Talk:Computable general equilibrium

Major reversion
25 June 2008; I return from overseas to find that the anonymous Econfusion has replaced the entire article mostly written by me. I have now reverted to May 2008. Econfusions replacement article had many virtues, but was (a) too long and not introductory; (b) pushed particular views (eg, the supposed contrast between AGE and CGE models) (c) rather informally and not very well written (d) contained errors: it is inaccurate to say that CGE models "are solved simultaneously by inverting a matrix", or that "every equation used in the SAM must necessarily have an exogenous and an endogenous variable.". GTAP is at Purdue (not Perdue). Not every CGE model uses a SAM. It is not true that "All CGE models are solved simultaneously for a single time period, despite claims to dynamic CGE models" [some CGE models solve for all periods at once].

I think that my original article provided a succinct introduction; Econfusion's longer article projects the reader into a breathless controversy.

Mark Horridge

Horridge (talk) 16:58, 24 June 2008 (UTC)

Major revision, vol. 2
Dear Mark and Wikipedians

Thank you for posting this, I think I have just had the same experience you said you had upon returning from abroad, and seeing all the work I had put into the page having been removed, even as I tried to include as much of your original article in the CGE article, and also the AGE article I have tried to contribute to this project. In response to your comments:

a) You are right that the entry was very long, but these are complicated models, and they have a large associated literature. Rather than create a "history of CGE models" wiki, I felt it might be more useful to include it in the CGE article. Maybe you would disagree with this, and in that case I think we should create a seperate wiki page for the history, and that might be useful?

b) The difference between AGE and CGE models are not 'a particular view' but a historical fact. AGE models were built in Yale to apply arrow-debreu general equilibrium in a numerical way, and was solved through a simplex method. CGE models however came from Leif Johansen and Chenery's work at the Bank, and Taylor and Robinsons work from there. But they are not the AGE models of Scarf and company, (which are technically unsolvable) but they are a derivative of input-outpout models and should be, I think, described as such, particularly in a introductionary or encyclopedic article. I am not sure what other 'particular views' you are refering to, as I felt the subjects I covered were quite balanced. What were missing were work on the modeling done in various countries and the projects of the various centers, and I would have thought that you would have some unique insights on this, which could be fascinating. I don't know if you are willing to write on this here, but I would be interested in opening a dialogue on this if you are?

c) "informal and not well written"... I was aiming for clear and concise, exactly because (as you pointed out) the readers may not be technical specialists, and I thought I was hitting the right balance in explaining technical terms, in readable english. How might we resolve this? This I feel should be resolved, in the context of the fact that CGE models are big and complex models, used across the world for much policy analysis - so it is important to cover them properly.

d) You note that the article "had many virtues" but that there were some errors. I would have thought it more useful to address those small issues, rather than delete the whole article, so again I think we can find a solution here. On the points you raise,

i. The matrix comment is inaccurate, and I was going to change it, but with travel and work I have not had the time. I think we could agree that a CGE model solves a set of equations which have been perturbed from the ones that generate the base solution. Since some of the equations are non-linear one generally has to do more to solve them than just invert a matrix, thus my inaccuracy. You can get a linear approximation to the solution of the perturbed system by using the Jacobian for the base solution but people usually want to be "exact".

ii. exogenous and endogenous variables: Presumably all relations in the system will have an exogenous and an endogenous element, but one could have equations with only endogenous variables, although on the whole, I do not think that in any way negates the point, it is a footnote to be added, with some suggestions as to why one would use these.

iii. yes it's Purdue not Perdue... simple typo.

iv. Again, solving all periods simultaneously as opposed to solving all sectors in one period is a question of choosing a time period of 1 or x, within a chain of CGE models. I think it is still questionable to what extent one can call these models 'dynamic', but that is a point of contention, which I thought I had raised while making the point.

v. regarding SAM's, I am unsure what to reply, as every CGE model is built on a 'empirical base' and that is rooted in a SAM. I agree that your article was succinct, but I think it needed some elaboration especially as it was somewhat loose in what it claimed CGE models could do and did, without explaining the processes and jargon of the models or how it performed those particualr functions. I think that may be of more interest to new users, while still keeping the material fresh enough for model builders to read and discuss, which is where we have to address the controversies that exist.

I really think that there is nothing fundamentally wrong with the piece I had added, and that it was by far more descriptive and explanatory than the existing record, and that it had included the parts which were the core parts.

As for the anonymity being an issue, I am unsure why it should be, as long as the articles are referenced properly.

Best Econfusion —Preceding unsigned comment added by Econfusion (talk • contribs) 21:24, 25 June 2008 (UTC) (Econfusion (talk) 21:30, 25 June 2008 (UTC)) (Econfusion (talk) 11:16, 3 July 2008 (UTC))

Reversion to June version
I have reverted to my June version, since I think that the version substituted by Benjamin Mitra-Kahn is inferior. As far as I am aware, Benjamin is NOT a CGE modeler; his 2008 "Debunking the Myths of Computable General Equilibrium Models", which he has excerpted to create his Wikipedia article, seems to be all he has written on the topic. I find no evidence that he has ever actually RUN a CGE model for practical purposes, for example by participating in one of the CGE-based studies which are often performed by all sorts of agencies. And this shows very clearly, since he writes all sorts of nonsense which no practitioner (such as myself) would ever write.

His article is largely a criticism of CGE models -- from a rather idiosyncratic point of view. Of course, idiosyncratic criticisms are a welcome part of debate, but it seems inappropriate for a one-eyed criticism to form the main article in an encyclopedia. The risk is that Wikipedia is captured by cranks presenting various alternative points of view. I think it is fairer to the reader to present a conservative or 'central-tendency' account of the field as practiced today.

Benjamin's article places far too much stress on the macro closure used in CGE models. For example, he opens with: "CGE models are based on the Keynesian set of macro balancing equations arranged within a social accounting matrix." The truth is that CGE models largely consist of equations modeling much more detailed behaviour, such as coal usage in electricity generation, and that this micro behaviour can be added up to give some (not especially Keynesian) macro relations. The macro assumptions are only a small (but important) part

I do not think many practicing CGE modellers would recognize their own activities in his description. For example, I would estimate that two well-known CGE frameworks, GTAP-based and MPSGE models account for 1/3 or 1/2 of CGE modelling today. But neither makes direct use of a SAM.

Or take again Benjamin's: "Importantly, the share of economic activity that each sector contributes or uses is predetermined by these equations and it is limited because of the numerous convexity assumptions. More specifically the input shares of sectors will not change if the elasticities of substitution are all equal to one[3], and similarly the consumption shares will not change if demands are homothetic with unit price elasticities (again Cobb-Douglas). So a CGE model could not predict, nor deal with any major structural changes like China’s recent boom in manufacturing, or India’s booming service outsourcing sectors. Simply because those productive parts of the economy are given a set percentage of the nations output in the benchmark, that will not change. To make adjustments to this, one would have to post facto change these shares exogenously, but it cannot be incorporated endogenously."

I am not aware of any widely-used CGE model which specifies only Cobb-Douglas for consumption and production.

Benjamin's article is also full of mistakes and inaccuracies. For example: ".....there are as many closures to a model as there are variables to the power of two" Since all variables will be either exogenous or endogenous, and since the number of endogenous variables must equal the number of equations in the model, the number of possible closures is bounded above by the combinatorial "choice" function:


 * $$ {n \choose k} = \frac{n!}{k! \cdot (n-k)!}. $$

where N would be the number of variables, and k the number of equations. (see http://en.wikipedia.org/wiki/Combination ) I say "bounded above" because many combinations would have to be excluded, for example combinations which made all variables in any one equation exogenous.

Or yet again, consider:

"Given the solution based nature of the CGE model, every equation used in the SAM must necessarily have an exogenous and an endogenous variable."

Let's take it apart:

"Given the solution based nature of the CGE model" ... meaning ?? if any??

"every equation used in the SAM" ... the SAM is a table of numbers, containing no equations

"must necessarily have an exogenous and an endogenous variable"

above is clearly false as many, perhaps most, equations in a CGE model follow one of the two patterns:

Demand (eg, for coal by steel industry) = F(prices)*steel output

or

Total Demand for Coal = sum of many individual demands

In each case, both LHS and RHS variables are typically endogenous.

Or consider again the discussion of calibration, which either quite misunderstands the usual procedure, or describes an unusual approach. Benjamin states: "the calibration results themselves are highly dependent on the form of closure" This is quite wrong, as a description of the typical CGE modelling exercise. The typical sequence is: The three steps are quite independent.
 * construction of a balanced flows database, from imperfect data
 * choice of elasticity values, either from external econometric studies, or from thin air
 * choice of closure for policy simulation

All these faults, I suspect, arise because, Benjamin has never done any CGE modelling, and has talked to only one (perhaps Lance Taylor) or a few CGE modellers. But I think he would find it very hard to get a representative sample of CGE modellers to agree that his Wikipedia article was at all balanced, or a useful introduction to contemporary CGE modelling.

I regret that my original article did not mention the early and important Taylor and Black (1974) contribution. That was an oversight -- but in any case my article is not mainly concerned with the history of the field.

I encourage Benjamin to prepare a separate article on "criticisms of CGE modeling" and perhaps another on its history.

Mark Horridge —Preceding unsigned comment added by Horridge (talk • contribs) 05:10, 27 July 2008 (UTC)

Dr. Savard's comment on this article
Dr. Savard has reviewed this Wikipedia page, and provided us with the following comments to improve its quality:

"I think it would be interesting to present a section on linking CGE models with microsimulation models. This is a trend that started at the end of 1990's after Decaluwé et al (1999) and has been growing since. The main objective is to allow analyst to analyse efficiency issues in combination with distributional issues. A few reviews of this combined analytical Framework have been published such as Bourguignon and Spadaro (2006), Davis (2009) and Cockburn et al (2014).

Decaluwe´, B., A. Patry, L. Savard and E. Thorbecke (1999) ‘Poverty Analysis Within a General Equilibrium Framework’, Working Paper 99–09, African Economic Research Consortium, Nairobi.

Bourguignon, F., and A. Spadaro (2006) “Microsimulation as a Tool for Evaluating Redistribution Policies ”, Journal of Economic Inequality, 4 (1); 77-106.

Davis, J.B. Combining Microsimulation with CGE and Macro Modelling For Distributional Analysis in Developing and Transition Countries, INTERNATIONAL JOURNAL OF MICROSIMULATION, 2(1); 49-65

Cockburn, J., Savard, L. and L. Tiberti,. (2014) Macro-Micro Models in C. O’Donoghue (ed.), Handbook of Microsimulation Modelling (Contributions to Economic Analysis, Volume 293), Emerald Group Publishing Ltd. Bingley, U.K..V., pp. 275-304."

We hope Wikipedians on this talk page can take advantage of these comments and improve the quality of the article accordingly.

Dr. Savard has published scholarly research which seems to be relevant to this Wikipedia article:


 * Reference 1: Dorothee Boccanfuso & Veronique Gosselin & Jonathan Goyette & Luc Savard & Clovis Tanekou Mangoua, 2014. "An impact analysis of climate change and adaptation policies on the forestry sector in Quebec. A dynamic macro-micro framework," Cahiers de recherche 14-04, Departement d'Economique de la Faculte d'administration a l'Universite de Sherbrooke.


 * Reference 2: Dorothee Boccanfuso & Marcelin Joanis & Patrick Richard & Luc Savard, 2012. "A Comparative Analysis of Funding Schemes for Public Infrastructure Spending in Quebec," Cahiers de recherche 12-10, Departement d'Economique de la Faculte d'administration a l'Universite de Sherbrooke.

ExpertIdeasBot (talk) 18:08, 1 July 2015 (UTC)

Dr. Roson's comment on this article
Dr. Roson has reviewed this Wikipedia page, and provided us with the following comments to improve its quality:

"Pasted: “A CGE model consists of (a) equations describing model variables and (b) a database (usually very detailed) consistent with the model equations.” Comment: This is not correct, both because all models have property (a) and databases are not part of any model. Should be replaced by: “A CGE model is a non-linear mathematical system of equations, providing a disaggregated representation of national, regional and multi-regional economies. The system includes market clearing conditions and accounting identities, to trace the circular flow of income and inter-sectoral linkages inside the economic system. Parameters describing the structure of the economy are typically estimated through calibration techniques from a database (usually very detailed) consistent with the model equations.”

Own citation after item 2 (imperfect competition (e.g., monopoly pricing)): Roson, Roberto “Introducing Imperfect Competition in CGE Models: Technical Aspects and Implications”, Computational Economics, vol.28, pp.29-49, 2006.

Pasted: “Other elasticities may belong to the constant elasticity of substitution class. Amongst these are Armington elasticities...” Comment: Incorrect, CES are production and utility functions, not elasticities. Should be replaced by: “Elasticity values are very important in determining the response to external shocks in CGE models, especially when technologies and consumers’ preferences are described by means of constant elasticity of substitution functions. In particular, the substitution possibilities among heterogeneous imported goods are determined by elasticity parameters referred to as “Armington elasticities”.

Add a paragraph: “CGE models are sometimes a part of a broader Integrated Assessment Model (IAM) framework, where socio-economic models are integrated with physical and natural sciences models. This is often the case in IAM of climate change, where an economic CGE module determines the level of greenhouse gases emissions, and a climate module simulates the change in the climate variables, affecting the economic impacts of climate change [link]” Possible (own) citation: Roson, Roberto, van der Mensbrugghe, Dominique “Climate Change and Economic Growth: Impacts and Interactions”, International Journal of Sustainable Economy, vol.4(3), pp. 270-285, 2012.

Pasted: “Recursive-dynamic CGE models are those that can be solved sequentially (one period at a time). They assume that behaviour depends only on current and past states of the economy.” Last part should be replaced by: “They assume that behaviour depends only on current and past states of the economy, or that future states can be perfectly inferred from currently available information.”

Pasted: “Within the latter group dynamic stochastic general equilibrium models explicitly incorporate uncertainty about the future.” Add: “However, DSGE models are typically much more aggregated than standard CGE models. They focus on dynamic paths of adjustment in the economy after changes in expectations, rather than on variations of the economic structure. For this reason, they should be regarded as a separate class of models.”

Solution techniques. Add at the beginning: “Even if a neoclassical general equilibrium state could be stated (in the absence of externalities or other market failures) as a solution of an optimization problem, CGE models are normally formulated as non-linear systems of equations. Therefore, in principle, any mathematical software that can solve non-linear systems could be employed to find a solution for a CGE model.”

Pasted: “Today most CGE models are formulated and solved using one of the GAMS or GEMPACK software systems.” Add: “A major advantage in using the latter packages is that the model description is made independent of solution algorithms.”

Add a new section “Key characteristics, advantages and disadvantages of CGE models”: “On the positive side, the high level of disaggregation of CGE models makes it possible to understand the complex structural interdependencies of globalized economic systems, and furthermore opens the way to a precise delineation of the effects of exogenous impacts on a finely detailed industrial and geographical scale. Also, the rich output of the simulation exercises can be further processed (within or outside the main model), which allows to undertake sophisticated analyses, e.g., on welfare impacts. On the negative side, CGE models are highly data-demanding, having been originally conceived for short-run policy analysis. Their use over a much longer time horizon may be problematic if significant changes in technology or consumer preferences substantially alter the economic structure from its current state. For this reason alone, they should not be regarded as forecasting tools or used to construct future economic scenarios.”"

We hope Wikipedians on this talk page can take advantage of these comments and improve the quality of the article accordingly.

Dr. Roson has published scholarly research which seems to be relevant to this Wikipedia article:


 * Reference 1: Roberto Roson & Kazuhiko Oyamada, 2014. "Modeling Firm Heterogeneity in International Trade: Do Structural Effects Matter?," Working Papers 2014:12, Department of Economics, University of Venice "Ca' Foscari".


 * Reference 2: Maria Berrittella & Katrin Rehdanz & Arjen Y. Hoekstra & Roberto Roson & Richard S.J. Tol, 2006. "The Economic Impact Of Restricted Water Supply: A Computable General Equilibrium Analysis," Working Papers FNU-93, Research unit Sustainability and Global Change, Hamburg University, revised Jul 2006.

ExpertIdeasBot (talk) 18:57, 27 June 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified one external link on Computable general equilibrium. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20090228052444/http://www.econ.cam.ac.uk/research/cgp/index.html to http://www.econ.cam.ac.uk/research/cgp/index.html

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 20:18, 11 August 2017 (UTC)

Feb 2024 edits and suggestions
- Edits to the the lead section - the aim here is to have 2 or 3 paragraphs summarising everything in the article below

- Reorganised text into main features, history, and Areas of use sections

- Added some new text on CGE modelling analysis of climate policy

Could anyone help to develop the 'Areas of use' - especially in climate policy ? (see also Economic analysis of climate change)

I also find that discussion about related modelling approaches ACE and DSGE and their connection with CGE are missing

(PS I'm not a CGE modeller but I am trying to improve articles related to climate change, economics and finance) Richarit (talk) 19:06, 27 February 2024 (UTC)