Policy Research Working Paper 8774 Pioneering Firms in Fragile and Conflict-Affected States Why and How Development Finance Institutions Should Support Them Paul Collier Neil Gregory Alexandros Ragoussis International Finance Corporation March 2019 Policy Research Working Paper 8774 Abstract The role of ‘first movers’ in fragile states is critical: they grow finance institutions to support them, and proposes new and diversify markets in ways that no other firms do, gen- ways to offset costs, risks, and the “unknown unknowns” erating disproportionate impact in terms of development that generate radical uncertainty. Through a process of and stability. But pioneer firms are rare in fragile states. social learning and resetting negative self-fulfilling inves- This study documents their profile, their challenges, and tor narratives, development finance institutions can help the barriers that prevent them from realizing their poten- pioneering firms shift the growth trajectory of fragile and tial. The study also explores the rationale for development conflict-affected states. This paper is a product of the Private Sector Development Vice Presidency, International Finance Corporation. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/ research. The authors may be contacted at paul.collier@bsg.ox.ac.uk; ngregory@ifc.org; and aragoussis@ifc.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Pioneering Firms in Fragile and Conflict-Affected States: Why and How Development Finance Institutions Should Support Them Paul Collier,1 Neil Gregory2 and Alexandros Ragoussis3 Acknowledgments. The authors would like to thank Djeneba Doumbia and Leo Bureau-Blouin for excellent research assistance; Kruskaia Sierra-Escalante and Paddy Carter for their thorough review and comments, as well as participants of the 2019 DFI Forum on Private Investment in Fragile Environments for helpful discussions. JEL Codes: O31, L2, O19, D81 Keywords: Firms; Innovation; Pioneering; Fragility; Risk; Development Finance. 1 Blavatnik School of Government, University of Oxford and International Growth Centre, London. 2 International Finance Corporation. 3 International Finance Corporation. 1. Introduction: Fragility and its implications for development finance Fragile states face multiple problems that reinforce each other, locking them in a recurring cycle of poverty and hopelessness. Typically, society is fractured into opposing groups, with little or no shared identity that might settle differences in a context of cooperation. The state lacks legitimacy with a substantial number of its own citizens, and importantly for the business environment, it lacks the capacity to perform basic functions such as taxation, and providing security, the rule of law, and economic infrastructure. Firms and households face existential uncertainties that discourage irreversible decisions and shorten horizons. There are few formal private enterprises and so the workforce cannot reap economies of scale and specialization, leaving the population unproductive and impoverished. The society is too poor to go through the demographic transition, so there are many youths facing few opportunities. Finally, the polity and the economy are frequently hit by shocks against which they have little resilience. These problems reinforce each other. Because the society is fractured into opposing groups, the state is typically regarded by parts of society as having been captured by one identity group, which undermines its legitimacy. Because it lacks legitimacy with many of its citizens, the state cannot rely on citizens’ compliance or community engagement, which means that basic functions of the state cannot be properly performed. This, in turn, further undermines the state’s legitimacy. The risks posed by lack of legitimacy, and the inadequacy of basic state functions, combine to discourage formal enterprises. The narrowness of the economy, the surfeit of young people, their lack economic opportunities, juxtaposed with the lack of state legitimacy, expose society to shocks, which the state struggles to cushion due to its limited capacity. In turn, the frequency and severity of adverse shocks keeps derailing attempts to escape from fragility. This cycle entraps the society in poverty and a sense of hopelessness, and is why fragility is highly persistent. As more successful emerging market countries continue to converge on the income levels of OECD societies, the gap between them and the poorest countries is widening. The concept of the ‘Bottom Billion’4 identified 60 countries that, as of the millennium, were mired in persistent poverty. The subsequent 2002–17 supercycle was a big opportunity for these countries, but even during this period of economic growth, fragile countries continued to rapidly diverge from emerging market countries (at 2.2 percent per year, per capita, population-weighted). A few of the bottom billion, such as Rwanda and Ethiopia, transformed; but many others such as the Central African Republic (CAR), the Republic of Yemen and the bulk of the Middle East, which were once seen on track of change, have not only diverged, but they have gone into absolute decline. While the rest of the world has made rapid progress in eradicating extreme poverty, fragile states are increasingly the locus of the remaining extremely poor. For the world to achieve the Sustainable Development Goal (SDG) of eradicating extreme poverty by 2030, most of the improvements will need to come in fragile states.5 Because rapid private enterprise growth has been central to reducing                                                              4 Collier, 2007. 5 World Bank Group, 2018a. 2    poverty elsewhere, we cannot expect to reduce poverty in fragile states without a better understanding of the constraints facing private enterprises, and the toolkit needed to address these. In an interconnected world, such persistent divergence of countries in the fragility spectrum is intolerable. This concentrated human tragedy will increasingly focus international public policy on how to address the problems of fragile states. Concerns about poverty will be compounded by fear of the regional and global spill-overs when fragile states implode, as with the recent flight from mass violence in the Syrian Arab Republic and South Sudan, and the risk of contagion from the Ebola epidemic in Sierra Leone, Liberia, and Guinea. Governments typically lack sufficient capacity to support private sector growth. These societies will continue to look to development finance institutions (DFIs) to create sustainable investment and growth. However, DFIs’ track record in fragile states suggests that they have not been very effective in stimulating private investment. Development finance cannot itself transform fragile states: unless political leaders want transformation, it is not feasible. Where leaders block change, the predominant role for international support lies with humanitarian aid. But precisely because fragile states are shock-prone, periodically, political opportunities do arise from crises or leadership change, and development agencies need to become better at building on these potentially pivotal moments. Once a political opportunity opens, there are two priorities. One is to generate jobs; the other is to strengthen the basic sinews of the state. In fragile states, with a gulf between young workers and new work, jobs are iconic. Jobs create a new and credible national narrative about the prospects for job growth that will stabilize society. But the generation of jobs is overwhelmingly a private sector activity, and whether jobs will be productive depends upon harnessing economies of scale and specialization. Only formal firms know how to organize a workforce to reap these gains, but all fragile states are chronically short of such firms. Once a fragile state faces an opportunity for transformation, the critical role of development agencies is to expand the formal private sector as rapidly as possible. Hence, the key areas for international support should shift from the humanitarian agencies to the development finance institutions (DFIs). If DFIs are to rise to this responsibility, they will need to overcome a fundamental problem—there are not enough large, formal firms operating in fragile states for them to invest in. Therefore, we need a better understanding of why there are so few formal firms—what are the obstacles to entry by firms, and what can be done to support their entry, survival, and growth. This paper focuses on these pioneer investors in fragile states—i.e. those firms that push the boundaries of markets in fragile states. The role of these ‘first movers’ is critical: they can grow and diversify the market in ways that no other firms do. But, much like everything else in fragile situations, pioneer firms are different from pioneers in other markets. They typically have additional experience, innovate at smaller sizes, and rely less on fixed assets, and on international standards, worker training, or corporate partnerships. The level of development and diversification in the markets suggests an intuitive explanation: the scope for innovation is much broader. Yet pioneers in fragile and conflict-affected states (FCS) are unable to capitalize on their potential through formal banking. Contrary to other developing countries, pioneers in fragile states finance 3    a higher share of investment through internal funds as they have less access to credit, and face a higher cost for capital. This suggests a clear role for DFIs in supporting their growth by financing them directly, and by supporting their access to finance from private equity and from local financial markets. The challenge of supporting pioneering firms in fragile states cannot be seen in isolation—it is part of a wider coordination problem, whereby it is difficult for any one firm to prosper in a context where other firms are absent. Pioneer firms will only succeed in an economic environment where multiple firms are succeeding, so creating value chains and markets is essential for their growth. But individual firms in fragile contexts face great risk and uncertainty in knowing what other firms will do, in fragile contexts. This suggests that an additional role for DFIs is addressing the lack of coordination, and the underlying risk and uncertainty that it creates. The rest of the paper is structured as follows: section 2 describes pioneer firms and the ways they benefit fragile states. Section 3 looks more closely into the barriers and costs pioneer firms face in fragile situations. Section 4 proposes what DFIs could do to help pioneer firms overcome the barriers they face and grow. Section 5 discusses how DFIs can achieve more through greater collaboration. Section 6 provides concluding remarks. 2. Pioneering firms: Who are they and why do they matter? The most useful firms for a fragile state are those that pioneer an activity that is new to the economy: this is what is meant by creating new markets. The herd is large—‘pioneers’, ‘first movers’ and ‘first entrants’ refer to the first firms to produce a new product, introduce a new process, or enter a new market.6 They can be in any sector of economic activity, and of any size, age, or origin; yet often they tend to be large and active in multiple markets. Multinational firms, by definition, push the boundaries of markets through exports or foreign direct investment.7 There are benefits to being a first mover in a fragile state. Pioneer firms can tap into unexploited natural resources, use low-cost labor, and provide missing basic services. Many markets may be open for entry with little existing competition. Initially markets in fragile countries are small and grow slowly, but during periods of conflict reduction and stabilization, when foreign aid pours in, fragile states often experience spurts in government spending, which can support enterprise growth if the pioneering firms come in to take advantage of it.8 If not, most of the spending leaks out into imports or inflation. Under pressure from their shareholders, DFIs are eager to finance pioneer                                                              6 Some distinguish between inventor, product pioneer, and market pioneer. An inventor “develops patents or important technologies in a new product category”. A product pioneer is “the first firm to develop a working model or sample in a new product category” while a market pioneer is “the first firm to sell in a new product category” (Golder and Tellis, 1993). Studies on market entry usually focus on a market pioneer and define it as the first firm to enter a new market (Kabuth, 2003). 7 Nakata and Sivakumar, 1997; Zashev and Ehrstedt, 2010. 8 International Finance Corporation, 2019.   4    firms, so that once these firms get established, they have good prospects of obtaining growth capital. Thus, it is important for DFIs to understand the obstacles that pioneer firms face when initially entering FCS, and to devise strategies to address these obstacles. First-mover advantages for firms and markets come with heavy costs and risks. In fact, the positive externalities of first entrants are often highlighted through the disadvantages they face: (1) free- riding; (2) market uncertainty; (3) shifts in demand, supply, and regulatory needs; and (4) incumbent inertia.9 The free-rider effect and the resolution of uncertainty have the biggest impact because of the institutional failures and pervasive risks: by competing solely on prices, followers can free-ride through imitation and respond to signals of market potential. And there is no universal answer as to whether costs outweigh the benefits – the context and timing matter. Thus, the typical late market entrant makes a strategic choice to hold back and allow the market pioneers to investigate consumers’ response to an innovation before they unleash the full force of their technological and marketing capabilities through replicating and elaborating on the market-tested features of a new product.10 Here lies the macroeconomic benefit of pioneering—by adding a new activity, a pioneer firm helps to broaden the economy, and open possibilities for other firms. Pioneers make a difference through clusters and webs of interdependence Firms benefit from the presence of other firms: cluster scale economies imply that firms benefit from others who are doing the same thing.11 If cluster economies matter, even if firms entered the cluster in a known sequence and at known times, it will be foolish to be first. If five firms are needed before any can break even, the fifth entrant will not be burdened with the accumulated losses of the first, and so will outcompete it. Production is commonly organized into a web of specialist firms that are interdependent. Where interdependence matters, the first firm in a web of interconnectivity will not be viable until the last piece of the web is complete, and so again, being first imposes avoidable costs.  The concept is not new in the development literature: that agents fail to coordinate their actions has been offered as an explanation for an array of failures from resource misallocation and poverty traps, to lack of infrastructure investment (Kremer, 1993; Murphy and Shleifer, 1989; Basu, 2014). New markets will not be created purely by conventional market forces. If being first simply imposes costs for a known period until other firms enter, this would be a known cost of business. But usually a pioneer firm does not know when other firms will enter the market. However, if a firm did have an approximate idea of when other firms will join the market then, with an appropriate adjustment for risk aversion, the firm would be compensated. But of course, there is no way of forming such an expectation: a firm simply does not know when, or even if, other firms will enter, as they themselves do not know, and so the only coherent assessment is ‘I don’t know’. This is termed radical uncertainty or Knightian uncertainty: the sophisticated decision-making procedures of a formal firm to filter out such problems from serious consideration, a phenomenon known as ‘ambiguity                                                              9 Lieberman and Montgomery, 1988.  10  Mathews, 2002.  11  Clustering is the phenomenon whereby firms from the same industry gather together in close proximity. Economists explain clustering as a means for small companies to enjoy some of the economies of scale usually reserved for large ones. For more detail, see https://www.economist.com/news/2009/08/24/clustering.  5    aversion’. Many formal firms do not even decide not to invest in fragile states: they know not to pose the question. This uncertainty also affects the investors that a firm needs to attract. In the absence of an established industry, it is difficult for investors to evaluate the risk profile of an investment. Investors like to model risk based on data. But at the point at which pioneering firms need to invest, there are little or no data to guide their investment decisions. Again, the activities of the pioneering firms that do go ahead benefit the firms entering later because the risk premium of investors can be better calibrated by looking at the experiences of the pioneer firms. Who pioneers in fragile states? The World Bank Enterprise Surveys (WBES) is a unique source of economic data to address this question. The WBES uses a pool of 131,000 firms from 139 developing countries, including many that are listed as fragile and conflict-affected. The data have been collected since 2006 using a harmonized methodology to ensure cross-country comparability.12 Pioneering firms in the survey are identified as those that have innovated by introducing new products or services to the fragile state over the last three years. These include new products or services for an establishment’s main market, or a new or significantly improved process during the last three years. The new firms among the pioneers, which are of particular interest because of their financial needs, can be identified as firms that are less than five years old. Since pioneer firms play such an important role in economic development, it is unsurprising that there is a clear relationship between the proportion of firms that are pioneering, and the income level that the country has reached. In states that are not fragile, this pattern emerges strongly: the proportion of firms that are pioneers is highest in low-income countries, where the opportunities for harnessing scale and specialization in production abound. Conversely, pioneers decline as income rises, and countries become lower-middle income countries and upper-middle income countries (see Figure 1). This is important because it implies that the impediments are not so overwhelming that feasible public policy interventions would be doomed to ineffectiveness. If pioneering is already common in fragile states, it could be increased to the level appropriate for the state’s level of income. But fragility systematically undermines the process through which opportunities are discovered. At each income level, fragile states have a substantially smaller proportion of pioneer firms. And, to the extent that fragile states are among the poorer within each income group, then the effect of fragility might be even more pronounced than what the summary statistics suggest. However, while fragility systematically discourages pioneering, it does not kill it off. Despite the disadvantages, the evidence suggests that almost half of firms innovate in one way or another (see Figure 1, left panel). This is important because it implies that the impediments are not so overwhelming that feasible                                                              12 The surveys only represent a country's non-agricultural economy. Manufacturing and services are the primary sectors of focus, with services including construction, retail, wholesale, hotels, restaurants, transport, storage, and telecommunications. Mining and financial services, which attract significant investment in fragile states, are excluded. This also excludes firms that employee fewer than five workers, as well as wholly state-owned enterprises. 6    public policy interventions would be doomed to ineffectiveness. If pioneering is already common in fragile states, it could be increased to the level appropriate for the state’s level of income. Figure 1. Distribution of pioneering firms by income group FCS Non-FCS UMIC 44.44 55.56 UMIC 46.47 53.53 LMIC 42.63 57.37 LMIC 54.43 45.57 LIC 55.05 44.95 LIC 63.23 36.77 0 50 100 0 50 100 Pioneers Non-pioneers Pioneers Non-pioneers Source: Calculations based on the World Bank Enterprise Surveys (2018) Note: LIC = Low Income; LMIC = Lower-Middle Income; UMIC = Upper-Middle Income Pioneers overall tend to be larger, more experienced, domestic rather than foreign-owned, male- owned and run, and a result of a partnership rather than a sole proprietorship. They also tend to be more productive, rely more on training and international certification, and are more frequently encountered in manufacturing. Simple averages across groups of all firms and pioneers in fragile and non-fragile developing countries highlight these differences, which are largely consistent between fragile and non-fragile countries (see Appendix Tables A2 and A3).13 Firms’ particularities, however, might have little to do with pioneering, and more to do with the sector characteristics where they are active, i.e., the structure of production in fragile markets. For example, there may be a lot of domestic pioneers because foreign ownership is not prevalent in the sectors and markets where they are active. Controlled averages in a simple linear framework allow to net out such influences by looking at deviations from sector averages across countries. Such a calculation highlights how pioneers deviate from their peers across different contexts, which provides some inkling about how fragility interacts with who the pioneers are, and how they innovate.                                                              13With the exception of ‘public ownership’ and ‘skill intensity’, all the mean differences between pioneers in FCS and non-FCS are statistically significant at the 10 percent level. 7    Table 1. How different are pioneers in FCS? Controlled averages across firms Pioneers in Pioneers in Pioneers in non-FCS vs. FCS vs. FCS vs. Others Pioneers in Others Non-FCS Firm Characteristics (β) (γ) (β + γ) Constant Observations R-squared Age 2.128*** -3.354*** -1.226 19.49*** 67,763 0.031 Size (number of workers) 38.61*** -49.71*** -11.1 85.05*** 68,181 0.032 Manufacturing (=1) 0.126*** -0.270*** -0.144 0.513*** 69,313 0.025 Purshase of fixed assets (% of firms) 22.79*** -2.158*** 20.632 27.36*** 68,389 0.058 Internationally-recognized quality certification (%) 13.76*** -13.48*** 0.28 25.20*** 67,131 0.062 Skilled workers (in percent of all production workers) -4.421*** 0.860 -3.561 74.53*** 34,803 0.005 Firms with formal training to employees (%) 0.244*** -0.179*** 0.065 0.259*** 68,631 0.063 Firms with a female top manager (%) -0.0105*** -0.0516*** -0.0621 0.137*** 69,038 0.014 Foreign ownership (%) 2.414*** 5.464*** 7.878 5.220*** 68,478 0.008 Public ownership (%) -0.121** -0.0668 -0.1878 0.705*** 68,512 0.000 Sole proprietorship (%) -0.0332*** 0.134*** 0.1008 0.376*** 69,075 0.007 *** p<0.01, ** p<0.05, * p<0.1 Source: Calculations based on World Bank Enterprise Surveys (2018). Note: The regression equation is specified as follows: FCS where indicates the characteristics of firms i; is a binary indicator for pioneering; FCS is a binary indicator for operating in a fragile state, and are sector fixed effects. Since the specification includes an interaction term with two binary indicators, coefficient captures the average difference between pioneers in non-FCS and others i.e. non-pioneers operating anywhere (Column 1 from left to right) controlling for the sector of economic activity. Coefficient (Column 2) captures what is distinctive about pioneers in FCS, relative to pioneers in non-FCS i.e. the effect of fragility on pioneering. The sum of the coefficients , whose statistical significance is not directly estimated, summarizes the differences between pioneers in FCS and others i.e. non-pioneers anywhere (Column 3). What is noteworthy in that respect is that pioneers in fragile states have additional experience, but also the fact that they can innovate at smaller sizes, and rely less on fixed assets, international certification, and worker training (see Table 1). The level of development and diversification of these markets suggests an intuitive explanation: the scope for innovation is much broader. More opportunities for simpler innovation can be found in fragile states, even for firms that are sole proprietorships. We encounter many more pioneers active in services, but this is largely due to the complete absence of manufacturing activities. And while foreign firms are a minority, they are significantly more likely to pioneer relative to domestic firms, independently of the state’s level of fragility. Table 2. Type of innovation by pioneers Pioneering firms in Pioneering firms in non- FCS countries FCS countries Innovation New products or services introduced during the last 3 years (%) 75.21 73.02 Improved process introduced during the last 3 years (%) 75.56 80.78 New products or services also new for the firm's main market (%) 69.30 67.02 Source: Calculations based on World Bank Enterprise Surveys (2018). But do pioneers innovate differently in fragile states rather than elsewhere? Many tend to innovate simultaneously: those that introduce a new product in the market tend to also introduce a new process and enter new markets. But, importantly, in fragile states the share of pioneers that 8    introduce new activities, products, and services is greater than in other markets, which highlights the scope for market diversification and the potential for positive impact (see Table 2). 3. Why is it hard to be a first-mover in fragile states? Market conditions in fragile states pose additional barriers, relative to other countries Firms require five things to enable them to operate at scale and grow. First, they need access to markets and the know-how to produce for those markets. Second, they need a business-friendly investment climate, encompassing policy and regulation (including the quality of implementation), and the provision of public services and infrastructure. Third, to finance their growth, firms need access to finance—both working capital (including trade finance) and long-term capital. Fourth, they need appropriately skilled workers that are available in sufficient numbers for the scale of production required. And lastly, firms require the management skill to combine capital, labor, finance, and technology to produce for the market. However, all these elements are disrupted by the forces of fragility.14 First, fragility disrupts market access. Internally, conflict and intra-group mistrust reduces internal trade, shrinking the size of domestic markets. Since most fragile states are low income and many have low populations, the size of the internal market is an important constraint to private enterprise. Externally, fragility inhibits the creation of international trade networks–counterparties are reluctant to trade and enter into financial relationships with counterparties in fragile states. International efforts to combat the financing of terrorism and prevent money laundering have increased the financial and reputational risks that international banks face in dealing with fragile states. Many opt to reduce their risk exposure by avoiding financial relationships with these states.15 This also deters foreign direct investment (FDI) and other close commercial relationships, which are critical to the transfer of technology. Levels of FDI to fragile states are lower than to other states, and even more so when the FDI in extractive industries is excluded.16 Second, fragile states typically have weak governments, with limited capacity to raise taxes and hire competent staff, both of which are needed for the provision of public services and infrastructure. Fragile states are clustered near the bottom of the Doing Business rankings on regulatory quality and efficiency. Poor infrastructure - lag in the provision of power, roads, ports, - can potentially be addressed, but much of it, such as electricity generation and distribution is intrinsically political, and the business market for its services is small. Credible firms may balk at irreversible investments and so the actors that stand to make the greatest difference may be aid agencies rather than the DFIs. As a result of such problems, a higher share of firms, and by extension pioneers in fragile and conflict-affected countries, operate in services, where infrastructure requirements are less                                                              14 International Finance Corporation, 2019. 15 World Bank Group, 2017. 16 World Bank Group, 2018b. 9    burdensome. For instance, 42 percent of pioneering firms in FCS operate in the services sector, compared to only 25 percent in non-FCS. Conversely, only 37 percent of pioneering firms in FCS are in manufacturing, which contrasts with 64 percent elsewhere. Third, banking systems struggle to develop in fragile states. Fiscally strained governments often crowd out credit to the private sector, and banks lack the capital to expand lending. Reflecting this, only 27 percent of pioneering firms in FCS have a line of credit or a loan when compared with 41 percent of pioneering firms in countries which are not FCS. Consequently, a higher proportion of firms in FCS rely on internal funds to finance their investments and working capital. Unsurprisingly, a much higher proportion of pioneering firms in FCS identify access to finance as a ‘major constraint’ when compared with firms elsewhere (36 percent, as compared to 21 percent). Fourth, weak education systems and a large population of children lead to low educational outcomes, which limits the availability of an adequate labor pool.17 That constraint affects pioneers significantly more than any other constraints, since firms cannot rely on education systems for functional skills, which by the definition of innovation do not exist in the market. This is a problem for pioneers regardless of fragility. Hence, in both fragile and non-fragile contexts they try to economize on the need for skill, and employ a significantly lower proportion of skilled workers. However, due to the nature of pioneering, they are unable to find many of the skills they need, and so must invest significantly more in training programs than is the case with other firms. Training is a form of investment that must somehow be financed, and the distinct lack of access to finance in fragile states presumably curtails the ability to provide the training programs that all pioneers need. This is consistent with our finding that although pioneers everywhere invest more in training, those in fragile states provide significantly less training. But perhaps most crucial is the lack of entrepreneurial capacity to pioneer new enterprises and expand existing ones. Why is this missing? Enterprise sponsors face higher costs and difficulties when operating in fragile states, so they usually go in one of three directions. They may move their assets and business out of the country, joining the diaspora. They may limit the size of their activities to small, informal businesses that can ‘fly below the radar’ and do not need a lot of government engagement or support. Or they may engage in rent-seeking opportunities to take advantage of the many economic distortions in fragile states. The last response is particularly problematic for DFIs supporting enterprise growth, as once sponsors become associated with illegal behavior, most DFIs will not be willing to work with them. Netting out sectoral characteristics allows us not only to assess more precisely the ‘constraint premium’, but also to assess the relative severity of the constraints facing pioneers in FCS (see Figure 2), as opposed to pioneers in other developing countries. In fragile states, the top-ranked constraint facing pioneers is political instability. In addition, this is highly distinctive to fragile states; in other states, such a concern is negligible and insignificant. Will governments over- regulate or under-regulate? Will they  impose unpredictable and predatory taxes? Will they be pressured to make corrupt payments? Moving down from political instability, several constraints are ranked as approximately equally important. These are access to power, land, and finance; and                                                              17 World Bank Group, 2016. 10    the problems posed by corruption and disorder. Again, there is a large difference from pioneers in other markets: the proportion of those in fragile states reporting the matter as a ‘major constraint’ is two or three times as high. Strikingly, however, one potential constraint which donors have strongly emphasized—education—appears not to be important for pioneers, both in fragile and non-fragile environments. To the extent that an expansion of pioneering is critical to the generation of productive employment, this suggests that the greater emphasis put on schooling, training, and technical advisory in comparison with that put on other sources of comparative advantage for enterprise development, may need to be reassessed. Figure 2. Pioneers face more severe constraints than other firms in fragile situations  Inadequately educated workforce 20 Water insufficiencies 15 Access to land 10 5 Access to finance 0 Transportation -5 Crime, theft and Electricity disorder 28.45 Corruption Political instability Non-FCS FCS Source: Calculations based on the 2018 World Bank Enterprise Surveys. Note: The chart reports coefficients on the linear probability that a pioneer identifies the category as a major constraint, conditional on sector fixed effects. Empty field markers indicate coefficients that are not significant at the 1 percent level. The black box of missed opportunity Ultimately the fact that there are not enough large, formal firms in fragile states, with good access to markets, know-how, and financing, as well as with strong management, is reflected in the constraints that pioneers face, and in a missed opportunity in fragile markets. 11    Figure 3. The scope for market creation is large in fragile states Share of products absent from customs data, in percent 90 80 70 60 50 40 30 20 10 0 Average, FCS Average, non‐FCS Firm entry rate in an import market, Firm exit rate from an import market, weighted averages, in percent (1998–2014) weighted averages, in percent (1998– 2014) Source: The data employed in this paper are transaction-level customs data for the period 1997–2014. The data were collected by the Trade and Integration Unit of the World Bank Research Department as part of their efforts to build the Exporter Dynamics Database (EDD) described in Fernandes, Freund, and Pierola (2016). The sources for the data for each country are detailed at http://econ.worldbank.org/exporter-dynamics-database. Note: Destination countries in the EDD raw dataset follow the United Nations’ guidelines which recommend that countries of origin record their destinations as defined by the destination countries themselves. Therefore, as of end- 2011, each country has a potential set of 247 destinations (see Fernandes, Freund and Pierola, 2016). The group of FCS includes 36 countries. Fragile and conflict-affected states are defined as those that were FCS in 2014, which was the last year of the EDD. The firm entry rate equals the number of entrants (firms that did not import in year t-1 but did import in t) as a share of the number of importers. The firm exit rate is calculated as the number of exiters (firms that import in year t-1 but do not import in t), divided by the number of importers. The average rates are weighted using import values. 12    Customs data allow assessment of the degree of unfulfilled potential in fragile states or, in other words, the scope for market creation. These data enable identification of the sectors in which no cross-border transactions—whether imports or exports—are recorded for years. The World Bank Exporter Dynamics Database is a unique source of information to assess that gap.18 Drawing on data sets covering the universe of cross-border transactions that are obtained directly from customs agencies in 70 countries allows assessment of the extent to which products are missing from fragile states. The differences therein are striking: an average of 50–70 percent of product lines are missing from custom transactions in fragile states, as opposed to only 20–30 percent in the rest of the world. The gaps are persistent over time, and illustrate the absence of large-scale domestic production, as well as a missing market for foreign suppliers. In less distorted markets, such gaps reveal investment opportunities that result in high entry. That is not the case with fragile states. The level of their failures and risks in fragile situations is prohibitive for most foreign suppliers, as illustrated by lower import entry rates, relative to other markets. As fragility imposes fierce selection at the borders, only the exceptional few make it in, which, in turn, translates into greater resilience post-entry for those that are able to overcome the sizeable entry barriers (see Figure 3). What these dynamics suggest is also a significant ‘survivor bias’ in what we know about pioneers in fragile situations. The information we have comes from the few that succeed and survive, and who capitalize on capabilities that deviate substantially from entrepreneurs in the rest of the world. The characteristics of those who did not survive are unobserved, and, importantly, so are the characteristics of those that could potentially create markets with the support of financial institutions. First-mover advantages are not sufficient to secure credit As corroborated by an array of studies, there are benefits to being a first entrant in a market. In high-income markets, pioneers enjoy, on average, as much as 16 market-share points over late entrants, and 10 market share points over early entrants.19 Yet the verdict is still out: a growing body of empirical studies lends support to the claim that following is better than pioneering due to the costs outweighing the benefits.20 However, the sector and the market clearly make a difference. In fragile states, first-mover advantages can be significantly higher than in other markets due to the lack of diversification: simple innovations can be greatly rewarding. But costs can be more constraining too. In fragile states, the fixed costs of creating a new market, or restoring one disrupted by conflict, are significantly higher. Pioneer firms thus face a time inconsistency problem: the firm that bears the fixed costs by entering first will become the least competitive once the market is established. That is because such firms will be carrying the legacy of their market creation or restoration costs. This explains why, in fragile states, we mostly see                                                              18 Fernandes, Freund, and Pierola, 2016. 19 Tellis and Golder, 2001. 20 Golder and Tellis, 1993; Schnaars, 1994; Shankar, Carpenter, and Krishnamurthi, 1998; Boulding and Christen, 2003; and Hauser, Tellis, and Griffin, 2006. 13    pioneering firms enter in sectors where competition is limited, such as extractive industries and mobile telephony, where government controls market access through concessions and licensing. If a pioneering firm is able to get a concession or a license for a particular resource (mineral deposit, bandwidth, and so on), then the firm may be willing to bear the market creation costs, as it can recoup these costs later from the rents it will earn on the resource. It is much harder to attract pioneer firms to sectors where rents are liable to be competed away by later entrants. Yet it is this process of subsequent entry, the shrinking of rents, and the expansion of the sector that is the key public benefit from pioneering, which means that guaranteeing that a pioneer will be able to preserve a monopoly is not a solution. What do these fixed costs comprise? They are partly research costs—about the market, the regulatory framework, the quality and cost of labor, the quality of infrastructure, and so on which are unobservable costs until the first firm attempts production. Then other firms can free-ride on the market information revealed by the entry of the first firm. Fixed costs are also partly relationship-building costs. The first firm to enter bears the costs of helping regulators, customs agents, banks, trading partners, and others to climb the ‘learning curve’. Governments frequently do not know how to tax or regulate a sector until the first business starts up. Or even if they have taxes and regulations on paper, there are start-up costs as they learn to implement. The weak market connections and financial market connections make these start-up costs much higher in fragile states. Table 3. Access to finance by young pioneers (aged 5-years or less): Controlled averages across firms Firms with a bank Proportion of Proportion of Proportion of Proportion of Cost of capital per loan/line of credit investments investments working capital working capital unit of value added (%) financed by internal financed by banks financed by internal financed by banks funds (%) (%) funds (%) (%) Pioneers 8.917*** -8.600*** 5.388*** -9.257*** 6.934*** -0.556** (1.078) (1.564) (1.202) (0.807) (0.580) (0.267) Pioneers in FCS -13.72*** 13.37*** -11.33*** 13.46*** -9.894*** 0.657 (1.691) (2.244) (1.724) (1.267) (0.910) (0.484) Constant 26.12*** 74.54*** 12.77*** 76.11*** 11.76*** 1.786*** (0.911) (1.380) (1.060) (0.687) (0.493) (0.179) Observations 7,508 2,757 2,757 7,499 7,499 1,390 R-squared 0.023 0.021 0.021 0.038 0.045 0.004 Sector fixed effects Yes Yes Yes Yes Yes Yes Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Source: Calculations on World Bank Enterprise Surveys (2018). Note: The regression equation is specified as follows: FCS where indicates the figures in columns reported by firm i; is a binary indicator for pioneering; FCS is a binary indicator for operating in a fragile state, and are sector fixed effects. Coefficient reported under ‘Pioneers’ captures the average difference between pioneers in non-FCS and others i.e. non-pioneers operating anywhere, while controlling for the sector of economic activity. Coefficient reported under ‘Pioneers in FCS’ captures what is distinctive about pioneers in FCS, relative to pioneers in non-FCS i.e. the effect of fragility on pioneering. 14    And they frequently include labor training costs – in the absence of other firms in the market, the specific skills required are unlikely to be available. But firms cannot prevent trained labor from being poached by later entrants, so they are left with a cost of training labor that is of benefit to all firms in the market. Access to finance can mitigate some of pioneers’ start-up costs if there is enough information and predictability for pioneers to capitalize on their potential. First-mover benefits in fragile states should be reflected in higher access to capital at lower cost. And this is the case for most developing countries, yet in fragile states the reality is much bleaker. The higher probability of young pioneers getting a bank loan or credit is reversed in fragile states where young pioneers are less likely, by half as much, to have access to credit (see Table 3). The shares of their investments and working capital financed by banks are disproportionally lower too, which forces them to rely on internal funds, and thus start up with a much higher probability of default. While pioneers in other countries are able to capitalize on their potential in terms of lower cost of capital per unit of value-added, pioneers in FCS face the same credit terms as everyone else. In both cases, first-mover costs are significantly higher. Nor is this surprising: banks avoid exposing themselves to radical uncertainty. Although all firms operating in fragile states face distinctively higher uncertainty due to political instability, pioneers face an additional layer of ‘unknown unknowns’. Figure 4. Labor productivity – pioneering versus non-pioneering firms Labor productivity Labor productivity (Average, in thousands of 2009 dollars) (Median, in thousands of 2009 dollars) 40 14 35 12 30 10 25 8 20 6 15 10 4 5 2 0 0 Pioneering Other firms Pioneering Other firm Pioneering Other firms Pioneering Other firm firms firms firms firms FCS Non-FCS FCS Non-FCS Source: Calculations using World Bank Enterprise Survey data. Note: Labor productivity is the ratio of value added by the number of workers. These conditions are reflected in lower productivity premia for pioneers in FCS: higher returns are offset by important costs and uncertainty. Estimates from the World Bank Enterprise Survey suggest that the productivity of pioneering firms in FCS is lower than that of their peers in non- FCS. As illustrated in Figure 4, the median value added per worker in pioneering firms in FCS ($6,986) is lower than that of pioneering firms in non-FCS ($11,651). 15    4. Expanding the DFI toolkit to support pioneering firms in fragile states The fragile state needs to overcome these impediments to building clusters and production webs. While this is not the only role of DFIs in fragile states, it is the fundamental one, and it is difficult. From the above characterization of the problem, we can split it into three components. One is the need to offset the known costs of being a pioneer, rather than entering once the cluster or web has reached commercial viability. The second is offsetting the known risks that can be expressed in some insurable form of probability. The third is offsetting the ‘unknown unknowns’ that generate radical uncertainty. The analysis above suggests three key roles for DFIs to support greater entry of pioneer firms. The first is to develop subsidy mechanisms that defray some of the initial entry costs, thus correcting for the externality problem described above. The second is to provide technical assistance and capacity building for market creation activities which range from labor skills development to support for government regulators and banks so that they better serve the market. The third is to be willing to confront the radical uncertainty of pioneering investments in these markets, and to invest in pioneer firms despite the absence of information concerning the risk profile of the activity. DFIs are currently organized around doing deals with firms that come with proposals: the core activities are selection, structuring, and risk assessment. This is a useful skill set for working in fragile states, but not by any means the most needed skill set: a demand-led approach misses the fundamental need to support the entry of pioneering firms that might be ambiguity-averse, so that deals can be generated. Notably, all these activities require DFIs to move beyond their traditional range of activities—into the blending of concessional and commercial financing, capacity building, greater coordination across investments, and investing in the face of unquantifiable risk and uncertainty. Addressing the sequential costs in new clusters and webs To offset the problem of known sequential excess costs of pioneers, the first requirement is to improve the knowledge base for DFI decisions. DFIs need sector diagnostics. By this we mean an estimate for each activity that is potentially a candidate for an early cluster, or an early production web, in the country. A useful guide in determining the possible candidates for becoming early clusters and webs is to find a few countries that one or two decades ago had economies somewhat similar to that of the fragile economy now, but which have been through a significant transformation. The sequence that they followed, with its successes and setbacks, can serve as a candidate list to be evaluated. This approach has two advantages over more complex techniques. One is realism: the nature of the problem is one of radical uncertainty— each national transition is unique, and all attempts to reduce it to probabilities are illusory. The other is that such a concrete comparison can more readily serve as the basis for a narrative that investors can readily assess for credibility. This is better than a technique too complex and proprietary for them to trust. We take this up further below. 16    Assessing the public benefit from new clusters and webs The same sector diagnostic that estimates the costs of establishing a new cluster or web can be used to assess its public benefit. The key non-commercial benefits in a fragile state are jobs and tax revenues. The former stabilizes the society; the latter strengthens the sinews of the state. Whereas the costs will need to be estimated for each firm in the sequence (since each firm will need to be compensated), for the benefits, what are likely to be the most pertinent of those generated once the sector reaches viability? This is an important point: the procedure for estimating the public benefit generated by each individual pioneer grossly understates the true benefit because the sector will not scale up until it has reached the point of commercial viability. The first entrant to the Bangladeshi garment industry employed, at most, a few hundred workers. This was not the public benefit: the public benefit was that the first entrant triggered what became a vast sector. In the end, what matters is a comparison of the pump-priming public costs of getting the cluster or web established relative to the public benefits in terms of jobs and tax revenues. In assessing the benefit from jobs, some valuation has to be placed on them. Again, comparison with some country that has already gone through transformation can be helpful. What matters is not the initial wage, but the likely trajectory of wages in a transforming economy, in comparison with the likely earnings trajectory of near-stagnation if fragility persists. As indicated in the preceding figure (see Figure 4), in countries that have stabilized, labor productivity is markedly higher. This puts a lower bound to the estimate of the public benefit, but an additional element should be added —something for the contribution of employment growth to reducing the existential dangers of fragility. In effect, this comes from a socio-political assessment by appropriate country and conflict specialists. Any quantification will be notional, but this is better than assuming that the benefit is zero. Sometimes, a useful exercise will be to ask how big would the contribution need to be to justify public money being used to establish the sector. But although, in principle, the nature of the decision seems difficult, in practice it will usually be straightforward: at any one time, only a few sectors will be candidates, and DFI capacity in the country will limit what can be attempted. The diagnostic is not a once-and-for-all guide: it launches a process of learning by doing, and so will need to be revised periodically. Suppose that a proposed project meets the requirements that once a cluster or web has reached scale it will be commercially viable, and that the public benefit of establishing it exceeds the costs of covering first-mover disadvantage. Evidently what is then needed is a public subsidy of pioneer firms to cover these costs. Supporting pioneers with subsidy mechanisms We can characterize the situation as pioneering firms face heightened private costs which generate social benefits beyond the firm – their activities and the information on risk that they generate by initial entry benefit the entire market (see Carter, 2015). The discovery process they undertake generates information on risk and reduces uncertainty, benefitting all firms and investors. This provides a rationale for subsidizing initial entry costs to correct for this externality. Such subsidies could in theory come from governments in these states, but they rarely have the fiscal or administrative capacity to do so in a transparent way. Since the governments of fragile states lack 17    the finance to provide the necessary subsidy, the only way they can provide compensation for initial losses is to grant tax waivers. Such tax waivers only help the firm if it becomes profitable, and even if everything were known, this would take time. So, the tax waivers must extend for a long period. But in the context of a fragile state, distant future commitments are heavily discounted. The only recourse of the government is to offer larger and yet more distant, tax concessions. Further, such concessions are very easily gamed by firms: when a firm reaches the time that tax becomes due, it could close and set up a new firm that is eligible for a concession. In the process, the ‘solution’ to this problem undermines the other overriding priority of escaping from fragility and strengthening the key sinews of the state. There is, therefore, a role for DFIs, which are often the main source of capital for pioneering firms. That role is providing subsidies as part of DFIs’ financing role. Blended finance mechanisms, through which DFIs combine commercial-term financing with subsidies linked to specific costs, benefits, or risks, are a suitable mechanism for providing this subsidy. The DFIs have advanced rapidly in developing and deploying blended finance tools over the past few years, and have developed common guidelines on how to do it in a disciplined way.21 The rationales that we have discussed here for subsidizing pioneering firms in fragile states fit well into the objectives of the significant donor funding pools for blending. Most notably, the last International Development Association (IDA) replenishment allocated $2 billion to IFC and $0.5 billion to the Multilateral Investment Guarantee Agency (MIGA) to cover the concessional component of various blended structures. IDA donors have put a strong priority on using these funds in fragile low-income states. One approach would be to actively seek out firms willing to be pioneers with subsidy support from DFIs. This requires an active phase of search, most likely through a combination of an open call for proposals and pro-active invitations to those firms likely to be suitable. This is well within the skill set of someone with an MBA, but it implies a switch in DFI activity from passive evaluation to pro-active opportunity generation. However, the switch could create further delays in a process that is already too long relative to the political time horizons of client governments. Once an activity has been selected and agreed with the government, to impose greater discipline on each stage, the entire process from calls for proposals and search for investors to a ‘gathered field’ and the first signed deal, should be time-bound. Subsidy mechanism design It is important to design the subsidy allocation mechanism to support market creation, and not undermine it. Current DFI processes for allocating a subsidy can be criticized as opaque and idiosyncratic.22 This only adds to the uncertainty facing pioneer firms. Subsidy should also avoid creating an advantage for the recipient that goes beyond the ‘first mover’ costs that it bears. Over the past two years, DFIs have converged on a set of common guidelines for the use of blended finance, which aims to bring proportionality, minimum subsidy, and transparency to the process.23                                                              21 DFI Working Group on Blended Concessional Finance for Private Sector Projects, 2017.  22 Kenny, 2018. 23 DFI Working Group on Blended Concessional Finance for Private Sector Projects, 2017. 18    However, there is need for further innovation in the design of blending structures to ensure that subsidies are well targeted to the upfront costs faced by pioneering firms. Subsidy should be frontloaded to reflect upfront costs, and taper off so that they do not give ongoing competitive advantage to these firms after their initial costs have been met. Options include: (a) a capital cost subsidy—as capital investment is frontloaded, this clearly meets the design criterion; (b) lower interest rates on debt—if borrowing tapers off due to amortization, this meets the design criterion, and is the most commonly used structure by DFIs; and (c) patient capital—equity with low expectations of return in the initial years. Compared to debt, the implicit subsidy in patient equity is less transparent. On the other hand, equity infusions improve the capital structure of the firm, and give it the opportunity to leverage DFI financing with additional debt on commercial terms. Whichever structure is used, the question arises of how to select pioneer firms to benefit from the subsidy. Where there is a defined market/business opportunity (such as for supplying power to a government utility) the government can run an auction, based on bids for the amount of capital subsidy or tariff enhancement (for example, the World Bank Group’s Scaling Solar program in Africa24). More active search processes, as described above, can bring greater transparency to firm selection. Where there is no defined market/opportunity, DFIs need to screen possible pioneer firms for their potential to contribute to market creation and provide tailored support that will promote market development. Technical assistance and capacity building DFIs can also provide support to firms through providing subsidized advisory services that support job training, technology transfer, market assessment, and other costly activities which pioneering firms need to undertake. They can also provide technical assistance and capacity building to governments to help create a more certain market for pioneering firms. This can include the establishment of necessary market regulations, tax regimes, administrative processes, and so on. This all serves to reduce the cost and uncertainty that pioneer firms face in entering a new market or providing a new good or service. Addressing the problem of risk and uncertainty Addressing the high risks inherent in individual investments in fragile states is the challenge most familiar to DFIs. The conventional DFI modalities for covering risks are risk-sharing through equity investment, insurance instruments such as first-loss provisions, various types of risk insurance and hedging mechanisms, and loan rates set below risk-corrected market rates. To the extent that these are not financially sustainable for DFIs, they depend on the provision of grant funding or risk-bearing capacity from beyond their balance sheets. For example, the IDA FCS facility bears the risk of forex and government performance guarantees, and enables MIGA to insure otherwise uninsurable political risks. Currently, the money provided from the IDA Private Sector Window to IFC is designed to be used for risk-reduction, in the form of guarantees, hedging, and risk insurance. However, risk is only one of the impediments: first-mover disadvantage and                                                              24 For more information, see Scaling Solar website at https://www.scalingsolar.org/. 19    radical uncertainty each need to be offset. DFIs need more than this one tool of risk-reduction to achieve the objective of creating markets in fragile states. For example, IFC can combine PSW guarantees with concessional loans and equity structured to defray upfront costs and delay the returns to the PSW. One of the major risks for private investors is government behavior. Although ostensibly MIGA is an insurance agency, since it recovers from governments most of the money it pays out in claims to investors, MIGA insurance is primarily a mechanism by which governments are able to reduce investor risk by making credible commitments to provide compensation for breaches of contract without recourse to the courts. Addressing radical uncertainty Radical uncertainty abounds, and not only in fragile states. A key implication of uncertainty for DFIs is that too little is known to be confident that any particular intervention will work: a new industrial zone can appear to be promising, be well-equipped, and yet firms do not come to it. This poses a simple problem for potential investors who can simply opt not to invest, but a complex problem for DFIs that are mandated to find investment opportunities in this context. First, consider the simple problem faced by potential investors. As discussed above, they typically resolve radical uncertainty by avoiding a decision. Yet in many contexts of radical uncertainty, decisions still get taken. Faced with an uncertain situation, people are eager to understand it: they pose the question ‘what is going on here?’25 This appetite for explanation is met by narratives that seem to provide a coherent and credible account of the recent past (see Shiller, 2017 for a good discussion on narrative economics). The narrative explanation circulates within a network, and as people start to believe it, this collective understanding gives reassurance. Even highly-sophisticated financial markets function like this, but in fragile states the process is of the essence. Not only do fragile states abound in uncertainty; but the small size of their economies does not justify substantial expenditures to forge an independent, evidence-based, understanding. These are the archetypical environments in which firms rely upon narratives as the basis for their decisions. So, instead of avoiding a decision altogether, potential investors often base a decision on some simplifying narrative that appears to account for ‘what is going on?’ The recent history of investor sentiment in developing countries abounds in such narratives: investor-oriented phrases such as ‘Asian Tigers’, ‘BRICS’, ‘Emerging Markets’, ‘Frontier Markets’, ‘the Hopeless Continent’, and ‘Africa Rising’ were simplified explanations of the recent past that became widely accepted in investor circles, and for a while were presumed to have predictive power. Most such narrative explanations are merely approximations that eventually so mis-predict events that they are abandoned. The ‘Dot Com’ bubble, and ‘the Great Moderation’ are recent examples of narratives that took hold in highly sophisticated markets and yet rapidly proved embarrassingly flawed. But narratives about future spatial decisions have the unusual feature of potentially being self-fulfilling. This is because the location of future agglomerations is highly indeterminate: there are thousands of ‘rational choice equilibria’, any one of which will happen if people believe that it will. Further, since firms can readily understand that this is the case, what                                                              25 Kay and King, forthcoming. 20    they need to know is not whether the narrative is correct in some objective, technical sense, but rather whether it is the narrative that other firms believe. If others believe it, then it is rational to accept it. When firms look at a fragile state, there is already a dominant narrative that provides a reasonably good explanation of the recent past: the country is a ‘basket case’. This means that there is some enduring characteristic of the country, which is not further specified other than in vague terms such as ‘corruption’ or ‘violence’, that causes repeated failure. This explanation is repeatedly reinforced by the extensive marketing of images of catastrophe on which the business models of the development NGOs are dependent. The ‘basket case’ narrative is the polar opposite to that of ‘Asian tigers’, and each is self-fulfilling. In Asia, investor confidence resulted in high growth rates, which confirmed the confidence. In fragile states, a lack of investor confidence results in stagnation which confirms the lack of confidence. Now we turn to the complex problem faced by DFIs. In order to find firms to invest in, DFIs need entrepreneurs and investors for whom the simple narrative of ‘basket case’ is false, but they need to do this in the context of themselves not knowing what they need to do in order to be sure of overcoming first-movers disadvantage. The solutions to these intertwined problems are to embark upon a phase of discovery as to what instruments work best in which context, while adopting a range of measures to reset investor narratives. Launching a phase of discovery Faced with the mandate to generate job growth in fragile states, a task for which the knowledge base is currently inadequate, the responsible action is evidently to ignite a process of rapid social learning. Governments usually resist admitting that they do not know what to do, and are afraid of learning through experimentation because it inevitably brings some failures and exposes them to criticisms from civil society. It is essential that the leaders of DFIs educate their stakeholders about the true nature of the task, and explain that the only way that governments can protect themselves is to grant DFIs sufficient operational independence that they can take risks and make mistakes, and give them resources not linked to immediate investment deliverables, to enable experimentation and upstream work to develop investment opportunities. An important role of a DFI with independence to deliver a mandate is that of taking responsibility for and incurring the costs of this process of social learning. Absorbing pertinent knowledge can help, but DFIs learn primarily from their own experience in experimenting with different approaches. Academic knowledge is of some use: the four pertinent fields are the political economy of fragility; the economics of spatial agglomeration; the industrial economics of low-income countries; and the macroeconomic analysis of radical uncertainty. In all four there have been useful recent advances. Escaping the Fragility Trap26 emphasizes the importance of ‘scaffolding’: supporting measures that can work in the context of getting started, even though they are not found in contexts where many firms are already thriving. Venables27 shows why market forces tend to perpetuate spatial failure: low wages are usually not enough to compensate for the lack of a cluster of firms, and the                                                              26 LSE-Oxford Commission on State Fragility, Growth, and Development, 2018. 27 Venables, 2018. 21    location fills up with low productivity activities. Starting from the recognition that in low-income countries a salaried wage job lifts people from poverty to the middle class, Woodruff summarizes the evidence on the impediments to private-sector job-creation.28 Kay and King provide an outstanding analysis of radical uncertainty.29 The collective experience of other DFIs is potentially of considerable use. Indeed, in business, learning from peers is the primary dynamic of technical progress. With over 40 DFIs, several with distinctive specializations, there is a substantial pool of variation. As discussed below, social learning among DFIs should be an entirely cooperative process, and go faster than among commercial firms, but it requires stronger mechanisms that provide effective opportunities for learning. But inevitably, learning will occur primarily through each DFI taking the risk of trying new approaches in new places. Since some of these will fail, the essential steps for DFI owners to take are authorizing this risk-taking, encouraging it by rewarding innovative failure as well as success, and building a systematic process in which staff who manage innovations share what they have learned. Resetting investor narratives The task facing DFIs is how to reset negative and self-fulfilling investor narratives such as the ‘basket case’ narrative. Three practical responses by DFIs can complement each other. One is to make a commitment. A second is to disrupt damaging narratives with evidence that is incompatible with them. A third is to replace a damaging narrative with a better one. We consider them in turn. Commitments that reset beliefs A powerful demonstration of the potency of a commitment was that of President Draghi of the European Central Bank (the ECB) during the euro currency crisis. He publicly announced that the Bank would do ‘Whatever it takes,’ to secure the future of the euro. By making it a public statement, Draghi left himself no personal room for failure: the statement was simple and completely unambiguous. Hence, it was at once apparent that if the euro fell apart, having made this commitment, he himself would have to resign. This, in turn, reassured asset holders that he would use his considerable power to intervene in markets. As a consequence, these three words rapidly transformed the trajectory of the euro even before the intervention: it worked by resetting expectations. Further, the phrase acknowledged ignorance: the unambiguous implication was that the ECB would experiment with innovations until it found something that was sufficiently effective to accomplish the goal. The ECB duly innovated with ‘Outright Monetary Transactions’. An analogous announcement was Deng Xiaoping’s 1978 narrative that ‘it doesn’t matter whether a cat is black or white as long as it catches mice’. In the context, this was readily understandable as an acknowledgement that the government was determined that China would catch up, but was not committed to any particular means of doing so. The Chinese government ignited a process of regional experimentation that has generated rapid social learning.                                                              28 Woodruff, 2018. 29 Kay and King, forthcoming.   22    What might be an equivalent commitment by DFIs toward fragile states? It has to be simple and unhedged: for example, had Draghi said ‘We will do whatever we can’, instead of ‘Whatever it takes’, it would have had little or no effect. The former phrase, though apparently a commitment, would clearly have left the president free, in the event of a failure of the euro, to exonerate himself by saying ‘We did all we could’. So, the equivalent DFI commitment has to leave the DFI with an unqualified exposure to achieving the commitment. This tells us that the commitment must be within the power of the DFI. The temptation is for the DFI to respond to this by committing to actions rather than outcomes, but this is likely to be ineffective since it leaves firms facing uncertainty as to whether the actions will succeed. Hence, the commitment must be to some outcome that matters for the firm, but is believed by the firm to be within the power of the DFI. The italicized qualification is important: Draghi’s phrase worked so well partly because asset holders, although unsure of his exact powers, recognized that whatever his powers were, he now had a personal interest in pushing them to the limit (as indeed proved to be the case). Firms will similarly be unsure of a DFI’s powers: most obviously, how much money will a DFI be willing to spend in trying to make a commitment work? Suppose, for example, that the DFI’s public commitment is ‘We will do whatever it takes to get this industrial zone up to 10 operating firms by 2023’. Precisely because this puts the DFI so firmly on the hook, it may well be sufficient to anchor expectations: a firm investing in 2020 can be highly confident that it will only have three years to wait before the cluster is at a viable scale. In turn, this puts bounds on the compensation required for first mover disadvantage. To make the commitment viable, the DFI may need to get commitments from the government that it will implement supporting actions, and more importantly, refrain from actions that might cause damage. As discussed above, for commitments by DFIs to work, they need to be credible and so plausibly within their powers. This, in turn, confines DFI commitments to specific, time-bound goals such as the minimum size of a cluster. An overarching goal, such as escape from the fragility trap, cannot become either a DFI or even a donor commitment, since it depends upon events beyond their powers. Falsifying a damaging narrative The ‘basket case’ narrative is very far from being a profound analysis of fragility. The notion that there is some unchangeable characteristic that dooms the society has repeatedly been falsified by evidence that provided a more credible narrative. In the 1950s, the country that was most commonly the recipient of the ‘basket case’ explanation was China.30 The rise of a few East Asian countries during the 1970s decisively falsified the ‘basket case’ narrative for the entire region. This helps to explain why, in the past decade, Asian investment in fragile states has increased far more rapidly than European investment. The ‘This country is a basket case’ narrative is much less credible to Asian firms, partly because within living memory it was also applied to their own countries, and partly because they are relatively new arrivals in the current fragile states. Having been falsified at home, they are somewhat immunized against the ‘basket case’ account of events. In its place, they are more inclined to adopt an equally simple                                                              30 See Collier (2017) for a survey of cultural explanations for national failure. 23    narrative explanation of ‘what is going on here?’ Their default option narrative is ‘this country is like us 40 years ago, so we know how things will evolve’. In contrast, European firms, with long exposure to persistent fragility, are more likely to adopt the stance ‘I’ve seen all this before, and listened to all the false hopes.’ The rapid economic turnaround in Myanmar has evidently been helped by the obvious narrative that it is ‘the latest East Asian transformation’. Hence, the most effective way of falsifying the ‘basket case’ narrative for Africa, is a ‘big push’ to generate some star performers. Many have played with the idea, resorting to governments, international trade, or the development industry to achieve that (see Hoff, 2000). One approach for DFIs is to concentrate resources on a few countries in the region that are at a pivotal political moment that makes rapid change feasible. An analogous approach for a single country is to focus on one city or zone. In each case, once some momentum is achieved, it needs to be sustained and reinforced until it becomes seen by investors as irreversible. This strategy of creating ‘stars’ has the added advantage of generating social learning among the peer group of countries. Launching a new narrative DFIs communicate with investors both by what they do and what they say. Narratives are the most effective mode of speech because they are memorable. Well-chosen costly actions can reinforce narratives: costly actions generate credibility (as explained by the Spence Theory of Signalling). 31 Combined effectively, narratives convey precision of meaning, while actions convey credibility. Evidently, a communication strategy to reset investor expectations is best agreed with the government, with a common narrative message matched by supporting actions both by DFIs and government. 5. Competing and cooperating for a common objective Competition between DFIs is healthy. It is an important stimulus for energy and discipline, and drives the innovation which is vital for social learning. But not all forms of competition are healthy. The public purposes of DFIs are not identical, but they have a considerable overlap. This makes some forms of competition wasteful, and provides scope for cooperation. If a firm is willing to do a deal with one DFI, no public purpose is served if another DFI undercuts it: indeed, the additional public resources used in a more generous deal are entirely wasted. As discussed above, social learning is a natural public good for the DFI community. Each DFI that operates in a country will need a diagnostic about the opportunities for clusters and production webs, and there is much to be gained through this being a public good. Indeed, only if DFIs have a shared analysis of the opportunities in a country can they work together to get the clusters and webs to grow to the necessary minimum scale: each DFI has a distinctive network of firms potentially willing to invest. By their nature, these relationships cannot be shared, and so the initial diagnosis is the appropriate common resource. IFC has recently pioneered a process for such private sector                                                              31 Spence, 1973. Spence, Michael, 1973. "Job Market Signalling", Quarterly Journal of Economics, Vol. 87, No. 3. (Aug., 1973), pp. 355-374. 24    diagnostics, conducted in collaboration with the World Bank. It would be useful for the DFI community to determine whether this is suitable to function as a common methodology, with each DFI having the opportunity to undertake a country or sector diagnostic on behalf of the community. This would provide a common format for the diagnostic that all pertinent countries could perform much more rapidly than IFC could on its own. One of the markets that DFIs are collectively trying to create is for the asset class of illiquid projects in low-income markets. Through such an asset class, illiquid investments become liquid and so much more attractive. But for the projects generated by different DFIs to gel into a common asset class, they need to conform to some standard features such as the instruments used to reduce risk. Currently, there are a plethora of small schemes that greatly increase the information costs to potential purchasers, and the idiosyncratic features are not amenable to price discovery. To address this, multilateral development banks are currently exploring ways to standardize appraisal standards, documentation, and so on. This builds on past efforts to standardize environmental, social, and governance standards based on IFC’s Performance Standards, which are now the market standard for project finance (the Equator Principles). 6. Concluding remarks International attention on fragile states tends to focus on the difficult challenge of building stable states. But without a growing economy that generates jobs and tax revenue, this is likely to continue to be a futile effort. Large, formal firms are critical to the stabilization and growth of fragile states, but are lacking in most places. Substantial costs, risks, and uncertainties confront pioneering firms, preventing them from filling this gap. It is therefore important that DFIs collaborate to bring sufficient resources to bear to overcome these costs, risks, and uncertainties. It is not enough to support individual firms, or individual reforms, or individual pieces of infrastructure; DFI interventions need to be sufficiently deep and wide to foster the creation of new markets, with their clusters of firms and supporting infrastructure, value chains, and trading networks. To do so, DFIs must help dispel the radical uncertainty concerning whether markets can develop, and states can escape the fragility trap. In short, DFIs need to play a big enough role to change the narrative about the opportunities for private investors in these states. DFIs have done it before—IFC rebranded low- and middle-income economies from ‘Third World’ to ‘Emerging Economies’ in the 1980s, triggering a wave of private capital inflows. DFIs can have the same transformative effect on perceptions of fragile states today. To do so, they need to scale up the use of blended finance to support pioneering firms, following the shared guidelines developed in 2017; build a common understanding based on country and market diagnostics; and undertake complementary investments, technical assistance, and risk mitigation to enable the creation of industrial clusters and production webs. Initial steps to coordinate and harmonize approaches need to go farther, faster. Stakeholders must recognize that this is a risky endeavor, and provide the political backing, resources and space for experimentation and error. Also, the risk- 25    taking appetite of DFI boards and shareholders must match the risk involved in changing the narrative and trajectory of fragile states. Only this will reduce the risk that fragile states remain an obstacle to achieving sustained global prosperity. Bibliography Basu, Kaushik. 2014. “Fiscal Policy as an Instrument of Investment and Growth.” World Bank Policy Research Working Paper No. 6850, Washington DC: The World Bank Boulding, William, and Markus Christen. 2003. “Sustainable Pioneering Advantage? Profit Implications of Market Entry Order.” Marketing Science 22 (3): iii–436. Carter, Paddy. 2015. “Why subsidize the private sector?”, London: Overseas Development Institute. Collier, Paul. 2007. The Bottom Billion: Why the Poorest Countries Are Failing and What Can Be Done About It. Oxford: Oxford University Press. ———. 2017. “Culture, Politics, and Economic Development.” Annual Review of Political Science 20 (1): 111–25. DFI Working Group on Blended Concessional Finance for Private Sector Projects. 2017. Summary Report. October. Washington, DC: International Finance Corporation (IFC). LSE-Oxford Commission on State Fragility, Growth and Development. 2018. “Escaping the Fragility Trap.” LSE-Oxford Commission on State Fragility, Growth and Development. London and Oxford, UK. Golder, Peter N., and Gerard J. Tellis. 1993. “Pioneer advantage: Marketing logic or marketing legend?” Journal of Marketing Research 30 (2): 158–70. Fernandes, Ana Margarida, Caroline Freund, and Martha Denisse Pierola. 2016. “Exporter Behavior, Country Size and Stage of Development: Evidence from the Exporter Dynamics Database” Journal of Development Economics 119 (C): 121–37. Hauser, John, Gerard J. Tellis, and Abbie Griffin. 2006. “Research on Innovation: A Review and Agenda for Marketing Science.” Marketing Science 25 (6): 551–765. Hoff, Karla. 2000. “Beyond Rosenstein-Rodan: the modern theory of underdevelopment traps.” Washington, DC: The World Bank. Kabuth, Oliver. 2003. “Pioneering Versus Following in Emerging Markets: The Case of the Automotive Industry in China and Brazil.” Doctoral Dissertation, University of St. Gallen, Switzerland. Kay, John and Mervyn King. Forthcoming. Radical Uncertainty. Kenny, Charles. 2018. “Aid Transparency and Private Sector Subsidies at the IFC.” Center for Global Development (blog). January 17. https://www.cgdev.org/blog/aid-transparency-and-private-sector-subsidies-ifc. 26    Kremer, Michael. 1993. "The O-Ring Theory of Economic Development". The Quarterly Journal of Economics. Oxford University Press. 108 (3): 551–575. International Finance Corporation (IFC). 2019. Generating Private Investment in Fragile and Conflict-Affected Areas. Washington DC: IFC. Lieberman, Marvin B. and David B. Montgomery. 1988. “First‐mover advantages.” Strategic Management Journal 9 (S1): 41–58. Mathews, John A. 2002. “Competitive Advantages of the Latecomer Firm: A Resource-Based Account of Industrial Catch-Up Strategies.” Asia Pacific Journal of Management 19 (4): 467–88. Murphy, Kevin M, Andrei Shleifer, and Robert W Vishny. 1989. “Industrialization and the Big Push.” Journal of Political Economy, 97 (5): 1003-1026. Nakata, Cheryl and K. Sivakumar. 1997. “Emerging market conditions and their impact on first mover advantages: An integrative review." International Marketing Review 14 (6): 461– 85. Schnaars, Steven P. 1994. Managing Imitation Strategies. New York: Free Press. Shankar, Venkatesh, Gregory S. Carpenter, and Lakshman Krishnamurthi. 1998. “Late Mover Advantage: How Innovative Late Entrants Outsell Pioneers.” Journal of Marketing Research 35 (1): 54–70. Shiller, Robert J. 2017. Narrative Economics. AEA Presidential Address for January 7, 2017, Chicago. Yale University. Spence, Michael. 1973. "Job Market Signaling." Quarterly Journal of Economics 87 (3): 355–74. Tellis, Gerard J. and Peter N. Golder. 2001. Will & Vision: How Latecomers Grow to Dominate Markets. New York: McGraw-Hill. Venables, Anthony J. 2018. “Globalization and urban polarization.” Review of International Economics 26 (5): 981–96. Woodruff, Christopher. 2018. “Addressing constraints to small and growing businesses.” Working Paper, International Growth Centre, Oxford, UK. World Bank Group. 2016. World Bank Group Engagement in Situations of Fragility, Conflict, and Violence. Washington, DC: World Bank. ———. 2017. De-risking and other challenges in the emerging market financial sector: findings from IFC’s survey on correspondent banking. Washington, DC: World Bank Group. ———. 2018a. Poverty and Shared Prosperity 2018: Piecing Together the Poverty Puzzle. Washington, DC: World Bank Group. ———. 2018b. Global Investment Competitiveness Report 2017/2018 : Foreign Investor Perspectives and Policy Implications. Washington, DC: World Bank Group. ———. 2018c. “Enterprise Surveys (database).” World Bank Group, Washington, DC. http://www.enterprisesurveys.org/. 27    ———. 2019. “Doing Business (database).” World Bank Group, Washington, DC. http://www.doingbusiness.org/en/data. Zashev, Peter and Stefan Ehrstedt. 2010. “The First Mover in New Emerging Markets: Balancing Risks versus Opportunities in the Case of Belarus.” Journal of East-West Business 16 (3): 201–30.   28    Appendix: Results from the World Bank Enterprise Surveys, 2010–2017 Table A1. Pioneer firms by country Number of  Percentage  Number of  Percentage  Income  Year of  FCS  pioneers  of pioneer  pioneers aged  of pioneers  Country Code Region group Observations survey (Yes/No) firms firms  5 or less aged 5 or less Afghanistan AFG SAR LIC 410 2014 Yes 271 66.58 61 15.25 Albania ALB ECA UMIC 360 2013 No 42 11.67 7 1.98 Antigua and Barbuda ATG LAC HIC 151 2010 No 11 32.35 0 0.00 Argentina ARG LAC HIC 991 2017 No 565 57.42 24 2.45 Armenia ARM ECA UMIC 360 2013 No 71 19.72 6 1.69 Azerbaijan AZE ECA UMIC 390 2013 No 45 11.54 5 1.29 Bahamas BHS LAC HIC 150 2010 No 27 65.85 4 9.76 Bangladesh BGD SAR LMIC 1442 2013 No 911 63.18 52 3.64 Barbados BRB LAC HIC 150 2010 No 45 64.29 3 4.48 Belarus BLR ECA UMIC 360 2013 No 167 46.52 28 7.87 Belize BLZ LAC UMIC 150 2010 No 24 33.33 1 1.39 Benin BEN SSA LIC 150 2016 No 46 30.87 1 0.68 Bhutan BTN SAR LMIC 253 2015 No 165 65.22 25 9.96 Bolivia BOL LAC LMIC 364 2017 No 267 73.35 39 10.80 Bosnia and Herzegovina BIH ECA UMIC 360 2013 Yes 160 44.44 6 1.68 Bulgaria BGR ECA UMIC 293 2013 No 95 32.42 0 0.00 Burundi BDI SSA LIC 157 2014 Yes 115 73.25 22 14.01 Cambodia KHM EAP LMIC 373 2016 No 149 41.27 13 3.63 Cameroon CMR SSA LMIC 361 2016 No 158 43.89 10 2.93 Central African Republic CAF SSA LIC 150 2011 Yes 110 73.33 29 19.33 Chile CHL LAC HIC 1033 2010 No 537 69.29 6 0.78 China CHN EAP UMIC 2700 2012 No 1658 61.48 93 3.60 Colombia COL LAC UMIC 942 2010 No 543 77.02 24 3.40 Costa Rica CRI LAC UMIC 538 2010 No 237 72.92 16 4.97 Croatia HRV ECA HIC 360 2013 No 169 46.94 16 4.48 Czech Republic CZE ECA HIC 254 2013 No 141 55.51 3 1.22 Côte d'Ivoire CIV SSA LMIC 361 2016 Yes 151 41.94 9 2.56 DRC COD SSA LIC 529 2013 Yes 279 52.94 87 17.09 Djibouti DJI MENA LMIC 266 2013 Yes 135 51.92 27 10.59 Dominica DMA LAC UMIC 150 2010 No 4 14.29 0 0.00 Dominican Republic DOM LAC UMIC 359 2016 No 166 46.37 8 2.26 Ecuador ECU LAC UMIC 361 2017 No 315 87.50 33 9.17 Egypt EGY MENA LMIC 1814 2016 No 218 12.02 6 0.34 El Salvador SLV LAC LMIC 719 2016 No 303 42.14 6 0.84 Estonia EST ECA HIC 273 2013 No 99 36.40 5 1.92 Ethiopia ETH SSA LIC 848 2015 No 425 50.12 59 6.96 FYR Macedonia MKD ECA UMIC 360 2013 No 136 37.78 10 2.80 Georgia GEO ECA LMIC 360 2013 Yes 50 13.89 14 3.92 Ghana GHA SSA LMIC 720 2013 No 516 71.77 79 11.17 Grenada GRD LAC UMIC 153 2010 No 18 75.00 1 4.55 Guatemala GTM LAC UMIC 590 2010 No 244 68.73 7 1.98 Guinea GIN SSA LIC 150 2016 Yes 50 33.33 10 6.85 Guyana GUY LAC UMIC 165 2010 No 44 61.11 3 4.17 Honduras HND LAC LMIC 332 2016 No 183 55.12 6 1.85 Hungary HUN ECA HIC 310 2013 No 84 27.18 8 2.64 India IND SAR LMIC 9281 2014 No 6334 68.31 569 6.18 Indonesia IDN EAP LMIC 1320 2015 No 274 20.79 5 0.39 Israel ISR MENA HIC 483 2013 No 146 30.23 3 0.63 Jamaica JAM LAC UMIC 376 2010 No 59 48.76 2 1.71 Jordan JOR MENA UMIC 573 2013 No 190 33.57 24 4.31 Kazakhstan KAZ ECA UMIC 600 2013 No 160 26.67 24 4.04 Kenya KEN SSA LMIC 781 2013 No 655 84.19 61 8.10 Kosovo XKX ECA LMIC 202 2013 Yes 125 61.88 21 10.40 Kyrgyz Republic KGZ ECA LMIC 270 2013 No 132 49.07 17 6.32 Lao PDR LAO EAP LMIC 368 2016 No 95 25.82 3 0.84 Latvia LVA ECA HIC 336 2013 No 89 26.81 11 3.37 Lebanon LBN MENA UMIC 561 2013 No 326 58.11 35 6.25 Lesotho LSO SSA LMIC 150 2016 No 13 8.67 0 0.00 Liberia LBR SSA LIC 151 2017 Yes 71 47.02 7 4.76 29    Table A1. Pioneer firms by country (Continued) Lithuania LTU ECA HIC 270 2013 No 85 32.08 8 3.07 Malawi MWI SSA LIC 523 2014 Yes 378 72.97 28 5.51 Malaysia MYS EAP UMIC 1000 2015 No 492 49.55 5 0.52 Mali MLI SSA LIC 185 2016 Yes 88 47.83 5 2.78 Mauritania MRT SSA LMIC 150 2014 No 113 75.33 19 12.84 Mexico MEX LAC UMIC 1480 2010 No 675 58.64 58 5.07 Moldova MDA ECA LMIC 360 2013 No 129 35.83 4 1.13 Mongolia MNG EAP LMIC 360 2013 No 165 45.83 24 6.76 Montenegro MNE ECA UMIC 150 2013 No 26 17.33 1 0.71 Morocco MAR MENA LMIC 407 2013 No 200 49.63 14 3.60 Myanmar MMR EAP LMIC 607 2016 Yes 155 25.54 32 5.33 Namibia NAM SSA UMIC 580 2014 No 464 80.70 135 24.19 Nepal NPL SAR LIC 482 2013 Yes 366 75.93 25 5.20 Nicaragua NIC LAC LMIC 333 2016 No 234 70.27 5 1.51 Niger NER SSA LIC 151 2017 No 57 38.00 10 6.90 Nigeria NGA SSA LMIC 2676 2014 No 1756 66.84 153 6.39 Pakistan PAK SAR LMIC 1247 2013 No 524 42.16 12 1.08 Panama PAN LAC HIC 365 2010 No 15 13.04 1 0.92 Papua New Guinea PNG EAP LMIC 65 2015 No 56 86.15 1 1.56 Paraguay PRY LAC UMIC 364 2017 No 246 67.58 10 2.75 Peru PER LAC UMIC 1003 2017 No 761 75.87 11 1.11 Philippines PHL EAP LMIC 1335 2015 No 704 53.29 39 3.01 Poland POL ECA HIC 542 2013 No 213 39.30 1 0.19 Romania ROU ECA UMIC 540 2013 No 319 59.07 10 1.88 Russia RUS ECA UMIC 4220 2012 No 1414 33.56 270 6.52 Rwanda RWA SSA LIC 241 2011 No 211 87.55 74 31.22 Senegal SEN SSA LIC 601 2014 No 392 65.44 29 5.15 Serbia SRB ECA UMIC 360 2013 No 150 41.67 15 4.18 Sierra Leone SLE SSA LIC 152 2017 Yes 53 34.87 14 9.21 Slovak Republic SVK ECA HIC 268 2013 No 76 28.46 0 0.00 Slovenia SVN ECA HIC 270 2013 No 103 38.15 4 1.51 Solomon Islands SLB EAP LMIC 151 2015 Yes 110 73.33 30 20.41 South Sudan SSD SSA LIC 738 2014 Yes 483 65.54 333 45.80 Sri Lanka LKA SAR LMIC 610 2011 No 319 52.30 26 4.40 St Kitts and Nevis KNA LAC HIC 150 2010 No 14 48.28 1 3.45 St Lucia LCA LAC UMIC 150 2010 No 10 15.87 1 1.59 St Vincent and Grenadines VCT LAC UMIC 154 2010 No 27 56.25 2 4.35 Sudan SDN SSA LMIC 662 2014 Yes 399 60.27 39 6.17 Suriname SUR LAC UMIC 152 2010 No 52 69.33 1 1.33 Swaziland SWZ SSA LMIC 150 2016 No 43 28.86 3 2.05 Sweden SWE ECA HIC 600 2014 No 538 89.67 39 6.67 Tajikistan TJK ECA LIC 359 2013 Yes 96 26.74 17 4.76 Tanzania TZA SSA LIC 813 2013 No 539 66.54 57 7.54 Thailand THA EAP UMIC 1000 2016 No 190 20.21 7 0.77 Timor Leste TLS EAP LMIC 126 2015 Yes 83 65.87 2 1.64 Togo TGO SSA LIC 150 2016 Yes 59 39.33 6 4.03 Trinidad and Tobago TTO LAC HIC 370 2010 No 60 48.00 4 3.20 Tunisia TUN MENA LMIC 592 2013 No 267 45.10 23 3.91 Turkey TUR ECA UMIC 1344 2013 No 270 20.35 38 2.93 Uganda UGA SSA LIC 762 2013 No 579 76.18 95 13.46 Ukraine UKR ECA LMIC 1002 2013 No 303 30.27 17 1.77 Uruguay URY LAC HIC 347 2017 No 293 84.44 21 6.07 Uzbekistan UZB ECA LMIC 390 2013 No 24 6.15 3 0.77 Venezuela VEN LAC UMIC 320 2010 No 50 58.82 3 3.66 Vietnam VNM EAP LMIC 996 2015 No 498 50.20 73 7.43 West Bank And Gaza PSE MENA LMIC 434 2013 Yes 139 32.10 20 4.66 Yemen YEM MENA LIC 353 2013 Yes 184 52.12 6 1.73 Zambia ZMB SSA LMIC 720 2013 No 529 73.88 78 11.11 Zimbabwe ZWE SSA LIC 600 2016 Yes 192 32.00 24 4.11 Note: The percentages of pioneer firms presented in this table are calculated using the total number of firms that reported if they have innovated or not by introducing either new products or services; new products or services for the establishment's main market; or new and new products or services for the  establishment's main market; or new and significantly improved process during the last three years. Firms with missing values in the three variables used  to define pioneer firms are not considered. The note considers a country as in a FCS if it was a FCS during (i)  the year when the WBES was carried out or  (ii)  the three years prior to the survey. This is an important aspect aimed at capturing the potentially long‐lasting impact of conflicts.  30    Table A2. Characteristics of pioneering versus non-pioneering firms in countries which are FCS and non-FCS Fragile and conflict affected states Other countries Pioneering  Non‐pioneering  Pioneering  Non‐pioneering  firms firms firms firms Workforce     Number of workers (average) 40.09 29.19 104.65 63.36     Permanent workers (percent of total workers) (%) 94.05 94.59 94.59 96.31     Skilled workers (out of all production workers) (%) 71.00 69.81 70.00 74.96     Firms with formal training to employees in last fiscal year (%) 33.42 16.64 50.97 28.12     Firms identifying inadequately education as biggest obstacle (%) 2.83 3.12 7.14 6.09 Innovation     Internationally‐recognized quality certification (%) 16.42 9.44 33.86 19.47     Firms spend on R&D during the last fiscal year (%) 26.64 4.25 35.63 5.20     New products or services introduced during the last 3 years (%) 75.21 0.00 73.02 0.00     Improved process introduced during the last 3 years (%) 75.56 0.00 80.78 0.00     New products or services also new for the firm's main market (%) 69.30 ‐ 67.02 ‐ Finance     Firms with line of credit or loan ( %) 26.58 16.66 41.32 28.18     Firms identifying access to finance as a major constraint (%) 35.75 33.51 20.71 18.91     Investments financed by internal funds (%) 75.73 79.55 66.24 71.46     Investments financed by banks (%) 10.51 8.07 19.44 15.77     Working capital financed by internal funds (%) 77.73 81.16 67.17 76.24     Working capital financed by banks (%) 9.00 6.34 17.22 11.31     Firms identifying access to finance as biggest obstacle (%) 13.66 17.61 14.54 14.18 Firm performance     Value added per worker, USD 2009 (average)            26,824.72                      20,347.84            34,681.62                      24,558.92     Cost of labor per unit of sales (average) 0.21 0.27 0.35 0.33     Real annual sales growth (%) 0.03 0.82 0.88 ‐0.04     Annual labor productivity growth (%) ‐4.82 ‐2.74 ‐2.91 ‐2.62     Annual employment growth (%) 5.42 4.68 4.53 3.21     Pushase of fixed assets in last fiscal year (%) 47.58 32.23 49.99 26.20 Infrastructure      Number of water insufficiencies in a typical month (average) 1.84 1.10 0.99 1.18     Percent of firms experiencing water insufficiencies (%) 19.65 12.29 11.74 8.34     Number of electrical outages in a typical month (average) 7.99 7.44 11.60 8.06     Duration of a typical electrical outages (hours) (average) 2.81 2.83 2.48 1.44     Firms experiencing electrical outages during the previous fiscal year (%) 72.38 69.23 62.41 42.43     Delay in obtaining a mainline telephone connection (days) (average) 7.63 18.50 13.59 12.15     Firms identifying electricity as biggest obstacle (%) 14.21 11.36 12.56 10.17     Firms identifying transport as biggest obstacle (%) 2.59 3.34 3.30 4.20     Firms identifying access to land as biggest obstacle (%) 4.02 4.39 3.45 2.93 31    Table A2. Characteristics of pioneering versus non-pioneering firms in countries which are FCS and non-FCS (Continued) Fragile and conflict affected states Other countries Pioneering  Non‐pioneering  Pioneering  Non‐pioneering  firms firms firms firms Other constraints     Firms identifying political instability as a major constraints (%) 55.08 48.66 26.22 23.88     Firms identifying political instability as biggest obstacle (%) 21.02 19.43 7.87 10.03     Firms identifying corruption as biggest obstacle (%) 9.47 6.68 9.21 8.04 Other firm characteristics     Age (average) 15.53 14.88 20.05 17.74     Firms of small size (<20) (%) 55.97 67.68 37.20 51.74     Firms of medium size (20‐99) (%) 31.96 25.03 37.15 33.10     Firms of large size (100 and over) (%) 12.06 7.29 25.65 15.16     Firms with a female top manager (%) 10.91 14.45 14.54 16.70     Firms with female participation in ownership (%) 24.05 23.58 35.47 31.70     Firms with majority female ownership (%) 8.34 12.01 10.35 11.52     Firms formally registered at the start of operations (%) 87.96 88.00 88.45 89.56     Number of years firms operatied without formal registration (average) 0.66 0.74 0.72 0.66 Ownership status     Proportion of private domestic ownership in a firm  (%) 77.94 84.24 89.65 93.09     Proportion of private foreign ownership in a firm (%) 12.75 8.73 7.43 4.37     Proportion of government/state ownership in a firm (%) 0.53 0.34 0.59 0.77     Proportion of other ownership in a firm (%) 8.78 6.69 2.33 1.77 Legal and export status     Sole proprietorship (%) 47.32 53.24 34.03 35.05     Shareholding company with trade shares (%) 6.22 5.15 4.30 3.40     Shareholding company with non‐trade shares (%) 24.02 25.99 34.28 42.55     Partnership (%) 13.45 10.81 10.33 8.99     Limitied partnership (%) 8.28 4.56 15.03 8.07     Other status (%) 0.70 0.26 2.00 1.94     Exporter (%) 11.09 7.36 18.76 11.30     Proportion of total sales that are exported directly (%) 5.16 4.13 9.31 6.01 Sectors     Manufacturing (%) 36.91 33.55 63.91 53.94     Retail (%) 21.48 26.54 11.51 15.91     Other services (%) 41.61 39.91 24.58 30.15 Observations 4,302                              4,295                  31,486                            29,230                      Countries 25 93 Source:  Calculations based on World Bank Enterprise Survey. Note:  Some variables exhibit missing values. 32    Table A3. t-tests on mean differences between pioneering versus non-pioneering firms in countries which are FCS and non-FCS Pioneers Pioneers Pioneer Pioneers Mean p-value Variable non-FCS non-FCS FCS FCS (SD) Difference (Mean) (SD) (Mean) Firm age 20.051 15.989 15.533 13.931 4.519 0.000 Size (number of workers) 104.646 244.903 40.098 72.106 64.548 0.000 Manufacturing (=1) 0.639 0.480 0.369 0.483 0.270 0.000 Purshase of fixed assets (% of firms) 49.992 50.001 47.582 49.947 2.410 0.003 Internationally-recognized quality certification (%) 33.864 47.326 16.417 37.048 17.447 0.000 Skilled workers (in percent of all production workers) 70.110 29.545 71.003 29.305 -0.893 0.291 Firms with formal training to employees (%) 0.510 0.500 0.334 0.472 0.176 0.000 Firms with a female top manager (%) 0.145 0.352 0.109 0.312 0.036 0.000 Foreign ownership (%) 7.439 24.089 12.744 30.313 -5.306 0.000 Public ownership (%) 0.592 5.514 0.530 5.153 0.062 0.491 Sole proprietorship (%) 0.340 0.474 0.473 0.499 -0.133 0.000 33