Multiple equilibria in economic systems and stochastic dynamics or: - - PowerPoint PPT Presentation

multiple equilibria in economic systems and stochastic
SMART_READER_LITE
LIVE PREVIEW

Multiple equilibria in economic systems and stochastic dynamics or: - - PowerPoint PPT Presentation

Potsdam Institute for Climate Impact Research PIK Member of the Leibniz Association Multiple equilibria in economic systems and stochastic dynamics or: prices as conventions in agent-based models of growing economies Sarah Wolf Work with the


slide-1
SLIDE 1

Potsdam Institute for Climate Impact Research PIK

Member of the Leibniz Association

Multiple equilibria in economic systems and stochastic dynamics

  • r: prices as conventions in agent-based models of growing

economies

Sarah Wolf

Work with the Lagom group at PIK: Prof. Carlo C. Jaeger Steffen F¨ urst, Wiebke Lass, Daniel Lincke, Antoine Mandel

ECCS, Lisbon, September 16, 2010 Satellite meeting: Social Complexity of Informal Value Exchange

slide-2
SLIDE 2

Background SCIVE Ideas in progress References

Outline

Background SCIVE Ideas in progress

slide-3
SLIDE 3

Background SCIVE Ideas in progress References

Economics and Climate Change

◮ Standard: Theory of general equilibrium ◮ supply = demand; several equilibria possible; static view ◮ models compute a unique equilibrium

slide-4
SLIDE 4

Background SCIVE Ideas in progress References

Economics and Climate Change

◮ Standard: Theory of general equilibrium ◮ supply = demand; several equilibria possible; static view ◮ models compute a unique equilibrium ◮ Climate policy context: intertemporal optimization models ◮ one welfare function optimized, value of a sequence of

situations society finds itself in is summarized in a global utility function

◮ “details” disputed (discounting issues etc)

slide-5
SLIDE 5

Background SCIVE Ideas in progress References

Economics and Climate Change

◮ Standard: Theory of general equilibrium ◮ supply = demand; several equilibria possible; static view ◮ models compute a unique equilibrium ◮ Climate policy context: intertemporal optimization models ◮ one welfare function optimized, value of a sequence of

situations society finds itself in is summarized in a global utility function

◮ “details” disputed (discounting issues etc) ◮ emission reductions are considered a deviation from the

business as usual path

◮ deviation from optimal path has a cost

slide-6
SLIDE 6

Background SCIVE Ideas in progress References

Economics and Climate Change

◮ Standard: Theory of general equilibrium ◮ supply = demand; several equilibria possible; static view ◮ models compute a unique equilibrium ◮ Climate policy context: intertemporal optimization models ◮ one welfare function optimized, value of a sequence of

situations society finds itself in is summarized in a global utility function

◮ “details” disputed (discounting issues etc) ◮ emission reductions are considered a deviation from the

business as usual path

◮ deviation from optimal path has a cost ◮ climate problem = problem of distributing mitigation costs ◮ e.g. COP 15, Copenhagen 2009, deadlock situation

slide-7
SLIDE 7

Background SCIVE Ideas in progress References

Alternative approach

◮ switch focus: consider emission reduction goals as focal point ◮ emissions as an externality, internalization yields benefits

Foley [2007]: The economic fundamentals of global warming

slide-8
SLIDE 8

Background SCIVE Ideas in progress References

Alternative approach

◮ switch focus: consider emission reduction goals as focal point ◮ emissions as an externality, internalization yields benefits

Foley [2007]: The economic fundamentals of global warming

◮ leaving the one equilibrium or BAU path, theoretical issues

arise

◮ theory of general equilibrium does not say anything about

choice of equilibria or out-of-equilibrium behaviour of the economy

slide-9
SLIDE 9

Background SCIVE Ideas in progress References

Alternative approach

◮ switch focus: consider emission reduction goals as focal point ◮ emissions as an externality, internalization yields benefits

Foley [2007]: The economic fundamentals of global warming

◮ leaving the one equilibrium or BAU path, theoretical issues

arise

◮ theory of general equilibrium does not say anything about

choice of equilibria or out-of-equilibrium behaviour of the economy

◮ agent-based economic models

aims: investigate system behaviour and generate more complete theory

slide-10
SLIDE 10

Background SCIVE Ideas in progress References

Economic ABM

◮ Aspects of the economy, such as financial markets ◮ economy as a whole, e.g. EURACE model

slide-11
SLIDE 11

Background SCIVE Ideas in progress References

Economic ABM

◮ Aspects of the economy, such as financial markets ◮ economy as a whole, e.g. EURACE model ◮ Gintis [2007]: The dynamics of general equilibrium

agent-based dynamics to replace Walrasian auctioneer; agents have private prices, limited information about other agents, imitate and mutate strategies

slide-12
SLIDE 12

Background SCIVE Ideas in progress References

Economic ABM

◮ Aspects of the economy, such as financial markets ◮ economy as a whole, e.g. EURACE model ◮ Gintis [2007]: The dynamics of general equilibrium

agent-based dynamics to replace Walrasian auctioneer; agents have private prices, limited information about other agents, imitate and mutate strategies

◮ Gintis observes convergence to equilibrium prices in

simulations

◮ theory? simple mathematical description of what happens in

the model?

slide-13
SLIDE 13

Background SCIVE Ideas in progress References

Economic ABM

◮ Aspects of the economy, such as financial markets ◮ economy as a whole, e.g. EURACE model ◮ Gintis [2007]: The dynamics of general equilibrium

agent-based dynamics to replace Walrasian auctioneer; agents have private prices, limited information about other agents, imitate and mutate strategies

◮ Gintis observes convergence to equilibrium prices in

simulations

◮ theory? simple mathematical description of what happens in

the model?

◮ Bilancini and Petri [2008] caution: capital used in production

is ‘land’, equilibrium notion becomes irrelevant when using capital

slide-14
SLIDE 14

Background SCIVE Ideas in progress References

Lagom models

◮ Swedish word lagom ≈ just right

slide-15
SLIDE 15

Background SCIVE Ideas in progress References

Lagom models

◮ Swedish word lagom ≈ just right ◮ Lagom models based on Gintis’ model, add heterogeneous

capital goods, economic growth

◮ sector structure, several regions, labour productivity grows

with investment, financial system determines interest rate etc.

slide-16
SLIDE 16

Background SCIVE Ideas in progress References

Lagom models

◮ Swedish word lagom ≈ just right ◮ Lagom models based on Gintis’ model, add heterogeneous

capital goods, economic growth

◮ sector structure, several regions, labour productivity grows

with investment, financial system determines interest rate etc.

◮ “proof of concept”: large-scale agent-based macro-model with

capital accumulation

◮ simulations suggest that agent-based dynamics can drive the

economic system to a stochastically stable state with equilibrium features [Mandel et al., 2010]

◮ again, theory seems out of reach right now

slide-17
SLIDE 17

Background SCIVE Ideas in progress References

Social Complexity of Informal Value Exchange

◮ social phenomena that involve value-exchange ◮ social processes and mechanisms other than those usually

considered by economists social norms, reputation, trust, group membership, expectations

slide-18
SLIDE 18

Background SCIVE Ideas in progress References

Social Complexity of Informal Value Exchange

◮ social phenomena that involve value-exchange ◮ social processes and mechanisms other than those usually

considered by economists social norms, reputation, trust, group membership, expectations

◮ The economic processes of price, supply/demand and varieties

  • f economic rationality (e.g. bounded rationality, optimisation

etc.) are relatively well studied – this workshop aims to concentrate on the other social aspects involved.

slide-19
SLIDE 19

Background SCIVE Ideas in progress References

Social Complexity of Informal Value Exchange

◮ social phenomena that involve value-exchange ◮ social processes and mechanisms other than those usually

considered by economists social norms, reputation, trust, group membership, expectations

◮ The economic processes of price, supply/demand and varieties

  • f economic rationality (e.g. bounded rationality, optimisation

etc.) are relatively well studied – this workshop aims to concentrate on the other social aspects involved.

◮ work in progress in the spirit of the Dahlem Conference

“New Approaches to Economics after the Financial Crisis” (Berlin August, 2010)

◮ central idea: insights from the social sciences, a variety of

models, and judgement needed for “better” economics, in particular concerning policy making

slide-20
SLIDE 20

Background SCIVE Ideas in progress References

Value

◮ Value formalized at different levels in standard economic

models: money, utility, welfare make more abstract ideas comparable, reduce to 1 dimension

◮ aggregate indicators of value considered, GDP etc.

slide-21
SLIDE 21

Background SCIVE Ideas in progress References

Value

◮ Value formalized at different levels in standard economic

models: money, utility, welfare make more abstract ideas comparable, reduce to 1 dimension

◮ aggregate indicators of value considered, GDP etc. ◮ simple model with several goods: compare vectors ◮ even GDP not necessarily comparable unless changes are small

slide-22
SLIDE 22

Background SCIVE Ideas in progress References

Value

◮ Value formalized at different levels in standard economic

models: money, utility, welfare make more abstract ideas comparable, reduce to 1 dimension

◮ aggregate indicators of value considered, GDP etc. ◮ simple model with several goods: compare vectors ◮ even GDP not necessarily comparable unless changes are small ◮ standard economic models consider cases in neighbourhood of

  • ne equilibrium
slide-23
SLIDE 23

Background SCIVE Ideas in progress References

Value

◮ Value formalized at different levels in standard economic

models: money, utility, welfare make more abstract ideas comparable, reduce to 1 dimension

◮ aggregate indicators of value considered, GDP etc. ◮ simple model with several goods: compare vectors ◮ even GDP not necessarily comparable unless changes are small ◮ standard economic models consider cases in neighbourhood of

  • ne equilibrium

◮ when looking at several equilibria it is not so clear how to

compare “value”

slide-24
SLIDE 24

Background SCIVE Ideas in progress References

Value

◮ Value formalized at different levels in standard economic

models: money, utility, welfare make more abstract ideas comparable, reduce to 1 dimension

◮ aggregate indicators of value considered, GDP etc. ◮ simple model with several goods: compare vectors ◮ even GDP not necessarily comparable unless changes are small ◮ standard economic models consider cases in neighbourhood of

  • ne equilibrium

◮ when looking at several equilibria it is not so clear how to

compare “value”

◮ reduction to 1 dimension desirable

slide-25
SLIDE 25

Background SCIVE Ideas in progress References

Stochastic dynamics

◮ investigate economic dynamics using simple stochastic models

slide-26
SLIDE 26

Background SCIVE Ideas in progress References

Stochastic dynamics

◮ investigate economic dynamics using simple stochastic models ◮ dXt = f (Xt)dt + √εdBt

slide-27
SLIDE 27

Background SCIVE Ideas in progress References

Stochastic dynamics

◮ investigate economic dynamics using simple stochastic models ◮ dXt = f (Xt)dt + √εdBt ◮ random perturbations allow transitions between different

meta-stable states

◮ dynamics: long stays around one meta-stable state and quick

switches to another

slide-28
SLIDE 28

Background SCIVE Ideas in progress References

Stochastic dynamics

◮ investigate economic dynamics using simple stochastic models ◮ dXt = f (Xt)dt + √εdBt ◮ random perturbations allow transitions between different

meta-stable states

◮ dynamics: long stays around one meta-stable state and quick

switches to another

◮ conventions, e.g., Young [1993]: repeated n-person game,

finitely many strategies, best reply and mutation Mandel and Botta [2009]: exchange economy

slide-29
SLIDE 29

Background SCIVE Ideas in progress References

Stochastic dynamics

◮ investigate economic dynamics using simple stochastic models ◮ dXt = f (Xt)dt + √εdBt ◮ random perturbations allow transitions between different

meta-stable states

◮ dynamics: long stays around one meta-stable state and quick

switches to another

◮ conventions, e.g., Young [1993]: repeated n-person game,

finitely many strategies, best reply and mutation Mandel and Botta [2009]: exchange economy

◮ Ormerod et al. [2009]: cluster analysis of US, British and

German data (1871-2009) identify different regimes in inflation/unemployment space major shifts between regimes and small fluctuations within

slide-30
SLIDE 30

Background SCIVE Ideas in progress References

Interpretation

◮ potential function as a tool for thinking about value

1 dimension to make situations commensurable; “social energy landscape”

slide-31
SLIDE 31

Background SCIVE Ideas in progress References

Interpretation

◮ potential function as a tool for thinking about value

1 dimension to make situations commensurable; “social energy landscape”

◮ dynamics driven by social phenomena instead of single agent

  • ptimizing
slide-32
SLIDE 32

Background SCIVE Ideas in progress References

Interpretation

◮ potential function as a tool for thinking about value

1 dimension to make situations commensurable; “social energy landscape”

◮ dynamics driven by social phenomena instead of single agent

  • ptimizing

◮ derive potential function from a representation of agents’

expectations (feedback)

◮ social process of how expectations are formed

slide-33
SLIDE 33

Background SCIVE Ideas in progress References

Interpretation

◮ potential function as a tool for thinking about value

1 dimension to make situations commensurable; “social energy landscape”

◮ dynamics driven by social phenomena instead of single agent

  • ptimizing

◮ derive potential function from a representation of agents’

expectations (feedback)

◮ social process of how expectations are formed ◮ “next-best try” towards some understanding, replacing welfare

function optimization

slide-34
SLIDE 34

Background SCIVE Ideas in progress References

Interpretation

◮ potential function as a tool for thinking about value

1 dimension to make situations commensurable; “social energy landscape”

◮ dynamics driven by social phenomena instead of single agent

  • ptimizing

◮ derive potential function from a representation of agents’

expectations (feedback)

◮ social process of how expectations are formed ◮ “next-best try” towards some understanding, replacing welfare

function optimization

◮ stochastic dynamic model as a simpler pre-step towards

agent-based model that could represent the observed dynamics

◮ get an idea of which details might be the important ones from

the micro level that influence the macro level

slide-35
SLIDE 35

Background SCIVE Ideas in progress References

Link: stochastic dynamics - agent dynamics

◮ S´

ebastian Grauwin et al.: Effective free energy for individual dynamics

slide-36
SLIDE 36

Background SCIVE Ideas in progress References

Link: stochastic dynamics - agent dynamics

◮ S´

ebastian Grauwin et al.: Effective free energy for individual dynamics

◮ Wyart and Bouchaud [2007]: Self-referential behaviour,

  • verreaction and conventions in financial markets

◮ chartist agents use (alledged) correlation between some

information (index) and stock price to predict stock price; their behaviour impacts the price

◮ noise arises from other agents trading randomly

slide-37
SLIDE 37

Background SCIVE Ideas in progress References

Link: stochastic dynamics - agent dynamics

◮ S´

ebastian Grauwin et al.: Effective free energy for individual dynamics

◮ Wyart and Bouchaud [2007]: Self-referential behaviour,

  • verreaction and conventions in financial markets

◮ chartist agents use (alledged) correlation between some

information (index) and stock price to predict stock price; their behaviour impacts the price

◮ noise arises from other agents trading randomly ◮ impact function: price change as function of what agents do

estimation procedure used by chartist agents combined into an Itˆ

  • -Langevin equation, with double-well

potential function

slide-38
SLIDE 38

Background SCIVE Ideas in progress References

Link: stochastic dynamics - agent dynamics

◮ S´

ebastian Grauwin et al.: Effective free energy for individual dynamics

◮ Wyart and Bouchaud [2007]: Self-referential behaviour,

  • verreaction and conventions in financial markets

◮ chartist agents use (alledged) correlation between some

information (index) and stock price to predict stock price; their behaviour impacts the price

◮ noise arises from other agents trading randomly ◮ impact function: price change as function of what agents do

estimation procedure used by chartist agents combined into an Itˆ

  • -Langevin equation, with double-well

potential function

◮ stable conventions appear for correlations of price change and

index change (positive/negative) quick switches from one to the other

slide-39
SLIDE 39

Background SCIVE Ideas in progress References

Goals

◮ climate policy context: get out of deadlock situation ◮ identify concrete win-win options for emission reduction and

labor market/economic growth (e.g. energy-efficiency improvements to buildings)

slide-40
SLIDE 40

Background SCIVE Ideas in progress References

Goals

◮ climate policy context: get out of deadlock situation ◮ identify concrete win-win options for emission reduction and

labor market/economic growth (e.g. energy-efficiency improvements to buildings)

◮ theory: explain such options in an economic framework ◮ not completely scrapping general equilibrium theory:

transitions between different equilibria

slide-41
SLIDE 41

Background SCIVE Ideas in progress References

Goals

◮ climate policy context: get out of deadlock situation ◮ identify concrete win-win options for emission reduction and

labor market/economic growth (e.g. energy-efficiency improvements to buildings)

◮ theory: explain such options in an economic framework ◮ not completely scrapping general equilibrium theory:

transitions between different equilibria

◮ potential function model ◮ very basic Lagom model, i.e. agent-based but tractable

slide-42
SLIDE 42

Background SCIVE Ideas in progress References

References

http://european-climate-forum.net Duncan K. Foley. The economic fundamentals of global warming. Santa Fe Working Paper, 2007. Herbert Gintis. The Dynamics of General Equilibrium. Economic Journal, 117(523): 1280–1309, 2007. Ennio Bilancini and Fabio Petri. A Comment on Gintis’s ”The Dynamics of General Equilibrium”. Economics Bulletin, 2:1–7, 2008. Antoine Mandel, Carlo C. Jaeger, Steffen F¨ urst, Wiebke Lass, Daniel Lincke, Frank Meissner, Federico Pablo-Mart` ı, and Sarah Wolf. Agent-based dynamics in disaggregated growth models. In preparation, 2010.

  • H. Peyton Young. The Evolution of Conventions. Econometrica, 61(1):57–84, 1993.

Antoine Mandel and Nicolca Botta. A note on the stochastic stability of equilibrium in some exchange economies. Documents de Travail du Centre d’Economie de la Sorbonne, 2009.84, 2009. Paul Ormerod, Bridget Rosewell, and Peter Phelps. Inflation/unemployment regimes and the instability of the philipps curve. Economics E-Journal Discussion Paper,

  • No. 2009-43, 2009. URL

http://www.economics-ejournal.org/economics/discussionpapers/2009-43. Matthieu Wyart and Jean-Philippe Bouchaud. Self-referential behaviour, overreaction and conventions in financial markets. Journal of Economic Behaviour and Organization, 63:1–24, 2007.

Thank You!