputs’ usage
2 If institutions are well respected by the society due to their reputation of being
efficient organizations, they can induce positive mechanism of relationship with important stakeholders’ activity in the territory
3 If the university is productive at its maximum level (i.e. is efficient in the
technical sense), it would encourage efficiency in the other institutions with which it interact
4 To the extent that the outputs of universities positively affect local economic
development, the higher-than-proportional production from efficient universities can have positive effect local economic development
Data and empirical approach
Data
The dataset refers to 53 public universities for the period 2006-2012
Being representative of the higher education system in Italy, corresponding to almost 90% of total number of public universities in the country
The analysis is made at labour market areas (similar to UK’s Travel-to-Work-Areas) made by ISTAT (Italian Statistical Office)
Labour market area is the place where individuals live and work and where their economic and social relationship take place
Data and empirical approach
Map of Italy with the localization of universities
A – Whole Italy B – Our dataset
Data and empirical approach
Map of Italy with the localization of regions and provinces
A – Whole Italy B – Provinces where at least a university included in our dataset is located
Data and empirical approach
Map of Italy with the localization of regions and labour
market areas
A – Whole Italy B – Labour market areas where at least a university included in our dataset is located
Empirical framework
Empirical framework
The analysis is performed in two stages:
Firstly, we use a Stochastic Frontier Analysis (Kumbhakar et al. 2014)
To calculate an index of efficiency for each university
Secondly, a growth model is tested, through a system generalized method moment (sys-GMM) estimator
Empirical framework
First step: Efficiency of the universities (SFA) and BoD
model
We analyze the efficiency of universities considering all of their activities: teaching (graduates), research (publications) and third mission (spin-offs) We construct a composite indicator of the three output variables to avoid the common subjectivity of weighting selection - e.g. use and endogenous weighting system (Benefit Of Doubt model - Melyn and Moesel, 1991)
Using Bod, each university gain its own weights that maximize (or minimize) the impact of the criteria where the university performs relative good (or poor) compare to the others (De Witte and Rogge, 2001; De Witte and Hudrikova, 2013)
Empirical framework
First step: Efficiency of the universities - Production set
INPUTS:
Number of academic (professors, associate professors, assistant professors and lectures) and non-academic staff (Johnes, 2014; Agasisti and Dal Bianco, 2009) Number of total students weighted by the percentage of enrollments with a score higher than 9/10 in secondary school (Agasisti et al. 2016)
OUTPUT:
Number of graduates weighted by their degree classification (Agasisti and Perez-Esparrells, 2010; Thanassoulis et al. 2011)
Number of spin-offs (Berbegal-Mirabent et al. 2013; Calcagnini et al. 2016) Number of publications (Wolszczak-Derlacz and Parteka, 2011; Duh et al. 2014)
CONTROLS:
Year of existence of university (Wolszczak-Derlacz and Parteka, 2011) Presence of Medical School (Kempkes and Pohl, 2010)
Number of inactive students (Barra and Zotti, 2016) University ordinary financing funding
Empirical framework
Second step: the effect of EFF on local growth - GMM
We specify the following two-step system GMM estimator with Windmeijer (2005) corrected standard error dynamic panel model:
lnGDPCijt= αlnGDPCij,t−1+ β1lnEFFijt+ β2lnEFFijt· W+
+β2lnGDPCijt· W + β3lnMKijt+ β4lnLGijt+ it
(1)
GDPCij,tis the worker productivity (output per worker) - LMA level
EFFij,tis the efficiency of the university - University level
MKij,tis the market share - University level
LGijtis the number of employed individual at time t minus the number of
employed individual at time t-1 - LMA level
EFF· W is the weighted average of human capital development proxies across Jj
areas neighboring area j
We use an inverse distance weighed matrix to weight EFF and GDPC of all neighboring areas
Empirical framework
Second step: the effect of EFF on local growth - GMM
(cont’d)
In order to deal with suspected endogeneity problem between EFF and local economic development we include lagged levels and differences as instruments of EFF
We check the correctness of the model through the Hansen test of over-identifying restrictions for validity of the instruments, while the
Arellano-Bond test is, instead, used for testing the autocorrelation between the errors terms over-time
Results
Empirical evidence: Direct effect of universities’ efficiency
on local economic development
The lagged value of GDP per capita (GDPC) has a significant coefficient with positive sign in all models
Supporting the idea that the presence of more efficient universities fosters local economic growth
Results
Empirical evidence: Indirect effect of universities’
efficiency on local economic development
We find evidence that the average productivity of labour is higher in areas that are supported by the presence of universities which well-contribute to the supply of high level human capital
Suggesting the presence of knowledge spillovers within areas having virtuous institutions
Data and empirical approach
The effect of universities’ efficiency on local economic development – Without and with spatial spillovers
Mod_1 Mod_2 Mod_3 Mod_4 Mod_5 Mod_6 Mod_7 Mod_8
GDPC(t-1) 0.707*** 0.718*** 0.709*** 0.714*** 0.704*** 0.716*** 0.698*** 0.705*** (0.014) (0.016) (0.015) (0.001) (0.011) (0.001) (0.015) (0.015) EFF (t-1) 0.051*** 0.047*** 0.0555*** 0.049*** (0.005) (0.005) (0.005) (0.005) EFF (t) 0.013*** 0.010** 0.016*** 0.011*** (0.003) (0.004) (0.004) (0.003) EFF*W (t-1) 0.258*** 0.387*** (0.091) (0.123) EFF*W (t) 0.257*** 0.379*** (0.087) (0.112) GDPC*W(t) 0.317 0.195 0.636** 0.503* (0.231) (0.232) (0.268) (0.268) MK (t) -0.007*** -0.008*** -0.008** -0.009*** -0.007*** -0.008*** -0.008*** -0.010*** (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.002) LG (t) -0.054** -0.047* -0.049* -0.050* -0.061** -0.053** -0.056* -0.057** (0.026) (0.025) (0.027) (0.026) (0.027) (0.024) (0.028) (0.0269) AB(2) 0.356 0.402 0.361 0.409 0.357 0.408 0.367 0.417 Hansen 0.317 0.334 0.274 0.334 0.270 0.299 0.218 0.264 N 318 318 318 318 318 318 318 318
Notes: Standard errors in brackets; * p < 0.1, ** p < 0.05, *** p < 0.01
Results
Empirical evidence: Distribution of universities’ efficiency
scores
We divide the efficiency scores in quartile in order to explore whether the main results are driven by the university with the lowest or highest efficiency level, i.e. at the tails of the efficiency distribution
We repeat the analysis first removing from the sample those universities with an efficiency score in the first quartile, that is, taking out the less efficient
universities and then we remove those universities with an efficiency score in the fourth quartile, that is, taking out the most efficient universities
Data and empirical approach
The effect of universities’ efficiency on local economic development– Without and with spatial spillovers using quartile university efficiency scores – Excluding the lowest efficiency universities
Mod_1 Mod_2 Mod_3 Mod_4 Mod_5 Mod_6 Mod_7 Mod_8
GDPC (t-1) 0.871*** 0.799*** 0.874*** 0.771*** 0.870*** 0.794*** 0.860*** 0.783*** (0.020) (0.026) (0.020) (0.025) (0.022) (0.026) (0.023) (0.034) EFF (t-1) 0.043*** 0.056** 0.066*** 0.074*** (0.014) (0.021) (0.020) (0.023) EFF (t) 0.018 0.002 0.014 0.031 (0.022) (0.022) (0.025) (0.027) EFF*W (t-1) 0.058 0.170** (0.072) (0.081) EFF*W (t) 0.146 0.264*** (0.119) (0.096) GDPC*W(t) 0.330 0.199 0.527** 0.518** (0.206) (0.148) (0.225) (0.247) MK (t) -0.005*** -0.008*** -0.005*** -0.007*** -0.004*** -0.008*** -0.005*** -0.009*** (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) LG (t) -0.040 -0.041 -0.039 -0.010 -0.037 -0.045 -0.034 -0.062* (0.028) (0.031) (0.028) (0.035) (0.029) (0.032) (0.029) (0.033) AB(2) 0.680 0.838 0.664 0.750 0.649 0.842 0.645 0.835 Hansen 0.723 0.660 0.756 0.894 0.729 0.705 0.727 0.796 N 232 232 232 232 232 232 232 232
Notes: Standard errors in brackets; * p < 0.1, ** p < 0.05, *** p < 0.01
Data and empirical approach
The effect of universities’ efficiency on local economic development – Without and with spatial spillovers using quartile university efficiency scores – Excluding the highest efficiency universities
Mod_1 Mod_2 Mod_3 Mod_4 Mod_5 Mod_6 Mod_7 Mod_8
GDPC (t-1) 0.667*** 0.717*** 0.695*** 0.688*** 0.628*** 0.634*** 0.636*** 0.672*** (0.017) (0.010) (0.0173) (0.020) (0.018) (0.021) (0.022) (0.021) EFF (t-1) 0.048*** 0.043*** 0.058*** 0.0509*** (0.004) (0.003) (0.006) (0.005) EFF (t) -0.003 0.002 0.015** 0.004 (0.003) (0.005) (0.006) (0.003) EFF*W (t-1) 0.139* 0.534*** (0.082) (0.157) EFF*W (t) 0.164** 0.449*** (0.078) (0.108) GDPC*W(t) 1.126*** 1.091*** 1.680*** 1.275*** (0.186) (0.186) (0.335) (0.320) MK (t) -0.008*** -0.011*** -0.007*** -0.009*** -0.008*** -0.011*** -0.008*** -0.0103*** (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) LG (t) -0.023 -0.008 -0.020 -0.014 -0.0206 0.001 -0.027 -0.016 (0.027) (0.014) (0.025) (0.026) (0.032) (0.030) (0.029) (0.023) AB(2) 0.337 0.395 0.336 0.398 0.323 0.389 0.342 0.401 Hansen 0.259 0.604 0.258 0.289 0.316 0.334 0.358 0.558 N 244 244 244 244 244 244 244 244
Notes: Standard errors in brackets; * p < 0.1, ** p < 0.05, *** p < 0.01
Results
Empirical evidence: Distribution of economic development
measure (GDP)
We divide the measure of economic development in quartile in order to explore whether the main results are driven by the university located in areas
characterized by a high level of economic development
We repeat the analysis first removing from the sample those universities with a GDP in the first quartile, that is, taking out areas with low development levels and then we remove those universities with a GDP per capita in the fourth quartile, that is, taking out those areas which grow the most
Data and empirical approach
The effect of universities’ efficiency on local economic development – Without and with spatial spillovers using quartile of GDPC – Excluding the areas with the lowest GDPC values
Mod_1 Mod_2 Mod_3 Mod_4 Mod_5 Mod_6 Mod_7 Mod_8
GDPC (t-1) 0.549*** 0.594*** 0.566*** 0.639*** 0.559*** 0.592*** 0.567*** 0.647*** (0.026) (0.032) (0.024) (0.026) (0.027) (0.036) (0.023) (0.030) EFF (t-1) 0.029*** 0.020*** 0.025*** 0.019*** (0.004) (0.004) (0.005) (0.005) EFF (t) 0.0003 -0.003 0.007** 0.004 (0.002) (0.002) (0.003) (0.002) EFF*W (t-1) 0.343*** 0.459*** (0.073) (0.135) EFF*W (t) 0.335*** 0.351*** (0.055) (0.123) GDPC*W(t) -0.288 0.187 0.247 0.212 (0.194) (0.293) (0.319) (0.397) MK (t) -0.014*** -0.013*** -0.013*** -0.012*** -0.012*** -0.0138*** -0.012*** -0.010*** (0.002) (0.002) (0.002) (0.001) (0.002) (0.002) (0.002) (0.002) LG (t) -0.006 -0.001 -0.017 -0.002 0.010 -0.011 -0.0005 -0.024 (0.025) (0.021) (0.022) (0.023) (0.024) (0.023) (0.021) (0.021) AB(2) 0.325 0.281 0.316 0.327 0.331 0.290 0.336 0.329 Hansen 0.421 0.555 0.390 0.685 0.315 0.592 0.252 0.544 N 241 241 241 241 241 241 241 241
Notes: Standard errors in brackets; * p < 0.1, ** p < 0.05, *** p < 0.01
Data and empirical approach
The effect of universities’ efficiency on local economic development – Without and with spatial spillovers using quartile of GDPC – Excluding the areas with the highest GDPC values
Mod_1 Mod_2 Mod_3 Mod_4 Mod_5 Mod_6 Mod_7 Mod_8
GDPC (t-1) 0.530*** 0.518*** 0.522*** 0.518*** 0.544*** 0.547*** 0.549*** 0.553*** (0.014) (0.018) (0.019) (0.020) (0.017) (0.022) (0.024) (0.025) EFF (t-1) 0.023*** 0.020*** 0.024*** 0.024*** (0.005) (0.006) (0.005) (0.006) EFF (t) -0.006 -0.008 -0.006 -0.008 (0.004) (0.005) (0.004) (0.005) EFF*W (t-1) 0.114 0.189 (0.115) (0.177) EFF*W (t) 0.092 0.263* (0.117) (0.155) GDPC*W(t) 0.682*** 0.736*** 0.397 0.799*** (0.179) (0.175) (0.455) (0.239) MK (t) -0.001 -0.003 -0.001 -0.002 -0.002 -0.003 -0.001 -0.003 (0.002) (0.002) (0.003) (0.003) (0.002) (0.002) (0.002) (0.002) LG (t) -0.064*** -0.043* -0.069*** -0.052** -0.072*** -0.052** -0.063*** -0.046* (0.021) (0.021) (0.021) (0.022) (0.024) (0.023) (0.023) (0.023) AB(2) 0.410 0.421 0.422 0.433 0.411 0.413 0.398 0.406 Hansen 0.440 0.425 0.454 0.427 0.369 0.341 0.535 0.483 N 238 238 238 238 238 238 238 238
Notes: Standard errors in brackets; * p < 0.1, ** p < 0.05, *** p < 0.01
Results
Robustness checks
We use the same composite index consisting of three output variables as measure of the university performances without calculating the ratio at which they are able to convert inputs into outputs
In other words, we do not calculate the level of efficiency at which they operate, but only the association between universities’ output level and local economic performance
Data and empirical approach
The effect of universities’ performances on local economic development – Using the composite index consisting of the teaching, research and knowledge transfer missions
Mod_1 Mod_2 Mod_3 Mod_4 Mod_5 Mod_6 Mod_7 Mod_8
GDPC (t-1) 0.757*** 0.768*** 0.799*** 0.797*** 0.739*** 0.753*** 0.788*** 0.784*** (0.018) (0.008) (0.015) (0.008) (0.021) (0.010) (0.017) (0.010) UNI_PERF (t-1) 0.010*** 0.006** 0.013*** 0.008*** (0.003) (0.002) (0.003) (0.002) UNI_PERF (t) 0.011*** 0.009*** 0.014*** 0.012*** (0.002) (0.002) (0.0024) (0.002) UNI_PERF *W (t-1) 0.018 0.040 (0.030) (0.034) UNI_PERF *W (t) 0.0435 0.075** (0.025) (0.030) GDPC*W(t) 0.251 0.295 0.253 0.400* (0.261) (0.206) (0.233) (0.201) MK (t) -0.009*** -0.010*** -0.008*** -0.009*** -0.010*** -0.010*** -0.008*** -0.008*** (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) LG (t) -0.063*** -0.045** -0.068*** -0.052** -0.066*** -0.050** -0.069*** -0.052** (0.021) (0.020) (0.020) (0.020) (0.022) (0.020) (0.020) (0.019) AB(2) 0.462 0.386 0.435 0.389 0.484 0.389 0.449 0.386 Hansen 0.228 0.336 0.221 0.351 0.221 0.336 0.210 0.349 N 318 318 318 318 318 318 318 318
Notes: Standard errors in brackets; * p < 0.1, ** p < 0.05, *** p < 0.01
Summary and Conclusion
Summary and Conclusion
Examine the relationship between university level of efficiency and economic growth
We use SFA to calculate an index of efficiency for each university and then, a growth model is tested, through a sys-GMM estimator, to evaluate the relationship between human capital development and local economic growth. The nature of spatial spillover has been also into account
Summary and Conclusion
Summary and Conclusion (cont’d)
Our findings demonstrates that it is not only university performances that matters, but especially their efficiency
Their ability of making the most with available resources
The presence of spillovers suggests that the geography of production is affected Productivity gains are larger in areas in which efficient institutions are located when all the three missions of universities are taken into account
Implications
Implications
1 The empirical evidence validates the use of university efficiency as an instrument
able to capture the impact on the community of the ability of universities of making the most with the available resources
2 Results confirm the importance of measuring the development of human capital
and skills, the technology transfer activities, new product development and research activities to better understand the mechanisms behind the local economic development
3 Results support the necessity to take into account the presence of spatial
dependence, when explaining regional income per capita and development differences
4 The findings reveal that presence of virtuous circles characterized by highly
efficient universities, located in the territories which grow more, which in turn stimulate to reach higher levels of operations’ efficiency
5 Economic development at local level benefits from an environment where
Final comments
A final comment
Our findings point at the use of the educational instrument for fostering economic development
The importance of the spatial effects leads to a call for more investments in the tertiary education system given that they would affect not only the performances of the universities but also the economic conditions of the areas where the institutions are located
This is not a secondary issue considering the substantial reforms that have been taken place in the last years in Italy and that the basis for allocating core funding to HEIs has become more output-oriented.