Public Procurement and the Private Business Sector: Evidence from Firm-Level Data

The quality of the public procurement system of an economy can have far-reaching effects on the private sector. This paper empirically explores several of these effects using two rich data sets. An overall indicator of public procurement quality is created from the World Bank?s Benchmarking Public Procurement project that is then combined with firm-level data from the World Bank Enterprise Surveys. The analysis includes more than 59,000 firms spanning more than 109 economies. The paper finds that firms in economies with good public procurement systems are more likely to participate in public procurement, face lower losses from shipping to domestic markets, and experience lower incidence of bribery than economies with poor public procurement systems. Similarly, better public procurement systems are positively correlated with more engagement in innovation, research and development, international certification, foreign technology adoption, and online connectivity.


Introduction
Governments set and implement the rules that guide the economy. Governments also use resources to provide public goods such as infrastructure. Public procurement is the one area where both government spending and government rule setting come into play. The size of public procurement in most economies is significant. On average public procurement makes up about 14.5 percent of GDP, with countries such as Eritrea and Angola going up to as high as 33% and 26% respectively (Djankov et al., 2017). In developing economies, public procurement can go up to 50% or more of total government expenditure (Knack et al., 2017). The quality of public procurement can have far-reaching effects throughout the economy given its magnitude.
Poor public procurement systems can incur significant costs on the economy. Opaque systems can increase rent-seeking behavior. Favoritism in awarding contracts can increase corruption, discouraging fair competition that could otherwise drive prices down and increase quality (World Bank, 2016). A fair and transparent public procurement system can encourage greater firm participation, decrease corruption, and improve the quality of public goods such as infrastructure. The use of online procurement portals can also encourage online connectivity in the private sector. Public procurement systems can also push the boundaries of innovation in the private sector. Using rich firm-level data we explore the far-reaching effects of public procurement systems throughout the business sector. We test whether public procurement systems are correlated with firm engagement in procurement, the prevalence of corruption in the business sector, the losses faced by firms when shipping domestically, firm innovation, and online connectivity.
The public procurement literature has explored certain aspects of public procurement and linked it to a number of outcomes. Several studies have explored the role of discriminatory public procurement policy (Vagstad, 1995;Krasnokutskaya and Seim, 2011;Nakabayashi, 2013). Branco (2002) explores the effects of favoritism in procurement on technology adoption. Strands of the literature have focused on other elements of public procurement such as audits (Di Tella and Schardgordsky, 2003), competition (Estache and Iimi, 2008;Ohashi 2009), and reputational mechanism (Spagnolo 2009). Very few studies have explored the effects of the overall public procurement system.
The study builds on two recent articles in the literature that study the overall public procurement system.
We use the public procurement score created by Djankov et al, (2017), which explored the effects of public procurement on road-quality outcomes at the country-level. The data source is the World Bank's Benchmarking Public Procurement database (BPP). The public procurement score adopts a lifecycle approach. Main stages of the procurement lifecycle are identified -bid preparation, bid and contract management, and payment of suppliers. An aggregated score of public procurement (PP Overall Index) is crafted that incorporates important aspects of these stages. A higher score implies a higher quality of the public procurement system. The second closely related study is by Knack et al. (2017) that also uses country-level measures of public procurement and links it to firm-level engagement and corruption. We build on this study in two ways. First, we explore a larger range of outcomes capturing infrastructure quality and innovation. Second, Knack et al. (2017) utilize the public procurement data from the Public Expenditure and Financial Accountability (PEFA) assessments.
PEFA assessments are built on the likelihood of accomplishing desirable outcomes based on seven pillars of performance. There is some thematic overlap between the BPP and PEFA assessments, particularly with regards to two of PEFA's pillars: predictability and control in budget execution and accounting and reporting. The BPP database is far more extensive in covering specific details of the public procurement system, going beyond the aggregate assessments in PEFA. Whereas BPP presents an assessment of the regulatory framework applicable to public procurement, PEFA does not measure the legal framework nor institutional capacity affecting public financial management. The PEFA has some comparability challenges given that assessments are done at the national level for some countries, and at the sub-national level for others (e.g. Afghanistan as opposed to Albania). Finally, although the quality of PEFA assessments has improved over time, only 18% of draft reports submitted in FY 2015 were awarded a PEFA CHECK, a non-mandatory yet indicative quality endorsement of the assessment requiring it to undergo a multi-step peer review process. This is the first study to combine the World Bank's Benchmarking Public Procurement database and firmlevel data from the World Bank Enterprise Surveys. We find that better quality of public procurement systems is correlated with positive firm engagement, infrastructure, innovation, and internet connectivity outcomes. Better quality of public procurement systems is also related to lower levels of corruption in the private sector. The rest of the paper is organized as follows. Section 2 lays out the conceptual considerations and section 3 describes the data. Section 4 provides the empirical estimation strategy, while section 5 presents the results and section 6 concludes.

Conceptual Considerations
In this section, we describe the number of ways the quality of public procurement can influence the five outcomes of interest: firm participation, infrastructure, corruption, innovation, and online connectivity.

Firm participation
The goal of good procurement systems is to encourage competition between firms bidding for contracts.
Transparency and accessibility can lead to greater firm participation in the bidding process. Transparency and accessibility will inform a larger number of firms of procurement opportunities, and also encourage productive firms to participate given the trust generated from an open process. Greater participation of firms has several benefits, both pecuniary and ones related to the well-functioning of the procurement system.
Non-competitive procedures in the procurement process can lead to increases in costs by 30 percent or more (Hoekman, 1998). But cost is not the only factor of concern. Better value for money and improved quality of goods and services are also results of healthier competition. Increased competition can reduce the chances of bidder collusion. It gives the public more confidence in the way public funds are spent and equips them with a powerful accountability tool. It is the main pathway through which most of the other outcome variables addressed below are affected. Thus, the empirical analysis must establish the first-order effect of increased participation of the private sector due to high-quality public procurement systems. Initial evidence is suggestive that this is the case (Knack et al., 2017).
Infrastructure A good procurement system increases the chances that productive firms will be awarded the contract, leading to delivery of high-quality products. Given that government procurement projects involve large infrastructure products, hiring productive firms will lead to timely and better quality infrastructure projects. Lewis-Faupel et al. (2016) find that the use of e-procurement for example leads to increases in road quality in India. Djankov et al. (2017) find a positive correlation in cross-country data between better quality public procurement systems and infrastructure quality.

Corruption
Poor public procurement systems characterized by a lack of transparency can be a channel through which corruption permeates throughout an economy. The lack of transparency and competition can allow public officials to use public procurement as a means of eliciting bribes. When procurement is less transparent, government officials use discretion to decide which firms receive the contract, creating a breeding ground for corruption (Ohashi, 2009). Poor public procurement systems have been found to be vulnerable to corruption (Auriol et al., 2016). In addition, opaque public procurement systems can also set the tone for other transactions between the government and the private sector and promote inefficiencies. Thus, if a poor public procurement system signals lower costs of rent-seeking behavior, other arms of the government may also engage in rent-seeking activities. Open competition may not be sufficient to deter corruption.
Contracts can be awarded to the firm offering the highest bribe instead of the firm offering the highest 6 quality or lowest price (Knack et al., 2017). Mironov and Zhuravskaya (2016) show that when procurement contracts are determined by bribes, less productive firms are awarded contracts.

Innovation
Public procurement systems can be used as a tool to improve innovation and technology adoption in the private sector (Branco, 2002;Hommen and Rolfstam, 2009). To be competitive in bidding for procurement contracts, firms may have to be innovative and adopt technologies to reduce costs. Public procurement can directly require innovation to be a criterion in winning the bid, thereby promoting innovation in the private sector. Furthermore, focusing on innovations in the final product can induce innovation in the private sector (Edler and Georghiou, 2007). Rothwell and Zegveld (1981) found that state procurement triggered greater innovation impulses in more areas than did R&D subsidies. Geroski (1990) determined that procurement policy "is a far more efficient instrument to use in stimulating innovation than any of a wide range of frequently used R&D subsidies." As recent policy reviews have shown, public procurement innovation is at the heart of many innovation policy initiatives across the OECD and at the EU level (Izsak and Edler, 2011;OECD, 2011;Rigby et al., 2012;Uyarra, 2016). Moreover, early engagement of suppliers is an important element in procurement for innovation. Through foresight effort and other joint activities, a common identification of needs can be shared between the demand and supply sides. When used as a policy to promote innovation, public procurement will generate varying degrees of collaboration and interactive learning (among procurers, suppliers and -sometimes -other organizations), which is a central determinant of the development and diffusion of innovations (Edquist et al.2015). Finally, the public sector can lower the risk for the developing firms and subsequent customers by acting as a launching customer for innovative technologies and solutions (European Commission 2005).

Online connectivity
Online connectivity is a more direct outcome from public procurement than innovation, although the mechanisms at play are similar. External pressure is a key motivator in the adoption of internet technologies 7 by firms (Mehrtens et al, 2001). Public procurement, through e-procurement, can directly lead firms to adopt internet technologies as it is a requirement to fully harness the procurement process. Finally, eprocurement can generate competitive pressures as firms compete with each other by adopting internet technology in order to outdo each other to win the contract. The latter is more likely to be true for small and medium-sized (SME) enterprises in developing economies.

Data
The analysis is based on two data sets -the World Bank Group's Enterprise Surveys (ES) and the World Bank Group's Benchmarking for Public Procurement databases. The ES consist of firmlevel data that capture a firm's business environment. The respondents are typically managers or owners of the business. In addition, a firm's characteristics and performance are measured. The ES conducted between 2006 and 2016 used a common questionnaire and sampling methodology (stratified random sample) across economies, thereby allowing for cross-country comparisons, which is a rarity in most data sets. The surveys are representative of the formal (registered) private sector of the economies excluding extractive sectors such as mining as well as agriculture. 2 Measures of participation in public procurement, corruption, infrastructure, innovation and online engagement available in the ES are utilized for the analysis in this study. The sample of ES firms in this study includes over 59,000 firms across 109 mostly developing economies. The list of countries is provided in table A1.
The public procurement data are based on structured expert surveys. This database has also been used by Djankov et al. (2017). Respondents were chosen based on their expertise in public 8 procurement law as well as advisory experience for businesses willing to provide services to the government. The respondents range from private sector companies, professionals in law firms, accounting firms, business advisory firms, chambers of commerce, legal bar associations, to public officials dealing with public procurement. Over 1,900 experts provided information that was coded by the World Bank team. To enable cross-country comparisons, a hypothetical scenario was developed to anchor survey responses, similar to the approach by Djankov et al. (2002). The standardized case study entails assumptions on three elements: (i) the procuring entity, (ii) the bidding company, and (iii) the public call for tender. The procuring entity is restricted to a local authority located in the economy's largest business city, and is planning to resurface a flat twolane road with asphalt. The bidding business is assumed to be a limited liability company that also operates in the economy's largest business city, and is 100 percent domestically and privately owned. The bidding business is assumed to have previously responded to public calls for tender and won similar-size service contracts. The following assumptions are made regarding the public call for tender. First, it is initiated by the procuring entity. Second, it follows an open and competitive process. Third, the public tender concerns the resurfacing with asphalt of a flat twolane road. The value is defined as the greater of: (i) 91 times the economy's income per capita or (ii) $2 million.
The methodology of the BPP has a few limitations. First, the surveys are not based on a representative sample. The assumption is that public procurement is within the scope of experienced experts and therefore a small number of experts would be able to respond with precision to the survey. Second, the data are cross-sectional for a single year (2016). Finally, the data focus on a set of procurement indicators in the largest business city, thereby ignoring the heterogeneity of public procurement within economies, especially large federal states.

Empirical Estimation
By combining firm-level heterogeneity in outcome indicators with country-level variation in public procurement, we estimate the following equation for firm i in country j and sector r: We use five main types of firm-level outcome variables to capture different dimensions of the private sector ( ) obtained from the Enterprise Surveys (ES). These include firm participation in public procurement, road infrastructure, corruption, innovation, and online engagement. For public procurement participation, we use the variable capturing whether or not a firm attempted or secured a government contract in the last 12 months. About 19 percent of firms attempted or secured a government contract. The same measure was used by Knack et al. (2017).
For road infrastructure quality, we use the measure of the percentage of products lost to breakage or spoilage during shipping to domestic markets. A similar measure has been used by Alterido et al. (2011). Around 1 percent of product value was lost for firms in the sample due to breakage and spoilage during domestic shipping. Corruption is measured as a binary variable capturing whether a firm experienced at least one bribe request across size public transactions. About 18 percent of firms faced corruption.
Innovation is captured through five variables from the ES data set. These include: whether the firm has engaged in product innovation, process innovation, spent on R&D, used technology licensed from foreign firms (manufacturing firms only), and whether the firm has an internationallyrecognized quality certification. Paunov (2016)  whether or not firms experienced losses from crime. Access to finance is proxied using two variables-whether the firm has a checking or savings account and whether the firm has a line of credit or loan.
We control for the current state of labor markets by capturing aggregate demand through the growth rate of GDP per capita ( ). Finally, we also account for the level of development ( ) and land area ( ) following Knack et al. (2017) . We also worry about industry-specific factors and region (continent) specific factors. We account for sector using a dummy variable for the service sector ( , with manufacturing being the omitted sector. Similarly, we use continent fixed effects ( to account for time-invariant regional factors. Finally, certain countries with common law systems may adopt a different public procurement system as the scope of the public procurement regulations may be reduced. This accounted for using a dummy variable for common law countries ( ). Summary statistics can be found in table 1, with data description and sources provided in table A2.

Results
Table 2 presents the findings for firm participation in public procurement, road infrastructure quality, and corruption. Better public procurement systems are positively associated with higher participation of firms in public procurement. The coefficient of the public procurement variable is positive and statistically significant at the 1% level. This finding is consistent with Knack et al.
(2017). Higher public procurement scores (better quality) are negatively correlated with corruption and products lost to breakage or spoilage during shipping in domestic markets. The coefficient of public procurement is negative and statistically significant at the 1% level for the corruption estimation, and 5% for the domestic infrastructure quality estimation. Table 3 presents the findings for public procurement quality and innovation. Better public procurement is positively associated with all five proxies for innovation -product innovation, process innovation, R&D spending, technology licensed from foreign firms, and internationally recognized quality certification. The coefficient of public procurement quality is statistically significant at the 1% level for all types of innovation with the exception of R&D spending, where it is statistically significant at the 5% level. The two other consistent results across all types of 13 innovation are the positive associations with firm size and formal training. Large firms and firms providing formal training for their employees tend to be more innovative.
Findings for the relationship between public procurement quality and online engagement are provided in table 4. Better quality of public procurement is associated with a higher probability of firms engaging suppliers through e-mail and having their own website. For both measures, the coefficient of public procurement quality is statistically significant at the 1% level. The implication may be that public procurement encourages online engagement through online procurement portals thereby encouraging firms to be more engaged online.
Public procurement may have heterogenous effects in the private sector conditional on the size of the firm. Small and large firms may react differently to the quality of the public procurement system. Thus, we split the sample into two: small and medium (SME) size firms (5 to 99 full-time employees) and large firms (100 plus full-time employees). We repeat the estimations separately for each group. The results from table 2 are split by small and large firms and provided in table 5.
Public procurement leads to greater participation of firms in public procurement, lowers corruption, and reduces losses from domestic shipping regardless of firm size. The coefficient of public procurement is statistically significant, at least at the 10 percent level.
In table 6 we repeat the estimations in table 3 by SME and large firms. Public procurement quality encourages process innovation, R&D spending and internationally recognized quality certification regardless of firm size. However, for product innovation and technology licensed from foreign firms, the coefficient of public procurement is statistically significant only for SMEs. The implication may be that the quality of public procurement has a greater influence in the incidence of production innovation and foreign technology among SMEs than large firms. In table 7 we   repeat the estimations of table 4 by SME and large firms. The quality of public procurement has a positive effect on the probability of a firm having its own website or engaging with clients or suppliers via e-mail regardless of firm size. The findings are statistically significant at least at the 10 percent level.
Finally, we explore the relationship between public procurement quality and firm outcomes by sector -manufacturing versus services firms. Table 8 repeats the results of table 2 by sector. Public procurement quality increases firm participation in public procurement and reduces corruption regardless of whether the firm is in the manufacturing or service sector. However, the negative association between public procurement quality and products lost in domestic shipping is only statistically significant for manufacturing firms. This finding could be because manufacturing firms are more likely to ship larger amounts of goods domestically.
The findings for innovation by sector are presented in table 9. Note that adoption of technology licensed from foreign firms is omitted as the survey question was only asked of manufacturing firms. The findings indicate that public procurement quality has a positive influence on product innovation, process innovation and R&D spending for manufacturing firms, but no statistically significant effect for service firms. However, public procurement quality has a positive relationship with the presence of internationally recognized quality certification for both manufacturing and service firms. The data seem to indicate that in terms of innovation, the quality of public procurement has a far greater effect on manufacturing firms than services firms. With 15 regards to online engagement, as reported in table 10, public procurement quality has a positive coefficient regardless of the sector of the firm.

Conclusion
Good quality public procurement systems may have direct and indirect effects on private businesses. They can generate a domino effect by raising certain aspects of firms that may lead to productivity increases. In this study, through two unique data sets, we find strong positive correlations between good public procurement systems and firm engagement, infrastructure quality, innovation and online connectivity. Good procurement systems are negatively correlated with corruption faced by the business sector. Given that public procurement is sizeable in many economies, and its effects may be multifaceted, reforming public procurement systems is an important endeavor to improve the business environment in the economy.
This study has some limitations. It is difficult to argue for causality given the data at hand. It is not possible to disentangle the direction of causality in many of the estimations. While the current study is rich in terms of country coverage and detailed in terms of the wealth of information on firms, future studies may adopt a less holistic approach and explore exogenous changes in aspects of public procurement systems. This could allow for some specific causal statements that would complement the current study.            Explores elements that form part of the bid preparation phase, such as the existence of procurement portals, the cost and accessibility of relevant information, and the openness and transparency on how this preparation phase is conducted.

Djankov et al., 2017
Bid and Contract Management Score Combination of the following elements of procurement: Bid submission, bid opening, evaluation and award, and the content and management of procurement contract. Bid submission measures the ease of submitting bids, including the procedures and costs involved in the process and the availability of electronic means to submit the bids. It also measures that the legal framework provides a minimum time to submit the bids and regulates the amount of bid securities. Bid opening, evaluation, and award assesses whether the bid opening, evaluation and contract award are conducted through an open and fair process in order to guarantee bidders that the process follows the best standards of transparency and that losing bidders are timely informed on the procuring entity's decision. Content and management of procurement contract examines the procedures involved during the execution of the contract until its completion or its termination. It also examines the existence of controls regarding modifications of the contract, including communicating those variations to other interested parties.

Djankov et al., 2017
Payment of Suppliers Score Examines whether the legal framework regulates the payment of suppliers. It also assess the time needed for the purchasing entity to start processing the payment once the invoice is submitted as well as the time in practice for suppliers to obtain payment once they submit their invoice. It also examines whether interests/penalties are paid in case of payment delays, whether they are automatic and the method for determining them