THE EFFECT OF MANAGEMENT PRACTICES ON INNOVATION

(1)

THE EFFECT OF MANAGEMENT

PRACTICES ON INNOVATION

E

VIDENCE FROM TRANSITION ECONOMIES

MSc Thesis

June 2016

Author: Fanni Gaál

University of Groningen

Faculty of Economics and Business

Student number: S2997940

Email:

f.e.gaal@student.rug.nl

Supervisor: M.S.S. Krammer,

University of Groningen

(2)

1

ABSTRACT

This thesis aims to examine the effects of certain management practices on innovation. In order to do so, I scan the literature and identify job autonomy, performance feedback and performance-based incentives (pay and promotion) as potentially important factors in promoting innovation. In the analysis I use a sample of 1266 manufacturing firms from transition economies of Central and Eastern Europe and Central Asia, from the Management, Organisation and Innovation (MOI) Survey, conducted by the EBRD in 2008-2009. The cross-country analysis confirms the positive and significant effect of the above practices on innovation, with a similar impact achieved when applied together.

Keywords

(3)

2

1. INTRODUCTION

Innovation is an important driver of competitiveness at the firm level, and as such has been in the focus of strategic management decisions and related academic research (Volberda, Van Den Bosch and Heij, 2013). There is extensive literature theorizing and analyzing the types and scope of innovation, and there is growing interest surrounding the organizational factors stimulating innovative activities within firms. What are the conditions for encouraging innovation? Is there a way to secure innovative performance through adopting certain practices? Which organizational structures and practices can foster innovation, and which are the ones that might hold it back?

The possible answers to such questions are manifold, but theories and research conducted so far have provided some viable approaches. One branch of these studies has its roots in creativity literature. Although technological advancement is as fast as ever, it is the creative minds of humans that need to make the best use of new technologies. Behavioral scientists have built up extensive theoretical background, in many cases also covered by analyses to explain the factors influencing creativity at the level of the individual (e.g. Shalley, Zhou and Oldham, 2004; Wang & Cheng, 2010; Sears & Baba, 2011). As also noted, in order to let the creative ideas flow, certain organizational practices must be in place (Amabile 1988). Human resource management practices aim to find a middle ground between employee and organizational goals, which work best together when aligned to avoid conflict of interest (Jensen & Meckling, 1976). To achieve this, it takes careful organizing and management support for an effective, cooperative and innovation-fostering culture to bloom. Research so far has covered various areas within the field of human resource management (HRM), examining the potential practices that induce innovation activities within the firm. The results are ranging on a wide scale, often bearing contradictions and reflecting differences in approaches. Incentive (or performance-based) pay, for example, has been found to have a strong positive impact on innovation in some cases (e.g. Bloom & Van Reenen, 2011; Barros & Lazzarini, 2012), while in other studies it proved to be insignificant (e.g. Zoghi, Mohr and Meyer, 2010). Such uncertainties show that research in this topic is far from complete – new variables, new mediating factors keep emerging as potential elements of a more integrated solution.

(5)

4

innovation (e.g. Lee & Kelley, 2008; Barros & Lazzarini, 2012). To the best of my knowledge, no studies have focused on the set of transition economies so far in this specific relationship of management practices and innovation. The regime changes of the 1990s left the transition economies of Central and Eastern Europe and Central Asia with unique economic structures. Different types of transitions brought about different outcomes, and even early studies discuss the possible impacts of the process on the management practices of firms operating in these economies (Frydman & Rapaczynski, 1997). Since those years of early transition, many external factors have shaped these economies through foreign ownership, FDI, the expansion of multinational enterprises, and most recently, (prospective) membership in regional economic formations like the European Union. These characteristics allow us to treat these countries as a unique group with distinctive paths of development.

The goal of this analysis is to contribute to the literature in two ways: on the one hand, by unfolding the effects of different management practices on innovation in the special context of transition economies. I use the Management, Organisation and Innovation survey dataset gathered by the EBRD in partnership with the World Bank, covering 10 transition economies, and also Germany and India. On the other hand, it gives a fresh perspective to the ongoing hunt for innovation-fostering HRM practices. Based on the literature, I select four management practices (job autonomy, performance feedback, performance-based promotion and performance-based pay) and analyze their impact on innovation. The results prove their positive and significant effect on innovation, which is in line with the literature regarding the first two practices, confirming a positive relationship. The findings for the performance-based incentives, pay and promotion also give support to their positive impact on innovation, thus eliminating doubts originated in previous analyses with ambiguous results.

(6)

5

2. LITERATURE REVIEW

In order to understand the relationship between management practices and innovation in the context of transition economies, I first examine the literature about the two topics separately, and also those covering the impact of management practices on innovation. Both fields are highly studied areas on their own, and it is also possible to look at some related empirical analyses regarding transition economies. This section highlights some of the most important theories and findings related to these fields, starting from innovation, then dealing with human resource management practices as possible means of influencing innovation activities, then putting it all in the context of transition economies.

2.1 Innovation

In today’s constantly changing business environment and growing competition, innovation has become an important indicator and a potential driver of firm success, giving rise to complex economic, strategic, social, organizational and technological value (Schumpeter, 1934; Cohen & Levinthal, 1990; Sears & Baba, 2011). One widely used definition of innovation is given by the Oslo Manual, saying that innovation is “the implementation of a new or significantly improved product (good or service), or process, a new marketing method, or a new organizational method in business practices, workplace organization or external relations” (OECD, 2005:46). As the Manual also indicates, innovation may come in many forms – it ranges from product innovation through process innovation to marketing or organizational innovation, affecting the work of small units, a whole industry or a chain of activities across industries.

(7)

6

Having accepted the inevitable need for innovation on the road towards success, financial or otherwise, arises the question: what are the factors that facilitate innovative activities? One indisputable answer to this question is human creativity. Previous research on organizational behavior yielded several theories and observations on creativity as a means to innovation. On the individual’s level, creativity can be considered the starting point of the innovation process (Sears & Baba, 2011). The literature on attracting and managing creative people provides evidence for the positive effect of many practices establishing the fit between the individual and the environment, such as job complexity, job autonomy, trainings, certain leadership styles or information-sharing patterns (see e.g. Amabile, 1988; Mumford, 2000; Shalley, Zhou and Oldham, 2004; Wang & Cheng, 2010). Among many others, these practices form the basis of the first stage of innovation, which is the generation of a creative idea (Shipton et al., 2006). To do so, employees need certain skills and expertise, but also working conditions that enable them to exploit their creative potential. Once such individuals are employed after selective hiring procedures, proper induction and trainings, new ideas can flourish if the organization has an established risk-taking culture (Barros & Lazzarini, 2012; Engel et al., 2015) – this is referred to as exploratory learning by Shipton et al. (2005).

As the individual learns and solves problems, the organization also needs to learn and adapt. Out-of-the-box thinking needs encouragement by the management, which leads to the next point in discussing the way towards innovation.

2.2 Human resource management practices

It has become clear that creativity cannot be the single most important determinant of innovation. As much as innovation depends on the performance and creativity of the individuals, it also has important organization-level components (Amabile, 1988). Firms, however, show a great heterogeneity, some of them are more successful than others, and this may be reflected by these organizational components as well.

(8)

7

Reenen, 2011). Others examine the impact of certain HRM practices on productivity through mediators, like innovation (Zhou, Hong and Liu, 2013). One comprehensive study of HRM practices as predictors of productivity was proposed by Bloom and Van Reenen (2011), who concluded that besides incentive pay, performance feedback, teamwork and similar direct measurements of HRM, the complementarities between these practices also have a positive impact on productivity.

Moving towards more details about HRM practices and their relationship with innovation performance, a review of theories building and reflecting on such practices can be of guidance. Below I will introduce four of these theories.

First I discuss the resource-based view, according to which firms need to analyze their internal strengths and weaknesses, and can only gain competitive advantage if they own strategic resources that are valuable, rare, imperfectly imitable and have no strategically equivalent substitutes (Barney, 1991). The resource-based view thus defines efficiency as the most important source of competitive advantage. The resource-based view has been extended to HRM practices arguing that strategic HRM practices can enhance creativity, and hence increase innovation activities that result in solutions unique to the firm (Beugelsdijk, 2008). Theory suggests that the design of work structures and job autonomy, flexible work schedules, training and performance-based pay, especially when combined, are expected to support the integration of knowledge, and thus creativity and innovation within the organization (Amabile et al., 1996; Bandiera, Barankay and Rasul, 2005; Kang, Morris and Snell, 2007; Wang & Cheng, 2010). However, empirical findings so far have only found sound support for some of these practices, like task autonomy, training or performance-based pay, which support incremental innovation in Beugelsdijk’s study of Dutch firms (2008), while he found that radical innovation is more likely to be enhanced by task autonomy and flexible working hours.

(9)

8

the field of HRM that relational support role of managers such as facilitating communication and giving autonomy for members, as well as the decision support role of managers such as monitoring and motivation are of vital importance for innovation. Zollo and Winter (2002) use the dynamic capabilities approach in the context of learning mechanisms, arguing that well-defined incentives and performance monitoring can encourage deliberate learning within the organization. Shipton et al. (2005) uncover that sophisticated HRM practices supporting exploration and risk taking, like formal appraisal schemes, trainings, recruitment and selection practices and induction, all facilitate product and production technology innovation. To the contrary, Santiago (2013) finds no empirical support for a significant relationship between HRM practices and innovation in Mexican pharmaceuticals – all the variables on training, remunerations, worker’s empowerment and rules of staff hiring turned out to be statistically insignificant.

Third, another broadly applied approach is the agency theory, which also has many implications to HRM practices. According to the agency theory, the separation of the ownership and the management of firms may lead to conflicts of interest between the two parties, which need to be tackled by introducing management incentives (Jensen & Meckling, 1976). Following this line of reasoning, if agents (here employees) receive a risk premium, they will be more inclined to engage in innovative activities (Barros & Lazzarini, 2012). As a result, certain practices have been found to foster innovation, like concentrated ownership, based pay, performance-based promotion, or decentralization (see e.g. Huselid, 1995; Bloom, Sadun and Van Reenen, 2010b; Barros & Lazzarini, 2012). Choi, Lee and Williams (2011) also give empirical evidence in Chinese firms that business group affiliation has a positive effect on innovation performance, although they could not find a statistically strong effect of ownership concentration on reduced agency costs. In an empirical study of Canadian workplace organizations, Zoghi, Mohr and Meyer (2010) proved that decentralized decision making and information sharing procedures were positively correlated with innovation.

(10)

9

contemporary changes in the organization of the employment relation” (Laursen & Foss, 2003:248). This means that certain practices are believed to work better than others, and even better when clustered together, since they tend to be highly correlated (Ichniowksi, 1990). The universalistic approach has probably been most common among researchers, collecting the best practices to be applied by everyone aiming for success: formal training systems, result-oriented appraisal, performance-based compensation, employment security, internal career ladder, etc. (Delery & Doty, 1996). Some empirical analyses provided evidence for the existence of such high performance work practices (Ichniowski, 1990; Huselid, 1995; Cappelli & Neumark, 2001). However, the findings of other empirical studies showed that these are not related to superior performance in all scenarios, so it is possible that there is an infinite number of “best practices”, depending on the individual firms and their environment (Dyer & Reeves, 1995). Regarding the impact of these best practices on innovation, Laursen & Foss (2003) observe that vertical as well as horizontal linkages are of great importance in assisting innovation.

In an attempt to synthesize the insights from the above described theories (and some more), Zhou, Hong and Liu (2013) clustered the HRM practices into two categories: commitment-oriented HRM (e.g. performance-based pay, selective hiring, team work, job enrichment, job rotation) and collaboration-oriented HRM (alliances, partnerships with professionals, businesses and academic institutions). They found positive impacts of both clusters on innovation and firm performance, although since they seem to compete for the same resources, the interaction between the two systems carried a negative impact. This study will also build on a more synthesized approach, using a selection of the HRM practices listed above in order to examine their effects on innovation.

2.3 The context of transition economies

(11)

10

As already mentioned, the context of transition economies provides an opportunity to examine the relationship of management practices and innovation performance of firms under very specific circumstances. Bloom, Van Reenen and colleagues, who ran several empirical analyses on the relationship between management practices and productivity (see e.g. Bloom & Van Reenen, 2007; Bloom & Van Reenen, 2011), also use a sample of Central and Eastern European and Central Asian transition economies in a recent study to examine the quality of HRM practices, but they do not reflect on their impact on innovation (Bloom, Schweiger and Van Reenen, 2012).

In summary, both innovation and HRM literatures have diverse and complex ideas about the factors that foster innovation at the firm level. These ideas have evolved over time, and capture different aspects and theories. As opposed to previous studies covering mostly developed countries (e.g. Bloom & Van Reenen, 2007; Beugelsdijk, 2008; Zoghi, Mohr and Meyer, 2010), analyzing the impact of management practices on innovation in transition economies is definitely a worthwhile exercise, given the special development path of these countries.

3. HYPOTHESES DEVELOPMENT

Based on the review of the literature, innovation seems to be an important factor that is able to boost and sustain high performance within firms. The role of management practices, however, is less clear in promoting innovation. There is no consensus between theories on exactly which types of HRM practices can best induce innovative activities, but it is possible to point out some of the usual suspects.

(12)

11

Another positive aspect of job autonomy is decentralization and flattened organizational hierarchy, which is included in many forms of the new HRM systems. The more autonomy the workers gain, the more control they have over the decisions regarding tasks and working conditions (Bloom & Van Reenen, 2011). Determining the degree of decentralization, thus the hierarchy of the firm carries an important decision, because giving up authority at the management level means that “problem-solving rights are delegated to the shopfloor” (Laursen & Foss, 2003:248). If employees enjoy managerial support in finding their own solutions to problems, it will lead to employees collecting, organizing and most importantly using more and more information and knowledge from different parts of the firm (Laursen & Foss, 2003; Lee & Kelley, 2008). Ultimately, this process fosters not only effectiveness, but also innovation along the way (Zoghi, Mohr and Meyer, 2010). So the first hypothesis is the following:

Hypothesis 1: Firms promoting and granting a higher level of job autonomy to workers are more likely to engage in innovation activities.

The positive effect of performance feedback (or appraisal) on innovation is also supported by the resource-based view and appears among the new HRM practices as well. In order to not lose sight of the target, feedback gives employees a hint on two important things: first, how their goals are aligned with the organizational objectives (as outlined in the agency theory), and second, it might shed light on some discrepancies between the applied practice and the most efficient practice available (Shipton et al., 2006). Also, it is important for employees to get feedback about their achievements from time to time, in order to encourage the link between better firm performance and more content employees. As a condition though, these feedbacks cannot be judgmental – they need to be informative, developmental and result-oriented in order to stimulate better performance and to spur creativity (Delery & Doty, 1996; Mumford, 2000).

(13)

12

Hypothesis 2: Organizations with managers giving performance feedback to employees frequently are more likely to engage in innovation activities.

The next HRM practice to be seemingly connected to innovation concerns the reward policy of the firm. Promoting employees based on merit and achievements is a direct way of rewarding above-average performance. According to Huselid (1995), promotion systems that reward workers based on merit do not only decrease employee turnover rate, but they also increase financial performance of the firm. Barros and Lazzarini (2012) find that performance-based promotion, being an intrinsic motivation, has significant effects in fostering innovation.

Performance-based pay is another way to incentivize work by making the progress more result-oriented. Unfortunately, research so far has yielded ambiguous findings about its effects on innovation. On the one hand, evidence from previous analyses suggests that performance-based pay directly motivates employees to focus on the target, and the recognition of their accomplishments results in a positive relationship between performance-based pay and firm innovation (Bandiera, Barankay and Rasul, 2005; Inderst, 2009). In one of their studies, Bloom and Van Reenen (2011) conclude that the importance of individual incentive pay lies in its selection effect: the least productive workers quit, and only the most productive ones stay, so basic behavioral patterns of those who want to join and those who want to quit, will change. Barros and Lazzarini (2012) confirm the positive relationship between performance-related pay and innovation (though only marginal, like in case of Laursen & Foss, 2003). Inderst (2009) conducted research on the conflict of interest potentially arising from employee compensation, proving that incentive pay linked to the performance of the innovation ensures that only those ideas will become elaborate new innovations that are also able to increase efficiency and profitability.

Note that in some of these cases, the positive effect is only present under special circumstances. In his study, Beugelsdijk (2008) observes that performance-based pay enhances incremental innovation, but not radical, and he also points out how size shows a negative moderating effect on performance-based pay, implying that the sound assessment of individual performance is probably a pre-condition for safe and fair application of this practice.

(14)

13

most crucial problem with performance-based pay is that it may undermine intrinsic motives and push people towards extrinsic motivation and well-tried methods instead of innovative approaches (Amabile, 1988; Shipton et al., 2006). Shalley, Zhou and Oldham (2004) point out that applying extrinsic rewards may hamper creativity of people in complex jobs, since their performance can hardly be measured appropriately. Another pitfall of performance-based pay is its effect on group performance: if pay is based on individual achievements, but work is group-based, this artificially generated competition may severely damage innovation and general performance of the group (Bandiera, Barankay and Rasul, 2005). But if incentive is group-based, the problem of free riders emerge, which again will not result in proper distribution of rewards (Bloom & Van Reenen, 2011).

In any case, performance-based pay and promotion, individual or group-based, clearly are important components in the HRM and innovation literature. Since it also plays a center role in the agency theory as well as in the new HRM systems, both suggesting a positive relationship, it will be included in the analysis. So the third hypothesis is the following:

Hypothesis 3: Firms rewarding employees with promotion or pay based on merit and performance are more likely to engage in innovation activities.

Finally, it seems to be of vital importance to take into consideration the fact that HRM practices are often correlated and thus have a larger impact when clustered than individually (Ichniowski, 1990; Dyer & Reeves, 1995; Laursen & Foss, 2003; Shipton et al., 2005; Bloom & Van Reenen, 2011; Zhou, Hong and Liu, 2013). Some of the studies include interaction variables for the HRM practices that are likely to have increased effects together, like applying profit sharing (as an extrinsic motivation) and autonomous teams together have been found to positively affect productivity (Cappelli & Neumark, 2001). However, mostly these simple interactions turn out to be statistically insignificant. An example is the interaction of decentralization and individual incentive pay: if workers are given more autonomy, individual incentive pay is expected to be more efficient since effort and results should be easier to determine and measure for the individual, but evidence does not support a significant relationship (Zoghi, Mohr and Meyer, 2010).

(15)

14

cluster affecting productivity positively. Dyer and Reeves (1995) administered similar results regarding firm performance when contrasting a bundle of new HRM practices against a traditional and a mixed system. Laursen & Foss (2003) argue that systems of HRM practices exhibit complementarities that make imitation hard for the competition, and such systems of vertical linkages and knowledge linkages between HRM practices are important for innovation. Shipton and colleagues (2005) find that effective HRM systems with sophisticated recruitment, training, appraisal and mentoring schemes play an important role in providing a supporting climate for the creation and transfer of knowledge, enhancing innovation in UK manufacturing firms.

Based on evidence from previous research in the HRM and innovation literature, it is suspected that the management practices included in this analysis also work better when clustered than individually, with a significant positive effect on innovation performance. The final, fourth hypothesis is the following:

Hypothesis 4: When applied together, clustered HRM practices have a significant positive impact on innovation activities of firms.

4. DATA AND VARIABLES

4.1 Database and sample

For the empirical analysis I use the EBRD–World Bank Management, Organisation and

Innovation (MOI) Survey, conducted by the European Bank for Reconstruction and Development

(16)

15

factory managers who are familiar with day-to-day work at the shop floor, and are also able to answer detailed questions about the management practices of the establishment.

Besides the Central and Eastern European countries (CEECs), the survey also covers Kazakhstan and Uzbekistan, two large Central Asian transition economies that also became independent of the (then dissolved) Soviet Union in 1991. Both are significant economies of the region and followed a path similar to CEECs towards democratization and switching to a market economy in the 1990s. Germany serves as a developed country benchmark. India was added to the analysis as a developing country benchmark, because although not under the USSR, a similar command regime was in place from its independence in 1947 until the 1980s and 1990s, when deregulation started (Bloom, Schweiger and Van Reenen, 2012). After cleaning the database from missing values, I have answers from 1266 establishments from 12 countries for the analysis. Tables II-III in the Appendix give a more detailed description of the sample.

4.2 Dependent variable

To analyze the four hypotheses, an R&D dummy determining whether the establishment had or had not spent on R&D in the fiscal year previous to the year of the interview, is used as the dependent variable for several reasons. R&D spending is a widely-used indicator of innovation (e.g. Wakelin, 1998; Santiago, 2013). The willingness of engaging in R&D marks a firm’s intentions regarding the future directions in which it hopes to find more space for development and profitability. As a qualitative survey, there were only a few questions in the MOI Survey targeting financials, and the answering rate for these was very low – only 23% of the respondents reported the actual amount of R&D spending. Having spent on R&D at all as a dummy variable turned out to be a powerful distinguisher in this dataset of transition economies, as the results will show.

4.3 Explanatory variables

(17)

16

variable with values between 1 and 5, measuring who sets the pace of work at the establishment. It uses the same scale taskallocation: it takes the value of 1 if only factory managers make the decision, and 5 if it is the decision of the workers only (3 being when the two parties participate equally in the decision-making). I included both of these variables for two reasons, first, the presence of either represents delegation of decision-making to the workers’ level regarding their own work, and second, based on the theory of new HRM practices there is reason to suspect that the more such practices are implemented, the better the firm’s innovation performance.

Hypothesis 2 establishes a connection between innovation and performance feedback. In

order to see how frequently firms collect production performance indicators at the establishment,

freq_ind is introduced to the analysis. Its value varies from 1 to 6, 1 indicating a yearly, 2 a

quarterly, 3 a monthly, 4 a weekly, 5 a daily and 6 an hourly collection of production performance indicators. For feedback, I use the variable perftoworkers, which measures the frequency of actually showing the collected production performance indicators to workers, on a scale of 0 (never) to 7 (hourly).

(18)

17

that the original scaling needed adjustments in some questions (see Table I of the Appendix with the original questions and answers from the MOI Survey).

Based on the MOI Survey, 2008

Table 1: Dependent and explanatory variables

Name Meaning Scale

Dependent variable

RD_d Presence of R&D spending in the last fiscal year dummy, 0–1

Explanatory variables

taskallocation

Allocation of tasks determined by:

1–5 1 – Only factory managers

2 – Mostly factory managers

3 – Factory managers and workers equally 4 – Mostly workers

5 – Only workers

paceofwork

Pace of work determined by:

1–5 1 – Only factory managers

2 – Mostly factory managers

3 – Factory managers and workers equally 4 – Mostly workers

5 – Only workers

freq_ind

Frequency of performance indicator collection:

1–6 1 – Yearly 2 – Quarterly 3 – Monthly 4 – Weekly 5 – Daily 6 - Hourly perftoworkers

Frequency of showing performance indicators to workers:

0–7 0 – Never 1 – Annually 2 – Half-annually 3 – Quarterly 4 – Monthly 5 – Weekly 6 – Daily 7 – Hourly

reward_d Reward for achievement of production target dummy, 0–1 allstaffrewarded_d All staff is rewarded for achieving production target dummy, 0–1

promotion

Employees are promoted based on:

1–3 1 – Factors other than individual’s effort and ability, such as

tenure.

2 – Partly on individual’s effort and ability, partly on other factors, such as tenure.

3 – Solely on individual’s effort and ability.

underperf

How underperformers are dealt with:

1–3 1 – Rarely or never moved from their positions

2 – Usually stay in position for at least a year before action is taken

(19)

18

Finally, since Hypothesis 4 is based on the assumption that the more “good” HRM practices, the better the innovation performance of the firm, I will test the joint effect of the above indicators by introducing them in the same regression.

4.4 Control variables

In order to control for some other effects on R&D not captured by the above listed management practices, I add a set of control variables to the analysis that have been found or are suspected to affect innovation activities of a firm by previous research. Larger firms, for example, may naturally have more resources to allocate for R&D, so size is measured here by the number of permanent, full-time employees at the establishment, log transformed to have a normal distribution (ln_size). Similarly, older firms might have had more chance to accumulate such resources, and since firm age also varies largely within the sample, I use ln_age to control for the age of the establishment. Regarding ownership, foreign ownership (foreign_own) is included in the analyses as a dummy variable indicating whether the largest owner of the firm is a foreign individual/firm/state, while state ownership (state_own) is another dummy variable to show if it is a state-owned enterprise (at the time of the interview). Both have been found to have significant impact on both financial and innovation performance (e.g. Bloom & Van Reenen, 2011; Zhou, Hong and Liu, 2013). Involvement in international trade and the level of competition that the firm’s main product faces can also induce innovation, so the following variables will also be used: export_d, measuring whether the main product is mainly sold outside the country, and competition, measuring on a scale of 0-3 the number of competitors the establishment’s main product has on its main market (0 for none, 1 for one, 2 for two to five, and 3 for more than five competitors). Finally, there is reason to assume that establishments that are part of a larger firm have more access to additional funding for research and development, so I also introduce a control variable for these firms, under the name of

multiest, referring to being a member of a multi-establishment firm.

(20)

19

5. METHODOLOGY

5.1 Model specifications

All the hypotheses use RD_d, a dummy variable taking the values of 0 or 1 as a dependent variable, so I use probit models in order to estimate the effects of the explanatory variables on R&D spending.

Hypothesis 1 examines the relationship between job autonomy and innovation, hypothesis 2 the effects of giving performance feedback to workers, while hypothesis 3 the impact of

performance-based pay on innovation. Translating the effects of the examined management practices on innovation into an equation, the output is the following:

𝑅𝐷_𝑑_𝑖 = 𝛽₀+ 𝛽₁𝑡𝑎𝑠𝑘𝑎𝑙𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛_𝑖+ 𝛽₂𝑝𝑎𝑐𝑒𝑜𝑓𝑤𝑜𝑟𝑘_𝑖+ 𝛽₃𝑓𝑟𝑒𝑞_𝑖𝑛𝑑_𝑖+ 𝛽₄𝑝𝑒𝑟𝑓𝑡𝑜𝑤𝑜𝑟𝑘𝑒𝑟𝑠_𝑖 + 𝛽₅𝑟𝑒𝑤𝑎𝑟𝑑_𝑑_𝑖+ 𝛽₆𝑎𝑙𝑙𝑠𝑡𝑎𝑓𝑓𝑟𝑒𝑤𝑎𝑟𝑑𝑒𝑑_𝑑 + 𝛽₇𝑝𝑟𝑜𝑚𝑜𝑡𝑖𝑜𝑛_𝑖+ 𝛽₈𝑢𝑛𝑑𝑒𝑟𝑝𝑒𝑟𝑓_𝑖 + 𝜆_𝑛Σ_𝑛=17 𝑐𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 + 𝜎_𝑘Σ23_𝑘=1𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦 + 𝜑_𝑙Σ_𝑙=112 𝑐𝑜𝑢𝑛𝑡𝑟𝑦 + 𝜀_𝑖,

where the number of observations is i=1, …, 1266. The coefficient β1 shows the effect of who

allocates tasks to workers. It is expected to be have a positive sign, with firms allowing workers to participate more in the decision to be more likely to spend on R&D. Similarly, β2 gives the effect

of whether workers are allowed to set the pace of work and to what extent. It is also expected to have a positive effect on RD_d, with workers participating more actively in the decision increasing the probability of spending on R&D. β3 is the coefficient for measuring the effect of how often

production performance indicators are collected at the establishment, with the expected sign being positive, so the more frequent the collection, the more likely it is for the firm to spend on R&D. The next coefficient is β4 with a similar meaning, only it is connected to the frequency of showing

production performance indicators to workers: the more often they are provided feedback, the more likely it is for the firm to engage in R&D. The next two variables are related to performance-based pay, with the coefficients β5 and β6 measuring the effects of whether there is a reward at all, and if

yes, whether it is divided among all the staff, with the signs expected to be positive. The coefficient β7 measures performance-based promotion, and is expected to have a positive impact with

promotions granted based more on individual achievements increasing the possibility of engaging in R&D. Finally, β8 shows the effect of how underperformers are dealt with, expected to appear

(21)

20

5.2 Principle Component Analysis

This leaves us with a lot of explanatory variables to be included in the analysis. Since there is reason to believe that these eight variables selected to measure the management practices (job autonomy, performance feedback and performance-based pay and promotion) are closely related, they may measure similar things.

In an attempt to decrease the number of variables in the equation, I ran a principle component analysis (PCA) to see if the variables load up to a fewer number of factors. This method is used specifically when there is need for data reduction. To do so, PCA measures the variance of the variables and extracts factors based on the amount of common variance they can account for (Abdi & Williams, 2010). This method is not new to the topic of management practices: Beugelsdijk (2008) also performed a principal component analysis on 12 human resources items in his analysis of Dutch firms, which all loaded up to 4 factors that he used in his analysis. Laursen and Foss (2003) analyzed the effect of 9 HRM practices on innovation in Danish firms, and in their case PCA rearranged their variables into two factors that they consequently implemented in their analysis.

The factor loadings of the PCA performed in Stata are displayed in Table 2. Since the test is run on the eight explanatory variables, PCA gives eight factors (components) as an outcome, but only the first four are included in the table, because these are the ones with an eigenvalue larger than 1. It means that these are the components that explain at least as much variance as the variables

Table 2: Principal component analysis, rotated (orthogonal varimax)

Factor1 Factor 2 Factor 3 Factor 4

Performance-based pay Job autonomy

Performance feedback

Promotions policy

Variable (Eigenvalue 1.76) (Eigenvalue 1.41) (Eigenvalue 1.40) (Eigenvalue 1.07)

taskallocation 0.0165 0.6999 -0.0232 -0.0366 paceofwork -0.0261 0.7043 0.0142 0.0267 freq_ind -0.0593 -0.0744 0.6980 -0.0033 perftoworkers 0.0463 0.0669 0.6897 -0.0349 reward_d 0.7019 -0.0127 -0.0143 -0.0131 allstaffrewarded_d 0.7002 0.0027 0.0011 -0.0032 promotion -0.0503 -0.0304 -0.0946 0.7934 underperf 0.0891 0.0543 0.1651 0.6058

(22)

21

do. An orthogonal varimax rotation was also applied, which does not change the content of the variables, but seem to fit the dataset better than the unrotated version and also gives more pronounced results, as all the factor loadings within each factor became either large or small. Numbers in bold indicate the largest factor loadings within the factors.

According to the results, reward_d and allstaffrewarded_d, two dummies measuring financial rewards are explained by a common underlying component. I named this component “Performance-based pay”, and will use it to refer to Factor 1 in the analysis. There are two variables that are highly correlated with Factor 2, taskallocation and paceofwork, so I will refer to this component as “Job autonomy”, since these were originally the variables selected to measure this HRM practice. The variables freq_ind and perftoworkers are both highly correlated with Factor 3, so in the analysis this component will be referred to as “Performance feedback”. Finally, the last two variables, promotion and underperf are highly correlated with Factor 4, measuring how above-average performance and underperformance are dealt with in terms of promotion and demotion, so I will refer to it as “Promotions policy”.

6. ANALYSIS AND RESULTS

6.1 Descriptive statistics

Summarized in Table 3 are the descriptive statistics of the sample. As already mentioned, after cleaning the database from missing answers (“don’t know” or refusal) and observations with missing values for the relevant variables, I have a total of 1266 observations.

The mean values are most interesting for variables that are not binary, with several values to take. The average firm size is the average number of employees, which is 274 in the sample, with a relatively high standard deviation of 428.62, hence the logarithmic transformation of this variable before using it in the probit model. The average firm age is 37 years (note that this refers to the age of the specific establishment participating in the survey). From the very low point for

taskallocation we can see that in most firms the participation of workers in allocating the tasks is

(23)

22 Table 3: Mean, standard deviation and correlations (i=1266)

Variable Mean SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 1. RD_d .43 .49 1.00 2. firmsize 274.38 428.62 .15* 1.00 3. firmage 36.98 37.23 .08* .20* 1.00 4. multiest .30 .46 .09* .14* .04 1.00 5. foreign_own .13 .34 .01 .08* -.05 .14* 1.00 6. state_own .08 .27 -.01 .12* .11* -.05 -.10* 1.00 7. export_d .25 .43 .09* .14* .08* .02 .18* -.07* 1.00 8. competition 2.48 .72 .04 -.08* -.05 -.02 -.02 -.04 .03 1.00 9. taskallocation 1.62 .65 .05 .01 -.03 .03 .00 -.01 -.04 -.03 1.00 10. paceofwork 1.78 .79 .07* .04 .07* .06* .03 -.02 -.02 -.05 .40* 1.00 11. freq_ind 3.90 1.11 .04 .06* .02 .10* .05 -.05 .07* .05 -.05 -.05 1.00 12. perftoworkers 3.84 1.94 .07* .06* -.05 .16* .06 -.04 .04 .01 .01 .04 .38* 1.00 13. reward_d .87 .34 .08* -.03 -.09* .06* -.05 -.04 -.01 .03 .02 -.01 .06* .13* 1.00 14. allstaffrewarded_d .78 .41 .05 -.04 -.14* -.03 -.08* -.01 -.03 .03 .02 .00 .06* .15* .73* 1.00 15. promotion 2.54 .60 .00 -.03 .00 -.06* .01 -.04 .01 .00 -.01 .01 -.01 -.06* .01 .01 1.00 16. underperf 2.52 .71 .11* -.02 -.10* -.00 .06* -.11* .06* .03 .02 .03 .05 .11* .11* .13* .08* 1.00 * Significant correlations at 5% level

(24)

23

regards to production performance indicators: the frequency of collection has its average somewhere around monthly and weekly, also showing these to workers at least quarterly, on average. The variable reward_d has a mean of 0.87, indicating that 87% of the establishments in the sample reward the achievement of production targets, but a little less give these rewards to all staff members, as seen in case of the variable allstaffrewarded_d (mean 0.78). Giving promotion based on individual merit and dealing with underperforming employees are definitely two relatively popular practices among firms in the sample. The average firm seems to at least take individual effort into account together with tenure, and also deals with underperformers sooner or later.

As shown in Table 3, there are some logical correlations between the control variables that are significant, like the correlation between firm age and size, firm size and competition or firm size and being part of a larger firm (multiest), or firm age and export. Some of the correlations look rather interesting, like the one between state ownership and export, which is negative, while foreign ownership and export are positively correlated. There is less correlation between the explanatory variables listed in the second half of the table. Those present are mostly the pairs that measure similar effects, as already established by the principal component analysis. Not surprisingly, many of the indicators for management practices are positively correlated with multiest, used to indicate multi-establishment firms. The reason for it can be that these establishments may need to implement practices that the parent company also uses.

6.2 Results: The effect of management practices on innovation

In order to examine the effect of the selected management practices on innovation, I use probit models, since the dependent variable RD_d is a dummy variable. Following the results of the principal component analysis, I use the four factors introduced in section 5.2 as explanatory variables instead of putting the individual variables in the model. Since RD_d takes the value of 0 in all observations with the industry codes 16 and 37, 14 observations are dropped from the regression, leaving a total number of 1251 observations. The results of the probit are shown in Table 4. Margins are reported in the appendix (Table IV). In Models 1-4, the four factors (job

autonomy, performance feedback, performance-based pay and promotions policy) are applied

(25)

24

Table 4: Management practices and innovation, probit

Model 1 Model 2 Model 3 Model 4 Model 5

Variables RD_d RD_d RD_d RD_d RD_d Job autonomy 0.060* 0.055* (0.033) (0.033) Performance feedback 0.089** 0.078** (0.034) (0.035) Performance-based pay 0.092*** 0.077** (0.030) (0.031) Promotions policy 0.091** 0.081** (0.038) (0.039) ln_size 0.191*** 0.189*** 0.190*** 0.195*** 0.182*** (0.047) (0.047) (0.047) (0.047) (0.048) ln_age 0.010 0.009 0.015 0.016 0.030 (0.045) (0.045) (0.045) (0.046) (0.046) multiest 0.166* 0.158 0.170* 0.178* 0.170* (0.101) (0.101) (0.101) (0.101) (0.102) foreign_own 0.008 -0.006 0.039 0.006 0.005 (0.123) (0.123) (0.123) (0.123) (0.124) state_own -0.027 -0.027 -0.012 0.009 0.007 (0.164) (0.164) (0.164) (0.165) (0.166) export_d 0.168* 0.130 0.150 0.160 0.144 (0.101) (0.101) (0.101) (0.100) (0.102) competition 0.088 0.075 0.075 0.080 0.076 (0.057) (0.057) (0.057) (0.057) (0.057) _cons -1.533*** -1.479*** -1.522*** -1.572*** -1.540*** (0.396) (0.397) (0.396) (0.398) (0.400)

Industry FE Yes Yes Yes Yes Yes

Country FE Yes Yes Yes Yes Yes

N 1251 1251 1251 1251 1251

Standard errors in parentheses * p<.10, ** p<.05, *** p<.01

Model 1 measures the effect of job autonomy, which has a positive and statistically significant effect on R&D, though only at a 10% level. The variables taskallocation and

paceofwork were highly correlated with this factor, which suggests that involving workers in

(26)

25

in R&D by 2.3%. This gives confirmation to Hypothesis 1, stating that granting higher levels of autonomy to workers encourages R&D spending, thus innovation activities at the establishment.

The next component, performance feedback is included in Model 2, and its positive effect on innovation is confirmed at a 5% level, which means that giving feedback to workers enhances engagement in R&D spending at the firm. In terms of numbers, it means that a 1 unit increase in the factor Performance feedback increases the chances of investing in R&D by 3.5%. Note that the variables freq_ind and perftoworkers are highly correlated with this factor, so the results also suggest that the higher the frequency of collecting production performance indicators and the frequency of showing them to workers in the form of feedback, the more likely the firm will spend on R&D, which also means confirmation to Hypothesis 2.

Model 3 in Table 4 contains the probit results for the effect of performance-based pay on innovation. The coefficient of performance-based pay enters the regression with a positive sign and is statistically significant at a 1% level. This supports the idea that firms applying performance-based pay schemes are more likely to spend on R&D. More precisely, it means that a 1 unit increase in the factor Performance-based pay increases the chances of spending on R&D by 3.6%. This component contains the variables on giving reward upon production target achievement, and giving it to all of the staff members, both being dummy variables, so the results refer to the establishment switching from no reward to distributing rewards. So applying rewards upon target achievement and rewarding all members of staff increase the probability of spending on R&D, which partly confirms Hypothesis 3.

The second half of Hypothesis 3 is resolved by Model 4. It estimates the effect of the fourth component, promotions policy on innovation, and the results show a positive significant effect of the factor at a 5% level. More specifically, a 1 unit increase in the factor Promotions policy increases the chances of investing in R&D by 3.6%. Referring back to the variables that were highly correlated with this factor, promotion and underperf, it indicates that firms that consider individual effort and merit when deciding about promotion and deal with underperformers as soon as possible are more likely to engage in R&D activities. These two variables together with the dummies for reward and whether all staff is rewarded incorporated in Model 3 were originally chosen to measure

Hypothesis 3. Both Models 3 and 4 confirmed the positive effects of the factors Performance-based

(27)

26

Finally, the last column reports Model 5, which contains all the four factors, showing what happens when all practices are applied together, incorporated in the same model. Since all four remain positive and statistically significant, we can confirm Hypothesis 4, according to which applying job autonomy, performance feedback, and performance-based pay and promotion at the establishment together has a significant positive effect on R&D spending, and thus on innovation.

Regarding the control variables, most of them remain statistically insignificant through the analyses. The only very strong significance is shown not surprisingly by firm size, which is positive and significant at a 1% level in all five models. Being part of a larger firm (multiest) also positively affects R&D spending, although it is only marginally significant at a 10% level in Models 1,3,4 and 5. Other than these two, the only control variable that turns significant is the export dummy, but only marginally and only in Model 1 with job autonomy.

6.3 Robustness check

In order to check the reliability of the analysis, I introduce a new dependent variable to replace the R&D dummy, also offered by the MOI Survey to measure innovation. The variable is

patent_home_d, which is a dummy variable taking the value of 1 if the establishment has any

patents registered in its home country, and 0 if it has no patents.

Patents had been discussed as a measurement of innovation in several previous studies (see e.g. Adams, Bessant and Phelps, 2006), and is often included in complex indicators as well (Furman, Porter and Stern, 2002; Hollanders & Esser, 2007). To avoid any inconsistency, I compared the definitions of R&D and patents used in the MOI Survey questionnaire to see how they refer to the newness or novelty content of these variables. The instructions for R&D in the manual are the following: “Research and development (R&D) is defined as creative work

undertaken on a systemic basis in order to increase the stock of knowledge. Research and development is distinguished from market research and product testing by the presence in research and development of an appreciable element of novelty.” Similarly to this, the definition of patents

according to the questionnaire manual states clearly that the invention it protects “must be new,

(28)

27

The outcome of the PCA are similar to the original analysis, and the variables load up to the same four factors (see Table V in the Appendix). The results of the probit are reported in Tables VI and VII in the Appendix. Table VI gives support to the results received in the original analysis to three out of four management practices: job autonomy, feedback and performance-based pay all appear with a positive sign and are statistically significant, though at a lower level than in the original model described in section 6.2. As shown by the margins (Table X), a 1 unit change in the variables increase the probability of having patents registered in the home country by 2.5% in case of both Job autonomy and Performance feedback, and 2.4% in case of Performance-based pay. They also keep their signs and level of significance when clustered together in the last model.

However, the coefficient for Promotions policy turns negative, though it is not statistically significant. Such results – even ones that suggest a significant negative effect of extrinsic rewards, like promotion – are not rare in psychology and behavioral science literature. In their analysis, Grabner and Moers (2013) examine retail banks and find evidence that when making promotion decisions, current job performance carries less information for decision-makers because they rather consider the capabilities of the individual based on subjective opinions of how effective that individual would be in the next position. According to Kohn (1993), the problem with rewards, especially with promotion is that they do not promote team work, but push the individual to act as best for them as opposed to the “competitors” for the promotion – mostly fellow co-workers. By intuition, this behavior also discourages risk-taking, which loops the problem back to creativity and innovation (Amabile, 1988; Amabile et al., 1996; Lee & Kelley, 2008).

All in all, there is still confirmation for Hypotheses 1,2 and 4, and also partly for Hypothesis

3 through the factor Performance-based pay. This is promising in light of the definitions introduced

above, confirming that the two dependent variables, R&D and patents both seem to control for the novelty content, and in doing so the results strengthen the notion that these management practices do support innovation. The lack of significance in case of promotion points back to the ambiguity in the literature already mentioned in sections 2 and 3 regarding extrinsic motives.

6.4. Discussion

(29)

28

bonuses and promotions based on the individual’s performance are more likely to engage in R&D spending and thus can be considered more innovative than firms not applying any of these practices. In this section I discuss the importance of these findings comparing them to the results of previous analyses in this field.

First, the findings regarding job autonomy are fully in line with previous studies. Besides being an important part of creativity literature, its relevance is also established by the resource based view and the agency theory, both applied in analyses related to management practices (e.g. Mumford, 2000; Beugelsdijk, 2008; Zoghi, Mohr and Meyer, 2010). It is important to see the means through which increasing autonomy in this case encourages innovation: decentralizing decision making regarding certain work practices is related to two areas in my analysis, decisions about allocating tasks and about setting the pace of work. This adds to the existing literature by breaking down job autonomy so that we know which forms of autonomy leave enough space for workers to be creative and innovative. Dissolving the rigidities of splitting responsibilities for certain tasks in a pre-defined manner, and allowing for a little flexibility in setting the pace of work can give the confidence to workers to try new ideas or explore new possibilities (Shipton et al., 2005). However, the positive effect here is only supported at a 10% level. The relative weakness of the significance level may come from the fact that the sample contains manufacturing firms, where there is probably less chance for flexibility, with all the strict production regulations and deadlines.

(30)

29

by the results that showing these performance production indicators and thus reflecting on one’s work increases the chances of innovation activities at the firm, as also suggested by e.g. Delery and Doty (1996).

Third, the importance of performance-based pay is also confirmed in the analysis at a 1% level. It is probably the most surprising and most robust finding of my analysis, because although it is included in all theories discussed in the literature review, studies including performance-based pay are far from being unified, with several empirical analyses reporting ambiguous or contradictory results (e.g. Shalley, Zhou and Oldham, 2004; Bandiera, Barankay and Rasul, 2005; Inderst, 2009). The strong positive results of my analysis add yet another aspect to the literature and the debate about how extrinsic motives hinder creativity of workers by pushing them to safe solutions. Now it looks like they might act as intrinsic motives fostering creativity, as also suggested by Barros and Lazzarini (2012). At least this is the conclusion for manufacturing firms of transition economies subject to this study.

Fourth, promotions policies of the firm seem to act in a similar vein as performance-based pay. The literature does mention promotion as a possible motivation for creativity and innovation, for example within the agency theory, but there are only a few studies that include it in their empirical analyses (e.g. Huselid, 1995; Barros & Lazzarini, 2012). Not only can we confirm its positive effect on innovation, but the PCA outcome also establishes that it is necessary to treat promotions policy separately from performance-based pay, instead of referring to them under the umbrella term of performance-based incentives. Some studies suggest that the two are much alike since promotion is likely to be accompanied by a pay raise almost every time (Gibbons & Waldman, 1999). However, as Auriol and Renault (2008) also make a point, promotion stands out as an incentive because it gives a certain status to the worker, which is a visible sign of social recognition that is much desired by employees and thus – with proper encouragement – it can be converted into an inner driver of performance and creativity.

(31)

30

other similar practices not measured in this analysis that can positively influence innovation performance, or it could be another set of practices that foster innovation in for example the service sector.

7. CONCLUSION

This study aims to test the effect of certain management practices on the innovation performance of firms. To do so, I first scanned the existing literature to identify practices that received theoretical support, possibly from more directions. The first step of the empirical analysis was a principal component analysis, a widely used approach in fields relying heavily on qualitative surveys like behavioral science or psychology. The eight variables (each measuring management practices) introduced to the PCA loaded up to four components that were tested in a probit model to analyze their impact on innovation, measured by an R&D dummy.

The results of the analysis gave support for all Hypotheses 1-3 that anticipated the positive effects of job autonomy, performance feedback and performance-based pay and promotion on innovation. Hypothesis 4 was also confirmed as the joint presence of these practices in one model indicated that their effects remained solid.

The study adds to the existing literature in two ways. First, it complements the previous empirical analyses carried out to measure the impact of certain HRM practices on innovation. By identifying and testing variables related to job autonomy, performance feedback, performance-based pay and performance-performance-based promotion, my analysis gives confirmation to the positive impact of these practices on innovation, either reinforcing previous results (like in case of job autonomy and feedback) or dissolving ambiguity (pay and promotion). Second, the study examines these practices in a special context, using data from transition economies, a yet under-researched context of innovation and management practices.

(32)

31

(33)

32

APPENDIX

Table I: Original questions for the explanatory variables in the MOI Survey

Variable Original question

taskallocation S.8 Who decides how tasks are allocated for workers in this establishment?

Only factory managers. 1

Mostly factory managers. 2

Factory managers and workers equally. 3

Mostly workers. 4

Only workers. 5

Other (SPECIFY - SPONTANEOUS) 6

Don't know (SPONTANEOUS) -9

Refusal (SPONTANEOUS) -8

paceofwork S.7 Who sets the pace of work in this establishment for workers?

Only factory managers. 1

Mostly factory managers. 2

Factory managers and workers equally. 3

Mostly workers. 4

Only workers. 5

freq_ind R.2b How frequently are these production performance indicators collected

in this establishment? Yearly 1 Quarterly 2 Monthly 3 Weekly 4 Daily 5 Hourly 6

perftoworkers R.2d How frequently are production performance indicators shown to

workers? Never 1 Annually 2 Half-annually 3 Quarterly 4 Monthly 5 Weekly 6 Daily 7 Hourly 8

(34)

33

Table I (cont.): Original questions for the explanatory variables in the MOI Survey Variable Original question

reward_d

allstaffrewarded_d R.7 How do you reward this establishment's production target achievement?

There are no rewards. 1

Only top and middle management is rewarded. 2

All staff is rewarded. 3

promotion O.14 Which if the following best corresponds to the main way employees

are promoted at this establishment?

Promotions are based solely on individual's effort

and ability. 1

Promotions are based partly on individual's effort and ability, and partly on other factors such as tenure (how long they have worked for the firm).

2 Promotions are based mainly on factors other than on individual's effort and ability, such as tenure. 3

Does not apply (SPONTANEOUS) -7

underperf O.15 Which of the following best corresponds to this establishment's main

policy when dealing with employees who do not meet expectations in their position?

They are rarely or never moved from their position. 1 They usually stay in their position for at least a year

before action is taken. 2

They are rapidly helped and re-trained, and then

dismissed if their performance does not improve. 3

Does not apply (SPONTANEOUS) -7

Notes: The original numbers assigned to the answers have been changed in the analysis for perftoworkers and promotion so that they reflect the changes from "bad" to "good" management practices more.

The dummy variables reward_d and allstaffrewarded_d have been formed based on the same question (R.7), the former being 1 in case of answers 2 or 3 (otherwise 0), the latter being 1 in case of answer 3 (otherwise 0).

(35)

34

Table II: Description by countries and main variables (Based on MOI Survey) _{R&D spending in the}

last year

Who decides how tasks are allocated in this establishment?

Who sets the pace of work in this establishment for workers? Country 0 1 1 2 3 4 5 1 2 3 4 5 Belarus 32 32 32 30 2 0 0 24 33 6 0 1 2.53 2.53 2.53 2.37 0.16 0.00 0.00 1.90 2.61 0.47 0.00 0.08 Ukraine 68 38 52 45 9 0 0 53 42 11 0 0 5.37 3.00 4.11 3.55 0.71 0.00 0.00 4.19 3.32 0.87 0.00 0.00 Uzbekistan 75 13 44 37 7 0 0 44 36 7 1 0 5.92 1.03 3.48 2.92 0.55 0.00 0.00 3.48 2.84 0.55 0.08 0.00 Russia 74 79 68 70 15 0 0 65 56 30 1 1 5.85 6.24 5.37 5.53 1.18 0.00 0.00 5.13 4.42 2.37 0.08 0.08 Poland 26 28 21 27 6 0 0 19 24 11 0 0 2.05 2.21 1.66 2.13 0.47 0.00 0.00 1.50 1.90 0.87 0.00 0.00 Romania 62 28 40 49 1 0 0 46 36 8 0 0 4.90 2.21 3.16 3.87 0.08 0.00 0.00 3.63 2.84 0.63 0.00 0.00 Serbia 56 36 40 48 4 0 0 39 46 6 1 0 4.42 2.84 3.16 3.79 0.32 0.00 0.00 3.08 3.63 0.47 0.08 0.00 Kazakhstan 65 33 42 49 7 0 0 36 46 15 0 1 5.13 2.61 3.32 3.87 0.55 0.00 0.00 2.84 3.63 1.18 0.00 0.08 Lithuania 40 33 33 31 9 0 0 31 29 13 0 0 3.16 2.61 2.61 2.45 0.71 0.00 0.00 2.45 2.29 1.03 0.00 0.00 Bulgaria 50 48 49 43 6 0 0 50 31 15 2 0 3.95 3.79 3.87 3.40 0.47 0.00 0.00 3.95 2.45 1.18 0.16 0.00 Germany 75 97 91 51 27 2 1 51 50 63 6 2 5.92 7.66 7.19 4.03 2.13 0.16 0.08 4.03 3.95 4.98 0.47 0.16 India 103 75 84 85 9 0 0 73 76 27 0 2 8.14 5.92 6.64 6.71 0.71 0.00 0.00 5.77 6.00 2.13 0.00 0.16 Total 726 540 596 565 102 2 1 531 505 212 11 7 57.35 42.65 47.08 44.63 8.06 0.16 0.08 41.94 39.89 16.75 0.87 0.55

Note: Numbers in Italic indicate percentages within the question. Plain numbers stand for number of observations.

THE EFFECT OF MANAGEMENT PRACTICES ON INNOVATION