Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Labour Behaviour, Basic Income and ,Social Influence: A Simulation Experiment
Sandra Gonzalez‐Bailon, University of OxfordJose A. Noguera, Universitat Autònoma de BarcelonaJurgen De Wispelaere, Université de Montréal Jordi Tena‐Sánchez, Universitat Autònoma de Barcelona J ,
Unconditional benefits: Basic IncomeUnconditional benefits: Basic Income
• Unconditional benefits• Basic Income: a cash benefit paid to all citizens on an individual basis irrespective of their income, work record, and disposition to work.
Basic Income and Social Dilemmas
Th i i f BI b th i d• The provision of a BI can be theorised as a collective action problem (Most) individuals are better off with the provision of public good, but fall in the trap of selfish rationality –
b d f h k h ff b deverybody prefers others to make the effort, nobody makes the effort, and the public good is not provided
• The policy relevant question is how to solve this dilemmathis dilemma
Theoretical Approaches to CAppOlson’s approach (Olson 1965)
Private vs public interests
selective incentivesinteraction in small groups
reputation
The sociological approach (Macy 1991, Gould 1993)
the efficacy question and fairness norms
The sociological approach (Macy 1991, Gould 1993)
sequential decisionsheterogeneity of motivations
social influence
Why an Agent‐Based Model?sequential decisions
heterogeneity of motivationsg ysocial influence
a counterfactual use of ABM
Intrinsic Inclination to Work
Modelled as a parameterModelled as a parameter
0free‐riders
1suckers
(or surfers)
Intrinsic Inclination to Work
Modelled as a parameterModelled as a parameter
0free‐riders
1suckers
(or surfers)
0.0 0.2 0.4 0.6 0.8 1.0
learnerslearners
Features of the Simulated Labour Market
Fixed ParametersFixed Parameters
a)Number of agents (1000)gb)Minimum salary (5 units/hr)c)Exponential distribution of salaries ( d 20% f t f ll b l(around 20% of agents fall below minimum salary)d)Max hrs work/week (80)d)Max. hrs. work/week (80)e)Subsistence level (200 units)f)Intrinsic inclination to work (0 for free‐riders, 1 for suckers, and in between for learners)
Features of the Simulated Labour Market
Manipulated ParametersManipulated Parameters
• Population composition p pHow many free‐riders, suckers and learners?
• Strength of social influenceNone moderate or full?None, moderate or full?
Features of the Simulated Labour Market
Outcome VariablesOutcome Variables
• Total number of hours contributed tothe labour market
A t f BI i d b t• Amount of BI received by agents
The Dynamics of the Model
Random salarysalary
allocation
Agents calculate number of working hours needed to reach subsistence level
Redistribution of collected taxes as BI reach subsistence level
Agents decide how many Tax applied g yof the remaining hours they want to work
Tax applied to salaries
If All Agents are Learners…If All Agents are Learners…
k k h
90
avg week working hrs Basic Income amount300
50
60
70
80
200
250
20
30
40
50
gamma = 1gamma = 5
100
150
0
10
time
gamma = .5gamma = 0
0
50
With full social influence (gamma = 1), agents work about 25% more and receive a BI about 20% higher. g
If 20%Are Free‐Riders and 5% Suckers…If 20% Are Free Riders and 5% Suckers…
k k havg week working hrs Basic Income amount
50
60
250
300 75‐20‐5
30
40
50
150
200
250
10
20 gamma = 1gamma = .5gamma = 0 50
100
0time
0
Social influence still has a (slight) positive influence on number of hours contributed, but overall agents work now about 22%
less and receive a BI about 23% lowerless and receive a BI about 23% lower
If 45%Are Free‐Riders and 5% Suckers…If 45% Are Free Riders and 5% Suckers…
k k havg week working hrs Basic Income amount
40
45 300
25
30
35
40
200
250
10
15
20
gamma = 1gamma = .5 50
100
150
0
5
time
ggamma = 0
0
50
time
Social influence has now the opposite effect: free‐riders make learners work about 50% less; agents receive a BI about 44%
lowerlower
If All Agents are Free‐Riders…If All Agents are Free Riders…
avg week working hrs Basic Income amount
25
30
250
300
15
20
150
200
5
10gamma = 1gamma = .5gamma = 0
50
100
Social influence does not make a difference. Agents only work
0
time0
time
g ywhat is necessary to reach the subsistence level. They adjust their behaviour to fund a BI that allows them to work the
minimum hours. This BI is about 43% of what learners received in conditions of full social influence
The Impact of Free‐Riders (full social influence) e I pa o ee i e s ( u so ia i ue e)
‐10
0
ked
40
‐30
‐20
10
Hours W
or
‐60
‐50
‐40
ecrease of H
‐80
‐70
0 10 20 30 40 50 60 70 80 90 100
% De
% of Free Riders in the Population
The first 10% increase of free‐riders in the population implies a d b f k d h h fl f
% of Free‐Riders in the Population
17% decrease in number of worked hours – the influence of free‐riders decreases with their relative number because they
still have to reach subsistence levels!still have to reach subsistence levels!
The Impact of Suckers (full social influence)The Impact of Suckers (full social influence) 60
ked
40
50ou
rs W
ork
10
20
30
crease of H
0
10
0 10 20 30 40 50 60 70 80 90 100
% Inc
The first 10% increase of suckers in the population implies a
% of Suckers in the Population
30% increase in number of worked hours. The ceiling of their influence is determined by the parameter specifying the
i h t k k (80)maximum hours agents can work per week (80)
The Importance of Making BI Amount C diti l W ki B h iConditional on Working Behaviour
all learners all learners
80
90
endogenous BI exogenous BI
0
50
60
70
‐10
‐5
s worked
20
30
40
gamma = 1gamma = .5 ‐20
‐15
% of h
ours
0
10
time
gamma = 0‐25
%time
If the amount of BI is independent of working behaviour (exogenous and equal to subsistence level), agents work about 23% less time with full social influence and about 7% less without influencewith full social influence and about 7% less without influence
ConclusionsConclusions• How much a BI might impact on labour behaviourHow much a BI might impact on labour behaviour depends on our assumptions on
(a) The stability of people’s preferences (fixed or interdependent)
(b) Their relative weight in the population
If h i h i• If we assume agents heterogeneous in their predisposition to work, and susceptible to be influenced by others (as in the form of theinfluenced by others (as in the form of the proportionality norm), introducing a BI does not necessarily imply working lessnecessarily imply working less
ConclusionsConclusions
• Even when all agents are free‐riders, the model stabilises with a ‘partial BI’ – agents adapt their p g pbehaviour to maximise the unconditional benefits whilst minimising number of hours worked
• Similar results apply when agents manage global k l d h th th th l f i lknowledge or when the other other rules of social influence are introduced.
DebateDebate
• How far can we take experiments like this?How far can we take experiments like this?
• Can we calibrate it empirically?Can we calibrate it empirically? ‐ Can we determine the right population composition (in terms predispositions tocomposition (in terms predispositions to work) using empirical data?
• How do exercises like this contribute to the theoretical debate on the feasibility of socialtheoretical debate on the feasibility of social policy reforms?