Econ_letters.dvi

The private contractual funding of academic Rachid Boumahdiγ, Nicolas Carayolφ∗ and Patrick Llerenaφ γ GREMAQ and LIRHE, Université des Sciences Sociales de Toulouse 61 avenue de la Forêt Noire, F-67085 Strasbourg Abstract : The paper presents new evidence explaining contractual pri- vate funding of academic laboratories. We find that public funding crowds out private one. While private funding increases with publications it decreases with publications corrected for impact.
Keywords : Contractual Funding ; Academic Laboratories ; Panel Data ; JEL classification : C33 ; L30 ; H4 ; O31 ∗ Corresponding author. Tel. : +33(0)390242104 ; fax : +33(0)390242071.
E-mail address : [email protected].
The last two decades have been a very disruptive context for academic research funding : Shortening of public funding, evolving rationales and chan- ging regulatory environments. One of the crucial issues relates to the rising share of public research being funded through contractual relations with pri- vate firms. Scholars mainly concentrate their attention on the consequences of private contractual funding on research production : Expected benefits are an increase of academic research due to extra funding and a shift in its econo- mic relevance. Risks reside in a slowing down in the dissemination of findings, a shift toward applied research and a decrease of research productivity1.
The present paper focuses on the reverse causality : What stimulates the attractivity of academic research for private funding ? What is meant here by contractual funding is all the funds received on a contractual basis, including grants. Typically, all funds received from private sources are included while recurrent public funding is excluded.
The main originality of our study is that it is micro-based, i.e. conduced at the academic research laboratory level while the literature on academic funding focuses on the university level of investigation. The laboratory is emphasized as the relevant level of analysis of scientific activity (Crow and Bozeman, 1987) especially in the Continental European style of academic research organization and as long as funding issues are concerned (Stephan, Our data allows us to examine three series of influences.
i) Publication signaling : Funds managers are subject to high uncertainty and asymmetries of information on academics. In such a situation (adverse selection), they may use publications to better evaluate laboratories compe- tencies in specific domains, the quality of their scientific production or even their ability to collaborate with industrial partners.
ii) Crowding : An important literature in public economics aims to evi- 1 See Dasgupta and David (1994), Cohen et al. (1998), Blumenthal et al. (1996).
dence the effects of public funding on other sources for non profit institutions.
While theoretical models differ in their predictions, Connolly (1997) shows that external funding and internal support of US universities research crowd in mutually and Payne (2001) finds that federal research funding of US uni- versities crowds in simultaneous private donations.
iii) Matching : The characteristics of laboratories also may have a great influence on private funds managers and on laboratories’ own willingness to provide the specific efforts associated with getting such funds.
The next section presents the data and the methodology. Section 3 pre- sents the results while the last section concludes.
Our data cover the period 1993-2000 and concern the research activity of 76 laboratories which belong to one single university, namely Louis Pasteur University (ULP) of Strasbourg. This university is quite large, diversified, has an old tradition of fundamental research and a long standing of scientific excellence2. The Third European Report on Science and Technology Indica- tors (2003) ranks it first among French universities in terms of impact and For each laboratory, we have eight annual observations on the follo- wing time-variant variables : Private contractual funding (PRIVF (it)), Pu-blic contractual funding (PUBF (it)), Publication performance (PPER(it)),Publication performance corrected for impact (PIMP(it))3 and two dummyvariables indicating whether at least one paper is coauthored with a re- searcher employed by a private firm (INDUS (it)) or by a foreign institution 2 ULP researchers have received numerous scientific prizes (six Nobel prices and one Field Medal). Active researchers count one Nobel laureate, eleven members of the Institut Universitaire de France and eleven members of the French National Academy of Science.
3 We collected more than 26,000 occurrences of published articles of all permanent researchers (using SCI and SSCI databases of ISI). PUBF (it) sums and corrects for coau-thorship. PPIMP (it) in addition weights each item by the impact factor (given in ISI-JCI).
(INTER(it)). Time-invariant variables relate to the personnel employed in thelaboratories4 : Number of full professors or researchers (SENIOR(i)), assistantprofessors or researchers (JUNIOR(i)), PhD students (PHD(i)), non resear-chers (administrative personnel, technicians and engineers) (NONRES (i)),number of national (NPOST (i)) and foreign post-docs (FPOST (i)).
where i = 1, 2, . . . , 76; and t = 1, 2, . . . 8. Combining all 608 observations, we where X = (X1 : X2) and Z = (Z1 : Z2). We assume that εit are iid N(0, σ2) uncorrelated with both the explanatory variables Xit and Zi. Following Haus-man and Taylor (1981) - hereafter HT -, Amemiya and MacCurdy (1986) - hereafter AM - we assume that the individual effects αi are iid N(0, σ2 ) cor- related with X2 and Z2, uncorrelated with X1 and Z1. For estimating theequation (2), we use the Instrumental Variable (IV) method described in HT In Table 1 (columns 1 and 2) we consider, first, the conventional proce- dures Within and GLS estimates. The Within estimates are unbiased whether or not the effects αi are correlated with the explanatory variables5. Howeverthe GLS estimates are biased if the individual effects are correlated with some explanatory variables. The assumption that αi are uncorrelated with (X, Z) 4 These variables come from standardized administrative reports completed by all labo- ratories in 1996 which are both a précis of the past four years and a project for the next 5 All time-invariant variables are eliminated by the data transformation.
is rejected in a Hausman test of the difference between the GLS and Wi- thin estimates. The Hausman test is χ2 = 21.55 which is significant at 5 per cent level. The HT and AM estimates are presented in columns 3 and 4. In this regression we let X1 =(PUBF, INDUS, INTER), X2 =(PPER, PIMP),Z1 =(PhD, JUNIOR, SENIOR, NONRES ) and Z2 =(NPOST, FPOST ).
The set of instruments proposed by HT is legitimate and supported by a Hausman test of the difference between the Within estimator and the HT estimator. This test is χ2 = 0, 32 and is insignificant at five per cent le- vel. Similarly, the additional exogeneity restrictions allowed by AM are not rejected6. The Hausman test which compares AM and HT is χ2 = 2, 22.
Our first result shows that present public contractual funding coefficient is negative and significant. This supports the crowding out hypothesis. Nei- ther the dummy for international collaborations, nor the one for industrial collaborations are significant. It may indicate that such signals are either not used by private funds managers or ambiguous. The level of publication performance has a positive and significant effect (as in Payne, 2003). The la- boratories that publish more are more attractive in the eyes of private funds The coefficient of publication performance corrected for impact is nega- tive. This result may seem counter-intuitive at first glance since the impact factor of scientific journals may signal the quality of research. Nevertheless, other phenomena seem to be predominant here. Let us suggest two of them.
First research appearing in the most prominent journals is likely to be more fundamental while private contractual funding may seek research closer to ap- plications. Secondly contracts with private partners often generate dedicated and specific efforts from faculty members (ex ante for attracting funds and ex post for meeting the requirements). Laboratory managers may consider the opportunity costs of contracts which are not independent of laboratory characteristics. The alternate use of their time and efforts in the academic 6 As the set of instruments proposed by AM is legitimate, we concentrate our attention sphere is better valued by the ones who are publishing in the best ranked journals (Carayol, 2003 finds similar results).
We now turn toward the effects of labor force composition. Post-docs ef- fects are positive and significant (the coefficient for the domestic ones is three times higher). This indicates that they strongly contribute to the research efforts implied by contracts signed with private partners. The coefficient for PhD students is also positive but significance is much lower : They are less intensively involved in contractual research being also strongly driven by doc- torate accomplishments. The effect of non researchers is also positive : Suffi- ciently large administrative and engineering staff are important for attracting interest and/or for meeting requirements (organization, delays, instrumenta- tion, etc.). Associate professors and researchers are negatively correlated with private funding : Not yet promoted researchers tend to concentrate on pure academic work because of career concerns.
The paper proposes an analysis of the yearly arrival of private contrac- tual funding of academic laboratories. We find that contractual public fun- ding crowds out simultaneous private funding. Private funds are attracted by the most active laboratories within the academic sphere while best ran- ked publishing laboratories attract less. Post-docs increase private support substantially because of “who is doing the job” issues.
This work is part of a larger project on knowledge production at ULP.
We are grateful to all members of the team. Acknowledgements extend to the administrative departments and the Technology Transfer Office at ULP, and to the CNRS Industrial Liaison Office.
Amemiya, T., Macurdy, T.E., 1986. Instrumental—Variable Estimation of an Error—Component Model. Econometrica 54, 869-880.
Arora, A., David, P.A., Gambardella, A., 1998. Reputation and compe- tence in publicly funded science : Estimating the effects on research group productivity. Annales d’Economie et de Statistique 49/50, 163-198.
Baltagi, B.H. and Khanti-Akom, 1990. On efficient estimation with panel data : An empirical comparison of instrumental variables estimators. Journal Blumenthal, M.D., Causino, N., Campbell, E.G., Louis, K.S., 1996. Par- ticipation of life-science faculty in research relationships with industry. The New England Journal of Medicine 335, 1734-1739.
Carayol, N., 2003. Objectives, agreements and matching in science indus- try collaborations : Reassembling the pieces of the puzzle. Research Policy Cohen, W.M., Florida, R., Randazzese, L., Walsh, J., 1998. Industry and the academy : Uneasy partners in the cause of technological advance. In : R. Noll, ed., Challenge to the University, (Brookings Institution Press, Wa- Connolly, L.S., 1997. Does external funding of academic research crowd out institutional support. Journal of public Economics 64, 389-406.
Cornwell, C., Rupert, P., 1988. Efficient estimation with panel data : An empirical comparison of instrumental variables estimators. Journal of Applied Crow, M., Bozeman, B., 1987. R&D laboratory classification and public policy : The effects of environmental context on laboratory behavior. Re- Dasgupta, P., David, P.A., 1994. Toward a new economics of science.
Hausman, J.A., 1978. Specification Tests in Econometrics. Econometrica Hausman, J.A., Taylor, W.E., 1981. Panel Data and Unobservable Indi- vidual Effects. Econometrica 49, 1377-1398.
Payne, A., 2001. Measuring the effect of federal research funding on pri- vate donations at research universities : Is federal research funding more than a substitute for private donations ?. International Tax and Public Finance 8, Stephan, P.E., 1996. The economics of science. Journal of Economic Li- Table 1. The dependent variable is PRIVF (it) (Private contractual funding) Standard errors are in parentheses. The ∗ and ∗∗ indicate that coefficients are statistically significant at the 0.05 and 0.10 levels respectively.

Source: http://ncarayol.u-bordeaux4.fr/priv_ac_funding.pdf