The Rise and (Potential) Fall of U.S. Cartel Enforcement

Cartels are a group of two or more producers that conspire to fix prices instead of competing with each other, resulting in consumer harm. Consumers can be either individuals buying products or services, or firms in the marketplace who are buying inputs from downstream suppliers. For example, if two suppliers of automotive components were to engage in price-fixing, companies such as Toyota Motor Corporation would have to pay higher prices for those components, raising the cost of the cars they sell to you and me. The economic harm caused by price fixing can be passed to several levels.

The idea of forming a cartel is not new. The history of cartels is complicated in that they haven’t always been considered bad. For example, cartels were once legal in many countries in Europe.

That said, as early as the beginning of the 1900s, the U.S. Supreme Court argued that cartels are illegal and caused supreme harm to consumers. Fines were created to discourage the formation of cartels.

However, as of 1955, the maximum fine was only $50,000 per cartel, regardless of how much the cartel raised prices or how wide the collusion was. There were many companies that paid fines repeatedly in this period; for example, General Electric Company (GE) and Westinghouse Electric Corporation were caught many times price fixing for the same products, because every time they did so, they gained profits. The fine was too small.

In the 1970s, there were changes in penalties: cartel regulation violations became a felony with a maximum prison term of one year, and the maximum fine went from $50,000 to $1 million.

The most significant U.S. innovation coming in 1978 was the "leniency program," which allowed companies participating in cartels to partially or wholly avoid fines and other penalties by providing the government with information about the cartel. There are now leniency programs in many countries, including Japan and the countries belonging to the European Commission (EC).

That first U.S. program did not work very well. Many felt that the incentives were not set up properly or transparent. A major revision was undertaken in 1993 with the creation of the new leniency program. Over time, the United States has substantially increased penalties and improved the enforcement of cartel regulations.

Alongside regulatory changes, over the history of the United States, we have seen a major shift in the way that people view cartel conduct and crimes.

At one time, scholars like George Stigler, Gary Becker, and others talked about antitrust in terms of Oliver Williamson's transaction cost efficiencies—by which I mean the idea that mergers and agreements, such as vertical mergers, can be beneficial to firms in terms of costs. For example, Toyota can either buy brake parts from a company upstream, or it can buy that company and make its own brake parts at a lower cost. There was a time when U.S. anti-trust laws viewed this kind of upstream-downstream vertical merger negatively. However, in Williamson's Nobel Prize-winning work, he argued that there are sound economic reasons why a company like Toyota should be allowed to have a vertical contract which resembles a vertical merger.

With cartels, on the other hand, although these agreements might sometimes resemble this kind of contract, there is really no justification as to why two companies should get together and raise prices.

The big law and economics thinkers in this area, such as Robert Bork, Richard Posner, among others, started to discuss how antitrust enforcement was sub-optimal: the laws were too rigid in the non-cartel area to allow for efficiency-enhancing mergers, and appeared too easy on companies engaging in cartel behavior.

Cartel enforcement is potentially influenced by two issues: policy innovations and politics.

Over time, we have seen policy innovations in terms of new amendments to cartel regulations, the new leniency program, and enhanced monetary fines and jail terms.

These innovations have overlapped with certain presidential administrations, and so, as a policymaker or researcher, it is hard to know what proportion of the cartel prosecutions were driven by policy or by political preferences.

Looking at trends in the total number of cartels prosecuted, we see a sharp change in prosecutions when Ronald Reagan became president in January 1981. The average number of cartel cases prosecuted prior to Reagan was about 20 cartels per year. This number jumped to close to 75 or 80 per year in the Reagan period. Over the eras of Bill Clinton, George W. Bush, and Barack Obama, we see some interesting shifts, with a low point in terms of the number of cartel prosecutions during the George W. Bush administration.

Looking at fines as well, we again see big differences. In the pre-Reagan period, fines were small and effectively not considered to be deterrents. Fines went up at the time of the Lysine cartel case in 1996 and 1997. This was the first big multinational cartel case prosecuted after the enactment of higher fines, longer jail terms, and the new leniency program. This prosecution was a test ground for what the new cartel regime would look like.

From the late-1980s to the mid-1990s, the number of cartels prosecuted rose steadily. However, it is not until 1996 that the total amount of fines increased substantially. The increase in the number of cartel cases was a result of a Reagan-era policy of prosecuting bid-rigging cases at the federal level, so even bid-rigging on a highway contract given out by a state government would be prosecuted. That is why we see a lot of prosecutions while still having relatively low fines in the Reagan era.

There have been considerable increases in the fines levied on individuals, as well as the number of people receiving jail time for cartel regulation violations. In the United States, executives of firms engaged in forming and maintaining cartels are convicted and get jail sentences. In recent years, the average number of days in jail per individual (executives of firms) prosecuted has reached around 550 to 600 days.

Many interesting patterns can be seen in U.S. cartel prosecutions. The total number prosecuted was low early on, went up for a while, and is again now quite low. However, fines—both fines per corporation and fines per individual—have increased dramatically. Finally, the number of jailed days per individual has gone up. Fewer cartels are being detected and prosecuted in recent years, but the penalties are higher. That said, I do not want you to think that U.S. prosecution rates are low. In recent years, the United States has still prosecuted an average of 50 cartels per year, which, in comparison, is a big difference from the EC, which prosecutes only about six cartels a year.

What is interesting in looking at these trends is in answering the question of whether policies or politics have had more of an impact on cartels. The problem is that policy periods have big overlaps with the presidential regimes.

There are two ways to deal with this issue. One would be to model the impact of policies and politics on cartel cases using linear regression. For this model, we could create a dummy variable for presidential administrations, with 0 assigned to Democratic presidents and 1 assigned to Republican presidents.

What happened between the 1950s and the present? Fundamentally, recent enforcement efforts and policies seem to be targeting catching fewer yet bigger fish. The question intellectually and policy-wise is "What is the deterrence effect of that?" What we really care about is whether the policy instruments are leading to less or more price fixing.

If we look at the political impact (a basic Democrat versus Republican difference) on cases, we find no effect. That is, the dummy variable is statistically insignificant in the model. The problem is that there in fact could be quite significant differences between presidents within the same party. Nobody would say that Clinton was the same type of Democratic president as Obama was—many view Clinton to be more pro-market compared to Obama. Republican presidents also vary in terms of their propensity to intervene in the markets and regulations.

If we remove the Democrat versus Republican presidential regime dummy variable and re-estimate the model, we can see the policy effects. Our results show an overall case average of 45 cartels per year. On the overall enforcement data on cartel prosecutions and penalties, we superimpose the mean levels of these variables for each presidential administration. This shows that George W. Bush had the lowest number of cartels prosecuted in recent history, while the Obama mean was about the same as the Clinton mean. There is something going on which can be attributed to the presidential regime, even after controlling for policy effects.

We see a different picture when it comes to penalties. There is no difference in the mean penalties between Obama and George W. Bush; both are very high, even though there are major differences between those periods in terms of actual prosecutions. For policymakers and researchers, it would seem that under George W. Bush, the government went after the really big fish, not worrying about the small ones, while under Obama, they continued going after the big fish and also went after the smaller ones. In terms of jail days, the same patterns can be seen under George W. Bush, Clinton, and Obama. The sentencing guidelines have largely taken away the discretionary power that judges had to give a stricter or softer penalty, and created a clearer template for fines and jail terms.

There seem to have been three broad stages of cartel enforcement over the history of the United States, starting from stage one (low prosecutions and low penalties), going to stage two (higher prosecutions, low penalties), and ending in stage three (fewer cartels prosecuted, but very high penalties).

This concept presents a big policy question for researchers. If the government's resources are targeted towards catching the big fish, what is happening with smaller domestic cartels? Suppose the government decided not to go after 100 smaller domestic cartels creating a total of $100 million in economic harm because that would take a lot of time and energy, and instead went after one cartel creating total economic harm in the area of $100 million? Given what the government is signaling with that kind of behavior, the domestic cartels might say, "The government is employed in catching these big fish and has no time to look at what we are doing," which might actually incentivize domestic cartels and lead to higher prices and welfare loss. Because cartels are covert, we really do not have a sense of whether the net effect of this kind of policy shift is beneficial or not.

The cartel prosecution trends in Europe provide an interesting contrast to the U.S. data. From 1954 to the present; the number of prosecutions per year in Europe was on average about 5.5 to 6, while the equivalent U.S. number was close to 50. European Union (EU) member states sometimes do their own cartel enforcement, but this number is typically small. If EU member states were to be included, then we would also have to include all cartel and bid-rigging cases prosecuted by U.S. states, and this number can be quite high.

In terms of fines, we do see a broadly similar pattern to the United States. Fines are low for a very long period of time, and then go up. As in the United States, EU fines went up after the Lysine case. Currently, EU fines are higher per cartel, but keep in mind that they prosecute only about six per year.

It is hard to compare the deterrent effect of policies between the United States and the EU. A big difference is that while the United States will send CEOs and executives to jail, the EC does not.

Analyses of regulations in the international arena highlight the importance of questioning what really impacts total deterrence: fines or fines plus jail terms? Once regulations become stricter, companies start to find more complicated ways to cheat, such as the use of big data, analytics, etc. to restrict output and raise prices. There is some evidence that financial markets in London or Zurich are hiding behind massive data and complicated algorithms to fix prices. If we are catching fewer cartels but the crooks are getting smarter, we do not know whether regulators are staying on top of this issue or not.

Cartels can be very complex. Overall, we really do not know the net effect of having penalties going up—are we really seeing the creation of fewer cartels? That question is almost impossible to answer.

Q1. The number of individuals and corporations fined has gone down, and while the number of cartels prosecuted on average has changed from administration to administration, there seems to be a steady decline in that average. Does that mean that cartels are getting smaller, or that the prosecution picks and chooses which cartels to prosecute, or who to go after within a cartel?

Vivek GHOSAL
In recent years, there has been a decline in the number of individuals prosecuted, and this is broadly consistent with the fact that the number of cartels that are prosecuted has declined. If fewer cartels are prosecuted, there would typically be fewer corporations prosecuted in the aggregate. Another factor is the number of firms that are fined when cartels are caught. There must be at least two, but the Lysine cartel, for example, spanned many companies. This is a challenging question, because with a single cartel involving two firms with a big global footprint, for example, Apple Inc., their market share might be massive, meaning huge fines. Or a cartel might involve six small companies, but with a smaller economic impact.

Q2. Do you think there has been any notable change in strategy in terms of whether regulators just go after leaders when they catch a cartel?

Vivek GHOSAL
If two firms are colluding and one files for leniency, they will be asked who else is in the cartel with them, and if they lie, they can face larger penalties. When a cartel is caught, basically everyone in that cartel is caught.

Q3. How did the second set of leniency rules make it easier to go after everyone in a cartel? How has the complexity of cartels increased over time?

Vivek GHOSAL
Leniency is complicated. Think of two people who have had a very good relationship who go out for dinner every week and talk about their families, but might also discuss increasing prices and have been doing this for years. The government has changed the rules so that the first person who comes to the government and confesses will benefit from leniency. In the 1978, with the initial leniency program, it was not very clear how protected the first person to confess would be, and other incentives to confess to colluding. The 1993 revision lays out clearly what protections the first person can expect provided he/she is not the leader of the cartel. If someone is leader of the cartel, then no concessions are offered.

Has the complexity of cartels increased since the 1980s? That is a difficult question to answer. A complex cartel might not involve 10 firms globally; or it could be just two firms that collude in a complex way, such as by using big data algorithms.

Q4. You mentioned the potential of using information from non-cartel investigations for cartel cases. Could you talk more about that?

Vivek GHOSAL
Here is an example. The United States in 1997-1998 deregulated the electricity market to allow companies to compete. We were evaluating a merger of two big electric companies, and, in the data, we started to see some strange patterns in terms of the price. As the electricity flows from state to state, you can observe the minute-by-minute prices in the energy exchanges, and we saw suspicious movements of prices and other behaviors. If I generate electricity, this means that the supply of electrons I put onto the grid has to match demand identically, otherwise the system is going to collapse (as historically there was no storage). What we know is that some regions such as New York/Northeast region had little or no excess capacity. If a generator reduces their generation while demand remains the same, they would create a supply shortage, and the prices buyers would need to pay would go up. We observed all of these changes in generator output correlated with transmission bottlenecks, and then we considered which companies should be allowed to merge. We saw strange patterns which pointed to the possibility of collusion. That is an example of a non-cartel investigation (i.e., a merger) providing evidence and information that companies may be engaging in some form of collusion.

Legally, this is a big problem. In the EC and the United States, it is not possible to mix a non-cartel investigation with a cartel investigation. There is a complete legal separation between cartel investigations (as they are criminal investigations) and merger investigations (as non-criminal investigation). If it is detected in a merger case that companies are colluding, the government must launch a completely separate investigation.

Q5. One hypothesis on the decline in the number of prosecutions in the United States is that stronger deterrence measures have resulted in fewer cartels, except for maybe big cartels, because there is a statutory limit on fines. But you emphasized the discretion of prosecutors, i.e., when the government thinks it is wiser to prosecute big cartels. Which idea is responsible for the decline?

Second, some prominent economists say that because of the strong regulations against cartels, there are now more merger and acquisition activities in the United States. Do you agree?

Vivek GHOSAL
Regarding statutory limits to fines, we cannot impose a fine of $6 trillion because nobody is going to pay that, so there is a realistic limit on fines. If a cartel were to fix prices and receive an extra $1 trillion in profit from that, fines are not going to deter them. The one counterpoint to that is jail terms; in the United States, multiple executives have gone to jail for many years. Apple was accused of anti-competitive conduct (in e-books), and the government embedded monitors in the company. Apple complained about that. I do think it had a deterrence effect.

To me, a big problem is that of "domestic versus international" case prosecution. There is evidence that the United States is focusing less on smaller local cartels because it has shifted its focus onto big international or big domestic cartels. What is the net effect on the cartel issue if 100 domestic cartels do not get prosecuted while one international one does? I do not know. It is a problem.

It is true that if a company cannot do things on the collusion side, it might try to take greater control of the market and competitors in other ways. One option would be through mergers, both vertical and horizontal. If I ask whether an increase in cartels results in more mergers, or whether more mergers brings about an increase in cartel activities, the little evidence we have based on some of my earlier research shows that greater merger and other forms of non-cartel antitrust activity reveals itself in more cartels prosecuted later on, and not the other way round. However, this is an open question, because the evidence is not conclusive.