Computable General Equilibrium Models: Training ...
Post on 29-Dec-2021
3 Views
Preview:
Transcript
Computable General Equilibrium
Models: Training Opportunities and
Skill Development
Scott McDonald (Humboldt)
&
Karen Thierfelder (USNA)
UNFCCC, Bonn, 2018 1© cgemod
Outline
• Introduction
• Types of User
• Training Decisions
– Software
– Models
• Providers of CGE Training
• Outcomes
• Support and Mentoring
• Concluding Comments
UNFCCC, Bonn, 2018 2© cgemod
Introduction
• Excess demand for CGE analyses– Why?
• Skill set: economics, coding, data, policy, ……..
• Types of Users– Modellers – ‘those who change the behaviours in models’
– Users – ‘those who code and interpret experiments’
– Clients – ‘those who commission CGE analyses’
• Training Decisions– Software
– Models• Global
• Single country
• CGE Training
• Support
UNFCCC, Bonn, 2018 3© cgemod
CGE Models and SAMs
• The data used for all CGE models can be present in a SAM format
– Transactions (pi*qij) matrix (SAM)
• Agents
– Commodities/activities
– Factors
– Institutions – household, government, enterprises, investment_saving
– Taxes
» Active and potential
– Rest of the World
– Satellite Accounts
• Labour quantities
• Energy quantities and emissions
• ……
– Alternative (Green/Brown) TechnologiesUNFCCC, Bonn, 2018 4© cgemod
Data and model are separable
CGE Model Components
• Markets – prices and quantities– Commodities/Activities
– Factors
– Trade
• Competition– Perfect/Imperfect, e.g., Cournot, monopolistic competition, …..
• Technologies
• Preferences
• Market clearing conditions– Without restrictions – prices
– With restrictions – quantities, etc.
• Macroeconomic closures– From new classical to structuralist
UNFCCC, Bonn, 2018 5© cgemod
Flexible price models
CGE Models and Time
• Comparative static– Typical constraints
• Factor supplies fixed
• Technology fixed
• Time path of shock does not (overly) determine the outcome
• Recursive dynamic– Backward looking – adjustments based on past outcomes
• Partial adjustment mechanisms
– ‘Baseline’ needs to be defined• Technologies
• Factor supples
• Forward-looking dynamic– Full - ‘Ramsey saving model’
– With terminal constraints
UNFCCC, Bonn, 2018 6© cgemod
Types of Users
‘Types’
• ‘Modellers’
– The developers of models
• Skill set: advanced economics and programming
• ‘Users’
– The Users and Adapters of models
• Users: coding, data and interpreting experiments
• Adapter: modifying behavioural relationships
• ‘Clients’
– Good clients
• An understanding of the strengths, weaknesses and limitations of CGE models
– Strong economics and understanding of policy
– Poor clients
• Those seeking results that confirm their priors
– *******!!
Experience
• ‘Modellers’
– 10+ years
– PhD (Econ, OR, ..)
– ‘blinkers’
• ‘Users’
– MSc/PhD (Econ)
– 3+ years
– proto modellers
• ‘Clients
– MSc/PhD (Econ)
– c10 years
– knowledge/insight
UNFCCC, Bonn, 2018 7© cgemod
Training Decisions - SoftwareGAMS
• General Algebraic Modeling
System
– GAMS Corp (USA & Germany)
• General
– Optimisation problems
• Strengths
• Weaknesses
• User base
GEMPACK
• General Equilibrium Modeling
PACKage
– KPSoft and CoPS (Australia)
• Specialist
– CGE problems
• Strengths
• Weaknesses
• User base
UNFCCC, Bonn, 2018 8© cgemod
CGE models can be done equally well in
GAMS or GEMPACK
Arguments about solution algorithms are futile
Training Decisions - ModelsGAMS
• Global Models
– GTAPnGAMS
– LINKAGE
– GEME3
– E3ME
– GLOBE/GLOBE_EN /…
– PEP
• Single Country
– IFPRI Standard
– STAGE/STAGE_EN/….
– PEP
GEMPACK
• Global Models
– GTAP/GTAP_E
– MyGTAP
– MAGNET
• Single Country
– ORANI/MONASH/TERM
family
UNFCCC, Bonn, 2018 9© cgemod
Global and Single country model families:
PEP
GLOBE & STAGE (cgemod)
CGE Training - GEMPACKGLOBAL
GTAP (USA)
www.gtap.agecon.purdue.edu
• GTAP model
• Training– Online
• GTAP101
• GTAP PTA
– On site• Annual Short Course
• Dynamic Short Course
SINGLE Country
CoPS (Australia)
www.vu.edu.au/cops
• TERM and ORANI models
• Training
– On site
• Introductory Practical GE course
• Various Dynamic courses
• Various GEMPACK courses
UNFCCC, Bonn, 2018 10© cgemod
CGE Training – GAMS (PEP)GLOBAL
PEP (Canada)
www.pep-net.org
• No Training identified
SINGLE Country
PEP (Canada)
www.pep-net.org
• PEP Standard Model
• Training
– On line
• Introduction to CGE modelling
(NEW)
UNFCCC, Bonn, 2018 11© cgemod
CGE Training – GAMS (EcoMod)GLOBAL
EcoMod (Belgium)
ecomod.net
• NO models identified
• Training (??)
– On line
• Trade Policy modelling (??)
SINGLE Country
EcoMod (Belgium)
ecomod.net
• NO models identified
• Training– On line and On site (inter alia)
• Practical GE modelling
• Energy & Environment GE Modelling
• Advanced Techniques in GE Modelling
UNFCCC, Bonn, 2018 12© cgemod
CGE Training – GAMS (cgemod)GLOBAL
cgemod (USA & UK)
www.cgemod.org.uk/global_cge.html
• GLOBE models
• Training
– Online
• Practical CGE Modelling
• Intermediate CGE Modelling
– On site
• Advanced Training/Support
SINGLE Country
cgemod (USA & UK)
www.cgemod.org.uk/single_int.html
• STAGE models
• Training
– Online
• Practical CGE Modelling
• Intermediate CGE Modelling
– On site
• Advanced Training/Support
UNFCCC, Bonn, 2018 13© cgemod
CGE Training – cgemod Open Source
UNFCCC, Bonn, 2018 © cgemod 14
Open source:
1. SAMs
2. GAMS
CGE Training – cgemod MOODLE
UNFCCC, Bonn, 2018 © cgemod 15
video
documents
code library
CGE Training – GAMS
• The GAMS corporation provides a list of courses using GAMS
www.gams.com
UNFCCC, Bonn, 2018 16© cgemod
Outcomes from ‘Short’ Courses
YES
• Better client
– More practice at interpreting model results needed
• Basic user
– Mentoring and a lot more practice
• Trainee modeller
– The next 3 years of learning and practicing will tell
NO
• Good client
– Experience and practice will be needed
• Good user
– ‘Rome was not built in a week’
• Modeller
– Why do you think there is excess demand?
UNFCCC, Bonn, 2018 17© cgemod
From a good programme of training
ALL available (short) courses can only provide a start.
Course graduates need a lot of time after the courses to
embed the materials.
Support and Mentoring
• Line managers
– A short course is NOT enough
– Allocate time (2 months +) for a post course project
– Arrange mentoring
• In-house mentors
– Allocate time (2 months +) for a post course project
• Ideal – IF they are experienced
• Useless (or worse) – IF they are NOT experienced
• External mentors
– Allocate time (2 months +) for a post course project
– Cost
UNFCCC, Bonn, 2018 18© cgemod
Concluding Comments
• Do not expect too much
• Separate research projects and training
– Employ teachers to teach
– Employ researcher to research
– Employ consultants to consult
• Choose courses that relate to the selected model family – chicken and the egg
• Allocate (lots of) time to post course practice
– Set up a project BEFORE the course period
• Plan the provision of support and mentoring
UNFCCC, Bonn, 2018 19© cgemod
‘Rome was not built in one day’
The End
UNFCCC, Bonn, 2018 20© cgemod
top related