Prsentation PowerPoint

1
Raluca PARVULESCU
Outline

• 1. An institutional perspective
• Experimental design
• Main result convergence
• "Clumsy" behavior
• Simulations
• Final remarks

EXPERIMENTAL AGENTS CLUMSINESS, THE MAIN REASON
WHY MARKETS CONVERGE TOWARDS AN INEXISTENT
EQUILIBRIUM
2
Main feature prices are decided by the market
operators themselves
1. An institutional perspective 1.1 Market
taxonomy 1.2 Price convergence issue
1.3 What predictions ?

• Double auction markets

• Posted-offer markets buyers "price takers"

• Production "to order"

• Production "in advance"

One important question Price convergence (or
its absence) towards the competitive issue
3

• D A markets
• prices converge towards the "competitive" level
  ( Nash equilibrium)

1. Theoretical predictions 1.1 Market
taxonomy 1.2 Price convergence issue
1.3 What predictions ?

• P - O markets with production "to order"
• prices converge towards the "competitive"
  level, be it a Nash equilibrium or not

• P O markets with production "in advance"
• in theory, there is no Nash equilibrium, so no
  convergence ?
• Alger 1979
• Friedman 1988

4
Does the Posted-Offer market with "advance
production" converge towards the "competitive"
issue, which is not a Nash equilibrium ?
1. Theoretical predictions 1.1 Market
taxonomy 1.2 Price convergence issue
1.3 What predictions ?
If this convergence is observed, what are the
forces accounting for it ?

• agents "clumsy" behavior observed in the
  experiments
• simulations

5

• Supply
• Every subject was a strawberry producer
• Inventories or carryover of unfilled orders from
  buyers are not permitted
• At the beginning of each period a price and
  output decision
• Decision support
• - table with all types of costs (total, average
  and marginal)
• - subjects own previous results (sold output
  and realised profit)
• - all posted prices on the market during the
  previous periods

2. Experimental design 2.1 General remarks
2.2 Specific details

• Demand
• computer simulated
• stable all along the experiment

6

• 2 experiments - 9 participants in each session
• - feb- april 2006
• - 2h30 on average
• - 15 / person, on
  average

2. Experimental design 2.1 General remarks
2.2 Specific details

• 2 phases - agents are alone on the market
• - agents play against each
  others on a competitive market

• Theoretical benchmarks
• - the "competitive" issue price 1.3 output
  200
• - the "monopoly/cartel" issue price 2
  output 85

7
Fig.1 Average price convergence
3. Main result convergence
8
Fig 2. Prices variation coefficient
3. Main result convergence
9
Fig 3. Evolution of the rationing phenomenon
4. "Clumsy" behavior
4.1 Criteria for "clumsiness "
4.1.1 Market adaptation
4.1.2 Overproduction 4.1.3 Profit
increase 4.2 Market response
10
Table 1. Impact of an incurred rationing (or its
absence) on a the posted price at the next round
Rationing phenomenon intensity
Decrease
Maintaining
Increase

• "Clumsy " behavior
• 4.1 Criteria for "clumsiness "
• 4.1.1 Market adaptation
• 4.1.2 Overproduction
• 4.1.3 Profit increase
• 4.2 Market response

Very important
91
4
5



(23 observations)


Significant
87
10
3
(32 observations)
Null or negligible
37
21
42
(289 observations)

• 50 of the decisions are "reasonable"
• 32 of the decisions are "unreasonable"

• Subjects are very heterogeneous

11

• Overproduction

• 28 of the decisions

4. "Clumsy " behavior 4.1 Criteria for
"clumsiness " 4.1.1 Market adaptation
4.1.2 Overproduction 4.1.3 Profit
increase 4.2 Market response
Fig 4. Evolution of the indicator of average
overproduction ( )
12

• Overproduction 2 kinds of effects
• Direct effect
• agents who disrespect this rule obtain
  smaller profits (44)
• Indirect effect
• agents who respect this rule obtain smaller
  profits too (40)

4. "Clumsy " behavior 4.1 Criteria for
"clumsiness " 4.1.1 Market adaptation
4.1.2 Overproduction 4.1.3 Profit
increase 4.2 Market response
Rationing phenomenon 18 less in the first
session
26 less in the second session
Agents are very heterogenous
13
Compare the policy adopted with a reference
policy

• when the subject sold the entire produced

maintain the same policy
4. "Clumsy" behavior 4.1 Criteria for
"clumsiness " 4.1.1 Market
adaptation 4.1.2 Overproduction 4.1.3
Profit increase 4.2 Market response

• when the subject got rationed

post the same price and reduce output to the
sales level
Ex ante and ex post comparaisons of respective
profits
Table 2. Only not "irrational" decisions
considered
Table.3 All decisions considered
14
Examine agents behavior impact on the market
2 individual indicators
4. "Clumsy " behavior 4.1 Criteria for
"clumsiness  " 4.2 Market response

• the proportion of reasonable decisions from a
  market adaptation criterion
• the proportion of reasonable decisions from a
  profit increase criterion (takes implicitly into
  consideration the tendency to overproduce)

Correlation coefficient 41
Regression equation
Av. P 33.4 31.53 Capacity to - 97.53
Capacity to adapt
increase P to the
market (t4.33) (t2.11)
(t-6.4)

15
2 types of agents

• "rational" type "best-response " strategy

5. Simulations

• "clumsy" agent inspired by the human subjects

5.1 Decision rules
5.2 Results
16
Fig 5. Average sales price evolution on a market
with 9 "clumsy" players
5. Simulations 5.1 Decision rules 5.2
Results
17
Fig 6. Average sales price evolution on a market
with 9 players adopting a "best response"
strategy
5. Simulations 5.1 Decision rules 5.2
Results
18
Fig 7. Average sales price evolution on a market
with 3 players adopting a "best response"
strategy and 6 "clumsy" players
5. Simulations 5.1 Decision rules 5.2
Results
19

• Average prices seem to be converging towards the
  competitve issue in the two sessions carried out

• Agents behavior seems to be far from the
  "rationality" canons

6. Final remarks

• When the number of automates adopting a "clumsy"
  behavior is predominant, markets stabilize.