Planning for Regional Development:
A General Equilibrium Analysis for
Erinç Yeldan (Yaşar University, Turkey)
Ebru Voyvoda (Middle East Technical University, Turkey)
Kamil Taşçı (Ministry of Development, Turkey)
Emin Özsan (Ministry of Development, Turkey)
In this study we investigate the macroeconomic effects of two
complementary policy environments to invigorate growth, employment
and income equality across two broadly differentiated regions in Turkey:
Poor and High/Mid-Income. With the aid of a regional computable
general equilibrium model that disaggregate the production structure into
thirteen sectoral activities and two geographical regions, we first study
the long run dynamic effects of a regional production and investment
subsidization programme. Second, we supplement this environment by a
productivity enhancement programme in the poor region.
Our results reveal that regionally differentiated productivity enhancing
measures coupled with a subsidized investment programme to facilitate
capital accumulation and reduce the outflow of factors out of the poor
region are of utmost importance in designimng a sustained growth path to
pull the aggregate economy from the dual traps of middle income and of
1 Support for this paper has been generously provided by the Turkish Enterprise and Business
Confederation (TURKONFED). Authors acknowledge their indebtedness to Süleyman Onat, Arzu
Turhan, Ümit Özmen, Haluk Tükel and the colleagues at the research department of TURKONFED for
their invaluable comments and suggestions in earlier phases of the study. All caveats apply; in
particular, we note that the views expressed herewith are solely those of the authors, and do not
implicate the institutions that the authors are affiliated with.
That “growth and development are by no means homogenous and unidirectional” is a
proposition well-known by the students of development economics. Inequalities
across regions, as well as people coming from different ethnic and social backgrounds
have attracted the attention of social scientists and economists alike. Aspects of
regional development, even though had been studied by anthropologists and other
sister disciplines; economists typically are in need of developing formal models to
study issues such as duality, poverty traps, and the determinants of migration of labor
and capital across regions.
Over a broader spectrum, economies that reached to the “middle income” status are
observed to suffer from a series of structural difficulties in sustaining their growth
performance. Many economists took note of the fact that as economies converge to
“middle income” level, the relatively “easily-found” sources of growth that are based
on the transfer of unlimited supplies of labor from rural agriculture to urban centers
and towards capital-investment-led high profit sectors, lose their stimulating impact
gradually; technologies grow mature and finally become worn out. After this
threshold, sources of growth must be derived from productivity gains, which can only
be achieved by the investments aimed at human capital, training and research-and-
development (R&D), and institutional reforms. This, however, is no easy task and
often countries get “trapped” at this stage of development, conceptually referred to as
the middle income trap.
Turkey offers a typical example of a middle income country together with its recent
fast, and yet very volatile and erratic patterns of growth. Accordingly, a 2012 report
released by the Turkish Enterprise and Business Confederation (TURKONFED):
“Escaping the Middle Income Trap: Which Turkey? argued that Turkey has reached
the middle income path from the lower end in 1955 and remained at this status for an
excessively long period. Turkey has graduated to the high middle-income status in
fifty years only after 2005. In contrast, this journey has taken quite short for certain
countries such as China (17 years), Malaysia (27 years) or Korea (19 years).
Yet, a very critical question to be answered urgently comes to the fore at this point:
which Turkey? In the above-referred TURKONFED report, Turkish economy is
divided into three sub-divisions by income brackets: With a regional income reaching
to $376 billion, which also exceeds those of European economies such as Norway and
Switzerland, the high-income Turkey that is led by Istanbul, Ankara and Izmir gives
the impression that it would have relatively powerful dynamics to escape the middle
income trap. This sub-division hosts Turkey’s administrative, political, commercial
and financial power centers and yet its links with the rest of Turkey are weakening
Apart from the “high-income Turkey”, there are two more sub-divisions: the one
which is exposed to the danger of being trapped in middle-income, and the one that
does not even have the opportunity to graduate to middle-income bracket (the “Poor
Turkey”). Being trapped in the poverty trap, the “Poor Turkey” includes 27 provinces,
all of which suffer from low level of education (the average period of education is
even less than five years; i.e. drop outs from elementary school), the lack of fixed
capital investment and infrastructure, and social exclusion with its seasonal and low-
skilled labor force.
The purpose of this study is to investigate and assess the sources of regional income
differences in Turkey on the basis of production and capital accumulation relations
and within the discipline of general equilibrium. The ”new” theories of growth that
emerged over the last two decades had underscored that sources of growth emanate
from various paths including human capital, factors unique to geographical location,
indigenous characteristics of the institutions, as well as the endowments of the factors
of production. Furthermore, the literature clearly underscores that a major explanatory
factor in assessments of per capita income differences across the countries and regions
are based on the initial income inequalities and the dualistic patterns of accumulation
of capital and of technology over the course of development (see, e.g., Combes et.al
(2008); Dicken, 2003. The earlier literature has, of course, rest on the now seminal
works of Frank, Amin, Dos Santos, and others.
Yet another purpose of the study is to provide a viable contribution to the general
equilibrium literature addressing issues of regional development and the economics of
geography with the exclusive emphasis on regional factor flows and regionally
differentiated economic equilibria. The study is organized in five sections following
this introduction. In the next two sections, we provide an overview of the structural
characteristics of the Turkish economy across two differentiated regions ( Poor and
Rich) and offer a regionally differentiated macroeconomic data set (a regionally
differentiated social accounting matrix) with a particular emphasis on their dualistic
endowments and paths. The (regional) computable general equilibrium model
structure is introduced in section three. In section four we carry out a series of
alternative regional development scenarios, and reserve section five fort he discussion
of results and concluding comments. Algebraic structure of the model is further
tabulated in an Appendix at the end of the paper.
II. Two Region Differentiation Based on Production
Turkey is a typical developing market economy with wide differences across regions
and social strata. Data on population censuses indicate that over the last decade
population of the rich/mid-income regions of Turkey has expanded, while that of the
relatively poor regions has remained almost stagnant. This fact was mainly due to the
vibrant migration ongoing across the two regions. Data based on the 2010 Household
Surveys of Population provided by TurkStat indicate that total population of Turkey is
73.7 million persons. Half of this population lives in the Level 2 regions encompassing
the rectangle of Edirne-Bolu-Antalya and Mugla. This region covers 29% of the total
land of the whole nation.
<Table 1 here>
Pertinent data of the nine Level 2 sub-regions are displayed in Table 1. The
municipality of Istanbul alone holds 37% of the population while providing about 45%
of the aggregate regional gross value added (RGVA). It further yields almost 60% of
the total income tax generated across the region.
One of the main reasons of this observation is due to the fact a variety of the
enterprises all over the whole country have their headquarters located in Istanbul.
Further due to this reason, Istanbul regions has a relatively high share of the aggregate
public investments. Half of the total public investments in the Rich/Mid Income
region is captured within the Istanbul province.
<Table 2 here>
Based on our differentiation, we distinguish 17 Level-2 regions classified under the
“poor region” under the threat of the “poverty trap”. This region overall covers 79%
of the geographical land, and about half of the total population. The observed
differentiation in the population densities across the two regions reflect, in many
respects, the structural imbalances across the local product and labor markets. The
relatively dispersed land and population structure of the “poor region” has direct
consequences for the presence of relatively weak external spillovers, otherwise
expected from public investments.
For many years the Turkish incentives structure for investment had been questioned
due to its lack of focus for regional and structural development. Structural bottlenecks
range from the general inability of the “West” to mobilize its scale economies and
simultaneously internalize the externalities of the “East”. Insufficient/non-existing
transportation networks; poor marketing infrastructure; and a relatively poorly
educated and unskilled labor force mark the parameters of this dual structure.
The fact that about 80% of the aggregate foreign trade is transacted within the
“western regions” is a manifest that the Eastern provinces face strong difficulties in
reaching marketing outlets, and are mostly restricted to the primary product space in
their production activities.
Over the course of the last decade Turkish economy experienced a vigorous expansion
in its capital base in line with rapid growth. Financial assets likewise expanded at a
very fast rate, wherein total deposits had reached to 500 billions TL in 2009 from the
2002 level of 300 billions (measured in fixed 2010 prices). The rate of expansion of
deposits was 60% in the East, in contrast to the 75% score achieved in the West. The
gap in asset formation follows the existing duality across the two regions.
III. Regional Macroeconomic Analysis
III-1. Construction of the Regional Social Accounting Data Base
In compilation of the regionally differentiated Social Accounting Matrix (SAM) data
base, we start with the 2002 input output data of the TurkStat. From the official 2002
I/O data we had re-constructed a more recent data base for 2010. This data set links
the 2002 I/O coefficients with macro aggregates maintaining demand and supply
consistency at the sectoral level. Based on our model specification the 2010 I/O data
set dis-aggregates the production activities into thirteen sectors and distinguishes the
domestic economy into two regions: High/mid Income Turkey and the Low Income
In what follows, we distribute wage and capital income, as well as factor flows across
sectors and across the two regions by utilizing production shares in gross regional
value added (GRVA). Thus, we observe that about 80% of agricultural and non-
agricultural output is produced in the High/mid Income Turkey, and the rest 20% is
produced in the Poor Turkey.
Another data disaggregation pertains to the indirect tax revenues by sectors. This is
done via sectoral employment shares across the two regions. In this vein, 33% of rural
employment, and 57% of non-agricultural employment are generated in the High/mid
Tables 3a and 3b introduce the structure and the data of the Regional I/O data
<Tables 3a and 3b here>
The remaining components of the input-output flows across the regions are the final
demand aggregates. The I/O structure distinguishes the final demand vectors across
sectors, but not over regions. The whole “national” demand is regarded as one single
entity at the aggregate macro demand level. We read that 93% of aggregate final
demand is met by the non-agricultural sectors and that of agriculture remains at 7%.
The sectoral structure of the economic activities is laid out in table 4.
<Table 4 here>
The regional Social accounting Matrix which is based on the 2010 I/O data, is
displayed in Table 5. The SAM discloses aggregate national accounts in line with the
sectoral production structure across the two regions identified: the High/mid Income
Turkey and the Poor Turkey. Income flows generated from the regionally
differentiated activities are collected within the single private household and the
central government. Expenditures, then, are carried out at the national level via the
private and the public agent.
Final component of the SAM is the rest of the world account. This is generated at the
national level for imports, but differentiated by regions as well as by sectors for the
export activities. We now introduce the distinguishing features of the analytic model.
III-2. Algebraic Structure of the Applied General
The model is built on the Social Accounting Matrix (SAM) introduced above. This
includes two regions (Poor, Rich) and thirteen aggregated sectors that is constructed to
represent the 2010 general equilibrium of the economy. Given the regional structure,
the model initially has to account for the flows between regions. Here, we follow an
approach that disaggregates the production processes in each sector at the regional
level, whereas we keep the demand side aggregated at the national level. In line with
this structure, the analytical construction of the model to represent the production,
employment and distribution activities at the regional level is based on the regional-
SAM (R-SAM) presented in the Appendix at the end of this chapter.
Building on this structure, we have total supply (absorption) at the national level as the
sum of the value added produced in each region of the economy. The model follows
the Armingtonian system of trade where the domestic production (DC), coupled with
the import demand (M) makes up the composite commodity at national level.
Following Armington (1969), we assume that the domestic and imported commodities
are imperfect substitutes through a constant elasticity of substitution (CES) function:
CC in Equation (1) represents total absorption in terms of the composite commodity;
DC is the level of domestic production and M is the level of imports in each sector i.
is parameter of constant elasticity of substitution between domestic
production and imports. Here, we assume that Equation (1) is representing the
relationship between domestic production and imports at the national level. Total
domestic production however, is also differentiated by the region of origin, DCr.
Therefore, the substitution possibilities represented in the equation above are among
the regional domestic production DCY, DCZ, and imports, M which make up total
domestic absorption. Figure 1 summarizes this relationship.
<Figure 1 here>
Here, the factors of production, capital (K) and labor (L) in each region produce the
output X of the region. The profit maximization behavior of the representative firm in
each region determines the regional wages (W) and the regional profit rate (rk). Output
in each region is either demanded domestically (DC) or exported (E). Total domestic
absorption at the nation-wide level (CC), on the other hand is further decomposed into
consumption (C), investment (I), government spending on commodities (G) and
regional intermediates (INTr). Under such a setting, the import price in each sector is
set at the national level; with no further differentiation at the regional level. Yet, based
on the resource availability and differences in factor prices, export price in each sector
is allowed to vary at the regional level.
The price of the composite commodity then is a function of the shares domestic
commodity and imports in the composite and the prices of domestic commodity and
imports in each sector i:
)1( ,,, ri
saltax in Equation (2) represents the sales tax rate. tm and te in Equations (3) and (4)
are tariff and export tax/subsidy rates.
Based on the characterization of the production technology at regional level, regional
unemployment rate is defined as the difference between regional labor supply and the
regional demand for labor. Likewise, total capital supply in each region is equated
with total capital demand to clear the capital markets at the regional level:
UNEMPr = LSUPr - (5)
rri KSUPPK ,
Net of tax factor incomes, along with transfers from the government, interest income
on domestic debt, factor income from the rest of the world net of interest payments on
foreign debt are the basic sources of income for the households in each region:
In Equation (7),
is the regional nominal wage rate,
is the parameter
representing the difference between the regional nominal wage rates. Similarly
is the profit rate differentiated at the regional level and
is the associated
difference in the regional profit rates.
represents the capital demand of each sector
at the regional level. GOVTRANS is total public transfer to the households, DomDebtG
is the stock of domestic public debt, ForDebtP is the stock of private foreign debt and
NPFI is the net factor income from abroad.
The government collects sales taxes (TOTSALTAX), production taxes
(TOTPRODTAX), tariffs (TARIFF), corporate taxes (TOTCORPTAX), income taxes
(TOTHHTAX), and export taxes (EXTAX):
On the expenditures side, we assume that the government follows a pre-determined
primary surplus target as its fiscal policy rule. Given the public revenues, the amount
of public transfers, the stock of domestic and foreign debts, it is the public investment
variable that adjusts to the balance of the public sector in the model economy2.
Accordingly, the public sector borrowing requirement is defined as:
PSBR = GREV – GCON –GINV – rFForDebtG - rDDomDebtG –GOVTRANS (9)
PSBR is either financed by domestic borrowing ΔDomDebtG, or foreign borrowing
Private households save a sp of their disposable income. The rest of the consumption
demand is distributed among the products of the sectors of the economy by constant
shares, clesi at the composite price PCi:
Similarly, total government consumption is distributed by constant shares among the
sectors of the economy:
We assume that as part of the fiscal rule, total government consumption, GOVCON in
Equation (11) is determined as a constant share of total revenues:
2 The fiscal rule of the 2010s is represented by a comprehensive primary surplus target in Equation (9).
Equilibrium and the Recursive Dynamic Structure of the Model
The general equilibrium of the macroeconomy is associated with the relative prices in
goods and factor markets and the real exchange rate that balances the goods markets,
the factor markets and the current account. In each period, we assume that the formal
real wage rate is constant and it is the regional unemployment levels that help the
regional labor markets clear.
The equilibrium condition of the goods market implies that total demand is equal to
total supply in each sector:
CCi = CDi + GDi + IDPi + IDGi + INTi(13)
The reflection of the goods and factor markets equilibrium at macro-level, implies that
total saving and total investments to equate:
PSAV + GSAV + e CAdef = PINV + GINV (14)
CAdef in Equation (14) represents the current account deficit of the national economy
in terms of foreign currency (US dollars). Here, CAdef is the difference between the
exports and workers’ remittances on the revenues side and the import bill, factor
income transfers abroad, and interest payments on (private and public) foreign debt on
the expenditures side:
In the model, we assume that the private and public components of the external capital
inflows follow a pre-determined path at a fixed level in foreign exchange terms.
Therefore, it is the real exchange rate, e that balances the current account each period.
The model updates the annual values of the exogenously specified variables and also
the policy ratios in an attempt to characterize the 2010 – 2025 growth trajectory of the
Turkish economy. Here we first update capital stocks with new investment
expenditures net of depreciation; and also increase the available labor supplies by the
population growth rates. Similarly, technical factor productivity rates are specified
exogenously in a Hicks-neutral manner.
In order to be able to represent the conditions of the labor markets at the regional level
in detail, we explicitly model the migration behavior between the regions of the
Here, MIG represents the labor migrating between regions; based on the value of this
variable, we find the total labor supply in regions Y (poor) and Z (rich) respectively. nY
and nZ are the population (labor supply accordingly) growth rates in regions Y and Z
respectively. We follow the traditional Harris-Todaro (1970) approach to model the
behavior of MIG through successive time periods. Given the elasticity parameter
migres to represent the sensitivity of the migration behavior to the difference between
the expected wage rate in the rich region (Z) and the actual wage rate in the poor
region (Y), we take on that migration of labor from poor region to rich region is a
function of this difference and the labor stock of the poor region:
We assume that the public and private sectors differ in terms of their investment
behavior. In the public sector, the distribution of total investments, GINV at the
regional and at the sectoral level (investment by destination) is determined
exogenously to represent its relevance as a policy tool. On the other hand, the sectoral
distribution of private investments in each region is formulated as a function of the
profit rates of the production sectors of the economy. Such a formulation is based on
the Tobin-q model of investment and helps one to determine the distribution of
private investments first at the regional and then, based on the difference between the
sectoral and (regional) average profit rates, at the sectoral level. Accordingly, in each
region we calculate the sectoral profit rates as the ratio of total value added net of
wage payments to the value of installed capital stock of sector i:
Once average profit rate (rAVG) in each region is determined, it becomes
straightforward to regulate sectoral investment demands through the difference
between the profit rate of the specific sector i (ri,R) and the average profit rate of the
DKi,R, in Equation (19) is the share of private investment of sector i in region R in total
regional private investment, SPi,R is the share of profits of the same sector in total
regional profits. Accordingly, if the profit rate of sector i is higher (lower) than the
average profit rate among the sectors of the region, the share that sector gets from the
regional total investment increases (decreases) through time.
The sensitivity parameter in Equation (19) is designed to reflect the effect of
expectations and future uncertainty on the distribution of total regional investment
among the sectors. Even though as mechanical as it may seem, the system designed in
Equation (19) emphasizes the “profit drive” as one of the main determinants of the
Finally in this stage we account for the evolution of debt stocks. First note that
government’s foreign borrowing is taken as a ratio of aggregate PSBR:
e ForBorG = (gfborrat)PSBR (20)
DomBor = (1 – gfborrat) PSBR (21)
Consequently, Government Domestic Debt accumulates via:
DomDebtt+1 = DomDebtt + DomBort(22)
Government Foreign Debt, on the other hand, becomes:
ForDebtGt+1 = ForDebtGt + ForBorGt (23)
Similarly Private foreign debt is found as:
ForDebtPt+1 = ForDebtPt + ForBorEt(24)
This completes the algebraic specification of the general equilibrium model. We now
turn to its use as an economic laboratory device to analyze various policy
environments over the 2010-2025 macro economic path.
IV. Policy Analysis
Now we will utilize our analytical structure to investigate alternative policy scenarios
at the regional/sectoral level. To this end we first start with the task of generating a
business-as-usual scenario, which we refer as the base path. This path yields the
historical trends of the Turkish economy, if segmented into the medium future without
any changes in the policy variables, nor the structural parameters. Then, at the second
stage we introduce alternative policy environments and contrast the model’s solution
paths with those of the base path to answer questions of the “what if?” type. This
exercise enables us to provide a cost-benefit analysis of the policies distinguished and
allows us to establish an objective ranking of the alternative policy environments that
could be envisaged from such an analysis.
We extend our time horizon over 2010 to 2025. Over this path we maintain our
business-as-usual stance and keep the policy variables intact without any change.
Also, we kept the structural parameters and their trends at their current levels.
Wherever possible we have incorporated the official programmed values (especially
with respect to components of final demand) into our trends estimates. Sure enough,
there had to be further hypotheses to be incurred regarding various exogenous flows
and parametric values. To this end we have adhered to the following specifications:
Total factor productivity rates are assumed to continue at 0.5% in agriculture
and 1.5% in non-agricultural sectors of the High/mid Income Turkey. These
are taken at 0% in agriculture and 0.1% in non-agriculture at the Poor Turkey;
Public investment by destination shares are kept constant at their current levels
with 48% in High/mid Income Turkey and 52% in the Poor Turkey.
Population (labor supply) growth is set at 1% in the High/mid Income Turkey,
and 1.6% in the Poor Turkey. The migration elasticity parameter is set 0.05.
The model solves the equilibrium real exchange rate endogenously to maintain
consistency at the balance of payments constraint; and sets the interest rate at 5%
(roughly at par with the trend growth rate of the GDP. The model generates the
regional as well as sectoral flows via Walrasian principles of general equilibrium in
response to the (endogenously solved) relative prices, wage rate and the exchange rate,
all in fixed 2010 prices.
All along the base path, the fiscal policy is conducted under the implicit constraint of
the “fiscal rule” so as to maintain a falling rate of public debt ratio. To this end the
primary budget balance is set at 3% as a ratio to the GDP over the base path. To
achieve the fiscal rule, public consumption expenditures is resolved as the balancing
variable in the public accounts, given the tax revenues and other sources of public
The base path “solution” of the model reveals that national income increases to 2,100
billions TL in 2025 (in 2010 fixed prices) from its level of 1,200 billion TL in 2010.
Yet, a closer look at the regional differentiation of the aggregate income suggests that
the existing gap across the High/mid Income Turkey and the Poor Turkey, continues to
widen. In other words, the poor Turkey remains trapped in the poverty trap. The end
result of this divergence is the overall slowdown of the national economy. The
workings of the poverty trap, thus, constitute an important attribute of the overall
middle Income trap for the national economy as a whole. Figure 2 portrays the
evolution of this hypothesis over the modeling horizon.
<Figure 2 here>
We further read that, under the quite optimistic assumptions of the base path,
aggregate export revenues reach to 443 billion $, with about two-thşrds of this sum
being generated by the High/mid Income Turkey. (Figure 3).
<Figure 3 here>
The base path indicates that the current account deficit is stabilized at a ratio of 4% to
the GDP; and thus, the aggregate saving-investment balance remains at that level by
2025 (Figure 4). Public debt as a ratio to the GDP falls secularly to reach 19% as a
ratio to the GDP by 2025. (Figure 5).
<Figure 4 here>
<Figure 5 here>
IV-1. Alternative Policies of Regional Development
The first scenario encompasses the analysis of a regional economic promotion exercise
by way of increasing production and investment subsidies (in the model’s technical
language reduction of indirect net production taxes) to the poor region. This scenario
in its essence tries to capture the essential elements of the incentive system introduced
in June 2012 to promote regional development. The programme entailed 6 groups of
regions across the whole Turkey and introduced differentiated sets of subsidies and tax
In addition, we do not envisage any further changes in the structural parameters, nor in
the other policy decisions regarding the public sector: the fiscal rule is maintained; the
division of public investments is not perturbed; and the direct income tax rates are not
The scenario results disclose many interesting results that for long captured the
development agenda of Turkey. First is the observation that even though the level of
regional value added rises in the Poor Turkey, its stimulus prove short-lived, given the
lack of structural changes that would boost productivity gains. If limited to price
incentives alone, the early gains from factor reallocation and income transfer soon
taper off, as the law of diminishing returns hit the logic of regional capital
accumulation. This leads to our second observation: the reallocation of factors to the
Poor Turkey region relatively deprives the High/Mid Income Turkey from the
resources that would otherwise accrue to the region. As the Rich Turkey region
stumbles, aggregate national income could not be invigorated at any significant sense.
The drag of the Poor Turkey continues to squander the aggregate national economy
within the middle income trap, in spite of the early progress in its regional income.
These results are portrayed in Figure 6 and summarized in tables 7 and 8 below.
<Figure 6 here >
<Table 7 here>
<Table 8 here>
Many mechanisms are at work for this dismal result, to be understood only within the
discipline of general equilibrium. The most obvious mechanism is the slowing down
of migration (of not only of labor, but also of capital, as well) from the Poor Turkey to
the High/Mid Income Turkey. The Poor Turkey, relatively reducing its role as the
suppler of unlimited supplies of cheap labor, now deprives the High/Mid Income
Turkey from a significant source of cost reduction. Furthermore, as income is
indirectly transferred back to the Poor region, aggregate private income relatively
declines as labor remunerations generated from the backward technologies generate
lower wage income. Thus aggregate savings dwindle, pulling down the rate of capital
accumulation as a whole.
These results signify the presence of duality drags reducing the rate of growth for the
whole national economy. The poverty of the region together with poor infrastructure,
limited linkages, widespread informalization and low skills lead to failures in
internalizing the spillovers and other externalities that would otherwise characterize
episodes of sustained growth. In the absence of a productivity driven programme
aimed at structurally transforming Poor Turkey as a region, the drag of the duality trap
continues to haunt the national economy.
Thus, as diminishing returns culminate, the weak spurts of growth achieved in the
Poor Turkey region leads only to a modest gain in GRVA of 4% over the 2025 base
path GDP. (Figure 7). This fails to be enough for pulling the national economy from
the middle income trap..
<Figure 7 here>
The main binding constraint faced in this scenario is that as the generous production
and investment incentives provided for the Poor Region induces an expansion of the
economic activity in the region; this relative re-positioning of the incentives,
nevertheless, divert resources away from the more productive High/Mid Income
Turkey and reduce factor utilization there. The relative slow down in the High/Mid
Income Turkey leads to proportionate decline of the aggregate economy as a whole.
Soon after when the early gains of capital accumulation in the Poor region starts to
falter as the diminishing returns to capital sink in, the national economy remains boud
to the middle-income trap.
Given the lessons of this scenario we deduce that the ongoing investment subsidization
program destined to the poor regions of Turkey, if not supplemented by interventions
to boost productivity and capture the spillover effects of the regional disparities, will
not be sufficient to invigorate overall growth in the long run. This leads us to
conceptualize a second scenario, one that complements the first one with productivity
enhancing reforms in the Poor Turkey. To do so, we first intervene with the
destination of public investments by way of increasing the share of the Poor region to
95% (from its base path value of 52%). The rest of the public investments is directed
to the High/Mid Income Turkey. In addition, we hypothesize that the public
investment programme in education, social capital and infrastructure in transport and
communications will yield an annual productivity gain of 0.5% in the non-agricultural
sectors; and 0.75% in the agriculture of the Poor Turkey. Thus, the productivity
enhancement programme is hypothesized to yield gains only in the Poor regions,
without any perturbations envisaged in the West. As such the interventions of the
scenario should be scaled quite modest, in comparison to the calculated historical path
of the aggregate TFP gains for Turkey, ranging from 0.5% to 2.5% (see, e.g. Yeldan,
2012; Kolsuz and Yeldan, 2013; Taymaz et. al 2008; and Saygılı et. al (2005). )
Thus, this second scenario aims at refurbishing the production-subsidization
programme with additional productivity inducement mechanisms at the regional level.
The comparative results of this scenario is displayed in Figures 8, 9 and 10, and its
sectoral results are tabulated in Table 8.
<Figures 8, 9 and 10 here>
The most important finding of the new policy scenario is that as the production
subsidization programme is supported via a productivity enhancement programme, the
regional value added in the poor Turkey reaches to 18% above the base path value by
2025. In the meantime, this enhancement further stimulates value added growth in the
High/Mid Income Turkey as well. Thus, the productivity-enhancement in the poor
region provides spill over effects over the High/Mid Income Turkey, creating a more
balanced growth path for the whole aggregate GDP. As a result, we obtain an
aggregate gain of 11% over the base path GDP.
Addition of the productivity enhancing measures to the production subsidies
interventions leads to higher factor demands in both regions and reduce pressures of
factor re-allocation away from the poor regions towards the High/Mid Income Turkey.
As this historical realignment is reduced, poor region is placed on a more vigorous and
sustained growth path, pushing the aggregate income to a higher plateau. In this
manner, we observe that formal employment expands by 5% over the base path. Per
contra, under the previous scenario such employment gains were virtually stagnant.
More employment demand for both regions (Figure 11) leads to higher wage income,
and enables an expansion in savings destined for capital accumulation.
<Figure 11 here>
In comparison to both the base path and the first policy scenarios, the second policy
environment means higher volume of trade. Based on the general equilibrium effects
reflected by the relative price changes, exports of the High/Med Income Turkey follow
a path that is 3.4% above the base path in the first half of the policy horizon, with a
reduction to 1.4% gain over the second half. Yet, given the fact that we do not
intervene in any manner to the trade policy instruments, the overall effect of this on
the current account balance remain modest across both policy scenarios.
Sectoral effects of these policy interventions depend, naturally, on the input-output
flows and characteristics of the regional production technologies as well as the
specifics of the labor markets across regions. We read, for instance, that under the
first scenario sectors that have a higher share in the poor regional value added –
Agriculture and Animal Husbandry, Medium Technology Manufacturing and Other
Services enjoy relatively high production and employment gains; while Energy and
High Technology Manufacturing are virtually remain stagnant with respect to the base
In the poor region production and employment levels rise by 5.7% and 5.7% in
Agriculture; by 4.2% and 3.9% in Medium Technology Manufacturing Industries; and
by 3.2% and 2.9% in Construction, respectively. In what follows, the ongoing transfer
of labor and capital flows across regions produce asymmetric effects within the
High/Med Income Turkey region. The region suffers production and employment
losses in Agriculture and Medium Technology industries. Yet, the sharpest losses in
the production levels in the High/Med Income Turkey are observed in Automative by
1.8% and in the High Technology Industries by 1.7%. These results underscore the
importance of regionally differentiated productivity enhancing measures in
accommodating growth targets with regional incomes policies.
V. Concluding Comments
Aiyar et.al ‘s 2013 IMF Staff Paper reveals that problems such as infrastructural
bottlenecks, limited trade openness and lack of regional integration count among the
main constraints reducing the rate of economic activity and trapping them into the
middle income stagnation. Aiyar and his colleagues suggest that, despite differences
in design, basic policy instruments such as openness in trade and rehabilitating
infrastructure investments towards a more balanced capital accumulation across
regions are important.
In this study we have investigated the macroeconomic effects of two complementary
policy environments to invigorate growth, employment and income equality across
two broadly differentiated regions in Turkey: Poor and High/Mid-Income. With the
aid of a regional computable general equilibrium model that disaggregate the
production structure into thirteen sectoral activities, we first study the long run
dynamic effects of a regional production and investment subsidization programme.
Second, we supplement this environment by a productivity enhancement programme
in the poor region.
Our results reveal that regionally differentiated productivity enhancing measures
coupled with a subsidized investment programme to facilitate capital accumulation
and reduce the outflow of factors out of the poor region are of utmost importance in
designimng a sustained growth path to pull the aggregate economy from the dual traps
of middle income and of poverty.
Appendix: List of Equaons of the Regional General Equilibrium Model
)1( ,,, ri
jryiryiryi IORYPCPXprotaxPVA ,,,, )).(1(
jrzirzirzi IORZPCPXprotaxPVA ,,,, )).(1(
Goods and Factor Markets
riRr LDLSUPUNEMP ,
rri KSUPPK ,
)1( iririr saltDCCC
PFORDEBT*INTFORP - NPFI GDOMDEBT*INTDOM
ii PROFGPROFPcorptTOTCORPTAX )(
iiiiii MPEPIDGIDGDCDPCGDP ])([
APPENDIX-2: Model Variables
PRODUCTION FACTORS BLOCK
MACROECONOMIC EQUILIBRIUM, FISCAL POLICY
APPENDIX-3: Model Parameters
Aiyar S., R. Duval, D. Puy, Y. Wu ve L. Zhang (2013) “Growth Slowdowns and the
Middle-Income Trap”, IMF Working Paper, No: WP/13/71
Combes, P. P., Mayer, T., Thisse, J.T. (2008).\Economic Geography: The Integration
of Regions and Nations. Princeton: Princeton University Press
Dicken, P.\(2003).\Global Shift: Reshaping the Global Economic Map in the 21st
Century. New York: Guilford.
Kolsuz, Güneş and Erinç Yeldan (2013) “1980-Sonrası Türkiye Ekonomisinde
Büyümenin Kaynaklarının Ayrıştırılması” Çalışma ve Toplum, forthcoming.
Saygılı, Ş., C. Cihan ve H. Yurtoğlu (2005) “Türkiye Ekonomisinde Sermaye
Birikimi, Büyüme ve Verimlilik, 1972-2003” Ankara: Devlet planlama
Teşkilatı No 2686.
Taymaz, Erol, Ebru Voyvoda ve Kamil Yılmaz (2008) “Türkiye İmalat sanayinde
Yapısal Dönüşüm, Üretkenlik ve Teknolojik Değişme Dinamikleri” METU
ERC Working Papers, No 08/01.
Yeldan, Erinc (2012) “Türkiye Ekonomisi İçin Beşeri Sermaye ve Bilgi Sermayesi
Birikimine Dayalı Bir İçsel Büyüme Modeli” Ekonomi-TEK, 1(1): 21-60,
Yeldan, Erinç, Kamil Taşçı, Ebru Voyvoda and Emin Özsan (2012) Escape from the
Middle Income Trap: Which Turkey? Turkonfed Publications, Istanbul
Yeldan, Erinç, Kamil Taşçı, and Emin Özsan (2012) “Türkiye’de İstihdam - Büyüme
İlişkisi Üzerine Bölgesel Hesaplanabilir Genel Denge Modeli Uygulaması”
Çalışma ve Toplum, 32(1): 11-50.
Figure 1: Domestic Production, Imports and Absorption
Table 1: Economic Indicators of the Rich/Mid-Income Regions Covering the
TurkStat Level 2 Region
Table 2: Economic Indicators of the POOR-Income Regions Covering the
TurkStat Level 2 Region
Table 3a: Rich/Mid Income – Poor Income Regional I/O Structure
Table 3b: Rich/Mid Income – Poor Income Regional I/O Flows (2010 Prices, Millions TL
Table 4 Sectoral Structure of the Model and the 2002 I/O Classification
Table 6 continued