Thursday, November 29, 2007

Some questions and answers on collusion

Q : Do cross-ownership always trigger collusive behavior ?

A : No.. Not Always. Some articles shows that cross ownership indeed could trigger collusion especially in static setting (Reynold and Snapp (1986) , Farrel and Shapiro (1990)). Yet in dynamics setting, Malueg (2002) found that cross ownership could lead into non-collusive behavior due to the inability of the firms to punish any defection behavior of the firms. (Malueg, 2002).
While that, Gilo, Moshe and Spiegel (2006), using also dynamics setting, found that the collusive outcome could be found in cross ownership under certain parameter vector.

It means that theoritically, there is no conclusive results on relationship between cross ownership and collusive behavior.

Q: If we can not use theory, what are the proofs of collusive behavior in cross ownership?

A : There are two possible forms of collusion. The first one is explicit collusion and another one is the implicit collusion. The main difference between these two collusive forms is the communication strategy and side payment. In explicit collusion, there is usually any forms of communication between the cartel members. Furthermore, they also have side payment as a forms of punishment from collusive agreement. While that, in implicit collusion, they do not have any of these two main forms on interaction.
Forms of collusive behavior determine the ability of Antitrust Authority (AA) to detect collusion. Most of the revealed collusion are the explicit collusion (vitamin collusion,lysine, etc). AA used any proofs of communication such as : meeting notes etc , as the evidence of the collusion. However, in implicit collusion, these can not be used since there is no proof of communication .

Q : Thus, in implicit collusion, is there powerful tools that can be used as signal of collusion, may be price cost margin, variation of price, comovement in prices, etc ?

A : We could discuss each of possible variables :
1. Price cost margin.
Traditionally, this variable is used as the signal of collusion. Yet it is also has a major set back. Collusion certainly means high price cost margin (PCM). But not all high PCM can be get from collusion. PCM is just an evidence of maket power (not more nor less). High PCM could also be the results of eficiency in cost structure. Konings,Cayseele,Wayzynski (2005) empirically found that high price cost margin in eastern europe after privatisation could be lead by cost efficieny ( due to competitive pressure).

2. variation in price.
Traditionally, we assume that low variation in price could be a signal of collusion. Yet, this view is refuted using the found of green and porter. Green and Porter (1984), and research on Joint Executive Commitee (198?), the variation in price could also be higher in collusion.

The intuition of these result is quite clear. If we used a trigger startegy ala Friedman (1972) or stick-and-carrot ala Abreu (1985-1986), price variation is low in collusive phase. But the price variation will high during punishment phase ( it is proven in price fluctuation at Joint Executives Committe). If we combine these two phase (collusion + punishment), the price variation could be higher rather than that of in competitive era.

Moreover, Rotemberg and Saloner (1986) also found that price ( thus its variation) is also depends on the demand fluctuation. If demand is low, the variation price could be low, yet if demand is high , thus there will be a price war, the variation of price can be high. If we combine these two condition, again we get that variation prices could also be higher in collusion. Hence, it means, using low variation of price could not be just used as the signal of collusion.

These two examples tells us that there is no a general variables that can be used as the clear signal of collusive agreement

Q : If that is the case, do we have a consistent method to detect collusive act?

A : As far I am concern, there is no a general method (which is at least consistent) that can be used to detect all implicit collusion. Bajari and Ye ( 2003), Porter and Zona (1993), Banerji and Meenakshi ( 2004),
Bos and Schinkel (2007) have developed some methods to theoritically and empirically detect implicit collusion. However, each method has its own set back (may be good to detect certain form of collusion but not that good to detect another forms of collusion, or the availability of data)

N.B : Congratulation for KPPU. Thank you for the effort.
Despite all controversies, this decision on Tamasek , could be a bad signal to all the cartel in Indonesia. Now, they should know that we have a good AA.


dendi said...

Hi Cauf,
I think what you are saying "No" on the first (A) Answer could be misleading because you clearly mention that the relationship between cross ownership and collusive behavior is not clear cut.

However, you give a very clear brief literature and taxonomy in collusive behaviors. Here, I want to add another problem for the anti-competition authority in making decision on collusive behaviors. The main problem in collusion is how to prove. If the AA can get meeting notes, it's very good because it is a hard proof to convince judges.

But most collusive behaviors are non-cooperative game. What AA can do by making research is only indication and it is hard to fine the firms based on this soft evidence. It may be not convincing because too many variables involved. Common practices after AA find the soft indication of collusive behaviors, AA give policy recommendation to reduce probability collusive behaviors repeatedly.

Back to the Temasek case, based on the information given on the KPPU's website the KPPU has indications that price fixing has been done by Telkomsel and Indosat by looking at increasing market share of both companies. I think it is not easy to prove relation between cross ownership and price fixing. On the other hand, no other variables can be suspected as the cause of price fixing except cross ownership.

In addition, again I have to say that how to reduce barriers to entry is very important. And, the main issue in telecommunication market is inter-connection. Big players often restrict connection from small player to their networks. I think KPPU should be aware on this issue.

The Dreamer said...
This comment has been removed by the author.
The Dreamer said...

Hi den.

Sorry. for the statement. What I meant is cross ownership do not always lead to collusive agreement. Thanks for the correction.

About the cartel. I think I have mentioned that most cartel are in forms of implicit collusion (there is no communication. The cartel only relies on the incentive of each member to stick with cartel agreement). I couldn't agree more with you that it is hard to prove this form of collusion.

Sharp increase of market share. Is it a sign of collusion? I have some comments on this part :
1. Increase in market share again could not be just assume as the signal of collusion per se. Demand, and cost structure, lower price or non-price advantage could also lead into an increase in market share. Again, we need again to prove that this increase of market share is caused by collusion.

2. If this is the case, it means that there are partial cartel (imperfect cartel). some literatures on merger or partial collusion say that in order to make a stable partial cartel, the cartel member should have at least 90% of the market share (see SALANT,SWITZER,REYNOLDS (1982)). This number becomes the rule of thumb of partial cartel. Yet, it is not take into account endowment constraint as in OPEC.

Again,we need a further proof of this.

3. Thanks for the additional points. :)

dendi said...

I agree with you that a further evidence is needed to prove collusion in the Temasek case.

Information given in the KPPU's website, the method how they concluded that collusion exists iis by comparing market share before and after cross-ownership period. The market share is rapidly increased. Yes, of course it could be caused by non-price advantage.

The KPPU also informs that the market share Telkomsel and Indosat is around 90%. It could make collusion easier and stable between them.

I think cross-ownership variable is very suspected in this case. But it is not only Temasek Holding ownership, but also the Government ownership.

Ps: this is link to the KPPU release on Temasek case

The Dreamer said...

Den, related to market share, I just want to ask :

1. do KPPU or you have the data on market share per firm (instead of total)?

2. Do you also have the data of market share per firms just only at place where telkomsel,indosat and pro-XL actually compete?(there are many places in Indonesia where they can only used telkomsel, due to network coverage)

dendi said...

Well... I am not a KPPU Commissioner :-). I quoted information from the KPPU website. Your questions are similar with mine. Especially I ask about how they define and calculate market share. It is a key point.

The information we ask may be available on the full report of the decision. But we have to wait until it is uploaded under "download" link on the website.

Berly said...

Can the two esteemed industrial economist tell me which one is more legitimate for government to step in:
1. Potential of cartel
Concentration ratio, cross holding, etc
2. Act of cartel
Increase price, lower competition.

and which one Temasek is guilty off?

dendi said...

Hi Berly,
You got interesting points. I post a thought based on your points on my blog at

javier said...

Do you think that the information that prieces give to the producers may lead them into collusion.