Policy Research Working Paper 9067 Management Practices in Croatia Drivers and Consequences for Firm Performance Arti Grover Leonardo Iacovone Pavel Chakraborty Finance, Competitiveness and Innovation Global Practice November 2019 Policy Research Working Paper 9067 Abstract Embedding management and operational practices survey by 36 percent, profits by 33 percent and the probability to in a broader firm capabilities survey, this paper finds that innovate by 11 percent. Likewise, better managed firms an average firm in Croatia scores 0.532 on structured man- more likely use sophisticated technologies and have a higher agement practices, which is farther from the frontier (0.615 probability of accessing external finance. What drives firms in the United States). This average, however, masks the to improve their management practices? As elsewhere in the wide heterogeneity in management practices among firms. world, global linkages of firms matter. However, unlike the Relative to advanced countries, a large share of firms in evidence in advanced countries, management capabilities in Croatia are badly managed. Management is particularly Croatia is negatively associated with firm age, especially in worse in services and more so in non-knowledge intensive services, indicating the possibility of allocative inefficiency, services. Better managed firms show superior performance: where learning and selection mechanism does not weed improving the management score from the 10th decile to out the badly managed firms perhaps due to the lack of the 90th decile is expected to improve sales per employee pro-competitive forces. This paper is a product of the Finance, Competitiveness and Innovation Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at agrover1@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Management Practices in Croatia: Drivers and Consequences for Firm Performance Arti Grover, Leonardo Iacovone and Pavel Chakraborty JEL Classification: D22, D24, F14, L2, M2, O33, Keywords: management, firm, productivity, competition, Croatia Acknowledgements: The team is grateful to Todor Milchevski (GFCEE) for leading the survey implementation and Šime Smolić (Survey specialist, consultant) for his careful work on data collection. The team thanks Kristen Himelein (GPV02) for her invaluable advice on sampling strategy and Andrea August (Consultant) for excellent research assistance. The paper has benefited from overall guidance of Marialisa Motta (Practice Manager, GFCEE) and insightful suggestions from Austin Kilroy (GFCEE). The work on this paper is partially the result of the World Bank team activities for the Government of Croatia, under the ongoing Eastern Croatia Growth and Jobs Reimbursable Advisory Service, co-financed by the European Union from the European Regional Development Fund.                                                              The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. This paper does not necessarily represent the position of the European Commission and of the Croatian Government.  Finance competitiveness and innovation Global Practice, The World Bank, The authors can be reached at: agrover1@worldbank.org, liacovone@worldbank.org; pavel.chakraborty@gmail.com. Section 1: Management in Croatia and Why it Matters? The persistence in productivity differences between firms, within the same industry, is striking and puzzling. For example, labor productivity for plants in the United States at the 90th percentile, within narrowly defined four-digit manufacturing industries, is four times as high as plants at the 10th percentile. Likewise, the difference in Total Factor Productivity (TFP) is twice as high, when controlling for other factors (Syverson, 2011). These differences persist over time and are robust to controlling for plant- specific prices in homogeneous goods industries. Such TFP heterogeneity is not an artifact of the United States establishment data but is evident in several countries. As an example, the average 90th to 10th percentile TFP ratios is over 5:1 in China and India (Hsieh and Klenow, 2009). Most often within industry productivity differentials are attributed to “hard” technological innovations, as embodied in patents or in adoption of advanced equipment. Recently, Bloom et al. (2016a) provide an alternative explanation, that is, productivity differences reflect variations in management practices, which is also the approach adopted in this paper to explain the heterogeneity in performance of firms in Croatia.1 Persistent differences in productivity can be explained by several other factors, such as higher quality labor and capital, differential investment in information technology and research and development (R&D), learning by doing, firm structure, productivity spillovers, regulatory behavior, and differences in competitive regime. Bloom et al. (2016a) propose that some forms of management practices are like “technology”, where management is perceived as an intangible capital in which output is monotonically increasing. Differences in managerial quality are critical in explaining the cross-country differences in income levels, productivity, innovation and firm dynamism noted in Hall and Jones (1999). For example, Bloom and Van Reenen (2010) find that differences in productivity across firms in Asia, Europe, and North and South America can be attributed to the heterogeneity in management practices across firms in these regions.2 Likewise, Bloom et al. (2016a) suggest that management practices account for nearly 30 percent of the differences in TFP across countries. An important finding here is that management quality also explains at least a large share of within country heterogeneity in firm performance as well. For instance, using evidence collected through the World management surveys, Bloom et al. (2016a) find that about one-fifth of the dispersion in TFP across firms within countries is explained by differences in management quality, thereby suggesting that firm- and sector-specific factors were at least as important as the general business environment in shaping managerial performance.                                                              1 Bloom et al. (2016) suggest that management practices account for about one-fifth of the dispersion in TFP across firms within countries, and nearly 30 percent of the differences across countries. Thus, firm- and sector-specific factors are at least as important as the general business environment in shaping managerial performance. 2 Other performance measures, such as profitability and survival rates are also positively correlated with good management practices. We focus on management for explaining the heterogeneity in performance of firms in Croatia for the following reasons: First, management practices are strongly correlated with firm productivity in several countries (e.g. US, Russia, Mexico, Pakistan), and is robust to controlling for many other factors associated with productivity (e.g. exports, innovation etc.). Second, while a frontier topic globally, there exists little evidence on managerial practices of firms in Croatia and the relationship between management and productivity. For instance, there is some evidence on management in countries from the Europe and Central Asia region, where Bloom et al (2011) find a strong association between managerial practices and firm performance but this study excludes Croatia.3 Three, given the strong association with firm performance, it remains a challenge to understand the drivers of management and why these practices do not diffuse more readily. We hope that our results will help shape a broader operational program in Croatia on firm-level interventions to improve management. To quantify the importance of management for the performance of firms in Croatia, we conducted a survey on firm capabilities in Croatia, including a module for capturing management capabilities. Through this survey, we collected information on 727 small and medium enterprises (SMEs) in manufacturing and services sector. The survey was conducted between January and June 2019. Financial Agency (FINA) database provides the sample frame for this survey, where the sample was stratified by region and sector. Firms were randomly sampled and interviewed face to face. Survey data on managerial practices, technology adoption and innovation is merged with other firm performance data such as profits, sales per employee and TFP from FINA database.4 This paper benchmarks management quality of firms in Croatia against other countries such as the United States and quantifies the link between management capabilities and firm performance (e.g. productivity, profits, adoption of technology). How are structured management practices measured? The module on Business Capabilities and Entrepreneurship includes 15 core questions of the United States Census Management and Organizational Practices Survey (MOPS) main module. The survey was deliberately designed to replicate the United States Census MOPS to maximize comparability with the United States as well as other countries where this survey has been conducted (e.g. UK, Canada, Russia, Mexico and Punjab province in Pakistan). Following the scoring matrix of Bloom et al. (2019a), firm responses to questions on managerial practices is aggregated into a single management score that ranges from between 0 and 1 with 1.0 reflecting perfect                                                              3 In 2008 and 2009, the European Bank for Reconstruction and Development (EBRD) in cooperation with the World Bank (WB) conducted a new survey, the EBRD-WB Management, Organization and Innovation (MOI) survey. 4 Financial Agency (Fina) is the leading Croatian provider of financial data services. The database comprises of all tax registered firms in Croatia. (more formal, frequent and explicit) adoption of all structured management practice, and 0 score reflects the other extreme. Analysis of this newly collected survey data in Croatia, suggests that, first, an average manufacturing firm in Croatia scores 0.538 in adoption of advanced management practices, a value behind the one in frontier economies such as the United States (0.615, Bloom et al., 2019a). Comparing Croatia’s management practices with countries for which comparable data is available5, we find that an average manufacturing firm in Croatia is better managed than its counterparts in Russia (average score, 0.432; Grover and Torre, 2019) and Mexico (average score, 0.42; Bloom et al., 2019b) (Tables 1a and 1b), although we are concerned that the relatively higher score in Croatia vis-à-vis Mexico, for instance, could be the result of low response rate, leading to a possible selection bias and due to differences in coverage with respect to firm size in the sampling frame.6 Moreover, the quality of management practices across firms in Croatia is highly heterogeneous, with a large share of poorly managed firms and a much smaller share of better managed firms, relative to advanced countries. For example, only 3.5 percent of the firms in Croatia have a management score of over 0.75 as compared to 18 percent in the US. Second, weaknesses in management practices, particularly relating to data driven performance monitoring, cripples the overall management score in Croatia. Monitoring includes specific practices that help managers define their key performance indicators (e.g. inventory, sales, absenteeism) and appropriately track them periodically over time. By comparison, Croatia’s average score on practices relating to target setting and incentives is relatively closer to that in the United States firms. Incentives practices relate to how managers and non-managers are awarded bonuses and promotion, for instance. The finding for Croatia regarding the weaknesses of these sub-indices is similar to that in Russian manufacturing, that is, monitoring aspect of structured management is weaker. By comparison, in the United States and Mexico                                                              5 Apart from data availability, it makes sense to compare Croatia with Russia due to comparable per capita GDP in recent years and also the fact that both countries are transition economies in the same region. Comparison with Mexico is revealing in the sense that Mexico has a lower per capita GDP than Croatia, but has a strong manufacturing base, and thus offers another perspective for comparison (lower bound among emerging economies), aside from the US, which is the benchmark frontier economy (upper bound). 6 For comparator countries, the sample frame varies significantly. For example, MOPS in the United States used the sample frame of the Annual survey of manufacturing, comprising of around 50,000 plants in each wave for the years 2010 and 2015. Nearly 74,000 responses for the two years were received, that is, an average response rate of 74 percent. Likewise, in Mexico the survey was conducted for nearly 25,000 firms in manufacturing and services, with a response rate of 96 percent. The survey in Russia is perhaps closest to that in Croatia where 727 firms were interviewed face to face across five broad regions, with response rate similar to that in Russia (about 17 percent). Although the survey in Croatia covered both manufacturing and services, in this paper we present comparisons with manufacturing only. It is also worth reminding that although much of the lagging regions in Russia is not surveyed, the survey also skips bigger cities and agglomerations such as Moscow, Moscow Oblast, Saint Petersburg and Leningrad Oblast. Due to restricted coverage on location and size of the firms in Russia, management score could go in either direction. For example, we do not cover smaller firms (less than 25 employees) in Russia (which are covered in both Mexico and the US), and likewise, we do not cover much of the lagging regions in Russia as well. A broader coverage could therefore also pull down the current management score, depending on the share of firms that is added on either end of the location or size distribution. Furthermore, as we will learn later in the paper, a low response rate in Russia is likely leading to an overestimation of the observed management score. manufacturing firms are stronger in monitoring vis-à-vis structured management practices relating to workers’ incentives and target setting. Third, relative to manufacturing, services firms are slightly farther from the frontier when it comes to adoption of structured management practices. In Croatia, an average services firm scores 0.527 in adoption of the most advanced management practices, compared to 0.538 in manufacturing. Although the difference in average scores for the two sectors is not statistically significant, the distribution of management score shows that services has a larger share of badly managed firms, and a smaller share of the ones closer to the frontier. This finding is similar to that in Mexico (Bloom et al., 2019b), where firms from services sector were included in MOPS framework for the first time. The authors attribute the lower management score in services to a lack of pro-competitive environment, which lowers the pay-off from adopting better management practices. As in Bloom et al. (2019b) we also conjecture that a lack of market competition in Croatia may partly explain badly managed firms in services. In addition, the lower average score in services could also be a manifestation of a heavy left tail in size distribution of services sector firms. Given a small size, services firms may be less motivated to upgrade their management capabilities. Interestingly, services firms with 10-20 employees have a higher management score relative to their counterparts in manufacturing.7 Table 1a: Management in Manufacturing is farther from the frontier, but higher than other developing countries                                                              7Table A.1a presents the overall management score for Croatia, combining both manufacturing and services sector, while Table A.1b presents the average score for each individual question on management practices. Table 1b: Management in Services is farther from the frontier, but higher than other developing countries Fourth, management shows a wide dispersion across firms in Croatia, a finding similar to that observed in other countries. The distribution of management score in Croatia has a thicker left tail and a thinner right tail, that is, relative to frontier countries, a larger share of firms in Croatia are badly managed and a much smaller share is better managed. In particular, 38 percent of firms in Croatia score below 0.5, while only 3.5 percent of these firms have a score over 0.75 relative to 18 percent in the US. Nonetheless, the finding on the distribution of management score in Croatia is similar to that noted in Russia (Grover and Torre, 2019) and Mexico (Bloom et al., 2019b). Fifth, consistent with the results in other countries, we find that structured management practices is consistently and strongly associated with superior firm performance. Figure 1 shows that at higher quartiles of management firms demonstrate an ability to generate higher sales per employee and profits, propensity to adopt of more sophisticated technology for quality controls, innovate (e.g. patents) and train staff as well as amass external finance. Improving management score from the 10th percentile to the 90th percentile is expected to improve sales per employee by 36 percent, profit margin by 32 percent and is associated with 11 percent higher probability of introducing new products or processes. The likelihood of adopting sophisticated technology increases by 20 percent with similar jumps in management practices, while the probability of accessing external finance increases by 14 percent.8 While monitoring practices (e.g. tracking key performance indicators) are more important for labor productivity and adoption of sophisticated technology, management practices relating to incentives and target setting (e.g. hiring practices) have a larger and significant association with profits, innovation and access to external finance.                                                              8These results are robust to controlling for several firm level characteristics as well as sector and region fixed effects in a regression setting. Figure 1: Management matters for firm performance outcomes How do these results inform firm interventions designed for enhancing aggregate productivity? A simple answer to this question is that management can be improved with concerted effort. For example, an experiment with 17 textile firms in India provides a proof-of-concept that intensive individualized consulting can deliver lasting improvements in the practices of badly managed firms, where management practices improved, on an average by 38 percent in a 12 month period, resulting in productivity improvements by 17 percent (Bloom et al., 2013). Although this intervention improved management and firm performance, it came at a huge cost of approximately $75,000 per treated firm. This high cost is likely to be prohibitive for many small and medium enterprises (SMEs) to finance themselves, and for governments seeking to scale this up for assisting a large number of firms. In another experiment, Bruhn et al. (2018) shows that using local consultants can also improve business practices and performance and is less costly (US$12,000 per firm) than hiring internationally renowned consulting firms as in Bloom et al. (2013). Iacovone et al. (2019) tested two alternative approaches to improving management among firms producing auto parts in Colombia. The first uses intensive and expensive one-on-one consulting, while the second draws on agricultural extension approaches to provide consulting to small groups of firms at approximately one-third of the cost of the individual consulting services. Their results show that both approaches lead to improvements in management practices of a similar magnitude (8-10 percentage points), so that the new group-based approach dominates on a cost-benefit basis.9 This points to the potential of group-based approaches as a pathway to scaling up management improvements. Finally, given the evidence that management can be improved, our analysis of the survey data points to some of the shortcomings in the incentive system for firms that inhibit adoption of structured management practices. For example, management practices in Croatia are negatively correlated with firm age, and especially so in low-tech manufacturing and non-knowledge intensive services. This finding is surprising when compared with the United States or Mexico, where firms tend to improve their management practices as they age due to the “up-or-out” dynamics. The hypothesis is that as firms grow their production processes become more complex and in order to stay put in the market, they should have to upgrade their management capabilities. However, in Croatia, firms continue to survive without having the need to do so. This indicates the possibility that in Croatia learning and market selection is not operating perhaps due to the lack of pro-competitive forces.10 Likewise, the finding that exporting firms are better managed confirms that selection and learning mechanism can work in Croatia if firms operate in competitive conditions of an international market. More precisely, management score of a manufacturing exporter in Croatia is likely to be higher by 7 percent than an average firm. The fact that this result is mainly driven by high-tech manufacturing exporters, where management score is likely to be 13 percent higher, points to the potential for improving management capabilities among other firms and sectors. The paper beyond this point is organized as follows: Section 2 introduces the methodology of the management survey and presents some descriptive statistics for firms in Croatia. Section 3 presents the methodology for computing management scores, and the main descriptive results on the relationship                                                              9 Their results suggest that group-based intervention led to increases in firm size over the next 1.5 years, including a statistically significant increase in employment, while the impacts on firm outcomes are smaller and statistically insignificant for the individual consulting. 10 In the United States it is well known that “up-or-out” dynamics in which unsuccessful firms exit or are bought out by more successful competitors who absorb the labor and capital released by exiting firms drives productivity and business dynamism (Haltiwanger et al., 2013). In the case of Croatia, such selection dynamics does not seem to work in a similar way, and that firms that are badly managed still continue to survive. between various firm characteristics such as size, age, export status, ownership and so on with the observed management score. Section 4 presents a simple regression framework for assessing the link of management practices with firm performance. Section 5 explores the drivers of better management practices. The last section concludes with policy recommendations. Section 2: Critical Firm Capabilities: Management in Focus 2.1 Why do we need a management survey? Although managerial and organizational practices have been conceived to be critical in affecting firm outcome since the time of Adam Smith’s pin factory and Walker’s 1887 paper on business performance, these practices are not well measured in administrative data on firms or the commercial data on companies’ accounts. Some studies have used information on publicly traded firms to link managerial style of senior managers and company performance (e.g., Bertrand and Schoar, 2003), this data is rather limited to a country such as the United States and there is evidence that suggests a company is far more than simply the identity of its most senior employee (Bender et al., 2015). Furthermore, this type of data on its own does not tell much about how firms are managed or organized. Therefore, over the last decade Bloom and his co-authors have attempted to fill this gap in data by collecting comprehensive information on management practices (see Bloom et al., 2014). In general, three broad management and organizational capabilities that are generally recognized to be important for firm productivity (Bloom et al., 2016b): (i) performance monitoring: collecting and analyzing information, on daily activities of the firm for continuous improvement, (ii) target setting: using and stretching short and long run targets, tracking outcomes, and taking appropriate action, and (iii) performance incentives: rewarding high-performing employees, and retraining or moving underperformers and careful hiring. To collect information on these practices, two alternative strategies could be deployed: (i) Open Ended questions (e.g. world management survey, WMS)11 and (ii) Closed Ended questions (e.g. Management and Organizational Practices Surveys, MOPS).12 Table 2 summarizes the comparison of the two approaches on a number of dimensions.                                                              11 Designing these surveys take some expertise in terms of selecting questions and response grids and hence intense training and monitoring of the interview team. The full questionnaire is available on www.worldmanagementsurvey.com. 12 The full questionnaire is available on http://bhs.econ.census.gov/bhs/mops/form.html Table 2 – Comparing open-ended versus closed-choice methods for collecting information on management practices Source: Bloom et al. (2016b) Being interactive in nature, WMS elicits better quality of responses, MOPS, on the other hand, is cheaper and scalable. Conducting WMS requires highly trained interviewers, which is expensive and often difficult to organize. Based on Bloom and Van Reenen (2007), MOPS uses a multiple choice based evaluation tool that defines 16 key management practices in the above mentioned three areas –monitoring, setting targets and talent management. For each management practice, the responses are bounded by the choices in the questionnaire. 2.2 Sample and firm capabilities survey in Croatia The survey in Croatia includes the following modules: (i) Working Time, Employed Personnel and Remunerations; (ii) Training; (iii) Business Capabilities and Entrepreneurship; (iv) Government Programs and Funding; (v) Global Production Chains; (vi) Science, Technology and Innovation and (vii) Access to Finance. To collect data on managerial practices, we follow the exact questionnaire of management practices module of the MOPS framework for a survey in Croatia. The reason for replicating the MOPS framework is to maximize comparability of firm in Croatia management practices to the US13 and other such countries where such a survey has been conducted (e.g. Russia, Mexico, Pakistan). In addition, we                                                              13MOPS is comprised of 36 multiple choice questions about the establishment. The survey is split in three broad sections: management practices (16 questions), organization (13 questions) and background characteristics (7 questions). also included other modules on firm capabilities to collect more information the drivers and consequences of structured management practices. The survey was targeted at company managers, who are senior enough to have an overview of management practices but not so senior as to be detached from day-to-day operations. 2.3 FINA as Sampling frame and stratification The sample was stratified according to the following characteristics:  Sector: (i) High-tech manufacturing (NACE codes: 20, 21, 26, 27, 28, 29, 30) (ii) Low-tech manufacturing (NACE codes: 10, 11, 13, 14, 16, 17, 18, 22, 23, 25, 31, 32, 33) (iii) Knowledge intensive services (NACE codes: 50, 58, 60, 61, 62, 63, 64, 66, 69, 70, 71, 72, 73, 74) (iv) Non-knowledge intensive services (NACE codes: 45, 46, 47, 49, 52, 55, 56, 68, 79)  Regions: Counties in Slavonia are mapped to five broad regions. These include: (i) Zagreb and Zagreb county (ii) Northern Croatia (iii) Dalmatia (iv) Croatian Coast and Istria (v) Slavonia or Eastern Croatia. Below is the mapping of counties and Zagreb to the five regions. Although firms were not stratified by size, the survey was conducted only for small and medium firms (10-250 employees in manufacturing and 6-250 in services).14 The rationale for the chosen size category is the following: For smaller firms, adopting identified areas of management practices, as measured by MOPS, does not make much sense. By comparison, for larger firms, the distinction between establishments becomes blurred. This is critical because performance information from FINA database is                                                              14The distribution of services firms is comparable to that of manufacturing firms with 10 or more employees in terms of the sector’s contribution to firm count, employment and sales when the lower threshold is set to 6 employees. available at the firm, while for larger firms there may be vast differences in management scores across plants. Of a total of 122,111, active firms recorded in FINA for the year 2017, a total of 13,905 firms meet the above criterion and have non-missing values of employment and sales such that a basic measure of firm performance, sales per employee, can be computed. The data on firm performance in Croatia are basic and does not, for instance, have information on prices which substantially limits the ability of researchers to apply the frontier techniques for measuring firm productivity. These datasets allow for analysis of productivity using measures such as output per worker or TFP (e.g. using Levinsohn and Petrin methodology).15 Table 3 provides the distribution of 727 surveyed firms by region and sector. Table 3: Distribution of firm count by region and sector 2.4 Response rate and selection bias Bloom et al. (2016) note that for these types of surveys private sector firms often only have response rates of 5 to 10 percent, which raises an obvious concern about selection on unobservables. To reach a target of 727 manufacturing firms in the selected sector and size categories, we contacted 4,807 firms during January-June 2019. Of these firms, more than 95 percent of the firms could be contacted, with 54 percent of the firms making hard refusals to participate, while another 1.5 percent of the firms were removed from the sample not fitting the sampling frame. Lastly, of the firms, those were contacted, about 15 percent of the firms refused to get interview temporarily (the survey agency is still in contact); for about 3.5 percent                                                              15Due to unavailability of product level price information by firms, total factor productivity (TFP) is often measured in revenue (known as TFPR) rather than quantity terms (known as TFPQ). This is not ideal because firms may charge higher prices—and therefore appear to be more productive in TFPR terms—for reasons that are unrelated to technical efficiency. Foster et al. (2008) show that TFPR is positively correlated with TFPQ but confounds the “true” measure with idiosyncratic demand and factor prices effects. For a set of 11 highly homogeneous manufactured products in the United States (for example, ready-mixed concrete, raw cane sugar, boxes, and the like), the authors report a correlation of 0.75 between the two measures. of firms, the interview was scheduled, but cancelled just before it was supposed to take place. Overall, the response rate among firms those were interviewed is around 17 percent. It is worth noting that an average firm in our survey is similar in terms of employment (within the SME category) and age as compared to the sampling frame. The average size in the sampling frame is 25 employees, whilst in our survey it is 24 employees (See Table 4). Likewise, the average age in the sampling frame is 16.6 years, whilst in our survey it is 17 years. Our survey has 28 percent of firms reporting to have exported in 2017 as compared to the observed 25 percent in the sampling frame. Although, the firms in our survey are more or less similar to firms in the sampling frame, we still check for the selection bias using inverse probability selection analysis. We run a simple regression of a binary variable capturing whether a firm is interviewed or not as a function of firm characteristics such as – size, age and export status.  Table A.2 (in the Annex) presents the required result. The estimates do not point towards any clear evidence of selection bias for Croatia, at least on the observables captured by the data. Table 4: Descriptive statistics of surveyed firms We also checked our results using the maximum entropy analysis. The mean management score after correcting for the selection bias is 0.535, whereas without the correction, it is 0.532. Therefore, we can safely argue that our results are not affected by selection bias based on firm’s observed characteristics. 2.4 Computing management score Firm responses to questions on managerial practices are aggregated into a single management score following Bloom et al. (2019a) in two steps. First, the responses to each of the management practice question are normalized on a 0-1 scale: the response which is associated with the most structured management practice is normalized to 1, and the one associated with the least structured is normalized to zero. More structured management practices are those that are more specific, formal, frequent or explicit.16 Second, the management score is calculated as the unweighted average of the normalized responses. Thus, the final score on a firm’s management practice is scaled from 0 to 1. Firms with the extreme score 0 are those that selected the options that show little structure around performance monitoring, targets and incentives in the firm, while those with perfect score of 1 represent an establishment that selected the top category revealing an explicit focus on performance monitoring, detailed targets and strong performance incentives. We also separate the overall management index into two sub-indexes separately assessing monitoring practices and practices relating to incentives and targets. Section 3: Management in Croatia: Some descriptive results 3.1 How well do firms in Croatia adopt structured management practices? As in Russia (Grover and Torre, 2019), Mexico (Bloom et al., 2019b) and the United States (Bloom et al., 2019a), adoption of structured management practices across firms in Croatia is quite heterogeneous. Although the average management score in Croatia is 0.532, that is, on an average the surveyed firms adopted 53 percent of the overall structured management practices, quality is highly heterogeneous, with a large share of poorly managed firms and a much smaller share of better managed firms, relative to advanced countries (Figure 2).17 For example, only 3.5 percent of the firms in Croatia have a management score of over 0.75 as compared to 18 percent in the US. Figure 2: Relative to the US, Croatia has a lot of badly managed firms, and fewer of the better managed ones 2.5 2 1.5 1 .5 0                                                                  16 Some examples are in Bloom et al. (2019a) while detailed instructions for assigning scores are available on the United States census website. Available on http://bhs.econ.census.gov/bhs/mops/SUR766_9.html 17 Of the 727 manufacturing firms surveyed in the five regions in Croatia, all of them have more than 10 non-missing responses that allows for computing a more precise management score (Bloom et al., 2019a). 3.2 On which management practices is Croatia doing better? Which practices are worse? Weaknesses in management practices particularly relating to data driven performance monitoring cripples the overall management score in Croatia. Data driven performance monitoring includes specific practices that help managers define their key performance indicators (e.g. inventory, sales, absenteeism) and appropriately track them periodically over time. Incentives management practices relate to, for example, processes by which managers and non-managers are awarded bonuses and promotion. The finding for Croatia regarding the weaknesses in practices relating to objective and frequent monitoring of production indicators is similar to that observed in Russian manufacturing, that is, monitoring aspect of structured management is weaker relative to incentives and targets. Figure 3: An average firm in Croatia is weaker at practices relating to monitoring performance indicators 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Performance Monitoring Targets & Incentives Croatia Russia Mexico Pakistan US By comparison, this finding is in contrast with firms in the United States and Mexico which are stronger in monitoring vis-à-vis structured practices relating to workers’ incentives and target setting (Figure 3, left panel; also see tables 1a and 1b). Underlying the lower average score on practices relating to data driven performance monitoring is the large heterogeneity across firms. Notably, the distribution of monitoring practices has a thicker left tail, thereby suggesting that a large share of firms in Croatia do not explicitly, or frequently track performance indicators. Likewise, the thin right tail of the distribution is indicative of the problem that a very small share of firms have specific and formal practices for monitoring performance indicators (Figure 3, right panel). 3.3 Which sectors have better management practices? a. Manufacturing versus services Manufacturing firms are better managed than services firms (Table 5a). However, the difference is not significant. Table 5b reveals that the average difference between services and manufacturing is explained by the scores at the ends of the distribution. Average management score for services firms between 10-20 employees is actually higher than the corresponding firms in the manufacturing sector. Given that services is dominated by smaller sized firms, the sector hosts a larger share of badly managed firms and a smaller share of the ones that are closer to the frontier (Figure 4). Table 5a: Relative to manufacturing, services in Croatia poorly managed Table 5b: Lower average management score in services is possibly due to smaller firm size Notes: The table presents mean management score across firm size categories. For the survey, the size threshold for services firms is 6 employees, relative to 10 in manufacturing. Figure 4: Services has a larger share of badly managed firms and a smaller share of frontier ones 3 2 Density 1 0 0 .2 .4 .6 .8 1 mscore manufacturing services b. Technology and knowledge intensity The adoption of structured management practices is around 6 percent lower for low-technology sector relative to high-technology sectors in Croatia (Table 6); but, the difference is not statistically significant. By comparison, firms in non-knowledge intensive services are significantly poorly managed relative to those in knowledge intensive services. Table 6: Relative to other sectors, high-tech manufacturing and knowledge-intensive sectors are better managed c. By ownership Family owned firms have lower management capabilities than non-family owned firms in Croatia. There is a significant difference between family-owned firms and non-family-owned firms in the manufacturing sector, but in case of services, there is no difference between these two sets of firms. Professionally run firms are better managed in general (Table 7); more so in case of manufacturing but, not services. Table 7: Relative to family firms, professionally run firms are better managed d. Across regions Management practices are higher in case of Zagreb and lowest for Northern Croatia. Slavonia and Croatian Coastline have similar average score of management practices (Table 8). Firms in Zagreb have 7.7–11 percent higher management practices than firms of other regions in Croatia. Management practices in Zagreb are significantly higher than other regions of Croatia. Table 8: Firms in Zagreb are better managed e. Age Unlike the results from prior studies on management practices in the United States, firm’s age is negatively correlated with management practices in Croatia. In the case of Mexico and U.S., manufacturing firms improve their management scores with age – presumably through a combination of learning such that firms get better with age and selection, whereby badly-managed firms exit over time (Bloom et al., 2014; Figure 5, left panel). This results in a decline in dispersion of management practices as firms age, because of the exit of poorly managed firms at an early stage of their life cycle, implying that older firms in the United States are more similar in their management capabilities. (Figure 5, left panel). By comparison, such learning and selection mechanisms do not seem to be playing out in Croatia where firm’s management capabilities decline with age while the spread does not shrink either (Figure 5, right panel).   Figure 5: In the US, firms’ management score rises with age and spread falls, while the reverse is true for Croatia United States Croatia       Note: The left panel plots the dispersion and the level (normalized to 0 at the mean) of the management score for firms in the United States; source is Figure 8 in Bloom et al (2013a). f. Size In US, Bloom et al. (2014) show that better managed firms are significantly larger than poorly managed firms: In their case, a one standard deviation of management is associated with a 40 log point increase in employment size. We find similar results in Croatia, where firm size is significant and positively correlated with management score (Figure 6). Figure 6: In Croatia, firms’ management score rises with size g. Exporters In manufacturing, exporters have adopted significantly higher management practices. On average, an exporter adopts 57 percent of the management practices, whereas a non-exporter has only 52 percent. However, in case of services the difference is not very clear whether exporters are better managed (Table 9). Table 9: Exporters are better managed, especially in Manufacturing Section 4: Do management capabilities matter for firm performance? Differences in management practices are meaningful, from an economy-wide point view, insofar as they make a difference on firm performance. Recent evidence in countries such as the US, Russia and Mexico have established that management practices are tightly linked to firm performance in these countries, it is therefore imperative to explore such a connection in the context of Croatia as well. The plots of firm performance with quartile of management score in Figure 1 is indicative of the importance of management capabilities for firm’s sales per employee and engaging in innovation, and technology adoption. This section analyses the relationship between management and firm performance in a more robust way, controlling for observable firm characteristics as well as sector and region dummies. In order to systematically test this relationship between management capabilities and firm performance, we follow Bloom et al. (2019a) approach and start from a standard production function: , Where Yi is output (or firm’s total sales), Ai is total factor productivity (excluding management practices), Ki denotes the firm’s capital stock (in our case, its fixed assets), Li is employment, Ii are intermediate inputs, Xi is a vector of additional factors such as the employees’ education and Mi is the management score. Dividing by employment and taking logs we obtain the following equation that can be estimated with our dataset: 1 Where we have substituted Ai by a set of sector (NACE 2-digit) fixed effects and district fixed effects . and a stochastic residual ui. Standard errors are clustered by the broad sector-district at which our sample was stratified. Tables 10a-10c present the corresponding regression estimations for various measures of firm performance for all sectors, while Annex tables A.3a and A.3b splits the analysis by manufacturing and services.18 Column (1) in Table 10a regresses natural logarithm of sales per employee on the overall management score of a firm controlling for region and industry fixed effects. Our coefficient estimate demonstrates that if a firm improves its management score from the 10th percentile to the 90th percentile, the sales per employee or labor productivity of a firm increases by 36 percent.19 Column (2) additionally controls for firm size (number of employees), inputs used (capital and material) and education of workers and managers; our coefficient of interest remains positive and significant. The increase in sales per employee of a firm as a result of the increase in the management score from the 10th to the 90th percentile even after controlling for other important firm characteristics is around 10 percent. In other words, our estimations suggests that management score plays an important role for firm performance. Columns (3) and (4) repeat the regression of column (2), but by replacing the overall management score by score according to monitoring and targets and incentives. Following Bloom et al. (2018), we divide our overall management score into (i) monitoring, and (ii) targets and incentives. Substituting the overall management score with its components does not qualitatively change the outcome. Both the components are significantly                                                              18 Tables A.3a and A.3b present the same results, but by dividing the sample into manufacturing and services. 19 This is computed by multiplying the coefficient on the key driver variable (e.g., management in column 1) with the change in management score from the 10th percentile to the 90th percentile. correlated with labor productivity of firm; the conditional effect of monitoring being 7.5 percent while that of incentives and targets being 6.3 percent. Thus, although Croatia’s firms is weaker in monitoring practices relative to incentives and targets, it stands to gain slightly more by improving these practices. Columns (5) – (8) repeat the estimations in columns (1) – (4) for profits of a firm. Higher management score is associated with higher levels of profits. Upon dividing the aggregate score, the estimates show that overall result is driven mainly by targets and incentives component of the overall management score. In a conditional regression, improving overall management practices overall from the 10th to the 90th percentile is associated with 33 percent increase in profits, most of which is derived from practices relating to better target setting (29 percent associated increase). Table 10a: Management and Firm performance (Sales per Employee and Profits) Table 10b uses two other key firm performance measures – innovation and technology adoption. Columns (1) – (4) use innovation as the dependent variable. It is a dummy – it takes value 1 if a firm registers itself for a trademark, patent, utility and industrial designs. Management score of a firm is significantly correlated with the probability of innovation. Unconditionally, improving management score from the 10th to the 90th percentile is associated with 11 percent increase in probability of the firm registering innovation. The number drops to 7 percent, when controlling for other factors, but the estimate remains strongly significant. Columns (3) and (4) show that probability of innovating is strongly correlated with targets and incentives mechanism of management practices. In addition, our results show that higher the number of workers with a college degree, the higher is the probability of innovating a new product or a process, highlighting the importance of skills for innovation. We find similar results when using technological adoption as the outcome of interest. The value takes 1 when a firm has adopted sophisticated technology for quality control. However, we find that both the components of management score have a similar relationship with the probability of adoption of high technology. More specifically, improving management score from the 10th percentile to the 90th percentile is associated with 20 percent higher probability of adopting sophisticated technology in unconditional regressions, where both monitoring and incentive and targeting practices are statistically significant, although the former have a higher marginal effect. Table 10b: Management and Firm performance (Innovation and Technology Adoption) Lastly, Table 10c examines whether access to external finance and participation in training programs is correlated with management practices of a firm. We find strong evidence of positive association with respect to both the outcomes of interest. Management is associated with a greater probability of accessing external finance; improving management score from the 10th percentile to the 90th percentile is expected to increase this likelihood by 14 percent.20                                                              20These results are robust to controlling for several firm level characteristics as well as sector and region fixed effects in a regression setting. Table 10c: Management and Firm performance (Access to External Finance and Participation in Training Programs) For simplicity, we summarize the most interesting results using rope ladder plots in Figure 7. Figure 7: Management, particularly relating to Targets and Incentives practices matter more for firm outcomes Notes: The figure is a rope-ladder representation of estimated coefficients and 90 percent confidence bands from regressions of the observed firm performance measure on management score controlling for employment, intensity of inputs used, exporter status, education of the managers, sector and region fixed effects. Due to limitations in our survey data (e.g. information on management practices are based on one year of data collection), we are unable to test for the causality of management practices with superior firm performance in Croatia. Nonetheless, the literature on other countries has shown that indeed there is a causal relationship going from improved management practices to increased firm performance. For example, in the case of post-war Italy, Giorcelli (2018) finds long term effects of better management practices on performance of firms (e.g. productivity, sales, and survival) relative to machinery purchases or technology. The key channel through which skills improved firm performance was by helping managers make better investment decisions—investing in new plants or new machines, for example—which made their production more efficient. The fact that the coefficients obtained from our analysis are similar to those found in the existing literature is reassuring in that better management practices in Croatia will likely drive positive effects on firm performance. Section 5: What drives firm-level differences in management practices? The realization that firms in Croatia and particularly the smaller and non-knowledge intensive firms in the services sector are far from the frontier in adoption of structured management practices, what can we really do about this? This section analyzes the factors that matter for management capabilities, with the objective of bridging the gaps to help firms upgrade their management quality. Firm characteristics play a key role in explaining variations in management score. Table 11 presents the regression estimations of these factors. Table 11: Drivers of management The regression estimations suggest that, first, in line with prior studies, firm size plays a key role in adoption of management practices as evident in both unconditional (column 1) and conditional correlations (column 6) – doubling firm size is associated with approximately 0.047 point increase in management score. Going across columns, this relationship appears to be stronger for manufacturing and particularly low-tech. Services, and especially knowledge intensive services, have a weaker relationship of management capabilities with firm size. This result is analogous to the finding in Mexico where the relationship between firm size and management is weaker in services relative to that in manufacturing (Bloom et al., 2019b). Second, as noted in the descriptive in Section 3, regressions confirm that firm age in Croatia is negatively correlated with management score, that is, adoption of structured management practices is lower among older firms (column 2). Following similar analysis in Bloom et al. (2019b) on Mexico and the work of Hsieh and Klenow (2014) regarding the life cycle of firms, this finding appears to be indicative of poor allocative efficiency among firms in Croatia where clearly, market selection is not operating perhaps due to the lack of pro-competitive conditions, especially in services. Nevertheless, the finding in Croatia is somewhat similar to that observed in a survey of manufacturing firms in Russia such that firms also do not improve with age (Grover and Torre, 2019), however, in Croatia this negative association is primarily driven by the services sector. Third, demonstrate that management score is positively and significantly correlated with the likelihood being an exporter, and having higher a share of tertiary educated workers (columns 3-4). Exporters are more likely to be better managed, and significantly so in manufacturing. More precisely, the management score of a manufacturing exporter in Croatia is higher by 7 percent than a non-exporter. This result is mainly driven by high-tech manufacturing firms, where the same difference is likely to around 14 percent, thereby suggesting that management capabilities can be improved among other firms in all sectors. Column (5) uses an indicator for ownership of firms. It takes value 1 if a firm is family-owned and run by a family member of the founder and 0 otherwise. We do not find any relation between ownership and management practices of a firm. Column (6) puts all these together and the results remains the same, qualitatively, that is, – firm size, export status and education of employees are significantly correlated with management score. The overall explanatory power of firm characteristics in explaining management score is still quite limited – in the most complete specification (column 6) explains only 20 percent of the overall variation in the management score, implying that nearly 80 percent of the variation in management practices is explained by idiosyncratic, firm-specific factors. Columns (7) – (12) replicates column (6), but by separating he sample into manufacturing, services, high- tech, low-tech, knowledge intensive, and non-knowledge intensive sectors. Overall, results remain the same across sectors. Using rope ladder plots Figure 8 shows the results of the estimations from Table 11. Figure 8: Larger firms and those with more educated workers have better management Notes: The figure is a rope-ladder representation of estimated coefficients and 90 percent confidence bands from regressions of the estimated management score controlling for employment, age, exporter status, education of the managers, sector and region fixed effects. Section 7: Policy conclusions and discussion This paper presents novel evidence on managerial capabilities of firms in Croatia and benchmarks it relative to other countries for which comparable data is available. Information collected from the firm capabilities survey with 727 firms suggests that: first, an average firm in Croatia scores 0.532 on adoption of structured management practices, a score which is higher than that observed in Mexico but is farther from a frontier economy such as the United States, which has a score of 0.615. Second, the weaknesses in management practices particularly relating to data driven performance monitoring pulls down the overall score in Croatia. Three, there is much heterogeneity in adoption of structured management practices among firms in Croatia. As in Russia, Mexico and Pakistan, the distribution of management score among firms in Croatia have a thick left tail, implying that a large share of firms in these countries, including are badly managed, when compared with advanced countries such as the US. Four, management score is lower for services relative to manufacturing, implying that there is much potential to improve the management among services firms, especially those in non-knowledge intensive services. Five, our work confirms that better management practices is associated with superior firm performance. Firms with more structured management practices have higher sales per employee, profitability, are more innovative and more likely to adopt sophisticated technology. This result is highly robust to a numerous checks and controls as well as varying measures of firm performance. Finally, while explaining the differences in management practice across firms, it becomes clear that among the firm characteristic that explain at least a portion of the variation in management practices, firm size, foreign linkages and the education level of non-managers are the most relevant. Surprisingly firm age is negatively associated with better management practices, especially in services. This implies that unlike in the United States and Mexico, firms do not learn to be better managed as they age. Thus, it is possible that a lack of pro-competitive forces in Croatia and more so in the services sectors do not promote learning and possibly hinder market selection. If management matters for firm performance, then why firms do not adopt better management practices? Our survey also includes a question relating to the firm’s own perception of its management practices, which was originally a part of WMS survey questionnaire. More specifically, we ask the following question: “Excluding yourself, how well managed is your firm on a scale of 1 to 10, where 1 is worst practice, 10 is best practice and 5 is average”. Prior studies using WMS data that compared self-perception of management capabilities with the objective management score found a huge divergence between observed management capabilities and perceived one. In line with these findings, we uncover that firms in Croatia also have a higher perception on their management practices (Figure 9). Given that firms do not realize that there is a problem with respect to their management practices, it does not get the required attention. Our survey objectively measures management capabilities among firms, and also provides an understanding of the capabilities that these firms are weak at. Figure 9: Firms perceive their management to be better than it actually is Difference vs. Real Management Score Croatian Firms (Manufacturing and Services) 0 .2 .4 .6 .8 1 Real Management Score Diff = Perceived - Real Fitted values Following the ABC framework for firm growth proposed in Grover, Medvedev and Olafsen (2019)21, we make the following recommendations for enhancing productivity and business dynamics in Croatia: 1. Allocative efficiency: The fact that management capabilities are negatively associated with firm age is an indication of a lack of learning and selection mechanism operating in Croatia, especially in the services sector. Comparison with peer countries reveals slow pace of business environment reforms in Croatia. The 2018 Global Competitiveness Report of the World Economic Forum ranks Croatia at 68th place, the lowest among EU Member States. Croatia also continues to lag Central, Eastern and Southeastern Europe (CESEE) countries in key Doing Business indicators, including those relating to starting a business, dealing with construction permits, access to credit and resolving insolvency. Croatia’s overall ranking on Doing Business 2019 is 58th, falling 7 places compared to last year. Firm entry is poorly ranked, at 123rd place, with 8 procedures that typically last 22.5 days and costs of 6.6 percent of income per capita. This is more burdensome than what is required in the ECA region and double the average of OECD countries. Relative to its CESEE peers, Croatia is not doing better on resolving insolvency matters either. The inefficient insolvency framework obstructs exit and re-entry of businesses into markets, which could potentially contribute to significant misallocation in Croatia. The Doing Business Report 2019 ranks Croatia at 59th place out of 190 economies on the ease of resolving insolvency. Policies to support healthy business dynamics need to establish the basic enabling conditions and facilitate reallocation of resources (e.g., labor, capital, land) across firms, for example, by easing the business environment for entry and exit of firms. Such policies will increase competition and market will ensure that firms are motivated to upgrade and invest in their capabilities. 2. Business to Business linkages: Our paper finds that firms with foreign linkages, that is, exporters are more likely to be better managed. In this context, policies to connect firms to export markets can encourage learning and hence better management practices. Such firms are more likely to engage in quality upgrading, leading to higher productivity and firm growth. 3. Capabilities of firms: More critically, our work confirms the tight association of managerial capabilities with firm performance for firms in Croatia. Management matters not only for productivity but also for profits and innovation. Thus, firms in Croatia are more likely to benefit from training on structured management relating to these practices.                                                              21This framework proposes that policies for improving firm dynamism and supporting job creation should steer away from picking potential winners and focus on improving Allocative efficiency, strengthening Business-to-business spillovers, and building firm Capabilities. These policies rely on encouraging innovation, network and agglomeration economies, global linkages, worker skills and managerial capabilities, and financial development to rebalance growth/productivity policies. References Bender, S., N. Bloom, D. Card, J. Van Reenen, and S. Wolter. (2015). “Management Practices, Workforce Selection and Productivity”, Mimeo. Bertrand, M. and A. Schoar. (2003) “Managing with Style: The Effect of Managers on Firm Policies”, Quarterly Journal of Economics, 118 (4): 1169-1208 Bloom, N. and J. Van Reenen. (2007) “Measuring and Explaining Management Practices Across Firms and Countries”, Quarterly Journal of Economics, 122 (4): 1351-1408 Bloom, N. and J. Van Reenen. (2010) “Why Do Management Practices Differ across Firms and Countries?”, Journal of Economic Perspectives, 24 (1): 203-24 N. Bloom, H. Schweiger and J. Van Reenen. (2011), “Quality of management practices in transition countries”, NBER Working Paper No. 17231. Bloom, N., B. Eifert, A. Mahajan, D. McKenzie and J. Roberts. (2013). “Does management matter? Evidence from India." Quarterly Journal of Economics, 128 (1): 1-51. Bloom, N., R. Lemos, R. Sadun, D. Scur, and J. Van Reenen. (2014). The New Empirical Economics of Management”, Journal of the European Economic Association, 12 (4):835–876 Bloom, N., R. Sadun, and J. Van Reenen. (2016a). “Management as a Technology.” Harvard Business School Working paper # 16-133 Bloom, N., R. Lemos, R. Sadun, D. Scur, and J. Van Reenen. (2016b). “International Data on Measuring Management Practices”, American Economic Review: Papers & Proceedings, 106 (5): 152–156 Bloom, N., E. Brynjolfsson, L. Foster, R. Jarmin, M. Patnaik, I. Saporta-Eksten. and J. Van Reenen. (2019a) “What Drives Differences in Management Practices?”, American Economic Review, 109 (5): 1648-1683 Bloom, N., L. Iacovone, M. Pereira-López and J. Van Reenen. (2019b) “Spillovers, Market Size and Misallocation: Management in Mexico”, Mimeo Bruhn, Miriam, Dean Karlan and Antoinette Schoar (2018). “The Impact of Consulting Services on Small and Medium Enterprises: Evidence from a Randomized Trial in Mexico.” Journal of Political Economy, 126(2): 635-687. Foster, L., J. Haltiwanger, and C. Syverson. (2008). “Reallocation, Firm Turnover, and Efficiency: Selection on Productivity or Profitability?” American Economic Review, 98 (1): 394–425. Grover, A., D. Medvedev. and E. Olafsen. (2019) “High-Growth Firms: Facts, Fiction, and Policy Options for Emerging Economies,” World Bank, Washington, DC Hall, R. and C. Jones. (1999). “Why Do Some Countries Produce so Much More Output per Worker than Others?" Quarterly Journal of Economics, 114 (1): 83-116. Haltiwanger, J., R.S. Jarmin and J. Miranda. (2013) “Who Creates Jobs? Small versus Large versus Young”, Review of Economics and Statistics, 95 (2): 347-361 Hsieh, C.-T., and P. J. Klenow. (2009). “Misallocation and Manufacturing TFP in China and India.” Quarterly Journal of Economics, 124 (4): 1403–48. Hsieh, C.-T. and P. J. Klenow. (2014) “The Life Cycle of Plants in India and Mexico”, Quarterly Journal of Economics, 129 (3): 1035-1084 Iacovone, L. and M. Pereira-López. (2017) “Management Practices as Drivers of Innovation: New Evidence from Mexico”, Mimeo Iacovone, L., M. Maloney and D. McKenzie. (2019). Improving Management with Individual and Group- Based Consulting: Results from a Randomized Experiment in Colombia, World Bank Policy Research Working Paper 8854. Lemos, R., A. Choudhary, J. Van Reenen, and N. Bloom. (2016) “Management in Pakistan: First Evidence from Punjab”, International Growth Centre Working Paper, London, UK. McKenzie, D. and C. Woodruff. (2017) "Business Practices in Small Firms in Developing Countries," Management Science, 63 (9): 2967-2981. Syverson, C. (2011). “What Determines Productivity?” Journal of Economic Literature, 49 (2): 326–65. Walker, F. (1887) “The Source of Business Profits.” Quarterly Journal of Economics. 1 (3): 265-288 Annex: Table A.1a – Management Score in Croatia  Table A.1b –Score by all dimensions of structured management  Table A.2: Selection Bias Analysis (Inverse Selection Sales per Employee and Profits)  Table A.3a: Management and Firm Performance: Sales per Employee, Profits, Innovation, and Technology  Adoption  Table A.3b: Management and Firm Performance: Access to External Finance and Participation in Training  Programs  33