• Non ci sono risultati.

Foreign participation in public procurement and firm performance: evidence from sub-Saharan Africa

N/A
N/A
Protected

Academic year: 2021

Condividi "Foreign participation in public procurement and firm performance: evidence from sub-Saharan Africa"

Copied!
41
0
0

Testo completo

(1)

Foreign Participation in Public Procurement and Firm Performance:

Evidence from Sub-Saharan Africa

Bernard Hoekmana and Marco Sanfilippob

a Robert Schuman Centre for Advanced Studies, EUI (Florence, Italy) & CEPR b University of Bari (Italy) and Institute for Development Policy, University of Antwerp

Abstract

This paper exploits a firm-level dataset for nineteen Sub-Saharan African countries that provides information on the share of total sales to government entities to provide new insights into the relative importance of participation in public procurement activity for different types of firms. The data suggest that foreign-owned firms often account for a large share of total procurement, illustrating the importance of FDI as a channel to contest procurement markets. Participation in public procurement is positively associated with different measures of firm performance. This is most strongly the case for domestic firms, especially small companies, and firms engaged in manufacturing activities. The positive association increases with the share of foreign firms in public procurement, suggesting more open policy environments enhance local competitiveness.

Keywords: public procurement; firm productivity; FDI; WTO Government Procurement

Agreement; Sub-Saharan Africa

JEL Classification: H57; O12

Acknowledgements

We are grateful to Richard Newfarmer, Ritwika Sen and two anonymous reviewers for helpful comments on previous drafts. We thank UNIDO for granting access to the African Investor Survey data. Financial support for this research was provided by the International Growth Centre (Project N. 43421).

(2)

1. Introduction

Governments around the world purchase a wide range of products to provide public goods and services to citizens. Such public procurement often accounts for a significant share of GDP and aggregate demand. In low-income countries, public procurement constitutes 14.4 percent of GDP on average, with even larger values recorded in some of the poorest regions of the world, including South Asia and Sub-Saharan Africa (Djankov et al., 2016). Most national public procurement systems embody procedures to ensure value for money: award of contracts to the lowest cost suppliers able to meet technical specifications required by a project. Value for money is not the only goal of public procurement systems, however. Public procurement may also be a tool of industrial policy or be used to attain sustainable development goals (Kattel and Veiko, 2010; OECD, 2013; UNIDO, 2017), reflected in “buy national” policies, local content or technology transfer requirements for foreign bidders (e.g., Altenburg and Lütkenhorst, 2015). As government procurement can represent a meaningful source of demand for firms, a government contract may encourage firms to invest more, expand employment and increase productivity (Geroski, 1990; Acemoglu et al., 2013; Ferraz et al., 2015).

Governments generally source most goods and services from national (domestic) firms: the import share of procurement in OECD countries is less than the import share of private sector demand (e.g., Shingal, 2015). The home bias in procurement awards has incentivized countries to negotiate international agreements on public procurement. The main example is the WTO Agreement on Government Procurement (GPA), but procurement is increasingly also included in bilateral or regional preferential trade agreements (PTAs). Such agreements aim to enhance market access for foreign bidders through a mix of transparency-related disciplines and commitments to open contracts above a minimum threshold value to international competitive bidding. Because foreign direct investment (FDI) may be a more important channel for firms to contest procurement markets than trade (Evenett and Hoekman, 2005), the GPA covers both trade and FDI. Despite efforts by signatories of the GPA to expand membership, very few developing countries have joined the GPA, in part because of concerns regarding the implications for their ability to allocate government contracts to domestic firms.1

1 The GPA is a Plurilateral Agreement under the WTO, i.e., participation is voluntary. It has 45 signatories, counting the 28 members of the EU separately. Most members are high-income economies.

(3)

There has been relatively little research on the extent of home bias in public procurement in low-income developing countries or the relationship between public procurement (government demand for goods and services) and the productivity performance of firms. As discussed in Section 2, most empirical research on public procurement centres on whether contract award procedures and mechanisms result in (support) realization of value for money goals, with a specific focus often being on control of corruption and anti-competitive behaviour. Moreover, most extant research focuses on advanced economies; empirical analysis of public procurement in low-income developing countries is limited due to absence of comparable, disaggregated data on procurement awards.

In this paper, we employ firm-level data from a survey covering roughly 6,700 companies based in 19 countries in Sub-Saharan Africa (SSA). The survey includes information on the share of total sales of each firm to the Government and thus the shares of foreign-owned as opposed to national firms in total reported sales to the government – a proxy for participation in public procurement by these two groups of firms. We use the information on the shares of foreign-owned or controlled firms in public procurement to analyse the relationship between sales to the Government (participation in procurement) and firm performance in the countries concerned. To the best of our knowledge, this is the first effort to explore empirically the nexus between public procurement and firm performance in a cross-section of low-income African countries, taking into account the degree of foreign firm participation (i.e., the revealed openness of procurement markets to foreign competition).

The case of SSA is particularly relevant for such an analysis given arguments in the literature that one factor impeding structural transformation in Africa is weak demand (McMillan et al., 2014; UNIDO, 2017). A focus on participation in public procurement in SSA is also of interest from the perspective of international trade cooperation. Governments may source from domestic or foreign firms. Given that public procurement policy often is skewed towards domestic sourcing, governments have negotiated trade agreements that require non-discrimination in procurement. The main example is the GPA. A distinguishing feature of SSA is that countries in this region have not (yet) signed the GPA, nor have they included procurement in their preferential trade agreements. The major reason suggested in the literature for this is a lack of export interest, i.e., limited capacity of firms in these countries to contest foreign public procurement markets. However, the literature has also pointed to the

(4)

potential importance of inward FDI as a channel to contest public procurement opportunities.2 Given that SSA countries tend to be open to FDI, another reason for lack of participation in the GPA may be that governments already source from foreign firms through this channel.3 The availability of data on FDI and sales by foreign firms to the government motivate our focus on SSA.

The analysis reveals that public procurement is a significant source of demand for many firms in SSA. Government contracts account for a larger share of sales of domestic firms than foreign-owned ones, and are more important for larger and older firms. We find a strong positive relationship between sales to Government and the performance of firms. Increasing the share of total output sold to the government by 10 percentage points is associated with a 4 percent higher productivity.4 Although there is substantial heterogeneity across firms, the correlation between sales to the government and performance is more evident for smaller and domestically owned firms, and for firms that are at the bottom of the productivity distribution. Public procurement also relates positively with other dimensions of firm performance, not just productivity, e.g., product innovation by domestic firms.

The correlation between procurement participation (sales to Government) and indicators of firm performance is greater in countries that are more open to foreign participation, measured by the share of total sales to the Government accounted for by foreign firms (FDI), suggesting domestic environments more open to foreign participation in public procurement are more likely to enhance overall local competitiveness. Note that, due to data limitations, foreign participation is defined as sales to the government by foreign-owned firms based in each country. This excludes other forms of foreign participation in procurement (i.e., imports). Such restricted definition of foreign procurement fits with the overall aim of this work, which is to examine whether and to what extent participation in public procurement can result in positive outcomes for locally based (domestic and foreign) firms.5 Foreign participation (the share of foreign firms in total sales to the government) in turn correlates

2 Kutlina-Dimitrova and Lakatos (2016), for example, use very detailed panel data on EU procurement and find that trade and FDI policies as well as the quality of ‘behind-the-border’ product market regulation has an impact on the allocation of procurement.

3 See e.g., Hoekman (2017) for references to the literature.

4 As discussed further below, this is a correlation and does not imply causation.

5 While imports by the government might be relevant as an indirect mechanism influencing productivity (by increasing competitive pressures)in some sectors, the available evidence for a broad cross-section of countries has shown that imports account for only a very small share of total procurement (e.g., Gourdon and Messent, 2017). As we discuss further in Section 3, this import channel is captured in part in our data through sales by wholesalers to the government.

(5)

with the quality of procurement institutions (as measured by the World Bank’s Benchmarking Public Procurement indicators) as well as economy-wide measures of governance such as control of corruption. Given that SSA countries have not signed the GPA but are nonetheless relatively open to foreign participation, our results suggest that the benefits of GPA membership may be associated more with improvement in procurement institutions than with market access liberalization.

Our analysis uses cross-sectional data. Given that we have no information on the timing of access to procurement, i.e., when firms start selling to the government and how long they have been doing so, we cannot draw any conclusions regarding causal effects of participating in public procurement: the results may be driven by selection of better firms into procurement contracts or by unobserved characteristics of the firms (e.g., connections with government officers) we cannot control in our analysis. We try to address this concern through a selection model exercise that matches firms that participate in procurement with firms that do not but share a common set of observables. This exercise suggests that larger and better performing firms are more likely to sell to the government, and our main findings continue to hold. The results we obtain are consistent with recent developing country-specific panel studies using detailed administrative (contract-specific) data that establish a causal relationship between participation in public procurement and different dimension of firm performance (e.g. Ferraz et al., 2015 for Brazil and Facic 2018 for Ecuador).

The paper proceeds as follows. Section 2 briefly reviews related literature. Section 3 discusses the data and provides some descriptive statistics for our sample of 19 countries. Section 4 describes the empirical framework. Section 5 presents the results and several robustness checks, including the matching exercise. Section 6 explores the role of foreign participation in public procurement in influencing our findings. Section 7 concludes.

2. Related literature

The analysis relates to two strands of the research literature on public procurement, trade (market access) and economic development.6 The first strand focuses on participation in public procurement by different types of firms, motivated by the fact that many procurement regimes seek to steer government expenditures to local firms (as opposed to foreign companies), and often target specific types of domestic firms, notably small and medium-sized enterprises (SMEs). From an economic welfare or efficiency perspective questions of 6 Carbone et al., (2018) is a recent survey of market access related studies; Hoekman (2017) reviews the literature on the effects of trade agreements.

(6)

interest in this line of research generally concern the extent of discrimination against foreign firms (home bias) and whether trade agreements are effective in reducing home bias (Evenett and Hoekman, 2005; Evenett and Shingal, 2016; Gourdon and Messent, 2017; Shingal 2011; 2015).

A stylized fact emerging from this research is that home bias is very strong.7 Evenett and Shingal (2006) conclude that Japan’s membership of the GPA did not increase the share of contracts awarded to foreign suppliers, comparing data for 1998-99 and 1990-91. Using a longer time series, Shingal (2015) finds that GPA membership had no effect in reducing home bias in Japan and Switzerland. Similarly, Shingal (2011) concludes that the extension of the GPA in 1996 to cover services procurement had no impact on sourcing patterns in Japan or Switzerland: the share of services contracts awarded to foreign suppliers actually declined over time for both countries. Rickard and Kono (2014), in a more comprehensive assessment of the effect on home bias of trade agreements spanning procurement find these have no impact on the procurement elasticity of trade (the ratio of imports to government demand), i.e., are not associated with a reduction in home bias.8

A second strand of the literature focuses on government contracts (participation in public procurement) and the performance of firms. A hypothesis informing research in this area is that public procurement can complement supply-side policies aimed at enhancing competitiveness of firms by generating additional demand for domestic firms that supports their growth and overall productivity performance.9 For example, Hebous and Zimmerman (2016) investigate the nexus between public procurement and capital investment for a sample of US firms and show that sales to the government relax financial constraints, permitting firms to increase capital investment. Ferraz et al. (2015) exploit a large dataset on procurement tenders in Brazil and show that winning a government contract increases overall

7 See, e.g., Branco (1994), Francois et al. (1997), Trionfetti (2000), Brulhart and Trionfetti (2004), Evenett and Hoekman (2005), Ramboll and Chur (2011), Kutlina-Dimitrova and Lakatos (2016), Gourdon and Messent (2017) and Cernat and Kutlina-Dimitrova (2017).

8 Recent research by Gourdon and Messent (2017) and Tas et al. (2018) analyzes the determinants of foreign sourcing in EU procurement awards using the EU Tenders Electronic Daily database (which reports information on procurement awards) and conclude that GPA membership increases the probability that firms win contracts in the EU. While these results are interesting, identifying the independent effect of the GPA in the context of the EU is difficult given that EU procurement law and policy is both broader and deeper than the GPA. Dengler and Hoekman (2018) find evidence that GPA membership is associated with a smaller decline in the foreign (import) share of procurement following the 2008 financial crisis than is observed for non-GPA countries, suggesting that the GPA may be an effective commitment mechanism.

9 See Edler and Georghiou (2007) for a discussion of the role public procurement as a demand-side policy and Georghiou et al. (2014) for analysis of the design of procurement policy if governments desire to use procurement to promote innovation.

(7)

employment growth of the firms concerned. They argue that learning processes help the firms to compete on new markets and to develop new products, with the government contracts acting as the trigger for such “downstream” effects. Lee (2017) obtains similar results using Korean procurement data: firms allocated public procurement contracts experience increased growth and activity generally, over and above the effects of the activity associated with the public contract. This is the case in particular for small, young and financially constrained firms. The research analysing the effects of participation in procurement and allocation of procurement contracts does not consider the role played by foreign participation or the openness of the procurement market to foreign participation.

3. Data

We use the African Investor Survey (AIS), a large representative survey of firms operating in Sub-Saharan African countries administered by UNIDO. This covers some 6,700 firms based in 19 SSA countries (UNIDO, 2011).10 The data were collected in 2010 using a rigorous survey methodology, including stratified sampling (on three dimensions: sector, size and ownership of firms) and interview techniques (face-to-face interviews with top-level managers). The sample is representative of public and private for profit firms with 10 or more employees.

The richness of information provided by the AIS illustrates the heterogeneous dimensions of firms in the region. The sample includes both domestic (62.4% of the total) and foreign-owned firms (37.6% of the total).11 It covers mainly firms in the manufacturing and services sectors and provides detailed information on each company, including its 2-digit industry classification (based on the ISIC Rev. 3 classification),12 location, international trade position, and the financial variables necessary to compute standard indicators of firm performance. Most relevant for our purposes is that the survey includes questions asking firms to identify their main customers and the share of total sales going to six possible client groups: (1) 10 See http://investment.unido.org/imp/About/AboutOurData.aspx. The countries are: Burkina Faso, Burundi, Cameroon, Cape Verde, Ethiopia, Ghana, Kenya, Lesotho, Madagascar, Malawi, Mali, Mozambique, Niger, Nigeria, Rwanda, Senegal, Tanzania, Uganda, and Zambia.

11 Foreign owned firms are classified following the standard OECD definition based on the ownership of ten per cent or more of the equity of a firm. The AIS sample includes both wholly-owned affiliates of multinational corporations based abroad and stand-alone firms owned by individuals who are nationals of a foreign country. Over 80% of the sampled foreign firms entered the country via greenfield or joint ventures. Chapter 2 of UNIDO (2011) provides more details on the definitions and all the summary statistics for the sample.

12 Manufacturing (excluding construction) represents nearly half of the total sample, and is especially concentrated in the food industry. Services represent about 39%, and include mainly trade related activities. A small number of firms (5.5%) are in the primary sector (agriculture and mining).

(8)

retailers; (2) wholesale distributors; (3) manufacturers; (4) consumers; (5) government agencies; and (6) international organizations/NGOs. Among the 4,600 firms (68% of total sample) that responded to this specific question, 29.6% reported at least some sales to the Government.13 This information is roughly comparable with the World Bank Enterprise Surveys (WBES). For the sample of 19 SSA countries surveyed by UNIDO, the WBES data covering the same time span indicate that 23.2% of firms report securing government contracts during the last year or attempting to do so. Despite the differences in sample definitions and specific questions included in the UNIDO AIS and the WBES,14 the information concerning participation in procurement correlates quite well (see Figure A1 in the Appendix).

FIGURE 1 ABOUT HERE

Based on this information, we use the share of total sales to the government as our variable of interest.15 For the full sample of firms and countries, on average, government accounted for about 8.1% of firms’ total sales (Figure 1). More disaggregated descriptive statistics reported in Table 1 reveal substantial heterogeneity in firm- and industry-specific characteristics that affect the extent to which government matters as a source of demand. On average domestic firms sell larger shares of their output to the government than foreign-owned firms. Among foreign-owned firms, those from OECD economies report a higher share of total sales to the state. Although Southern multinationals have been successfully investing and contracting in SSA (Zhang and Gutman, 2015), it appears that such firms are less focused on the local government procurement market. The data also indicate that larger firms tend to sell a greater 13 As some 32% of the firms do not respond to this specific question, attrition could represent a source of selection bias in our analysis. We explored the characteristics of the group of non-respondents, as compared to the firms with procurement contracts, and they seem to show that on average the former group comprises smaller, less productive and younger firms compared to the latter. While there are no systematic differences in their geographic distribution (with the partial exceptions of Senegal and Madagascar), non-respondents are prevalently service firms, while firms selling to the government are mostly engaged in industrial (manufacturing) sectors.

14 The UNIDO AIS includes a larger share of foreign firms in its sample than the WBES. In addition, the AIS question on public procurement differs from that in the WBES. The WBES asks firms to report whether they secured or attempted to secure public procurement over the last year, with the main goal of determining if kick-backs or bribes were paid when seeking such contracts. The AIS asks about actual levels of current sales to government entities. The AIS does not provide information on when tenders were submitted by firms or contracts were awarded.

15 The same question was asked to all types of firms (domestic and foreign) in all the countries, so that it is perfectly comparable among them. The figure refers to the latest financial year available. A main limitation is that we do not have information on the exact year in which firms were awarded a government contract and through which procedure. Owing to the generic formulation of the question, it is not possible to distinguish between sales to different layers of government (i.e. local or national).

(9)

share of their output to the government, as do firms that have been in operation for more than 10 years. Family-owned businesses in contrast appear less likely to take advantage of government demand.

TABLE 1 ABOUT HERE

The relative importance of government demand for firms varies across sectors. Not surprisingly, firms in the utilities and the construction sector report significantly higher shares of total output sold to governments. Manufacturing firms sell proportionally less to government than firms in service sectors16, especially those engaged in traditional, relatively unskilled activities. Producers of higher-tech products are likely to sell relatively less to government entities.17 There are large cross-country differences in the relative importance of sales to the government and the share of Government demand in total sales of local firms (Figure 2).

FIGURE 2 ABOUT HERE

Table 2 reports descriptive evidence on the importance of foreign firms in procurement markets in Africa. We do this by constructing indicators of implied de facto openness to foreign firm participation in national public procurement markets, which we back out from our firm-level data. This is relevant per se since, to the best of our knowledge, there is very little data or comparative evidence available from SSA countries on the share of total public procurement that is awarded to (executed by) foreign-invested firms (remember that this does not include foreign procurement directly supplied from abroad). We use this information in

16 Services include a variety of activities. Among them, trade firms and wholesalers are a specific case since they do not produce directly but can sell other firms’ goods to the government. There are a total of 389 wholesalers in the data, 323 of which responded to the question on procurement. About 27% of the latter group report having sold something to the government. The survey does not provide details on which goods they actually sell to the government (or whether they are imported), but 86.5 percent of the wholesalers who sell to the government are either foreign affiliates or importers (sourcing part of what they sell as direct imports). Aside of being a small number, their economic weight is also small. Taken together, the group of wholesalers who import account for small values (7% on average or less than 2% if the median value is considered) in terms of total sales to the government (note that these values include both imported and locally sourced goods by wholesalers, as the data does not allow to distinguish among the two). This suggests that our analysis incorporates to some extent also the propensity of governments to source from foreign firms through importation through locally-established wholesalers. Yet, due to the small weight of wholesalers in terms of procurement, this is a less relevant channel to explore as compared to FDI (see Section 6).

17 The mapping of manufacturing activities into technology categories uses a classification developed by OECD (2005). See UNIDO (2011, p. 62).

(10)

Section 5 to test whether cross-country heterogeneity in terms of the share of foreign firms in total reported sales to the government affects the core results of our analysis.

Information on the total value of sales to the government was obtained following two simple steps. First, for each firm for which we have non-missing information, we calculate the absolute level of sales to the government. Second, we aggregate the data at the country level by summing up individual firms’ totals. In doing this we use sample weights provided with the survey to increase the representativeness of the information provided.18 Table A1 in the Appendix shows that, relative to GDP, the constructed data on total sales to the government retrieved from the AIS falls within the order of magnitude of country level official data on total government expenditures on goods and services reported by the IMF.19 For many countries in the AIS sample the overall magnitude of sales to the government reported by the surveyed firms account for a significant share of total. This provides some confidence in the use of the absolute value of reported sales to the government by firms in our sample to assess the relative openness of countries to foreign firm participation in public procurement.

TABLE 2 ABOUT HERE

The first column of Table 2 reports the share of foreign firms in total reported sales to the government in the 19 AIS countries; columns 3-5 of Table 2 report more disaggregated sectoral shares. The picture that emerges is quite heterogeneous across countries and across sectors within the same country. Foreign firms (i.e., FDI) are important for public procurement. Foreign companies account for a large share of reported total sales to the government. Although there is significant variance, for most of the SSA countries the foreign share is above 30%. In about half of the countries in the sample foreign firms account for

18 The representativeness of the AIS is guaranteed by the stratification strategy, which is based on firms’ sector, size and ownership (domestic/foreign) (UNIDO, 2011). The ownership dimension, which is important in assessing the openness of public procurement and the potential role played by foreign competition as a factor influencing the performance of firms awarded contracts by the government, is a specific feature of the AIS that distinguishes it from other surveys (e.g. the World Bank Enterprise Surveys). That said, while unique, caveats apply that need to be considered in interpreting results. One of these is that the business directory used to sample firms includes only firms with more than 10 employees.

19 When confronting the two series, it should be noted that IMF data record total government expenditure on products and are based on the national accounts, while our survey data record (self-reported) sales to the Government. Our estimates are at least within an order of magnitude of official data, and provide some indication of the extent to which they correlate with total government spending. As our goal is not to assess the determinants of total procurement or its allocation but rather the relationship between sales to the government and firm performance and how the latter is associated with the extent of foreign competition for government projects the extent of the mismatch is not very salient.

(11)

over 50% of the total sales to the government reported by all firms. These ratios appear substantially higher than those observed in OECD countries.20 One reason for this is the much more limited capacity of domestic firms to provide some of the products and services demanded by governments. From the perspective of the main motivation underlying inclusion of procurement disciplines in trade agreements, the data suggest markets are already substantially open to foreign competition.

Higher foreign shares of total sales to the government reported by all firms in the sample correlates positively with levels of economic development (per capita income), negatively with the quality of relevant domestic institutions, proxied by perceived levels of corruption, and negatively with a more narrowly defined average indicator of the quality of procurement institutions and processes (Figure A3). Higher average levels of procurement system quality, as measured by the World Bank Benchmarking Public Procurement indicators (World Bank, 2016), are associated with a lower foreign share of total sales to government, whereas foreign firm shares are higher in environments that are more corrupt. While these are just simple correlations, they suggest foreign firms may be better able to operate in corrupt environments and that better procurement processes are associated with domestic firms capturing higher shares.

4. Empirical framework

Following the voluminous literature on heterogeneous firms (e.g. Helpman et al., 2004), our empirical framework is based on a general functional relationship linking firm performance to a vector of firm-specific factors and a measure of government demand:

yijx=

βZi+γ sharegovi+θj+δx+εijx (1)

In this model, y is a performance indicator for any firm i in country j and sector x. In our main estimates, we use an indicator of labour productivity (lab_prod) constructed as the ratio of sales on employees, but we also use alternative performance indicators, including value added per worker, capital intensity, wages, and innovation21. Z is a vector of standard controls 20 In the EU, for example, foreign-invested firms (FDI) accounted for 13.4% of total EU procurement in 2010 (Ramboll and Chur, 2011).

21 While a standard proxy for firm performance, commonly used in empirical work, including those based on the AIS data (Foster-McGregor et al. 2014; Sanfilippo and Seric, 2016; Gold et al., 2017), the measure of labour productivity we adopt in our baseline analysis (sales on employees) has also the advantage of being the one that maximizes the number of observations. Number of employees is used instead of hours worked as information on

(12)

used in the literature to account for firm heterogeneity. More specifically, we add the age (age) and the size (size_class) of the firm, both expected to be positively correlated with productivity (Helpman et al., 2004), and whether a firm is family owned (family), which usually has a negative impact on firm performance. We account also for export status (exporter) and foreign ownership (foreign), both of which are generally significant predictors of superior performance, including in the African context (Foster-McGregor et al. 2014), and control for the skill intensity of workers in a firm (skill_ratio). In addition, we include a full battery of country and industry fixed effects to account for all unobserved contextual factors that may influence the relationship between Government demand and firm performance. They account for both cross-country differences in levels of development, institutions and regulation of the procurement process, as well as for the fact that some industries (such as construction, as shown in Table 1) may rely more on procurement markets and have different levels of technology intensity. A full description of the variables introduced in equation (1), together with their descriptive statistics, is provided in Appendix Table A2. Table A3 reports the correlation matrix, suggesting no major concern regarding collinearity among the independent variables.

The coefficient of interest, γ , is the share of firm i total sales to government entities. We expect this variable to be positively related with indicators of firm performance. Effective public procurement regimes that emphasize value for money mimic the market by generating competition between suppliers, the aim being that the most efficient firms win contracts. Insofar as this is the case, there should be no difference between firms that sell (more) to the government and those that sell to other customers (private buyers). However, this need not be so if public procurement is used as an instrument of industrial policy or if it is distorted by corruption or collusion between bidders.22 In low-income SSA economies, firms generally have low levels of productivity and are often growth constrained, reflecting poor business environments, limited access to credit and structural characteristics such as small size and limited international exposure (Iacovone et al. 2014).

the latter is unavailable. See Section 5.2 and Table 5 for a description of alternative measures of performance.

22 There is a substantial literature on public procurement in developing countries that looks at it through the lens of corruption and governance. See e.g., Auriol et al. (2016), Peireira and Schwind (2017) and Knack et al. (2019) for recent contributions. Tas et al. (2018) find that in the case of the EU, GPA membership reduces the risk of corruption by decreasing the number of contests with single bidders and the number of wins by a single firm.

(13)

Public procurement may help offset such supply side constraints by increasing demand for firms’ output. Such “demand shocks” have been identified in the literature as a major channel through which success in a procurement bid leads to improvements in firm performance. In response to the demand shock, firms become able to mobilize additional resources to invest (Habous and Zimmerman, 2016; Lee, 2017) and learn about markets and products (Ferraz et al., 2015). Government demand may incentivize pursuit of more risky activities, such as development and introduction of new (differentiated) products and new investments in R&D (Edler and Georghiou, 2007; Aschhoff et al., 2009; Slatchev and Wiederhold, 2016). Of course, these potential channels may not apply equally to all types of firms and sectors – and so we expect the relationship to be heterogeneous on firm characteristics. While the literature suggests that government demand may be important for firms that are smaller and younger and/or have limited access to finance and high innovative potential (Acemoglu et al., 2013; Ferraz et al, 2015), the effect of public procurement (government demand) need not to be positive. In practice, the design of procurement policy and its implementation will matter. If procurement reflects a process of generating and sharing rents between “connected” firms and government officials, positive performance effects either may not be observed or be smaller than they would otherwise be.

A preliminary look at our data suggests there is a positive procurement/performance nexus in low-income SSA countries. Figure 3 reports the kernel distribution of labour productivity, distinguishing between firms that sell to the government (independently of the share of total output being sold) and firms that do not. The graph shows that the distribution of the former group tends to dominate the latter, reflected in a rightward shift. In addition, Figure 4 reports the relationship between productivity and the size of government demand, suggesting a positive association, albeit a weak one.

FIGURE 3 ABOUT HERE

FIGURE 4 ABOUT HERE

5. Results

Table 3 reports the results of estimating equation (1) using OLS. The model generally performs well, and all the major controls included have the expected sign. As indicated in Column 1, we replicate the well-known existing empirical evidence on the performance of

(14)

heterogeneous firms: larger and older firms are significantly more productive, as are those employing a greater number of skilled workers. Higher levels of productivity are also observed among firms that export and those that are foreign owned. On the other hand, we find that family-owned firms are less productive compared to others.

TABLE 3 ABOUT HERE

Moving to our variable of interest (Table 3, column 1), we find that the share of sales to government is positively correlated with firm performance. The estimate is statistically significant at the 1 percent level, demonstrating a strong correlation between government demand and firm performance. The size of the coefficient is not only statistically, but also economically significant. Increasing the share of output sold to the government by 10 percentage points is associated with higher levels of productivity of about 4 per cent.

While selling to the government may help firms improve their performance, dependence on the State may also represent an obstacle to firm growth if firms lack alternative markets. For this reason, in column (2) we test the hypothesis that the relationship between government demand and firm performance may be non-linear. We find that the squared coefficient of procurement turns negative (and is statistically significant) as the share of total output sold to the state rises, with a turning point reached when two-thirds of total sales are to the government23. To complete this first set of results, in column (3) we replace our variable of interest with a dummy taking the value of 1 if firms sell to the government and 0 otherwise. The coefficient estimate is again positive and significant, indicating there is on average a 22 per cent (exp0.202-1) difference in productivity levels between the two groups of firms.

Finally, in column (4) we explore whether government demand is a specific mechanism that enhances firm performance as opposed to demand by non-commercial entities more generally. We replace our coefficient of interest with the share of output sold by each firm to international organizations and non-profit organizations (NGOs). These types of buyers are more likely to have as overriding objective maximizing value for money and are much less likely to have industrial policy type objectives when engaging in procurement of goods and

23 An alternative explanation is that due to cronyism in most corrupt systems governments’ contracts might concentrate on a smaller number of firms, which could explain that after a certain threshold the correlation becomes negative.

(15)

services.24 The coefficient for this variable is positive but not statistically significant, providing suggestive evidence that government procurement may be acting as an industrial policy tool.

The results reported in this Section are robust to inclusion of country-industry fixed effects and a jack-knife exercise in which we estimate our model dropping one country at the time to determine the influence of potential outliers.25

5.1 Results: firm heterogeneity and other potential mechanisms

We next investigate whether the positive relationship between procurement (sales to the government) and productivity holds across different sub-samples of firms. Dividing the sample by firm characteristics helps reducing heterogeneity (procurement may include specific provision for different types of firms), allowing at the same time other controls to be more flexible to different specifications.

In columns (1)-(2) of Table 4 we divide the sample into domestic and foreign-owned firms. The coefficient estimate on the share of output sold to the government is statistically significant only for domestic firms. In columns (3)-(4) we differentiate between firms according to size (SMEs vs. larger companies, see Table 1 for the definitions used). The positive correlation is confirmed only within the group spanning the smaller firms. Interestingly, if we further disaggregate the group of SMEs considering only small (less than 50 employees) and micro26 (less than 20) enterprises, the size of the coefficient becomes larger the smaller the size of the firm, confirming that the demand shock represented by receiving public procurement has proportionally higher potential on smaller, demand constrained, firms (see Figure 5). The lack of statistical significance of the estimates for larger and foreign-owned firms could be due to these firms already having higher levels of productivity. Conversely, the additional demand stemming from government contracts may positively affect the performance of the set of firms that are most likely to be growth constrained, i.e. domestically owned firms and SMEs.

TABLE 4 ABOUT HERE

24 In addition, international and non-governmental organizations may impose more regulations and stricter standards in their procurement processes. Insofar as this places a greater burden (costs) on firms, this may offset or preclude the realization of higher productivity benefits.

25 Results of these sensitivity tests are not reported in the paper for reasons of space, but are available upon request from the authors.

26 To identify micro enterprises we need to use a larger threshold than usual (<10) because the AIS data cover a representative sample of registered firms employing more than ten workers (UNIDO, 2011: 183).

(16)

FIGURE 5 ABOUT HERE

Results in columns (5)-(6) of Table 4 show that the positive association between government demand and performance mostly applies to firms in the manufacturing sector (excluding construction). There is no statistically significant effect for firms in services sectors. Why this is observed is an interesting research question. Possible reasons may include differential potential for productivity improvements across sectors; the non-tradability of many services, which may reduce the scope for foreign firms to bid on contracts, or smaller average contract sizes for services.27

In the remaining columns of Table 4, we check whether our main result is affected by specific characteristics of the context in which firms operate. To do this, we interact our variable of interest with variables that vary at the country and industry level. First, we control for the level of corruption prevailing in each country using the World Governance Indicators (WGI). The procurement literature notes that weak public sector governance can discourage competitive firms from participating in procurement processes, reflecting expectations that bids will not be awarded on the basis of published requirements (Evenett and Hoekman, 2005; Best et al., 2017; Knack et al., 2019).28 One potential consequence is to reduce any potential effect of government demand on productivity as winning contracts will be determined in part by the ability and willingness to provide side-payments. We find that the interaction term is not statistically significant, suggesting that higher levels of overall corruption do not affect the mechanisms through which procurement relates to performance.29 Second, we interact the government demand variable with average import tariffs for the products being procured and an indicator measuring whether sectors have been targeted by national investment promotion activities (both indicators exclude services).30 These variables

27 This is one of many questions suggested by our findings calling for collecting information on contract sizes and the composition of procurement across sectors in our sample countries.

28 World Bank (2016) provides more detailed indicators of the quality of procurement systems but these are available for only one year and do not include any information on the objectives of procurement regimes, the extent to which there are preferences for certain types of bidders or the prevalence of kick-backs and other corrupt practices.

29 The lack of statistical significance obtains as well if alternative WGI indicators of institutional quality are used, such as the rule of law or the political stability index.

30 Sector targeting is a dummy variable taking the value of 1 if the industry has been targeted prior to 2003 for specific promotion towards foreign investors (Harding and Javorcik, 2012). For the purposes of this paper, it can either be intended as a measure related to industrial policy, aimed at increasing the competitiveness of the industry, or as a measure of competition, due to the likely large presence of foreign investors.

(17)

can be regarded as rough indicators of industrial policy. The results indicate that the performance-procurement relationship is bolstered in the presence of complementary sectoral policies.

5.2 Additional results

In this sub-section, we implement additional empirical tests to our results to different definitions of the dependent variable. Table 5 reports the results of our main specification using alternative indicators of firm-level performance, including value added per worker31 (leading to a reduction in the number of observations) and a measure of total factor productivity (TFP).32 Columns (1)-(2) reveal that, in spite of a reduction of the sample size, results do not change much; the signs and magnitudes of the estimates are very similar to our main findings.

TABLE 5 ABOUT HERE

The same result obtains for alternative indicators of performance, including the capital labour ratio (measured dividing total assets by the number of employees) or the level of wages per worker (wage bill divided by the number of employees). Interesting results obtain when measuring outcomes related to the introduction of product and process innovation on the left-hand side.33 Note that such outcomes are only available for the sample of domestic firms. Firms that sell a greater share of their output to the government are more likely to introduce new products as opposed to engage in process innovation. This result is consistent with findings in the extant literature showing a nexus between public procurement and the development of innovative capacities by firms in OECD economies (Ashhoff and Sofka, 2009; Slavtchev and Wiederhold, 2016; Czarnitzki et al., 2016). To the best of our knowledge there is no evidence regarding government demand as a driver of innovation processes in low-income economies in general, and in SSA countries in particular. As noted earlier, the literature highlights several channels through which government demand can incentivize 31 Value added is measured by subtracting the value of purchased inputs and the costs of advertising to total sales.

32 Due to the lack of time series information, TFP is not estimated but it is constructed with a constant return to scale Cobb–Douglas production function, assuming a conventional share of two-thirds for the labour (the wage bill) component and one-third for capital (fixed assets).

33 More specifically, product innovation is a dummy taking one if the firm introduced any new or significantly improved products/services into the market during the last three financial years; process innovation is a dummy taking one if the firm introduced any new or significantly improved production processes including methods of supplying services and ways of delivering products

(18)

innovation and help to offset supply-side constraints, including by creating new markets for products, providing a testing ground for innovative products and learning by doing (Kattel and Lember, 2010). While data limitations prevent investigating the salience of such mechanisms in SSA, it is noteworthy that this relationship holds in our sample countries.

5.3 Robustness checks and endogeneity

The cross-section nature of our data precludes any claims or inference of a causal relationship between the intensity of public procurement participation and firm performance. Our results may be affected by endogeneity as a result of possible self-selection of better performing firms into government procurement, or government entities selecting more productive firms when undertaking procurement. Recent papers that have estimated the effects of procurement on different dimensions of firm performance are able to address such potential endogeneity issues by having panel data that permit conditioning on the characteristics of the firms before winning procurement bids, or exploiting a quasi-experimental setting as a result of specific features of the procurement process analysed (Ferraz et al., 2015; Hebous and Zimmerman, 2016; Lee, 2017).

The cross-country, cross-sectional nature of our data impedes the use of standard approaches to address endogeneity in a robust manner. Instead, we try to address the selection issue by constructing a control group of firms that display as much as possible the same characteristics as firms receiving the treatment (selling to the government). A basic limitation here is that we do not have information on the exact time firms began to sell to the government. This means that we cannot select, as it would ideally be the case (Blundell and Costas Dias, 2009), firms based on their characteristics before receiving the treatment (selling to government). An additional limitation is that unobservable variables can play an important role in the context of our analysis. The literature on procurement is rich in documenting how, for instance, connections with government officials or payments of bribes are relevant determinants of access to procurement, or how much such connections can affect competition for firms excluded from procurement (see, e.g., Best et al., 2017; Knack et al., 2019).

Nonetheless, this approach is used frequently in cross-sectional studies to disentangle potential bias arising from the heterogeneity of the firms in our sample.34 We implement a matching methodology by selecting among the 3,268 firms that do not sell to the Government

34 For example, in an empirical setting that is very similar to ours, Czarnitzki et al. (2018) apply matching methods to construct a control sample of firms not receiving procurement to evaluate the effects of procurement on innovation for a cross section of German firms.

(19)

a control group based on characteristics that help predicting selection into procurement. The latter are found by running a probit model in which the probability of finding firms with positive shares of sales to the government depends on the same factors that we include in our OLS model (size, ownership, skills) plus the level of productivity in the two years preceding the survey.35 Pairs of country-industry fixed effects are added to constrain control group firms to be in the same country/industry as the treated firms (those that sell to the government). Results of the selection model are reported in Appendix Table A4. We use these results to construct inverse probability weights (IPW) that we use in our main regressions to provide a better comparison between controls and treated firms. To assess the results of the matching procedure just described, Figure A2 plots the distribution of the IPW of the two groups before and after weighting. This shows that these are almost overlapping following the weighting exercise. Results of the weighted regression are reported in the second column of Table A4. Coefficient estimates remain very similar to our main results (note, however, that the sample is smaller after selection is controlled for); the estimate of the share of procurement has a magnitude only slightly smaller than the baseline finding (0.36 instead of 0.4).36

A concern regarding the extent to which the matching technique controls for selection is that it does not correct for selection on unobservable factors. After all, the dynamics of the public procurement process and the impact on firm exit and firm survival cannot be observed or controlled for using cross-section data, so that the sample of firms itself may be a selected group of survivors, with the selection process potentially being influenced by the award of public procurement contracts. Thus, firms awarded contracts may well have pushed out competitors that were more productive before the award of the contracts. If this were the case, it would appear in the cross-sectional data that government contracts are associated with higher levels of productivity. While these are salient considerations, we do not believe they are a serious concern in the present context as public procurement by itself is unlikely to be large enough to drive entry or exit decisions.37

35 The questionnaire asks firms about the levels of sales and employment in the preceding two years (i.e. in 2008), so that we can reconstruct labour productivity accordingly.

36 As an additional test, we also used a standard matching estimator based on the Leuven and Sianesi (2003) algorithm, comparing directly how differences in firms’ productivity are affected by the simple treatment (a dummy indicating if firms sell to the government, without accounting for the shares). Results remain consistently similar to those reported in column (3) of Table 2, with an average treatment effect on the treated firms of about 0.2 when using the nearest neighbour estimator with common support. Results are available upon request.

37 A potential exception could be the case of firms selling larger shares of their sales. To control if this is the case, we have run the matching exercise excluding firms for which government procurement accounts for over 30% of their total sales (corresponding to the 90th percentile of the distribution). Results – available upon request

(20)

6. Participation by foreign firms in public procurement

So far, our analysis has mostly centred on the potential role played by government procurement in realizing typical industrial policy objectives, measured by firm-level performance indicators. As noted in the Introduction, there is also a trade and FDI policy dimension to public procurement. The main goal of the GPA and preferential trade agreements that include procurement disciplines is to reduce home bias, i.e., to reduce the prevalence and extent of discrimination in favour of domestic firms in the award of procurement contracts. The presumption is that greater transparency and associated due process rules, combined with requirements to employ open international bidding for contracts above a certain value threshold, will increase foreign competition and help attain value for money.

The AIS includes information on sales to the government by both foreign invested firms based in a country and national firms, allowing thus to calculate the share of the former in total reported sales (see Section 3). The relative importance of foreign owned firms in total reported sales to the government is a proxy for openness of the procurement system and a measure of the prevailing extent of home bias. As noted in Section 3, the AIS data reveal that on average foreign shares are significant and that there is substantial variance across countries in such shares. We cannot assess to what extent foreign participation helps governments achieve value for money goals, as we have no information on procurement transactions, but we can use the survey data to investigate whether the extent of foreign participation (competition) has a bearing on the association between selling to the government and firm performance. This is salient for several reasons. It is relevant to the concern that many developing countries have that accession to the GPA will lead to greater foreign competition and have negative consequences for local firms’ ability to obtain government contracts. It can also be a concern insofar as it is perceived to undercut the realization of industrial policy type goals – i.e., to use procurement as a means to bolster firm performance. In what follows we use the foreign share data to shed some light on whether and to what extent the relative openness to foreign participation moderates the association that we observe between selling to the government and firm performance.

To do this we run our baseline regression again and include an interaction term combining our variable of interest (the share of each firms’ sales directed to the government) with a

(21)

variable measuring the share of foreign value in total procurement in each country (see Section 2). We do this for both the whole sample of firms (as in column 1 of Table 3) and for the sub-sample of foreign firms only (column 2 of Table 4). The results for the whole sample are reported in Table 6; those for the foreign sub-sample are reported in Appendix Table A5. The marginal effect of procurement participation on firms’ productivity is plotted in Figure 6 at different levels of foreign shares of total procurement. The positive correlations obtained in the previous Sections are confirmed.

TABLE 6 ABOUT HERE

FIGURE 6 ABOUT HERE

The added insight obtained is that the size of the correlation increases with the degree of openness to foreign firms (the share of total sales to the government captured by foreign firms). Comparing the two panels of Figure 6, observe that the coefficient is statistically significant for most of the distribution in the whole sample (meaning that the result we found previously is not conditional on foreign shares of procurement being high). If the sample is restricted to foreign firms only, this result obtains only for the highest levels of foreign shares. Although we cannot dig further into these results due to the lack of more precise information at both the firm and the country levels, they seem to suggest that domestic environments more open to foreign participation to public procurement are more likely to enhance local competitiveness.

Given that procurement markets already appear to be rather open in many of the SSA countries included in the AIS, it may well be that GPA membership mostly will have implications for the quality of the procurement process quality. The World Bank BPP average indicator for the SSA countries in the sample range between 50 and 68 out of 100, suggesting there is a lot of scope for improvement. The AIS data do not provide support for the concern that greater competition from foreign owned firms will have adverse consequences for domestic firms – to the contrary. Such competition is already in place – through FDI – and appears to strengthen the association between participation in public procurement and the performance of the firms that do so. Moreover, the simple correlations between measures of corruption and quality of procurement processes reported in Section 3 indicate that foreign firms may be better able to operate in corrupt environments and that better procurement processes are associated with domestic firms capturing higher shares of government demand.

(22)

This also suggests that the institutional dimension of GPA membership may be as if not more important than the market access liberalization dimension.

7. Conclusion

Much of the research and policy literature on public procurement focuses on the extent of home bias and whether value for money goals are achieved, with a strong presumption that encouragement of foreign competition will help do so. In this paper we exploit firm-level data for SSA countries that provide information on the extent to which different types of firms sell their output to government entities. This provides new insights into how overall procurement activity across SSA countries is allocated, and whether participation in public procurement is associated with realization of the types of goals that underlie industrial policy – an improvement in measures of firm performance. It also provides new insights into the role of foreign participation in public procurement in SSA countries.

We find that firms that sell more of their output to government entities tend to have higher levels of productivity. This is most strongly the case, both in terms of statistical significance and the economic magnitude of the relationship, for domestically-owned firms, SMEs and companies engaged in manufacturing activities. A positive relationship between procurement and firm performance is not observed among foreign-owned firms or companies that are in the service sector. The results are suggestive in that they indicate that whatever the underlying policy objectives that guide the allocation of government contracts in a given country, public procurement activity in our SSA sample countries is associated with outcomes that motivate effective industrial development policies.

As noted in the Introduction, the nature of our data preclude any firm conclusions given that we cannot address fully selection issues. This requires firm-level panel data as well as detailed information on government contract awards that is not available for SSA. We regard our analysis as opening up an important, neglected research agenda, one that is highly policy relevant given both the magnitude of government procurement in most countries and the provisions found in many procurement laws seeking to ensure that SMEs have opportunities to participate in, and benefit from, procurement processes.38 What is needed first and foremost is to extend the time dimension of firm-level data on participation in public

38 The World Bank report, Benchmarking Public Procurement (World Bank, 2016), notes that many governments seek to “level the playing field” for SMEs through various mechanisms – e.g., requiring less in the way of bid security or by setting aside a share of total procurement for SMEs, but this is unfortunately not an indicator for which data are reported in this benchmarking exercise.

(23)

procurement. We hope our findings will help stimulate greater interest in collecting longitudinal information on public procurement participation by firms and sourcing behaviour of government entities.

Better data will allow a better understanding of the role of foreign-owned companies in procurement, e.g., the finding that greater foreign participation in public procurement is associated with stronger positive performance measures for those firms that sell to the government. Better understanding of the interrelationships between explicit FDI barriers and indirect barriers (quality of procurement process; corruption); the channels through which higher foreign shares impact on domestic productivity performance (e.g., whether there are learning spillovers); and the design/role of investment promotion agencies could complement research on the effects of procurement processes narrowly defined. The large positive and significant coefficient estimate for the interaction between the share of total sales to the government and the prevalence of inward investment promotion activities in the relevant sectors suggests there may be important complementarities. These and other questions – such as our finding that procurement is not correlated with higher productivity for firms in services sectors – all require additional data on overall procurement activity at the national level and its allocation by sector and type of firm.

(24)

References

Acemoglu, D., Akcigit, U., Bloom, N., and W. R. Kerr (2013) “Innovation, Reallocation and Growth,” NBER Working Paper 18993.

Altenburg, T. and W. Lütkenhorst (2015) Industrial Policy in Developing Countries: Failing Markets, Weak States, Edward Elgar: Cheltenham

Aschhoff, B. and W. Sofka (2009) “Innovation on demand—Can public procurement drive market success of innovations?,” Research Policy, 38: 1235-47.

Auriol, E., S. Straub and T. Flochel (2016) “Public Procurement and Rent-Seeking: The Case of Paraguay,” World Development 77(C): 395–407.

Best, M.C., Hjort, J. and D. Szakonyi (2017) “Individuals and Organizations as Sources of State Effectiveness, and Consequences for Policy,” NBER Working Paper 23350. Blundell, R. and M. Costa Dias (2009) “Alternative Approaches to Evaluation in Empirical

Microeconomics”, Journal of Human Resources, 44(3): 565-640.

Branco, F., (1994) “Favouring domestic firms in procurement contracts,” Journal of International Economics 37, 65–80.

Brulhart, M., Trionfetti, F. (2004) “Public expenditure, international specialization and agglomeration,” European Economic Review 48, 851–881.

Carbone, C., Iossa, E. and G. Mattera (2018) “Barriers towards foreign firms in international public procurement markets: a review,” Economia e Politica Industriale 45: 85-107. Cernat, L. and Z. Kutlina-Dimitrova (2017) “International Public Procurement: From Scant

Facts to Hard Data,” in A. Georgopoulos, B. Hoekman and P. Mavroidis (eds.) The Internationalisation of Government Procurement Regulation. Oxford University Press.

Czarnitzki, D., P. Hünermund and N. Moshgbar (2018) “Public procurement as policy instrument for innovation,” ZEW - Centre for European Economic Research Discussion Paper No. 18-001

Dengler, B. and B. Hoekman (2018), “The WTO Government Procurement Agreement as a Commitment Device: A First Appraisal,” CEPR Discussion Paper 13266.

Djankov, S., Islam, A. and F. Saliola (2016) “How Large Is Public Procurement in Developing Countries?” Peterson Institute for International Economics: Washington, DC.

Edler, J and L. Georghiou (2007) “Public procurement and innovation—Resurrecting the demand side,” Research Policy 36: 949–963.

Evenett, S. and B. Hoekman (2005) “Government Procurement: Market Access, Transparency and Multilateral Trade Rules,” European Journal of Political Economy, 21(1): 163-183.

Evenett, S. and A. Shingal (2006) “Monitoring Implementation: Japan and the WTO Agreement on Government Procurement,” in S. Evenett and B. Hoekman (eds.), Economic Development & Multilateral Trade Cooperation, London: Palgrave-McMillan.

Evenett, S. and A. Shingal (2016). "Gauging Procurement Policy Change During the Crisis-Era: Evidence from the Global Trade Alert," PRONTO working paper, Bern: World Trade Institute.

Ferraz, C., Finan, F. and D. Szerman (2015) “Procuring Firm Growth: The Effects of Government Purchases on Firm Dynamics,” NBER Working Paper No. 21219. Foster-McGregor, N., Isaksson, A. and F. Kaulich (2014) “Importing, Exporting and

Performance in sub-Saharan African Manufacturing Firms,” Review of World Economics, 150(2): 309-36.

(25)

Francois, J., D. Nelson, and D. Palmeter (1997) “Government Procurement in the US: A Post-Uruguay Round Analysis,” in B. Hoekman and P. Mavroidis (eds.), Law and Policy in Public Purchasing: The WTO Agreement on Government Procurement, Ann Arbor: University of Michigan Press, 105–24.

Georghiou, L., J. Edler and J. Yeow (2014) “Policy instruments for public procurement of innovation: Choice, design and assessment,” Technological Forecasting & Social Change 86: 1–12.

Geroski, P.A. (1990) “Procurement policy as a tool of industrial policy,” International Review of Applied Economics, 4 (2), 182-198.

Gold, R., Görg, H, Hanley, A. and A. Seric (2017) “South-South FDI: is it really different?” Review of World Economics, 153 (4), 657-673.

Gourdon, J. and J. Messent (2017) “How government procurement measures can affect trade,” OECD Trade Policy Papers No. 199, OECD Publishing: Paris.

Harding, T. and B. Javorcik (2012) “Foreign Direct Investment and Export Upgrading,” The Review of Economics and Statistics, 94 (4): 964-80.

Hebous, S. and T. Zimmerman (2016) “Can Government Demand Stimulate Private Investment? Evidence from U.S. Federal Procurement,” IMF Working Paper No. 16/60

Helpman, E., M. Melitz, and S. Yeaple (2004) “Export versus FDI with heterogeneous firms,” American Economic Review, 94(1): 300-316.

Hoekman, B. (2017) “International Cooperation on Public Procurement Regulation,” in A. Georgopoulos, B. Hoekman and P. Mavroidis (eds.), The Internationalisation of Government Procurement Regulation, Oxford University Press.

Iacovone, L., Ramachandran, V. and M. Schmidt (2014) “Stunted Growth: Why Don’t African Firms Create More Jobs?” Center for Global Development Working Paper 353

Kattel, R. and L. Veiko (2010) “Public procurement as an industrial policy tool: an option for developing countries?,” Journal of Public Procurement 10(3): 368–404

Knack, S., Biletska, N. and K. Kacker (2019) “Deterring Kickbacks and Encouraging Entry in Public Procurement Markets: Evidence from Firm Surveys in 90 Developing Countries,” World Bank Economic Review, DOI: 10.1093/wber/lhy016

Kutlina-Dimitrova, Z. and C. Lakatos (2016) “Determinants of direct cross-border public procurement in EU Member States,” Review of World Economics, 152: 501-28.

Lee, M. (2017) “Government Purchases, Firm Growth and Industry Dynamics,” University of San Diego (mimeo).

Leuven, E., and B. Sianesi (2003) “psmatch2: Stata module to perform full Mahalanobis and propensity score matching, common support graphing, and covariate imbalance testing,” Boston College Department of Economics, Statistical Software Components. At: http://ideas.repec.org/c/boc/bocode/s432001.html.

McMillan, M., Rodrik, D. and I. Verduzco-Gallo (2014) “Globalization, Structural Change, and Productivity Growth, with an Update on Africa,” World Development, 63: 11-32. OECD (2005) OECD Science, Technology and Industry Scoreboard 2005. Paris: OECD. OECD (2013) Perspectives on Global Development 2013: Industrial Policies in a Changing

World, Paris: OECD.

Ramboll Management Consulting and HTW Chur (2011). Cross-Border Procurement Above EU Thresholds. Copenhagn: Ramboll.

Rickard, S. and D. Kono (2014) “Think globally, buy locally: International agreements and government procurement,” Review of International Organizations, 9(3): 333-52. Sanfilippo, M. and A. Seric (2016) Spillovers from Agglomeration and Inward FDI: A

Riferimenti

Documenti correlati

Notably, the so-called focal adhesion density, i.e., the number of focal adhesions per surface unit, increased on both the alumina surfaces significantly compared to the

Recently, the councillors for housing of the cities of Rome, Milan and Naples, the three largest urban areas in Italy, wrote a joint open letter in order to ask the Government to

The proposed compact antenna represents a good candidate for a number of wireless applications, especially in the automotive industry where the available space under

Seasonal operating (ΦPSII, left) and maximum potential (Fv/Fm, right) efficiency of photosynthesis of the three species studied (Achillea maritima, Achmar;

We employed Mendelian randomization (MR) to evaluate the causal relationship between leukocyte telomere length (LTL) and amyotrophic lateral sclerosis (ALS) with summary

A Milano, il collegio degli avvocati e giudici (via via si uniformano a quello della capitale gli sta- tuti delle altre città nel ducato) nella sua prima normativa trecentesca si

Nel ‘500 il cronista lucchese Giuseppe Civitale percepiva ancora la Garfagnana, da Sillano ai monti pistoiesi ormai sottoposti a Firenze, come «l’antico e naturale

In this new dimension of subjectivity, the Weberian theme of contractual freedom takes on an unprecedented rationalizing value: on the one hand, because one can