Report No: AUS0001387 . Brazil Procurement Reforms in Amazonas . February 22, 2020 . GOV / DIME . © 2017 The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved This work is a product of the staff of The World Bank. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Attribution—Please cite the work as follows: “World Bank. {YEAR OF PUBLICATION}. {TITLE}. © World Bank.” All queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. 2 Who Chooses to Sell to the Government? Understanding the Determinants and Consequences of Participation in Brazilian Public Procurement Brazil P156357 Keywords: Firm Behavior, Organizational Behavior, Transaction Costs, Regulation, Procurement JEL Codes: D21, D23, H32, H57, L51 3 TABLE OF CONTENTS 1. EXECUTIVE SUMMARY .................................................................................................... 1 2. BACKGROUND AND KEY INSTITUTIONAL FEATURES.......................................... 2 3. LITERATURE REVIEW ..................................................................................................... 2 4. POLICY RELEVANCE ........................................................................................................ 4 5. DATA ..................................................................................................................................... 4 5.1 QUANTITATIVE LICITACAO (BIDDING) DATA ...................................................................... 4 5.2 BANCO DE PRECOS (BENCHMARK PRICE) DATA .................................................................. 5 5.3 SURVEY OF POTENTIAL FIRMS ............................................................................................. 5 6. RESULTS AND DISCUSSION .......................................................................................... 5 6.1 PARTICIPATION IN PROCUREMENT ..................................................................................... 5 6.2 IMPACT OF PARTICIPATION .............................................................................................. 12 6.2 EFFECTS OF "MINI-AUCTIONS" (CEL) REFORM ............................................................... 14 6.2 CAUSES OF NON-PARTICIPATION ..................................................................................... 17 7. CONCLUSION AND DIRECTIONS FOR FUTURE RESEARCH ............................ 18 REFERENCES ............................................................................................................................ 18 1. EXECUTIVE SUMMARY This report responds to the question “who sells to the government?”. Using a large administrative dataset on the universe of procurement auctions for the State of Amazonas from 2010-2016, this report documents several stylized facts on firm participation in public procurement. Despite a large pool of firms that could potentially supply products to the government, participation in procurement auctions is relatively low and decreasing over time. This trend significantly impacts the state’s ability to procure the items needed for government functions at a reasonable price. The study reveals that an increase in the number of firms that participate in a given auction reduces both the probability of failure and the final price for a good. Consequently, incentivizing firms to enter into public procurement is a first-order concern for the state government. The project conducted a survey with 150 suppliers in Amazonas state to better understand the main barriers to participating in government procurement auctions, discussing the factors that led them to choose (or not) to sell to the government. The survey revealed two main causes for non-participation in government procurement: issues with receipt of payment; and bureaucratic hurdles in preparing a bid. This report is the result of work on an impact evaluation (IE) aimed at studying the effects of lowering transaction costs to participating in public procurement. The initial project proposed in March 2016 was focused on procurement reforms introduced by the state of Amazonas and its capital city, Manaus, with the aim to reduce the excess of bureaucracy and foster competition in procurement processes. The IE proposed a randomized rollout of an eDocuments platform at the state level aimed at streamlining firm registration by migrating the process online, as well as a randomized rollout of an electronic auction (eAuction) platform for procurement at the municipal level. Despite close collaboration with these governments, logistical and institutional hurdles made implementation of these reforms unfeasible as initially proposed. Nonetheless, the study generated valuable insights from descriptive analysis of the administrative data provided by the Amazonas government. 1 2. BACKGROUND AND KEY INSTITUTIONAL FEATURES Within Brazil, governments utilize a variety of different procurement methods. Procurement systems in the state of Amazonas are relatively sophisticated with respect to other states in Brazil. The state government has been using an eAuctions platform for procurement auctions since 2006. However, other aspects of the procurement process have not matched the online auction platform in terms of sophistication and ease of use. For any bidding process, bidders are required to present four sets of documents: legal documents proving the existence of the company, tax documents for all three levels of government, technical documents with the details of the product offered and financial documents showing the financial status of the company. These documents must be collected from several different government agencies, have rolling expiration dates, and must be presented at every auction. In addition, many firms find the time allowed for bid preparation is insufficient to gather all required documents, with some firms resorting to hire specialists to work through all of the government regulation. These upfront bureaucratic hurdles present a high cost to entry for firms wishing to supply to the government. The current high barriers to entry possibly explain the higher prices paid by governments vis-a vis the private sector. Tripadalli (2013) finds that, compared to the private sector, the Federal government and Amazonas government pay an average mark-up of 49% and 27%, respectively. State government agencies wishing to procure items are required to tender through the public procurement process, with purchases procured in different methods depending on the nature of the purchase: - Pregao Eletronico (PE): full procurement auction open to all firms properly registered with the government. Procurement is done through a second price auction with a reserve price and additional negotiation with the winning bidder. - Registro de Compra Direta (RCD) / Registro de Dispensa de Licitacao (RCL): noncompetitive procurement purchases for small-value items, up to R$8000. For purchases below this threshold, agencies can choose not to run a full procurement auction and instead request up to three bids from firms (chosen by the purchasing agency), and choose the most attractive bid. - Compra electronica (CEL): A “mini-auction” process for small-value items, up to R$8000. For purchases below this threshold, agencies can use the mini-auction which greatly simplifies the paperwork required by firms and the process to publish the initial tender. CEL purchases are open to all bidders that meet the requirements, and a winning bid is chosen through an online auction. The process was created as an alternative to the noncompetitive RCD/RDL processes. 2 3. LITERATURE REVIEW Government regulation imposes significant costs on businesses around the world, and Brazil is no exception: managers report spending an average of 14% of their time on dealing with government regulation (World Bank, 2009). Businesses that wish to sell to the government face additional hurdles. These range from inflexibility by the government on adherence to technical specifications (Asker & Cantillon, 2008), to time costs of preparing bidding documents, complying with documentary requirements and delays (Krasnokutskaya & Seim, 2011), to the costs of sustaining collusion (Conley & Decarolis, 2015; Kawai & Nakabayashi, 2015), and outright bribery (Tran, 2008). Which firms the government allocates contracts to has important implications both for the firms themselves (Ferraz, Finan & Szerman, 2015, show that as a result of winning Brazilian federal government contracts, firms grow by 2.2 percentage points more) and for the overall efficiency of the economy (Hsieh & Klenow, 2009, find that misallocation of resources to inefficient firms, possibly driven partly by regulation and institutions, can explain around half of the total factor productivity difference between India and China and the United States). Public procurement auctions have been studied extensively in the empirical industrial organization literature (see e.g. the survey by Hendricks & Porter, 2007). A number of papers have studied policies to reduce corruption in the procurement system (Olken, 2007; Litschig and Zamboni, 2011; Gerardino et al. 2015; Lichand et al. 2015), In addition, there has been a recent surge of work using public procurement as a context in which to study the effectiveness of the state. The most prominent of these has been Bandiera et al. (2009) who show that quality adjusted prices paid for off the shelf goods in Italy vary dramatically from office to office. Of the waste from paying higher prices 83% is due to bureaucratic inefficiency as opposed to active waste driven by corruption. While the burden of such inefficiencies can be substantial, it is a priori ambiguous whether lowering transactions costs of entry into public procurement auctions improve procurement outcomes. In order to avoid corruption, procurement systems are highly regulated and the bid selection criteria is usually rules-based to remove discretion from the bureaucrats. Often governments use a lowest bid rule, 1 in which any firm that meet the eligibility criteria can participate and the lowest bid is picked. Since quality cannot be fully contracted ex ante, a price competition can screen out high quality suppliers. In this context, lowering transaction costs to increase competition may not necessarily increase value-for-money. Therefore, on the one hand, if firms that have a comparative advantage in bearing the burden of the hurdles associated with dealing with the government are also the most efficient producers, then government hurdles would help the government screen suppliers. On the other hand, if low quality of suppliers are better at 1 All levels of government in Brazil, India, China, Mozambique, Colombia, Mexico, among other countries, use a “two step rule” in public procurement: (i) an eligibility criteria based on a number of financial qualification requirements and tax documents; (ii) pick the lowest-price bid among qualified suppliers. 3 dealing with these costs of doing business with the government, these hurdles may screen out good suppliers, so more competition could indeed increase value-for-money. The existing literature on this question is inconclusive. On the one hand, seminal theoretical work by Bulow & Klemperer (1996) argues that encouraging entry into auctions should be expected to lead to large improvements in auction outcomes. Empirically, Coviello and Mariniello (2014) show that lowering barriers to entry by publicizing procurement auctions more widely increased participation in procurement auctions in Italy and reduced procurement costs. Lewis-Faupel et al. (2016) study the impact of the adoption of electronic procurement in India and Indonesia and find that it led to increased participation in auctions and improvement on the quality of the procured projects and but not on prices. On the other hand, recent work has suggested that procurement outcomes are only affected by entry decisions when the entry decisions of the most competitive bidders are affected. For example, in a descending auction, only the lowest bid and the second-lowest bid affect the outcome of the auction, and so entry decisions only matter to the extent that they are likely to affect the lowest and/or second-lowest bids in the auction (Coey et al. 2015). Kang & Miller (2015) take this argument even further. They argue that if processing bids is costly for the government, and/or if the government seeks to favor particular types of bidders, it is likely to be optimal for the government to put in place barriers to entry to restrict entry and competition. Using a structural model, they find that the government could save up to 22% of its costs by restricting entry further, without significant sacrifices to quality. Relatedly, Coviello et al. (2015) provide quasi- experimental evidence from Italy that giving procurement officers discretion to restrict entry into procurement auctions leads to less entry and a greater chance that the same firm wins repeatedly, but does not lead to worse procurement outcomes. It is, therefore, an empirical question how barriers to entry affect the pool of suppliers and procurement outcomes. This report adds to the literature by providing insights into the role that participation plays in the procurement process in Amazons. Using rich administrative data and a survey of firms, the report explores how participation by firms have changed, the size of the pool of potential firms that could supply to the government, and the reasons that firms choose not to sell to the government. By providing new evidence on the selection of suppliers into procurement in a developing country context, this report can contribute to the debate on how procurement regulation can affect businesses decision to become a government supplier that directly affects value for money and state effectiveness. 4. POLICY RELEVANCE Inefficiency in public procurement is a serious obstacle to growth in emerging economies. While public procurement is central to a government's ability to provide vital goods and services to the public, it is also a government’s most vulnerable activity to waste and corruption due to the volume of financial flows it generates (IMF, 2016). It is estimated that procurement is one of the largest 4 government spending activities and accounts for up to 15 percent of global GDP. As procurement comprises a large share of the government expenditures, improvements in the effectiveness of government procurement can have first order impacts on the effectiveness of service delivery and public good provision. 5. DATA Analysis for this report relies on three main datasets. Two of the datasets are administrative data on the universe of procurement processes provided by the Amazonas government and cover many aspects of both procurement and the activity of firms in the private sector. The third dataset was generated by the research team through a survey of firms in Amazonas to gauge the barriers to participation by these firms in the procurement process. 5.1 LICITACAO (BIDDING) DATA The Licitacao data is the primary administrative dataset used to track procurement in the state of Amazonas This file contains all bidding-related data such as number of bidders, products, and procurement modality for the universe of procurement activity from 2010-2016. Data is organized at the process-item level: for each item in a given procurement process, the data contains information on the number and identity of bidders, their bidding behavior, the selected winner, how many suppliers downloaded information about the auction items, all the details of the awarding phase (including information about non-winning bids), and information of the payment and delivery phase. 2 This report restricts the analysis to materials (as opposed to services), as the technical specification and selection criteria are more standardized than requests for services. The licitacao dataset contains information on over 41,000 distinct items and approximately 4,700 unique firms for the 6-year period. 5.2 BANCO DE PRECOS (BENCHMARK PRICE) DATA For a subset of items in the procurement data, the Amazonas state government has a benchmark price directory (Banco de Precos) to aid in price management during the bidding process. Any item in the banco de precos has a “max price” per unit as determined by the Amazonas government, with any bidding process failing to produce a bid lower than this max price resulting in a failed process. The calculation of this max price is accomplished by the government’s access to rich tax data and details of transactions between firms (in the state) through their electronic receipts (Nota Fiscal Eletronica). By collecting information on similar items sold between firms in the private market, the government can calculate the maximum price that it is willing to pay for 2 The following provides an interactive look at this data: http://dataviz.worldbank.org/views/Licitacoes/1_Inicieaqui?:iid=1&:embed=y 5 an item. 3 Importantly for this report, this data makes it possible to identify the potential pool of suppliers for a given item, which is discussed further in Section 6.1.2. 5.3 SURVEY OF POTENTIAL FIRMS In order to better understand the main barriers to participating in government procurement auctions, the research team conducted a survey with 150 suppliers in Amazonas state. The survey was implemented over the phone in the period December 2015 and January 2016. 4 While the survey was qualitative in nature and covered a small number of firms, Section 6.4 provides important insights in the decision process for firms in choosing not to enter into the state procurement process and informs many of the stylized facts presented in the next section. 6. RESULTS AND DISCUSSION This section presents stylized facts on participation in the state government’s procurement process. The report first looks at the variation in participation in procurement by item and over time. It then studies the effects of increased participation by firms on various outcomes. Following the literature, the outcomes of particular interest are the final prices for goods and the failure rate. These outcomes have been highlighted by the state government as important indicators to gauge the improvement in the procurement process. The report next analyzes a recently launched government program aimed at reducing bureaucratic red tape – the “mini-auction” (CEL) - and how it affected these outcomes. This section concludes with a discussion of the survey of firms conducted by the research team that highlights the reasons firms choose not to sell to the government, and provides insights into potential policies that could increase participation by firms. 6.1 PARTICIPATION IN PROCUREMENT This section looks at the participation of firms in the procurement process. Using the licitacao data, one can calculate the number of unique firms that entered a given procurement process. The histogram below of the number of participants across all procurements from 2010-2016 displays a strong right skew, with the majority of auctions having fewer than 10 firms bidding on the lot. 3 The government’s max price usually includes a mark-up on the private market price to account for the additional costs of supplying to the government and to incentivize greater participation in procurement. 4 The questionnaire is available at http://pesquisacompras.com/ 6 This figure hides heterogeneity in the number of participants by procurement type. In particular, the large number of processes with one and three participants is driven by participation in the RCD/RDL processes. These types of “auctions” are designed for non-competitive small value (<$8000 Reals) procurement, where the purchasing agency collects a maximum of three offers from firms to choose from without a formal auction process. As RCD/RDL purchases are capped at three firms, they account for the large spikes on the left-hand side of the histogram. The standard online procurement auction (pregao eletronico) displays a smoother distribution of firm participation, although the majority of auctions have less than ten firms bidding on the items. 7 As a given procurement process is composed of lots with numerous items for which a firm can bid, it is revealing to look at participation across a given product, as opposed to the entire lot. The figure of products by firm participation largely mirrors the histogram of participation in auctions, with a spike at three firms being driven by the RCD/RDL process. Of greater interest is how participation changes by lot size. Below is a plot of the number of participants as a function of the lot size. There exhibits a strong increasing relationship in lot size and participation: the larger a lot (by value), the more firms participate in the auction. The largest 8 lots concern procurement of items in the medical field, and in particular hospital equipment (e.g. stretchers, medical utensils, bedding). That these are specialized items in addition to the fact that auctions with greater participation also have larger lot sizes suggest a large fixed cost to participation, and that firms are only willing to bear the bureaucratic costs to entry when the payoff is sufficiently large. 6.1.1 CHANGES IN PARTICIPATION OVER TIME The report next analyzes how participation in procurement has changed over the study period. The figure below plots the average number of participants in a given auction from 2010 to 2017. One can see that there has been a gradual decrease in participation over time, with a reduction of one participant on average over the seven-year period. 9 The decrease observed in the above figure may be due to changes in products offered as opposed to a decreased propensity for firms to enter into procurement. Controlling for the product being procured, a much stronger trend emerges, as evidenced by the figure below. Two characteristics of the figure bear noting. First, there is a pronounced seasonality to firm participation, with spikes in participation in January of a given year. This is most likely due to government agencies procuring framework agreements to set a price for certain goods at the beginning of the year. Second, after controlling for product, the decrease in participation over time is even larger, with an average decrease of two firms over the study period. Given the low participation numbers discussed in the previous subsection, this decrease represents a significant percentage drop of firms choosing to sell to the government. 6.1.2 POTENTIAL POOL OF SUPPLIERS The previous subsection presented evidence of a decrease in participation by firms in government procurement over time. Of importance for increasing firm engagement in procurement is to understand the size of the pool of firms that could potentially enter the public procurement process. 10 To understand the ecosystem of firms in Amazonas state, this report first looks at the linkages between firms that have made bids in a government procurement auction. The left panel of the above figure displays the connections between the various subgroups of products procured by the state government from 2010-2017. Each node represents a product subgroup (a collection of similar items, e.g. aparelhos e utensilios hospitalares) with two subgroups sharing a connection (the black line connecting the nodes) if a firm has bid on an item in both subgroups. From this figure, it appears that nearly all product types are connected, implying a densely interconnected network of firms supplying to the state government. However, many of these connections are due to so-called “middlemen” firms. These are wholesale firms that purchase a wide range of items with the intention of solely providing these items in public procurement auctions. These middlemen firms generally charge a higher markup than identical items sold in the private market, with anecdotal evidence suggesting middlemen firms include a markup to compensate for the inherent riskiness of selling to the government when payment schedules are not always fixed and transparent. If one removes the connections from “middlemen” firms, a more segregated network appears (right panel of the above figure). Firms generally cluster into providing similar classes of items, such as automotive products (blue circle) or apparel (teal circle). These smaller networks of firms providing similar products can be utilized in targeting firms that provide similar goods in the private market who may be incentivized to enter into public procurement. Utilizing the data from the Banco de Precos dataset, it is possible to gauge the size of the potential pool of suppliers. For a subset of products in the licitacao dataset, the government has calculated a max price to be used as a reservation price for the procurement auction. In effect, the max price 11 is a benchmark for how much the product is sold for on the private market. The calculation of this max price is accomplished by the government’s access to rich tax data and details of transactions between firms (in the state) through their electronic receipts (Nota Fiscal Eletronica). By collecting information on similar items sold between firms in the private market, the government can calculate the maximum price that it is willing to pay for an item. Importantly for this report, the electronic receipts data allows one to observe firms that have not participated in public procurement but have sold similar items in the private market. Using data from the max price creation, this report can analysis - for a subset of products (~9.84% of all products and 26.55% of auctions) - the universe of firms that can potentially supply a particular item to the government. As max price calculations became more common later in the study period, the subset of products for which it is possible to calculate the pool of potential supplier increased to 44.82% of all procurement auctions in 2015 and 2016. Restricting attention to those items that appear in the licitacao and banco de precos data for 2015- 2016, the analysis generated a histogram of the share of firms that participate in the Amazonas procurement process for a given product: The most striking feature of the above figure is the wide range of shares across products, with some products having virtually no participants of a potential pool of firms and other products having nearly universal participation by potential firms. The variation in firm shares is in part due to the specialization of products. As show in the tables below, the products with the smallest supplier pools are specialty products related to the biomedical field (e.g. surgical instruments, liquid nitrogen, pharmaceutical chemicals), whereas the products with the largest pool of potential 12 suppliers are common commodities or off-the-shelf consumer goods (e.g. sugar, fruit juice, plastic cups, ink cartridges). One can alternatively depict the pool of potential suppliers by plotting the total suppliers of a given product against the number of firms that participated in the procurement process. The figure below plots this relationship for total number (left panel) and average number (right panel) of participants, with the forty-five-degree line depicting products with complete participation by potential firms. The figure demonstrates that there is significant room for increased participation of firms in the procurement of most products requested by the state government. 13 6.2 IMPACT OF PARTICIPATION This report has demonstrated that participation in procurement run by the Amazonas state government was small (usually <10 firms) for a given product, decreasing over time, and has a potential to increase the pool of firms willing to supply products to the government. The analysis now studies the consequences of increased (or decreased) participation on procurement by focusing on two well-studied outcomes in the literature: the failure rate and winning bid price. Standard theory on second price auctions with a reservation price predicts that both the winning bid and probability that the auction fails to meet the reservation price should decrease with an increase in the number of bidders. Using data from all procurement auctions, this report tests the prediction by plotting both failure rate and winning bid as a function of the number of participants, as seen in the figures below. 14 The key feature of the figures is that as participation increases, both the failure rate and winning bid decrease. The magnitude of these decreases is quite large, with each additional participant reducing the failure rate approximately 0.5%. This finding should be interpreted cautiously however, as the figures depict an observational study of the cross-section of procurement auctions, and absent an impact evaluation that increases participation, these relationships cannot be interpreted as definitively causal. However, these findings are in line with standard intuition that increased competition in auctions should decrease final prices and failure rates. The relationship between failure rate and the winning bid is very different if we plot the value of a procurement lot as the running variable: 15 The most striking feature of these figures as compared to those on the previous page is the presence of a sharp discontinuity at R$8,000. The above figures show that failure rate increases for lot sizes above the threshold, with a gradual reduction as lot size increases. The decreases on either side of the threshold are in line with the fact that participation in auctions increase with lot size. The relationship between the winning bid and lot size is less intuitive, with the price of the winning bid increasing in lot size before a small jump at R$8,000 and stable prices on the right-hand-side of the cut-off. The discontinuity in both the failure rate and winning bid can be attributed to the discontinuous change in procurement process after this threshold (see figure below). For small purchases under R$8,000, government agencies can forgo a full procurement auction and opt to purchase goods through the noncompetitive RCD/RDL process instead. The RCD/RDL process has the advantage of minimizing the risk of failure as agencies can request tenders from well-known firms, but there is a concern that the noncompetitive nature of these process lead to inefficient prices, as firms do not have an incentive to compete along this dimension. The government has attempted to address this issue through the introduction of the“mini-auction” (CEL - compra electronica), which is discussed in more detail in the next subsection. 16 6.3 EFFECT OF “M INI-AUCTIONS” (CEL) REFORM In order to balance the need for competitive auctions with reducing the administrative burden for small purchases that firms may not be willing to supply given a large cost to entering into procurement, the state government attempted a reform by introducing the “mini-auction” (CEL - compra electronica) format. This new format was introduced with the initial aim to replace the non-competitive RCD/RDL processes for procurements under R$8,000, as allowing greater discretion in procurement has generally led to less favorable outcomes (Palguta and Pertold, 2017). The mini-auctions were structured to have a very low bureaucratic burden – firms do not have to submit the paperwork required for a full auction – and to be quick, with request-to-awarded- contract taking less than 24 hours. Unlike full auctions that are required to be published in newspapers for an extended period of time, CEL procurements publish the tender online and all firms that are registered in the state system are informed about it by email. Once the proposal is made, the tender is run through a live mini auction similar in structure to the fell electronic auction. In this way the government attempted to retain the best aspects of non-competitive procurement – the large reduction in bureaucratic burdens – while maintaining the competitive auction format to ensure the most competitive bids from firms. The state government initially attempted to require the new CEL process for all small-value purchases, but received push-back from many agencies with frequent small purchases, such as 17 hospitals and schools. In response, the government adopted an encouragement strategy to educate agencies on the benefit of the new process. This new strategy was implemented in tandem (albeit unrelated) with the increased use of the max price database to control the cost of products. Below is the share of procurement processes by auction type as a function of changes to the banco de precos database. The vertical dashed line indicates when a max price for a product was entered into the price benchmarking database. We can see that after the introduction of max prices for products, there is increased take-up of items being bought through CEL instead of the non- competitive RDL process, with traditional electronic auctions for larger value lots remaining unchanged. Focusing on the share of procurements utilizing the CEL process, this report observes a significant increase in small-value CEL purchases after the introduction of a max price in the banco de precos system. While this relationship is not causal – the CEL is available to all items, not just those that have a listed max price – it is still useful to use the introduction of max prices as a proxy for CEL usage to analyze the effects (if any) on the outcomes of interest from section 6.2. 18 The figures below plot the outcomes of interest – failure rate and final price – against the days before and after the introduction of a max price in an event study. 19 Interestingly, this report does not find any evidence that the introduction of max prices had a discernable effect on the probability of an auction ending in failure or on the winning bid. This is particularly surprising as the introduction of a reservation price should in theory lead to reductions in the final price at auction. The lack of an effect may be due to a few reasons. First, the max price calculated by the government may not have been an accurate proxy for the price supplied by firms in the private market, and therefore did not induce any change to the structure of the auction. Second, the increase in CEL auctions – while notable – did not completely replace the noncompetitive RCD/RDL processes, leading to little movement in the average failure rate and final bid. Third, since CEL auctions are not compulsory for all small value purchases, there may exist selection bias in which agencies choose to utilize (or not utilize) the CEL auction instead of the noncompetitive RCD/RDL purchases. Unfortunately, these potential explanations cannot be teased out from the observational cross-section of procurement data for which this report has access. A rigorous randomized roll-out of CEL auctions across products and agencies would be required to causally estimate the effect that reducing bureaucratic burdens while promoting competitive auctions would have on small value government purchases. 6.4 CAUSES OF NON-PARTICIPATION The research team conducted a survey of 150 firms in Amazonas state from December 2015 to January 2016 to better understand why firms choose to sell (or not) to the government (the complete survey can be found in the Appendix at the end of this report). The in-depth surveys took place over the phone and asked the firms to discuss the main barriers to participating in government procurement auctions. A summary of the main factors listed by firms can been seen below. 5 5 A summary of the results can be found at: http://dataviz.worldbank.org/views/Pesquisa_Fornecedores/Inicio?%3Aembed=y&%3AshowShareOptions=true&%3Adisplay_count=no#5 20 The survey revealed two main causes for non-participation in government procurement: issues with receiving payment, and bureaucratic hurdles in preparing a bid. In regards to the second issue, suppliers consider preparing a bid for a government contract to be expensive and time- consuming. In addition, the time allowed for bid preparation is insufficient to gather all required documents. For any bidding process, bidders are required to present four sets of documents: legal documents proving the existence of the company, tax documents for all three levels of government, technical documents with the details of the product offered and financial documents showing the financial status of the company. In addition, the survey provides useful examples of the cost of bureaucracy to the firm. For instance, 68 percent of the respondents hired an employee dedicated to filing the information and documents required to participate in government procurement. Indeed, the large number of documents required throughout the procurement process rank among the top 4 difficulties faced by suppliers. 7. CONCLUSION AND DIRECTIONS FOR FUTURE R ESEARCH Using a large administrative dataset on the universe of procurement auctions for the State of Amazonas from 2010-2016, this report has documented several stylized facts on firm participation in public procurement. Despite a large pool of firms that could potentially supply products to the government, participation in procurement auctions is relatively low and decreasing over time. This trend significantly impacts the state’s ability to procure the items needed for government functions at a reasonable price (27% average mark-up relative to private market prices). As this report observes that an increase in the number of firms that participate in a given auction reduces both the probability of failure and the final price for a good, incentivizing firms to enter into public procurement is a first-order concern for the state government. 21 Results from a survey of 150 firms in Amazonas reveal the main barriers to participating in public procurement is issues with receiving payment and bureaucratic hurdles in submitting a bid. As payment for procured items requires the cooperation of several government agencies, any structural reform of the politically-fraught management of public finances – while leading to large potential gains for increased firm participation – would likely take a long time to implement and would not be completely in the control of the procurement office. For shorter term policies that could induce increase participation, the government could focus on decreasing the bureaucratic hurdles faced by firms, thereby lowering transaction costs to participating in public procurement. The Amazonas government has proposed an ambitious program to introduce an eDocuments platform to lower transaction costs for participating firms, which will facilitate the submission of a wide range of documents required for each bidding process. The eDocuments platform allow potential suppliers to submit, sign, and keep track of procurement related documents online, replacing the existing paper-based system. Several documents will be available in their “supplier cadaster” online profile, reducing the cost of submitting documents, and aiding management of legal documents (e.g., document checklists in preparation for biding). Initially proposed to roll out in mid-2016, the reform faced numerous obstacles to implementation, and any future analysis of the completed roll-out may identify a causal effect of lowering transaction costs to participating in public procurement. While this report focused on the number of firms participating in procurement auctions, another potentially salient dimension is the quality of firms that supply to the government. Increased access to the rich tax invoice data from the state combined with federal RAIS dataset on registered firm characteristics could provide another avenue to studying the role of firm quality plays on selection into public procurement. REFERENCES Asker, John & Estelle Cantillon (2008). “Properties of Scoring Auctions”, RAND Journal of Economics, 39, 69-85 Bandiera, Oriana, Andrea Prat & Tommaso Valletti (2009). “Active and Passive Waste in Government Spending: Evidence from a Policy Experiment”, American Economic Review, 99, 1278-1308 Bulow, Jeremy & Paul Klemperer (1996). “Auctions versus Negotiations”, American Economic Review, 86, 180-194 Coey, Dominic, Bradley Larsen & Kane Sweeney (2015). “The Bidder Exclusion Effect”, Mimeo: Stanford University Conley, Timothy G. & Francesco Decarolis (2015). “Detecting Bidder Groups in Collusive Auctions”, American Economic Journal: Applied Economics, forthcoming 22 Coviello, Decio & Mario Mariniello (2014). “Publicity Requirements in Public Procurement”, Journal of Public Economics, 109, 76-100 Coviello, Decio, Andrea Giuglielmo & Giancarlo Spagnolo (2015). “The Effect of Discretion on Procurement Performance”, Mimeo: HEC Montreal Ferraz, Claudio, Federico Finan & Dimitri Szerman (2015). “Procuring Firm Growth: The Effects of Government Purchases on Firm Dynamics”, Mimeo: PUC-Rio Gerardino, Maria Paula, Stephan Litschig & Dina Pomeranz (2015). Can Audits Backfire? Evidence from Public Procurement in Chile, Mimeo. Lewis-Faupel, Sean, Yusuf Neggers, Benjamin A. Olken, & Rohini Pande (2016). “Can Electronic Procurement Improve Infrastructure Provision? Evidence From Public Works in India and Indonesia”, American Economic Journal: Economic Policy, forthcoming Lichand, Guilherme, Marcos F M Lopes, Marcelo C Medeiros (2006). “Is Corruption Good for Your Health?” Working Paper, Harvard University. Litschig, Stephan and Yves Zamboni (2011). “Audit Risk and rent Extraction: Evidence from a Randomized Evaluation in Brazil”, Working Paper, Pompeu Fabra Hendricks, Ken & Robert H. Porter (2007). “An Empirical Perspective on Auctions”, Chapter 32 in M. Armstrong and R. Porter (eds), Handbook of Industrial Organization Volume 3, North Holland International Monetary Fund (2016). “Public financial Management in Latin America – the key to efficiency and transparency”. Kang, Karam & Robert A. Miller (2015). “Winning By Default: Why is There So Little Competition in Government Procurement?”, Mimeo: Carnegie Mellon University Kawai, Kei & Jun Nakabayashi (2015). “Detecting Large-Scale Collusion in Procurement Auctions”, Mimeo: UC Berkeley Krasnokutskaya, Elena & Katja Seim (2011). “Bid Preference Programs and Participation in Highway Procurement Auctions”, American Economic Review, 101, 2653-2686. Olken, Ben (2007). “Monitoring Corruption: Evidence from a Field Experiment in Indonesia”, Journal of Political Economy, 115, 200-249. Palguta, Ján, and Filip Pertold (2017). "Manipulation of Procurement Contracts: Evidence from the Introduction of Discretionary Thresholds." American Economic Journal: Economic Policy, 9 (2): 293-315. 23 Tran, Anh (2008). “Can Procurement Auctions Reduce Corruption? Evidence from the Internal Records of a Bribe-Paying Firm”, Mimeo: Harvard University World Bank (2009). “Enterprise Surveys, Brazil: Country Profile 2009”, technical report, available at http://www.enterprisesurveys.org/~/media/GIAWB/EnterpriseSurveys/Documents/Profiles/Engli sh/Brazil-2009.pdf accessed on 21 Jan 2016 24 APPENDIX – SUPPLIER S URVEY Instructions to enumerator: At the start of the interview, the enumerator will state clearly whether the questions are about the municipal government or about the state government, and will adjust the first statement and Section III accordingly. Instructions to enumerator: Please, read: “The objective of this questionnaire is to understand public procurement in [Manaus city, Amazonas state]” PART I: PARTICIPATION IN PROCUREMENT PROCESS Has your firm ever participated in a public Yes, No procurement process (local, state or federal)? If NO, please jump to Part II and read to associated instructions. Which levels of government has your firm ever Federal government, Amazonas state, Other supplied for? (Circle all that apply) state:____________, Manaus municipality For which level of government does your firm Federal government, Amazonas state, Other most often supply? (Circle all that apply) state:____________, Manaus municipality, N/A When did your firm last participate in a public Current year, last year, 2 years ago, more than 2 procurement process?(Circle one) years ago (specify year):___________, N/A What percentage of firm revenues in the last 12 100%, 99-75%, 74-50%, 49-25%, 25%-1%, 0% months came from public procurements? (Circle one) Which procurement processes has your firm In-Person auction, Online auction, Direct ever participated in? (Circle all that apply) Purchases (<8k Rs), Framework agreement, Other:_______________ Which procurement process does your firm most In-Person auction, Online auction, Direct often use? (Circle one) Purchases (<8k Rs), Framework agreement, Other:_______________ Does your firm currently use the services of a Yes, No, Not currently but have in the past consultant to help navigate the government procurement process? Does your firm currently have staff in the office Yes, No, Not currently but have in the past that is specialized in government procurement? Does your firm currently sell some Yes, No, Not currently but have in the past goods/services to another company so that they could sell them on to the government? 25 Does your firm currently buy some final Yes, No, Not currently but have in the past goods/services (i.e. not inputs of production) from another company in order to sell them to the government? Does your firm currently sometimes subcontract Yes, No, Not currently but have in the past the whole or part of a government contract to another firm? PART II: ISSUES IN THE PROCUREMENT PROCESS PARTICIPATION Instructions to enumerator: Please read the opening script to the respondent. Then ask them to write down the 5 options. Then ask them to think about which ones are most important to them and to tell you what their top 3 options are. If the respondent uses the “Other” option, write down what they say. Then please ask them to double-check that their answer doesn’t fit into any of the pre-specified options, but don’t pressure them to fit it into the pre-specified options. Many things make participating in supplying goods and services to the government challenging. We’d like to ask some questions about what in particular makes it difficult for your firm. While answering the questions we’d like to know about the representative experience that your firm has had in dealing with the [STATE/MUNICIPAL] government. [If the respondent answered “NO” to question 1, read the following passage instead and continue with the rest of the survey: Many things make participating in supplying goods and services to the government challenging. Since you have not participated in a procurement process with the government, we’d like to ask some questions about what in particular makes your firm decide to not participate in govrnment procurement. Even though your firm hasn’t previously sold to the [MUNICIPAL/STATE] government, we’d like to hear the impressions of what it would be 26 like if your firm did. While answering the questions we’d like to know about the representative experience that your firm would have in dealing with the [STATE/MUNICIPAL] government.] Please rank the top 3 options from those listed below that are the most important in creating challenges for your firm in supplying to the [STATE/MUNICIPAL] government. Rank the most challenging option to your firm as 1, the second-most important challenge as 2, and the third most important challenge as 3. An open-ended option is also available if there is another challenge that faces your firm that is not included in the list. If applicable, please specify this challenge and then rank it (from 1-3) along with the other options. Rank Information on participating in a procurement process is hard to find and/or unclear It is difficult to place a competitive bid There is a lot of paperwork involved There are issues in the government making award decisions and/or payments Successfully fulfilling a winning contract is difficult Other (please specify): Instructions to enumerator: For each of the 3 options that the respondent chose in the previous question, now please ask the corresponding detailed question. Ask the questions in the order that the respondent ranked them, starting with the option that they chose as most important. Option 1: Finding Information on Procurement Now we’d like to ask about the representative experience of your firm in getting and understanding information provided by the government on participating in public procurement. 27 Please choose among this list which are the top 3 barriers that are the most challenging for your firm in finding information about the procurement process. If your firm experienced a top 3 barriers that is not mentioned in this list, you can also name a different barrier. Now, rank the top 3 barriers for your firm from 1 to 3, with 1 being the biggest obstacle, 2 the second and 3 the third. Rank Finding opportunities of government purchases that the firm can bid on is difficult Documents and rules that the government provides on participating and submitting a bid are hard to understand and requires the help of a specialist It is hard to obtain information because some government officials demand informal payments in order to provide information or documents that are needed to participate in procurement The amount of time between publishing the information of an upcoming procurement and the actual bidding is too short The documents provided by the government are unclear about how bids will be evaluated Government officials don’t have the training and knowledge required to respond to queries Other (please specify): Option 2: Competing for Contracts Now we’d like to ask about the representative experience of your firm in competing to receive government contracts. Please choose among this list which are the top 3 barriers that are the most challenging for your firm in competing to bid and win [STATE/MUNICIPAL] government contracts. If your firm experienced a top 3 barriers that is not mentioned in this list, you can also name a different barrier. Now, rank the top 3 barriers for your firm from 1 to 3, with 1 being the biggest obstacle, 2 the second and 3 the third. Rank It is hard to comply with the technical specifications of the bid 28 The government restricts participation in the auctions via red-tape (e.g. registration, pre- qualification, etc.) Technical specifications are made to unfairly favor certain firms The government doesn’t provide enough time to prepare all the necessary documents Some businesses make side payments to government officials in return for contracts Having price as the only determinant in choosing a winning bid makes it difficult to provide a high quality good competitively Other (please specify): Option 3: Paperwork involved Now we’d like to ask about the representative experience of your firm in filling out the paperwork involved in the public procurement process. Please choose among this list which are the top 3 barriers that are the most challenging for your firm in filling out procurement documentation. If your firm experienced a top 3 barriers that is not mentioned in this list, you can also name a different barrier. Now, rank the top 3 barriers for your firm from 1 to 3, with 1 being the biggest obstacle, 2 the second and 3 the third. Rank Submitting bids requires many government-issued documents and licenses A lot of time is needed to manually fill out all of the documents Claiming payment for the goods/services requires many documents Amending contract terms after award of a contract requires many documents Government officials demand informal payments in order to approve/supply documents Confirming delivery of the product requires many documents Other (please specify): 29 Option 4: Government Award Decisions and Payments Now we’d like to ask about the representative experience of your firm in waiting for the award decisions and payment from the government for a procurement process. Please choose among this list which are the top 3 barriers that are the most challenging for your firm in obtaining award decisions and payments from the [STATE/MUNICIPAL] government for a procurement process. If your firm experienced a top 3 barriers that is not mentioned in this list, you can also name a different barrier. Now, rank the top 3 barriers for your firm from 1 to 3, with 1 being the biggest obstacle, 2 the second and 3 the third. Rank The time between submitting bids and the government making an award decision is very long Government officials require informal payments to make decisions or process payments The time between contract fulfillment and payment is very long The government payment system is not transparent The government does not pay the full amount of the contract The government cannot pay for any amount of the contract Other (please specify): Option 5: Fulfilling a Contract Now we’d like to ask about the representative experience of your firm in being able to successfully fulfill the terms of a winning procurement contract. Please choose among this list which are the top 3 barriers that are the most challenging for your firm in renegotiating a procurement contract with the [STATE/MUNICIPAL] government. If your firm experienced a top 3 barriers that is not mentioned in this list, you can also name a different barrier. 30 Now, rank the top 3 barriers for your firm from 1 to 3, with 1 being the biggest obstacle, 2 the second and 3 the third. Rank It is difficult to change the final quantity of the products that your firm is able to supply It is difficult to change the final price at which your firm supplies products It is difficult to change the quality of products that the government requires your firm to supply It is difficult to change the delivery schedule of the contract It is difficult to change the payment schedule of the contract Penalties for not fulfilling the entire contract are very high Other (please specify): PART III: COMPARING STATE AND MUNICIPALITY EXPERIENCES So far we have asked about the representative experience of your firm in dealing with the [STATE/MUNICIPAL] government. Now we’d like to compare the experiences with [STATE/MUNICIPAL] government to the [MUNICIPAL/STATE] government. Even if your firm hasn’t previously sold to the [MUNICIPAL/STATE] government, we’d like to hear the impressions of what it would be like if your firm did. Please compare the state and municipal government according to the following criteria. An open-ended option is also available if there is another challenge that faces your firm that is not included in the list. 31 Comparison Information on participating in a State is worse Both the same Municipality is procurement process is hard to find and/or worse unclear Placing a competitive bid is difficult State is worse Both the same Municipality is worse There is a lot of paperwork involved State is worse Both the same Municipality is worse The government is slow to make award State is worse Both the same Municipality is decisions and/or payments worse Fulfilling a winning contract is difficult State is worse Both the same Municipality is worse Other (please specify): State is worse Both the same Municipality is worse PART IV: GENERAL FIRM INFORMATION Name of Firm Address Street: City: State: Postal Code: Contact Information Phone number: Email Address: CNPJ Number Number of Full-Time Employees Number of establishments Main Business Area (circle one) Agriculture, Mining, Manufacturing, Construction, Transportation, Consulting, Services, Wholesale, Retail, Other:_________________ 32