Management Practices, Competition, Innovation and the Performance of Enterprises in Canada

The views and opinions expressed in the research paper are those of the authors alone and do not represent, in any way, the views or opinions of the Department of Innovation, Science and Economic Development or of the Government of Canada.


Innovation, Science and Economic Development CanadaFootnote *


University of Toronto

Abstract

Using data from the Survey of Innovation and Business Strategy 2009, a management practices index ( MPindex) à la Bloom and Van Reenen (2007) is built to analyze the relationships between management practices, competition, innovation and firm performance in Canada. The results show that the distribution of MPindex varies among industries and that large firms tend to have more structured management practices. A positive correlation between ( MPindex) and the intensity of sales and profits is found for manufacturing firms. There is also a positive correlation between ( MPindex) and business innovation, a result that holds for enterprises in all sectors. Finally, the importance of competition depends on how it is measured: the number of competitors in the enterprise's main market is associated with a higher intensity of sales and profits, while it is the entry of new competitors in the main market that matters for innovation.

Final version:

Table of Contents


1. Introduction

A number of recent reports by industry observers highlight that Canada's lagging performance in terms of productivity growth is associated with the low levels of business innovation (Competition Policy Review Panel, 2008; Expert Panel on Business Innovation, 2009; Drummond and Bentley, 2010; Expert Review Panel on Research and Development, 2012). Several possible causes have been put forward such as industry structure, subpar investments in tangible capital, small market size, "business complacency" and the low importance attributed to education by Canadian managers. The last two factors are of particular interest as they are intangible in nature. Also falling in this intangible assets category, are management practices (MP).

The role of MP in the success of enterprises has long been recognized in economics and other social sciences (see for example Huselid, 1995). And put by Alchian and Demsetz (1972), "efficient production… is a result not of having better resources but in knowing more accurately the relative productive performances of those resources."

The recent empirical work by Bloom and Van Reenen (2007) defines a tractable framework for measuring MP and assessing their relationship with firm performance. MP, and more generally intangible assets, may be an important factor explaining the persistent productivity differentials observed across countries, but also within a country in narrowly defined industries (Bartelsman and Doms, 2000; Syverson, 2004a,b, 2011). If firms in a same industry are observed to use the same inputs and technologies and face the same level of competition, then there must be something else affecting their productivity. Supporting the view that MP matter for productivity, Bloom et al. (2012b) estimated that better MP of US firms accounts on average for 30 percent of the total factor productivity differential between the US and other countries such as France, the UK, Sweden and Germany.

Manufacturing enterprises in Canada fare well compared to their counterparts located in other countries. Only US enterprises have better MP compared to those in Canada (Institute for Competitiveness and Prosperity, 2009). Bloom and Van Reenen (2010) showed that the US has the lowest proportion of badly managed enterprises compared to any other countries, including Canada. Moreover, there is no statistical difference between average MP in Canada, Germany, Sweden and Japan, while Canada is outperforming the UK and France, among others. From a broad perspective, Bloom (2010) mentioned that the marketplace framework in Canada is favourable to the implementation of good MP due to, for example, the high levels of competition and low market regulation. Firms in Canada, however, should make better use of skilled workers.

The objective of this paper is to determine the relationships between MP, firm performance and innovation using a framework similar to the one developed by Bloom and Van Reenen (2007). There are three main research questions: i) what are the firms' characteristics associated with better structured MP; ii) what is the relationship between MP and firm performance; and iii) what is the relationship between MP and innovation. Answering the last question is one of the main contributions of the paper as very little evidence exists on the link between MP and innovation. In addition to MP, the role of competition in firm performance and innovation is also considered. This analysis is based on the data from the Survey of Innovation and Business Strategy 2009 (SIBS) and other administrative sources (Statistics Canada).

The main results can be summarized as follows: i) the distribution of the management practices index ( M P i n d e x ) differs by sector; ii) the variables positively correlated with M P i n d e x include firm size, percentage of workers with a university degree, the presence of a multinational in the enterprise's main market and whether the head office is located in the United States; iii) M P i n d e x is positively correlated with sale and profit intensity for manufacturing enterprises only; iv) M P i n d e x is positively correlated with innovation—no matter how innovation is defined—for enterprises in all industries; and v) the links between competition, performance and innovation depend on the nature of competition. More precisely, the number of competitors in the enterprise main market is positively correlated with higher performance but it is the entry of new enterprises that matter for innovation.

The rest of the paper is organized as follows. The analytical framework is presented in the second section while the data discussion is in the third one. Section 4 shows the results and Section 5 concludes.


2. Analytical Framework

2.1 Construction of the Management Practices Index

The M P i n d e x used in this paper is based on the framework introduced by Bloom and Van Reenen (2006, 2007) (BVR). SIBS provided an opportunity to collect additional MP data for Canada and linked them to economic indicators such as sales, profits and innovation. Most SIBS questions on MP were derived from the BVR survey on MP. This allowed to build a similar but not identical M P i n d e x .

The main contribution of BVR was to develop a comprehensive and tractable framework to analyze MP. The BVR index is based on four types of indicators referring to a specific aspect of MP: operational practices, monitoring of performance, target setting and incentives. See the Appendix in Bloom and Van Reenen (2007) for more details.

The BVR work can be linked to several strains of the economic literature: management style (Bertrand and Schoar, 2003); innovative work practices (Ichniowski et al., 1997; Macduffie, 1995; Osterman, 1994); workers empowerment (Cappelli and Neumark, 1999); incentives pay (Lazear, 2000; Bandiera et al., 2007); hiring practices (Oyer and Schaefer, 2011; Autor and Scarborough, 2008); total quality management (Powell, 1995); and information and communication technologies use (Black and Lynch, 2001; Bresnahan, 1999; Bresnahan et al., 2002; Bloom et al., 2012a).

The SIBS MP index ( M P i n d e x ) is built for 2009 using 19 indicators. Each indicator has been normalized between zero and one, the latter denoting the best practices and zero the worst. A simple average is taken, as shown by Equation (1), so M P i n d e x also ranges from zero to one. M P i n d e x has at least one element in each type of indicator of the BVR index, but as shown in Section A.1 of Appendix A, the coverage is unequal.

All SIBS MP questions were used, except for the share of workers with a university degree (Q63) and the one about who set the pace of work to achieve production performance targets (Q57). Exclusion of these variables are based on the fact that it is unclear which response should be attributed a higher score. Q63 was nevertheless included as a separate variable in the regression analysis.

M P i i n d e x = Σ j = 1 19 I n d i c a t o r s i j 19
(1)

The normalization process is specific to each indicator and is consistent with the spirit of the BVR framework. See Section A.2 for more details on the normalization rules used (Appendix A).

Apart from the contents, the other main difference between the BVR survey and the SIBS is the way data were collected. BVR used face-to-face interviews to collect in-depth information on the firm MP. In contrast, paper/electronic questionnaires were used for the SIBS. Because of their collection method, BVR can argue that their index is effectively a proxy for the quality of a firm's MP, but this is less clear for the SIBS. Therefore, the results are going to be interpreted as "more structured MP" rather than as "better MP." This interpretation is borrowed from Bloom et al. (2013), who also used MP data from a paper/electronic MP survey in the US.

The differences listed in this section raises the question of comparability between the BVR and the SIBS MP index. Comparison of the non-parametric distributions from Figure 2 in Bloom (2010) and Figure 2 shows that, overall, both indices yield similar distributions for manufacturing enterprises. The only noticeable difference is the seemingly fatter tails in Figure 2.


2.2 Determinants of MP

The first part of the regression analysis examines the relationship, at the firm level, between M P i n d e x and a number of characteristics such as firm size and the degree of competition the firm faces. The relationship takes the following linear form (for 2009):

M P i = α 0 X i + α 1 C O M P i + α 2 L E R N E R i + i
(2)

Firms are denoted by i and the usual error term. X is a vector of firm characteristics that includes workers education and binary variables for firm size, firm structure, industry and province. See Appendix B for a detailed description of all variables used in this analysis. The inclusion of most of these variables is justified given results from the BVR literature. Bloom (2010) and Institute for Competitiveness and Prosperity (2009) are especially relevant because of their focus on Canada.

Bloom (2010) found that large firms in Canada have the best MP, a evidence supported by SIBS data as shown in Figure 1. There is a clear pattern suggesting that the larger the enterprise, the more structured their MP. M P i n d e x average is 0.40 for small enterprises and 0.66 for the x-large, while the average for medium and large ones are around 0.55. It is however unclear which factor is the cause. Large enterprises may need to implement good MP to conduct efficiently their operations, but it may be that good MP are required for growth. As for the other variables, the results must be interpreted as correlations, not causal effects. This applies to all relationships estimated in this paper and to other works in the BVR literature apart from Bloom et al. (2011).

Figure 1: Distributions for MPindex by firm size
– Weighted densities, total sample N = 4,227 –

Graph of distributions for MP index by firm size – Weighted densities, total sample N = 4,227 (the long description is located below the image)
Source: Survey of Innovation and Business Strategy, 2009
Note: Firm sizes are based on individual labour units.
Description of Figure 1

This figure describes the distribution of enterprises based on their MP index score across the potential values, which are bounded between 0 – 1 on the X-axis. Note that the Y-axis is suppressed for confidentiality reasons. The figure contains four lines, which represent small size enterprises (20–49 employees), medium size enterprises (50–99 employees), large size enterprises (100–249 employees) and extra-large size enterprises (250+ employees). The small size enterprise distribution line shows that most are concentrated on the low-end of the MP index – the line shows the largest concentrations of small size enterprises have values between 0.2 and 0.4. Moving across the X-axis, the distribution line for small size enterprises tapers after its max point just below 0.4; this steady decline shows that small size enterprises have the lowest concentration of observations with MP index scores above 0.6. The medium size enterprise distribution line shows two large humps; the first is just beyond small size enterprise max point – roughly at an MP index value of 0.4 ­– and the second is centered at a MP index value of approximately 0.7. This suggests that a significant percentage of medium size enterprises are relatively poorly managed with a low MP index value, and a significant percentage are relatively well managed with a high MP index value. The large size enterprise line gradually increases along the X-axis until reaching an MP index value of close to 0.4; from this point the line decreases slightly until reaching 0.6 on the X-axis; then the line follows a slight upward trend until approximately 0.85 on the x-axis; finally, the declines sharply until reaching 1. This suggests that while there are large enterprises with low MP index values, a larger concentration of large size enterprises have high MP index values relative to small or medium size enterprises. Lastly, the distribution line for extra-large size enterprises gradually rises moving left to right along the x-axis until reaching 0.6 on the x-axis; from there, the line dips slightly before rising again and reaching a new high-point at approximately 0.85 on the x-axis, then following a sharp decline until reaching 1 on the x-axis. This line shows that extra-large size enterprises have the smallest concentration of poorly-managed enterprises and the highest concentration of well-managed enterprises relative to the other enterprise size classes.

Structure and ownership of the enterprise also matters for MP. Multinational enterprises (MNE) were found by Bloom and Van Reenen (2010) to have better MP compared to other enterprises. In this paper, a set of binary variables for the location of the headquarters (HQ) outside Canada is used as a proxy for MNE status. A country of control variable was also available, but it was so closely correlated with HQ variables that it could not be included. Note that the drawback of using the HQ variables is their inability to distinguish between a MNE and a non-MNE that both have their HQ in Canada. Another binary variable indicating the multi-establishment status of the enterprise was included as Bloom et al. (2013) reported that multiple establishment enterprises tend to have better MP. Finally, family-owned and managed enterprises are the worst managed compared to dispersed shareholders and private equity ownership (Van Reenen, 2011; Bloom and Sadun, 2009). Unfortunately, ownership information was not available in the SIBS.

Education is summarized by the percentage of workers with a university degree ( % U N I V ) . Unfortunately, the SIBS did not collect data on managers' education. Bloom et al. (2013) showed that the magnitude of the correlation between MP and the workers' education is similar to the one between MP and managers' education. This means that the estimated coefficient of %UNIV may be overestimated.

The original BVR framework focussed on manufacturing enterprises, but a project was developed in collaboration with the Institute for Competitiveness and Prosperity to extend their framework to the retail sector. They found that in Canada, the US and the UK, manufacturing enterprises are better managed compared to retail sector enterprises (Institute for Competitiveness and Prosperity, 2010). SIBS data are again consistent with this evidence as shown in Figure 2. It is shown that the left tail of the distribution for non-manufacturing enterprises is much fatter compared to one for manufacturing enterprises. Given the difference between sectors, all descriptive statistics and regression analyses are going to be presented for manufacturing and non-manufacturing sector separately.

Figure 2: Distributions of MPindex by sector
– Weighted densities, total sample N = 4,227 –

Graph of Distributions of MP index by sector – Weighted densities, total sample N = 4,227 (the long description is located below the image)
Source: Survey of Innovation and Business Strategy, 2009
Description of Figure 2

This figure describes the distribution of enterprises based on their MP index score across the potential values, which are bounded between 0 – 1 on the X-axis. Note that the Y-axis is suppressed for confidentiality reasons. The figure contains 3 lines: one line describing enterprises operating in the manufacturing sector, one line describing enterprises operating in non-manufacturing industries, and one line describing all enterprises. The all enterprises line and the non-manufacturing line are almost identical. Moving from left-to-right along the x-axis, the all enterprises and non-manufacturing lines show a sharp increase until reaching a high-point just below 0.4 on the x-axis. Following this high-point, both lines show a sharp decline losing approximately half their height by near 0.5 on the X-axis. After this initial decline, both lines follow a more gradual decline until the end of the x-axis, which corresponds with an MP index value of 1. The line describing enterprises operating in manufacturing industries shows a less steep incline at the beginning of the X-axis, reaching its high-point at a slightly higher MP index value at around 0.5 on the X-axis. From the high-point, the line follows a steady decline until the end of the X-axis. This figure suggests that manufacturing enterprises have less poorly-managed firms and more well-managed firms than enterprises operating in non-manufacturing industries.

Caution must, however, be exercised with non-manufacturing data because the SIBS coverage for this sector is not as comprehensive as for the manufacturing sector. For example, out of a population of 13,280 enterprises in the retail trade sector, only 26 were sampled (for more details see Appendix A of Industry Canada, 2011). In contrast, a third of all manufacturing enterprises in the target population (4,394 out of 12,846) were included in the sample. This implies that some non-manufacturing enterprises have large sampling weights that should be kept in mind when analyzing the results.

Despite this potential issue with sampling weights, it is possible to further decompose the M P i n d e x distribution by two-digit NAICS industry. Two conclusions can be drawn from Figures 5, 6, 7 and 8 in Appendix C. First, there is a lot of variation in M P i n d e x within a sector, no matter which sector is considered. Second, there is much more variation among the distributions of non-manufacturing sectors compared to two-digit manufacturing sectors. As mentioned in the previous paragraph, part of these results are due to small sample sizes and high sampling weights for non-manufacturing industries.

The ICAP report on retail also reported that the relationships between MP and size, MNE status, education and ownership have the same sign as in the manufacturing sector. In addition, the authors found that US-owned retail enterprises in Canada are better managed compared to Canadian-owned. Other non-manufacturing sectors studied by BVR, in separate initiatives, are health care services and school (Bloom et al., 2010).

Competition was found to be an important determinant of MP by BVR. Bloom et al. (2012b) mentioned that competition may affect MP through at least two mechanisms. First, this can happen through reallocation of resources toward better managed enterprises. In other words, competition drives badly managed enterprises out of business. Second, competition may reveal information about competitors' MP. This causes managers to revise their "over-optimistic" perceptions of their own performance and to increase managerial efforts. Overall, BVR found that more competition is associated with better MP.

Competition in the BVR literature is measured by the number of competitors, the penetration rate of imports and the Lerner index. In this study, the vector COMP includes four indicators of competition related to the enterprise's main market: the number of goods that compete against the enterprise's highest selling product; the presence of a MNE; the number of competitors; and the entry of new competitors. The main market of an enterprise is defined as the geographic area from which the highest share of revenue from its highest selling product is derived.Footnote 1 Figure 3 shows that enterprises facing a MNE in their main market tend to have more structured MP. This is, again, consistent with the BVR results mentioned previously. In addition to COMP, a Lerner index is also included in Equation (2) as done in Bloom and Van Reenen (2006). A sensitivity analysis was also conducted for the Lerner index and it was bounded to a -10 to 1 interval.

Figure 3: Distributions of MPindex by presence of MNE
– Weighted densities, total sample N = 4,227 –
Graph of distributions of MPindex by presence of MNE – Weighted densities, total sample N = 4,227 (the long description is located below the image)
Source: Survey of Innovation and Business Strategy, 2009
Description of Figure 3

This figure describes the distribution of enterprises based on their MP index score across the potential values, which are bounded between 0 – 1 on the X-axis. Note that the Y-axis is suppressed for confidentiality reasons. This figure contains three lines: the first line represents enterprises operating in the manufacturing sector (same line as described in Figure 2), the second line represents firms with a MNE (multi-national enterprise) operating in their main market, and the third line represents firms that do not have an MNE operating in their main market. Starting with the line describing enterprises with a MNE operating in their main market, this line follows a similar path as the line for enterprises in the manufacturing sector. In particular, the line follows a general upward trend as we move across the X-axis until reaching its high-point just under 0.6. Then, the line follows a gradual decrease until reaching 1. The line is slightly higher than the line for enterprises in manufacturing industries near the beginning of the X-axis (i.e., around 0.15 - 0.3) and is slightly higher near the end of the X-axis (i.e., 0.7 – 0.9). The line describing enterprises not facing competition from an MNE in their main market is quite different from the other two lines. Specifically, the line is reasonably similar at the beginning of the X-axis (i.e., between 0 – 0.2), but shows a rather dramatic increase with a high-point centered on an MP index value just under 0.4. Following the high-point, the line decreases sharply remaining under the other two lines for the latter half of the X-axis. This suggests that enterprises that do not face competition from an MNE in their main market have less structured management practices.


2.3 MP and Labour Productivity

This part aims to estimate the relationship between MP and performance of enterprises operating in Canada, controlling for environmental factors and firm characteristics. The relationship estimated is given by:

P e r f i = γ 0 M P I i n d e x + γ 1 X i + γ 2 C O M P i + υ i
(3)

Perf denotes the performance of the enterprise in 2008, either measured by sales over employment (sales) or profits over employment (profits). The estimated parameter MPindex, is expected to be positive as in the BVR literature (see for example Bloom and Van Reenen, 2010). Note that while SIBS data are only available for 2009, sales and profits for 2009 were not timely available. It is thus assumed that MP were consistent between the two years. In addition to the variable in X, the ratio of capital over employment (cap) is also included. It is as the sum of tangible and intangible assets over employment.

There is some evidence in the empirical literature that competition increases productivity. For example, Nickell (1996) reported that the increase in the number of competitors is associated with significantly higher total factor productivity growth. Aghion et al. (2009) suggested that entry of new firms may increase productivity growth, but only in industries that are close to the technology frontier. Griffith et al. (2010) also show that the increased competition in the European Union, measured by the level of profitability, had a positive impact on productivity growth. However, Blanchflower and Machin (1995) did not find much evidence to support a positive relationship between competition and productivity growth. This suggests that different measures of competition matter for firm performance, which justify the inclusion of the COMP variables in Equation (3).

Equation (3) is estimated using the ROBUSTREG SAS procedure which uses residuals from an initial linear regression to identify the outliers and leverage points. Based on these residuals, a set of weights is produced and applied to the data and the model is estimated using an iterative algorithm.Footnote 2 Consequently, each outlier and bad leverage point receives a lower weight, which can be set to zero in extreme cases. Finally, observations for which the sales to employment ratio was equal to zero were removed, but negative profits were allowed.

2.4 MP and Innovation

There is some evidence in the non-BVR literature that other intangibles, such as human capital and organizational changes, are positively related to innovation (Becheikh et al., 2006). Using the Workplace and Employee Survey from 1999–2006, Dostie and Paré (2013) showed that both firm-sponsored classroom and on-the-job training lead to more innovation in Canada. Arvanitis et al. (2013) found similar results for Switzerland for human capital and organizational changes. There are no results on the relationship between innovation output and MP in the BVR literature, but Bloom et al. (2013) reported a positive correlation between R&D (innovation input) and MP for US establishments.

SIBS data suggest that innovation and MP are indeed positively associated. Figure 4 clearly shows that enterprises that have introduced innovation tend to have more structured MP compared to non-innovators. To further investigate this relationship, the following empirical specifications were estimated:

I i   =   δ 0 M P i + δ 1 X i + δ i C O M P i + μ i
(4)
I i k   =   δ 0 k M P i + δ i k X i + δ i k C O M P i + μ i k
(5)

In Equation (4), I is a binary variable capturing whether the enterprise innovated from 2007–09. An innovator is defined as an enterprise that introduced any of the usual four types of innovation: process (PRCS), organizational (ORGZ), product (PRDT) or marketing (MRKT). The vector X contains the same variables as for Equation (2) plus R&D expenditures in 2004—results are robust to other definitions of R&D, 2004–06 average for instance—and the number of advanced technologies used by the enterprises in 2009. As μi is assumed to be normally distributed, so a simple Probit is used to estimate the relationship.

Figure 4: Distributions of MPindex by innovator status
– Weighted densities, total sample N = 4,227 –
Graph of distributions of MPindex by innovator status – Weighted densities, total sample N = 4,227 (the long description is located below the image)
Source: Survey of Innovation and Business Strategy, 2009
Note: Firm sizes are based on individual labour units.
Description of Figure 4

This figure describes the distribution of enterprises based on their MP index score across the potential values, which are bounded between 0 – 1 on the X-axis. Note that the Y-axis is suppressed for confidentiality reasons. This figure contains three lines: the first line represents enterprises operating in the manufacturing sector (same line as described in Figure 2), the second line represents enterprises that have introduced an innovation between 2007–09, and the third line represents enterprises that did not introduce an innovation between 2007–09. The line for innovators closely resembles the line for enterprises operating in the manufacturing industry, though it is less smooth. However, the line describing non-innovators is quite different from the other lines. In particular, this line shows a large peak centered just under 0.4 on the X-axis. Following its high-point, the line declines rapidly, dipping below the other two lines by approximately 0.45 on the X-axis, and remains below the other two for the latter half of the X-axis. This shows that non-innovators are more likely to have less structured management practices relative to innovators.

For Equation (5), three additional definitions of innovation were used: i) technological (PRCS–PRDT) versus non-technological (ORGZ–MRKT); ii) PRCS–ORGZ versus PRDT–MRKT; iii) and PRCS, ORGZ, PRDT and MRKT. Innovation variables remained binary variables, but k sets of parameters were estimated. It was assumed that μk were normally distributed and correlated with each other, so multivariate Probits were used to estimate Equation (5). It was expected that some of δ will be positive, but it was unclear whether this relationship were to change depending on the type of innovation.

The impact of competition on innovation is not clear from the literature. Aghion et al. (2001) showed that an increase in product market competition has a positive impact on innovation. Similarly, Aghion et al. (2009) noted that entry of competitors spurs innovation in industries close to the technology frontier, a result that reflects their result for productivity. However, Boone (2000) reported that the effect of competition on innovation depends on the relative efficiency of the enterprise compared to its competitors. Therefore, in light of this brief overview of the literature, it is unclear what to expect about the relationship between MPi ndex and COMP.


3. Data

Most variables were derived from the Survey of Innovation and Business Strategy 2009 (SIBS). The SIBS target population included all enterprises with at least 20 employees and $250,000 of revenue in NAICS 11 to 56, but manufacturing enterprises were oversampled. The statistics and results presented in this paper are weighted, so the results can be generalized to the target population. The SIBS final sample consists of 4,227 enterprises.

Data on MP, competition intensity (except the Lerner index), innovation and use of advanced technology were taken from the SIBS. In addition, some control variables such as location of head office and the share of workers with a university degree were also extracted from the SIBS. The other control variables were derived from the Business Register (province of location, industry and multi-establishment status).

SIBS data needed to be supplemented by information from other Statistics Canada administrative databases. Sales, profits, assets (tangible and intangible) and cost of goods sold variables were extracted from the General Index of Financial Information database. Dollars variables were expressed in $M and were deflated using the National Account Productivity price index (KLEMS) produced by Statistics Canada.

Employment was extracted from the Longitudinal Employment Analysis Program database. The individual labour unit (ILU) was used to construct all employment variables, including firm size. ILU is not a count of the number of employees, but a constructed variable reflecting the share of employment attributable to an enterprise. For example, if an employee works in two enterprises and derived half of his salary from each, ILU is equal to 0.5 for both enterprises for this employee.

The Research and Development in Canadian Industry database provides data on R&D expenditures. As a practical matter for the year 2004, this database was a census of all R&D performers in Canada, so any missing observation was considered to be a zero.


4. Results

4.1 Determinants of the MPindex

Table 1 summarizes the results from the regression of MPindex on firm characteristics and competition indicators. The full set of estimated parameters is presented in Table 4 of Appendix D. Overall, results are fairly robust across sectors and consistent with the BVR literature.

Firm size is one of the most important variables. Consistent with the evidence in Figure 1, there is a positive relationship between firm size and MPindex and the magnitude of the estimated parameters increases with size. These results hold for both sectors and estimated parameters for the non-manufacturing sector are larger. Enterprises with a higher percentage of workers with a university degree ( % U N I V ) tend to have more structured MP. The location of head office also matters, especially if it is located in the US. The multiple establishments variable is negatively correlated with MPindex for non-manufacturing enterprises only.

Table 1: Summary of Eq. (2): Determinants of MPindex
Dep. var.: MPindex, weighted OLS regressions
VariableManufacturingNon-manufacturing
– Characteristics –
MEDIUM0.10250.1544
LARGE0.10830.1827
XLARGE0.18730.2378
%LARGE0.00180.0026
HQ_US0.05860.0943
HQ_EU0.06680.0543
HQ_ROW0.0199−0.0994
MULTI0.0083−0.0211
Competition variables
MNE0.05870.1022
#COMP−0.0081−0.0006
ENTRY0.0143−0.0306
#GOODS0.00060.0106
LERNER−0.00630.0998
Estimated parameters significant at a 0.10 level or less are in bold.
Detailed results are presented in Table 4 of Appendix D.

In terms of competition, the presence of a MNE in the enterprise's main market for its highest selling product is positively correlated with MPindex. Few other competition variables are found to be significant. For the manufacturing sector, there is a negative relationship between the number of competitors in the main market ( # C O M P ) and MPindex as in Bloom and Van Reenen (2007). For non-manufacturing enterprises, a positive link has been identified between the Lerner index and MPindex. Exclusion of the Lerner index does not have much impact as shown in Table 4.


4.2 Relationships between Sales, Profits and MPindex

Table 2 summarizes the results for sales and profits regressions. The full set of estimated parameters is presented in Table 5 of Appendix D. The main difference between sectors relates to MPindex as the estimated coefficient for this variable is positive and significant for manufacturing enterprises, but not for non-manufacturing enterprises.

One possible explanation for this result is the heterogeneity among enterprises outside the manufacturing sector. Complementary evidence on this heterogeneity is provided in Figures 6 to 8 in Appendix C. These figures show that, at least in terms of MP distributions, non-manufacturing industries are less similar among themselves compared to the three-digit manufacturing industries (Figure 5). The results may also be driven by some non-manufacturing enterprises with large sampling weights—the SIBS coverage being much less comprehensive outside the manufacturing sector.

Table 2: Summary of Eq. (3): Economic performance and MPindex
– Dep. var.: sales and profits (intensity), weighted robust regressions –
 ManufacturingNon-manufacturing
VariableSalesProfitsSalesProfits
– Characteristics –
cap++++
MPindex++ no data no data
MEDIUM++ no data
XLARGE+ no data
LARGE+
%UNIV no data+++
HQUS++++
HQEU++ no data no data
HQROW no data no data no data
MULTI++++
Competition Variables
MNE+ no data
#COMP no data+++
ENTRY no data no data
#GOODS no data
+/−: significant at a 0.10 level or less.
Detailed results are presented in Table 5 of Appendix D.

To further investigate this issue, Equation (3) was estimated for two additional non-manufacturing sectors groupings: resources-based and best-managed. The first group includes NAICS 11 (agriculture), 21 (mining and oil and gas extraction), 22 (utilities) and 23 (construction). The second is a grouping of NAICS 52 (finance and insurance) and 54 (professional services), which are the two services sectors with the highest MPindex average (Figure 7). The estimated coefficient for MPindex (not shown here) is negative and significant in the sales regression for the best-managed group and in the profits regression for the resources-based group. While these results seems to confirm the importance of heterogeneity and the effect of the sampling strategy used, it is hard to explain why the sign for MPindex is negative.

A second explanation for the non-significance of MPindex in Table 2 is that the SIBS MP questions were less appropriate for non-manufacturing activities. Testing of the questionnaire revealed that respondents in non-manufacturing enterprises had more difficulty answering some of these questions. A different framework for the non-manufacturing sector would have been better, as in Institute for Competitiveness and Prosperity (2010) for the retail trade sector—but this was not possible to implement at the time SIBS was developed. In turn, this suggests that MPindex may not accurately measure the extent of MP in non-manufacturing industries.

Turning to the competition indicators, the results in Table 2 indicate that facing more competitors (#COMP) is associated with better performance while a higher number of goods competing with the enterprise's highest selling product (#GOODS) has the opposite effect. Presence of a MNE and entry of new competitors are also negatively correlated with profits in both sectors. Two conclusions can be drawn from the competition results. First, not all aspects of competition correlate in the same way with firm performance. Second, profits seem more sensitive to competition compared to sales.

Among the other variables, the estimated coefficient of asset-to-employment ratio cap has a positive sign in all specifications, which is also the case for HQ_US (head office located in the United States) and MUTLI (multi-establishments). The percentage of workers with a university degree is positively correlated with performance, except for the sales regression for manufacturing enterprises. Finally, performance increased with firm size only for manufacturing enterprises in the sales regression.


4.3 Relationships between Innovation and MPindex

Table 3 summarizes the results from the innovation regressions. Column (1) shows the results for the general indicator of innovation; Column (2) shows the ones for technological and non-technological innovations; Column (3) shows the results for the PRDT–MRKT and PRCS–ORGZ innovation groupings; and the last column shows them for the four types of innovation. All estimated parameters are presented in Tables (5) to (7) in Appendix D.

The most important result is the estimated positive correlation between MPindex and innovation. This holds for manufacturing enterprises no matter how innovation is measured and for most indicators for non-manufacturing enterprises. This is an important result because it shows how critical MP is for innovation even when accounting for other control variables such as innovation inputs, firm size, enterprise structure and competition indicators.

Table 3: Summary of Eq. (4) and (5): Innovation and MPindex
– Dep. var.: see table, weighted Probit –
Manufacturing
Variable(1) no data(2) no data(3) no data(4)
INNO no dataTECHNTECH no dataPD–MKPC–OG no dataPRCSORGZPRDTMRKT
– Characteristics –
MP index+ no data++ no data++ no data++++
RD no data no data no data no data no data no data no data no data no data no data no data no data
ADVTECH+ no data++ no data++ no data no data no data no data
MEDIUM no data no data no data no data no data no data no data no data no data no data
LARGE no data no data no data no data no data no data no data no data no data
XLARGE no data no data no data no data no data no data++++
%UNIV no data no data no data+ no data+ no data no data no data no data++
HQ_US no data no data no data no data no data no data no data no data no data no data
HQ_EU no data no data no data no data no data no data+ no data no data no data no data
HQ_ROW no data no data no data no data no data no data no data
MULTI no data no data no data no data no data no data no data no data no data no data no data
Competition Variables
MNE+ no data no data no data no data+ no data no data no data no data+ no data
#COMP no data no data no data no data no data no data no data no data no data
ENTRY+ no data++ no data++ no data++++
#GOODS no data no data+ no data no data+ no data no data no data no data++
Table 3: Summary of Eq. (4) and (5): Innovation andMPindex
– Dep. var.: see table, weighted Probit –(Non-manufacturing)
Non-Manufacturing
Variable(1)(2)(3)(4)
INNOTECHNTECHPD–MKPC–OGPRCSORGZPRDTMRKT
– Characteristics –
MP no data no data no data+ no data++ no data++ no data+
RD no data no data+ no data no data no data no data no data no data no data no data no data
ADVTECH+ no data+ no data no data no data+ no data+++ no data
MEDIUM  no data no data no data no data no data no data no data no data no data no data+ no data
LARGE+ no data no data no data+ no data no data no data+ no data+
XLARGE no data no data no data no data+ no data no data no data++ no data
%UNIV no data no data no data no data no data no data no data no data no data no data no data no data
HQ_US no data no data no data no data no data+ no data no data+ no data no data
HQ_EU no data no data no data no data no data no data no data no data no data no data no data no data
HQ_ROW no data no data no data no data no data no data no data no data no data no data no data
MULTI no data no data no data no data no data no data no data no data
Competition variables
MNE+ no data+ no data no data no data+ no data+ no data+ no data
#COMP no data no data no data no data no data no data no data+ no data
ENTRY+ no data++ no data+ no data no data no data+++
#GOODS no data no data no data no data no data no data no data no data no data
+/–: significant at a 0.10 level or less.
Detailed results are presented in Tables 5, 6 and 7 of Appendix D.

Of the two inputs to innovation, R&D expenditures (RD) and the number of advanced technologies used (ADVTECH), only the latter is found to be positively correlated with innovation. For manufacturing enterprises, this relationship holds for all aggregated measures of innovation as shown by Columns (1), (2) and (3). In contrast, for non-manufacturing enterprises, the results suggest that ADVTECH is positively correlated with PRCS, ORGZ and PRDT innovations.

RD is significant only for technological innovation introduced by non-manufacturing enterprises (Column (2)). Although a similar result was found in Brouillette (2013), R&D is nevertheless still important for innovation in Canada. It is likely that R&D expenditures affect indirectly innovation, for example, through the use of advanced technology.

The influence of the competition indicators depend on how competition is measured. The most striking result is the entry of new competitors in the enterprise's main market for its highest selling product. The estimated parameters for ENTRY is systematically positive and significant for all types of innovation in the manufacturing sector regressions and for almost all cases in the non-manufacturing sector regressions, a result consistent with Aghion et al. (2009). Another aspect of competition that is positively correlated with innovation is the presence of a MNE in the enterprise's main market although fewer significant relationships were found. Results are consistent for both manufacturing and non-manufacturing sectors. The link between the number of competitors (#COMP) in the main market is negative overall for both sectors. This contrasts with the results for the performance analysis where this variable was found to have a positive and significant sign.

The results also show that a small number of positive estimated correlations between innovation and %UNIV were found in the manufacturing sector, but no such association was found in the non-manufacturing sector. This is in contrast to the results of the performance analysis where %UNIV is more important for non-manufacturing enterprises.

The relationship between firm size and innovation is less clear. Compared to small ones, large and x-large non-manufacturing enterprises are more susceptible to innovate, but results depend on the innovation indicator used. Results are even more mixed for manufacturing enterprises. Finally, the links between innovation and the location of headquarters and multi-establishment status are mostly negative.


5. Conclusion

This analysis confirms the presence of a positive correlation between firms' financial performance and business innovation and well structured management practices (MP). This study also provides evidence on the role played by competition.

The results show large differences in MP distribution across industries. Firm size, presence of a multinational enterprise (MNE) in the enterprise's main market, head office located in the United States and the percentage of workers with a university degree are the main determinants of MP. In terms of firm performance, MP are positively correlated with the intensity of sales and profits in the manufacturing sector, but not in the non-manufacturing sector. In contrast, MP are important for all industries when it comes to innovation. Finally, the role of competition depends on its nature and to a certain extent of the industry considered. The presence of a MNE and entry of new competitors are positively correlated with innovation, while it is the number of competitors that is important for firm performance.

From a policy perspective, this analysis highlights the importance of intangible capital for firm performance and business innovation. While this analysis focusses on MP, other intangibles such as managers' education, investments in skills and intellectual property management are perhaps equally important for growth and innovation. More research at the enterprise-level on theses intangibles is needed to better understand their interactions with productivity growth. This would allow better support to enterprises so that they can achieve their full potential.

Two caveats should be mentioned. The first one relates to data limitation for performance indicators. As described in Section 3, the last year available for these data was 2008. Since most SIBS data cover 2009 only—this is the case for MP and competition indicators—the implicit assumption made is that these variables did not change between 2008 and 2009.

The second caveat is about the endogeneity of the MP index. Although this is acknowledged as a serious issue, it was not possible to address it in this paper for lack of instruments. This may change with the availability of the SIBS 2012 data in 2014. By combining the 2009 and 2012 waves, this will provide a measure of the change in management practices that could be compared to the change in firms' performance or innovation.


References

  • Aghion, Philippe, Christopher Harris, Peter Howitt, and John Vickers (2001) 'Competition, Imitation and Growth with Step-by-Step Innovation.' Review of Economic Studies 68(3), 467–492.
  • Aghion, Philippe, Richard Blundell, Rachel Griffith, Peter Howitt, and Susanne Prantl (2009) 'The Effect of Entry on Incumbent Innovation and Productivity.' Review of Economics and Statistics 91(1), 20–32.
  • Alchian, Armen A., and Harold Demsetz (1972) 'Production, Information Costs, and Economic Organization.' American Economic Review 62(5), 777–795.
  • Arvanitis, Spyros, Euripidis N. Loukis, and Vasiliki Diamantopoulou (2013) 'Are ICT, Workplace Organization and Human Capital Relevant for Innovation? A Comparative Study Based on Swiss and Greek Micro Data.' KOF Working Paper No. 333.
  • Autor, David H., and David Scarborough (2008) 'Does Job Testing Harm Minority Workers? Evidence from Retail Establishments.' Quarterly Journal of Economics 123(1), 219–277.
  • Bandiera, Oriana, Iwan Barankay, and Imran Rasul (2007) 'Incentives for managers and inequality among workers: Evidence from a firm-level experiment.' The Quarterly Journal of Economics 122(2), 729–773.
  • Bartelsman, Eric J., and Mark E. Doms (2000) 'Understanding Productivity: Lessons from Longitudinal Microdata.' Journal of Economic Literature 38(3), 569–594.
  • Becheikh, Nizar, Réjean Landry, and Nabil Amara (2006) 'Lesson from Innovation Empirical Studies in the Manufacturing Sector: A Systematic Review of the Literature from 2993–2003.' Technovation 26(5–6), 644–664.
  • Bertrand, Marianne, and Antoinette Schoar (2003) 'Managing with Style: The Effect of Managers on Firms Policies.' Quarterly Journal of Economics 68(4), 1169–1208.
  • Black, Sandra, and Lisa M. Lynch (2001) 'How to Compete: The Impact of Workplace Practices and Information Technology on Productivity.' The Review of Economics and Statistics 83(3), 434–445.
  • Blanchflower, David, and Stephen Machin (1995) 'Product Market Competition, Wages, and Productivity: International Evidence from Establishment-Level Data'.
  • Bloom, Nicholas (2010) 'Measuring and Explaining Management Practices in Canada'.
  • Bloom, Nicholas, and John Van Reenen (2006) 'Measuring and Explaining Management Practices Across Firms and Countries'.

    —(2007) 'Measuring and Explaining Management Practices Across Firms and Countries.' Quarterly Journal of Economics 122(4), 1351–1408.

    —(2010) 'Why Do Management Practices Differ across Firms and Countries?' Journal of Economic Perspectives 24(1), 203–224.

  • Bloom, Nicholas, and Raffaella Sadun (2009) 'Do Private Equity-Owned Firms Have Better Management Practice?' In Globalization of Alternatice Investments, Working Papers Volume 2 – The Global Economic Impact of Private Equity Report 2009, ed. Anuradha Guring and Josh Lerner (World Economic Forum).
  • Bloom, Nicholas, Benn Eifert, Aprajit Mahajan, David McKenzie, and John Roberts (2011) 'Does Management Matter? Evidence from India.' International Growth Center Working Paper no. 10/0873.
  • Bloom, Nicholas, Carol Propper, Stephan Seiler, and John Van Reenen (2010) 'The Impact of Competition on Management Quality: Evidence from Public Hospital.' Center for Economic Performance Discussion Paper No 983.
  • Bloom, Nicholas, Erik Brynjolfsson, Lucia Foster, Ron Jarmin, Itay SaportaEksten, and John Van Reenen (2013) 'Management in America.' Presented at the American Economic Association Meeting 2013.
  • Bloom, Nicholas, Raffaella Sadun, and John Van Reenen (2012a) 'Americans Do IT Better: US Multinationals and the Productivity Miracle.' American Economic Review 102(1), 167–201.

    —(2012b) 'Management as a Technology.' Presented at the American Economic Association Meeting 2013.

  • Boone, Jan (2000) 'Competitive Pressure: The Effects on Investments in Product and Process Innovation.' RAND Journal of Economics 31(3), 549–569.
  • Bresnahan, Timothy F. (1999) 'Computerisation and Wage Dispersion: An Analytical Reinterpretation.' Economic Journal 109(456), F390–415.
  • Bresnahan, Timothy F., Erik Brynjolfsson, and Lorin M. Hitt (2002) 'Information Technology, Workplace Organization, and the Demand for Skilled Labor: Firm-Level Evidence.' The Quarterly Journal of Economics 117, 339–376.
  • Brouillette, Dany (2013) 'Drivers of Innovation, Complementarity of Innovation, and Performance of Enterprises in Canada.' mimeo, Industry Canada.
  • Cappelli, Peter, and David Neumark (1999) 'Do "High Performance" Work Practices Improves Establishment-Level Outcomes?' NBER Working Paper 7374.
  • Competition Policy Review Panel (2008) 'Compete to Win.' Government of Canada, Ottawa.
  • Dostie, Benoit, and Jonathan Paré (2013) 'Formation Parrainée par l'Employeur et Performance en Innovation.' Centre sur la productivité et la prospérité.
  • Drummond, Don, and Alistair Bentley (2010) 'The productivity puzzle: Why is the Canadian record so poor and what can be done about it?' TD Economics Special Report.
  • Expert Panel on Business Innovation (2009) 'Innovation and Business Strategies: Why Canada Falls Short.' Council of Canadian Academies (CCA).
  • Expert Review Panel on Research and Development (2012) 'Innovation Canada: A Call to Action.' Government of Canada, Ottawa.
  • Griffith, Rachel, Rupert Harrison, and Helen Simpson (2010) 'Product Market Reform and Innovation in the EU.' Scandinavian Journal of Economics 112(2), 389–415.
  • Huselid, Mark A. (1995) 'The Impact of Human Resource Management Practices on Turnover, Productivity, and Corporate Financial Performance.' Academy of Management Journal 38(3), 635–672.
  • Ichniowski, Casey, Kathryn Shaw, and Giovanna Prennushi (1997) 'The Effects of Human Resource Management Practices on Productivity: A Study of Steel Finishing Lines.' American Economic Review 87(3), 291–313.
  • Industry Canada (2011) 'Business Innovation and Strategy: A Canadian Perspective.' Government of Canada, Ottawa.
  • Institute for Competitiveness and Prosperity (2009) 'Management Matters.' ICAP Working Paper 12, Toronto.

    —(2010) 'Management Matters in Retail.' ICAP Working Paper 13, Toronto. Lazear, Edward P. (2000).

    —'Performance Pay and Productivity.' American Economic Review 90(5), 1346–1361.

  • Macduffie, John Paul (1995) 'Human Resource Bundles and Manufacturing Performance: Organizational Logic and Flexible Production Systems in the World Auto Industry.' Industrial and Labour Relations Review 48(2), 197–221.
  • Nickell, Stephen J. (1996) 'Competition and Corporate Performance.' Journal of Political Economy 104(4), 724–746.
  • Osterman, Paul (1994) 'How Common Is Workplace Transformation and Who Adopts it?' Industrial and Labour Relations Review 47(2), 173–188.
  • Oyer, Paul, and Scott Schaefer (2011) 'Personnel Economics: Hiring and Incentives.' In Handbook of Labor Economics, Vol 4b, ed. Orley Ashenfelter and David Card (North Holland) pp. 1769–1823.
  • Powell, Thomas C. (1995) 'Total Quality Management as Competitive Advantage: A Review and Empirical Study.' Strategic Management Journal 16(1), 15–37.
  • Syverson, Chad (2004a) 'Market Structure and Productivity: A Concrete Example.' Journal of Political Economy 112(6), 1181–1222.

    —(2004b) 'Product Substitutability and Productivity Dispersion.' The Review of Economics and Statistics 86(2), 534–550.

    —(2011) 'What Determines Productivity.' Journal of Economic Literature 49(2), 326–365.

  • Van Reenen, John (2011) 'Boss-Onomics: Does Management Matter?' Industry Canada's Distinguished Speaker Series.

Appendix A. The Management Practices Indices

A.1 Comparison of MP Indicators from the BVR Survey and the SIBS

Contents comparison between BVR and SIBS indices
BVR index
(Bloom and Van Reenen, 2007)
BVR types of
indicators
SIBS index
(MPindex)
1 Modern manufacturing, introductionOperations No equivalent
2 Modern manufacturing, rationaleOperations No equivalent
3 Process documentationOperationsQ52The enterprise has a systematic process to resolve problems associated with production of goods or delivery of services
4 Performance trackingMonitoringQ53Number of key production performance indicators (KPPI) monitored in the enterprise
5 Performance reviewMonitoringQ54Frequency at which KPPI are shown to managers of operations
Q55Frequency at which KPPI are shown to workers
Q56Frequency of review of KPPI by top and middle managers
6 Performance dialogueMonitoringQ62Employees' involvement in the decision-making process on task allocation
7 Consequence managementMonitoring No equivalent
8 Target breadthTargets No equivalent
9 Target interconnectionTargets No equivalent
10 Target time horizonTargetsQ58Time frame of the enterprise production performance targets for its highest product
11 Targets are stretchingTargets No equivalent
12 Performance clarityMonitoringQ64dFormal performance agreements based on objective, quantifiable results are prepared for managerial/supervisory/executive employees at least annually
Q64e, Q64fFormal appraisals are conducted for the majority of managerial and non-managerial staff at least annually
13 Managing human capitalTargets No equivalent
14 Rewardinghigh-performanceIncentivesQ59How does the enterprise reward production performance target achievement
15 Removing poor performersIncentivesQ61Enterprise's main policy to deal with employees not meeting expectations
16 Promoting high performersIncentivesQ60Enterprise's main way to promote employees
Q64bFormal training programs are available to employees to teach them the skills required to perform their job
Q64cFormal training programs are available to employees to increase their promotability
17 Attracting human capitalIncentivesQ64aAt least one of the following selection methods to select candidates is used: personality/attitude tests, intelligence or aptitude tests, work sample
18 Retaining human capitalIncentivesQ64gIncentives programs such as employee stock ownership,profit-sharing,gain-sharing or merit bonus are available to non-managerial employees
Q64hIncentives programs are available to managerial, supervisory, or executive employees
Q64iIncentives programs are available to all employees

A.2 Details on MPindex

Indicators and score of MPindex
QUESTION SCORE QUESTION SCORE
Q52 Systematic process No 0 Q64a Use selection No 0
to solve problems Yes 1 methods for candidates Yes 1
Q53 #KPPI None 0 Q64b Training No 0
  At least one 1 for skills Yes 1
Q54 Frequency KPPI Never/don't know 0 Q64c Training for No 0
shown to managers Any frequency 1 promotion Yes 1
Q55 Frequency KPPI Never/don't know 0 Q64d Performance No 0
shown to workers Any frequency 1 agreements Yes 1
Q56 Frequency KPPI Don't know 0 Q64e Appraisal No 0
shown to executives Rarely 13 for workers Yes 1
  Periodically 23      
  Continually 1      
Q58 Time frame No targets 0 Q64f Appraisal No 0
for performance Short-term 13 for managers Yes 1
targets Long-term 23      
  Both 1      
Q59 Rewards None 0 Q64g Incentives No 0
  Managers only ½ for workers Yes 1
  All staff 1      
Q60 Promotion Tenure 0 Q64h Incentives No 0
based on … Effort and tenure ½ for managers and Yes 1
  Effort 1 executives    
Q61 If employees Never moved 0 Q64i Incentives No 0
don't meet Warned 13 for all Yes 1
expectations Warned, re-trained 23      
  Removed 1      
Q62 Employees No 0      
make decisions yes 1      

The SIBS questionnaire is constructed in such a way that Q54 to Q59 have to be answered only if the number of key production performance indicators monitored (Q53) is greater than zero. This means that for all enterprises not monitoring any, the maximum score is 14, not 19. A score of 0 was nevertheless assigned to Q54–Q59 for these enterprises so that an enterprise with fewer MP obtains a lower value of MPindex compared to one with more practices. Without this assumption, Enterprise A with MPindex = 13/14 (0.93) would rank higher than Enterprise B with MPindex = 17/19 (0.89). Enterprise B should, however, rank higher compared to Enterprise A because it has a broader set of MP and its performance for indicators other than those of Q54–Q59 is as good as for Enterprise A.


Appendix B. Variables Definition

Dependent variables
NAMEDESCRIPTIONSOURCE
MPManagement practice index (2009)SIBS
salesSales over employment ratio in $M (2008)GIFI, LEAP
profitsProfits over employment ratio in $M (2008)GIFI, LEAP
INNO= 1 if enterprise has innovated (2007–09)SIBS
IPRCS= 1 if PRCS innovation has been introduced (2007–09)SIBS
IORGZ = 1 if ORGZ innovation has been introduced (2007–09)SIBS
IPRDT = 1 if PRDT innovation has been introduced (2007–09)SIBS
IMRKT= 1 if MRKT innovation has been introduced (2007–09)SIBS
ITECH= 1 if PRCS or PRDT have been introduced (2007–09)SIBS
INONTECH= 1 if ORGZ or MRKT have been introduced (2007–09)SIBS
IPDMK= 1 if PRDT or MRKT have been introduced (2007–09)SIBS
IPCOG = 1 if PRCS or ORGZ have been introduced (2007–09) SIBS
Competition variables
NAMEDESCRIPTIONSOURCE
MNE= 1 if a MNE is present in the main market (2009)SIBS
#COMPNumber of competitors in the main market (2009)
Categories: 1 = 1 comp.; 2 = 2 comp.; 3 = 3 comp.;
4 = 4–5 comp.; 5 = 6–10 comp.;
6 = 11–20 comp.; 7 = 20+ comp.
SIBS
ENTRY= 1 if a new competitor entered main market (2009)SIBS
#GOODSNumber of competing products in main market (2009)
Categories: 1 = 1–2 prod.; 2 = 2–4 prod.; 3 = 5–7 prod.;
4 = 8–9 prod.; 5 = 10–19 prod.; 6 = 20–49 prod.;
7 = 50–100 prod.; 8 = 100+ prod.
SIBS
SALESSales in $M (2008)GIFI
COGS Cost of goods sold in $M (2008) GIFI
LERNER Lerner index (2008): ( S A L E S C O G S ) S A L E S GIFI
Control variables
NAMEDESCRIPTIONSOURCE
SIZEBinary variables for firm size (ILU) (2009)
SMALL: [20-50[ ILU; MEDIUM: [50-100[ ILU;
LARGE: [100-250[ ILU; XLARGE: 250+ ILU.
SMALL size is the reference.
LEAP
PROVBinary variables for the location of the enterprise (2009)
QC: Québec; ON: Ontario; AB: Alberta;
BC: British Columbia; ROC: Rest of Canada.
ON is the reference.
BR
NAICSBinary variables for industries (2009)
MANU 1: NAICS 31; MANU 2: NAICS 32;
MANU 3: NAICS 33; RES: NAICS 11, 21, 22 and 2
SERV : NAICS 41, 44–45, 48–49, 51, 52, 53, 54, 55 and 56.
MANU 1 is the reference for manufacturing and RES is the reference for
non-manufacturing.
BR
HQBinary variables for location of head office (2009)
HQ CA: Canada; HQ US: United States;
HQ EU: Europe; HQ ROW: All other countries.
HQ CA is the reference.
SIBS
%UNIVPercentage of workers with university degree (1009)SIBS
MULTI = 1 if the enterprise has multiple establishmentsBR
CAP Sum of tangible and intangible assets in $M (2008)GIFI
capCAP over employment ratio in $M (2008)GIFI
RD R&D expenditures in $M (2004) RDCI
ADVTECH Number of advanced technologies used (2009) SIBS

Appendix C. Descriptive Statistics (weighted)

Statistics by sector
– Manufacturing N = 2,890 –
MP0.54 [0.4]#comp (%)size (%)
salesFootnote b ($M)0.16 [0.2]13.9 [0.4]small54.3
profitsFootnote c ($M)0.04 [0.1]24.9 [0.5]medium24.3
INNO (%)81.2 [0.9]39.0 [0.6]large13.8
IPRCS 58.1 [1.2]4–524.4 [1.0]xlarge7.2
IORGZ 53.9 [1.2]6–1023.2 [1.0]prov (%)
IPRDT 48.6 [1.1]11–209.8 [0.7]on41.9
IMRKT 40.0 [1.1]20+24.8 [1.0]qc28.5
ITECC 70.3#goods (%)ab9.1
INTECH 64.50–218.4bc12.2
IPDMK 61.82–423.4roc8.3
IPCOG 71.55–715.1hq (%)
CAPFootnote b ($M)0.08 [0.2]8–92.3hq_ca88.2 [0.5]
RDFootnote e ($M)0.90 [12.6]10–1916.0hq_us7.6 [0.4]
advtech1.64 [3.8]20–4910.2hq_eu3.1 [0.3]
%univFootnote a (%)11.1 [2.7]50–1003.6hq_row1.1 [0.2]
multi (%)15.5100+11.2no data
MNE (%)66.0 [1.1]naics (%)no data
entry (%)31.1 [1.1]manu_116.7no data
lernerFootnote d0.28manu_230.4no datano data
no datano datamanu_352.9no datano data

When available, standard error is in brackets.

Sources: Statistics Canada SIBS, GIFI, LEAP, RDCI and br.

Statistics by sector (Non-Manufacturing)
– Non-ManufacturingFootnote f N = 2,890 –
MP0.45 [1.4]#COMP (%)SIZE (%)
salesFootnote g ($M)0.18 [1.0]12.9 [1.5]SMALL59.1
profitsFootnote h ($M)0.05 [0.3]26.1 [3.4]MEDIUM18.8
INNO (%)63.6 [5.4]313.9 [2.3]LARGE17.0
IPRCS 27.9 [3.9]4–59.8 [1.5]XLARGE5.1
IORGZ 30.2 [4.2]6–1028.9 [4.4]PROV (%)
IPRDT 31.6 [4.2]11–2017.4 [4.4]ON36.2
IMRKT 34.3 [4.3]20+21.1 [3.2]QC27.1
ITECC 47.3#goods (%)AB12.1
INTECH 47.00–224.4BC11.8
IPDMK 44.12–422.7ROC12.9
IPCOG 45.35–712.6HQ (%)
CAPFootnote g ($M)0.09 [1.2]8–94.1HQ_CA95.4 [0.9]
RDFootnote k ($M)3.26 [40.1]10–1918.2HQ_US3.8 [0.9]
advtech0.97 [9.5]20–499.2HQ_EU0.5 [0.1]
%univFootnote i (%)16.9 [7.7]50–1003.1hq_row0.3 [0.1]
multi (%)21.1100+5.7
MNE (%)46.4 [5.5]naics (%)
entry (%)32.7 [5.2]serv75.5
lernerFootnote j 0.45 RES 24.5

When available, standard error is in brackets.

Sources: Statistics Canada SIBS, GIFI, LEAP, RDCI and BR.

Sample size for sales and profits regressions are smaller than the full SIBS sample (4,227). There are four explanations to this. First, sales and profits (GIFI) have some missing records. Second, as mentioned in Section 2.3, the ROBUSTREG SAS procedure assigned a weight—not the sampling weights—of zero to some outliers and bad leverage points which effectively removed them. Third, enterprises with a sales to employment ratio equals to zero were removed. Negative profits, however, are permitted. Fourth, the Lerner index was bounded between -10 and 0.

Figure 5: Distributions of MPindex for NAICS 31, 32 and 33

Graph of distributions of MPindex for NAICS 31, 32 and 33 (the long description is located below the image)
Source: Survey of Innovation and Business Strategy 2009
Description of Figure 5

This figure describes the distribution of enterprises based on their MP index score across the potential values, which are bounded between 0 – 1 on the X-axis. Note that the Y-axis is suppressed for confidentiality reasons. This figure contains four lines. The first line describes enterprises operating in manufacturing industries (same line as described in Figure 2), and lines two through four describe the enterprises operating in three subsets of the manufacturing industry – NAICS 31, NAICS 32 and NAICS 33, respectively. The lines for the three subsets of the manufacturing sector closely resemble the line describing the overall manufacturing sector. The line describing NAICS 31 is slightly higher near the beginning of the X-axis and slightly lower near the end of the X-axis than the line describing manufacturing more generally. The lines for NAICS 32 and NAICS 33 are an even closer match to the manufacturing line – they are nearly identical. The figure suggests that there is little variation in management practices across manufacturing industries at the 2-digit NAICS aggregation level.

Figure 6: Distributions of MPindex for NAICS 21 and 22

Graph of distributions of MPindex for NAICS 31, 32 and 33 (the long description is located below the image)
Source: Survey of Innovation and Business Strategy 2009
Description of Figure 5

This figure describes the distribution of enterprises based on their MP index score across the potential values, which are bounded between 0 – 1 on the X-axis. Note that the Y-axis is suppressed for confidentiality reasons. This figure contains three lines. The first line describes enterprises operating in manufacturing industries (same line as described in Figure 2), and the other two lines describe to subsets of firms operating in the non-manufacturing sector – NAICS 21 (Mining, quarrying, and oil and gas extraction) and NAICS 22 (Utilities). The line describing NAICS 21, is slightly more flat than what is shown for the line describing the manufacturing sector. However, the line for NAICS 21 is noticeably higher relative to the line describing the manufacturing sector near the beginning of the X-axis, while the line is similar near the end of the X-axis. The lines show that while enterprises operating in NAICS 21 have a similar concentration of observations near the high-end of the MP index, they have a higher concentration in the low-end and a lower concentration near the middle. In contrast, the line describing NAICS 22 differs substantially from the other two lines; the line shows relatively high spike centered just above 0.4 on the X-axis. This suggests that enterprises operating in NAICS 22 have a much higher concentration of enterprises with an MP index value close to the average for non-manufacturing enterprises (0.48). This figure, as well as Figures 5, 7, and 8, highlight that there is more variation among the enterprise distributions across MP index values in non-manufacturing two-digit NAICS industries relative to manufacturing two-digit NAICS industries.

Figure 7: Distributions of MPindex for NAICS 51, 52 and 54

Graph of distributions of MPindex for NAICS 51, 52 and 54 (the long description is located below the image)
Source: Survey of Innovation and Business Strategy 2009
Description of Figure 7

This figure describes the distribution of enterprises based on their MP index score across the potential values, which are bounded between 0 – 1 on the X-axis. Note that the Y-axis is suppressed for confidentiality reasons. This figure contains four lines. The first line describes enterprises operating in manufacturing industries (same line as described in Figure 2), and the other three lines describe the three subsets of firms operating in the non-manufacturing sector – NAICS 51 (Information and cultural industries), NAICS 52 (Finance and Insurance) and NAICS 54 (Professional, scientific and technical services). The line for NAICS 51 increases near the beginning of the X-axis (from 0 - 0.2 on the X-axis), after which, the line is mostly flat showing little movement until decreasing near the end of the X-axis (from 0.8 – 1). This line shows NAICS 51 has the highest concentration of observations at the low-end of the MP index (between 0 - 0.3). The line for NAICS 52 increases steadily across most of the X-axis (until approximately 0.85) before declining. This line shows NAICS 52 has the highest concentration of observations at the high-end of the MP index (between 0.8 – 1). The line for NAICS 54 increases steadily across the X-axis until approximately 0.6 on the X-axis, at which point, the line steadily decreases for the remainder of the X-axis. This figure, as well as Figures 5, 6, and 8, highlight that there is more variation among the enterprise distributions across MP index values in non-manufacturing two-digit NAICS industries relative to manufacturing two-digit NAICS industries.

Figure 8: Distributions of MPindex for NAICS 48 and other

Graph of distributions of MPindex for NAICS 48 and other (the long description is located below the image)
* All covered industries except NAICS 21, 22, 31-33, 48, 51, 52 and 54.
Source: Survey of Innovation and Business Strategy 2009
Description of Figure 8

This figure describes the distribution of enterprises based on their MP index score across the potential values, which are bounded between 0 – 1 on the X-axis. Note that the Y-axis is suppressed for confidentiality reasons. This figure contains three lines. The first line describes enterprises operating in manufacturing industries (same line as described in Figure 2), and the other two lines describe the two subsets of firms operating in the non-manufacturing sector – NAICS 48 (Transportation), and other (category includes all 2-digit NAICS non-manufacturing industries other than 48). The line for NAICS 48 increases near the beginning of the X-axis reaching a high-point centered over approximately 0.4 on the X-axis, after which, the line gradually declines for the remainder of the X-axis. The line for the other non-manufacturing category is similar to the line for NAICS 48 except that it has a higher peak near the beginning reaching its high-point just before 0.4 on the X-axis – i.e., slightly before the line for NAICS 48. This line shows NAICS 48 has a higher concentration of observations at the low-end of the MP index (between 0 - 0.4) than in the manufacturing sector. This figure, as well as Figures 5, 6, and 7, highlight that there is more variation among the enterprise distributions across MP index values in non-manufacturing two-digit NAICS industries relative to manufacturing two-digit NAICS industries.


Appendix D. Detailed Results

Table 4: Detailed Results of Equation (2)
Dep. var.: MPindex, weighted OLS regressions
(Characteristics)

Table 4: Detailed results of Equation (2) – Dep. var.: MPindex, weighted OLS regressions – (Characteristics)
VariableManufacturingNon-manufacturing
– Characteristics –
CONS0.4032Footnote a0.4067Footnote a0.2514Footnote a0.3127Footnote a
(0.0001)(0.0001)(0.0002)(0.0001)
MEDIUM0.1025Footnote a0.1001Footnote a0.1544Footnote a0.1539Footnote a
(0.0004)(0.0004)(0.0003)(0.0001)
LARGE0.1083Footnote a0.1060Footnote a0.1827Footnote a0.1855Footnote a
(0.0001)(0.0001)(0.0001)(0.0001)
XLARGE0.1873Footnote a0.1851Footnote a0.2378Footnote a0.2465Footnote a
(0.0001)(0.0001)(0.0001)(0.0001)
%UNIV0.0018Footnote a0.0019Footnote a0.0026Footnote a0.0027Footnote a
(0.0003)(0.0001)(0.0001)(0.0001)
HQ_US0.0586Footnote a0.0602Footnote a0.0943Footnote a0.0606
(0.0003)(0.0001)(0.0081)(0.1111)
HQ_EU0.0668Footnote a0.0671Footnote a−0.01210.02690
(0.0006)(0.0004)(0.8326)(0.6497)
HQ_ROW0.01990.01990.05430.1040
(0.4080)(0.389)(0.5673)(0.2006)
MULTI0.00830.0080−0.0994Footnote a−0.1031Footnote a
(0.5735)(0.5698)(0.0188)(0.0046)
SERV−0.0211−0.0215
(0.6316)(0.5758)
MANU_20.0489Footnote a0.0459Footnote a
(0.0004)(0.0006)
MANU_30.0579Footnote a0.0550Footnote a
(0.0004)(0.0006)
QC−0.0331Footnote b−0.0303Footnote b0.03730.0292
(0.0660)(0.0863)(0.3890)(0.4438)
AB−0.007−0.00580.02100.0368
(0.7708)(0.7975)(0.6766)(0.3742)
BC−0.0212−0.0156−0.0530−0.0556
(0.3229)(0.4534)(0.3054)(0.2627)
ROC−0.0101−0.0118−0.01090.0293
(0.6409)(0.5826)(0.7510)(0.3104)

P-values are between parenthesis.

Table 4: Detailed Results of Equation (2) – (continued)
– Dep. var.: MPindex, weighted OLS regressions –
(Competition variables)

Table 4: Detailed results of Equation (2) – Dep. var.: MPindex, weighted OLS regressions – (Competition variables)
VariableManufacturingNon-manufacturing
Competition variables
MNE0.0587Footnote c0.0597Footnote c0.1022Footnote c0.0995Footnote c
(0.0001)(0.0001)(0.0029)(0.0011)
#COMP0.0081Footnote c0.0086Footnote c0.00060.0062
(0.0135)(0.0070)(0.9490)(0.4647)
ENTRY0.01430.01460.03060.0261
(0.3395)(0.3139)(0.4013)(0.4101)
#GOODS0.00060.00080.01060.0134Footnote c
(0.8277)(0.7718)(0.1104)(0.0354)
LERNER0.00630.0998Footnote c
(0.8168)(0.0129)
N2,7302,8901,1291,337
R-SQUARE0.21280.43610.20860.3951

P-values are between parenthesis.

Table 5: Detailed Results of Equations (3) and (4)
– Dep. var.: salesFootnote 3, profitsFootnote 3 and INNO regressions – Characteristics

Table 5: Detailed results of Equations (3) and (4)
– Dep. var.: sales, profits and INNO regressions –
VariableManufacturingNon-manufacturing
salesprofitsINNOsalesprofitsINNO
Characteristics
CONS0.0751Footnote d0.0162Footnote d−0.6298Footnote d0.1673Footnote d0.0314Footnote d−1.2375Footnote d
(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0170)
cap0.3117Footnote d0.0507Footnote d0.0939Footnote d0.0950Footnote d
(0.0000)(0.0000)(0.0000)(0.0000)
M P i n d e x 0.0494Footnote d0.0232Footnote d2.1554Footnote d−0.0116−0.0011Footnote d0.7406
(0.0000)(0.0000)(0.0000)(0.1928)(0.7063)(0.2590)
RD0.10600.0782
(0.2240)(0.3830)
ADV_TECH0.2320Footnote d0.4232Footnote d
(0.0000)(0.0000)
MEDIUM0.0145Footnote d0.0031Footnote d−0.02180.0021−0.0125Footnote d−0.1451
(0.0000)(0.0009)(0.8420)(0.4898)(0.0000)(0.6820)
LARGE0.0138Footnote d−0.0042Footnote d−0.0629−0.0034−0.0149Footnote d1.1507Footnote d
(0.0011)(0.0110)(0.5680)(0.3348)(0.0000)−0.0040
XLARGE0.0159Footnote e−0.0120Footnote d−0.2490Footnote e−0.1177Footnote d−0.0270Footnote d0.8465Footnote d
(0.0765)(0.0004)(0.0640)(0.0000)(0.0000)(0.0440)
%UNIV(0.0000)0.0002Footnote d0.00540.0004Footnote d0.0003Footnote d−0.0035
(0.8533)(0.0000)(0.1050)(0.0000)(0.0000)(0.4450)
HQ_US0.0057Footnote d0.0059Footnote d−0.10720.1082Footnote d0.0692Footnote e0.5129
(0.0000)(0.0107)(0.4130)(0.0000)(0.0000)(0.2020)
HQ_EU0.0102Footnote d0.0105Footnote d−0.11170.02410.00760.0171
(0.0000)(0.0100)(0.5620)(0.8543)(0.8071)(0.9780)
HQ_ROW0.0216−0.0188Footnote d−0.3737−0.1444−0.01770.1221
(0.2594)(0.0172)(0.1590)(0.3149)(0.6231)(0.8360)
MULTI0.0118Footnote d0.0045Footnote d−0.1773Footnote e0.3147Footnote d0.0140Footnote d−0.6923Footnote d
(0.0027)(0.0026)(0.0920)(0.0000)(0.0000)(0.0200)
SERV−0.0824Footnote d−0.00010.5240
(0.0000)(0.9095)(0.1190)
MANU_20.0172Footnote d0.0043Footnote d0.2798Footnote d
(0.0000)(0.0019)(0.0130)
MANU_30.0232Footnote d0.0048Footnote d0.2188Footnote d
(0.0000)(0.0001)(0.0370)
QC−0.0187Footnote d−0.0063Footnote d−0.01110.0177Footnote d0.0029Footnote d0.3758
(0.0000)(0.0000)(0.9140)(0.0000)(0.0153)(0.2080)
AB0.0067Footnote e0.0068Footnote d−0.4310Footnote d−0.0554Footnote d0.0091Footnote d0.3256
(0.0772)(0.0000)(0.0020)(0.0000)(0.0000)(0.4550)
BC−0.0115Footnote d−0.0026Footnote d−0.0121−0.0411Footnote d−0.0045Footnote d−0.1816
(0.0005)(0.0430)(0.9290)(0.0000)(0.0299)(0.6540)
ROC−0.0174Footnote d−0.0073Footnote d−0.1189−0.0517Footnote d−0.0188Footnote d−0.3666
(0.0000)(0.0000)(0.3960)(0.0000)(0.0000)(0.3870)

N w = 0 : Number of observations for which the ROBUSTREG procedure weight is set to zero
N REG : Final number of observations for ROBUSTREG regressions.
P-values are between parenthesis.

Table 5: Detailed Results of Equations (3) and (4) – (continued)
Dep. var.: salesFootnote 4, profitsFootnote 4 and INNO regressions
(Competition variables)

Table 5: Detailed results of Equations (3) and (4)
– Dep. var.: sales, profits and INNO regressions –
(Competition variables)
VariableManufacturingNon-manufacturing
sales profits INNO sales profits INNO
– Competition variables –
MNE0.0118Footnote e0.0046Footnote e0.2123Footnote e−0.0033−0.0097Footnote e0.8614Footnote e
(0.0000)(0.0000)(0.0180)(0.2134)(0.0000)(0.002)
#COMP−0.0029Footnote e−0.0003−0.0425Footnote f0.0047Footnote e0.0042Footnote e0.0417
(0.0000)(0.2062)(0.0980)(0.0000)(0.0000)(0.5850)
ENTRY−0.00110.00000.3650Footnote e−0.0036−0.0018Footnote e0.9936Footnote e
(0.6232)(0.9951)(0.0000)(0.1473)(0.0399)(0.0020)
#GOODS0.00030.00010.0222−0.0029Footnote e−0.0018Footnote e−0.1174Footnote e
(0.4532)(0.6915)(0.2440)(0.0007)(0.0000)(0.0280)
NTOTAL 2,6882,7071,1041,208
Nw=0 15412310366
NREG 2,5342,5841,0011,1422,8901,337

Nw=0: Number of observations for which the ROBUSTREG procedure weight is set to zero
NREG: Final number of observations for ROBUSTREG regressions.
P-values are between parenthesis.

Table 6: Detailed Results of Equations (5)
– Dep. var.: ITECH, INON-TECH, IPD-MK and IPC-OG weighted Biprobit –
(Characteristics)

Table 6: Detailed results of Equations (5)
– Dep. var.:ITECH,INON-TECH,IPD-MKandIPC-OGweighted Biprobit –
VariableManufacturingNon-Manufacturing
 ITECHINON-TECHIPD-MKIPC-OGITECHINON-TECHIPD-MKIPC-OG
– Characteristics –
CONS0.6765Footnote g1.6133Footnote g0.9958Footnote g0.9958Footnote g1.2980Footnote g1.4362Footnote g1.8312Footnote g2.2768Footnote g
(0.0000)0.00.00.0(0.0190)(0.0060)(0.0010)0.0
MP index 1.7467Footnote g2.6589Footnote g0.9284Footnote g2.2058Footnote g0.54561.8365Footnote g1.1644Footnote g2.3178Footnote g
(0.0000)0.0(0.010)0.0(0.3590)(0.0010)(0.030)0.0
RD0.03180.00600.00420.01950.1112Footnote g0.01190.00850.0009
(0.4630)(0.1850)(0.6680)(0.2840)(0.0460)(0.6550)(0.5960)(0.970)
ADV_TECH0.2234Footnote g0.1646Footnote g0.1824Footnote g0.2070Footnote g0.3061Footnote g0.08860.10990.4023Footnote g
(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.2430)(0.180)(0.0000)
MEDIUM0.05380.05330.2077Footnote h0.04010.17920.24150.23860.0027
(0.5970)(0.5750)(0.0740)(0.6820)(0.5960)(0.4150)(0.4720)(0.9940)
LARGE0.01300.1907Footnote g0.03250.03400.00760.7307Footnote g0.5528Footnote h0.4802
(0.8960)(0.0410)(0.7280)(0.730)(0.9810)(0.0270)(0.0960)(0.1240)
XLARGE0.03690.2931Footnote g0.16400.15250.35930.55370.9272Footnote g0.5116
(0.7560)(0.0090)(0.1360)(0.1950)(0.3750)(0.1570)(0.010)(0.1720)
%UNIV0.00410.0048Footnote g0.0114Footnote g0.00010.00670.00390.00350.0047
(0.1550)(0.0450)(0.0020)(0.9730)(0.1020)(0.2970)(0.2920)(0.3120)
HQ_US0.3205Footnote g0.06630.01370.11710.03130.56030.00740.9015Footnote g
(0.0130)(0.5640)(0.9090)(0.3480)(0.9460)(0.1480)(0.9870)(0.0110)
HQ_EU0.20370.02740.09010.01140.03340.26860.14170.4095
(0.2500)(0.8690)(0.5790)(0.9480)(0.9330)(0.5260)(0.7450)(0.460)
HQ_ROW0.4840Footnote g0.4585Footnote g0.18360.4398Footnote g0.03810.14240.72320.2596
(0.0240)(0.0310)(0.420)(0.0430)(0.9370)(0.8430)(0.3220)(0.580)
MULTI0.15370.02050.03950.11110.42030.21840.03701.3776Footnote g
(0.1050)(0.8220)(0.6770)(0.240)(0.1390)(0.5030)(0.9060)0.0
SERV0.36770.7463Footnote g0.8700Footnote g0.1605
(0.2490)(0.0250)(0.0150)(0.5560)
MANU_20.1751Footnote h0.05760.03880.2439Footnote g
(0.0820)(0.5630)(0.6860)(0.0170)
MANU_30.1560Footnote h0.04460.00610.1191
(0.0990)(0.6330)(0.950)(0.2110)
QC0.08100.1486Footnote h0.1962Footnote g0.07950.16340.5942Footnote g0.5393Footnote h0.1118
(0.3840)(0.0960)(0.0440)(0.3850)(0.5670)(0.0460)(0.0690)(0.6590)
AB0.4756Footnote g0.08380.2624Footnote h0.2100.02060.22660.44680.4930
(0.0000)(0.4910)(0.0530)(0.1010)(0.960)(0.5130)(0.2290)(0.1710)
BC0.06270.18240.10300.04780.06860.03630.10130.3182
(0.6180)(0.110)(0.4090)(0.6940)(0.8560)(0.9290)(0.7970)(0.3360)
ROC0.09550.06650.06030.11950.36370.51070.09390.4863Footnote h
(0.4560)(0.5850)(0.6510)(0.3380)(0.3310)(0.1150)(0.8030)(0.0950)

P-values are between parenthesis.

Footnotes

Footnote g

Significant at a 0.05 level.

Return to footnote g referrer

Footnote h

Significant at a 0.10 level.

Return to footnote h referrer

Appendix D (continued)

Table 6: Detailed Results of Equations (5) – (continued)
– Dep. var.: ITECH, INON-TECH, IPD-MK and IPC-OG weighted Biprobit –
(Competition variables)

Table 6: Detailed results of Equations (5)
– Dep. var.:ITECH,INON-TECH,IPD-MKandIPC-OGweighted Biprobit –
VariableManufacturingNon-Manufacturing
 ITECHINON-TECHIPD-MKIPC-OGITECHINON-TECHIPD-MKIPC-OG
– Competition variables –
MNE0.08820.09200.1353Footnote j0.02571.0230Footnote i0.13180.04960.8919Footnote i
(0.2770)(0.2440)(0.0940)(0.7530)(0.0000)(0.5940)(0.8410)(0.0000)
#COMP0.0582Footnote i0.02070.03350.00190.03580.1326Footnote j0.02260.0435
(0.0170)(0.3720)(0.140)(0.9350)(0.6710)(0.0520)(0.7640)(0.4580)
ENTRY0.3330Footnote i0.3186Footnote i0.3157Footnote i0.1446Footnote j0.9473Footnote i1.1182Footnote i1.1408Footnote i0.3007
(0.0000)(0.0000)(0.0000)(0.0770)(0.0020)(0.0000)(0.0000)(0.2060)
#GOODS0.0336Footnote i0.02640.0391Footnote i0.01510.0986Footnote i0.04500.07450.0116
(0.0440)(0.1160)(0.0220)(0.3710)(0.0340)(0.3250)(0.1190)(0.7920)
N2,8902,8901,3371,337
0.4132Footnote i0.3988Footnote i0.4007Footnote i0.3190Footnote i

P-values are between parenthesis.

Table 7: Detailed Results of Equations (5)
– Dep. var.: IPRCS, IORGZ, IPRDT and IMRKT weighted multivariate Probit –
(Characteristics)

Table 7: Detailed results of Equations (5)
– Dep. var.:IPRCS,IORGZ,IPRDTandIMRKTweighted multivariate Probit –
VariableManufacturingNon-manufacturing
 PRDTORGZPRDTMRKTPRDTORGZPRDTMRKT
– Characteristics –
CONS-1.0364Footnote k-1.9299Footnote k−0.9625Footnote k-1.2442Footnote k-1.9216Footnote k-2.5825Footnote k-1.7676Footnote k-1.9739Footnote k
(0.0000)0.00000.00000.00000.00000.0000−0.00100.0000
MPindex1.8407Footnote k2.6694Footnote k0.6937Footnote k1.1656Footnote k0.8562Footnote l2.7276Footnote k0.74091.2384Footnote k
(0.0000)0.0000−0.01400.0000−0.09700.0000−0.1830−0.0110
RD0.0076−0.00100.01930.00060.0000−0.00350.01470.0208
(0.4520)−0.8490−0.1510−0.8930−0.9960−0.7050−0.2210−0.1490
ADVTECH−0.0217−0.0090−0.1835Footnote k−0.11280.3478Footnote k0.2003Footnote k0.1951Footnote k−0.0174
(0.8220)−0.9240−0.0740−0.27800.0000−0.0040−0.0120−0.7810
MEDIUM−0.0467−0.1560Footnote l0.0525−0.0719−0.26730.31360.5614Footnote l−0.0559
(0.6040)−0.0800−0.5570−0.4130−0.4260−0.2880−0.0800−0.8490
LARGE−0.0679−0.2733Footnote *−0.0893−0.2020Footnote l−0.05850.5997Footnote k0.00010.6742Footnote k
(0.5330)−0.0150−0.4120−0.0640−0.8630−0.0230-1.0000−0.0290
XLARGE0.2315Footnote k0.1500Footnote k0.1738Footnote k0.1514Footnote k0.27610.7148Footnote k0.7894Footnote k0.1917
(0.0000)0.00000.00000.0000−0.4460−0.0280−0.0250−0.6100
%UNIV−0.00300.00280.0102Footnote k0.0056Footnote k−0.00490.00440.0020−0.0012
(0.2100)−0.18900.0000−0.0300−0.3080−0.1900−0.5100−0.7140
HQUS−0.4647Footnote k−0.06320.0092−0.16930.54801.2791Footnote k0.4407−0.2119
(0.0000)−0.5930−0.9360−0.1490−0.23900.0000−0.3250−0.6120
HQEU−0.2752Footnote l0.14380.0396−0.1822−0.05190.2798−0.2781−0.0202
(0.0840)−0.3590−0.7910−0.2180−0.8900−0.4840−0.4680−0.9630
HQROW−0.3801Footnote l−0.4642Footnote k−0.3653Footnote l−0.11090.28650.0554−0.8006−0.2358
(0.0620)−0.0140−0.0850−0.6110−0.4280−0.9250−0.2680−0.7020
MULTI−0.10300.0290−0.06780.0700−0.6982Footnote k−0.9863Footnote k−0.12150.0374
(0.2470)−0.7480−0.4660−0.4670−0.00300.0000−0.6680−0.9000
SERV−0.06050.4740Footnote l0.8704Footnote k0.9717Footnote k
(0.8330)−0.0920−0.0130−0.0060
MANU20.13130.13570.0643−0.1028
(0.1700)−0.1720−0.5070−0.2680
MANU30.04880.09550.0185−0.1863Footnote k
(0.5920)−0.3120−0.8460−0.0440
QC0.06420.1648Footnote l0.2001Footnote k0.1048−0.4710Footnote l0.4666Footnote k0.32140.4472
(0.4620)−0.0550−0.0250−0.2420−0.0970−0.0500−0.2400−0.1000
AB−0.2618Footnote k−0.1483−0.3305Footnote k−0.10850.2341−0.0632−0.1259−0.4815
(0.0410)−0.2190−0.0090−0.4240−0.5460−0.8320−0.7250−0.1500
BC−0.06960.11500.01460.1665−0.37080.51670.39470.0269
(0.5410)−0.3050−0.9020−0.1470−0.2420−0.1350−0.2990−0.9480
ROC−0.05830.0304−0.0279−0.0951−0.4455−0.21130.2105−0.4242
(0.6400)−0.8030−0.8230−0.4550−0.1380−0.4650−0.5860−0.1700

P-values are between parenthesis.

Table 7: Detailed Results of Equations (5) – (continued)
– Dep. var.: IPRCS, IORGZ, IPRDT and IMRKT weighted multivariate Probit –
(Competition variables)

Table 7: Detailed results of Equations (5)
– Dep. var.:IPRCS,IORGZ,IPRDTandIMRKTweighted multivariate Probit –
VariableManufacturingNon-manufacturing
 PRDTORGZPRDTMRKTPRDTORGZPRDTMRKT
– Competition variables –
MNE−0.05590.03880.1735Footnote m−0.01410.7933Footnote m0.2840.5025Footnote m0.1515
−0.478−0.613−0.027−0.86−0.001−0.145−0.029−0.514
#COMP−0.00580.032−0.0506Footnote m−0.00580.1699Footnote m−0.1275Footnote m−0.1682Footnote m−0.0552
−0.796−0.158−0.022−0.794−0.027−0.016−0.025−0.443
ENTRY0.2032Footnote m0.2623Footnote m0.3076Footnote m0.2975Footnote m−0.160.7168Footnote m1.3516Footnote m1.1618Footnote m
−0.008−0.00100−0.499000
#GOODS0.0140.010.0503Footnote m0.0366Footnote m−0.0731Footnote **0.0125−0.0553−0.0367
−0.367−0.545−0.002−0.023−0.077−0.75−0.222−0.404
N2,8901,337
120.4217Footnote m0.2087Footnote n
130.3842Footnote m0.2726Footnote m
140.2368Footnote m−0.0518
230.2547Footnote m0.5523Footnote m
240.3565Footnote m0.5062Footnote m
340.4040Footnote m0.6603Footnote m

P-values are between parenthesis.