A General Equilibrium Analysis of Income Distribution and Development Policy
Sibel Giiven
Middle East Technical University
Industrial Engineering Department, Ankara, Turkey
This paper presents an empirical analysis of the interaction between growth and income distribution for the
Turkish economy over the period 1973-1979. The analysis is based on a multi-period, multi-sector computable
general equilibrium model constructed for the Turkish economy, that permits the evaluation of distributional as well
as growth consequences of alternative policies. Along with empirical testing of the model, three sets of experiments
are conducted with the model. Counter factual trade policy experiments, alternative wage policy experiments and
simulations investigating the distributional and growth consequences of alternative potential redistributive policies
indicate that the size distribution of income is stable and very difficult to change with the policy interventions
considered. Agricultural terms of trade is one of the most important determinants of reducing overall poverty, and
policy instruments conducive to higher growth is not necessarily in conflict with the policies aiming at more even
distribution of income.
1. Introduction
The necessity of modeling distributional issues as ‘an integral part of policy evaluation and
development planning within the framework of multi-sector CGE models have been emphasized in
recent country specific empirical research. (Adelman et al 1978; Taylor et al 1980; Bourguignon
et al 1983) The major thrust of the emphasis is that fixed price input output models and linear
programming models do not permit unified, endogenous treatment of the determinants of both
growth and income distribution. Moreover, policy focus of development planning in the 60's,
where the initiation and achievement of high rates of growth in a consistent framework that would
lead to a better distribution of income, proved inadequate in the light of empirical findings that
followed. (Adelman et al 1973; Chenery et al 1975; Ahluwalia 1976) Controversial theoretical
viewpoints and empirical findings on the relationship-between growth and income distribution in
the short and medium runs (Cline 1972, 1975) further aggrevated the need for endogenous
treatment of the circular flow in the economy in the context of country specific CGE models.
With the same purpose this paper presents an empirical analysis of the interaction between
growth and income distribution for the Turkish economy over the period 1973-1979. The analysis
is based on a dynamic multi-sector CGE model constructed for the Turkish economy. Basic
features of the model is described in section 2. The model is calibrated to 1973 base year data and
empirically tested over the period 1973-1979. Section 3 presents validation results and a brief
historical background on Turkey clarifying the purpose of the experiments conducted.
Distributional implications of actual policies pursued are outlined in Section 4. Alternative trade,
wage and redistributive policies simulated by the model are reported in Section 5.
2. Basic Features of the Model
The current model is an extended and modified version of CGE models developed by Dervis
and Robinson (1978) and Urata and Lewis (1983) in conjuction with the World Bank to assess
Turkish economic performance. The modifications are on the treatment of national, public and
institutional accounts in a manner compatible with the flow of funds in Turkey's official planning
process. Extensions concern mapping functional incomes into household incomes that permit the
analysis of distribution of income to households. A thorough presentation of the model is
contained in (Giiven 1986, Celasun et al 1981).
The model is a simulation model that simulates the price-responsive optimizing behaviour of
decentralized decision making units such as households, firms and the government, and the
operation of product, factor and foreign exchange markets so as to determine outputs and product
prices, employment and factor prices, and the exchange rate that clears each market. Market
clearing or equilibrium is attained with prices varying in response to supply and demand
conditions in each market so as to reduce all excess demands to zero. Accounting consistency is
maintained between the production and expenditure accounts both at the micro and macro levels.
Foreign trade is explicitly modeled where equilibrium is sought between receipts and expenditures.
Page 228
System Dynamics '91 Page 229
The model is structured on the basis of input-output description of the economy, and consists
of eleven sectors, three labor categories (Agricultural labor, organized urban labor, unorganized
urban labor), five economically active groups (the three labor categories, agricultural and non-
agricultural capitalists), and five household categories (farmer 1, farmer 2, urban poor, urban
middle:and urban rich). Within this main framework product market clearing constitutes the thrust
of the formulation and factor and foreign exchange market clearing are attained in the order of
causality of the circular flow in the economy described below.
2.1. Product Supplies and Factor Market Clearing
Production in each sector is assumed to be carried out by several firms with the objective of
maximizing profits under perfect competition subject to multi-level production functions. Demand
for intermediate inputs are given by fixed input output coefficients: Within a period the supply of.
each labor type is fixed and agricultural labor is assumed to be employed only in agriculture and
urban labor categories in non-agricultural activities. Capital is. a fixed coefficient aggregation of
capital goods and assumed to be immobile across sectors in a period. For agriculture, composite
primary input is a CES (constant elasticity of substitution) aggregate of agricultural labor and
aggregate capital. In non-agricultural sectors, two urban labor categories are combined with. a
CES function to form aggregate labor and then combined with aggregate capital to form a CES
aggregate of composite primary input. For an arbitrary non-negative commodity price vector, the
solution to the optimization problem yields product supplies, employment by sector and labor
type, and factor incomes including residual value added accruing to capital.
2.2. Enterprise and Household Accounts
Disposable income of economically active labor categories and agricultural capitalists are
derived from factor incomes by making the necessary adjustments for direct taxes, social security
contributions to, and other budget revenue and transfer payments of the government. Non-
agricultural capital income is first transformed into institutional income by deducting social security
and foreign interest payments, from which institutional taxes are payed,.and.a fixed proportion
saved. Residual enterprise income accrue to non agricultural capitalists and determine their
disposable incomes after the adjustments for taxes and transfers. Income share of the
economically active groups in the disposable income of the identified household categories are
exogenous to the model. They are estimated by aggregating and rescaling the data provided in
Dervis and Robinson (1980), compiled from 1973 SPO-Hacettepe Survey of Household Incomes
for Turkey. Disposable income of each household category is then arrived at by applying these
exogenous shares to the disposable incomes of the economically active, summing over the
economically active groups, and adding workers’ remittances in fixed proportions. Thus
household incomes are endogenously determined as an algebraic_sum of the factor incomes solved
for by the model.
Following (Dervis et.al 1978; Celasun 1975) an "additional financing" mechanism is modeled
that is assumed to transfer proximate forced savings from households to public sector capital
accounts through monetary mechanism described below.
2.3. Public Sector Accounts and Additional Financing Requirement
Public sector collects direct tax, social fund contributions, factor income and other budget
revenue from the five economically active groups and enterprises and makes current transfers to
them in fixed proportions. When indirect taxes on domestic output and foreign. trade (net of
subsidies) are added and government share of foreign interest payments are deducted from this
flow, public disposable income is obtained. Since institutional, direct and indirect taxes are
functions of endogenously determined factor incomes and outputs, public disposable income is
also endogenous. In the official Five Year Plans and Annual Programs prepared for Turkey, both
public investment and public consumption are treated as target variables to be attained, and
therefore specified as exogenous variables of the model. Public savings, determined as the
residual public disposable income after public consumption, and exogenously specified levels of
domestic and foreign borrowing of the public sector need not equal public sector capital
expenditures defined as the sum of public investment and direct capital transfers. The difference
gives the additional financing requirement of the public sector and is assumed to be mobilized from
externally: specified reserve decumulation and from households in: the form of forced savings.
Page 230 System Dynamics '91
2.4, Treatment of Foreign Trade and Balance of Payments
In the model the commodities produced and consumed are assumed to be tradeable goods. In
determining export volumes from domestic production two specifications are used. For
agriculture and mining sectors the small country assumption is made where domestic price of
exports are determined from world prices. In these sectors the share of exports in domestic
production is a function of the ratio of domestic prices to exports. In other sectors it is assumed
that the world price of exports is an aggregate price and is different then the world price of Turkish
exports determined endogenously from domestic prices, exchange rate and export subsidy rates.
In these sectors the demand for Turkish exports is defined to be a function of the ratio of dollar
prices of Turkish exports to world prices, and the price elasticity of demand for exports.
On the import side, the small country assumption is made. Furthermore, domestically
produced commodities and imports of the same sector classification are treated as imperfect
substitutes. First order conditions of the CES trade aggregation function of imported and
domestically produced goods (composite good) determine the demand for imports as a function of
their relative prices and elasticity of substitution between them.
Import demands determined as such are desired imports. Actual amounts imported depend on
the availability of foreign exchange. Total foreign exchange available to be spent on imports is the
sum of foreign exchange receipts from exports, short and long-term borrowing and reserve
decumulation less the interest and project credit service payments of the private and public sectors.
Depending on the specification of the nominal exchange rate different adjustment mechanisms are
modeled that would clear the foreign exchange market. Under the flexible exchange rate
specification nominal exchange rate is allowed to vary to.equate the demand for foreign exchange
to the supply of it. Under a non-liberal trade regime, where the exchange rate is externally
specified, the adjustment mechanism is quantity rationing. Following Dervis et al (1978) a
rationing factor is determined as the ratio of available foreign exchange to total desired imports and
all sectoral imports are rationed by this ratio. This is the specification used in the historical
simulation since it resembles what has actually taken place in Turkey during the period under
consideration.
2.5. Final Demand and Equilibrium in Product Markets
Intermediate demand in product markets are determined by fixed coefficients. Domestic final
demand is comprised of private consumption, public consumption, fixed investment and stock
changes.
Total nominal investment (including stock changes) is savings driven and defined to be the
sum of public, institutional, external, and household forced and voluntary savings. Total private
investment is endogenously determined as the difference between total and public investments.
Forced savings are assumed to be mobilized from household incomes in fixed proportions
before voluntary savings. Private consumption expenditure is residual disposable income after
forced and voluntary savings. Sectoral private consumption demand by household type is based
on the linear expenditure system (LES), and sectoral public consumption is determined in fixed
proportions.
Following Urata and Lewis (1983) total investment funds are distributed among sectors by
differentiating between public and private investment allocation patterns. Sectoral public
investment shares are externally specified whereas private shares are endogenously determined in
the intertemporal phase of the model as a function of sectoral deviation of rental rates from average
rental rate in the economy. Investments by destination are transformed into investments by origin
through the use of predetermined capital composition coefficients. Sectoral investment in stocks
are determined using stock-output ratios.
Thus effective sectoral domestic demand is obtained as the sum of sectoral private and public
consumption, intermediate demand from each sector, less import demand for each commodity.
Domestic sectoral demand estimated as such is a function of the initial arbitrary nonnegative price
vector that determined sectoral supplies explained in Section 2.1. Since equilibrium in product
markets require sectoral supply to be equated to sectoral domestic demand, the problem reduces to
one of solving for the market clearing prices that reduce all excess demands to zero. The solution
System Dynamics '91 Page 231
algorithm employed is due to (Powell 1970) and uses information about the derivatives of excess
demand equations.
The model, being a Walrasian general equilibrium model, is homogenous of degree zero and
solves for relative prices of commodities and factors, relative to the specified aggregate price level.
This specification implicitly assumes that the determinants of the aggregate price level and relative
prices are independent of each other, and interacting only through monetary mechanisms via
additional financing requirement of the public sector and the transfer mechanism of forced savings.
The exogenous variables of the static CGE model outlined above are updated in the
intertemporal linkage model using exogenous projections,growth rates,and linear interpolation
between base and terminal year shares depending on the variable beign updated.
3. Model Validation
The CGE model presented in the preceding section is validated through simulating the model
over the period 1973-1979 in two year intervals and comparing models estimates of major
economic variables with their actual values. The purpose of the historical validation is twofolds;
(i) testing and demonstrating the capability of the model in simulating the effects of a medium-term
policy package on the sectoral and macro performance of the Turkish economy (ii) parameter
estimation.
A CGE model has substantial data requirements and not only the available data is inconsistent
and insufficient but most of the behavioral and structural parameters required by the model are not
available. Thus the base year data was calibrated so that inconsistencies were removed, and the
values of the parameters over time were estimated by solving the model over the period 1973-1979
several times, adjusting the values of the parameters at each iteration so that model's estimates of
major economic variables were closer to their actual values.
Table 1. Comparison of Actual and Model Estimates of GDP and GNP
1973 1975 1977 1979
Actual Model Actual Model Actual Model Actual = Model
Nominal Value Added
(millionTL current prices) 265661 265658 468382 467881 796123 796179 2015305 1981012
Sectoral Dist.of Nominal Value Added (%)
Agriculture 27.54 27.84 29.06 28.74 27.61 = 26.63 23.11 22.43
Mining 1.38 0.96 1.27 0.88 1.80 0.86 1.54 0.92
Manufacturing 17.38 17.50 17.05 1749 16.29 17.89 20.68 24.77
Construction 5.56 5.56 5.26 6.48 5.29 6.55 5.15 4,55
Infrastructure 14.98 14.98 14.71 13.92 15.12 14.95 15.76 13.57
Services 33.16 33.16 3265 3249 3390 33.12 33.75 33.77
GDP (f.c) 100.00 100.00 100.00 100.00 100.00 100,00 100.00 100.00
Index of real,GDP(1973=100) 100.0 100.0 118.2 118.7 134.0 135.2 136.6 135.5
Nominal GNP (m.p.
million TL) 309829. 309405 535771 533165 872894 878390 2199520 2190777
Structure of Nominal GNP (m.p.)
Private Consumption 67.94 68.03 70.32 70.13 68.89 68.72 66.65 68.93
Public Consumption 11.88 11.90 11.92 11.98 = 13.89 13.66 13.40 13.42
Total Private Investment 9.43 9.20 1064 10.74 11.00 11.52 8.86 7.06
Total Public Investment 8.62 8.63 12.29 12.34 13.24 13.02 13.10 12.71
Total Resources 97.87 97.76 105.18 105.19 107.02 106.93 102.01 102.13
Current Deficit 2.13 -2.24 5.18 5.19 7.02 6.93 2.01 2.13
GNP (m.p.) 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00
Public Disp.Income/GNP 20.67 20.76 = 20.88 »=— 20.95 20.23 19.83 16.02 15.85
Public Savings/GNP 8.79 8.86 8.95 8.96 6.34 6.17 2.62 2.43
Public Savings/Public I 101.95 102.64 7288 72.61 47.88 47.39 19.96 19.10
Public Cunsumption/GNP 11.88 11.90 11.92 11.98 13.89 13.66 13.40 13.42
Page 232 System Dynamics '91
As observed from Table 1 and 2 model estimates closely reproduce actual changes in major
sectoral variables and macro economic balances demonstrating the suitability of the model in
simulating the effects of medium term economic policies on the performance of the economy
despite the exogenous shocks of the seventies and the resulting imbalances experienced.
When the growth performance of the Turkish economy since the postwar era is analyzed it is
seen that growth (more than 6% per annum) has not been smooth and self-sustaining, but
disrupted almost every ten years due to the balance of payments problems.The inward oriented
trade strategy, emphasizing import substitution and heavy protection is seen to have played an
important part in the resulting structure of domestic production that is oriented to the large
domestic market, requiring high imports, but not externally competitive. Though this strategy has
been effective in the structural change of the economy, it has been extremely vulnerable to the
external shocks of the 1970's. With the rise in the world price of oil in 1974, recession in
developed countries, Turkey tried to maintain the growth momentum gained in the previous periods
by trying to insulate the economy from external shocks by subsidizing the domestic price of oil,
promoting investments and increasing public consumption. The widening savings.gap and foreign
exchange gaps were financed through reserve decumulation and massive external borrowing.
These measures were successful in postponing the slowdown in the growth process until 1977.
However with the drying up of reserves and the unforthcoming foreign borrowing in the
magnitudes desired,the economy came toa halt in 1979.
Table 2. Actual and Model Estimates of Balance of Payments Accounts (Current prices,Million $)
1973 1975 1977 1979
Actual Model Actual Model Actual Model Actual Model
Exports of Merchandise 1317 1310 1401-1431 1753-1742 2261 = 2259
Imports of Merchandise 2086 2088 4739 4618 5796 = 5813 5069 5002
Petroleum and products 222 276 812 894 1470-1213 1712-1518
Current Deficit AR -494 1880 1877 3425-3434 1239-1244
Workers’ Remittances 1183-1183. 1312 1312 982 982 1694 1694
Net Capital Inflow 233 233 1463. 1463 2874 = (2874 11641164
Reserve Accumulation(-) -728 = -729 417 414 551 560 75 80
As seen from Table 1 and 2, the swing in the growth rates achieved, the increasing share of
total and public investment and public consumption in GNP, the declining share of public savings
in public investment and the increase in current account deficit are captured by the model both in
magnitude and direction, indicating the success of the model in simulating the stance of the
economic policy followed during 1973-1979.
4. Income Distribution Analysis of the Historical Base Run
Turkey with a GNP per capita of $1110 (1977 US dollars) in 1977 compares well to other
middle income countries with an average GNP per capita of $1223 in 1977 (Urate et al 1983,13).
However three country wide size distribution of income studies conducted in years 1963, 1968
and 1973 indicate that the rise in income levels have not been accompannied by a better distribution
of income and that Turkey with gini coefficients of 0.55, 0.56 and 0.51 in the respective years
suffer a substantial amount of income inequality. The study exploring the sources and structure of
income inequality over the period 1950 to 1973 (Dervis et al 1980) point out that the major
determinants of overall inequality in Turkey are; (i) the gap between agricultural and
nonagricultural productivities, (ii) high inequality within the agricultural sector, and (iii) regional
income differences.
There are no other country wide income distribution surveys after 1973. with which we can
compare our estimates. However it is believed that the relative income distribution estimates
obtained for 1973, and the CGE model estimates that closely reproduce actual values provide
sufficient justification for the analysis of the size distribution presented below.
System Dynamics '91 Page 233
The CGE model generates the distribution of income to aggregate factors of production, to
economically active groups and to the five household categories. Using the mean household
incomes generated by the CGE model, the household size distribution can be obtained as the
numerical aggregation of a number of different within group income distributions whose
functional form and parameters are known (Robinson 1976). Aggregating and rescaling the data
provided in (Dervis et al 1980,103) the parameters of each within group distribution for the five
household categories distinguished in the model are estimated assuming that the within group
distribution functions are log-normal. Log-variances of each household category is assumed to
remain the same for the following years.
Table 3 presents the overall household distribution statistics for the four benchmark years
obtained by using the real disposable household incomes excluding remmittance income.
Table 3. Household Size Distribution Statistics for the Base Run
1973 1975, 1977 1977
Relative Distribution(%)
Bottom 40% 9.86 10.25 10.49 10.57
Middle 40% 31.89 31.56 31.45 31.13
Top 20% 58.25 58.19 58.06 58.31
Gini Coefficient 0.535 0.532 0.529 0.531
Log-Variance 1.061 1.007 0.973 0.956
Mean Incomes (thousands of 1973 TL)
Overall 31.85 34.80 38.29 37.80
Farmer 1 13.05 15.25 17.20 16.86
Farmer 2 79.34 89.62 96.21 92.75
Urban Poor 21.64 22.84 24.25 22.90
Urban Middle 32.89 33.77 36.05 34.92
Urban Rich 252.25 281.65 324.82 353.31
Ratio:Top Decile-Bottom Decile 40.04 37.33 35.83 35.75
Based on Table 3 the following conclusions can be drawn:
(i) The distribution of household incomes excluding remittances is more uneven for the base
year than suggested by 1973 income distribution survey results. Since the distribution
Statistics are obtained from the survey data, and the CGE model generates household incomes
from GDP accounts, the discrepancy must be due to the underestimation of both agricultural
and nonagricultural incomes in the survey.
(ii) In accordance with the GDP growth rates achieved the relative size distribution slightly
improves from 1973 to 1977 and shows a tendency to deteriorate after 1977. However,
when the relative size distribution is obtained using nominal incomes, the deterioration from
1977 to 1979 is more profound. The main reason for this is the substantially lower consumer
price index for farmer households, 5.22, as compared to the consumer price index for urban
households, 5.61, in 1979 (1973=1).
(iii)Despite the steady increase in the mean real incomes of all household categories, with the
rate of increase of farmer 1 incomes highest, the improvement in the relative distribution from
1973 to 1977 is very slight (Gini coefficient declines by only 1.12%) indicating that unless
measures aimed at reducing the within group log-variances are sought, it is very difficult to
reduce the extent of overall inequality.
(iv) The very uneven distribution of rural incomes seem to remain as one of the major sources of
overall inequality. However the increasing inequality in the urban sector, characterized by the
substantial mean income differentials among urban households, shows a tendency to be the
other source of overall inequality.
(v)When mean real household incomes including remittances are used to generate household
size distribution, it is seen that the Gini coefficient for 1973 alone reduces by 4.3%. Given
the insensitivity of the Gini coefficient to rising mean incomeés the role of remittances in
reducing relative inequality cannot be overlooked. , However to avoid introducing bias with
respect to different exchange rate parities and thus Turkish Lira equivalent of remittance
Page 234 System Dynamics '91
income, we refrained from reporting and using household income statistics including
remittances as the base run statistics with which the results of the experiments are compared.
5. Results of Experiments
The experiments conducted are analyzed under three groups. Group A focus on alternative
trade strategies with exchange rate the main policy instrument. Group B concer alternative wage
policies.. Group’C contain alternative redistributive experiments with a new base, where the oil
price rise and the balance of payments problems that occurred due to it is nonexistent, so that the
relationship between growth and income distribution can be analyzed without distortions. The
experiments are pure in the sense that all exogenous variables and parameter values of the base run
are preserved, and only a few of them changed in each experiment.
Group A:Definition of Experiments
Al: Same as historical base run. Instead of fixed exchange rate specification with quantity
rationing, the flexible exchange rate specification is used.
A2: Same as Al. Instead of actual borrowing and reserve decumulation, following (Dervis et al.
1978, 50) normal borrowing and reserve change has been assumed.
A3: Same as A2. Except that growth rates of labor categories, set at their actual political
employment rates in obtaining the base run, are set at their natural grouwth rates.
A4: Same as A3. Except that remittances are assumed to remain constant in dollar terms, or
increase at the average annual growth rate of world inflation (excluding oil prices).
AS: Same as A4. Except that,.tariffs are reduced by 10% per year.
The growth and distributional consequences of the experiments in Group A are summarized in
Tables 4 and 5.
Table 4. Comparison of Basic Economic Indicators - Group A
Average Annual Real Growth Rate (%)
BaseRun Al A2 A3 A4 AS
Total Consumption 2.54 5.97 4.09 4.53 5.08 5.00
Total Investment 4.82 5.09 3.25 3.66 4.23 440
Exports 0.54 8.84 19.03 20.37 17.10 17.97
Imports 0.68 3.19 -1,36 1.18 0.01 0.25
GDP (m.p.) 5.19 6.38 5.72 6.27 6.39 6.40
Organized labor wage 2.28 1.12 0.15 -1.49 -1.00 0.99
Unorganized labor wage 2.34 1.83 0.57 1.71 -1.02 -1.08
Table 5. Comparative Household Distribution Statistics (Group A 1979)
Averge Annual Real Growth Rate (%)
BaseRun Al A3 A4 AS
Relative Distribution(%)
Bottom 40% 10.57 10.99 10.88 10.96 10.92 10.92
Middle 40% 31,13 30.90 31.05 30.43 30.38 30.37
Top 20% 58.31 58.11 58.07 58.61 58.70 58.71
Gini Coefficient 0.531 0.525 0.526 0.529 0.530 0.530
Log-Variance 0.956 0.909 0.923 0.903 0.908 0.908
Mean Incomes (thousends of 1973 prices)
Overall 37.80 36.77 36.02 34.42 35.61 35.58
Farmer ] 16.86 19.14 17.98 18.43 18.81 18.83
Farmer 2 92.75 103.37 98.61 100.47 104.37 104.59
Urban Poor 22.90 21.98 21.97 19.33 20.15 20.09
Urban Middle 34.92 31.71 31.78 29.35 30.36 30.31
Urban Rich 353.31 298.57 295.22 297.35 308.61 307.46
Ratio:Top Decile-Bottom Decile 35.75 32.15 32.95 32.33 32.66 32.64
System Dynamics '91 Page 235
The results of the counter factual trade policy experiments summarized in Tables 4 and 5
indicate that the growth performance would have been more favorable and self-sustaining with the
exchange rate equilibrating balance of payments and avoiding the foreign exchange crisis.
Moreover by comparing experiments A1,A2 and A4 with each other it can be inferred that both
foreign borrowing and workers’ remittances have been important determinants of the high rates of
growth achieved, and that had it not been for the substantial foreign borrowing actually used, both
growth and income distribution would have been adversely affected despite the exchange rate
policy employed. Another point to note is that flexible exchange rate policy alone is not sufficient
in channeling industrial production to foreign exchange earning activites since export expansion is
seen to be the result of expenditure switching. The results further indicate that the initial
improvement in the overall distribution is due to the narrowing rural urban gap brought about by
the improvement in the agricultural terms of trade. But the unchanged within-group log variances,
and enlarging income gap among urban households deem further improvement almost impossible
with the policies considered. ‘
Group B : In this group experiment A4 is selected as reference.
B4A:Same as A4. Instead of the endogenous determination of the wages of all labor categories,
the growth rate of organized labor's real wage is set exogenously so that it remains constant at its
base year value throughout the period 1973-1979.
B4B and B4C: Same as B4A except that the growth rate of organized labor's real wage is set at
+3% per annum, and -3% per annum, respectively.
Table 6. Comparison of Group B Alternatives
Average Annual Real Growth Rate (%)
A4 B4A B4B BAC
Total Consumption 5.08 4.16 441 5.41
Total Investment 4.23 4.00 3.42 4.55
Exporis 17.10 16.61 15.27 18.04
Imports 0.01 -0.04 -0.24 0.13
GDP (m.p.) 6.39 6.16 5.55 6.79
Unorganized labor wage -1.02 -1.14 -1.19 1.09
As observed from Table 6 experiment B4C singles out as the most favorable wage policy
from the growth point of view. This result is due to the favorable agricultural terms of trade in
BAC with agriculture accounting for more than 30% of GNP.
Table 7. Comparative Household Distribution Statistics (Group B 1979)
A4 B4A B4B B4C
Relative Distribution(%)
Bottom 40% 10.92 10.81 10.43 11.06
Middle 40% 30.38 30.55 31.13 29.87
Top 20% 58.70 58.64 58.44 59.08
Gini Coefficient 0.530 0.530 0.532 0.531
Log-Variance 0.908 0.921 0.970 0.885
Mean Incomes (thousands of 1973 prices)
Overall 35.61 36.36 38.07 34.42
Farmer 1 18.81 18.79 18.14 19.37
Farmer 2 104.37 104.12 99.85 108.02
Urban Poor 20.15 20.02 19.77 20.21
Urban Middle 30.36 31.90 36.87 2744
Urban Rich 308.61 307.90 306.98 308.68
Ratio:Top Decile-Bottom Decile 32.66 33.16 35.06 31.92
Page 236 System Dynamics '91
Statistics provided in Table 7 demonstrate the difficulty of judgeing what a better distribution
of income is. The relative income distribution measures and the mean real incomes of households,
when used as ranking criteria, point out different wage policies as the most favorable from the
distributional aspect. Based on the mean real incomes of all household categories, except urban
middle households, B4C rates as the most favorable policy. However if the objective is to
improve the mean real income of a target group, B4B seems to be the most promising wage
policy, at the cost of lower growth, and more uneven overall distribution.
Group C-
C1: Same as the Historical Base Run with the changes being; (i) the increase in the world price of
petroleum set equal to the increase in the world price of other products.(ii) The values of
exogenous variables set at their targeted values in the Fourth Five Year Plan that did-not envisage
the oil price rise. (iii) Smoothed out growth rates of parameters from period to period preserving
cumulative growth rates. (iv)Price level deflated exchange rate held constant over the entire period.
C2:Same as C1 except that the direct tax rate of agricultural and nonagricultural capitalists
increased by 10% per annum and the additional tax revenue transfered to farmer 1 and urban poor
households in the proportions of 60% and 40% respectively.
C3:Same as C2 except that the additional tax revenue is not transfered but channeled to public
investment.
C4:Same as C1 except that institutional taxes are lowered by 10% per year.
Table 8. Comparison of Group C Alternatives
Average Annual Real Growth Rate (%)
ci c2 C3 C4
Total Consumption 6.61 6.60 6.17 6.61
Total Investment 13.44 13.51 14.56 13.46
Exports 3.77 3.79 4.04 3.78
Imports 5.92 5.94. 6.03 5.93
GDP (m.p.) 7.91 7.92 7.88 7.92
Organized labor wage 3.67 3.48 4.32 3.68
Unorganized labor wage 2.57 2.37 2.84 2.59
Table 9. Household Distribution Statistics (Group C)
cl 1979
1973 1979 Lev! C3 C4
Relative Distribution(%)
Bottom 40% 9.86 11.34 12.04 11.62 11.31
Middle 40% 31.89 31.23 32.27 31.93 31.21
Top 20% 58.25 57,43 55.69 56.46 57.48
Gini Coefficient 0.535 0.517 0.498 0.508 0.517
‘Log-Variance 1.061 0.878 0.820 0.860 0.879
Mean Incomes (thousands of 1973 prices)
Overall 31.85 36.03 35.28 33.77 36.23
Farmer 1 13.05 19.82 22.24 18.89 19.82
Farmer 2 79.34 105.64 104.17 98.38 105.63
Urban Poor 21.64 22.10 24.51 21.93 22.14
Urban Middle 32.89 30.88 29.13 29.36 31.10
Urban Rich 252.25 277.24 ~—-255.89 237.03 282.25
Ratio: Top Decile-Bottom Decile 40.04 30.09 26.20 28.55 30.24
System Dynamics '91 Page 237
The results of experiment C1 indicate that if the external shocks of the seventies had not been
experienced the targeted growth rates would have been achieved, and the high rates of GDP
growth would have translated into steady improvement of income distribution. Tax and transfer
policy simulated in C2 demonstrate that redistributive policies do not have adverse growth
implications while improving the distribution of income significantly. Experiment C3 indicate that
increased public investment induces no change in the growth performance and only slightly
improves overall distribution at the cost of reduced absolute income levels. Neither the
distributional nor the growth consequences of encouraging institutionalization is promising as it
results in lower disposable household incomes due to increased forced savings. So experiment C2
singles out as the most promising redistributive policy.
Overall it can be concluded that although the experimental results are specific to Turkey major
policy conclusions regarding income distribution and growth are in close conformity with the
empirical findings of research carried out in other country contexts using CGE models. (Adelman
et al. 1978,Taylor et al. 1980) Mainly; (i) It is very difficult to improve the overall relative
distribution of income, while functional distribution is quite sensitive to policy changes. (ii) Policy
instruments conducive to higher growth are not necessarily in conflict with the policies aiming at
more even distribution of income, and higher rates of growth result in more even distribution of
income. (iii) The most important relative price change effecting size distribution is the change in
agricultural terms of trade.
References:
1. Adelman,I., Morris,C.T. 1973. Economic Growth and Social Equity in Developing Countries.
Stanford:Stanford University Press.
2. Adelman,I., Robinson,S. 1978. Income Distribution Policy in Developing Countries: A Case
Study of Korea . Stanford: Stanford University Press .
3. Ahluwalia, M.S. 1976 . Income Distribution and Development: Some Stylized Facts. American
Economic Review . May .
4. Bourguignon,F.,Michel, G., Mique, D. 1983. Short-run Rigidities and Long-run Adjustments
in a Computable General Equilibrium Model of Income Distribution and Development. Journal
of Development Economics. 13: 21-43 .
5. Celasun, M. 1975. Policy Analysis with a Price Responsive Interindustry Model . Washington
D.C.: The World Bank Prototype Country Models. 6.
6. Celasun, M., Giiven, S., Yaprak,T. 1981. The Transfer of the General Equilibrium Modelling
Framework to Turkey; Final Report. M.E.T.U. Ankara; [BRD Research Project .
7. Chenery, H.B., Syrquin, M. 1975. Patterns of Development, 1950-70. Oxford: Oxford
University Press .
8. Cline, W.R. 1972. Potential Effects of Income Redistribution on Growth. New York: Praeger.
9. Cline, W.R. 1975. Distribution and Development: A Survey of Literature. Journal of
Development Economics. 1:359-400.
10.Dervis, K., Robinson, S. 1978. The Foreign Exchange Gap, Growth, and Industrial Strategy
in Turkey: 1973-1983. Washington D.C.:World Bank Staff Working Paper. 306.
11.Dervis, K., Robinson, S. 1980. The Structure of Income Inequality in Turkey (1950-1973) in
The Political Economy of Income Distribution in Turkey. New York/London: Holmes and
Meier Publishers Inc.
12.Giiven, S. 1986. Income Distribution and Growth in Turkey: A General Equilibrium Analysis
(1973-1979). Ankara: Unpublished Ph.D. Thesis M.E.T.U.
13.Lewis, J.D., Urata, S. 1983. Turkey, Recent Economic Performance and Medium Term
Prospects 1978-1990. Washington D.C. World Bank Staff Working Papers. 602.
14.Powel, M.J.D. 1970. A Hybrid Method for Non-Linear Equations in Numerical Methods for
Non-linear Algebraic Equations. New York: Gordon and Breach.
15.Robinson, S. 1976. Income Distribution Within Groups, Among Groups and Overall: A
Technique of Analysis. Princeton, N.J.: Discussion Paper No 65.
16.Taylor, L., Bacha, E., Cardoso, E., Lysy, F. 1980. Models of Growth and Distribution for
Brazil. Oxford: Oxford University Press.
