Economic Equilibrium Analysis with GAMS/MPSGE

Thomas F. Rutherford University of Colorado INFORMS Presentation November 18, 2002

GAMS/MPSGE: A Mathematical Programming System for General Equilibrium Analysis specically designed for applied general equilibrium analysis, including both models represented as systems of equations and those which involved complementarity between inequalities and bounded variables. system is particularly useful for large, complex models based on benchmark equilibrium datasets GAMS/MPSGE provides a highly structured framework for inexperienced analysts, yet GAMS/MPSGE models can be customized through the use of auxiliary variables.

Economic Models in the Policy Arena Economic models produce results which can play a central role in political dialogue. Although economic models are based on formal mathematics, it is important to recognize that the origins of economic analysis are in social philosophy rather than physical science. Everyone participates in economic transactions, so non-specialist audiences can be inuenced by populist appeals to common sense.

Economic Equilibrium Ideas Agents in an economic model include consumers, producers and governments, collectively representing all participants in market transactions in a given economy. A central concept in economic equilibrium models is that agents optimize subject to constraints. Systematic errors are logically inconsistent with individually rational choice. A typical starting point for dynamic economic models is that consumers are fully informed and hold consistent expectations of the future. This approach dierentiates economic models from models of physical systems.

GAMS/MPSGE Equilibrium Framework Variables

p RN Prices for all goods and factors (possibly indexed by commodity, sector, region, household, time period etc.)

y RM Production activity levels (also indexed) M RH Income levels for each consumer in the model

Given Functions and Data

pj is a vector of producer prices, net (gross) of applicable taxes for outputs (inputs, resp.) rj (j ) is the unit revenue function for sector j p p cj (j ) is the unit cost function for sector j dih(p, Mh ) is the demand function for household h, derived from budgetconstrained utility maximization. By denition, these functions satisfy Walras law: pi dih(p, Mh ) = Mh .
i

ih, jh are matrices of commodity endowments and tax revenue allocations.

Dual Equilibrium: Zero Prot cj (j ) rj (j ) p p Primal Equilibrium: Market Clearance rj cj pij pij yj 0

yj +
h

ih
h

dih(p, Mh)

pi 0

Income Balance Mh =
i

pi ih +
j

jh yj
i

(ij pij ) p

rj cj pij pij

Calibrated Functions Data for specication of functions is typically based on the rst two terms in a Taylor approximation: U = pj benchmark producer inputs (the use matrix) aij i p M = pj benchmark producer output (the make matrix), aij i p dih = dih(, Mh) benchmark consumer demands at reference prices p ih, jh benchmark initial endowments and tax shares.
r c

Mission for Public Income Colombian Fedesarrollo 1997 input-output table supplemented with additional data from 1999, 2000, 2001. 56 production sectors 6 categories of labor: ufs ufn utc umc rsw rnw Urban formal salaried work Urban formal non-salaried work Urban traditional contract work Urban modern contract work (consulting) Rural salaried work (organized farming work) Rural non-salaried work (farming)

 10 representative households h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 n 2980 1980 1410 800 520 340 210 180 100 90 n% 33 22 15 9 6 4 2 2 1 1 c 2980 2701 4444 5661 6785 8145 9931 11615 14936 22000 c/n 907 2245 4029 8522 15773 29615 54643 82257 226664 246362 $/day 2 4 7 16 29 54 100 150 414 901

Tax Revenue Revenue 5.6 4.3 2.1 1.4 1.0 0.9 0.8 0.7 -0.034 16.8 6.6 20.1 12.5 9.5 % 33 26 12 8 6 5 5 4 -0.2 100 Base 216 24 216 24 51 200 200 51 17 104 Tax Rates (%) Collected Posted 2.6 16.5 17.8 34 1.0 5.8 5-15 1.9 0.5 0.4 1.4 17-34 -0.2 6.3

VAT Corporate Income Excise Taris Payroll Indirect Output State/Local Individual Income Subsidies Total Social Security Central Govt. Income: Local Govt. Income: Soc. Security Expend:

Equilibrium Framework

Eg

m k

Ys

Ag

Ch
j

Government Mg
e f g h

Households

Key:

Physical flow of goods:: Financial flows of factor earnings, tax payments and transfers:

Model Formulation
Sets: s, g h Activity Levels: Ys Ag Ch K I G Production activity level Aggregate supply to domestic and export markets Aggregate consumption demand by household h Capital stock Investment Public demand Sectoral and commodity identiers Households Labor types

Prices: e rK w pY g pA g pC h pG pI Real exchange rate Rental price of capital Wage rate for labor type Supply price of good g (gross of indirect taxes) Market price of good g (gross of excise and VAT) Consumer price index Public provision price index Investment cost index

Other variables: GB Capital stock Government external balance

Leontief Demand and Supply Coecients aM gs aU gs aG g aI g a gg Output of good g per unit activity of sector s, the make matrix. Input of good g per unit activity of sector s, the use matrix. Demand for good g per unit of government activity Demand for good g per unit of aggregate investment Trade and transport margin net demand per unit aggregate supply of good g

Cost and Revenue Functions

cY ( w s , r K ) s
A p Rg (A , pX ) r g

Unit cost of value-added in sector s (Ys) Unit revenue per unit of aggregate supply (Ag ) Unit cost of aggregate supply (Ag ) Unit cost of nal demand for leisure and goods (Ch)

cA(pY , pM ) g g g cC (wh , pA ) h

Arbitrage (Zero-Prot) Conditions Domestic production (Ys ): pY aM = gs gs

g g

pA aU + cY (w s , r K ) gs gs s

Aggregate Supply (Ag ):

A Rg (A , pX ) = cA (Y , pM ) + pg g g pg g g

pA a g gg

Consumption Cost (Ch): pC = cC (pA , w h ) h h Cost of Investment (I): pI =

g

aI pA g g

 Cost of Public Provision (G): pG =

g

aG pA g g

Market Clearance Conditions Domestic Output Ys aM gs

s

= Ag

cA g pY g

Domestic Demand Ag =
g

a Ag gg

+
s

aU Ys gs

+
h

cC h + GaG + IaG g g pA g

Labor Markets L
h h

=
h

cC Ch h + w h

cY Ys s w s

Capital Market
h

Kh =
s

cY s Ys K r

 Household Demand Ch = Investment-Savings I=

h

Mh cC h

Sh + SG

Current account BG +
h

Bh +
g

Ag

A Rg

pX g

=
g

Ag

cA g pM g

Income Balance Household Income Mh = Government MG =

g

L h w h + K Kh + eBh Th pI Sh r

Ag +
g

(M pM ) + p g pM g g
A Rg

cA g

Ag

(X pX ) p g pX g g

+
s

cY Ys s (w s w s ) w s cC h Lh w h (w s w h )

+
h

+...

Auxiliary Constraints Steady-state model (): rK q= I =1 p Budget Balance (BG ) G=1

Data Management in GAMS

* * set Units of the 1997 SAM are Millions of 1997 Pesos (current price). After scaling, we get Billions of Pesos r SAM Rows /11*210/;

alias (r,c); sam(r,c) = sam(r,c)/1000; parameter samchk Check of SAM consistency; samchk(r) = round(sum(c, sam(r,c)-sam(c,r)), 5); display samchk; set Goods(r) /11*68/, Labor(r) /128*133/, Households(r) /186*195/, Firms(r) /196,200*201,202*206/, Investment(r) /208*210/; Sectors(r) /70*127/, Capital(r) /134/, Government(r) /135*158,159*185,197*199/, Row(r) /69,207/,

Aggregated Social Accounts

G Goods Sectors Labor Capital Households Government Firms row Investment 199.0 74.3 35.5 74.3 9.8 23.8 2.1 7.6 3.5 24.5 0.1 9.6 9.5 0.2 5.4 6.2 27.2 2.1 1.3 -0.2 10.7 8.7 15.3 2.8 14.5 4.8 0.8 0.2 5.8 S 87.1 L C H 78.8 G 24.2 F 0.2 R 16.8 I 25.5

25.5

$model:sam97 $sectors: y(s) x(g) a(s) gov inv con(h) $commodities: py(g) pa(g) pc(h) w(l) rk pfx $consumers: ra(h) govt

! ! ! ! ! !

Production Domestic supply Aggregate supply Public expenditure Investment Household consumption ...

! ! ! ! ! !

Output price Aggregate price Household consumption Wage rates Return to capital (gross of tax) Foreign exchange ...

! !

Households Government

Production Functions
$prod:y(s) s:0 o:py(g) i:pa(g) i:w(l) i:rk $prod:x(g) t:4 o:pd(g) o:pfx i:py(g) $prod:a(g) s:0 dm:4 o:pa(g) i:pmg(gg) i:pd(g) i:pfx va:1 t:0 q:make(s,g) q:use(g,s) q:ld0(l,s) q:kd0(s)

a:govt t:(tcm(s)+tif(s)-sub(s)+ty(s)) a:govt t:-vat(s) p:pl0(l,s) a:govt t:pyrl(l,s) va: va:

q:d0(g) p:1 q:x0(g) p:px0(g) q:y0(g) p:py0(g)

a:govt t:crt(g) a:govt t:tp(g)

q:a0(g) p:pa0(g) a:govt t:txs(g) t:vat(g) q:margin(gg,g) q:d0(g) dm: q:m0(g) p:pm0(g) dm: a:govt t:tm(g)

Dual Equilibrium: Zero Prot cj (j ) rj (j ) p p Primal Equilibrium: Market Clearance rj cj pij pij yj 0

yj +
h

ih
h

dih(p, Mh)

pi 0

Income Balance Mh =
i

pi ih +
j

jh yj
i

(ij pij ) p

rj cj pij pij

Equilibrium Framework

Eg

m k

Ys

Ag

Ch
j

Government Mg
e f g h

Households

Key:

Physical flow of goods:: Financial flows of factor earnings, tax payments and transfers:

Corresponding GAMS Algebra

* Demand for domestic input into Armington production:

DEF_A_Y(d,i,re)$(vd(d,i,re)).. A_A_Y(d,i,re) =E= vd(d,i,re) * {[ ((1- thetaa_m(d,i,re)) * py(i,re)**(1-esubdm(i,re)) + thetaa_m(d,i,re) * PM(i,re)**(1-esubdm(i,re)) )**(1/(1-esubdm(i,re))) ]/ (py(i,re)) }**esubdm(i,re);

Constraints Associated with Auxiliary Variables

$constraint:tau gov =e= 1; $constraint:kf pinv =e= pinv0 * rkf;

Proportional Tax Increases: Steady-State Model Value-Added Taxes ACF() Revenue 0 1 Pesos(T) %GDP 1.29 1.36 1.47 1.1 0.6 1.33 1.40 1.51 3.2 1.9 1.37 1.44 1.55 5.3 3.2 Import Taris ACF() Revenue 0 1 Pesos(T) %GDP 2.03 1.98 1.92 0.2 0.1 2.09 2.03 1.97 0.5 0.3 2.14 2.08 2.02 0.8 0.5

Rate 1.2 1.6 2.0

%Yield 95 95 94

Rate 1.2 1.6 2.0

%Yield 68 64 61

Tax Revenue Yield in the Rened Petroleum Market

3.5

Demand
3

2.5

Tax Rate=t0 2, Tax Revenue=T0(1+yield)

1.5

Tax Rate = t0, Tax Revenue = T0

Supply
0.5 0 0.2 0.4 0.6 0.8 1

Quantity index