Read without ads and support Scribd by becoming a Scribd Premium Reader.
 
Innovation and the determinants of companysurvival
By Hielke Buddelmeyer, Paul H. Jensen, and Elizabeth Webster 
Appendix 1 – Survey population and sampleTable A1
The number of companies by industry, ABS compared with ASICmatched to business directoriesMajor IndustryABS1999-2000ASIC_Business Directories
%%
Agriculture, forestry and fishing
3.22.7
Mining
0.41.0
Manufacturing
10.239.5
Electricity, gas and water 
0.00.1
Construction
16.916.3
Retailing and wholesaling
32.415.7
Accommodation, cafes and restaurants
6.82.0
Transport and storage
6.35.9
Communication services
0.90.5
Private community services
12.09.0
Cultural and recreational services; personal andother services
11.07.3
TOTAL
100.0100.0
Appendix 2 – Robustness checks
To check the robustness of our estimates, we experimented with other modelspecifications. Specifically, we estimated a standard Cox model, which does not parameterise or estimate a baseline hazard. The piecewise-constant baseline specificationand the Cox specification produced almost identical results, which is a manifestation of the flexibility of the piecewise-constant form. We also experimented with other popular  parametric specifications – e.g. Weibull, Gompertz and exponential – for the baselinehazard. These results were qualitatively in line with the results in Table 5.We also experimented with up to three lagged periods for the IP variables. Allestimations produced very similar results. While the three-year lagged variables had the
 
largest effects, we only present one- and two-year lags to clearly distinguish the measuresfor current innovative activities versus past innovative activity.In order to test for the robustness of our patent application and stock coefficientsacross industries, we estimated the model for each single-digit industry. Theseestimations showed that the patent coefficients were only statistically significant inindustries which both comprised a large number of companies or had high patentingrates. These industries – Mining; Manufacturing; and Property and business services – account for 74.1% of all patent applications and 57.9% of all observations. Results fromthese regressions, which are presented in Table A2, show that except for IP stocks inmining which was insignificant, we observe the same pattern of coefficients foinnovation investment and capital as for the whole sample.As a final robustness check we tested for the effects of unobserved heterogeneity. Weexperimented extensively with the model specification but found no evidence of unobserved heterogeneity, a result similar to other recent attempts (Hoetker 
et al.
2007;Ortega-Argiles and Moreno 2005; Audretsch and Lehmann 2005; Nkurunziza 2005;Esteve-Perez and Manez-Castillejo 2008; Fontana and Nesta 2007; and Bayus andAgarwal 2007). We were surprised by this given compelling evidence from studieshighlighting the importance of founder skills and managerial competence for companysurvival (such as Bruderl
et al.
1992; Klepper and Simons 2000; Klepper 2002;Thompson 2005), but cannot reconcile the two types of results.
1
It is possible thatunobservable heterogeneity is difficult to detect in single failure models.
1
While Jenkins (2005) notes that use of flexible base-line hazard specifications alleviates the need to usefrailty corrections to ensure the robustness of estimates, this does not satisfactorily explain why unobservedheterogeneity is not found when we have good economic and management literature priors for believingthat it exists.
Search History:
Searching...
Result 00 of 00
00 results for result for
  • p.
  • More From This User

    Notes
    Load more