CGE and DSGE Models: A Comparative Analysis
CGE and DSGE Models:
Reconciliation?
Sherman Robinson
Peterson Institute For International Economics (PIIE)
PIIE-UCB Macro Workshop, February 2020
Introduction: CGE and DSGE models
•
Computable General Equilibrium (CGE) models
•
Multi-agent, multi-market. Optimizing producers and households
interact across product and factor markets to determine
equilibrium prices and factor returns. Patron saint: Walras
•
Dynamics: multi-sector growth model. Patron saint: Solow-Swan
•
Dynamic Stochastic General Equilibrium (DSGE) models
•
Dynamic neoclassical growth models with forward-looking
optimizing agents. Few agents and markets.
•
Optimal saving behavior. Patron saint: Ramsey
2
CGE/DSGE: Similarities
•
General equilibrium: simulate operation of commodity
and factor markets
•
Equilibrium conditions are “descriptive” in that they are
assumed to describe the results of agent interaction
across markets in actual economies
•
CGE: profit-maximizing producers and utility-maximizing
consumers:
•
DSGE: add dynamic optimization with intertemporal utility
functions, endogenous savings behavior, and perfect
foresight
3
Descriptive Equilibria
•
Equilibrium concept: very powerful in an empirical model
•
Describes the results of a process that need not be specified in
the model
•
Need not describe or specify disequilibrium behavior. Need only solve for
equilibrium.
•
Need not discuss how individual agents interact to achieve
equilibrium
•
“Descriptive” if it can be validated empirically for the
domain of applicability of the model
•
Behavioral, theoretical, and statistical validation
4
Descriptive Equilibria
•
Use of equilibrium concepts greatly simplifies model
specification and, often, solution
•
Enhances model clarity and transparency: model behavior
consistent with economic theory
•
Facilitates validation: “predictable” empirical results from
“shocks”
•
Compare with “System Dynamics” models that specify
“rules of motion” but no equilibrium—very hard to tell
what is going on or to validate the model
•
Models written as difference equations, but no validation of
behavior or dynamic adjustment process
5
Equilibria: Forward-Looking Dynamic
CGE/DSGE Models
•
DSGE models—variants of the Ramsey model
•
Single household maximizes discounted utility
•
Optimizes savings rates over time
•
Producers maximize present value of discounted profits
•
All agents have perfect foresight
•
Models specified to have steady-state solutions
•
Multisectoral dynamic CGE models
•
Recursive dynamic models: common, no perfect foresight
•
Neoclassical optimal growth models
•
Roe et al.,
Multisectoral Growth Models: Theory and Application
•
Existence/stability of steady state solutions
6
Descriptive Dynamic Equilibria
•
Are Ramsey/DSGE/rational expectations model equilibria
“descriptive”? Specification of agent behavior is not
“realistic”.
•
Signals agents see: perfect foresight
•
Agents borrow only for intertemporal smoothing
•
No uncertainty. No “precautionary” savings motive.
•
Ability to adjust easily: no borrowing constraints (with
exceptions)
•
Unemployment only because of distortions in labor/leisure
choices by agents (e.g., Prescott)
•
Realism of assumption of steady state growth paths
7
DSGE Models
•
Kehoe and Prescott,
Great Depressions of the Twentieth
Century.
•
Blurb by Thomas Sargent: “Studying this book is an excellent
way to learn about how to apply and adapt the optimal growth
model to understand the most disturbing of macroeconomic
events of the twentieth century, great depressions. The book
bristles with intriguing stories, creative ways of expressing them
in terms of dynamic equilibrium models, and ambitious
attempts to compare them with data.”
•
Is “compare them with data” the same as “validation”? Are
they descriptive models? Very stylized models.
8
Macroeconomics: Keynesian
•
Short to medium-run focus: widen the domain of
applicability of simulation GE simulation
•
Capital “fixed” by sector: Marshallian short run
•
Want to analyze macro “shocks”:
•
Asian and other financial crises
•
Structural adjustment programs (World Bank)
•
Impact of stabilization programs (IMF)
•
Changes in trade policy: trade wars
•
Issue: Can factors be involuntarily unemployed?
•
Old and new Keynesian models
9
CGE/DSGE: Macro Flows
•
All economywide models include macro aggregates:
•
C, I, G, E, M
•
Macro flow equilibrium in CGE/DSGE models requires:
•
S-I balance
•
G-T balance
•
E-M balance
•
DSGE/CGE models: loanable funds market
•
Convert financial markets (assets, money) in a macro model into
a “shock” on the flow equilibrium in the loanable funds market in
the CGE/DSGE model
10
CGE/DSGE: E-M balance
•
All trade focused CGE/DSGE models have a functional
relationship between the trade balance (E-M) and the real
exchange rate (relative price of traded/non-traded goods)
•
Armington specification implies that all goods are imperfectly
tradable, so “non-traded” share is large (D = GDP – E).
•
In practice, CGE/DSGE models will solve for 2 out of 3
variables: trade balance (E-M), “nominal” exchange rate
(EXR), and an anchor price index (P).
•
Specify (E-M) and P, model solves for R
•
Specify R and P, model solves for E-M
•
Specify R and E-M, model solves for P
11
Global CGE/DSGE Models:
E-M Balance
•
Global CGE/DSGE models determine world prices that
clear world markets for commodities and determine real
exchange rates that equilibrate trade balances
•
Same mechanisms as in single-country models, but addition of a
global clearing requirement: Sum of E-M globally must be 0.
•
All CGE/DSGE models are Walrasian: only relative prices
•
Numeraire choice for countries: P or R are common
•
Global numeraire: $US is common, but also a basket of
OECD currencies
•
Choice matters for “valuing” E-M by countries. They represent a
“claim” on the exports of the numeraire countries.
12
GTAP Model: Questions
•
Does the GTAP model use CET functions for export supply?
This is an extension of Armington in CGE models.
•
Are the national trade balances (E – M) endogenous or
fixed? A standard version of GTAP endogenizes
international capital flows, and hence the trade balances.
•
What are the national numeraires? A version of GTAP uses
a wage as the national numeraire (which is Keynes in the
General Theory, but odd in CGE models).
•
Issue of defining the real exchange rate
•
Is GTAP using uniform CES elasticities for all countries?
13
Macro Closure: S-I, G-T, E-M
•
How to achieve macro balances
•
Walras Law: determine 2, the 3
rd
must balance
•
A K Sen article on macro closure in the 1960s
•
Heated debates in the 1970s in the CGE literature
•
Neoclassical, Structuralist, Johansen, Keynesian, Kaldorian
•
S-I balance: S determines I; I determines S
•
G-T balance: G or T adjust; G-T financed by “borrowing”
•
E-M balance: E-M exogenous, R adjust; R fixed, E-M adjusts
•
Issue: Is there involuntary unemployment?
•
Old and New Keynesian models
14
Full Employment Closures
•
“Compositional” macro
•
Factor markets clear, ensuring full employment and
essentially fixed GDP
•
Closure rules affects macro aggregates (C, I, G, E, and M),
but not aggregate GDP
•
Will affect “welfare” if E-M adjusts
•
Absorption (welfare) = GDP + M - E
•
Minimal strain on the neoclassical paradigm, but
inadequate for any Keynesian model
15
Unemployment Closures
•
Links between demand aggregates and real supply (GDP)
•
Must assume non-neoclassical factor markets
•
Must stretch the neoclassical paradigm for a Keynesian model
•
Common approach: fix wage/rental rate, model will respond
with quantity adjustment on employment/capital utilization
•
Factor markets do not clear by wages. Two common variants:
•
Firms on demand curve for labor
•
Firms not on demand curve for labor
•
Introduce a “wage curve” into the CGE model
•
Reduced form, but strong empirical support
•
Unemployment equilibrium imposed: consistent with Keynesian view
16
Conclusion
•
Keynesian macro closure is more difficult in a DSGE model
•
Endogenous optimal saving mechanism makes it difficult to
implement a Keynesian multiplier link between aggregate
demand and the factor markets
•
CGE models are more flexible in macro “closure” mechanisms,
but DSGE models are catching up
•
Both CGE and DSGE models involve “flow” equilibria in
macro balances, less treatment of asset market
•
Essentially a flow-of-funds model
•
Standard CGE models do not take account of asset holdings of
various agents. “Financial” CGE models do.
•
More asset “action” in DSGE models?
17
“Financial” CGE Models
•
CGE model as the core “supply side” in a broader financial
macro model
•
Need to have many sectors in order to analyze issues of
structural adjustment and trade shocks
•
Exchange rate and trade balance shocks
•
International price shocks (e.g., Oil)
•
Different modeling philosophies in the same neoclassical
theoretical framework
•
Implicit in DSGE models, which neglect the financial side
18
Orthodox School
•
“There is only one Model, and its prophet is Walras”
•
The Walrasian model is truth and should not be corrupted
to analyze macro issues
•
Stick to relative prices, resource allocation, and full employment
in the long run
•
Yale: Srinivasan, Kehoe, Whalley
•
Schizophrenia with DSGE models: neoclassical growth
model with a few features to incorporate macro concerns
19
Eclectic School
•
Integrate financial variables and asset markets into
neoclassical Walrasian CGE model. Draw from all schools
•
State of the art in CGE models:
•
Bourguignon, Branson, and de Melo
•
McKibbin-Sachs and McKibbin-Wilcoxen
•
Ag
énor (World Bank, Manchester)
•
All incorporate a standard CGE model (e.g., the 1-2-3 model) to
specify the supply side of a macro model
20
Ecumenical School: A Modular
Approach
•
“Render unto Walras the things which are Walras’, and
unto Keynes the things which are Keynes’.”
•
Separate real and financial models, CGE and macro models
•
Specify each model in its own theoretical framework
•
The issue is to link the models
•
Variables exogenous in one model are endogenous in the other
•
Need to have the CGE model incorporate the results of the
macro model
21
Ecumenical School: A Modular
Approach
•
Build two models: A modular approach
•
CGE supply-side model
•
Macroeconomic model
•
Formally link the models:
•
Robinson/Tyson.
•
Dixon et al.
•
Clean division between the models
•
Links to force consistency between macro and CGE models
22
Ecumenical School
•
Problems.
•
Same agents in both models doing very different things:
•
E.g., savings as flow equilibrium versus asset allocation equilibrium
•
Some variables jointly determined, with potential conflicts (e.g,
exchange rate)
•
Schizophrenia shared by literature on micro foundations of
macro.
•
Still no fundamental reconciliation
23
Macro Closure Approach: Implement
Modularity
•
Tell the macro story outside the CGE model.
•
Explicit macro-econometric model
•
Macro “story”: informal specification of a macro model
•
Impose the results on the CGE model.
•
No money, assets, or financial variables in the CGE model.
•
Define “macro closure” rules to ensure macro flow
equilibrium in the CGE model
24
References
•
Brenner, T. and C. Werker (2007). “A Taxonomy of Inference in
Simulation Models.”
Computational Economics
, Vol. 30, pp
227-244.
•
Edmonds, B. and S. Moss (2004). “From KISS to KIDS—an ‘Anti-
Simplistic’ Modelling Approach. International Workshop on
Multi-Agent Systems and Agent Based Simulation. Springer
Berlin/Heidelberg, pp. 130-144.
•
Gintis, Herbert (2007). “The Dynamics of General Equilibrium.”
The Economic Journal,
Vol. 117, No. 523, pp 1280-1309.
•
Gintis, Herbert (2009).
The Bounds of Reason: Game Theory
and the Unification of the Behavioral Sciences.
Princeton:
Princeton University Press, Second edition.
25
References
•
Gintis, Herbert (2000).
Game Theory Evolving: A Problem-
Centered Introduction to Modeling Strategic Interaction.
Princeton, NJ: Princeton University Press.
•
Kehoe, Timothy and Edward C. Prescott, eds. (2007).
Great
Depressions of the Twentieth Century.
Federal Reserve
Bank of Minneapolis.
•
Peters, Ole (2019). “The ergodicity problem in economics.”
Nature Physics
, Vol. 15 (December), pp 1216-1221.
26
References
•
Pyka, Andreas and Claudia Werker (2009). “The Methodology
of Simulation Models: Chances and Risks.”
Journal of Artificial
Societies and Social Simulation,”
Vol. 12, No. 4, p 1.
•
Robinson, Sherman. 1989. “Multisectoral Models.” Chapter 18,
in
Handbook of Development Economics
. H. Chenery and T.N.
Srinivasan (eds.). Amsterdam: North-Holland Publishing Co, pp.
885-947.
•
Robinson, Sherman. 2006 . “Macro Models and Multipliers:
Leontief, Stone, Keynes, and CGE Models.” Chapter 11 in Alain
de Janvry and Ravi Kanbur, eds.
Poverty, Inequality and
Development: Essays in Honor of Erik Thorbecke.
New York:
Springer Science, pp. 205-232.
27
References
•
Robinson, Sherman. 1991. “Macroeconomics, Financial
Variables, and Computable General Equilibrium Models.”
World Development
. Vol. 19, No. 11, pp. 1509-1525.
•
Weisberg, Michael (2013).
Simulation and Similarity:
Using Models to Understand the World.
Oxford: Oxford
University Press.
•
Devarajan, Shantayanan, Jeffrey D. Lewis, and Sherman
Robinson. 1993. “External Shocks, Purchasing Power
Parity, and the Equilibrium Real Exchange Rate.”
World
Bank Economic Review
. Vol. 7, No. 1, pp. 45-63.
28
Explore the similarities between Computable General Equilibrium (CGE) models and Dynamic Stochastic General Equilibrium (DSGE) models, their equilibrium concepts, and the use of descriptive equilibria in empirical modeling. Learn how CGE and DSGE models simulate the operation of commodity and factor markets, incorporating dynamic optimization and intertemporal utility functions in DSGE models for forward-looking analysis.
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. Download presentation by click this link. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
E N D
Presentation Transcript
CGE and DSGE Models: Reconciliation? Sherman Robinson Peterson Institute For International Economics (PIIE) PIIE-UCB Macro Workshop, February 2020
Introduction: CGE and DSGE models Computable General Equilibrium (CGE) models Multi-agent, multi-market. Optimizing producers and households interact across product and factor markets to determine equilibrium prices and factor returns. Patron saint: Walras Dynamics: multi-sector growth model. Patron saint: Solow-Swan Dynamic Stochastic General Equilibrium (DSGE) models Dynamic neoclassical growth models with forward-looking optimizing agents. Few agents and markets. Optimal saving behavior. Patron saint: Ramsey 2
CGE/DSGE: Similarities General equilibrium: simulate operation of commodity and factor markets Equilibrium conditions are descriptive in that they are assumed to describe the results of agent interaction across markets in actual economies CGE: profit-maximizing producers and utility-maximizing consumers: DSGE: add dynamic optimization with intertemporal utility functions, endogenous savings behavior, and perfect foresight 3
Descriptive Equilibria Equilibrium concept: very powerful in an empirical model Describes the results of a process that need not be specified in the model Need not describe or specify disequilibrium behavior. Need only solve for equilibrium. Need not discuss how individual agents interact to achieve equilibrium Descriptive if it can be validated empirically for the domain of applicability of the model Behavioral, theoretical, and statistical validation 4
Descriptive Equilibria Use of equilibrium concepts greatly simplifies model specification and, often, solution Enhances model clarity and transparency: model behavior consistent with economic theory Facilitates validation: predictable empirical results from shocks Compare with System Dynamics models that specify rules of motion but no equilibrium very hard to tell what is going on or to validate the model Models written as difference equations, but no validation of behavior or dynamic adjustment process 5
Equilibria: Forward-Looking Dynamic CGE/DSGE Models DSGE models variants of the Ramsey model Single household maximizes discounted utility Optimizes savings rates over time Producers maximize present value of discounted profits All agents have perfect foresight Models specified to have steady-state solutions Multisectoral dynamic CGE models Recursive dynamic models: common, no perfect foresight Neoclassical optimal growth models Roe et al., Multisectoral Growth Models: Theory and Application Existence/stability of steady state solutions 6
Macroeconomics: Keynesian Short to medium-run focus: widen the domain of applicability of simulation GE simulation Capital fixed by sector: Marshallian short run Want to analyze macro shocks : Asian and other financial crises Structural adjustment programs (World Bank) Impact of stabilization programs (IMF) Changes in trade policy: trade wars Issue: Can factors be involuntarily unemployed? Old and new Keynesian models 9
CGE/DSGE: Macro Flows All economywide models include macro aggregates: C, I, G, E, M Macro flow equilibrium in CGE/DSGE models requires: S-I balance G-T balance E-M balance DSGE/CGE models: loanable funds market Convert financial markets (assets, money) in a macro model into a shock on the flow equilibrium in the loanable funds market in the CGE/DSGE model 10
CGE/DSGE: E-M balance All trade focused CGE/DSGE models have a functional relationship between the trade balance (E-M) and the real exchange rate (relative price of traded/non-traded goods) Armington specification implies that all goods are imperfectly tradable, so non-traded share is large (D = GDP E). In practice, CGE/DSGE models will solve for 2 out of 3 variables: trade balance (E-M), nominal exchange rate (EXR), and an anchor price index (P). Specify (E-M) and P, model solves for R Specify R and P, model solves for E-M Specify R and E-M, model solves for P 11
Global CGE/DSGE Models: E-M Balance Global CGE/DSGE models determine world prices that clear world markets for commodities and determine real exchange rates that equilibrate trade balances Same mechanisms as in single-country models, but addition of a global clearing requirement: Sum of E-M globally must be 0. All CGE/DSGE models are Walrasian: only relative prices Numeraire choice for countries: P or R are common Global numeraire: $US is common, but also a basket of OECD currencies Choice matters for valuing E-M by countries. They represent a claim on the exports of the numeraire countries. 12
GTAP Model: Questions Does the GTAP model use CET functions for export supply? This is an extension of Armington in CGE models. Are the national trade balances (E M) endogenous or fixed? A standard version of GTAP endogenizes international capital flows, and hence the trade balances. What are the national numeraires? A version of GTAP uses a wage as the national numeraire (which is Keynes in the General Theory, but odd in CGE models). Issue of defining the real exchange rate Is GTAP using uniform CES elasticities for all countries? 13
Macro Closure: S-I, G-T, E-M How to achieve macro balances Walras Law: determine 2, the 3rd must balance A K Sen article on macro closure in the 1960s Heated debates in the 1970s in the CGE literature Neoclassical, Structuralist, Johansen, Keynesian, Kaldorian S-I balance: S determines I; I determines S G-T balance: G or T adjust; G-T financed by borrowing E-M balance: E-M exogenous, R adjust; R fixed, E-M adjusts Issue: Is there involuntary unemployment? Old and New Keynesian models 14
Full Employment Closures Compositional macro Factor markets clear, ensuring full employment and essentially fixed GDP Closure rules affects macro aggregates (C, I, G, E, and M), but not aggregate GDP Will affect welfare if E-M adjusts Absorption (welfare) = GDP + M - E Minimal strain on the neoclassical paradigm, but inadequate for any Keynesian model 15
Unemployment Closures Links between demand aggregates and real supply (GDP) Must assume non-neoclassical factor markets Must stretch the neoclassical paradigm for a Keynesian model Common approach: fix wage/rental rate, model will respond with quantity adjustment on employment/capital utilization Factor markets do not clear by wages. Two common variants: Firms on demand curve for labor Firms not on demand curve for labor Introduce a wage curve into the CGE model Reduced form, but strong empirical support Unemployment equilibrium imposed: consistent with Keynesian view 16
Conclusion Keynesian macro closure is more difficult in a DSGE model Endogenous optimal saving mechanism makes it difficult to implement a Keynesian multiplier link between aggregate demand and the factor markets CGE models are more flexible in macro closure mechanisms, but DSGE models are catching up Both CGE and DSGE models involve flow equilibria in macro balances, less treatment of asset market Essentially a flow-of-funds model Standard CGE models do not take account of asset holdings of various agents. Financial CGE models do. More asset action in DSGE models? 17
