Do Governmental Economic Institutions help to foster tacit collusion ?

Everything is in Hidden.

2023-07-10

2023-07-14

Our focus is on the results of the 5th supergame (in the instructions, subjects see the 5th and the 6th cycle) where the intervention occurs during this supergame. Only the data of the 5th supergame is to interest for us. The first 4 supergames are to be considered as trial periods (in order that subjects gain experience in the game and in belief elicitation).

BELIEFS: belief about cooperation of firms' competitors just before and after the statement
PRICES: individual, market prices and average prices at the matching group level just before and after the statement
DEGREE OF PROFITABILITY: degree of profitability of a market and average degree of profitability at the matching group level just before and after the statement

BELIEFS: we will use three differents measures of the belief of cooperation q (which is between 0 and 1) of the firms' competitors:
1.beliefs about the highest possible prices (which offers a very conservative measure of the belief of cooperation about competitors)
2.beliefs about higher or equal prices set by competitors at the actual period compared to last period's market price
3. All prices set above the Nash Equilibrium (common measure of collusion in the literature). This will be our main measure to analyze the changes in beliefs.

Each firm will give its estimation about the action of each competitor separately. We elicit the whole distribution of possible prices. To comply to our model that assumes that the belief q about its two competitors is the same, we will take the minimum, the maximum and the average of the two elicited beliefs. Finally, we will compute the change of beliefs between the first period after intervention and the last period before intervention. After this period, we will analyze the beliefs during the end of the 5th supergame given the intervention.

PRICES: Since the set of possible prices changes (due to the cost shock and the related change in reservation price), we will use a Diff-and-Diff regression to measure the impact of each treatment separately. Our main analysis will be on individual submitted prices with Block bootstrapped standard errors in brackets, where the blocks are the matching groups. Bootstrap with 9999 repetitions
DEGREE OF PROFITABILITY: The degree of profitability is a standardized measure of the (gross) mark-up of a firm. It allows us to compare the changes before and after the treatment directly. We will perform regressions correcting for multiple hypotheses (Roman-Wolf correction) panel data regressions.

risk preferences, attitudes towards strategic uncertainty, other standard control variables.

risk preferences might mediate the relationship between actions and beliefs. Attitudes towards strategic uncertainty seems to be linked with risky choices implying strategic interactions (Bruttel et al. 2022).

Everything is to be found in Hidden.

To answer our research question, we need three elements : an oligopoly market (we use Bertrand homogeneous good market with perfect information), we need to elicit beliefs (to measure the level of strategic uncertainty) and we need a cost shock followed by a stylized governmental statement (The intervention).

There are three parts in our main experiment :
1) An oligopoly game without belief elicitation (3 supergames). Firms have to submit integer prices between 11 and 15. (marginal cost 10)
2) An oligopoly game with belief elicitation about competitors' actions (4th supergame + begin of the 5th supergame). No other changes.
3) Announcement of the increase of the marginal cost (and reservation cost) as well as the applicable statement given the treatment and then continuation and end of the 5th supergame. Now Firms have to submit integer prices between 21 and 30. (marginal cost 20).

Between each supergame there is rematching. Supergame termination is random and we use a combination of the Block Random Termination with first a block of 8 periods and the random termination. If the match did not end within the block, then the supergame continues until the match end and subjects learn whether the match ends at the end of each period after the 8th.

For our analysis of the data, we use only the data of our 5th supergame because it is where our intervention occurs and to be able to measure the effect of a statement on tacit collusion, we need the same firms to interact in the same market before and after the intervention.

Our Hypotheses :
[For the first perod after intervention]
1. Beliefs of competitors' cooperation is higher in PESSIMISTIC (P) THAN in OPTIMISTIC (O) AND IN BASELINE (B) (P > O = B) in the first period after the statement disclosure.
2. Everything else being fixed (for the same level of beliefs, same level of degree of profitability before the intervention, etc.), higher prices are submitted in PESSIMISTIC and OPTIMISTIC than in BASELINE (P = O > B).
These two mechanisms, leads to the following ordering of the average submitted prices and average degree of profitability across treatments :
P>O>B.

Explanation:
The statement in P increases both the own belief of competitors' cooperation and for fixed level of beliefs make it more likely to cooperate than without statement.
The statement in O does not increase the own belief of competitors' cooperation but for fixed levels of beliefs, the statement make it more likely to cooperate than without statement.

[For all the periods after the intervention]
3. Beliefs of competitors' cooperation is higher in PESSIMISTIC (P) THAN in OPTIMISTIC which is in turn higher than in BASELINE (B) (P > O > B).
Explanation:
3. On average, the P condition yields higher beliefs in other firms' cooperation since the first period than in the other conditions and the higher prices submitted since the first period make it most likely to get higher average beliefs of competitors' cooperation than in the two other conditions. In O, the higher level of submitted prices than in B in the first period when everything is hold fix will push the average belief of competitors' cooperation higher than in B since higher prices have been submitted.

Remark: Two mechanisms are explained above: a pessimistic statement reduces the strategic uncertainty and correlate actions among firms (i.e. Firm 1 thinks that the Firm 2 and Firm 3 will cooperate with probability q at the same time), while the optimistic statement only correlate actions among firms. We are aware that only the first channel (reduction of strategic uncertainty) might be observed empirically. We think we might be able to explain the two following potential results (P>O>B) or (P>O=B) quite well with our model.

Computer in a lab

matching group

Yes

90 markets and 45 matching groups.

270 subjects

90 subjects per treatment.