Transcript How To Be Loyal, Rich And Have Fun Too

WEHIA KYOTO – 5 / 27-29 / 2004
Interaction
routines and
opportunistic
behaviours in an
artificial market.
FFV wholesale market: field
observation and model
Juliette Rouchier GREQAM - Marseille
1
General thematic and methods
To understand the organisation of supply dynamics for a market:
– Indiv.: which information do individuals need to organise their activity?
– Global: what shape does competition actually take?
• Are there market powers / coalitions?
• How do prices evolve in time?
• How do offer and demand evolve in time with market events?
The market we are dealing with:
– FFV – Fresh Fruits and vegetables
– Professional market (“MIN”)
Two tools that interact:
– Field observation and interviews with stakeholders and actors (the demand
originates in the market itself) – capture of the concrete market (Callon,
2004)
– Simulations to test assumptions about this system
2
« les Arnavaux »
• Marseille wholesale fruits and vegetables market.
– Outside the city, near the port, the airport, the motorway.
– Open from 3:30 until 8:30 am.
– 3 bars, 2 restaurants, a bank, tobacco retails
• Official rules
– Pair wise interactions
– “Mercuriales”: previous days’ average values for products
– Access for two types of sellers (local producers and wholesale sellers)
and several types of buyers (retailers, restaurant owners, collective
restaurants, supermarkets)
• Structural: position of the market
– There are other markets in the area, « producer markets ». This is the
one with imported goods (winter mainly).
– Formation of prices at the regional scale? (importers, producers,
supermarkets)
3
An interaction-focused study
• Relations between wholesale sellers and retailers
– No written information: negotiation > transaction – individual offers
– Buyers don’t make any prior offer, they ask for the price or declare a
quantity they want (Kirman)
– One should not interfere with others’ transactions
– Also: explicit information exchange
– Diverse services: priority on some goods, advises to anticipate on future
prices/ lacks
– Credit
• Wholesale sellers
– The wholesale sellers have networks of sellers
– Each have an “ecological niche” known by all – have a network of suppliers
• Retailers
– Two main types of behaviour (schematic): regular / faithful buyers and
opportunistic ones
– (Change of behaviours is apparently linked to age)
« Network vs market » / opportunism vs fidelity
4
Aspects emphasised in this work
Importance of the transactions for supply renewal
Reasons for fidelity:
interesting to get info and good prices
“character”
ethical speech
norms of behaviours, expectations
Some literature, and especially in simulation but with different
emphasis (Kirman et Vriend, 2000, 2001), (Brousseau, Codron,
1998) (Galtier et al., 2002) (Rouchier et al, 2001)
>> Economic function to these cohabitation of heterogeneous
behaviours?
Assumption: garbage reduction and possibility to get needed
goods
5
Real life observations: wholesale sellers
Numerous aspects to a relation for price setting
–
–
–
–
–
Regularity
Loyalty
Quantity
Feelings
Quality / age of the good
Variation of 5 to 50% off the « basic price »
Expectations after a good deal offer: carry on buying
Signs of good relation and trust
– Credit
– Information about prices for the next day
– Keeps some goods that are ordered by telephone
6
Real life observations: Retailers
Basic constraints (variable):
– List of goods to acquire (season and past sells)
– Limited time (employees)
– Type / quality of goods (situation of the shop)
Knowledge that determine choices
– « les mercuriales » and producers
– Ask wholesale sellers while negotiating
– Ask other retailers « at the bar » or friendly wholesale sellers
Search habits
–
–
–
–
First go to producers’ area
« Favourite » wholesale seller
Extensive search for some goods
Stick to the list or adapt to offer (time and space)
7
Model: wholesale sellers
10 wholesale sellers - 10 products
Supply:
Each one has a list of goods and prices
Each one has a « normal supply »: buy at the beginning of the day.
Probability to get goods: parameter of the simulations
Variation of price among sellers and in time: 20%
If unable to supply: increase normal supply ; if throw away: decrease normal supply
Normal supply product 1
(date, price) (date, price)…
Average price product 1
…
Normal supply product 10
(date, price)
Average price product 10
Goods aging
Price reduction of 20% for 3 days old product, 40% for 4 days old
Thrown away after 4 days
Transaction:
Gives cheapest products to regular / or give younger products to regular: parameter
10 % reduction for more than 3 products, and 10 % for regular
Regular client bought more than 10 products in the last 10 market days
8
Model: retailers
100 retailers – ratio of Loyal / Opportunistic as a parameter of the simulations
List of goods to acquire, come everyday
Time is structured in three time-steps
Time is used for: transacting or information search
Fixed attitude => a strategy to use time
Behaviour <= attitude + gathered information
Information = average price for all products for 5 wholesale seller
Request = list of goods, one unit per type of good
Different attitude induces:
Loyal
Time use
Choice process
buy – info – buy
First regular, then as few demands as
possible
Opportunistic info – buy – buy
Cheapest possible basket (up to 5
demands)
9
Simulations - Observation
• Simulations defined by:
– Repartition of retailers attitude: 50-50 / 100-0 / 0-100
– “Answer simulation”
• The wholesale seller gives cheaper price to regular
• The wholesale seller gives freshest products to regular
– Already led: “Price” and “Supply” simulations
• Supply probabilities = [40;60], [70-90], [80-100]
• Price variation: 5- 20 or 40 %
• Observation (global – average)
•
•
•
•
•
Number of thrown units
Quantity of missing products
Evolution of normal supply – average and msd
Average age for bought products
Average price depending on the type of retailer
10
11
average quantity of
garbage
simulations 20 / 60-80
16
14
12
10
8
6
4
2
0
0
20
40
60
80
100
120
number of loyal agents
12
missing products at the
end of the day
missing products for retailers
25
20
15
10
5
0
0
50
100
150
num ber of loyal
average lack
msd lack
13
normal supply simul 20/ 60-80
normal supply
1200
1000
800
600
400
200
0
0
20
40
60
80
100
120
number of loyal
14
average prices with best price for loyal
15000
10000
5000
49
45
41
37
33
29
25
21
17
13
9
5
0
1
price per unit
20000
m arket days
loyal
opportunistic
15
16
Results
• Supply and waste: retailers get more goods and wholesale
sellers throw less when there are more loyal
• This increase in waste is mainly due to wholesale agents
overstocking to attempt to satisfy sales that are subsequently
not made – since in previous days selfish agents ask several
wholesale agents to supplier various products. Importance of
loyal in the regulation of market
• If wholesale sellers give “best products” to regular, loyal pay
more and get fresher products and otherwise pay equivalent
• Better to use the “freshest products” logic because it is closer
to other researchers’ results
17
And then….
• With this model
– Test a higher number of ratio loyal / opportunistic
– Test more price variability
• New model
– Describe new procedure of supply buying: have a global quantity instead
and get it by queuing model
– Find a better way to describe the wholesale sellers’ answer
– Have a better description of reaction to prices by adding memory for
opportunists who don’t need to gather all information before choosing
– Have faithful agents evaluate their regular seller and change their habits
• New interviews to increase knowledge about
– Add price evolution of prices along the day (S. Moulet – A. Kirman)
– Model where wholesale sellers have more adaptive price policy with agents
(not just regular vs unknown) – hard to select the “most relevant” indicators
for agents
– Importance of the Global price knowledge (O. Chappuis)
18