Dutch Construction Cartel: As Bad As It Seems?
Elliott Terpstra 11913576
Supervisors: Prof. Dr. M.P Schinkel & L.M. Treuren 15th July 2018
Acknowledgements
For this thesis I would like to thank Ad Bos for his time and help in researching the construction cartel and the Koop Tjuchem shadow bookkeeping.
Verklaring eigen werk
Hierbij verklaar ik, Elliott Terpstra dat ik deze scriptie zelf geschreven heb en dat ik de volledige verantwoordelijkheid op me neem voor de inhoud ervan.
Ik bevestig dat de tekst en het werk dat in deze scriptie gepresenteerd wordt origineel is en dat ik geen gebruik heb gemaakt van andere bronnen dan die welke in de tekst en in de referenties worden genoemd.
De Faculteit Economie en Bedrijfskunde is alleen verantwoordelijk voor de begeleiding tot het inleveren van de scriptie, niet voor de inhoud.
Table of contents
1. Introduction 5
2. Dutch Construction Market 7
2.1 Market Structure 7 2.1.1 Size 7 2.1.2 Firms 8 2.1.3 Demand 13 2.2 Legal Background 17 2.3 Conclusion 18 3. Pre-auction mechanism 18
3.1 Koop Tjuchem Shadow Accounts Introductory Data 18
3.2 Shadow Bookkeeping Description 24
3.2.1 Example Sheet 27
3.3 System of Pre-Auctions 28
3.3.1 Step 1: Arranging the Pre-Auction Meeting 28 3.3.2 Step 2: Determining Pre-Auction Tender Price 31 3.3.3 Step 3: Determining Target-Auction Bid 35 3.3.4 Step 4: Determining the Pre-Auction Winner 37
3.3.5 Step 5: Allocation of Cartel Profits 40
3.3.6 Step 6: Target-Auction 42
3.3.7 Step 7: Settling of Cartel Profits 43
3.4 Conclusion 45
4. Koop Tjuchem Shadow Account Patterns 46
4.1.1. Pre-Auction Bidding Behaviour 46
4.1.2 Cartel Profits Over time 51
4.1.3 Per Firm Wins Over Time 54
4.2 Conclusion 55
5. Defense Argumentation 57
5.1.1. Rotation Mechanism Defense Argument 57
5.1.2. Bid Rotation Efficiency 59
5.2 Auction Criteria Defense Argument 60
5.3 Monopoly Money 63
5.5 Peddling Fright 64
5.6 Combination Works 65
5.7 Conclusion 66
6. Existing estimation 67
6.1 Pheijffer Parliamentary Enquiry 67
7. Conclusion 70
Literature 72
1. Introduction
In the beginning of the 21st century, in 2001, the people of the Netherlands were notified of a large collusionary practice that had been active for decades and had an impact on all of their lives. Harrington (2018) defines collusion as when some or all firms in a market coordinate to restrain competition with the intent of raising price and earning a higher profit. In this case within the Dutch construction sector, a large number of firms rigged the bids that they placed in procurement auctions held by their prospective customers. Both auctions in the public- as well as the private sector were rigged. The cartel primarily formed in the GWW (Ground-, Water- and Road Construction) sector. Given the public nature of works in this sector, governmental institutions function as the main auctioneers. Governmental institutions accounted for 49% of all GWW construction orders in 2000.1 Any negative effects of the cartel weighed strongly on taxpayers’ money.
The cartel came to light when whistle-blower Ad Bos turned in a bookkeeping secretly kept at the firm Koop Tjuchem, where he was active as a director, to the Dutch investigative journalism television programme ZEMBLA. This bookkeeping contained records of thousands of secret meetings among construction firms held before the actual auction, or target auction. Given that these records showed the extensiveness of the cartel over large as well as small construction works in the Netherlands, the Dutch Parliament was quick to set up an investigation into what had happened in the Dutch construction sector and specifically in the Dutch GWW sector by enacting a Parliamentary Enquiry Committee.
Over the course of 10 months representatives of the Dutch Parliament undertook an investigation into the workings of the Dutch construction sector and spoke to 67 representatives from the Dutch construction sector as well as representatives from institutional governmental agencies. In December 2002 the Parliamentary Enquiry presented their findings of the irregularities in the GWW sector and their recommendations to the Dutch government for future regulation. The Committee concluded that the irregularities in the Dutch
construction sector were widespread and diluted over many different types of construction- works and –firms. Furthermore, the committee concluded that clients of the construction firms had been negatively affected by the cartel with an estimated overcharge of 8.8%. Naturally, the overcharge a cartel imposes on a market is dependent on various different characteristics pertaining to the market where the cartel is operating in and the specifics and number of firms that form a cartel. However, given the current literature on cartel overcharge estimation, 8.8% seems rather conservative. A study conducted by Boyer and Kotchoni (2015) finds a median overcharge estimate of 15.47% - 16.01% over multiple sectors, while Conner and Lande (2008) found estimated overcharges in the range of 31% - 49% over various industries.
Moreover, the hearings of the Parliamentary Enquiry held with representatives of the Dutch construction sector provided multiple economic arguments on behalf of the construction sector arguing why the Dutch construction sector was not cartelised and whether this supposed cartelisation was as economically detrimental to society as it seemed. In the Final Report of the Parliamentary Enquiry these arguments are only briefly discussed before being eventually discarded by the Committee. Arguments presented by the defense include the rotation mechanism of the cartel, the auction criteria set by the auctioneer, the participants in private- versus public auctions, peddling fright, combination works and so-called ‘monopoly money’. The latter argument claims that the money involved in the system of pre-auctions did not represent any value in reality.
The aim of this thesis is to further investigate the economics behind the arguments presented by the Defense of the construction cartel and to gain a deeper understanding of the overcharge as estimated in the Final Report of the Parliamentary Enquiry. In order to conduct my research I have digitalised all available secret accounts from the Koop Tjuchem shadow bookkeeping from 1986 - 1998 on 4114 reports from pre-auction meetings. This data includes all public- and private auctions in which Koop Tjuchem took part during this period and the corresponding records that they held with 373 unique firms and subsidiaries. In addition, I have held multiple conversations with the cartel whistle-blower Ad Bos to clarify the inner workings of the cartel as well as the
contents of the shadow bookkeeping. The report of these conversations is featured in the appendix.
The thesis is structured as follows. First, I will discuss the Dutch construction sector and the historical background that directly led to the cartelisation of the industry. Second, I will enter into the system of pre-auctions particular to this cartel. Third, I will investigate the bidding behaviour of the construction firms. Given the nature of the cartel and the shadow bookkeeping of Koop Tjuchem, the cartel that I will research will focus on the GWW sector of the construction market. Fourth, I will research the arguments presented by the defense in the Parliamentary Enquiry. To conclude, I will discuss the existing overcharge estimation of the Dutch construction cartel before presenting my own findings.
2. Dutch Construction Market
2.1 Market Structure
2.1.1 Size
At the time of the Dutch construction cartel in the late 1990’s, the construction market in the Netherlands was one of the most important sectors in the Netherlands in terms of employment and economic activity. In the year 2000 the sector employed 400 thousand employees and accounted for 10% of Dutch GDP2. The construction sector is divided into multiple sub-sectors. These sub-sectors include the B&U (Residential- and Utility construction) sector and the GWW (Ground-, Water- and Road construction) sector.
The B&U sector consists out of the construction, maintenance and renovation of buildings, while the GWW sector focuses on infrastructural works. Table 1 provides an overview of the composition of the Dutch construction sector in 2000.
2 CBS (2002)
Table 1. Construction Production 2000 € in M. € in M. % Of Total Residential Construction 9.332 - New construction 5.343 - Reconstruction Total Residential 14.675 33.3% Utility construction - New construction 7.853 - Reconstruction 3.339 - Maintenance 7.479 Total Utility 18.671 42.4%
Ground-, Water-, and Road Construction
- New Construction 6.904
- Maintenance 3.824
Total Ground-, Water- and Road
Construction 10.728 24.3%
Total 44.074 100%
Source: EIB, CBS (2002)
At the time of the cartel the GWW sector was smaller than the B&U sector. In 2000 the GWW sector comprised of 24.3% of the total construction sector amounting to €10.7 billion. Historically, the growth rate of the total Dutch construction sector has been on the level of Dutch GDP growth. Only in 2000 did the Dutch construction sector lag behind GDP growth, with 2.7% construction production growth versus 3.5% GDP growth. During the final years of the cartel period the GWW sector has shown a higher growth rate than the B&U sector. In 2000 the growth in the GWW sector (9.6%) far outperformed the growth in the B&U sector (2.5%).
2.1.2 Firms
Total production in the Dutch construction market in 2000 was in majority controlled by a large number of firms that differed in size and production activity. Despite merger- and acquisition activity, the number of firms in the Dutch construction sector did not fall, but increased in the cartel period. Within the period 1995 – 2000 the number of firms with employees in the construction sector grew from 19 thousand to 20.5 thousand firms. More than 93% of these
firms consisted out of 5 or fewer employees. The growth in construction firms was mainly attributable to firms with 5 or less employees. The consistency of the firms employing more than a thousand employees, demonstrates the high barriers to entry in the market for large construction works. In 2000 only 1 firm grew sufficiently enough to become larger than a thousand employees. Five firms were able to grow into the category 501 – 1000 employees in that year. Figure 1 displays the number of firms active in the construction sector as classified by employee size.
Figure 1. Number of Construction Firms per Employee Size, 2000
Source: EIB, CBS (2002)
The table shows that only a small number of firms, 6, employed more than a thousand employees. These firms were BAM, Ballast Nedam, HBG, Heijmans, Koop Tjuchem and KWS. Only these firms were able to take on large-scale projects, given the production capacity with which smaller firms were faced due to their size.
The distribution in the size of firms becomes narrower when zooming in on the GWW sector. Figure 2 presents the number of firms active in the B&U and GWW sector per employee classification.
11943 3222 2562 1952 520 200 79 19 6 0 2000 4000 6000 8000 10000 12000 14000 1 - 5 6 10 11 20 21 50 51 100 101 -200 201 -500 501 -1000 > 1000 N um b er of F ir m s Employee Size
Figure 2. Number of Construction Firms per Sector, 2000
Source: EIB, CBS (2002)
Figure 2 displays that 5% of the firms in the GWW sector employ over a thousand employees. Within the same sector 50% of the firms employ between 1 – 5 employees. As mentioned earlier, these smaller firms are not able to take on large construction works. In effect, 50% of the construction firms in the GWW sector do not posses an effective competitive threat on large-scale construction works to the firms with 100 or more employees. This reduces the competitive market where the larger firms operate in and sustains potential coordination among these larger firms. The small concentration of larger firms can potentially serve as a feeding ground for collusion.
The same picture prevails when I look at the size of Dutch construction firms in terms of revenue size, 5% of the firms in the GWW sector attain revenue larger than €22 million per year. This revenue amount is obtained by 3% of the firms in the B&U sector. Larger firms in the GWW sector are more dominant in terms of revenue in the entire market than in the B&U sector.
At first glance, based on these statistics, the concentration of revenue and employee size does not seem to be disproportionate in both sub-sectors of the industry. Especially, when I compare them to international data of the concentration of construction markets. In 2009 the market share of the top 4
9553 2552 1979 1494 402 232 1330 435 381 342 95 76 1060 235 202 116 23 18 0 2000 4000 6000 8000 10000 12000 1 - 5 6 - 10 11 - 20 21 - 50 51 - 100 >100 N um b er of F ir m s Employee Size B&U GWW Other
international contractors was 27.4%.3 The top 4 largest construction firms in the Netherlands in 2000 had a combined market share of 16.6%.
Looking more closely into the ten largest construction firms in the Netherlands in 2000, the 10 largest firms represented €13 billion in revenues. This amounted to 30% of the total construction market in the Netherlands. In addition to the 6 largest firms by employee size, these 10 firms include Dura Vermeer, Strukton, TBI and van Wijnen. Table 2 presents the revenues of these firms as a total of the 10 largest construction firms in 2000 as well as the total Dutch construction sector in the Netherlands at that time.
Table 2. Top 10 Largest Firms by Revenue in 2000
Firm Revenue € in M. % Of Top 10 % Of Total
Ballast Nedam 881 6.6% 2.0% BAM 1.510 11.3% 3.4% Dura Vermeer 782 5.8% 1.8% HBG 1.511 11.3% 3.4% Heijmans 2.130 15.9% 4.8% Koop Tjuchem 805 6.0% 1.8% KWS 2.200 16.4% 5.0% Strukton 1.157 8.6% 2.6% TBI 1.474 11.0% 3.3% Van Wijnen 969 7.2% 2.2% Total 13.419 100.0% 30.4%
Source: Annual Reports (2000)
Standalone none of these firms represent a significant share of the construction market. However, together the firms possess a notable market share of 30.4%. A widely accepted measure to capture the extent to which a market is concentrated is the HHI-index (Herfindahl-Hirschmann Index). The HHI has been used extensively to assess a market’s competitiveness.4 Based on the market share of all firms within an industry, it measures the level of concentration from 0 to 1 by using the formula
HHI = ∑ 𝑠𝑁𝑖 𝑖2, (1)
3 Ye, Lu & Jiang (2009) 4 Stigler (1968)
where 𝑁 is the number of firms in the market and 𝑠 represents the market share of firm 𝑖 in the market. Using the information presented in table 2, I can construct the HHI-index for the sub-section of the 10 largest firms in the GWW sector in the Netherlands for the year 2000. The HHI-index for this sub-section was 0.113 in 2000. Given this number, the HHI-index does not seem to give an indication of a concentrated top 10 section of the market. Generally, only a market with an HHI-index above 0.25 is perceived as highly concentrated.5
However, here it is important to note that the sub-sectors within the construction sector consist out of different sections within the sub-sectors B&U and GWW. This makes it relatively simple for a firm to not appear dominant on the larger scale of the B&U or GWW sector, while in fact being dominant within a specific sub-section. For instance, throughout its existence Koop Tjuchem achieved a dominant position in construction works concerning asphalt- and asphalt related inputs to the point that other top 10 GWW firms were forced to cooperate with Koop Tjuchem in order to be able to produce asphalt related works.6 Furthermore, it is important to note that the market share of large firms within both sub-sectors had grown throughout 1996 – 2000. In 1996 the total revenue of the 10 largest firms amounted to €6.7 billion, compared to 13.4 billion in 2000.
The dominance of large firms in the GWW sector is exemplified via statements made by whistle-blower Ad Bos. Bos confirmed that smaller firms were dependent on the 10 largest firms. According to Bos this dependent relationship became clear by means of the inputs needed to produce certain construction works. Producing construction works in the GWW sector requires a complex set of production inputs. Given the complexity of combining these inputs and the large fixed costs associated with attaining this production capital, only large firms were able to attain these inputs and were able to attain them against a lower cost. This raised entry barriers into the GWW market to such a level, that it was virtually impossible to enter the market without the consent of one or more of the top 10 largest firms.
5 US Horizontal Merger Guidelines, 2010 6 Ad Bos, 2018
For instance, when Koop Tjuchem started as a small firm in the north of the Netherlands, in the beginning of the 1980’s, it tendered for a road construction work. Koop Tjuchem won the tender process. The firm had to be able to produce a sufficient amount of asphalt to complete the construction of the road. However, at that time Koop Tjuchem did not posses the production capacity to produce enough asphalt. Despite efforts of hiring a sub-contracter to produce the amount of asphalt that Koop Tjuchem came short, no firm was willing to work for Koop Tjuchem. The firm decided to ship in the needed production capacity from Germany in order to complete the work. The situation speaks to the character of the Dutch GWW sector. It was made difficult for an outsider, as Koop Tjuchem was at the time, to produce. In this analogy it was perceived by representatives of Koop Tjuchem that it was not the case that other Dutch construction firms were unable to assist Koop Tjuchem, but that other construction firms simply refused to work for them. This dependent relationship of small firms on large firms carried on throughout the years, as well as during the years of the data of the Koop Tjuchem shadow bookkeeping.
2.1.3 Demand
In the GWW sector the Dutch government is the largest component of demand for construction works. In 2000, the Dutch government and its related governmental institutions accounted for 49% of all GWW construction works. Other clients of GWW construction works were private- firms and individuals, other construction firms and a category of other diverse clients. Construction firms could also function as a client when a construction work is produced via a so-called ‘combination work’, or produced in commission of another construction firm. This will be discussed to more extent in section X. Figure 3 presents the distinction in clients for construction firms for the GWW sector in percentages for the year 2000.
Figure 3. Client Type GWW Sector, 2000
Source: EIB (2001)
The Dutch government commissioned nearly half of all GWW works in 2000. A little over twice as much as private firms commissioned in that same year.
Part of the dependent relationship discussed in section 2.1.2 is also exemplified by figure 3. Other construction firms accounted for 16% of total demand in 2000. Thereby, the demand of other construction firms was higher that the demand of private individuals and other non-classified institutions combined. It is important to note that construction firms that function as a client towards other construction firms have also attained the majority of these projects via government issued construction orders. Therefore, the data presented in figure 3 is negatively biased towards the percentage of works commissioned by the government. In reality, this number will have been higher.
The percentage of works commissioned by the government has left its mark on the GWW sector. Under EU legislation the Dutch government is obligated to appoint a construction work via a tender procedure based on the size of the work. Given the share of demand and GWW revenue from governmental institutions, construction firms are forced to cooperate in the system of governmental public- and private tenders in order to be able to operate in the GWW sector. Public tenders are open for all construction firms to submit a bid. On the other hand, only pre-selected construction firms were able
49% 23% 16% 4% 8% 0% 10% 20% 30% 40% 50% 60%
Government Private Firms Construction
Firms IndividualsPrivate Other
Cl ie n t P er ce n ta ge Client Type
to submit bids in private tenders. Given from the Koop Tjuchem data, in private tenders the auctioneer selected up to 25 firms to take part. In public tenders the number of firms who would submit bids could reach as far as 117 firms. Roadworks commissioned by the federal government of the Netherlands were exclusively appointed via tenders in 2000. The percentage of roadwork tenders for lower levels of government had increased from 73% in 1989 to 84% in 1999. Table 3 presents the channels of procurement in the GWW sector in 2000.
Table 3. Channel of Orders by revenue, 2000
Channel of Orders € in M. % Of Total
Public Tender 6.233 58% Private Tender 2.264 21% Client Work 676 6% Other 1.555 14% Total 10.728 100% Source: EIB (2001)
Public- and private tenders were especially important in the GWW sector. In 2000 47% of the revenue in the B&U sector was appointed via tenders. While, in the GWW sector 79% of all revenue was ordered via tenders. This underlines the importance for firms in the GWW sector to take part in tenders, both public- and private. The number of private tenders that are commissioned by auctioneers is also important to note. In 2000 a little over a fifth of the revenue was ordered via private tenders. Given the smaller amount of firms that are able to attend private tenders versus public tenders, private tenders potentially facilitate collusion more easily than public tenders. Auriol (2006) shows this by studying price captures and extortions in public sector tenders. He finds that, independent of country characteristics, to minimize dead-weight loss in society large public sector purchases should be commissioned by public tenders. Auriol shows that public tenders are not necessarily beneficial in terms of societal welfare in small tenders. In the Dutch GWW market large construction works were also commissioned via private tenders. The Koop Tjuchem shadow accounts show that in the period 1986 – 1998 11,8% of the private tenders involved construction works where the winning bid was higher than 10 million Dutch Guilders. The average number of firms attending these private auctions was 8.
Even though the government was obligated to appoint the works that they needed via a tendering process, it was not obligated to do so via a public tender. When the government did not deem it necessary to do so, it could organise private tenders or design the criteria of the public tender in such a way that only firms that they would otherwise invite for their private tender, were able to enrol in the public tender. The requirements that were set by the government were set in such a way that in many instances only large firms were able to sign up. This is exemplified by the shadow bookkeeping of Koop Tjuchem, where a large firm such as KWS attended over 500 private auctions. In the same time span a small firm such as Ouwejan only participated in 19 private auctions.
Considering the magnitude of works commissioned by higher levels of government, favouritism of large corporations over smaller firms is understandable up to a certain point. Moreover, lower levels of government tended to favour construction firms from their own regions. Combined, these preferences by governmental institutions of various sizes led to skewed private governmental tenders and even skewed public governmental tenders. The Final Report of the Dutch Parliamentary enquiry concluded that ‘apparently Dutch governmental institutions favoured personal bonds with construction firms over Dutch and European legislation’.
Moreover, at the time there had been a shift in the Dutch governmental procurement strategy to divert more of the risk involved in construction from the client itself (the government) to the contracter. The implementation of this diversification of risk had negative impacts on the transparency of the tender price of the construction firms. It became more difficult for clients to assess the level to which the tender price was sufficient for the contracter to produce to work and account for potential extra costs for which the contracter was responsible. Firms became increasingly responsible for the potential extra work associated with project itself. On the one hand this increased the potential for bankruptcy, especially for smaller firms (causing them not to be invited to the auction). On the other hand, it still left the government with a fair share of the risk of potential extra work. As in the auction the focus lies on the tendered price for the work at hand, not on the costs of potential extra construction, the firms that are able to tender with the lowest price need not necessarily be the cheapest
firm in the long run. The risk of potential extra work was already existent before this procurement strategy shift came in effect. However, the Final Report of the Parliamentary Enquiry concluded that the diversion of risk to the construction firms was not sufficient to offset the burden of costs of potential extra work on the government.
2.2 Legal Background
One of the most interesting characteristics of the Dutch construction cartel is its history before 1992. From here I will name the period before February 1992 the legal period and the period hereafter the illegal period.
The reason to do this is because the Dutch construction cartel actually finds its roots at in the Dutch government. In 1963 the SPO, the foundation of collaborating pricing organisation (in Dutch: Samenwerkende Prijsorganisaties) was founded. The purpose of this organisation was to minimise costs of calculating cost prices for works that a firm eventually would not win. Through this organisation, firms could share (expected) cost information and thus share costs and minimise the risk of miscalculations. With the latter being an aspect that would also benefit the client.
This foundation continued for an extended period of time until the European Commission started an investigation and eventually forced the Dutch government to shut it down in 1992 due to stricter cartel legislation. At this time even the Dutch prime minister took the side of the Dutch construction firms in talks with the EU.7 However, the Dutch government could not prevent the legislation from entering into effect in February 1992. From that point the SPO was illegal and it was no longer allowed for construction firms to meet ahead of target-auctions to discuss cost prices.
However, these meeting still proceeded, as the Koop Tjuchem shadow accounts shows us. From that point on the meeting were illegal and the cartel was in effect. It is important to stress that already in the legal period, the firms in the pre-auction arranged more than what was allowed in the SPO. In essence, the Dutch construction cartel began earlier than 1992, but a precise start date is
difficult to pinpoint. In this thesis I will take 1986 as a starting point, as this is the starting point of my data.
2.3 Conclusion
In this chapter I have shown the competitive setting of the Dutch construction market at the time of the revelation of the construction cartel. I have shown that the Dutch GWW sector is concentrated not only in terms of market share of the 10 largest firms, but also in the control of production inputs.
Furthermore, demand in the GWW sector consists mainly out of government driven demand. Given EU tender legislation it followed that the vast majority (79%) of the works that were commissioned in the GWW sector, were ordered via public- and private tenders.
Both the market concentration of the GWW sector as well as the one-sided government demand and wide use of private tenders have the potential to facilitate coordination among firms and stimulate collusion.
3. Pre-auction mechanism
The shadow accounts of Koop Tjuchem together with testimonies for the Parliamentary Enquiry and conversations with whistle-blower Ad Bos reveal an ingenious pre-auction mechanism. It was used to decide which firm was allowed to tender the lowest in the target auction and by what amount the initial price determined in the pre auction could be raised. In this section I will discuss the inner working of the system, complemented with data from the Koop Tjuchem shadow accounts.
3.1 Koop Tjuchem Shadow Accounts Introductory Data
Before I dive deeper into the workings of the pre-auction mechanism, I will discuss the shadow bookkeeping of Koop Tjuchem and how to interpret it. To begin I will present introductory data of the shadow accounts that I posses to increase the understanding of the foundations of the cartel. The data that I
posses from the shadow accounts contains information from the pre-auctions from 1986 – 1998. It includes 4114 reports of pre-auction meetings and the implications thereof. Considering that the first data stems from 1986, the data until April 1992 was gathered in a setting where the system of pre-auction was condoned by the Dutch government and not illegal. This subset consists out of 550 pre auction meetings.
The shadow bookkeeping was updated for individually for each firm Koop Tjuchem did business. For each work a representative of Koop Tjuchem updated the shadow bookkeeping separately for each firm that attended the target auction. In total 373 unique firms and subsidiaries are mentioned in the shadow bookkeeping and thus the bookkeeping contains equally as many separate bookkeepings. Corrected for double entries across the multiple sub-bookkeepings (when multiple firms attended the same target auction) the Koop Tjuchem shadow bookkeeping contains 2,499 individual entries of construction works from 1986 – 1998. Descriptive statistics of the data at hand are presented in table 4 and table 5. Table 4 presents the number of times the 10 largest firms, as of 2000, are counted for as a separate record in the Koop Tjuchem shadow bookkeeping. This includes all daughter companies of each firm, including Koop Tjuchem. Interestingly, HBG, Strukton and TBI account for only a few mentions. Van Wijnen does not appear in the secret accounts at all. The fact that Strukton, TBI and Van Wijnen are not mentioned often is not surprising considering the nature of works where these firms were most active in, in comparison to Koop Tjuchem. Strukton was mainly active in railroad construction while TBI and Van Wijnen were most active in the B&U sector of construction.8
Table 4. Top 10 Mentions in Shadow Bookkeeping
Firm Bookkeeping Accounts
Ballast Nedam 220 BAM 129 Dura Vermeer 241 HBG 8 Heijmans 134 Koop 4114 KWS 489 Strukton 11 TBI 1 Van Wijnen 0
Source: Koop Tjuchem Shadow Accounts
Table 5 presents a data overview of all separate construction works divided between the legal period (1986 – 1992) of the pre-auctions and the period thereafter.
Table 5. Data Description Shadow Accounts
1986 - 1992 % Of Period 1992 - 1998 % Of Period
Auction
- Number of private auctions 88 18,6% 804 39,7% - Number of public auctions 376 79,5% 1140 56,3%
- Unknown 9 1,9% 82 4,0%
Total number of auctions 473 2026
Payment - Work credit 23 4,9% 530 26,2% - Direct payment 450 95,1% 1496 73,8% Total 473 2026 Contractor - Sole contractor 325 68,7% 1576 77,8% - Combination 148 31,3% 450 22,2% Total 473 2026 Average attendance - Private auctions 5 6 - Public auctions 5 7 Total 5 7 Failed - Private auctions 0 9 - Public auctions 5 206 Total 5 215 Foreign - Private auctions 0 7 - Public auctions 3 15 Total 3 22
The shadow bookkeeping of Koop Tjuchem contains fewer accounts of pre-auctions in the legal period then in the period thereafter. This does not necessarily reflect the amount to which pre-auctions occurred within that period. It reflects the phase that Koop Tjuchem was in as a firm. Koop Tjuchem was founded in the 1980’s and did not yet fit into the ring of cartelised bidders to be invited to attend private auctions nor did it posses the production capacity to place many bids in public auctions. However, in total the shadow bookkeeping of Koop Tjuchem contains 473 records of pre-auction data during 1986 – 1992. Of which 22 construction works were of a revenue size larger than 10 million Dutch Guilders in the period 1987 – 1991. Therefore, the records still provide me with an extensive insight into the inner workings of the cartel in during the pre-1992 legislation.
Other than in the GWW market structure in 2000 (see section 2.1.3) the number of public- and private tenders is higher in the shadow bookkeeping. Other types used to appoint a contractor for a certain work account for 1.9% during 1986 – 1992 and 4.0% during 1992 – 1998. The most reasonable explanation for this is the nature of the other types of appointing clients. Koop Tjuchem only needed to keep records of meetings with other firms. Records of works that were appointed one-on-one via client work were not kept and therefore do not show up in the data.
Settlements of the cartel profits were conducted via two ways. The winning firm in the pre-auction could redistribute its cartel profits among the other cartelised firms via work credit or direct payments. Work credits functioned as a rotation indicator, meaning that once a firm had accumulated a sufficient amount of work credits, it could ask the other cartelised firms to act as the potential winning bidder in the target auction. Direct payments were paid out in a number of ways. Most payments were done under the table. This goes against several testimonies in the Parliamentary Enquiry, but it exemplified by Ad Bos and the shadow bookkeeping. Other forms of payment included wine packages, increased invoices, travel expenses and sub-contracting. In the latter a firm would pledge to construct a part of a work for another contracter. This could be either a firm who had won an auction and would construct a part of a work below a certain cost as well as a firm who had lost a pre-auction and could
still recoup a part of the cartel profits by sub-contracting for the cartel winner. Interestingly, the data in Table 4 shows that in the legal period more transactions were being done via direct methods of payment. A little over 95% of all transactions of cartel profits were undertaken via these direct payments. This figure declined to 73.8% within the years after. Naturally, it could be the case that the growth of Koop Tjuchem and subsequently also the growth in size of their construction works had led to this decline due to complications in direct forms of payment in transactions that large. However, given the nevertheless substantial part of direct payments including direct payments of work worth over 100 million Dutch Guilders, I find it reasonable to conclude that it could be that the cartel legislation did have an effect on the behaviour of the cartelised firms. These firms may have used fewer direct forms of payment in fear of being caught. However, as 26.2% of payments were conducted via work credits, this fear could not have been substantially high.
When a firm would offer their services to sub-contract for another firm as a means of redistribution of cartel profits, it would not tender for the auctioneer as a combination of both firms. These tenders were noted as ‘silent combos’, where it was clear for all other firms involved that in essence it was a combined tender. In Table 4 these silent combos are not accounted for to show the distribution in solo- and combined tenders as seen by the auctioneer. The percentage of tenders where firms submitted a bid as a combination is fairly stable between the legal- and illegal period, declining from roughly one third to a quarter of the data.
This also holds for the average attendance of auctions both public- and private. There is a slight increase in absolute numbers. Albeit, that the growth in attendance for public auctions larger than the growth for private auctions. This combined with the increase in the number of failed pre-auction may indicate an increase in competition after the construction legislation was imposed in 1992. Namely, the number of failed target-auction bids increased from 1.1% in the legal period to 10.6% in illegal period. In addition, whereas in the legal period in no circumstance a pre-auction held before a private auction led to a failed target auction bid, this did happen in the illegal period in 9 instanced. The sample of private auctions is lower in the illegal period compared to the legal period,
however considering the lower failed bids in the public auctions, it is fair to conclude that this same pattern also holds for private auctions.
Lastly, Table 4 provides information on the number of foreigners that are noted in the Koop Tjuchem shadow bookkeeping. A foreign entry in the secret accounts was kept when either a foreigner attended a pre- or target-auction in the Netherlands, or when Koop Tjuchem attended a pre-auction for a construction work outside of the Netherlands. During the legal cartel period this occurred three times in the shadow bookkeeping. In a work commissioned by Rijkswaterstaat North-Holland worth 2.4 million Dutch Guilders, a Belgian firm Doon entered and shared in the cartel profits. Both other instances were of an unnamed Italian firm who entered in two auctions in the province of Friesland worth 2 million Dutch Guilders. Foreign firms shared in the cartel profits for the pre-auctions that succeeded to determine the winning bid. In 7 instances, foreign firms won the target auction without being part of the cartel. In total the 22 foreign cases in the legal period account for 8 records of Belgian firms being present in Dutch auctions, of which two failed due to the Belgian presence. Furthermore, in one Belgian account Koop Tjuchem was present in a pre-auction for the Belgian Railways, in which Belgian firms won, but Koop Tjuchem still received a cartel payment of 750 thousand Belgian Francs on a total revenue of 1.5 billion Belgian Francs. In addition, there are 5 records of German firms entering in Dutch auctions. The cartel noted all 3 instances where a German firm won the target auction as a fail. However, in a construction work connected to the construction of the Schiphol tunnel, it was noted that Koop had to pay the German firm Teerbau 248 thousand Dutch Guilders. Teerbau accounted for 4 out of 5 failed pre-auction meetings due to German involvement, also after this payment agreement was made with Koop Tjuchem. From this I conclude that Teerbau played a double role, both as an outsider of the Dutch cartel, as well as being an insider by profiting of cartel winnings. All other instances of foreign involvement include Surinam and the Dutch Antilles. In Surinam the pre-auctions involved local Surinamese firms where Dutch firms joined pre-auction for large construction works, whereas all instances in the Dutch Antilles show records of pre-auctions of Caribbean daughter companies of Koop Tjuchem, together with other daughter companies of other Dutch construction firms.
3.2 Shadow Bookkeeping Description
In this section I will explain the workings of the shadow bookkeeping as recorded by the representatives of Koop Tjuchem. I will explain the different columns, sheets and accounts of firms in the bookkeeping via prominent examples of the bookkeeping. As in the beginning years of the data the system of pre-auctions was not illegal, these auctions were recorded via accounts in a computer. When the construction legislation came into force in 1992, Koop Tjuchem switched to recording the auctions in writing. Therefore, there is a slight difference between the bookkeeping during 1986 – 1992 and the period 1992 – 1998. I will account for this difference after I have explained the shadow bookkeeping accounts as in the illegal period of the cartel.
3.2.1 Example Sheet
Table 6 featured on page 24includes an example sheet of an account in the Koop Tjuchem shadow bookkeeping. This particular example is taken straight from construction work 7474, which was near the end of the dataset. Each shadow bookkeeping account included 13 columns filled with data particular to a certain pre- and target-auction. Firstly, a pre-auction was assigned a number that would correspond to the same number appointed by other firms whom attended the same pre-auction.9 In this way firms would be able to trace each deal made back to the corresponding pre-auction.
The following column included the date of the tender. This date of the tender corresponds to the date of the target-auction and not the date of the pre-auction. If information were recorded on the date of the pre-auction, this would be written down in the notes section of the account.
The third column included the auction code as determined by the auctioneer. In case of a private tender or when the auctioneer was an organisation with only a few auctions each year, this section would be left empty. The auctioneer was recorded in the fourth column, followed by the expected or mandated date of completion in the fifth column. The latter would be left empty
or would merely include a year, when the expected date of completion was not set in stone.
The sixth column included the winning tender price in the target auction. This column would also include the winning tender price when a firm outside of the pre-auction would win the target auction. In that case the pre-determined bid by the firms in the pre-auction would be recorded in the notes section of the account.
After the winning target auction price, the shadow bookkeeping included three columns specific to Koop Tjuchem. The first of the three was the debit account of Koop Tjuchem in relation to the firm for which the shadow account was kept. If Koop was the winner of a work, this column would include the amount to which Koop owed a monetary amount to another firm. In case of work credits, this column would remain empty, or a zero would be recorded. Naturally, the opposite followed for the Koop credit account. Again, in case of work credits, this column would record a zero or remain empty. The last of the three columns specific to Koop included a settlement, if applicable. At the time of the pre-auction this column could include a previous work with which Koop had agreed to settle a part or all of the cartel profits of the specific account. In that case, as is the case for 7474, Koop would note in that same column for the settled work when, and in some cases with which case, the account was settled. If the account was partly settled, the column included the amount that Koop had to pay or receive. Furthermore, if an account was settled via an increased invoice or a larger Equalisation Fund (see section 3.3.7), this was also recorded in this column.In addition, if a firm would be paid out via another firm this would also be recorded in this column. Lastly, if the pre-auction winner did not succeed to become the target auction winner, a ‘Mislukt’ (in Dutch: failed) would be recorded in this column.
The tenth column included the representative of Koop who was present at the pre-auction. The following column recorded the auction type of the target auction. All public auctions were recorded with a T, whereas all private auctions were noted with an F. The exact meaning behind this remains unclear, it is
mentioned that it found its origination in the digital accounts of the shadow bookkeeping.10
The second last column recorded the winner of the target auction. All other information was reserved for the last column. If a firm outside of the auction had won, it would be recorded in the notes what firm had won the pre-auction, but failed to win the target-auction. In addition, the notes section would include the runner-up price and third price above the winning, if a firm outside of the pre-auction had won. In other instances the notes section was used on a structural basis, but lacked information in 572 instances. Other information included in the notes section pertained to the method of payment, the percentage- or absolute value of price increase for a single firm, the expected date of settlement, whether a firm obtained preference during the pre-auction, the distribution of the cartel profits among the firms in the pre-auction, the sub- and silent-combinations, foreign companies and specific information pertaining to the developments of the pre-auction in preparation to the target-auction.
In this case of 7474, the notes section includes information on the deal made in the pre-auction about the minimum cartel increase per participant in the pre-auction. The ‘5k starting point’ meant that the increase in cartel profits was dependent on the competition in the target auction. The pre-auction winner, Cornelissen, would increase their pre-auction tender price with a minimum of 5k per head, but would increase further if they felt that they faced no competition from a firm outside of the pre-auction meeting. Considering the Koop credit recorded value of 5000 Dutch Guilders, Cornelissen did not increase further than the pre-determined heightening in price.
7474 Pr e-a uctio n co de Ta b le 6 . Ex am ple S he et 21 /0 8/ 96 Da te o f T end er 116 -1996 Tend er -au ctio n co de A m ste rd am No or d Client 15 /0 2/ 97 Da te o f Co m ple ti on 10 68 00 0 Win ni ng te nd er pr ice K oo p de bit 5000 K oo p cr ed it Se tt le d 6-11 -1998 Se tt le d Jd B Rep. T A uctio n T yp e Co rne lis se n Win ne r 5k sta rti ng po in t No te s
3.3 System of Pre-Auctions
In this chapter I will dive deeper into the workings of the system of pre-auction based on the depositions in the Parliamentary Enquiry, conversations with Ad Bos and the data from the shadow accounts. The system of pre-auctions can be distributed in 7 steps from the first point of contact to organise a pre-auction, to eventual settlement and distribution of the cartel profits. These 7 steps will be discussed in detail below.
3.3.1 Step 1: Arranging the Pre-Auction Meeting
The first step of the pre-auction was initiated by the issuance of a tender by a firm or governmental institution. In the case of public tender firms knew, based on previous construction works, which other firms would be most likely to submit a bid in the target-auction One firm would initiate the first contact towards other firms. This would spearhead a snowball effect of firms contacting each other leading to an eventual pre-auction meeting containing most or all of the firms that planned on submitting bids. Given the amount of firms attending public auctions, it was not always feasible to meet physically. Therefore, substantial parts of the pre-auction were discussed via other means of communication.
In the case of a private tender, it was more difficult for firms to determine which firms had been invited. Based on tight connections within the construction sector, the pre-auction system started with a few firms that would come together, knowing that they had been invited to the auction. From there the network grew from firms who had heard via unofficial channels which firms would attend the private target auction.11 In the case of a private auction with a maximum of 15 firms attending, the physical pre-auction would be held on the day of the target auction. If for example a target auction would take place at 12:00 in the afternoon, the pre-auction would take place only a few hours beforehand at 10:00.
11 Interview Ad Bos (2018)
Since the physical pre-auction took place shortly before the target-auction, most of the work to determine the pre-auction winner was conducted beforehand. This was done via communications between firms individually or in smaller groups of 3 to 4 firms. During these communications, firms tried to come to an agreement before the physical meeting to avoid any surprises at the target-auction.
A key importance when arranging the pre-auction meeting and deliberation of the cartel was the desire of cartel members to include all firms in the pre-auction.12 This desire stemmed both from the cartelised firms as well as the firms outside of the cartel. This is remarkable as in standard auction theory literature a larger and more heterogeneous cartel increases the instability within a cartel. A cartel is stable if all cartel members prefer to be in the cartel and all non-members prefer to be outside the cartel.13 Deviations from this increase the instability due to the incentives of in- and outside firms to disrupt the cartel. The stable collusive outcome is the outcome that maximises the joint profits of the cartelised firms.14 This leads to firms having the incentive to form only the smallest sustainable cartel. Since the smaller firms in the construction sector were not able to take on larger construction works by themselves, the theoretical outcome would be to not include these firms within the cartel.
So why did construction firms in the first step of the pre-auction mechanism differ from auction theory?
A first explanation stems from the failure risk firms inside of the cartel wanted to reduce. By including outsiders firms reduced the risk of an outside firm undercutting the cartelised pre-auction winner in the target-auction.
A second explanation lies within the assumptions of theoretical collusion models. In Vasconsoles (2005) firms are assumed to be homogeneous, which is not the case in the Dutch construction cartel. In the GWW sector firms differed greatly in their production capacity at any point in time, due to the large cyclicality of demand. When larger firms were unable to take on a large project
12 Interview Ad Bos (2018) 13 d’Aspremont et al. (1983) 14 Vasconcoles (2005)
by themselves, heterogeneity in available production capacity could incentivise larger firms to include smaller firms to produce for them.
Bos and Harrington (2010) correct for this homogeneity by introducing firms that are heterogeneous in their capacities into an infinitely repeated game of collusion. The authors find that in such a setting cartels would be incomplete in equilibrium as well. In this equilibrium large firms would be incentivised to join a cartel, while firms that are smaller in their production capacity would rather be excluded from the cartel. This is because the price increase of the cartel depends on the production capacity a firm is able to add to the cartel. Thus, a large firm is able to increase the cartel profits by a greater amount than smaller firms. This decreases the threshold for a larger firm to join. However, given the intertemporal occupation of production capacity for both type of firms, the production capacity a large firm adds to the cartel for a specific work may not be sufficient to increase cartel profits for this to be a stable outcome. In that case large firms will have to include smaller firms to increase cartel stability and decrease the risk of winning without the production capacity to produce the work.
On the other side, smaller firms were incentivised to join the cartel. This is not in line with what would be expected as well. Normally, the increase in price by the cartelised firms would act as an umbrella for smaller firms to increase their prices without having to share this increase in price with others.15 Thus, being outside of the cartel would generate higher profits than joining the cartel. So why did smaller feel the need to join the pre-auction? The answer to this question is three-fold.
Firstly, the largest firms in the GWW sector dominated on the market of the inputs needed to produce construction works.16 This meant that without the consent of large construction firms, smaller were unable to produce. Koop Tjuchem also exemplified this in section 2.1.2 where I discussed the construction of an asphalt work. This was almost made impossible due to the refusal of the needed inputs by larger construction firms.
15 Bos and Harrington (2010) 16 Interview Ad Bos (2018)
Secondly, the dominance of these large firms on production inputs enabled large firms to punish small firms severely when they did not join the pre-auction. Refusal to cooperate in a pre-auction, led to vertical exclusion in the chain of production.17 This potential punishment incentivised small firms to participate.
Lastly, small firms were incentivised to join the cartel, because they were able to share in the cartel profits without having the wish to produce. The mere attendance of a pre-auction was sufficient for a firm to share in the potential cartel profits. This meant that firms that did not want to produce a certain work, or were even unable to produce a certain work, were incentivised to behave as if they did want to produce. All firms, large and small, were incentivised to attend as many pre-auctions as possible to share in as many cartel profits as possible. This pre-auction characteristic led to a discrepancy in the private willingness to produce versus the public willingness to produce by firms. The possibility of sub-contracting if a firm would win a pre-auction against their own willingness decreased any negative potential side effects of this behaviour. These negative effects included winning the pre-auction below their own cost price or not having sufficient production capacity to produce. After all, if the win of a firm would strain a too heavy burden on their production capacity, they could hire other firms to produce for them, while still collecting a share of the cartel profits. In the worst case firms could use their share of the cartel profits to pay for the cost of sub-contracting. Thus, the most negative situation would be one of zero profits, making this behaviour relatively risk-free.
3.3.2 Step 2: Determining Pre-Auction Tender Price
Once the firms had decided on the participants in the pre-auction, it was important for each firm to determine their strategy to win the pre-auction. A firm would win a pre-auction based on a number of criteria. The most important of these criteria was the price that they submitted. A pre-auction would follow the same criteria as the target auction so that the firm with the lowest price would win the right to submit the potential winning bid in the target auction.
17 Parliamentary Enquiry (2002)
However, there is an exception to this. Based on the shadow bookkeeping of Koop Tjuchem this exception is the use of turn rights (staffels: in Dutch). In essence, turn right were used to bypass step 2. If a firm had accumulated a sufficient amount of turn rights, it could ask the other firms in the pre-auction for preference to produce a construction work. If agreed, the other firms in the pre-auction would submit bids higher than the bid of the firm who had been given preference. If a firm had asked and had been granted preference, this would be recorded in the notes section of the shadow bookkeeping. The shadow bookkeeping data shows 92 records of construction works where preference had been asked and granted.
In other cases the lowest bid determined the pre-auction winner. Therefore, it was important for all firms to calculate their preferred tender price for the pre-auction. This pre-auction bid was determined in a somewhat competitive setting. I include the word ‘somewhat’ because it must not be forgotten that this competition occurred within the setting of a cartel. However, representatives of construction firms have testified that the submissions of pre-auction tender bids were highly competitive.18 Ad Bos (2018) describes it as very strong competition until the moment it was determined which firm would receive to right to submit the potential winning bid of the pre-auction in the target-auction.
The determination of the pre-auction tender price consisted out two steps. The first step of each firm was to determine their cost price of the construction work. The second step was to add a profit margin that would not have to be shared with other in the cartel. The combination of the cost price and the profit margin would become the pre-auction bid. In essence, this profit margin will have been the same as in a competitive situation. However, due to the longevity of the construction cartel it is difficult to deduce any clear non-cartel information from this.
In addition, due to the heavy competition the pre-auction bid will have been dependent on the expectations of a firm of the bids of other firms. Here it is
18 Deposition 1 until 39: Ph. Groen, Director Projects Dura Vermeer: “(…) we have strong competition from each other, another firm keeps me (meaning: my price) low. (…) we are keeping each other poor.”
important to look at the system of pre-auctions as a repeated game. Because firms came in to contact with each other multiple times within a given period, they developed an understanding of the cost prices and profit margins of other firms. The pre-auction tender prices were dependent on these understandings of other firms.19
Representatives of the construction industry have declared in depositions that cartel profits from earlier works were sometimes used to decrease the pre-auction tender price below cost price.20 However, the data from the Koop Tjuchem shadow accounts complemented by conversations with Ad Bos are contradictory to this statement. Firstly, given the extensive records of the shadow accounts and the detail to which characteristics of each pre-auction was written down, it is remarkable that there exists no indication of Koop Tjuchem noting that a firm submitted an out of the ordinary low bid in the pre-auction. Remember that each firm could make clear indications of the cost price of other firms. Therefore, it is likely that firms would be able to notice deviations from their own expectations. In addition, there are no records that Koop Tjuchem itself noted that they had to reduce their pre-auction tender price below their cost price to win. There are notes that stress that the target-auction tender price was low, but not the pre-auction tender price. There are 50 instances in the shadow bookkeeping of a low price in the target-auction.
In addition, Ad Bos (2018) also confirmed that firms that not use earlier built-up cartel profits to lower their pre-auction tender price. Bos went on to say that if it happened, it happened only in rare cases. Moreover, in those rare cases it would be most likely to be a large firm that would such a strategy. A small firm would not be able to afford to use these built-up rights, as they are mere credits and do not reflect real liquid money before being paid out.
It is also questionable if it would be rational for firms to use earlier cartel profits to lower the pre-auction tender price below their cost price. Namely, any lowering of the price in the pre-auction would lead to an increase in the cartel profits. The difference between the submitted bid in the pre-auction and the bid
19 Ad Bos (2018): “(…) if my cost price is 600k but I know others are not as price efficient, I will tender just below the other. Everyone is very good with numbers.” 20 Ter Steege, NMA (2005): “(…) built-up cartel profits were used to lower the tender price extra if a firm really desired to produce the work.”
in the target auction. By lowering the bid in the pre-auction a firm would increase the cartel profits for other firms, while decreasing their own previous cartel profits and potentially accepting the construction work below cost price. This potentially loss-making behaviour would not benefit the cartelised contractor or the auctioneer. Only the other firms attending the pre-auction would benefit.
Bos (2018) did confirm that in exceptionally large works firms would use earlier cartel profits to lower the profit margin that was placed on top of their cost price. The same irrationality argument applies here as in the situation of lowering the pre-auction tender price below cost price. However, it is important to note that the division and settlement of the cartel profits was a game in itself. If a cartelised firm won the right to produce a work, it received all of the cartel profits that were to be redistributed among the other cartelised firms at a later time. There were no fixed rules in place as to when and where this redistribution had to take place. Furthermore, if the cartel profits were redistributed via work credits there were no direct monetary payments from one to the other. This meant that the winning firm could keep the cartel profits for periods of 12 months or longer.21 This amount could be very substantial for large construction works where the amount to be redistributed could reach as least as high as 27 million Dutch Guilders.22 Such an amount could be very welcome for a firm facing liquidity constraints and incentivise a firm to reduce their profit margin, knowing that it would take time before they would have to pay out the cartel profits. In addition, taking on a construction work could also potentially mean extra revenue from potential extra future work. This means that it could still be rational for a firm to decrease their profit margin if their expected value of future work together with the decrease in profit margin remained positive.
Furthermore, another reason for firms to occasionally use earlier cartel profits to reduce their profit margin could have been to settle cartel profits of earlier pre-auctions. By increasing the cartel profits for other firms without sharing in these profits, a firm would be able to make sure that their balances of rights to cartel profits remained fairly balanced. However, from the shadow
21 Deposition 39 until 67, Ad Bos (2002)
bookkeeping there is no clear indication that this happened frequently. In addition for construction works with revenues smaller than 5 million Dutch Guilders the use of turn rights (staffel) played a large role. Given the turn rights firms acquired on the ‘staffel’, there was no real need to use earlier cartel profits to balance the distribution of cartel profits.23 A large discrepancy in credit to- and debit of cartel profits meant that a firm would also have a turn right and could ask for preference in the pre-auction without having the need to reduce their pre-auction tender price.
Lastly, it is important to note that after all the above considerations for a firm to determine their optimal pre-auction tender price, the cartel allowed a firm to retract their bid in the pre-auction after submission.24 This potentially increased the incentive for firms to bid aggressively. If a firm would notice in the pre-auction that their bid was very low and this would indicate that they might have made a mistake in calculating their cost price, they could simply retract their bid and retract from the pre-auction. In this way the auction mechanism removed the winner’s curse from the pre-auction. The winner’s curse is a common phenomenon in common value auctions. It entails the chance that a firm places a bid higher (or in this case lower) than the actual value of the object under bid by competitive pressure. In that case the bidder is left with a negative value. In a regular auction setting this chance will ensure that bidders reign in their bids. Thus, by removing the winner’s curse bidders can bid more aggressively.
3.3.3 Step 3: Determining Target-Auction Bid
In the physical pre-auction, determining the price with which the cartel would bid in the target-auction was the first step that would take place. While in the period before the physical pre-auction the cartel had determined which firms would be present and what each firm would bid, the physical pre-auction started with determining the maximum willingness to pay of the auctioneer.
23 In 193 auctions agreements were made to record the auction on a ‘staffel’ in construction works worth up to 5 million Dutch Guilders
In a few instances the willingness to pay of the auctioneer was public information. However, in the majority of the auctions the willingness to pay could either be inferred by the construction firms or was secretly leaked by employees within the auctioneering institution.25 The latter case is corruption by government officials. Once a firm had obtained information on the willingness to pay of the auctioneer, this information was shared with other firms attending the pre-auction. Thus, it was only necessary for one firm to obtain this information to be able to let the whole of the cartel make use of this. According to depositions there was always at least one representative of a firm who would be able to obtain this information.26
If the willingness to pay was not made public, or could not be obtained by representatives within the auctioneering institution, construction firms were able to obtain the information by drawing inferences from earlier construction works by that same institution. Construction firms could establish a fair estimation of the new maximum willingness to pay based on this historical data. This stresses the fact that the auctions should be considered as a repeated-game. Information obtained in previous auctions was used to determine optimal strategies in the following auctions.
After an agreement was made on the most likely willingness to pay of the auctioneer, the cartel decided on the bid price in the target-auction. Firstly, the target-auction bid would be what the cartel firms assumed to be a ‘fair’ price for the auctioneer. In this case a ‘fair’ price would be a price that would not be substantially higher or lower than prices of previous comparable works. In practice, this meant that the target-auction bid would be as close as possible to the willingness to pay of the auctioneer as the cartel could get. In some instances it was even the case that the cartel increased their bid above the willingness to pay of the auctioneer. This occurred when the cartel knew that the auctioneer had no real alternative but to turn to one of the cartelised firms. In that case the cartel feared no competition from firms outside of the cartel. According to Ad Bos (2018) this happened most often in private auctions by lower levels of
25 Ad Bos (2018)
26 NMa (2005) Van Leeuwen: “There was always someone who knew or though to know the budget for the construction work.”
