46182 Understanding and Improving Data on Entrepreneurship and Active Companies June 2008 Jacqueline Coolidge Kusisami Hornberger Ronald Luttikhuizen Disclaimer The Organizations (i.e., IBRD, IFC and MIGA), through FIAS, endeavour, using their best efforts in the time available, to provide high quality services hereunder and have relied on information provided to them by a wide range of other sources. However, they do not make any representations or warranties regarding the completeness or accuracy of the information included in this report, or the results which would be achieved by following its recommendations. About FIAS For over 20 years, FIAS has advised more than 130 member country governments on how to improve their investment climate for both foreign and domestic investors and maximize its impact on poverty reduction. FIAS is a joint service of the International Finance Corporation, the Multilateral Investment Guarantee Agency and the World Bank. We receive funding from these institutions and through contributions from donors and clients. FIAS also receives core funding from: Australia New Zealand Canada Norway France Sweden Ireland Switzerland Luxembourg United Kingdom Netherlands Understanding and Improving Data on Entrepreneurship and Active Companies ii June 2008 Contents Executive Summary iv 1. Introduction 1 2. Data on entrepreneurship and its limitations 4 3. Implications for use of data on entrepreneurship 10 4. Conclusions and recommendations 12 References 19 Annex 1.Entrepreneurship Database including "implicit de-registrations" 21 Annex 2.Analysis of Cross-Country comparisons using Entrepreneurship Database 22 Annex 3.Country Case Studies: 25 Ukraine 26 Latvia 33 Peru 34 Macedonia 36 South Africa 37 Other African countries 39 Understanding and Improving Data on Entrepreneurship and Active Companies iii June 2008 Understanding and Improving Data on Entrepreneurship and Active Companies Executive Summary Over the past several years, there has been a growing interest in reforms to improve company registries in developing and transition countries, with a major focus on legal and institutional reforms. In close association, researchers have been interested to use the databases created by company registries for purposes of statistical and economic analysis. In this regard, the new World Bank "Entrepreneurship Database" is a very well-conceived, potentially very powerful new research tool to track a critical component of different economies - the number of formal firms (or "legal entities"), in 83 countries. The objective of this paper is to examine the quality of the data in the Entrepreneurship Database (which is being used increasingly in the World Bank Group for purposes of policy analysis and monitoring and evaluation), to identify the strengths and weaknesses of the data, and offer recommendations which may help, over time, to improve the quality of the data. While other indicators have routinely been employed to measure the growth or strength of economies (most obviously GDP, GDP per capita, investment rates, employment rates, etc.), there has been growing interest in measures of "entrepreneurship" within an economy. The number of formal firms or legal entities can add a new dimension to the analysis of an economy. More firms, ceteris paribus, may be indicative of stronger competition within the economy, or enhanced prospects for division of labor.1 However, while it could be tempting to use this type of data for economic analysis and/or monitoring and evaluation of business environment reforms, without correcting for inherent problems in both the wide variety of the sources of the data and for the exit of inactive firms, it would be a serious disservice to WB/IFC credibility and the quality of monitoring and evaluation work moving forward. To accept the Entrepreneurship Database as a tool for economic analysis and for Monitoring and Evaluation of IFC "Business Enabling Environment" projects, more needs to be done to improve the underlying data. While the current data available are problematic, we do see hope for improvement, which may allow for more reliable, robust and accurate analysis and comparisons in the future. However, such improvements will take investments of time and resources. The first requirement should be to ensure consistent definitions across countries, (preferably using a harmonized business registry database design), and confine analysis to countries that 1On the other hand, larger firms in an economy (which might imply relatively fewer firms) might also allow for greater economies of scale and thus improved efficiency. Understanding and Improving Data on Entrepreneurship and Active Companies iv June 2008 are broadly similar in terms of the quality of the relevant statistical data (and perhaps also in terms of legal definitions and level of income). As a second step, we would recommend targeted technical assistance to interested client governments in developing and transition countries aimed at improving the quality of the statistical data from the company and statistical registries. This would include the following elements: · Design of a harmonized database for business registries for statistical purposes, with related registration procedures · Improvement of the linkages between company registries, statistical agencies, and tax registries, to facilitate data sharing and updating of the core register; · Clear legal definitions including which types of businesses are required to be registered and different legal categories (e.g., sole proprietors, LLCs, etc.) · A requirement for periodic updating of information from companies listed in the company registry, with those not responding being moved to a category of "inactive" (but not necessarily removed from the registry entirely); · Computerization (where practical) of Company Registries for ease of updating; · Design of simple, user-friendly forms for company registration to facilitate development and maintenance of a sound statistical database (including simpler forms for periodic updating of information). · Household surveys to estimate the size and composition of the informal sector · A "quality assessment" approach to assessing business statistics, e.g., by using area business analysis methods, or business area sampling Many of these measures could be incorporated into projects to create or upgrade Company Registries, or as a component of "statistical capacity building" to improve the quality of national statistics used for national accounts and economic analysis. Over time, such improvement in more countries, including the use of standardized definitions and templates for data on companies should yield an international database that can be used with much greater confidence for both economic analysis and for monitoring and evaluation purposes. Understanding and Improving Data on Entrepreneurship and Active Companies v June 2008 1. Introduction Over the past several years, there has been a growing interest in reforms to improve company registries in developing and transition countries, with a major focus on legal and institutional reforms. In close association, researchers have been interested to use the databases created by company registries for purposes of statistical and economic analysis. In this regard, the new World Bank "Entrepreneurship Database" is a very well- conceived, potentially very powerful new research tool to track a critical component of different economies - the number of formal firms (or "legal entities"), in 83 countries. The objective of this paper is to examine the quality of the data in the Entrepreneurship Database (which is being used increasingly in the World Bank Group for purposes of policy analysis and monitoring and evaluation), to identify the strengths and weaknesses of the data, and offer recommendations which may help, over time, to improve the quality of the data. While other indicators have routinely been employed to measure the growth or strength of economies (most obviously GDP, GDP per capita, investment rates, employment rates, etc.), there has been growing interest in measures of "entrepreneurship" within an economy. The number of formal firms or legal entities can add a new dimension to the analysis of an economy. More firms, ceteris paribus, may be indicative of stronger competition within the economy, or enhanced prospects for division of labor.2 However, while it could be tempting to use this type of data for economic analysis and/or monitoring and evaluation of business environment reforms, without correcting for inherent problems in both the wide variety of the sources of the data and for the exit of inactive firms, it would be a serious disservice to WB/IFC credibility and the quality of monitoring and evaluation work moving forward. To accept the Entrepreneurship Database as a tool for economic analysis and for Monitoring and Evaluation of IFC "Business Enabling Environment" projects, more needs to be done to improve the underlying data. The next section describes the relevant definitions and the existing literature on the subject. Chapter II discusses the existing data on entrepreneurship and its limitations. Chapter III discusses the implications of the use of data on entrepreneurship for economic analysis and monitoring and evaluation purposes. Chapter IV contains conclusions and recommendations for future efforts to improve the quality of the relevant data. Background: Definitions and Literature One of the first problems we face is the need for clear and usable definitions for terms such an "enterprise" and "formal vs. informal", and to establish the relevant methodologies for measurement. The challenge here is that legal and administrative definitions and those used in economic and statistical literature differ from one another, and also vary widely across countries. 2On the other hand, larger firms in an economy (which might imply relatively fewer firms) might also allow for greater economies of scale and thus improved efficiency. Legal definitions focus on the rights and legal status of firms, which are important preconditions for formal or legal economic activity. The economic approach is important to understand how firms' economic decision-making and operations take place. It is in this context that the production factors first described by Schumpeter ­ land, capital, labor and entrepreneurship ­ are combined in order to generate value-added. Generally, there is a separate literature focusing primarily on "formal" firms and another body of literature focusing more on "informal" firms, but even these basic terms are often unclear. The broadest definition of "formal" enterprises might be those with at least some recognition by some governmental body, at any level (e.g., a municipal administration). In a more narrow sense "informal firms" refer to production units (mostly in the non- agricultural sector) that are not registered with the relevant authorities, for a wide range of reasons. For this reason the informal sector will differ from country to country, because of different rules and systems for registration of units. This paper will have more of a focus on "formal" firms and in particular legally- recognized and registered firms, as described in recent World Bank reports. Most relevant would be Klapper, et. al., "Entrepreneurship and Firm Formation Across Countries" (2007).3 Their definition of the "unit of measurement of entrepreneurship" is: Any economic unit of the formal sector incorporated as a legal entity and registered in a public registry, which is capable, in its own right, of incurring liabilities and of engaging in economic activities and transactions with other entities. This definition is clearly one based on legal status and the fact of registration, and measurement of it does not necessarily say anything about whether the firm is economically active.4 It could be, for example a "shell company" created by a law firm, waiting for a client to put it to active use, or a company that had once existed as a meaningful economic agent, producing and selling goods or services, but later went defunct and never de-registered. This concept is more legalistic than economic in its nature. The above definition also excludes the informal sector, presumably on the grounds that useful data are unavailable on that sector, to maintain some degree of consistency in definitions that can be used for cross-country comparisons, and to ensure a focus on entrepreneurship that is generally abiding by the laws and regulations of the country within which it operates. However, if we are interested in the economic activity of firms (e.g., production and sale of goods and services, employment, etc.), then we should also be looking at the informal sector (for more on the informal sector, see below). This is especially relevant for developing countries where most units are informal and most entrepreneurship is informal. Practitioners and researchers hope to show that improving the business enabling environment helps facilitate new business formation and the formalization of informal firms, as was originally suggested in The Regulation of Entry (Djankov et al, 2002). The hope is both to measure the impact of reforms to company registration and related procedures (e.g., the Doing Business indicators for "starting a business") and to measure indirect impacts of other reforms intended to improve the business enabling environment. 3Klapper, Leora, et. Al., World Bank Policy Research Working Paper no. 4313. 4Alternative definitions can be found in Section IV and in the various Case Studies in Annex 3. Understanding and Improving Data on Entrepreneurship and Active Companies 2 June 2008 New formal firms are expected to invest more, expand output and provide new employ- ment opportunities as has been shown in Canada (Baldwin and Picot, 1995), India (Besley and Burgess, 2004), Italy (Viviano, 2006), Mexico (Bruhn, 2008) and Sweden (Davidsson et al, 1998). These increases in turn appear to lead to higher economic growth. Djankov et al (2006) use the Doing Business data to show countries in the best quartile of the business regulations grow 2.3 percentage points faster than those in the worse quartile. Growth in the number of firms is also valued as an indicator of improved competition in an economy (see Doing Business 2007). Any economy where there are large and growing numbers of new firms may be distinguished, say, from an economy that may have relatively high rates of investment and economic growth but is dominated by a relatively small number of long-established, large firms. Competition is expected not only to help keep prices lower (Bruhn, 2008), but start-ups are expected to be more innovative than older firms (Porter, 1990) not only in terms of technology (Ruttan,1997), but also in terms of marketing and management (Nickell, 1996). The literature recognizes a "demography" of firms, knowing that many new firms ("births") fail to survive, and even older firms may close, go bankrupt, or be acquired by other firms. Theoretical models of firm entry and exit were originally done by Joyanovich (1982) and Hopenhayen (1992) which emphasize the importance of incorporating firm dynamics - entry and exit - as the rate at which firms "die" measures both the competitive nature of the local market and the potential advantage incumbents may have. However, up to this point, good empirical analysis of entrepreneurship incorporating both entry and exit has been limited, especially in developing and transition countries. A first strand of empirical research on firm entry and exit has shown rapid entry rates tend to be associated with rapid exit rates (Dunne et al, 1989). Case specific research done in the US has shown that the likelihood that a firm closes declines with age as well as with industry, size and location (Nucci, 1999). The"life expectancy" of a new firm in the United States was about 4.7 years.5 For firms that were "autonomous establishments" (i.e., not subsidiaries of an established firm), the survival rates were even lower, with barely 40% surviving to an age of five years.6 Bartelsman et. al. (2004) found "post entry performance" differing significantly between Europe and the US, with Argentina resembling continental Europe more than the US. They reiterated that death among firms is, of course, not necessarily a bad thing, as "a large fraction of total factor productivity and labor productivity growth at the industry level is accounted for by the reallocation of outputs and inputs from less productive to more productive businesses." Meanwhile, research done in Europe using data on corporate activity in 33 European countries from 1997 to 1998, shows that greater fairness (i.e., a more level playing field) and greater protection of property rights are associated with increased firm entry rates, and reduced firm exit rates (Desai et al, 2003). Regarding informal firms in developing countries, the literature mostly refers to the small units that offer services or goods from households or shops on the streets without permitted building structures, or to mobile traders (e.g., who just walk the streets). Of 5Nunci, Alfred, "The Demography of Business Closings," in Small Business Economics 12: 25-39, 1999; note these were firms with a payroll, and thus excluded self-employment and most sole proprietorships. 6Ibid Understanding and Improving Data on Entrepreneurship and Active Companies 3 June 2008 course, informality may have many aspects, including non-compliance with any of a number of different formal requirements such as: · Company registration · Tax registration · Payroll registration · Municipal permits · Operating licenses The ILO considers it too difficult to find and apply agreed criteria to identify these units. They therefore for operational reasons accept the outcomes of the Delhi City Group and have agreed to consider all small units of 5 and less workers as being informal.7 If we would apply this criterion, many firms in the developed countries would be defined as being informal, which could be misleading. In developed countries, there are relatively fewer businesses which are informal in the sense of being unrecognized by any government body, although there is often still a significant problem of "informal economic activity" such as unrecorded employment or revenue (including in-kind and barter transactions, e.g., see Schneider, 2004). Unrecorded revenue is defined as a form of underreporting by enterprises for all kinds of reasons. Unrecorded employment can be defined as employees without a proper legal (written) contract. The work of Mr. Schneider (2004) may in some respects be misleading for purposes of international comparison, as his definition includes tax evasion, tax evading labor inputs, and do-it-your-self activities. 2. Data on Entrepreneurship and its limitations There are a number of different sources of data about companies in countries around the world, including company registries, tax registries, statistical registries and others, at various levels of government, following different legal traditions and different administrative procedures. The Entrepreneurship Database (ED) has made a heroic attempt to consolidate such information across over 80 countries, but it exhibits a number of problems for our efforts to compare data across countries or even within a country over time. First we can look at the data sources and we see that of the 83 countries, in about 19 of them statistical registers are used, and the rest is a combination of administrative sources and legal sources. This mix of sources suggests that the units that are registered differ in type and definition. 7Measuring the Non-Observed Economy, A Handbook, OECD, IMF, ILO, CIS Stat, 2002, Chapter 10, pages 159-177. and Hussmanns, Ralf, "Statistical definition of informal employment: Guidelines endorsed by the Seventeenth International Conference on Labour Statisticians (2003)", 7th Meeting of the Expert Group on Informal Sector Statistics, New Delhi, 2-4 February 2004 Understanding and Improving Data on Entrepreneurship and Active Companies 4 June 2008 The biggest problem however, is a fundamental inconsistency in what might be called "gross" vs. "net" company formation. If we continue with the analogy with population demographics mentioned above, some countries in the database are clearly taking into account both "births" and "deaths" in the figures for the "total population" while others appear to use "births" only. Specific examples of such inconsistencies are as follows: Using the Entrepreneurship Database data from 2002-2005, we were able to calculate the implicit number of de-registrations for each country being included in the database. We calculate the implicit de-registrations by: (TotalCorporationst - TotalCorporationt ) - (NewCorporationst ) -1 Because the Entrepreneurship Database only has four years of data, this calculation is only possible for three years (2003-2005). A number of countries in the ED (data for 2002 ­ 2005, See Annex 1) are apparently counting ONLY "new corporations" to obtain the change in each year's "total corporations", for example Botswana8: Country Year Region Total New Implicit De- Ratio Ratio Corporat registra- new/total de- ions Corpora tions reg./total tions Botswana 2002 AFR 54,611 5,262 9.6% Botswana 2003 AFR 62,385 7,774 0 12.5% 0.0% Botswana 2004 AFR 72,242 9,857 0 13.6% 0.0% Botswana 2005 AFR 79,543 7,301 0 9.2% 0.0% The same pattern can be seen for the following countries in the Entrepreneurship Database: Congo, Rep. Georgia Jordan Latvia Lithuania9 Madagascar10 Malta Morocco Sri Lanka In the case of Latvia, the Entrepreneurship Database used data from the Ministry of Justice11, rather than the Company Registry itself, which has very detailed data on both 8First 5 columns: Entrepreneuship Database, last 3 columns calculated by authors. 9Except in 2004, when there were 81 implicit de-registrations (less than 0.1% of total corporations) 10In each year there were less than 25 implicit de-registrations, (less than 0.1% of total corporations) 11Klapper, et. al., 2007, Annex 1. Understanding and Improving Data on Entrepreneurship and Active Companies 5 June 2008 entry and exit as well as other changes in legal status (for details, see Annex 2, Case Studies). By contrast, all the "developed countries" in the Entrepreneurship Database (and some developing countries) implicitly include de-registrations in their figures for "total corporations".12 A good example is Finland13: Country Year Region Total New Implicit De- Ratio Ratio Corporat registra- new/total de- ions Corpora tions reg./total tions Finland 2002 DEV 112,106 7,226 6.4% Finland 2003 DEV 112,682 7,011 6,435 6.2% 5.7% Finland 2004 DEV 112,734 7,424 7,372 6.6% 6.5% Finland 2005 DEV 114,061 7,710 6,383 6.8% 5.6% Most of the Northern and Western European countries (and the US and Japan) show relatively high rates of "deaths" (implicit de-registations), usually over 4 - 5% of "total corporations" each year (up to about 15% as in Germany and Norway; and 8 - 11% in UK/US). Similarly, we see quite stable/consistent patterns in the developed countries of "births" versus "deaths", with births usually running about 30% to 70% higher than deaths.14 Other countries' data look frankly very confusing. Algeria, for example, has some years with "total corporations" higher than the previous year's "total" plus "new corporations": Country Year Region Total New Implicit De- Ratio Ratio Corporat registra- new/total de- ions Corpora tions reg./total tions Algeria 2002 AFR 69,692 13,770 19.8% Algeria 2003 AFR 79,908 10,123 -93 12.7% -0.1% Algeria 2004 AFR 92,930 12,494 -528 13.4% -0.6% Algeria 2005 AFR 103,482 12,164 1,612 11.8% 1.6% Some of this might be due to large firms re-registering as multiple entities (to gain abetter tax status or relief from labor legislation), but it might also be evidence of poor quality data. Other countries showing higher increases in "total corporations" than can be accounted for by "new corporations" in at least one year include the following: Albania Bangladesh Bosnia 12 We note, however, there is one negative figure, i.e., an increase in the "total corporations" greater than could be accounted for by "new corporations" - in Canada for one year. 13First 5 columns: Entrepreneuship Database, last 3 columns calculated by authors. 14 Exceptions, where births are more than double the rate for deaths (for more than a single year, according to the data) include Australia, Greece, Iceland, New Zealand, and Spain (these countries also show less than 4% deaths/total corporations). This may be indicative either of flawed data or (given the short number of years for which data is available) may be driven by business cycle fluctuations. Understanding and Improving Data on Entrepreneurship and Active Companies 6 June 2008 Canada15 Ghana Iceland16 India Indonesia Kenya Senegal Serbia Slovakia Slovenia Tunisia Finally we can see that most of the developing countries (including those in the list above) show very low ratios of implicit de-registrations to "total corporations" (e.g., less than 1%), while "new corporations" are often over 8 - 15% of "total corporations". Thus, their figures for "total corporations" seem to be getting progressively more inflated every year. The resulting distribution of the ratio of implicit de-registrations to total corporations (a ratio is used to normalize the distribution) for the full database is shown in Figure 1: Figure 1: De-Registration/Total Corporations Histogram with Normal Curve, 2003-2005 60 40 ycneuqe Fr 20 0 -.4 -.2 0 .2 .4 Ratio de-reg/total Observations Mean St. Dev Min Max 213 .033 .056 -.28 .31 Source: World Bank Entrepreneurship Database 2007 While one might expect a significant positive level of firm de-registration/exit across countries, Figure 1 based on the Entrepreneurship data suggests otherwise ­ it presents a central tendency around or close to zero for the ratio of implicit de-registrations to total corporations. There is a mean ratio in the data of only 3.3 percent of total corporations de-registering per annum, suggesting a survival rate of 96.7 percent per annum. If we were to project this number out over five years it would imply that roughly 85 percent of new firms would still be in operation five years after beginning operation, a dramatically 15One year only 16One year only Understanding and Improving Data on Entrepreneurship and Active Companies 7 June 2008 higher number that Nucci found for the United States in 1999: a survival rate of about 40 percent (noted above). When we disaggregate the ratio by level of economic development, and compare the implicit de-registrations of developed countries with the low and middle income countries of the world in the Entrepreneurship data, we can see the problem probably is in the quality of the data from the business registries of the poorer countries. In Figure 2 on the left hand side we can see that for the developing and transition economies in the dataset, the mean ratio of de-registrations to total corporations is 1.8 percent implying that after five years roughly 92 percent of firms which register are still listed in the relevant registry. To assume they are all economically active seems hardly plausible, given our understanding of actual firm demography in developing and transition economies. For example, Bartelsman et. al. found: "Looking at cross-country differences in survivor rates, about 10% (Slovenia) to more than 30% (Mexico) of entering firms fail within the first two years." (Bartelsman et. al., 2004, pg. 23). Figure 2: De-Registrations/Total Corporations Distribution (Developed vs. All Others), 2003-2005 60 15 40 10 yc yc quenerF quenerF 20 5 0 0 -.4 -.2 0 .2 .4 -.4 -.2 0 .2 .4 Ratio de-reg/total Ratio de-reg/total Observations Mean St. Dev Min Max Observations Mean St. Dev Min Max 140 .018 .058 -.28 .31 73 .062 .04 -.01 .16 Source: World Bank Entrepreneurship Database 2007 Supporting this conjecture, Figure 2 on the right hand side also shows the ratio of de- registration to total corporations for the 73 observations in the developed countries. What we see is that the distribution is centered well to the right of zero ­ very logical since we should expect a natural rate of firms exiting - and the mean ratio is 6.2 percent implying a survival rate of 93.8 percent per annum or that up to 70 percent of new firms survive for five years or longer (which still may be a reflection of a lag in the recording of "exit" data, and/or the up-side of a business cycle). A clear conclusion from this analysis is that in the current entrepreneurship database there is very little implied de-registration in most developing and transition economies, suggesting that either de-registrations are either (i) not being counted or (ii) miscounted. Why is this happening? An example might be the experience of Madagascar. In the case of Madagascar, the INSTAT statistical registry (the main company registry for the country) had a total of just over 500,000 "created" firms as of 2007, and a little over 16,000 de-registrations ("annulations uniquement"). After requiring legal re-registration Understanding and Improving Data on Entrepreneurship and Active Companies 8 June 2008 that year, and removing firms that failed to register or to object to being removed from the registry, the total number of firms was reduced to less than 104,000. Even for Societe a Responsabilite Limetee (S.A.R.L., roughly equivalent to limited liability companies), the totals fell from 19, 938 to 6226.17 Thus before the re-registration exercise, it is clear the data for corporations in Madagascar were polluted with corpses. Correcting for the exit of formal or legally-recognized firms in the data of most developing and transition countries in the ED would thus probably reduce the figures of "total firms" and all related calculations (e.g., "density"). On the other hand, these countries are, by most accounts, missing data regarding the informal sector, although there is no reason to believe that the overstatement of formal firms in the data is balanced by informal sector firms. For this reason, cross-country comparisons using "total corporations" (aside perhaps from those confined only to data from developed countries) are clearly invalid. The World Bank "Entrepreneurship and Firm Formation" paper notes explicitly that: ... although approximately 80% of surveyed countries require businesses to report closures, a significantly lower number were actually able to report the number of closed businesses mainly due to the fact that the registrars generally have no enforcement mechanisms to obligate businesses to report closures. ...18 What does this imply? A direct conclusion from the above analysis is that researchers and WB/IFC should be extremely cautious to use the ED "total company" data for any sort of analytical work of the business environment in developing and transition countries until the inherent limitations in the data are addressed. Not only would it not be right to infer causality from doing cross-country correlations as was suggested by Klapper et. al "Although we find significant relationships with these measures ­ i.e. more dynamic economies in countries with better business environments ­ we cannot postulate on the direction of causality."19 In addition, we also need be cautious about the results of the original theoretical suggestions from the cross-country correlations (for examples, pls see Annex 2, below). In fact, the data on the developed countries in the Entrepreneurship Database also shows anomalies. The number of units on the Netherlands is clearly not about the active units. That number is about 20% less.20 Germany appears to have half of the units registered (in terms of "density" or the ratio of total companies to working age population) compared with the Netherlands. The number of units per 1000 inhabitants of the US is half that of Russia and Serbia, with the implication that the entrepreneurship in the US is half that of Russia or Serbia. Changes in tax treatment of natural vs. juridical persons can drive "company formation" in the form of employees leaving an employer, forming a company, and returning to their former employer as a full time "contractor." 17 Madagascar INSTAT printout out; the data in the ED for Madagascar appears to be those for SARL, reporting 19,305 "Total Corporations" in 2005. 18Klapper, et. al., op. cite., pp 13 ­ 14. 19Klapper et. Al, p. 18 20See Annex 3, Case Studies Understanding and Improving Data on Entrepreneurship and Active Companies 9 June 2008 Some case studies can shed more light on data availability and data quality for Ukraine, Latvia, Macedonia, Peru, South Africa, and several African countries (See Annex 3, below). While it might be possible to argue that firm survival rates might be higher on average in developing countries than developed countries (e.g., if cumbersome bureaucratic requirements or high costs deter all but the "strongest" entrepreneurs from starting formal firms and therefore they are more likely to survive) the examination of the data from the case studies (and the Madagascar example cited above) suggests otherwise. Ideally, we would like to look at detailed and accurate entry and exit data to infer "age specific mortality rates" over a full business cycle before drawing any firm conclusions. 3. Implications for use of data on entrepreneurship Analysis of the Entrepeneurship Database and alternative sources of data suggests that while the data on "legal entities" for most developed countries appears reasonably sound, the data for the majority of developing and transition countries shows evidence of an upward bias, due primarily to the failure to de-register firms that have exited the market and effectively ceased to exist. If the data for developing and transition countries were corrected, "total corporations" would probably be significantly lower in most cases. By extension, the estimated "density" of firms (i.e., ratio of total corporations to working-age population) would probably be lower in most developing and transition countries because the numerator should be lower, while the "rate of new company formation" (i.e., the ratio of new corporations over total corporations) might be higher because the denominator should be lower. The direction of the bias in the numbers of legal entities (as opposed to active entities) is quite clear: most data on "total corporations" from developing and transition countries is overstated. The next question one might ask is what is the magnitude of the bias? This is much more difficult to assess. Data from a few middle-income countries (e.g., Latvia, Peru and S. Africa) suggests that data from company registries might be roughly double the number of active legal entities. However, as suggested in the previous section, the experience of Madagascar (a low-income country) suggests the magnitude of the bias might be a function of the age of the registry (or number of years since a serious effort to require re-registration of active firms). In the case of Madagascar, the overall number of firms in the INSTAT database fell by almost four/fifths after its re-registration exercise and even the number of limited liability companies (presumably more solidly established than sole proprietorships) fell by over two/thirds. However, if we are interested primarily in the economic activity of firms (as opposed to legal entities), we should also be interested in estimates of the number of informal firms, but estimates of the size and composition of the informal sectors of developing and transition countries faces even greater challenges than those plaguing company registries. Understanding and Improving Data on Entrepreneurship and Active Companies 10 June 2008 Another serious concern is the use of the Entrepreneurship Database for purposes of monitoring and evaluation. While the number of "new registrations" may be reasonably reliable on its own, any use of "total numbers of firms" is plagued by the problem of inflated numbers due to the accumulation of "dead firms" in the registry over time. Thus both the "density of firms" and the "rate of new firm creation" yield serious data quality problems. For example, suppose someone tried to monitor the impact of reforms in Madagascar over the period 2006 ­ 2008: While the numbers of newly registered firms might be useful, any effort to use the total number of firms, the "density of firms" or the "rate of new company formation" would find (even if the true situation were improving) a major decline in the data from almost 20,000 LLCs to just over 6,000, because the company registry went through an exercise that purged dead firms from the data. The box below about Ukraine provides another example. In Ukraine, regulatory barriers to exit have resulted in a considerable number of businesses, which are not active, but are still in the business registrar. This backlog of inactive enterprises in Ukraine (40% of all registered enterprises) thus hampers accurate tracking of business entry and growth rates and means that officially reported figures do not adequately reflect the true entrepreneurship picture. This means that the World Bank Group Entrepreneurship Database (WBG ED), while being the most comprehensive cross country firm entry dataset, provides information that does not accurately reflect business development in Ukraine. Not filtering out the "inactive but still registered" businesses means that the overall number of enterprises, and their density per 1,000 population, is overestimated. This is compounded by differences in definitions of enterprise that further contribute to make entrepreneurship in Ukraine appear more developed than it really is. At the same time, it means that the growth rate is substantially underestimated, making Ukrainian business look less dynamic than it is. IFC Ukraine BEE project estimates that the actual development of business in Ukraine in 2005-2006 was as follows: · Average entry rate was 7.1% ­ vs. 5.9% according to WBG ED · Average annual growth rate in active enterprises was 7.1% ­ vs. 4.3% according to WBG ED · The 1-year survival rate for Ukraine was 91% and the 2-year survival rate was 81%. Source: Annex 3 How would the data appear in an evaluation of a project? The actual impact of reforms to improve the investment climate may be positive, but if data have recently been corrected and purged of "dead firms", the indicator may appear negative. Comparing a project in Madagascar with a similar project in a country that did not improve its statistics would be misleading. And worst of all, the temptation might be for donor-financed projects to avoid improving company statistics (or even discourage it or lobby for a "low priority" for such an exercise) because it might not look as good as the situation with flawed data. A project that reduced barriers to entry might encourage new business formation and/or the formalization of informal firms, but we need to understand the longer-term dynamics: whether the new firms are as viable, on average, as those that formed earlier; whether new firms are displacing the market share of older firms, and ultimately whether the hoped-for impacts of increased value-added, employment opportunities and greater consumer choice Understanding and Improving Data on Entrepreneurship and Active Companies 11 June 2008 are being realized. An IFC project in Lima, Peru, which reduced entry barriers, clearly led to an increase in numbers of new firms registering, but the project is using "tracer studies" of the long-term impact on a representative sample of the new firms in order to gain a better understanding of such dynamics. For Latvia, someone trying to show more impressive results might be tempted to use the data in the Enterprise Database (based on data from the Ministry of Justice and excluding any "exits") because it shows higher numbers of "total corporations" than the more accurate data from the Company Registry and the Statistical Registry. Of course, as noted above for Ukraine, the "entry rate" using correct data would be higher than figures derived from the Entrepreneurship Database. Thus, robust M&E efforts should include an effort to obtain accurate data on active companies, or to correct for the distortions caused by the inclusion of inactive firms in the data. 4. Conclusions and Recommendations The idea of a WB database of enterprises is very useful and the motivations for creating one are clear. However, it is also apparent from inspection that the current data available in the ED are problematic. The ED needs improvement, to allow for more reliable, robust and accurate analysis and comparisons in the future. Moreover, to realize useful improvements in the ED, it will take necessary investments of time and resources. One idea considered was to obtain data on numbers of formal, "active" companies from tax registries, in cooperation with the IMF and OECD. Tax authorities should be able to see which companies have been filing regularly (including those reporting positive income, expenses, or payroll activity). Unfortunately, experience has shown that the tax authorities in most lower-income countries are not yet inclined to cooperate. While the tax authorities in Latvia routinely share their data with the national statistical agency, those in Macedonia and South Africa have barely begun the process of cooperation on company statistics, and the relationship between the tax authorities and the statistical agency in Peru has reportedly been inconsistent over the past few years. Thus, an important first step to improve ED will be to encourage and support building these critical relationships over time. When the tax authorities are not willing to cooperate other national institutions may be considered. Also more research should be done for the proper understanding of the reasons of this lack of cooperation. Another important step in the process to improve ED data quality should be to ensure consistent definitions across countries, (the only real solution would be to use a harmonized business registry database design), and confine analysis to countries that are broadly similar in terms of the quality of the relevant statistical data (and perhaps also in terms of legal status and size). As an example, EU regulations require harmonization of "statistical business registers" in large part to ensure a standard level of data quality "particularly as regards comparability."21 Finally, we would recommend targeted technical assistance to interested client governments in developing and transition countries aimed at improving the quality of the 21Eurostat/OECD "Manual on Business Demography Statistics", 2007Ibid., pg. 11. Understanding and Improving Data on Entrepreneurship and Active Companies 12 June 2008 statistical data from the company and statistical registries. This would include the following elements: · Clear legal definitions including which types of businesses are required to be registered and different legal categories (e.g., sole proprietors, LLCs, etc.) · Design of a harmonized database for business registries for statistical purposes · Improving the linkages between company registries, statistical agencies, and tax registries, to facilitate data sharing and updating of the core register; · A requirement for periodic updating of information from companies listed in the company registry, with those not responding being moved to a category of "inactive" (but not necessarily removed from the registry entirely); · Encouraging computerization of Company Registries for ease of updating; · Design of forms for company registration to facilitate development and maintenance of a sound statistical database (including simpler forms for periodic updating of information). Definitions The difference in definitions are important for statisticians and economists, as well as for lawyers, as they affect the way we measure entrepreneurship, and can impact our ability to compare entrepreneurship and business activity across countries. Below is brief discussion of some of the definitional challenges the ED faces and some potential ways forward. In discussions about the definitions, the core question is; what is a "business" or an "enterprise"?22 This may be a complicated question in the face of informality (common in developing countries) and various complex "group holdings" (more typically found in developed countries). An enterprise can be defined from the legal perspective as a legal unit, from the administrative perspective as a registered unit, and from the statistical perspective as a statistical unit. Legal units are defined on the basis of an existing legal classification (types) of units (e.g., "sole proprietor" or "Limited Liability Company"), and can be active or not active. Administrative units can be defined on the basis of the existing registration in a country, or a combination of registrations, provided that double entries are eliminated; they can also be active or not active. Statistical units can be defined according to the types of statistical unit: the enterprise, the kind of activity unit and the local unit; they can also be classified as "active" or "inactive" and units can move back and forth between these two categories depending on the definitions used (e.g., providing financial statements or filing for taxes on a stipulated, regular basis). Eurostat offers the following definition for a "statistical unit": 22"Enterprises" are commonly understood as production units. Production units are all those units that produce good and/or services for a market (and may also produce market equivalents for own consumption). Production units that only produce for own consumption or that do not enter the market are generally not thought of as enterprises. Understanding and Improving Data on Entrepreneurship and Active Companies 13 June 2008 The enterprise is the smallest combination of legal units that is an organizational unit producing goods or services, which benefits from a certain degree of autonomy in decision-making, especially for the allocation of its current resources. An enterprise carries out one or more activities at one or more locations. An enterprise may be a sole legal unit."23 Eurostat has guidelines on the use of primary sources, including tax registers, "compulsory registration systems (e.g., for limited liability businesses or those quoted on stock markets), social security sources and other public or private sector data holdings." They note that duplication and mapping of different data sources referring to the same enterprise need to be synthesized.24 They define the population of "active enterprises ... [as] all enterprises that had either turnover or employment at any time during the reference period."25 An example of good practice is Latvia, where the statistical registry draws data from the company registry, the tax registry, and the employment registry (with all firms using a "unique number" for all government interactions and reporting requirements). If a firm has not filed for taxes, or files showing no turnover and no employment for two years, it is moved to a category called "inactive." South Africa on the other hand provides an example of the reality of most low-middle income countries where, the company registry is computerized, but not yet fully harmonized with the statistical and tax registries. It's neighbor, Namibia has a requirement for updating company information in order to stay fully registered, but has not yet computerized its database so has not yet been able to update it. Both cases show examples of why it is difficult to know if a firm is active or not in many low-middle income countries. Eurostat is also very specific about definition whether an enterprise is active or not. They define the population of "active enterprises ... [as] all enterprises that had either turnover or employment at any time during the reference period."26 This distinction is critically important if we are using data for economic analysis. If one wants to understand the actual economic and entrepreneurial activity in a country it is relevant to know how many of the registered units are active (versus, e.g. a shell company or a tax vehicle or a dead firm). This point was made clear by Bartelsman et. Al (2004), "... in all countries net entry (entry minus exit) is far less important than the gross flows of entry and exit that generate it."27 In addition, a good statistical registry may further categorize and define the units in its database, and define whether the units are active or not. For example, if one wants to understand the entrepreneurial activity in a more limited sense one can try to understand how many units have declared their interest in becoming active. Or the registry could make a distinction between units that work in order to create profits and units that only 23Eurostat/OECD "Manual on Business Demography Statistics", 2007, pg. 8 24Ibid. 25Ibid., pg. 14. 26Ibid., pg. 14. 27Bartelsman, et. al., 2004, pg. 15. Understanding and Improving Data on Entrepreneurship and Active Companies 14 June 2008 want to have their costs covered.28 This can be done by combining different sources of information (e.g., results of surveys and information from tax registers.) Apart from the issue of whether enterprises may be active or not, there is also another aspect that needs to be reflected: A unit can be registered or not. In this case we speak about non-registered units. In most definitions these units are considered to be part of the so called "informal sector". This issue is complicated because there is considerable variation on the definition of an informal sector unit and how to measure it. According to the ILO definitions the informal sector is not defined by their level of registration but by their level of size and a series of specific characteristics, and by the fact whether they are registered or not. For example, units of 5 and less working persons are considered as informal sector units, or units with some relevant characteristics like operating from a household premise, not being registered and/or not keeping files.29 Many developing countries (typically with large informal sectors measured as a percent of GDP) have correspondingly large numbers of "informal sector production units." These are still unfortunately defined in different ways in different countries and by different researchers. This shows that there is no firm agreement about the definition of the concept of the informal sector. If it is defined, e.g., as the total of registered small units (e.g., those under five employees), it would mean that the "informal sector" could be a combination of registered small units and non registered units. If the criterion of registration is used, the "informal sector" could be defined as the total of not registered units. In both cases, these informal units could be considered at any point as being active or not. Because of the lack of agreed international definitions of the informal and the formal sector, it is important that each publication that uses these concepts should define them clearly, to explain how they are used and should be understood. This becomes even more important when we notice that some authors on the topic make estimations of the informal sector that include a wide range of activities, from tax avoidance (including, e.g., income tax of people with a job), to do-it-yourself-activities (Schneider 2004). Because of the wide discrepancies over definitions of informality and how to measure it in business statistics (especially in low and middle income countries) this issue needs to be addressed clearly when the topic is dealt with. In addition, countries may exclude certain activities from some of their formal registration, like certain agricultural activities and/or financial activities. In some countries, legal units are units with a legal status distinct from their owners (who may be physical persons). When units do not have a legal status they are considered to be natural persons. The difference between the two is usually that legal units have some form of limited liability and the natural persons do not. Thus, for practical reasons it is here proposed that all enterprise units that are not registered with at least one of the eligible formal registries in a country30, should be considered as being an informal sector unit. 28In the former we speak about profit oriented units (businesses), and in the latter about subsidized, non- profit or government financed organizations. 29ILO, op cite. 30E.g., Company Registry, tax registry, statistical registry, labor registry Understanding and Improving Data on Entrepreneurship and Active Companies 15 June 2008 Standard model for the registration of enterprises for statistical purposes EU regulations require harmonization of "statistical business registers" in large part to ensure a standard level of data quality "particularly as regards comparability."31 In this paper it is proposed to agree on a standard registration procedure for statistical purposes. The table below is a proposal for the agreement on a standard model for the registration of enterprises for statistical purposes. If such a proposal would be accepted we would be able to start with the harmonization of enterprises registration for all countries that would accept the idea. This would improve the information at the national levels and it would vastly increase the comparability of information of enterprises across countries. This approach would allow an understanding of the relationship between the national administrative registrations and the statistical registrations of enterprises. Table 4.1 below presents information on the main characteristics that can to be registered in a national database of formal enterprises. The table shows five columns. The first two represent the administrative units and the last three the statistical units. Administrative registers legal units as types, firms with limited liabilities, own account workers and others, and the addresses of these units. Statistical units used by statistical registers for statistical purposes are: · the enterprise unit: that is the unit responsible for the financial information and/or the financial management of business activity (note: some "enterprises" may have several "establishment" units, e.g., large conglomerates) · The concept of "establishment" is used to define a unit for which information is collected on the production process (output, costs and revenues), which may be a branch or subsidiary of a larger group or may be equal to an independent enterprise unit. (Usually the "enterprise" is the same as the "establishment unit" but for large conglomerates, the "enterprise" may have several "establishments"). In the European Union the concept of Kind of Activity (KAU) is used to obtain information from production units, about production, costs and revenues. · "Local units" are used to present information on the various different geographical locations of the enterprise Here is a list of suggested information that should be gathered in a good business registry based on the good practice of Eurostat: · Source key ­ the unique number for that unit with the government administration · General business registration number ­ unique number for each unit with the business register · Legal status ­ legal status of the person registering · SNA-sector code ­ which SNA sector that the unit belongs to · Activity code (ISIC) ­ · Size class ­ number of working persons (and / or turnover) · Change code: if change has occurred, which kind of change. · Survey code: for which kinds of surveys the unit is approached · Remaining codes: are self-evident. 31Ibid., pg. 11. Understanding and Improving Data on Entrepreneurship and Active Companies 16 June 2008 Table 4.1 Characteristics of a business register data base for statistical purposes. Legal units Local part of a Enterprise Establishment Local unit legal unit Source key X X General X Business Register number name /address X X X X X Legal status X SNA-sector X code Activity code X X X X X (ISIC) Size class X X X X X (code) Change (code) (X) Geo code X Survey code X X Date of entry X X Date of "birth" X X Date of "death" X X Date of de- X X register It would be useful to aim at obtaining this level of information, or part of it, for each registered enterprise in a business registry. The best time to obtain this information should be during the registration procedure. However, if there is no form of a formal registration, this information should be obtained by field visits (area approaches). Parts of the city can be visited by staff of governmental institutions (e.g., statistical agency). Also by using a GIS system, units can be identified and registered.32 But when a proper registration procedure is developed this information can be considered for inclusion in the registration form. A well-structured register can also record over time change of ownership, mergers/acquisitions, break-ups, split-offs, and/or creation/cessation of a joint-venture, as well as changes in principal activity, size and of main location.33 Informal Sector Finally, it would be desirable to get a better understanding of the size and importance of the informal sector in each country by applying household surveys (e.g., Bartelsman et al, 2004 and Bruhn, 2008). This is of particular importance in all of the developing countries (like IDA) and most of the middle income countries. The household survey is 32This information is normally of a lesser quality, however still better than no information at all. 33Eurostat manual, pp 18 ­ 19, with complete definitions on pp. 22 ­ 23. Understanding and Improving Data on Entrepreneurship and Active Companies 17 June 2008 the ideal tool to obtain a general impression of the number and size of these informal units. Such a tool, as part of the Labor Force Survey, has now been developed for Africa as part of the IMF/WB GDDS II project for Anglophone African countries by the DECDG division of the World Bank and by Statistics Canada. By creating a number of well-defined questions in a survey, the size and structure of the informal sector can be described in statistical terms. (The design of these questions has been completed and is made available.) In addition it is possible to target specific parts of cities, like public markets, to come up with estimates about the number of units that are active in trade on those markets, how many staff they have and the kind of trade they are involved in. * * * * * * * Ultimately, we believe it is both possible and desirable to help client governments invest in improving their data on active companies. The World Bank group has access to the resources and expertise both to improve the data derived from company registries (regarding formal firms, probably best in conjunction with projects to create or improve company registries regarding their legal and administrative bases) and to design and implement household surveys to learn the size and characteristics of the informal sector. Over time, such improvements, including the use of standardized definitions and templates for data on companies should yield an international database that can be used with much greater confidence for both economic analysis and for monitoring and evaluation purposes. Understanding and Improving Data on Entrepreneurship and Active Companies 18 June 2008 References Baldwin, J and Picot, G, "Employment generation by small producers in the Canadian manufacturing sector", Small Business Economics, 1995 Bartelsman, E., J. Haltiwanger and S. Scarpetta, "Microeconomic Evidence of Creative Destruction in Industrial and Developing Countries", 2004 Besley, T and Burgess, R, "Can Labor Regulation Hinder Economic Performance? Evidence from India", H Stree Technology, MIT Press, 2004 Davidsson et al, "Where do they come from? Prevalence and characteristics of nascent entrepreneurs", Entrepreneurship and Regional Development, Vol 12, Num 1, 2000 Desai et al, "Institutions, Capital Constraints and Entrepreneurial Firm Dynamics Evidence from Europe », NBER Working Paper No 10165, 2003 Djankov et al, "The Regulation of Entry", Quarterly Journal of Economics 117(1), February 2002, p. 1-37 Djankov, Simeon, Caralee McLiesh, and Rita Ramalho, "Regulation and Growth" 2006 http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V84-4KGG5RP- 5&_user=2052538&_rdoc=1&_fmt=&_orig=search&_sort=d&view=c&_acct=C0000553 00&_version=1&_urlVersion=0&_userid=2052538&md5=07fd5324bc587ae5c50d76d7b 1f7e7b1 Dunne et al, "The Growth and Failure of US Manufacturing Plants", The Quarterly Journal of Economics, 1989 Eifert, Benn, "The Economic Response to Regulatory Reform, 2003-06", University of California at Berkeley Work Paper, 2007 Hopenhayn, H, "Entry, Exit and Firm Dynamics in Long Run Equilibrium", Econometria, 60(5), 1992 Hussmanns, Ralf, "Statistical definition of informal employment: Guidelines endorsed by the Seventeenth International Conference on Labour Statisticians (2003)", 7th Meeting of the Expert Group on Informal Sector Statistics, New Delhi, 2-4 February 2004 ILO, Measuring the Non-Observed Economy, A Handbook, OECD, IMF, ILO, CIS Stat, 2002, Chapter 10, pages 159-177. and Joyanovic, B, "Selection and the Evolution of Industry", Econometria, 50(3), 1982 Klapper, Leora, et. Al., Entrepreneurship and Firm Formation Across Countries" (2007) World Bank Policy Research Working Paper no. 4313. Nickell, S, "Competition and Corporate Performance", Journal of Political Economy, 1996 Understanding and Improving Data on Entrepreneurship and Active Companies 19 June 2008 Nunci, Alfred, "The Demography of Business Closings," in Small Business Economics 12: 25-39, 1999 Porter, Michael, The Competitive Advantage of Nations, Harvard Business School Press, 1990 Ruttan, VW, "Induced Innovation, Evolutionary Theory and Path Dependence: Sources of Technical Change", The Economic Journal, 1997 Schneider, Friedrich, and Robert Klinglmair, "Shadow Economics around the World: What Do We Know?" 2004, Institute for the Study of Labor Discussion Paper IZA DC no. 1043. Schumpeter, Joseph. (1934/1912). The Theory of Economic Development. Cambridge, Mass.: Harvard University Press Viviano, Eliana, "Entery Regulations and labour market outcomes: Evidence from the Italian retail trade sector", Economic Research Department Working Paper, Bank of Italy, 2006 World Bank, Doing Business 2006 Understanding and Improving Data on Entrepreneurship and Active Companies 20 June 2008 Annex 1 Entrepreneurship Database including "implicit de-registrations" [Excel file: Entrep with calculated deregistrations.xls] Understanding and Improving Data on Entrepreneurship and Active Companies 21 June 2008 Entrepreneurship Database New Implicit Ratio Total De- new/de- Country Year Region Corporatio Corporatioregistra- Ratio Ratio de- registra- ns ns tions new/total reg/total tions Albania 2002 ECA 10,902 821 7.5% Albania 2003 ECA 16,553 952 -4,699 5.8% -28.4% Albania 2004 ECA 13,687 1,421 4,287 10.4% 31.3% Albania 2005 ECA 16,423 2,388 -348 14.5% -2.1% Algeria 2002 AFR 69,692 13,770 19.8% Algeria 2003 AFR 79,908 10,123 -93 12.7% -0.1% Algeria 2004 AFR 92,930 12,494 -528 13.4% -0.6% Algeria 2005 AFR 103,482 12,164 1,612 11.8% 1.6% Argentina 2002 LAC 343,848 21,000 6.1% Argentina 2003 LAC 359,192 40,000 24,656 11.1% 6.9% Argentina 2004 LAC 404,452 60,000 14,740 14.8% 3.6% Argentina 2005 LAC 450,535 53,000 6,917 11.8% 1.5% Armenia 2002 ECA 109,892 8,402 7.6% Armenia 2003 ECA 113,486 8,499 4,905 7.5% 4.3% Armenia 2004 ECA 118,596 9,089 3,979 7.7% 3.4% Armenia 2005 ECA 123,951 9,667 4,312 7.8% 3.5% Australia 2002 DEV 814,022 78,320 9.6% Australia 2003 DEV 870,963 75,946 19,005 8.7% 2.2% 299.6% Australia 2004 DEV 935,047 81,079 16,995 8.7% 1.8% 377.1% Australia 2005 DEV . . Austria 2002 DEV 158,590 12,564 7.9% Austria 2003 DEV 161,732 12,504 9,362 7.7% 5.8% 33.6% Austria 2004 DEV 166,487 13,476 8,721 8.1% 5.2% 54.5% Austria 2005 DEV 172,602 14,669 8,554 8.5% 5.0% 71.5% Bangladesh 2002 Asia 52,030 2,986 5.7% Bangladesh 2003 Asia 55,933 3,906 3 7.0% 0.0% Bangladesh 2004 Asia 59,975 4,042 0 6.7% 0.0% Bangladesh 2005 Asia 67,459 5,328 -2,156 7.9% -3.2% Belgium 2002 DEV 308,989 20,662 6.7% Belgium 2003 DEV 317,981 22,105 13,113 7.0% 4.1% 68.6% Belgium 2004 DEV 328,817 25,143 14,307 7.6% 4.4% 75.7% Belgium 2005 DEV 343,761 25,492 10,548 7.4% 3.1% 141.7% Bolivia 2002 LAC 20,149 1,836 9.1% Bolivia 2003 LAC 21,632 1,551 68 7.2% 0.3% Bolivia 2004 LAC 23,084 1,524 72 6.6% 0.3% Bolivia 2005 LAC 24,649 1,625 60 6.6% 0.2% Bosnia and Herzegovina 2002 ECA . . Bosnia and Herzegovina 2003 ECA 27,775 1,481 5.3% Bosnia and Herzegovina 2004 ECA 31,145 1,481 -1,889 4.8% -6.1% Bosnia and Herzegovina 2005 ECA 34,035 1,409 -1,481 4.1% -4.4% Botswana 2002 AFR 54,611 5,262 9.6% Botswana 2003 AFR 62,385 7,774 0 12.5% 0.0% Botswana 2004 AFR 72,242 9,857 0 13.6% 0.0% Botswana 2005 AFR 79,543 7,301 0 9.2% 0.0% Canada 2002 DEV 1,343,806 84,767 6.3% Canada 2003 DEV 1,378,405 79,892 45,293 5.8% 3.3% 76.4% Canada 2004 DEV 1,466,554 86,464 -1,685 5.9% -0.1% Canada 2005 DEV 1,357,881 85,083 193,756 6.3% 14.3% -56.1% Chile 2002 LAC 163,466 27,141 16.6% Chile 2003 LAC 167,800 28,123 23,789 16.8% 14.2% Chile 2004 LAC 171,497 29,824 26,127 17.4% 15.2% Chile 2005 LAC 170,636 31,088 31,949 18.2% 18.7% Colombia 2002 LAC 17,111 1,021 6.0% Colombia 2003 LAC 18,034 1,093 170 6.1% 0.9% Colombia 2004 LAC 19,092 1,182 124 6.2% 0.6% Colombia 2005 LAC 20,026 987 53 4.9% 0.3% Congo, Rep. 2002 AFR 28,083 2,441 8.7% Congo, Rep. 2003 AFR 30,316 2,233 0 7.4% 0.0% Congo, Rep. 2004 AFR 32,354 2,038 0 6.3% 0.0% Congo, Rep. 2005 AFR 34,514 2,160 0 6.3% 0.0% Costa Rica 2002 LAC . . Costa Rica 2003 LAC . . Costa Rica 2004 LAC 348,622 36,084 10.4% Costa Rica 2005 LAC 392,726 44,301 197 11.3% 0.1% Croatia 2002 ECA 97,491 5,354 5.5% Croatia 2003 ECA 101,939 6,756 2,308 6.6% 2.3% Croatia 2004 ECA 106,923 7,311 2,327 6.8% 2.2% Croatia 2005 ECA 113,708 8,733 1,948 7.7% 1.7% Cyprus 2002 ECA 109,061 8,496 7.8% Cyprus 2003 ECA 115,744 9,080 2,397 7.8% 2.1% Cyprus 2004 ECA 125,361 11,586 1,969 9.2% 1.6% Cyprus 2005 ECA 137,636 14,494 2,219 10.5% 1.6% Czech Republic 2002 ECA 235,721 . Czech Republic 2003 ECA 248,107 . Czech Republic 2004 ECA 260,940 . Czech Republic 2005 ECA 273,688 30,945 18,197 11.3% 6.6% Denmark 2002 DEV 194,425 15,837 8.1% Denmark 2003 DEV 202,008 17,526 9,943 8.7% 4.9% 76.3% Denmark 2004 DEV 211,871 21,263 11,400 10.0% 5.4% 86.5% Denmark 2005 DEV 234,432 33,047 10,486 14.1% 4.5% 215.2% Egypt 2002 AFR . . Egypt 2003 AFR . . Egypt 2004 AFR . . Egypt 2005 AFR 367,559 9,595 2.6% 0.0% El Salvador 2002 LAC . 1,801 El Salvador 2003 LAC . 1,328 El Salvador 2004 LAC 40,739 1,549 3.8% 0.0% El Salvador 2005 LAC . 2,617 Estonia 2002 ECA 58,371 6,471 11.1% Estonia 2003 ECA 63,056 6,813 2,128 10.8% 3.4% Estonia 2004 ECA 67,852 8,204 3,408 12.1% 5.0% Estonia 2005 ECA 73,999 9,945 3,798 13.4% 5.1% Finland 2002 DEV 112,106 7,226 6.4% Finland 2003 DEV 112,682 7,011 6,435 6.2% 5.7% 9.0% Finland 2004 DEV 112,734 7,424 7,372 6.6% 6.5% 0.7% Finland 2005 DEV 114,061 7,710 6,383 6.8% 5.6% 20.8% France 2002 DEV 1,102,943 110,782 10.0% France 2003 DEV 1,133,955 123,765 92,753 10.9% 8.2% 33.4% France 2004 DEV 1,182,941 142,625 93,639 12.1% 7.9% 52.3% France 2005 DEV 1,225,291 144,521 102,171 11.8% 8.3% 41.5% Georgia 2002 ECA 45,011 2,771 6.2% Georgia 2003 ECA 47,676 2,666 1 5.6% 0.0% Georgia 2004 ECA 51,805 4,129 0 8.0% 0.0% Georgia 2005 ECA 56,840 5,035 0 8.9% 0.0% Germany 2002 DEV 465,704 85,914 18.4% Germany 2003 DEV 464,172 71,259 72,791 15.4% 15.7% -2.1% Germany 2004 DEV 465,615 69,744 68,301 15.0% 14.7% 2.1% Germany 2005 DEV . 66,747 Ghana 2002 AFR 93,982 5,576 5.9% Ghana 2003 AFR 100,272 6,189 -101 6.2% -0.1% Ghana 2004 AFR . . Ghana 2005 AFR . . Greece 2002 DEV 29,941 2,315 7.7% Greece 2003 DEV 31,251 2,309 999 7.4% 3.2% 131.1% Greece 2004 DEV 32,356 2,151 1,046 6.6% 3.2% 105.6% Greece 2005 DEV 33,839 2,381 898 7.0% 2.7% 165.1% Guatemala 2002 LAC . 3,479 Guatemala 2003 LAC . 3,773 Guatemala 2004 LAC . 4,193 Guatemala 2005 LAC 68,451 4,251 6.2% Haiti 2002 LAC 385 2 0.5% Haiti 2003 LAC 337 7 55 2.1% 16.3% Haiti 2004 LAC 339 9 7 2.7% 2.1% Haiti 2005 LAC 300 9 48 3.0% 16.0% Hong Kong, China 2002 DEV 510,114 47,363 9.3% Hong Kong, China 2003 DEV 504,689 50,900 56,325 10.1% 11.2% -9.6% Hong Kong, China 2004 DEV 526,557 66,439 44,571 12.6% 8.5% 49.1% Hong Kong, China 2005 DEV 557,002 74,122 43,677 13.3% 7.8% 69.7% Hungary 2002 ECA 198,427 19,931 10.0% Hungary 2003 ECA 209,641 19,854 8,640 9.5% 4.1% Hungary 2004 ECA 226,143 24,301 7,799 10.7% 3.4% Hungary 2005 ECA 240,556 22,251 7,838 9.2% 3.3% Iceland 2002 DEV 18,519 3,120 16.8% Iceland 2003 DEV 20,160 2,389 748 11.9% 3.7% 219.4% Iceland 2004 DEV 22,062 2,517 615 11.4% 2.8% 309.3% Iceland 2005 DEV 25,223 2,938 -223 11.6% -0.9% India 2002 Asia 605,768 22,727 3.8% India 2003 Asia 636,461 28,024 -2,669 4.4% -0.4% India 2004 Asia 661,371 36,859 11,949 5.6% 1.8% India 2005 Asia 712,800 38,129 -13,300 5.3% -1.9% Indonesia 2002 Asia 232,771 6,901 3.0% Indonesia 2003 Asia 232,507 7,266 7,530 3.1% 3.2% Indonesia 2004 Asia 232,243 4,481 4,745 1.9% 2.0% Indonesia 2005 Asia 259,799 19,851 -7,705 7.6% -3.0% Ireland 2002 DEV 154,242 13,814 9.0% Ireland 2003 DEV 148,303 14,347 20,286 9.7% 13.7% -29.3% Ireland 2004 DEV 157,502 15,592 6,393 9.9% 4.1% 143.9% Ireland 2005 DEV 160,707 17,234 14,029 10.7% 8.7% 22.8% Israel 2002 DEV 340,850 17,197 5.0% Israel 2003 DEV 354,387 14,938 1,401 4.2% 0.4% Israel 2004 DEV 368,352 13,965 0 3.8% 0.0% Israel 2005 DEV 379,503 14,687 3,536 3.9% 0.9% Italy 2002 DEV 1,584,125 107,092 6.8% Italy 2003 DEV 1,625,707 97,777 56,195 6.0% 3.5% 74.0% Italy 2004 DEV 1,657,635 101,066 69,138 6.1% 4.2% 46.2% Italy 2005 DEV 1,688,198 104,364 73,801 6.2% 4.4% 41.4% Japan 2002 DEV 2,549,003 . Japan 2003 DEV 2,550,087 105,988 104,904 4.2% 4.1% 1.0% Japan 2004 DEV 2,553,135 112,872 109,824 4.4% 4.3% 2.8% Japan 2005 DEV 2,572,088 114,013 95,060 4.4% 3.7% 19.9% Jordan 2002 AFR 83,398 4,792 5.7% Jordan 2003 AFR 88,478 5,080 0 5.7% 0.0% Jordan 2004 AFR 95,010 6,532 0 6.9% 0.0% Jordan 2005 AFR 102,716 7,706 0 7.5% 0.0% Kazakhstan 2002 ECA 26,144 . Kazakhstan 2003 ECA 27,805 2,696 1,035 9.7% 3.7% Kazakhstan 2004 ECA 29,883 2,690 612 9.0% 2.0% Kazakhstan 2005 ECA 32,150 3,302 1,035 10.3% 3.2% Kenya 2002 AFR 101,582 4,760 4.7% Kenya 2003 AFR 107,490 5,943 35 5.5% 0.0% Kenya 2004 AFR 114,168 6,701 23 5.9% 0.0% Kenya 2005 AFR 125,102 7,371 -3,563 5.9% -2.8% Latvia 2002 ECA 165,447 6,178 3.7% Latvia 2003 ECA 172,996 7,549 0 4.4% 0.0% Latvia 2004 ECA 183,037 10,041 0 5.5% 0.0% Latvia 2005 ECA 193,893 10,856 0 5.6% 0.0% Lebanon 2002 AFR 57,079 2,853 5.0% Lebanon 2003 AFR 58,858 2,891 1,112 4.9% 1.9% Lebanon 2004 AFR 61,282 3,470 1,046 5.7% 1.7% Lebanon 2005 AFR 63,423 3,127 986 4.9% 1.6% Lithuania 2002 ECA 59,163 3,241 5.5% Lithuania 2003 ECA 62,897 3,734 0 5.9% 0.0% Lithuania 2004 ECA 66,578 3,762 81 5.7% 0.1% Lithuania 2005 ECA 71,085 4,507 0 6.3% 0.0% Luxembourg 2002 DEV 18,849 2,250 11.9% Luxembourg 2003 DEV 19,546 2,176 1,479 11.1% 7.6% 47.1% Luxembourg 2004 DEV 20,239 2,172 1,479 10.7% 7.3% 46.9% Luxembourg 2005 DEV . . Macedonia, FYR 2002 ECA 132,934 9,757 7.3% Macedonia, FYR 2003 ECA 141,611 9,698 1,021 6.8% 0.7% Macedonia, FYR 2004 ECA 148,950 8,584 1,245 5.8% 0.8% Macedonia, FYR 2005 ECA 157,973 10,814 1,791 6.8% 1.1% Madagascar 2002 AFR 15,823 667 4.2% Madagascar 2003 AFR 16,889 1,089 23 6.4% 0.1% Madagascar 2004 AFR 18,075 1,200 14 6.6% 0.1% Madagascar 2005 AFR 19,305 1,234 4 6.4% 0.0% Malawi 2002 AFR 5,194 329 6.3% Malawi 2003 AFR 5,214 339 319 6.5% 6.1% Malawi 2004 AFR 5,262 272 224 5.2% 4.3% Malawi 2005 AFR 5,595 420 87 7.5% 1.6% Malta 2002 ECA 30,835 1,651 5.4% Malta 2003 ECA 32,996 2,161 0 6.5% 0.0% Malta 2004 ECA 35,410 2,414 0 6.8% 0.0% Malta 2005 ECA 37,773 2,363 0 6.3% 0.0% Mexico 2002 LAC . . Mexico 2003 LAC . . Mexico 2004 LAC . . Mexico 2005 LAC 4,290,000 306,400 7.1% Moldova 2002 ECA 48,189 4,012 8.3% Moldova 2003 ECA 52,529 4,587 247 8.7% 0.5% Moldova 2004 ECA 56,881 4,874 522 8.6% 0.9% Moldova 2005 ECA 61,333 5,033 581 8.2% 0.9% Morocco 2002 AFR 119,942 9,362 7.8% Morocco 2003 AFR 130,730 10,788 0 8.3% 0.0% Morocco 2004 AFR 142,540 11,810 0 8.3% 0.0% Morocco 2005 AFR 155,947 13,407 0 8.6% 0.0% Netherlands 2002 DEV 941,000 88,000 9.4% Netherlands 2003 DEV 955,000 88,000 74,000 9.2% 7.7% 18.9% Netherlands 2004 DEV 986,000 102,000 71,000 10.3% 7.2% 43.7% Netherlands 2005 DEV 1,030,000 116,000 72,000 11.3% 7.0% 61.1% New Zealand 2002 DEV 275,813 42,976 15.6% New Zealand 2003 DEV 307,461 54,861 23,213 17.8% 7.5% 136.3% New Zealand 2004 DEV 345,702 62,468 24,227 18.1% 7.0% 157.8% New Zealand 2005 DEV 388,846 62,695 19,551 16.1% 5.0% 220.7% Norway 2002 DEV 268,491 39,041 14.5% Norway 2003 DEV 290,432 38,747 16,806 13.3% 5.8% 130.6% Norway 2004 DEV 289,955 43,068 43,545 14.9% 15.0% -1.1% Norway 2005 DEV 298,360 47,436 39,031 15.9% 13.1% 21.5% Pakistan 2002 Asia 38,893 1,371 3.5% Pakistan 2003 Asia 40,631 1,738 0 4.3% 0.0% Pakistan 2004 Asia 40,670 2,576 2,537 6.3% 6.2% Pakistan 2005 Asia 44,897 4,227 0 9.4% 0.0% Peru 2002 LAC 469,692 23,136 4.9% Peru 2003 LAC 494,449 25,696 939 5.2% 0.2% Peru 2004 LAC 521,765 28,302 986 5.4% 0.2% Peru 2005 LAC 554,135 33,349 979 6.0% 0.2% Poland 2002 ECA 478,972 23,247 4.9% Poland 2003 ECA 489,738 23,938 13,172 4.9% 2.7% Poland 2004 ECA 498,920 23,683 14,501 4.7% 2.9% Poland 2005 ECA 509,894 23,864 12,890 4.7% 2.5% Portugal 2002 DEV 252,827 15,076 6.0% Portugal 2003 DEV 262,686 16,770 6,911 6.4% 2.6% 142.7% Portugal 2004 DEV . . Portugal 2005 DEV . . Romania 2002 ECA 629,410 50,129 8.0% Romania 2003 ECA 686,451 73,850 16,809 10.8% 2.4% Romania 2004 ECA 768,056 89,244 7,639 11.6% 1.0% Romania 2005 ECA 851,562 91,386 7,880 10.7% 0.9% Russia 2002 ECA 3,845,000 311,339 8.1% Russia 2003 ECA 4,150,000 362,887 57,887 8.7% 1.4% Russia 2004 ECA 4,417,000 369,476 102,476 8.4% 2.3% Russia 2005 ECA 4,767,300 446,605 96,305 9.4% 2.0% Senegal 2002 AFR 987 31 3.1% Senegal 2003 AFR 1,015 57 29 5.6% 2.9% Senegal 2004 AFR 1,115 24 -76 2.2% -6.8% Senegal 2005 AFR 1,000 23 138 2.3% 13.8% Serbia 2002 ECA 199,572 7,644 3.8% Serbia 2003 ECA 204,891 6,721 1,402 3.3% 0.7% Serbia 2004 ECA 263,251 8,832 -49,528 3.4% -18.8% Serbia 2005 ECA 270,872 14,608 6,987 5.4% 2.6% Singapore 2002 DEV 87,490 11,294 12.9% Singapore 2003 DEV 91,600 13,544 9,434 14.8% 10.3% 43.6% Singapore 2004 DEV 98,025 17,513 11,088 17.9% 11.3% 57.9% Singapore 2005 DEV 102,662 19,501 14,864 19.0% 14.5% 31.2% Slovakia 2002 ECA 57,338 2,372 4.1% Slovakia 2003 ECA 62,368 4,029 -1,001 6.5% -1.6% Slovakia 2004 ECA 72,191 6,199 -3,624 8.6% -5.0% Slovakia 2005 ECA 81,775 7,507 -2,077 9.2% -2.5% Slovenia 2002 ECA 31,401 2,003 6.4% Slovenia 2003 ECA 33,974 2,559 -14 7.5% 0.0% Slovenia 2004 ECA 37,078 2,994 -110 8.1% -0.3% Slovenia 2005 ECA 40,560 3,237 -245 8.0% -0.6% South Africa 2002 AFR 507,813 29,590 5.8% South Africa 2003 AFR 509,815 29,343 27,341 5.8% 5.4% South Africa 2004 AFR 529,028 33,645 14,432 6.4% 2.7% South Africa 2005 AFR 553,425 41,356 16,959 7.5% 3.1% Spain 2002 DEV 1,840,645 117,780 6.4% Spain 2003 DEV 1,952,789 124,088 11,944 6.4% 0.6% 938.9% Spain 2004 DEV 2,067,703 131,685 16,771 6.4% 0.8% 685.2% Spain 2005 DEV 2,193,691 139,119 13,131 6.3% 0.6% 959.5% Sri Lanka 2002 Asia 45,673 3,116 6.8% Sri Lanka 2003 Asia 49,612 3,939 0 7.9% 0.0% Sri Lanka 2004 Asia 53,764 4,152 0 7.7% 0.0% Sri Lanka 2005 Asia 58,518 4,754 0 8.1% 0.0% Sweden 2002 DEV 287,126 16,649 5.8% Sweden 2003 DEV 289,132 15,684 13,678 5.4% 4.7% 14.7% Sweden 2004 DEV 295,538 20,245 13,839 6.9% 4.7% 46.3% Sweden 2005 DEV 301,814 21,695 15,419 7.2% 5.1% 40.7% Switzerland 2002 DEV 127,388 16,719 13.1% Switzerland 2003 DEV 131,638 12,221 7,971 9.3% 6.1% 53.3% Switzerland 2004 DEV 134,287 13,186 10,537 9.8% 7.8% 25.1% Switzerland 2005 DEV 140,580 8,998 2,705 6.4% 1.9% 232.6% Syria 2002 AFR 2,192 201 9.2% Syria 2003 AFR 2,168 173 197 8.0% 9.1% Syria 2004 AFR 2,299 178 47 7.7% 2.0% Syria 2005 AFR 2,268 216 247 9.5% 10.9% Tanzania 2002 AFR . 3,623 Tanzania 2003 AFR . 2,675 Tanzania 2004 AFR . 3,242 Tanzania 2005 AFR 59,163 3,933 6.6% Tunisia 2002 AFR 47,816 5,460 11.4% Tunisia 2003 AFR 53,665 5,332 -517 9.9% -1.0% Tunisia 2004 AFR 57,682 5,883 1,866 10.2% 3.2% Tunisia 2005 AFR 62,563 6,353 1,472 10.2% 2.4% Turkey 2002 ECA 585,981 56,285 9.6% Turkey 2003 ECA 605,020 56,804 37,765 9.4% 6.2% Turkey 2004 ECA 632,093 82,250 55,177 13.0% 8.7% Turkey 2005 ECA 593,166 86,900 125,827 14.7% 21.2% Uganda 2002 AFR . 5,106 Uganda 2003 AFR . 4,857 Uganda 2004 AFR . 7,221 Uganda 2005 AFR 89,503 8,096 9.0% United Kingdom 2002 DEV 1,658,200 225,500 13.6% United Kingdom 2003 DEV 1,804,100 325,900 180,000 18.1% 10.0% 81.1% United Kingdom 2004 DEV 2,016,700 390,200 177,600 19.3% 8.8% 119.7% United Kingdom 2005 DEV 2,160,000 333,700 190,400 15.4% 8.8% 75.3% Ukraine 2002 ECA 411,094 27,361 6.7% Ukraine 2003 ECA 430,796 27,877 8,175 6.5% 1.9% Ukraine 2004 ECA 451,167 26,724 6,353 5.9% 1.4% Ukraine 2005 ECA 471,839 28,716 8,044 6.1% 1.7% United States 2002 DEV 4,921,000 . United States 2003 DEV 4,960,000 618,503 579,503 12.5% 11.7% 6.7% United States 2004 DEV 5,052,000 657,195 565,195 13.0% 11.2% 16.3% United States 2005 DEV 5,156,000 676,830 572,830 13.1% 11.1% 18.2% Yemen 2002 AFR . . Yemen 2003 AFR . . Yemen 2004 AFR . . Yemen 2005 AFR 21,332 1,800 8.4% Zambia 2002 AFR . 2,350 Zambia 2003 AFR . 2,443 Zambia 2004 AFR . 3,078 Zambia 2005 AFR 65,155 3,389 5.2% Annex 2 Analysis of Cross-Country comparisons using Entrepreneurship Database Let's reexamine the correlations done by Klapper et. al in light of what we know about the potential problem with firm demography. Using the ED data, Klapper et al, ran various correlations with indicators such as (i) the business environment using indicators from the World Bank Doing Business report, such as the 178 country rank34, # of procedures to start a business as well as (ii) the level of economic and financial development using income and private credit/gdp measures. Running a simple correlation of the entry rate, or the number of new companies this year over the total number of companies last year on the ease of doing business indicator rank (a simple weighted average of ranks of ten business related indicators) and the # of procedures to start a business, Klapper et. al found statistically significant evidence that a higher entry rate35 was associated with a more competitive business environment and with fewer number of procedures to start a business. We followed the same procedures and in Figure 3 you can see the results confirm the original hypothesis ­ higher entry rates are negatively correlated with DB rank and # of procedures to start a business - without identifying the direction of causality. Figure 3: Entry Rate versus Ease of Doing Business Rankings & # Start-Up Procedures, 2005 .2 Singapore .2 Singapore New Zealand Chile Albania New ZealandChile United Kingdom Albania Norway Denmark United Kingdom 5 Norway e .1 e 5 Denmark HongEstoniaChina Germany Turkey atR .1 Hong EstoniaChina atR UnitedKong, IcelandStates Kong, Germany Costa Rica AlgeriaArgentina United States Iceland Turkey Romania France Czech Republic ry Argentina CostaAlgeria Rica Netherlands Kazakhstan Tunisia ntE/ Ireland Cyprus Netherlands Czech Republic FranceRomania yrtnE/ Pakistan Ireland Kazakhstan Tunisia Hungary Botswana Morocco Georgia .1 Georgia Botswana Pakistan Hungary Bangladesh Sri LankaMoldova Austria Slovenia SouthCroatia Indonesia ueslav Morocco Austria Bangladesh Sri Lanka Moldova Belgium Armenia Jordan SloveniaJordan Malawi Indonesia esluav .1 Sweden Africa Macedonia, FYR Greece SwedenSouth Belgium Africa Armenia MalawiCroatia Finland Spain Bolivia Macedonia, FYR Greece Switzerland Kenya Bolivia Latvia LithuaniaItaly Madagascar Congo, Rep. El Salvador Ukraine edttiF Malta SwitzerlandSpain Finland Lithuania El Salvador Kenya Ukraine Madagascar Congo, Rep. PeruIndia Serbia Colombia 5 Canada Latvia Peru Italy Lebanon Serbia India edttiF 5 Canada .0 Japan Poland Bosnia and Herzegovina .0 Israel Japan Poland Colombia Lebanon Bosnia and Herzegovina Israel Haiti HaitiSenegal Senegal 0 0 Luxembourg 0 5 10 15 20 0 50 100 150 200 Start-up procedures to register a business (number) (2006) Ease of Doing Business 2006 Source: World Bank Entrepreneurship Database 2007, World Bank Development Indicators 2007 However, if we disaggregate the sample and look at the 26 countries (listed above) that have data which suggests they are not collecting firm "deaths" or firms are not de- registering at all and do the same correlations, we see that the relationship significantly weakens for the ease of doing business rank and reverses for the number of start up procedures to register a business ­ implying more procedures increases firm entry rate (or the reverse; see Figure 4 below). 34Since the Doing Business Report did not produce a rank in 2005, we used the Ease of Doing Business Rank from 2006, something which was not clear from the Klapper et. Al paper 35 We define entry rate as: NewCompaniest TotalCompaniest -1 Understanding and Improving Data on Entrepreneurship and Active Companies 22 June 2008 Figure 4: Entry rate versus Ease of Doing Business Rankings & # Start-Up Procedures (Both) .2 .2 Albania Albania e 5 atR .1 etaR 5 .1 yr Iceland Algeria yr Iceland Algeria ntE/ Tunisia ntE/ Tunisia Botswana Pakistan .1 Pakistan Georgia ueslav Morocco SloveniaJordan Bangladesh Sri Lanka Bangladesh Indonesia esulav .1 Botswana Morocco Georgia Sri Lanka Slovenia JordanIndonesia edttiF Malta Lithuania Kenya Madagascar Congo, Rep. Lithuania Congo, Rep.Madagascar Kenya 5 Canada Latvia Serbia India .0 Bosnia and Herzegovina edttiF 5 Canada Latvia SerbiaIndia .0 Bosnia and Herzegovina Senegal Senegal 0 0 0 50 100 150 200 0 5 10 15 Ease of Doing Business 2006 Start-up procedures to register a business (number) (2006) Source: World Bank Entrepreneurship Database 2007, World Bank Development Indicators 2007 We also checked by grouping three consecutive years of data (2003, 2004 and 2005) to increase the sample size to 78 observations and the same puzzling results were found (Figure 5). Figure 5: Entry rate versus # start-up procedures (Both), 2003-2005 .2 Albania'05 Botswana'04 e 5 Algeria'04 .1 rat Botswana'03 Algeria'03 ry Iceland'05 Algeria'05 ntE/ Tunisia'03 Tunisia'045 .1 Pakistan'05 ueslav Morocco'05 Georgia'05 Botswana'05 Morocco'04 Bangladesh'05 05 Sri Lanka'05 Sri Lanka'03 Morocco'03 Sri Lanka'04 Slovenia'04 Congo,Slovenia'03 Georgia'04 Albania'03 Rep.'03 Jordan'05 Indonesia'05 04 Bangladesh'03 Bangladesh'04 Jordan'04 edttiF Madagascar'05 Madagascar'04 Madagascar'03 Canada'04 Lithuania'05 5 Congo, Rep.'04 Lithuania'03 Pakistan'04 Ghana'03 Lithuania'04 Kenya'04 Kenya'05 5 Canada'03 5 Latvia'05 4 Georgia'03 Senegal'03 Kenya'03 Serbia'05 India'04 Bosnia and Herzegovina'04 5 Jordan'03 .0 Latvia'03 India'03 Pakistan'03 Serbia'04 Bosnia and Herzegovina'05 Serbia'03 Indonesia'03 Senegal'04 Senegal'05 Indonesia'04 0 0 5 10 15 Start-up procedures to register a business Source: World Bank Entrepreneurship Database 2007, World Bank Development Indicators 2007 It is important to remember when looking skeptically at these small sample size correlations, that in fact the whole database suffers from a sample size that may be too small to infer anything meaningful from the data. Even if we assumed that the data were accurate, with just 84 countries in the entrepreneurship data set, and in many cases with lack of data in other areas the correlations are being made with just 60-80 observations - strongly increasing the likelihood that outliers are driving the correlations. If we remove the top two (Singapore and New Zealand) and bottom two (Haiti and Senegal) countries for business entry rate, then the relationship between things like entry rate and average of Kaufmann et al Governance indicators (Figure 9 p. 21 Klapper et. Al) becomes insignificant. Another curiosity emerges when we compare the entry rates with level of economic development and disaggregate the data by income levels. Klapper et al. correlated entry rate with GDP per capita and found a significant positive relationship ­ suggesting, as they say, either "positive economic growth is determinant for the creation (i.e. registration) of new businesses or ...greater entrepreneurship leads to economic growth Understanding and Improving Data on Entrepreneurship and Active Companies 23 June 2008 and innovation" (p. 20). We repeated this correlation using GNI per capita, Atlas method and only with 2005 data (Figure 7 left hand side). The result is consistent with Klapper et. al's finding. Thus, it is possible to suggest that greater business opportunities are related to a more robust private sector (without inferring causality). However when we disaggregate the data something very interesting happens: for the 25 high income countries36 in the dataset, the relationship disappears and becomes slightly (but not statistically significantly) negative37. This in turn suggests, for these 25 high income counties, either that as they get richer they produce less fewer businesses, or that more new businesses lead to decreases in income per capita. Figure 4: Entry Rates versus GNI per capita (All vs. 25 High Income Only) .2 Singapore .2 Singapore Chile New Zealand Albania New Zealand United Kingdom Norway etaR 5 Denmark United Kingdom Norway .1 Estonia 5 Denmark Germany Turkey Hong Kong, China .1 UnitedIceland States Hong Kong, China Germany rytnE/ AlgeriaArgentina Costa Rica etaR UnitedIceland States RomaniaCzech Republic France Netherlands ry France Tunisia Kazakhstan Ireland Netherlands .1 Pakistan Ireland Georgia Botswana ntE/ ueslav Morocco Hungary Bangladesh Moldova Sri Lanka .1 Indonesia Slovenia Austria MalawiSouth ArmeniaCroatia Jordan Africa Belgium Austria Sweden esluav Belgium edttiF Madagascar Bolivia Macedonia, FYR Kenya Congo,Lithuania UkraineRep. El Salvador Malta GreeceSpainItaly Finland Switzerland Sweden 5 India PeruLatvia Canada Malta Finland Switzerland Colombia Serbia GreeceSpainItaly .0 Lebanon Canada Bosnia and Herzegovina Poland 5 Israel Japan dettiF .0 Israel Japan Haiti Senegal 0 Luxembourg 0 Luxembourg 0 20000 40000 60000 80000 GNI per capita, Atlas method (current US$) 0 20000 40000 60000 80000 GNI per capita, Atlas method (current US$) Source: World Bank Entrepreneurship Database, World Bank Development Indicators 2007 Perhaps for the higher income countries, differences in legal form and financial practice may dominate over "entry procedures" and `ease of doing business" in determining the number and size of corporations. As mentioned earlier, some economies with relatively fewer (and larger) firms may be reaping economies of scale and efficiencies that might have a larger effect than that for increased competition between a relatively larger number of firms. At a minimum, it suggests that there are no absolute linear relations in the data, even when the data (as for developed countries) is presumed to be relatively accurate. 36We define high income under the World Bank definition as those countries with GNI per capita of $11,116 or more. In ED database there are 25 high income countries. 37Spearman rho of .1008 Understanding and Improving Data on Entrepreneurship and Active Companies 24 June 2008 Annex 3 Country Case Studies · Ukraine · Latvia · Peru · Macedonia · South Africa · Other African countries Understanding and Improving Data on Entrepreneurship and Active Companies 25 June 2008 "Sorting out the active from the inactive" How a backlog of inactive but still registered enterprises is hiding the true entrepreneurship picture in Ukraine Working paper ­ by Yuriy Kuzmyn and Florentin Blanc, IFC Ukraine BEE Project Significant differences between the officially reported figures on entrepreneurship (which get reflected in international indices) and the real situation means that the design of reform interventions and solutions can be seriously misguided. The government as well as the donor community and not least the IFC need to take this into account when designing programs aimed at fostering private sector development. Paraphrasing the well known slogan: "what gets measured, gets done" ­ "wrong measurement will lead to wrong actions!" Executive summary In Ukraine, regulatory barriers to exit have resulted in a considerable number of businesses, which are not active, but are still in the business registrar. This backlog of inactive enterprises in Ukraine (40% of all registered enterprises) thus hampers accurate tracking of business entry and growth rates and means that officially reported figures do not adequately reflect the true entrepreneurship picture. This means that the World Bank Group Entrepreneurship database (WBG ED), while being the most comprehensive cross country firm entry dataset, provides information, which does not accurately reflect business development in Ukraine. Not filtering out the "inactive but still registered" businesses means that the overall number of enterprises, and their density per 1,000 population, is overestimated. This is compounded by differences in definitions of enterprise that further contribute to make entrepreneurship in Ukraine appear more developed than it really is. At the same time, it means that the growth rate is substantially underestimated, making Ukrainian business look less dynamic than it is. IFC Ukraine BEE project estimates that the actual development of business in Ukraine in 2005-2006 was as follows: · Average entry rate was 7.1% ­ vs. 5.9% according to WBG ED · Average annual growth rate in active enterprises was 7.1% ­ vs. 4.3% according to WBG ED · The 1-year survival rate for Ukraine was 91% and the 2-year survival rate was 81%38. 38This was calculated using specially requested data from the State Statistics Committee ­ WBG ED does not have a "survival rate" indicator. Using the methodology suggested by Luttikhuizen, Hornberger and Coolidge in their draft paper "Understanding and Improving Data on Entrepreneurship and Active Companies" (May 2008 ­ see pages 4-8) one can calculate the 1-year survival rate suggested by the WBG ED data as being over 98%. Understanding and Improving Data on Entrepreneurship and Active Companies 26 June 2008 Number of registered and active enterprises in Ukraine This document sheds light on the business statistics in Ukraine. We compare the data coming from two sources: State Statistics Committee of Ukraine and WBG Entrepreneurship database. We intend to show the differences in methodology behind the datasets and how they influence business demography statistics for Ukraine39. Despite the fact that international definition of businesses does exist, the rules that govern what statistical offices do largely reflect institutional and administrative arrangements that exist in their country. Therefore indicators of businesses demography may differ from country to country. For the purpose of this study, we base ourselves on the definition used in the EU `The enterprise is the smallest combination of legal units that is an organizational unit producing goods or services, which benefits from a certain degree of autonomy in decision-making, especially for the allocation of its current resources. An enterprise carries out one or more activities at one or more locations. An enterprise may be a sole legal unit.'40 This definition is consistent with the one used in the 1993 System of National Accounts and International Standard of Industrial Classifications. With the goal to have indicators, which could be comparable across countries, from the universe of all legal entities registered in Business registrar in Ukraine we selected those, which correspond to the above definition and local practice. Specifically, the following rules are applied to derive the final dataset: 1. Legal forms: The indicators presented below include market oriented legal forms (e.g. limited liability companies, partnerships) but exclude business units in the central and local government sectors, associations and unions. This is partly because the births and deaths of enterprises in the latter sectors are typically determined by very different factors than those that govern births and deaths in the market sector41. Due to sub-optimal quality of data we also exclude sole proprietors42 (which are also excluded from the WBG ED data for Ukraine). 2. Ownership: 39Interestingly, at least for corporations, Ukraine official statistics actually maintain a register of active enterprises vs. "registered but inactive". This is far from being the case in every country, as shown in Luttikhuizen, Hornberger and Coolidge (op.cit.). Thus this suggests that one key way to improve the quality of data in WBG ED can be to exercise stricter control of the quality of the responses provided by local institutions. 40Council Regulation (EEC) No 696/93 of 15 March 1993 on the statistical units for the observation and analysis of the production system in the Community. 41Specifically, companies accounted for here are legal entities (a) of selected organizational types [Private enterprise, Associated company, Foreign enterprise, Enterprise of association of citizens, Enterprise of consumer co-operation, Joint-stock company, Public corporation, Closed joint-stock company, State joint- stock association (society), Limited (liability) company, Company with additional liability, Complete partnership, Special partnership, Cooperative (society), Associations of legal entities], Parties, religion organizations, etc (in total 41 types of establishments were excluded) and of (b) private ownership. 42According to results of a population survey conducted by IFC in April-May 2007, number of active sole proprietors is about a quarter of the official data on registered sole proprietors. Understanding and Improving Data on Entrepreneurship and Active Companies 27 June 2008 For the above reason we exclude business units owned by central or local government. 3. Activities: Activities relating to production, construction, distributive trades and services are covered, but agriculture, public administration, non-market, and extra-territorial activities are not. This is mainly to comply with the current coverage of statistical business registrars in most OECD and EU countries43 - agriculture is typically excluded because of its specificities which mean it is difficult to aggregate "farms" and "enterprises" in a meaningful way. The exclusion of public administrations etc. from entrepreneurship data is relatively self-explanatory. As a result of application of the above mentioned criteria about sixty percent out of 1,130,456 Ukrainian legal entities (not counting sole proprietors) satisfied this internationally accepted definition of enterprise in 2006. Chart 1. About half of registered legal entities satisfy the international definition of enterprise. Registered Excluded legal entities Included (1,130,000) Do not Satisfy satisfy property property criteria criteria (259,000) (871,000) Do not Satisfy legal Do not Satisfy legal satisfy legal form criteria satisfy legal form criteria form criteria form criteria (257,000) (2,100) (126,000) (745,000) Do not Satisfy type Do not Satisfy type Do not Satisfy type Do not Satisfy type satisfy type of activity satisfy type of activity satisfy type of activity satisfy type of activity of activity criteria of activity criteria of activity criteria of activity criteria criteria criteria criteria criteria (57,000) (200,000) (400) (1,700) (63,500) (62,900) (112,000) (633,000) 43 Specifically, companies accounted for here are filtered by their main type of activity [Industry, Construction, Trade, Public catering and hotels, Transportation, Services are included; Agriculture / Forestry, Public administration and extra-territorial activity are excluded. See also EUROSTAT - OECD Manual on Business Demography Statistics, OECD, p. 13 http://epp.eurostat.ec.europa.eu/cache/ITY_OFFPUB/KS-RA-07-010/EN/KS-RA-07-010-EN.PDF Understanding and Improving Data on Entrepreneurship and Active Companies 28 June 2008 to rate la r (see 9t eredt l ered)t in aniesp and 29 2-yea Exit (de- regis tota regis C/A 1% 2% 2% born surviv abou com the is of ises sesi Ukraine the the ni nda es ris many pute Enterpr 91% Ukraine enterprlal how com in Liquidated (# enterp de- registered during year) C 4,969 11,001 11,527 fo of e business was to of entm nitorom cane rate ed)r on enterprise stei gistereR percentag to is,th Ukraine an eredt a as elop to for regl State of helps Entry (newly regis tota B/A 6% 7% 7% dev rate ater ition age (# s xxraey the the Unified ot ity add - track activ In. survival to so averageeth eredt wly enterprise title devivrus Ne registered of regis during year) B 36,578 40,201 46,686 The 1-year dna tors doing that the te to Official indica ice. rises.pr of ra finds e. n-xxraey of explanatory: 2006, ered)t ni Serv ente In one l seta self Tax rate Activity (active tota regis D/A 59% 58% 60% Ukrain nrob al fo mpanies ted are difficultyyr Co nstruc others Stateeth registered atistics.st year of# es of surviv in regulato on e ris co to the official and rainekU mmitteeoC Actived we rns sharea in Year change activ enterp dA x 4% 10% an retu the rate tics explanation, tax ass ofe of tax (# es erewtaht,sesirpretneforeb from Statis ial enterprises ilef becaus activity ris s) mun spec tea enterprises the enterprise Active enterp filing return D 333,604 347,196 382,210 State which requested on ehtswo epreneurship Entr gisteredre serve liquid active the sh on de active year by of tities, not of do Based. and of# in es ichhw them enl Data 44 on ber eredt ris held number ofo but specifically lega­ 81% steredi oving Tw num­ Year change regis enterp dA x 5% 6% ert cator,idni tive,ca data was reg pr tities on of (# es as regis rate' Im rate derived en ing ratel Ukraine. be steri end) of val and the details). based Number enterpris reg siness `survi year on for1 Bu Active Activity stop rate surviva years. 1: Total registered of in of A 565,113 595,915 632,759 dardn n. · · sta xx- 2008 Based Table Table Source: Organizations A 44 year Understanding June Year 2004 2005 2006 Comparison with the World Bank Group Entrepreneurship Database (WBG ED)45 The World Bank Group Entrepreneurship database, a joint effort led by the IFC SME Department and the World Bank Development Research Group, is known as the most comprehensive dataset on cross-country firm entry data available today. It includes cross- country, time-series data on the number of total and newly registered businesses and was collected directly from Registrars of Companies via questionnaires46. Judging from the total number of entities, WBG ED provides data on registered entities and corporations in Ukraine despite the fact that its questionnaire contains the requirement that the data supplier should report only on businesses, which are considered to be active47. As specified in the methodological note accompanying Entrepreneurship Database48, for the sake of comparisons across countries, WBG ED covers only registered companies ­ but the Ukrainian State Statistical Committee clearly did not understand correctly or overlooked the request to exclude non-active entities. By contrast, the IFC Ukraine BEE Project dataset analyzed here provides data separating "registered and active" from "registered but non-active" entities. Despite the fact that WBG ED and IFC Ukraine BEE data come from the same source and aim to provide the same kind of information, they present some differences. A major difference comes from the definitions used. The IFC Ukraine BEE dataset was prepared by the State Statistics Committee of Ukraine (holder of the Business Registrar) according to the rules specified above (definition of enterprise that is built on legal forms, ownership and activities criteria, see also Chart 1). On the contrary, the dataset provided to WBG ED was not filtered to eliminate categories that are usually excluded from the internationally accepted definition of enterprise (see Table 2 for details). This also indicated that the Ukrainian State Statistical Committee does not do it as a matter of usual practice. The differences apply both to "enterprises" and to "corporations", as defined by the WBG ED questionnaire (see below). 45http://www.ifc.org/ifcext/sme.nsf/Content/Entrepreneurship+Database 46http://www.ifc.org/ifcext/sme.nsf/AttachmentsByTitle/Enterprise+Database+Survey+English/$FILE/Engl ish.pdf. 47See footnote 4 to questionnaire. 48See p.16 of `Entrepreneurship and Firm Formation across Countries', Policy Research Working Paper 4313, The WB Development Research Group, August 2007. 30 Table 2: Comparison with WBG Entrepreneurship Database Indicator Of those in practice: Registered Active49 Total entities in business registrar as of January 1, 1,133,200 53% 2007 of them: Total enterprises (as per WBG ED) 830,719 57% Total corporations (as per WBG ED) 494,730 58% Total enterprises (as per IFC Ukraine BEE) 632,759 60% Total corporations (as per IFC Ukraine BEE) 443,046 60% Implications for business demography statistics The above differences in methodologies have serious implications for understanding business demography in Ukraine. 1. The data reported by WBG ED more than twice (217%) overestimates the number of actually active enterprises. This is mainly due to: a. Use of registered, rather than active entities. The number of active entities in Ukraine is less than two thirds (60%) of the number of registered entities. b. Variation in the definition of enterprise. In accordance with a commonly internationally accepted definition of enterprise, the IFC Ukraine BEE dataset does not cover certain types of activity. As a result, the number of registered enterprises reported by IFC Ukraine BEE is about one quarter less (24%) than the same indicator reported by WBG ED. This means that the level of entrepreneurship development, as measured by the number of enterprises per 1,000 active population, was actually lower than currently reported by WBG ED. 2. Because WBG ED data contains a large amount of non-active entities as well as entities, which should not be considered as enterprises, the analysis based on this total underestimates the actual growth rate of business in Ukraine over the last years. The average annual year on year change (growth rate) of the number of active enterprises over 2005-2006 was 7.1% comparing to 4.3% reported by WBG ED. This means that entrepreneurship in Ukraine was more dynamic than currently reported by WBG ED. 49Since WBG ED does not provide data on active entities, the respective number was derived from IFC Ukraine BEE dataset. 31 Table 3: Implications for business demography statistics Indicator Legal Legal Year on year Entry rate entities, entities per change (average for 2006 1,000 active (average for 2005, 2006) population50 2005, 2006) Total registered enterprises 830,719 25.65 4.3% 5.9% (WBG ED) Total registered corporations 494,730 15.27 4.7% 6.4% (WBG ED) Total registered enterprises 632,759 19.53 5.8% 7.1% (IFC Ukraine BEE) Total registered corporations 443,046 13.68 N/A51 N/A (IFC Ukraine BEE) Total active enterprises (IFC 382,210 11.80 7.1% ca. 8%52 Ukraine BEE) Total active corporations (IFC 267,727 8.27 N/A N/A Ukraine BEE) 50Enterprise density per 1,000 people aged 15-64 as suggested in Luttikuizen, Hornberger and Coolidge 51Because the IFC Ukraine BEE project uses "enterprises" as per commonly accepted definition and not "corporations", its historical dataset does not have data on "corporations" as per WBG ED definition ­ this was only calculated specially for the latest year, to show that the differences apply regardless of whether one looks only at "corporations" or at "enterprises". 52The average entry rate of active enterprises is not computed since data on enterprises, which were newly registered and active in 2005, is not currently available. Given that entry rate of active enterprises was 8.1% in 2006, our rough estimate of it is 8%. 32 · Latvia As of the late 1990s, Latvia was preparing itself for EU accession, and was undertaking a broad range of reforms. In 1999, the Government of Latvia committed itself, inter-alia, to the principle of sharing information across government ministries and agencies, developing e-government solutions, and reducing the compliance burden for businesses (i.e., by avoiding requirements for them to provide the same information multiple times to different government agencies). In particular, they combined company registration and tax registration, and introduced a "unique identification number" for both physical and legal persons, to be used for all interactions with government. The Central Statistical Bureau has a database representing the intersection of businesses (legal persons) registered in the Commercial Registry and those "active" in the sense of showing economic activity in their tax filings to the State Revenue Service (SRS). They receive data regularly from the Commercial Registry, SRS, and other sources (e.g., re physical persons, farms, NGOs, public sector agencies, etc.). They see the balance sheet and financial information submitted by registered firms, and can update their own databases regularly. They have their own database of "economically active" firms, by type of organizational form, by "main activity" (four digit NACE), by size of firm, by ownership (public/private, foreign/domestic) etc. Firms that show no economic activity (i.e., tax filings all "zero" or failing to file) for two years are considered "dead" in the Statistical Business Registry.53 If it re-starts after that, it is counted as a "new company." They do structured annual surveys of about 15,000 firms (out of about 55,000 economically active firms as of 2007). The Statistical agency also has detailed data on investment, broken down by source of financing and "main activity" and other categories. The commercial registry in Latvia was computerized and outsourced, and companies can update their information on-line. The Company Registry has a contractual relationship with Lursoft until 2010, when it will be re-bid.54 There are varying definitions of an "active" company in Latvia. The Commercial Registry and Lursoft use "active registrations in previous year + new registrations - business terminations in the same year". Lursoft also can show the companies that filed their annual financial reports in a particular year (noting that their data may show an increase for a past year if, e.g., someone files it late). Thus for 2005, the Commercial Registry has over 80,000 firms designated as "active" by their definition. Lursoft shows slightly over 55,000 companies filed financial statements in 2005, while the Statistical Business Registry shows 49,881 "active" by their definition (following Eurostat: if a company fails to file for taxes two years in a row they are considered "dead.") 53They have reports since 2004 on "business demography," available at http://data.csb.gov.lv. (but only in Latvian language) 54Their web-site is http://www.ur.gov.lv. 33 The State Revenue Service (SRS) sees about 65% of firms that are registered are "active" and paying taxes. SRS closely tracks the amount of debt owed by companies to the state, but does not pay as much attention to the actual number of firms that are active or inactive. Data on that come from the SRS regional offices and is shared with the Statistical Business Registry, but is not actively tracked by the central SRS. They currently have 141,000 registered, "active" (i.e., non-liquidated) tax payers in 2007 including sole proprietors. Summary comparison of business statistics in Latvia for 2005: World Bank Entrepreneurship Database55: 193,893 Company Registry (incl. sole proprietors): 105,690 Company Registry (firms/legal entities): 81,128 Company Registry (firms filing financials): 55,103 CSB "active firms" (firms filing for taxes): 49,881 · Peru: Peru has a number of registration organizations at both the municipal and national level, including the national registry of "legal persons" (SUNARP), the Lima Municipality registry of firms, the national tax registry (SUNAT), the national statistical agency (INEI) and the Ministry of Labor. Of course, all these agencies have different mandates and therefore have their own definitions and classifications of firms. SUNARP focuses on legal entities and ensures their legal status within Peru. They report they have a problem of many "inactive" firms still in the registry, and they can't be sure which firms in the registry are active or not. Fees for de-registration are not expensive, but there is a procedure that has to be followed ­ the applicant needs to file the "minutes" of dissolution and get them notarized (cost about S30 - 50). All firms that are registered in SUNARP must register with SUNAT in the RUC. They expect that in the future, it may be possible for firms to bypass notaries and register themselves on-line, and simultaneously register in the RUC. Sole proprietors are generally not registered in SUNARP, but are supposed to register with SUNAT. There was a consensus among all the agencies listed above that the most accurate and up to date data about active companies in Peru would be the tax database (SUNAT). They reported they have 4.5 million registered taxpayers, of which 3.8 million are currently "active". Of the latter, almost 75% are natural persons. The other 25% (about 1 million) are businesses, including mostly natural persons (sole proprietors and professionals). Their database is called "RUC" (Unified Taxpayers Registry or Registro Unico de Contribuyentes in Spanish). 55Source cited by ED is "Ministry of Justice". 34 They get about 45,000 new registrations (of all kinds) per month. Firms that fail to file for six months are automatically de-registered (roughly 8000/month, or a total of about 700,000 since RUC was created in 1993). Firms can also ask to be "suspended" for up to a year (if they don't re-activate within a year, they are automatically de-registered). There had been about 10,000 suspensions by JPs and 37,000 by NPs for 2007 as of October. If they file "0" income, and if this continues for more than a year they are declared "inactive." Business/Vendor/Company REGIME 2004 2005 2006 2007 * GENERAL REGIME 219,327 251,191 207,782 221,349 RER 16,569 24,298 17,130 21,145 RUS 1 13 2 2 OTHER REGIMES 3/ 2,777 3,705 4,042 3,912 TOTAL 238,674 279,207 228,956 246,408 STATE 2004 2005 2006 2007 * ACTIVE 2,917,980 3,283,378 3,482,079 3,807,617 TEMPORARILY SUSPENDED 266,843 206,454 202,421 208,136 Voluntary liquidations 220,995 260,314 301,389 329,034 Involuntary liquidations 1,109,784 1,162,458 1,370,817 1,375,540 TOTAL 4,515,602 4,912,604 5,356,706 5,720,327 Firms that are de-registered either voluntarily or involuntarily have their banks notified by SUNAT, and their accounts are de-activated. The State Statistical Agency (INEI) was able to verify the accuracy and currency of the RUC database at SUNAT. They used it last year, starting with a "universe" in the RUC of about 800,000 "active" companies (including sole proprietors), and drawing a sample of 20,000 from 13 different sectors. Most large firms (about 5000) were included in a mandatory survey. There were about 8000 medium firms. Of the (about) 7000 "small" firms, at least 10% were "lost or closed" and up to a maximum of 20% were inactive or had moved to another address. A smaller percentage had changed their activity. 35 Summary comparison of business statistics in Peru, 2005: World Bank Entrepeneurship database (2005): 554,135 Tax Registry (companies) 279,207 Estimate of "active companies" (max)56 237,000 Estimate of "active companies" (min)57 186,573 · Macedonia The Government of Macedonia recently enacted a number of reforms to overhaul their registries and consolidate them in a "Central Registry". They report they have regular sharing of data between the Central Registry, Statistical office, Public Revenue office, and related agencies (e.g., customs). They state they are getting harmonized with Eurostat, re "main activities" (NACE 4 digit), and other relevant definitions and categorization. Before 2006, most company registrations took place in the courts, and companies were also required to register with the Public Revenue Office and the Statistical Agency. As of Jan. 1, 2006, there is a "one stop shop" for basic registration, a unique ID shared across government agencies (with various "sub-numbers" as relevant for specific purposes), and a protocol for sharing information across the relevant government agencies. The courts are no longer part of the process of company registration, and the Central Registry is the body responsible for registration (it was built up from the former "ZPP" central payment office). Prior to 2006, Macedonia suffered the usual problems of company statistics. There were few legal or economic incentives for companies to report changes in their address, status, etc. Now there is a requirement for annual submission of financial reports, and a legal protocol to remove a company that fails to report for three years in a row. They are also introducing full electronic registration and will soon allow electronic submission of annual financial reports. Before the reform, the statistical database had about 180,000 entities (including private commercial companies, but also public sector entities such as schools, non-profit entities such as churches and associations, etc.). Only about 55,000 submitted annual accounts in 2005; of which about 4000 submitted financial reports indicating "no financial activity." There were over 84,000 firms entered into the electronic registry through conversion of court files in 2006 (a number of which were subsequently deleted as inactive). The Central Registry reported that both registration and de-registration activity is increasing. De-registration is still a somewhat cumbersome process, including public announcement and a waiting period. It takes about 3 - 4 months. 56Estimate based on statistical agency's recent survey sample 57Estimate based on tax registry's active/total (including sole proprietors) 36 The Public Revenue Office (PRO) confirmed that they share data on companies with the Central Registry and can alert the Central Registry if, e.g., a company has been audited and found to have submitted incorrect financial information. More commonly, the two agencies compare the financial information they receive from a company (which is supposed to be in agreement). If a taxpayer submits contradictory information to the Central Registry and the PRO, they are requested to make necessary corrections and can be pursued by the PRO. The statistical agency promised to send historical data from about 2000 - 2005 (before the reform) but do not yet have a complete database ready for the new data after the reform. They plan to do surveys to verify data (e.g., address, main activity, etc.). Summary comparison of business statistics in Macedonia for 2005 and 2007: World Bank Entrepeneurship database (2005): 157,973 Of which filed annual accounts (2005, estimate) 55,000 Company Registry (thru conversion of court files) 2007: 84,089 No. "private trade companies", sole proprietors in reg.2007: 71,118 · South Africa In South Africa, there is an official policy of cooperation between the relevant government agencies to combine their data into an "integrated business registry" and to share information on a regular basis, but this is still a work in progress. For example, new companies who register with the Company and Intellectual Property Registry (CIPRO) are automatically passed along to South Africa Revenue Service (SARS), and de- registrations are similarly shared, at least in theory. CIPRO registers "legal entities", but there is no formal requirement for registration of "sole proprietors" or "partnerships" with CIPRO. As far as SARS is concerned, sole proprietors and partnerships are liable for income tax as individuals, and their companies are not recognized as tax payers in their own right. CIPRO readily admitted that they have a lot of old data, and that companies lack an incentive to formally "de-register." While they receive notice of "deregistration" from SARS, they are also taking steps to introduce a new "duty system" that will require firms to pay a small fee each year to remain on the "active list" of the registry, as a way to help ensure that the registry is more reliable and up to date. An amendment to the company law is expected to be enacted later this year. CIPRO also records liquidations and conversions (e.g., from a closed corporation to a private company), but not "mergers," which are apparently intended only for public corporations. SARS has its own database of business taxpayers (including sole proprietors), and defines the following categories: 37 · Active and still trading · De-registered · Dormant · In liquidation · In suspension · Unknown It follows established policies and procedures for moving taxpayers from one of these categories to another, and it has annual data on each of these categories, going back several years. The table below shows SARS taxpayers for 2006, (including sole proprietors as well as legal entities), those counted as fully "active" and the total in their registry, by turnover band.58 TAXPAYER_STATUS Turnover bracket (Rands) ACTIVE Grand Total59 > 500,000,000 4 5 50,000,000 - 100,000,000 36 50 30,000,000 - 50,000,000 55 71 20,000,000 - 30,000,000 68 90 10,000,000 - 20,000,000 160 232 5,000,000 - 10,000,000 314 455 2,500,000 - 5,000,000 543 790 1,000,000 - 2,500,000 1,307 1,929 300,000 - 1,000,000 3,975 5,982 20,000 - 300,000 29,066 45,892 0 - 20000 550,010 1,207,146 (Turnover unavailable) 442,020 752,216 Grand Total 1,027,558 2,014,858 Summary comparison of business statistics in South Africa, 2005 and 2007: World Bank Entrepeneurship database (2005): 507,813 CIPRO (excluding "close corporations", 2007): 425,107 CPRO (including "close corporations", 2007): 1,701,264 SARS total (including close corporations): 2,014,858 SARS "active companies" including close corporations: 1,027,558 58Data provided by SARS, Feb. 2007. 59Includes, in addition to "active" taxpayers, those who are "dormant" (not trading), "suspended" (return to the taxpayer is not being issued", "address unknown" and those in the process of closing down. 38 Other African Countries In general a precondition of comparing countries is that use is made of harmonized data that means that the data set of one country is at the level of the definitions and the individual units comparable with the other country. Therefore it can make a difference which kind of sources is being used, because different sources in different countries are likely not to be harmonized. The matter is whether active and not active units are being used in one data set, whether the informal sector is part of the data, and if so whether similar definitions are used, or completely different ones. An example can help us to understand this. It is possible to compare information from a developed country with lesser developed countries, even when the data is completely incomparable. The differences in data show at the same time differences in realities and in registration methods, between the benchmark, the developed country, and between the selected African countries. The Netherlands was chosen as a benchmark in this exercise, to compare with African countries since it is a middle sized developed country with many small firms and a well developed registration system for enterprises. The African countries were chose because they were easy to visit by one of the authors of this paper We want to emphasize that the data of all countries in principle cannot be compared, but nevertheless we can draw lessons from looking at this data. Table 1: Comparison of not comparable information from The Netherlands with 6 selected African Countries. Countries Registered Registered Popu- reg. units reg. units reg. units reg. units Differences units in stats units in stats lation per 1000 per per inhabi per with Neth. = inhabitants inhabi tant inhabitan Active tant t Index total index NL=100 active total active *1 million total tot. active NL=100 factors Netherlands 900000 719405 16 56.3 45 100 100 Tanzania 27962 NA 38 0.7 1 2 61 Kenya 43057 NA 34 1.3 2 3 36 Uganda 160883 NA 28 5.7 10 13 8 Ghana 26493 NA 22 1.2 2 3 37 Namibia 40000 NA 2 20.0 36 44 2 Botswana 20000 NA 2 10.0 18 22 4 Total of 6 318395 126 2.5 4 6 18 The number of units compared are, by definition, not comparable. In The Netherlands agricultural units (95.000) are excluded. In Tanzania are registered all ACTIVE units that are profit making and not profit making (mostly governmental), of at least five employed persons.60 In Uganda the information of 2002 is about all registered ACTIVE 60In the cases of Tanzania, Uganda and Kenya, "active" firms were identified on the basis of surveys by the statistical agency for firms with more than five employees (including government-owned enterprises). 39 firms, divided between formal (5 and more employed persons) and .informal, defined by having less than 5 working persons. The size of the included informal sector (as defined here) in Uganda is 150,000 units. For Kenya (2006) information is presented of all active firms of what they call the Modern sector, that excluded all small firms of agricultural and non agricultural nature (small is not defined). Included are all firm with a limited liability, including those of the government and all units in the urban areas. The Ghana data of 2006 is limited to all Mining and Industry units, etc. For Namibia and Botswana we only have general estimates of speakers of the statistical offices. This information shows that only in The Netherlands can an explicit difference be made between active and not active units. The data allows a comparison between the African countries and the benchmark as far the number of registered units per 1000 inhabitants is concerned. This is called the density of units. The index for the active units in the Netherland (=100) shows the number of active units in the compared country, per 1000 inhabitants. These differences also can be expressed as a factor. This means that in The Netherlands the density of commercial enterprises is 61 times higher compared with the fdata on Tanzania. The data also shows that we have an extreme problem in comparing these countries. Table 2: Statistical Business Register information compared between The Netherlands and 6 African countries for 2005. (The Netherlands is about employees in stead of working persons.) Country Total units workforce units with units with units 100 plus, units <5 >5 - 99 with 100 % of with <5 employed employed plus share of employed employed workforce as a share of total units Netherlands 719,365 8,799,000 614,215 91730 13420 60 85 = 2005 Tanzania 27,962 26995 967 Uganda 160,883 444,118 150,138 10745 254 93 Ghana 26,493 275,495 14,438 14135 303 54 Kenya 43,057 1,807,712 14,073 33 Namibia 40,000 Botswana 20,000 Total 318395 2,527,325 178,649 51875 1524 This table is first of all useful to show the differences. These differences show not only numerical differences but also the enormous differences between the types of registers. Only the data from Uganda looks comparable with the Netherlands in a very limited way. But this is just a first impression. The number of units in Uganda cannot really be compared with The Netherlands, nor can the other countries be compared, because 40 Uganda consists in fact of almost 90% "counted informal units." The result is that the share of the small units (as a ratio of the total) seems to be comparable with the share of small units in the Netherlands. However, the underlying figures are not comparable and therefore the ratios are not comparable. For Ghana and Kenya it shows that the share of the small units is also different. The conclusion is that all three countries in fact have only information over a very limited part of their economic activities. In the table below we present information on the growth of the population and of the labor force. That is essential to understand the relation between enterprises and the labor force. Table 3: Comparisons of the development of the population with the labor participation rate since 1990 for The Netherlands and six African countries. Country Population in Population growth since labor labor change 1990 in 2005 1990 in % participati participatio rates on rate in n rate in Males 1990 Male 2006 Male since 1990 Netherlands 15 16.3 8.7 71 73 2 Tanzania 26.2 38.3 46.2 91 90 -1 Kenya 23.4 34.3 46.6 90 90 0 Uganda 17.8 28.8 61.8 92 86 -6 Ghana 15.5 22.1 42.6 80 75 -5 Namibia 1.4 2 42.9 65 63 -2 Botswana 1.4 1.8 28.6 77 70 -7 Average 85.7 127.3 48.5 83 79 -4 This table shows the enormous growth of the population in the African countries compared with the growth in The Netherlands. Apart from Botswana is the growth of the population in these countries more than 5 times higher compared with the Netherlands. Further we see in the African countries a decline of the labor force participation, and an increase for The Netherlands. This can be understood in different ways. It is important to note that only a very small part of the labor force in African countries work in the formal sector, see next table. Tables 4 shows that compared with the benchmark country (the Netherlands) only a very small part of the workforce in the African countries is working in these registered units. This is particularly relevant for the workforce in the age group 15-64. Of the 1000 persons in the Netherlands, where 801 work in registered firms, the situation is different in Africa. In Kenya only 90 out of 1000 work in registered firms, in Uganda 32 out of 1000 (including the "informal" sector), and in Ghana 21 out of 1000 work in registered units. 41 Table 4. Comparison of the benchmark with three African countries for the workforce in registered units as part of the total population and the age group 15-65. Country Information on Population age group 15-workforce in worksforce worksforce in businesses in 2005 64 in 2005 registered units in registered registered units per units per 1000 1000 inhabitants inhabitants in age group 15- 65 Netherlands all businesses 16.3 10986200 8,799,000 540 801 Tanzania all units 5 and plus 38.3 Kenya modern sector 34.3 20020200 1,807,712 53 90 Uganda formal and informal 28.8 13910400 444,118 15 32 Ghana only industry 22.1 13271000 275,495 12 21 Namibia 2 Botswana 1.8 Total/Average 127.3 47201600 2,527,325 The size of the labor force in the Netherlands in registered units is, roughly comparable with Uganda and Ghana, although those countries have has a much larger population. This shows the demographic differences with The Netherlands and the share of the workforce in the registered sector as part of the total workforce. In other words, they key question is: what are the rest of the people in Kenya and Uganda and Ghana doing, those who are not working in registered units. Table 5: Indexes that present information on the size of the workforce in registered units, compared with the benchmark for the total population and the age group 15-65 (per 1000). Country Information on population age group 15-workforce in Index Index businesses in 2005 64 in 2005 registered units 540=100 801= 100 Netherlands all businesses 16.3 10986200 8,799,000 100 100 Tanzania all units 5 and plus 38.3 Kenya modern sector 34.3 20020200 1,807,712 10 11 Uganda formal and informal 28.8 13910400 444,118 3 4 Ghana only industry 22.1 13271000 275,495 2 3 Namibia 2 Botswana 1.8 Total/Average 127.3 47201600 2,527,325 This table shows again, but now as an index, that the share of the labor force in African countries that works in registered units is extremely small compared with the benchmark 42 country. The workforce in the registered units in the three countries (Kenya, Uganda and Ghana) for which we have this data is much less compared with The Netherlands. But even between the African countries there are major differences. In Kenya the level of the workforce in registered units as part of the age group 15 ­ 65 years, is higher compared with the two other countries. This information reinforces the question about the quality of this data and the nature of the work that is done by the not-registered workforce. Are they working in the informal sector, in not-registered units as in Uganda, in the very small units in Kenya and Tanzania, or in subsistence activities in agriculture? Only a labor force survey can inform us about how the massive labor force in Africa is involved in production processes. 43 3. Structural differences and a conclusion. The countries compared show that in Uganda where the "informal" sector (150,000) is included, only a fraction of the workforce is working in registered units. In Kenya a larger part of the relevant age group works in what is called the modern sector. That leaves the informal sector in Kenya, as defined in Uganda, out of the picture. The workforce in Ghana is only related to a part of the economy. The actual workforce in the formal sector in Ghana for this reason needs to be much larger. This means that the data presented in these tables strongly underestimates the number of firms and the workforce that are active in these three African countries. 44