It is argued that the better quality of evidence obtained through Leniency Programs reduces the ambiguity of proofs which in turn allows to increase maximum fines for cartelization. The trend to set higher fines for this unlawful practice reinforces the deterrence effect of antitrust laws instead of diminishing. The article also presents the current state of cartel prosecution in Chile. Recent cases of alleged collusion brought by the competition agency before the tribunal and supported only with circumstantial evidence, has been dismissed either in first or second instance by the courts. A reform in competition law, currently under discussion in the parliament, will introduce Leniency Programs and also will give power to the competition agency to obtain material evidence. Increments in fines are also included in the bill. These reforms are aimed to improve the instruments to effectively fight collusion.
The introduction of Leniency Programs (LP) has been an important milestone in the prosecution of cartels around the world. The large number of cartels unveiled through the implementation of these programs has caused widespread impact on the antitrust activity.
According to the OECD, following the entry into force of the new directive on corporate leniency in the United States in 1993, applications to qualify for the benefit of this program increased from one case per year to more than twenty. In terms of fines, the total amounts paid by firms rose from 23 million in 1993 to 1,100 million in 1999. Importantly, this latter figure is higher than all the fines accumulated in more than a century of application of the antitrust laws in the United States. In the European Union, the number of sanctioned cartels has increased from one per year during the period 1989-1997 to five per year during the period 1998 -2005.
Entry into force of Leniency Programs
Country or Jurisdiction
European Union – Korea
Brazil – Canada
Norway – Switzerland
Japan – México
The success achieved in a short time span by the LP on the detection of cartels has led several countries to introduce amnesty programs in their local legislation. Nowadays, all member countries of the OECD, with the exception of Turkey, have explicit laws on amnesty for collusion (see Table 1).
An important component of the success of LP is the quality of evidence obtained through this mechanism and its effect on the litigation of cases and the application of fines. The great revelation of information achieved by the LP on the working of cartels (geographic markets involved, duration of the collusive agreement, companies involved, methods of monitoring and enforcement, etc) makes easier to establish a case of illegal action before the courts dispelling any ambiguity about the existence of a cartel. The better quality of information gives greater certainty to the courts at the moment of deciding on the basis of the available evidence. When it is not clear that firms had been colluded explicitly, the judicial system will be reluctant to sanction them. When the evidence that courts have allows multiple interpretations, besides the collusive one, it is unlikely that courts will opt for applying any sanctions at all.
With the information revealed through the LP, the likelihood of error in a judgment about cartels has been virtually eliminated. Moreover, the magnitude of fines has been higher due to the application of aggravating circumstances. The evidence shows us that the entry in force of LP in Europe led to the application of sanctions in average higher in comparison to the previous period. This would imply that although LP selectively offers a reduction in the sanctions, on average the punishment for cartels is bigger now, which undoubtedly strengthens the deterrent effect of such measures
2. Circumstantial Evidence vs. Hard Evidence
Historically, there have been two ways to prove the existence of collusion, the so-called circumstantial evidence and the hard evidence. The first employs the commercial behavior of firms in the market, which is presumed to be explained by an explicit agreement between the firms. It is generally believed that parallel behavior both in price and types of bids or negative sales would be indicative of a coordinated behavior among participants in the same market. The hard evidence, in turn, corresponds to physical evidence, such as documents, minutes, recordings, e-mails that show clearly that there has been direct communication between companies to agree on prices or market sharing.
Qualitatively both types of evidence are not equivalent. The courts tend to give more probative power to hard evidence than the behavioral one. The reason is that the former clears any doubts beyond what is reasonable about the existence of the cartel, while the latter always raises questions about whether the allegedly unlawful conduct is due to a competitive behavior. For example, the parallel conduct in prices could be explained by both that the firms have been colluded to adjust their prices simultaneously, or that some of the fundamentals in the market such as costs or demand have changed. Moreover, in oligopolistic markets, where there is interdependence between competitors, is expected that movements in prices of a firm are followed in the same direction for the rest of the participants. Thus, in principle, the parallel behavior would be consistent with both an explicit coordinated conduct and individual profit maximizing behavior.
Often cases based on circumstantial evidence are accompanied by additional factors, such as the exchange of business information between firms and some other facilitating mechanisms of coordination. Structural aspects of the market in question such as the level of concentration, the existence of entry barriers, transparency in prices and abnormal levels of profitability of firms are usually presented as evidence to reinforce the hypothesis of collusion.
Courts must weigh, using the available evidence, the error they commit punishing firms that did not act unlawfully (type I error) versus leave unpunished firms that have been cartelized (type II error). When evaluating the options, the courts not only consider the probability that the observed behavior is due to collusion, but also takes into account the damage caused to the parties and the working of the market if they punish wrongly. Note that the weaker is the support to the hypothesis of collusion by the circumstantial evidence, the greater is the type I error and the lower will be the willingness of the court to impose sanctions. This reluctance to apply sanctions based only on behavioral evidence is exacerbated by the fact that in many jurisdictions collusion is considered as a criminal offense. Thus, the requirement of no ambiguity of the evidence becomes more important.
The hard evidence does not suffer the problem of ambiguity in the interpretation, which certainly facilitates decision. Furthermore, given the nature of offense per se given to the explicit collusion in many jurisdictions, it reduces litigation because the structural and behavioral aspects of the market play a secondary role.  Although we still may observe sanctions to companies for colluding, based only on circumstantial evidence, in recent years there has been a tendency in the United States and Europe to apply more stringent standards to condemn firms without material evidence. 
3. Relationship between Fines and Quality of Evidence
Under the traditional paradigm of “public enforcement” to prosecute crimes, a company compares the benefits of acting illegally versus the costs that this action involves, when deciding whether to collude or not. The cost of the illegal action is equivalent to the chance of being detected and punished (p) multiplied by the magnitude of the fine (F), which amounts to pF.
Clearly, a higher probability of detection or a higher fine makes greater the cost of colluding. The parameter p depends on the effectiveness of the mechanism of persecution. It is expected that the more resources are devoted to the task of prosecution, the more likely is the capture of wrongdoers. On the other hand, increasing the fine, although costly for the transgressor, has no social costs because it is a mere transfer between agents. Thus, the most efficient solution from the social point of view is to allocate the minimum amount of resources to increase p and fix the fine F to the maximum level. This recommendation came from the classic result of Becker (1968) who suggests that in order to deter crimes it is optimal to set fines equal to infinity. As to the relationship between the two variables p and F, these would substitute. In other words, to achieve a certain level of deterrence, we can augment F and diminish p or vice versa.
In this model, the parameter p is the probability of being detected committing the crime and hence sanctioned. It assumes that once detected the offense, with probability one the company will be fined. This definition implies that the evidence obtained by the enforcement agency is perfect in the sense that there will be no doubt that a court of justice, after observing the evidence will condemn the accused. However, when the evidence contains some ambiguity about what can be inferred from it, and leaves some level of doubt as to whether or not the defendant acted unlawfully, the result of the traditional model of “enforcement” explained above is lost.
To illustrate this, we will introduce an intermediate stage between the gathering of evidence and application of the sanction, which will be the evaluation of the evidence by a court or a jury, who ultimately will decide whether the evidence is clear enough to condemn.
If we accept that judges assess the error they may commit when deciding in one way or another, there will be a cost benefit analysis applied, by assessing the magnitude of the error being committed by each decision. Andreoni (1991) presents a model of behavior of judges which includes the evaluation of possible incorrect decisions, so that those who decide get a negative value if they make a wrong sentence.
To further explain this reasoning, we denote as U1 (X + F) the utility that obtains a judge (or jury) for punishing a company that has not colluded (Type I error), while U2 is the utility of the judge if he does not condemn a guilty firm (type II error). Both utilities are negative, and U1 is growing in absolute terms in its argument (X + F). This subjective utility includes both the social damage caused by an erroneous decision, and the personal cost that the judge perceives of being unfair when applying the rules.
In the specific case of collusion, the parameter X may represent the distortion introduced in the market by forcing a company to not behave in a manner that is allegedly anticompetitive when it is not. This factor is relevant especially when the proofs are based on indirect or circumstantial evidence. The magnitude of the fine F also affects the disutility of a wrong verdict, since the larger penalty the greater is the injustice being committed with a wrongly condemned accused.
Suppose that, based on the available evidence, there is a probability q that a company has explicitly engaged in collusion and by consequence a probability 1-q that is innocent. A judge decides upon the expected value of the utility of each alternative, so he chooses the option that minimizes the error in expected terms. Thus, a firm is sanctioned if and only if:
Which is equivalent to the following condition:
This result, obtained by Andreoni (1991) tells us that in order to condemn, the quality of proof, represented by q, must exceed a certain threshold which depends on F, the magnitude of the fine. As we can see, the term of the right hand side is increasing in F, which means that a larger fine, demands a higher standard of proof by the courts to sentence an offender. An interesting fact that emerges from this result is the relationship of complementarity that would exist between q and F, which contradicts the result of substitutability predicted by the traditional model of enforcement.
While in this model q is exogenous and depends on the characteristics of the case under analysis, it can be affected by the type of information that brings the prosecutor before the judges. With evidence of high quality, there will be a greater likelihood of condemnation if the firm is guilty, but also there will have more room to increase the fines, which strengthens the deterrent effect of the global mechanism of persecution. The important lesson to draw is that fines are dependent on the quality of evidence available. Increasing fines without improving the quality of evidence may even be counterproductive, because those who decide may be more reluctant to apply sentences.
Table 2 shows four scenarios depending on the quality of evidence available and the magnitude of the monetary penalties, for the case of cartels prosecution. We have two extreme cases, one in which there is no possibility of detection and monetary penalties are low and other in which the quality of the evidence is high and the fines are significant. In the latter case, the goal of deterring crime is achieved.
Quality of Evidence
Level of Fine
Effective Detection and Deterrence
When the quality of proof is good but the maximum fine is low, companies will choose anyway to collude and then pay the fine once discovered. This may be the case of LP that are overly generous in granting amnesty. As noted by Spagnolo (2005), this scenario apart from being ineffective is inefficient because antitrust agencies must spend resources in the task of obtaining evidence without satisfying the goal of deterring collusion.
In the opposite case -poor quality of evidence but high level of fines- there is impunity in the sense that companies will never be punished for colluding due to the impossibility of proving the unlawful act with the available evidence. Increasing fines is not a solution to this problem, and as we have explained above it may worsen the situation. This scenario may be representative of countries that having severe penalties for collusion, do not have the power to obtain direct evidence, either by LP or by other mechanisms and where most of cases brought to the court are based uniquely on circumstantial evidence.
Leniency Programs and the Quality of Evidence
Authors like Motta and Polo (2002) have mentioned the inherent trade-off between desertion and deterrence of LP. The reduction or annulment of the fine offered by the LP while destabilizing the formation of cartels, reduces the cost in expected value of colluding, weakening the deterrent power of the anti-cartels policy. This result is based on the assumption that the fine applied either by direct inspection or by the LP are equivalent in magnitude, which in turn implies that the quality of the evidence obtained by both mechanism are equivalent as well.
If penalties are dependent on the quality of information that can be extracted about the characteristic of a crime, then fines may differ upon the mechanism employed to obtain the evidence. Leniency Programs have led to the disclosure of a wide range of information on the details of the working of cartels. This includes duration of the agreement, the markets involved, sales volumes, participating companies, and so on. Such wealth of information enables the courts to increase fines by applying the aggravating circumstances. In Europe, the Commission sets fines depending on the number of years the cartel operated. It may also apply a higher fine to the firm that played the role of ringleader or instigator of the agreement. The more information is available to the courts the higher the probability that they can justify an increase in fines.
Although the evidence obtained either by direct inspection or by self-confession using LP constitutes proofs of high quality, it is relevant to ask whether LP, besides helping in the detection of cartels, produce evidence of higher quality. Is it expected that those who desert form a cartel through LP provide better information than can be obtained by direct inspection? The answer is yes, since under LP the firms have all incentives to provide as much information as possible when decide to defect, while under the threat of inspection firms will tend to hide the evidence. Moreover, LP introduces additional incentives to keep evidence in the first place, which can be used in case that one of the companies opts for deserting later on.  In the absence of LP, there is no benefit of maintaining information about the characteristics of a cartel because it can be used against the firm itself.
There is some evidence suggesting that the introduction of LP have tended to increase fines on average rather than to diminish them, respect to the period prior to the introduction of such programs. Brenner (2005) obtains results along this line for the case of Europe, indicating that while the LP selectively reduces fines for certain participants of a cartel, in expected value fines are higher, which would tend to reinforce the deterrence effect of the anti-cartels policy. In addition, different economies have tended to increase the maximum fines established for the case of cartelization. Such is the case of the European Union which in the year 2006 increased the maximum fine for collusion of 10% of annual sales to 30%.
4. Institutional and Legal Framework for Prosecuting Collusion in Chile
The Chilean competition law establishes that collusion is an anti-competitive practice. According to Article 3rd a: It shall be considered as a practice that hinders competition and therefore subject to punishment: “Any explicit or tacit agreements between traders, or concerted practices between them, intended to fix prices of sale or purchase, limit production or allocate areas or market shares, abusing the power that such agreements or practices confer to them ”
There are two elements to be highlighted in the text of the law. First, both types of collusion tacit and explicit are considered as illegal, and second, it is mentioned the term “abuse of market power” for committing an offense. The way that the anticompetitive practice is characterized in the law, mentioning the concept of abuse of market power, leave room for multiple interpretations at the moment of prosecuting a case of collusion. In a more permissive view, collusion would not have the status of offense per-se as it is needed to prove that competitive harm occurred in the form of abuse of market power. From the point of view of prosecuting cases, the agency competition not only must have evidence that demonstrate that the practice effectively occurred but also should prove that the colluding companies posses market power and that its action allowed them to exercise such an abuse of power.
The legal category given to collusion in Chile differs from the approach of per se anticompetitive that applies in the rest of the world. As is clear from the text of the Chilean law, collusion would have a similar status of the most of the so-called exclusionary or monopolizing practices. In those cases, along with proof that the practice existed, the agency must demonstrate that it lead o would probably lead to competitive damage. International jurisprudence has tended to give the character of offense per-se to collusion because it is estimated that situations where collusion is harmless or even beneficial are exceptional and rare. Therefore courts are willing to sacrifice any cases of collusion beneficial for greater clarity and ease of interpretation of the rule. By contrast, the theory shows that the practices called exclusionary such as bundling, tying, price discrimination or other are not always harmful from the point of view of competition and not in few circumstances may even be beneficial to final consumers. That is why they are analyzed according to the criterion of the rule of reason, where the agency tries to determine whether or not the practice tends to have the effect considered negative.
The Chilean Competition Tribunal (TDLC)  however has converged towards another interpretation, as evidenced by its recent sentences. This Tribunal has established as a necessary and sufficient condition that firms that engaged in collusion must possess collectively market power regardless of its actions led or not to competitive injury. While in the case of Chile the TDLC is close to internationally applied criteria and the recommendations of experts, the lack of clarity in the law leaves open the possibility that the interpretation changes. Moreover, we do not know whether the Supreme Court will endorse the approach employed by the TDLC for cases of collusion or will make a less restrictive interpretation, requiring the demonstration of competitive harm form the part of colluding firms. So far, we have not reached this stage because the discussion has focused on an earlier stage which is the quality of proof.
Chile has almost no record of successful prosecution of collusion. In general there is not widespread consciousness in the business community that collusion is a serious offense to competition. For example, the president of a medium-sized company of retail asked on whether he feared anti-competitive actions from its larger rivals said to the press “to deal with this problem I trust more in a frank conversation between competitors, instead of relying on the antitrust institutions ” Neither it is unusual that local government authorities sponsor agreements among small public transportation operators to raise their fares in order to avoid price wars or conflict scenarios.
The current antitrust institutions in Chile dates back to 2003. The main agencies are the “Fiscalía Nacional Económica” (FNE) and the “Tribunal de Defensa de la Competencia.” The first is the agency responsible for representing the public interest in competition cases, acting as prosecutor and reporting cases to the Tribunal. It has no sanctions power except in very limited areas. The TDLC is a tribunal specialized in resolving matters of competition. It is composed by five members, three lawyers and two economists. This tribunal has the same rank as the Court of Appeal in the Chilean legal system and its decisions are appealed to the Supreme Court.
Reforms to the law
There is currently a bill for amending the Chilean competition law, which is being discussed in the Congress. Among other things, the reform introduces Leniency Programs and also gives special powers to the competition agency (FNE) to investigate anti-competitive actions, such as collusion. Specifically, it will allow the FNE, with the permission of a judge of the level of the Court of Appeals, to perform searches in the private premises of companies in order to seize records that may serve as proof. Similarly, it authorizes the agency to intercept all forms of communications.
According to the text accompanying the bill, both Leniency Programs and the power given to the FNE of obtaining direct evidence are complementary instruments that make more effective the investigation against collusive agreements. As recommended by the theory and shown by the practice, it is unlikely that LP by itself will induce firms to disclose information accusatory if there is no a real threat that agency can obtain evidence by direct inspections.
Another important change in the law is the increasing of 50% on maximum fines, from $ 16.2 Million to $ 24.4 Million. Such an increase, as explained in section 3 is consistent with a better quality of the evidence that is expected due to the reforms mentioned above. Also, the concept of competitive damage has been introduced as a reference to fix the level of fine to the condemned firms.
5. Collusion Prosecution in Chile
Between 2006 and 2007 there have been three rulings about collusive practices in Chile: Shipping Agencies, Oxygen Medicinal and Health Insurance. In these three cases, the accused were acquitted either in the first instance, by the TDLC or in the second instance by the Supreme Court. The critical element of the decision was the quality of evidence. i.e whether the circumstantial evidence was a sufficient demonstration that there was agreement (explicit or implied) between the players in each case. The disparity in the way that the circumstantial evidence was interpreted has been reflected both in the existence of divided decision within each court as to the difference in the decisions between the TDLC and the Supreme Court. Below are discussed and the most important elements of each of the cases.
At the beginning of 2003 the Exporters Association filed a complaint against a number of shipping companies for an agreement in prices among them and discriminatory practices applied to their clients. The plaintiffs argued that the defendants arbitrarily applied a charge for unsolicited services called “Integrated Services Documentaries” associated with the processing of export documents.
The main element of suspicion of a concerted practice was the simultaneity in the application of the new charge by the shipping agencies (between March and April 2002) and the similarity of tariffs among them, which vary between 25 and 27 dollars. The accused firms pointed out that the charges were based on the costs involved in such services. With regard to the similarity of tariff changes among competitors, the agencies pointed out that being all in the same market and performing similar functions in a scenario of fierce competition, it is normal to observe such similarity among competitors.
In a unanimous verdict, the Tribunal condemned the shipping agencies for concerted behavior. According to the TDLC the charges applied by the defendants were unjustified, and could only be sustained over time if several companies agree to fix them at same time and under similar conditions. The proof employed by the TDLC was the behavior of the accused firms on the market and no evidence demonstrating that the firms were communicating each other to coordinate its business strategy. As stated in the ruling: “….the Tribunal has formed the conviction to be in the presence of a concerted imposition.” For the Court, the actions undertaken by the companies such as type of charge implemented, his simultaneous and the equivalent levels were highly unlikely to be the outcome of the action of individual agents and therefore were almost exclusively explained by concerted action.
The ruling was appealed to the Supreme Court, who revoked the verdict. According to the Supreme Court, the facts cited as evidence-commercial behavior-represented only a hypothesis of collusion, but under no circumstances it constitutes conclusive evidence of unlawful conduct. For the Court, collusion requires the existence of a joint will and determination to carry out this practice, which would not be demonstrated with a simple similarity and simultaneity of tariffs change. The parallel conduct, according to the verdict, could just be explained by the similarity in the type of services provided by the shipping agencies and the competitiveness of the market, which leads to involved companies to quickly imitate the strategies of competitors.
In August 2005, the FNE filed an injunction against several companies in the business of medical oxygen for concerted actions aimed to restrict competition and discriminate against clients. Specifically, the FNE accused the companies Air Liquide SA Chile, Indura SA, AGA SA and Praxair Chile of dividing the market of procurement to public hospitals and to seek the failure of the bidding organized by the agency in charge of the purchases of supplies for hospitals (CENABAST).
The Tribunal found that companies were guilty of collusion, with one vote of dissension. The main piece of evidence for the court was the behavior of firms in the bidding process called by the CENABAST. The purchasing agency devised a complex system of auction in order to avoid collusion among the participants. It should be mentioned that there were serious suspicions of the existence of collusion among the major providers of oxygen in their direct sales to government hospitals in the previous years.  So, it was crucial to design a mechanism proof of collusion among bidders. The process consisted of a descending auction with sequential offers and a reservation price kept secretly by the agency. In a first stage each of the participating firms offers a price for supplying each of the geographical zones. Once first stage prices were offered, firms were asked in a successive way whether they were willing to bid $ 10 less than the lowest bid existing so far. Finally, if prices were above the reservation value, firms were asked whether they were willing to match the reservation price set in advance. During the bidding, the companies offered in the first and second phase values above the reservation prices that three of them finally accepted.
For the Tribunal the fact that the initial offers of the participants were almost 50% above the reservation value was an evidence of collusive behavior. Not offering in the first round the minimum payment acceptable to provide the service would be contrary to the individual interest of each company, and therefore it would be indicative of collusion.
According to the TDLC, the oxygen companies had clear incentives for not being aggressive in their bids during the auction. Offering low prices would introduce a gap with respect to the prices prior charged to other customers and would make transparent a market that was considered opaque at that time that allowed firms to discriminate prices among different clients. The transparency caused with the auction would induce the clients of the companies -who have contracts containing higher prices- to demand reductions in the rates they have to pay, something that is clearly detrimental to the oxygen firms. The existence of a kind of implicit “Most Favored Customer” clause would incentive the firms to soften his behavior in the auction, a strategy –in opinion of the Tribunal- that would only succeed under a coordinated action between the companies.
In the opinion of the dissident judge, the conduct of the firms in the auction is not a sufficient evidence of collusion. It may also correspond to the action of individual companies that operate in an environment where there is interaction between the parties as is the case of an auction with multiple stages. Of particular importance is the fact that the alleged agreement finally failed and some firms accepted the reservation price set by the principal.
Moreover, the dissenting vote pointed out that in order to have a case of collusion, all the conditions need to be satisfied, not only the existence of an agreement. In other words, although we have proofs that there was coordination among firms, we also must demonstrate that such agreement allowed those firms to abuse of market power conferred by the collusion. According to the judge, this abuse or harm would not have happened since the bidding was successful. The reasoning of the dissenting judge employed in this case would not sustain the interpretation of collusion as a per se offense, giving to this practice a legal status equivalent to that of other anti-competitive practices, usually subject to the rule of reason.
The Supreme Court unanimously overturned the ruling of the Tribunal, noting that the submitted evidence was totally inadequate to prove the existence of a collusive agreement aimed at the failure of the bidding. Moreover the fact that the auction ended up awarding contracts with the reservation price fixed by the buyer, would be a sign rather of success than of failure. In its ruling, the highest Court also pointed out that participant firms hired experts for the purpose of wining the contracts, fact that would be inconsistent with the hypothesis of the boycott.
There are several relevant facts to remark in this case. With regard to the conduct of the firms in the tender and its relationship to the collusive hypothesis, it must be mentioned that in a bidding mechanism with multiple stages, it is normal that companies do not reveal all the information in the early stages of the process. We should expect that the revelation occurs as the other participants are also deploying their strategies. If the auction was designed in a way that in the first stage companies show their lowest willingness to provide the service, one must wonder what is the purpose of having multiples stages in the tender. Even if the auction is designed such that the best strategy from the point of view of each firm individually –the subgame perfect equilibrium in technical terms- is to reveal full information in the first offer, it is also possible that participants prefer to wait if they have conjecture that the rest will do the same, as evidenced by results in experimental games.
Secondly, it is true that is costly for companies to make transparent the market due to the high prices charged earlier to other customers, if the tender leads to lower prices. However, this cost of being aggressive in competing for winning the tender exists both at individual and at collective level. In other words, there is no need to collude in order to have incentives to offer high prices at the auction. This is so because individually and without coordination, each firm loses if offers a very low price in the tender as it involves returning money to those clients previously supplied at higher prices. As explained by the article of Cooper (1986), the existence of most favored customer clauses allows firms to sustain higher prices than the competitive level without need to have collusion among competitors.
Finally, invoking the success of the tender from the point of view of buyers, as evidence against the hypothesis of collusion is questionable. The fact that companies have accepted the reference price set by the buyer does not necessarily imply success. For instance, it is plausible that without the alleged collusion, the companies have offered lower prices, as noted by the court’s ruling. Note that the reservation price was kept secretly by the principal, and it was intended to be used only in the case that bids were above that value. If the buyer were confident that this price was the competitive one, and there was almost no possibility of having better offers, then there is no point in keeping the price secret.
In 2005, the competition agency –FNE- accused five health insurance companies-known in Chile under the name of ISAPRES-to collude in order to phase out their plans for maximum coverage, known as 100/80 and replacing them with others of lower coverage called 90/70. This replacement of plans that occurred between 2002 and 2003 was not followed by a reduction in prices, despite the lower level of coverage that was offered to users. Also, in the same period, the profits of the allegedly colluding companies increased significantly, some of them even being 200% higher in comparison with the previous period. Another evidence of concerted behavior, submitted by the FNE, was the parallel reduction in sales effort by the companies during the period where the replacement occurred.
Accused firms noted that the change in health plans is explained by the increase in costs due to the moral hazard problem inherent in the health market, which was exacerbated by having one hundred percent coverage in some services. The insurers also argued that the scenario of regulatory uncertainty that supposedly existed at that time, had contributed to the replacement of the plans. Additionally, firms invoked the theory of parallelism of strategies that is intrinsic to competitive markets, where there is a constant monitoring and copying of strategies between companies. They also pointed out that the high turnover of clients among companies was contradictory with a hypothesis of collusion. Another line of defense employed by the firms was to discredit the argument that they collectively had market power. They mentioned that the industry had no barriers to entry and the relevant market was a broader, which includes also the public health insurer.
The TDLC in a narrow verdict, three against two, rejected the accusation. While acknowledging that there was a parallel behavior between the required companies-both in the timing and type of the replacement plans and in the reduction of the intensity of competition through the sales effort-did not interpreted such parallelism as a product of a concerted action. According to the majority vote, this parallel behavior can be consistent with both hypotheses the collusive and the competitive. Thus, the ambiguity of the evidence would not be sufficient evidence to condemn the firms.
The minority vote, on the other hand, said that given the facts and information of the case, the only plausible explanation to the observed behavior would be the collusion among the accused, either explicit or tacit. The dissident judges mentioned that the majority vote does not present an explanation that makes consistent the facts credited in the case with a competitive and independent conduct from the part of companies.  For instance what was the fundamental parameter of the market that motivated the change on behavior? Assuming that the change in fundamental was real: Why only some companies modified the plan while others do not?
The ruling of the TDLC was appealed by the FNE to the Supreme Court, which upheld the decision of the Tribunal, again in a divided verdict, three against two. The highest court dismissed the evidence of collusion because it does not allow concluding unequivocally that the behavior was the product of concerted action between the companies. According to the ruling “runs around potential uncertain that could lead one way or another, which makes it necessary to dismiss their foundations and carries this Court to leave established with clarity and certainty, that the evidence added to the process are insufficient to prove the existence of collusion.” Later, the verdict states that “.. thus appears clearly that the proofs that are intended to demonstrate the existence of collusion in question is not direct because the fact referred to can be explained for a variety of reasons. ”
The majority vote endorses the explanation of competitive parallel behavior, presented by the defense, noting that the participants in the market are subject to intense mutual surveillance with the aim of preventing the advantages that some of them may have. On the contrary, the vote of minority believes that the use of logical reasoning, known as healthy criticism, leads to logically deduce that there was collusion. The coincidence in the conduct of business, therefore, could not be explained by individual action motivated by some exogenous change, such as the regulatory uncertainty in the sector at that time.
The validity of the use of so-called indirect or circumstantial evidence is expressly mentioned in the rulings. In particular, the dissenting vote of TDLC mentions that given the practical impossibility of having direct evidence -because there are no powers to seize documents or raid offices in search of proofs – the evidence of conduct constitutes the element used to decide on accusations of collusion, despite its inherent ambiguity.
6. Conclusions and Final Remarks
The three cases on prosecution for collusive practices clearly illustrate the inherent difficulty in punishing firms based only on circumstantial evidence. Even if one refines the economic analysis, there are always areas of doubt or some positive probability that the observed behavior has been the result of individual action. The divergent judgments both among courts and inside them, is an expression of the importance of subjectivity in the evaluation of behavioral evidence.
The Supreme Court in its three pronouncements has been skeptical about arguments claiming that the observed parallel behavior can be explained only by concerted action of the companies. In two judgments -Navieras and Isapres- the Court notes that the parallel conduct would also have a competitive explanation. Thus, has rejected the allegations on the grounds that the behavioral evidence is not sufficiently clear to apply sanctions, even in cases as the ISAPRES, where there is a sound analysis about why the observed conduct of the defendant is unlikely to be the outcome of individual actions. Such ruling put a high standard of proof for the indirect evidence in the sense that the probability of the mistakenly condemn must be very low (type I error).
The possibility of having physical evidence would dispel doubts beyond what is reasonable as to whether or not there was collusion between companies. The legal reforms being currently implementing, such as the introduction of Leniency Programs and the power of obtaining direct evidence, will improve the quality of the evidence and the decisions. It consequently will remove any doubts and likely errors in the decisions taken by courts, in case the agency gets material evidence for a case.
It remains an open question the interpretation given to law about what are the conditions to be met in order to condemn firms for collusion. So far the discussion has been focused on the quality of the available evidence. If the amendments to the law allow the FNE to obtain direct evidence, and the agency submit this material evidence in a case. Would also be required to prove that there has been competitive damage to impose sanctions? The TDLC in its last verdicts has evolved into consider collusion as anticompetitive per se as long as the involved firms have collectively market power. Nevertheless, we do not know whether the Supreme Court will take the same approach as the TDLC, given the ambiguity of the law on the legal status of collusion.