III. AGE analysis of trade, policy reform and environment

III-A

1

III. AGE analysis of trade, policy reform and environment

A. Core concepts and structural features of AGE models

B. Country case studies

Sources:* Shoven and Whalley 1984* OEE Chapter 5Lee and Roland-Holst 1997 & other CGE studies

III-A

2

Concepts and structural features

1. Overview and a simple AGE structure2. Solution procedures for ‘Johansen’

models3. Incorporating environmental analysis4. Dealing with institutional issues5. Dealing with political economy

III-A

3

Overview of AGE models

• Describe Walrasian equilibria in fairly detailed manner--sufficient to support policy claims– Too large to be solved analytically; must use

numerical solutions instead– But structure and results depend on same

theoretical foundations• Advantages and disadvantages of size.

III-A

4

Pros and cons of AGE modeling

• √ Capture economy-wide mechanisms and implications

• X limitations– ‘Time’ is not explicitly taken into account: major limitations for

analysing impact of exchange rate changes– ignores risk/uncertainty issues; credit market imperfections

ignored– aggregations can mask important differences; AGE models must

be used in conjunction with in-depth, ‘micro’ research and analyses

– resource intensive: only worthwhile when economy-wide effects are deemed sufficiently important!

III-A

5

An N-good, F-factor economy

• General structure• Equilibrium conditions• Closure rules and decisions

III-A

6

P commodity prices (N) W mobile factor prices (F)

R sector-specific factor prices (N) Y dom. commodity supplies (N)

X mobile factor demands (N×F) D dom . final dem ands (N)

S net im ports (N) V factor endowm ents (F)

U aggregate utility (1). φ Foreign currency exch. rate (1)

Variables in the basic model

III-A

7

-- Suppose V and P are given, and let φ = 1 be the num é raire price.

-- Aggregate revenue is given by G(P,V ) = max{P⋅Y | V } . From FONC:

Yφ = Yφ (P, V ) (φ = 1, ..., N), (5.1)

and the prices of m obile and specific factors:

Wi = Wi(P, V ) (i = 1, ..., F), (5.2)

Rφ = Rφ (P, V ). (φ = 1, ..., N), (5.3)

-- Each sector is a price-taker in factor m arkets. Therefore, the output level that

m aximizes revenue is also the cost-mi nimizing level, and from FONC

of the sectoral cost mi nim ization problem Cφ (W , Yφ) = min{W ⋅X | Yφ),

we obtain demands for intersectorally m obile factors:

Xiφ = Xiφ (W , Yφ) (i = 1, ..., F; φ = 1, ..., N). , (5.4)

III-A

8

Domestic final demands for each commodity are found from the

expenditure minimization problem E(P,U) = min{P⋅D | U}:

D φ = D φ (P, U) (φ = 1, ..., N). , (5.5)

Net trade volum es are determ ined by market-clearing conditions:

Sφ = D φ – Yφ (φ = 1, ..., N), , (5.6)

where Sφ > (<) 0 indicates a net import (export) good. Im port prices are set

in world m arkets, while for M exportables (M ≤ N), prices are set by

inverse foreign dem and functions:

Pk = Pk (Sk) (k = 1, ..., M). (5.7)

Finally, the m odel is closed by an aggregate budget constraint:

E(P,U) = G(P,V ) (5.8)

III-A

9

Closure

• No. of equations must match endog. vars.• In (5.1)-(5.8): 4N + F + FN + M + 1 eqns.• But we have 5N + 2F + FN + 2 variables.• Must choose N - M + F + 1 exog. vars

– Declare V exog; (N - M) elements of P, and f.• Now (5.1)-(5.8) solve for Y, W, R, X, D, S

and U as endogenous variables.

III-A

10

Closure rules and decisions

• Other closures are possible– ‘Neoclassical’ closure has all domestic prices

flexible– Alternatives: e.g. fix wages, allow

unemployment in labor market. • These choices reflect our beliefs or

observations about the real world.

III-A

11

Other features

• Can add in– Intermediate inputs– Products distinguished by source– Different kinds of labor– Many sources of final demand– Trade and transport ‘margins’– Tariffs, taxes, and other policies … etc.

• Again, real-world conditions should motivate these.

III-A

12

Solving the model: the ‘Johansen’ AGE structure

• First-order approximations to changes in variable values

• Models solved in proportional (percentage) changes of variables, or ‘hat calculus’.

• Advantages: – Models are linear in variables– Parameter values are intuitive and accessible (shares,

elasticities)– Simulation results are additive in separate shocks

III-A

13

Features of Johansen models

• Parameter values are shares and elasticities• Quick checks:

– Homogeneity & ‘balance’ of underlying data base.• Solution is by matrix inversion

– Entire model is a system of linear equations• Examples of Johansen-style models:

– ORANI (Australian economy)– GTAP (international agricultural trade)– Model OEE, Ch.6

III-A

14

The AGE model

1. Output supplies and factor demands (N+FN)2. Zero pure profits in production (N)3. Factor market clearing conditions (F)4. Consumer demands for goods (N)5. Net trade in commodities (N)6. Export prices (M)7. Aggregate budget constraint (1)

III-A

15

Variables solved within model

1. Output supplies and factor demands (N+FN)2. Returns to sector-specific factors (N)3. Mobile factor prices (F)4. Consumer demands for goods (N)5. Net imports (N)6. Export prices (M)7. Aggregate real income (1)

III-A

16

Data base

• The model in proportional change form uses data on production, consumption, trade,… all in the form of – Shares (e.g. employment shares by sector)– Elasticities (e.g. parameters of demand and

supply functions). • Easy to check ‘balance’ of data base• Easy to interpret results.

III-A

17

Environmental analysis in GE

• Most AGE models constructed for more general analytical purposes: environmental structure is added later

• Given uncertainty about env. variables and valuations this may be appropriate!

• Industrial emissions• Natural resource degradation• Questions about institutions.

III-A

18

Industrial emissions: AHTI

0 20 40 60 80 100 120

Feed MillingMeat & Meat Prod.

Beverages and TobaccoAnimal Feeds

Sugar Milling/RefiningOther Foods

Transport EquipmentMilk and Dairy

Electrical MachineryOils and Fats

Textile & KnittingMetal Products

Coal & Petroleum Prod.Misc. Manufacturing

Other TextilesCement & Non-Metallic

Wood Products Garments

Non-Ferr. Basic MetalsPaper Products

Rubber/Plastic/Chem Prod.Fertilizer

Linear AHTI value.

III-A

19

Deforestation & land degradation

• Commercial and non-comm’l deforestation: does timber have market value?– Non-comm’l deforestation is driven by search

for land, and responds to changes in the marginal valuation of land in agricultural production...

– … although institutional setting also matters (more later)

III-A

20

Land degradation

• Hard to measure, and problems of aggregation.

• Can use information on erosion rates by crop, together with land use data, to build ‘baseline’ data set.– Then erosion changes can be inferred from

changes in land use• Production externalities: technical ‘regress’.

III-A

21

Institutional issues

1. Trees may be cut (or planted) to establish property rights over land.

2. In open access forests (non-commercial), opportunity cost of forest is set by ag. land values and clearing costs.

3. In commercial forestry, timber harvesting/replanting also depends on property rights.

• Will an increase in timber prices promote or retard tree-felling in aggregate? Depends on prop. rts.

III-A

22

Institutions in AGE models

• Can incorporate open access (quantity vs. price adjustment in market clearing)

• Distinguish between commercial forestry and land colonization by farmers

• In latter case, implied land values indicate pressures for deforestation.

III-A

23

Dealing with political economy

• Economists’ welfare weights seldom coincide with those of policy-makers!– This extends to valuations of environment-

economy tradeoffs. • AGE results can be re-cast to reflect

alternative sets of priorities…