HOW TO GO GREEN: a general equilibrium investigation of environmental policies for sustained growth with an application to Turkey’s economy

Abstract

Green growth is a relatively new concept aimed at focusing attention on achieving sustainable development through the efficient use of environmental assets without slowing economic growth. This paper presents a real-world application of the concept, and identifies viable policy options for achieving a complementary environmental regulatory framework that minimizes output and employment losses. The analysis utilizes macro level data from the Turkish economy, and develops an applied general equilibrium model to assess the impact of a selected number of green policy instruments and public policy intervention mechanisms, including market-based incentives designed to accelerate technology adoption and achieve higher employment and sustainable growth patterns. Overall, our results indicate that an integrated employment and urban greening policy strategy that combines a green jobs programme with a set of earmarked tax-cum-innovation policies towards R&D-driven growth, mainly targeted to strategic industrial sectors and agriculture, developing market economies can achieve significant reductions in gaseous emissions and urban waste while maintaining significant gains in productivity and employment.

Appendices

Appendix 1: algebraic structure of the CGE model

The model is in the Walrasian tradition with optimizing agents against market signals and a simultaneous resolution of market equilibrium of commodity prices, the wage rates and the real rate of foreign exchange. “Dynamics” into the model is integrated via “sequentially” updating of the static model into a medium-run of 20 years from 2010 through 2030. Economic growth is the end result of (1) rural and urban labor population growth; (2) investment behavior on the part of both private and public sectors; and (3) TFP growth performance of the Turkish economy.

The supply-side of the economy is modeled as twelve aggregated sectors. See Addendum 2 for the aggregation scheme. Labor, capital and a composite of primary energy inputs, electricity, petroleum and gas and coal, together with intermediate inputs are the factors of production. For modeling agricultural production activities, the model further accommodates rain-fed and irrigated land as additional factors. Water and fertilizer use (nitrate and phosphorus) are explicitly recognized as part of land usage in rural production.

In algebraic terms, for the non-agricultural sectors the production technology is given as follows:

$$\begin{aligned} XS_i =AX_i \left[ {K_i^{\lambda _{K,i} } L_i^{\lambda _{L,i} } \left( {\prod \limits _j {ID^{\lambda _{ID,j,i} }} }\right) ENG_i^{\lambda _{E,i} }} \right] \end{aligned}$$

(6a)

$$\begin{aligned} i = CO, PG, RP, EL, CE, IS, MW, ET, AU, CN, OE \end{aligned}$$

whereas in agriculture, production entails land aggregate ($N_{A})$ as an additional factor of production:

$$\begin{aligned} XS_A =AX_A \left[ {K_A^{\lambda _{K,A} } L_A^{\lambda _{L,i} } N_A^{\lambda _{N,i} } \left( {\prod \limits _j {ID^{\lambda _{ID,j,A} }} }\right) ENG_A^{\lambda _{E,A} } } \right] . \end{aligned}$$

(6b)

In Eqs. 6a and 6b, AX is the technology level parameter, $\lambda _{K,i} ,\lambda _{L,i} ,\lambda _{N,i} \lambda _{E,i}$ denote the shares of capital input, the labor input, aggregate land input (only for agriculture) and the energy input in gross output in sector $i$. Under the assumption of constant returns to scale (CRS) technology, for every sector $i$:

$$\begin{aligned} \lambda _{K,i} +\;\lambda _{L,i} +\lambda _{N,i} +\sum \limits _j {\lambda _{ID,j,i} } +\lambda _{E,i} =1. \end{aligned}$$

(7)

At the lower stage of the production technology, the primary energy composite is produced along a constant elasticity of substitution (CES) production function using the primary energy inputs, coal, petroleum and gas and electricity:

$$\begin{aligned} ENG_i =AE_i \left[ {\kappa _{CO,i} ID_{CO,i}^{-\rho x_i } +\kappa _{PG,i} ID_{PG,i}^{-\rho x_i } +\kappa _{EL,i} ID_{EL,i}^{-\rho x_i } } \right] ^{-1/\rho x_i }. \end{aligned}$$

(8)

Under the above production technology, differentiation of the minimum cost per unit of primary energy inputs gives the sectoral demand for coal, petroleum and gas and electricity:

$$\begin{aligned} \frac{ID_{CO,i} }{ENG_i }&=\left[ {\frac{\kappa _{CO,i} PEG_i }{AE_i^{-px_i } ({1+CO_2 taxN_{CO} +PM10taxN_{CO} })PC_{CO} }} \right] ^{1/({1+\rho x_i })}\end{aligned}$$

(9)

$$\begin{aligned} \frac{ID_{PG,i} }{ENG_i }&=\left[ {\frac{\kappa _{PG,i} PEG_i }{AE_i^{-px_i } ({1+CO_2 taxN_{CO} +PM10taxN_{CO} })PC_{PG} }} \right] ^{1/({1+\rho x_i })}\end{aligned}$$

(10)

$$\begin{aligned}&\frac{ID_{EL,i} }{ENG_i }&=\left[ {\frac{\kappa _{EL,i} PEG_i }{AE_i^{-px_i } ({1+CO_2 taxN_{CO} +PM10taxN_{CO} })PC_{EL} }} \right] ^{1/({1+\rho x_i })} \end{aligned}$$

(11)

where PEG is the cost of energy input composite, and $CO_{2}\,taxN_{j}$ and $PM10taxN_{j}$ are the pollutant’s fees (carbon, and particular matter-10 tax rates, respectively) on input $j$.

Sectoral demands for labor, capital, and energy composite and intermediate inputs arise from the profit-maximization behavior of the representative firm in each sector:

$$\begin{aligned} K_i&=\lambda _{K,i} \left[ {\frac{({1-t_{\Pr od,i} -CO_2 taxP-PM10taxP})PX_i XS_i }{r}} \right] \end{aligned}$$

(12)

$$\begin{aligned} L_i^D&=\lambda _{L,i} \left[ {\frac{({1-t_{\Pr od,i} -CO_2 taxP-PM10taxP})PX_i XS_i }{({1+pyrltax})\overline{w} }} \right] \end{aligned}$$

(13)

$$\begin{aligned} ID_j&=\lambda _{ID,j,i} \left[ {\frac{({1-t_{\Pr od,i} -CO_2 taxP-PM10taxP})PX_i XS_i }{({\text{ CO }_2 taxN_j +PM10taxN_j})PC_j }} \right] . \end{aligned}$$

(14)

In agriculture, land aggregate is demanded in relation to its factor intensity as above:

$$\begin{aligned} N_A =\lambda _{N,A} \left[ {\frac{({1-tax_A -CO_2 taxP-PM10taxP})PX_A XS_A }{RN_A }} \right] \end{aligned}$$

(15)

where RN $_{A}$ is the average land rental rate. This average is obtained from the weighted average of the rental rates on irrigated land and rain-fed land:

$$\begin{aligned} RN_A N_A = RNRF({1+taxF})NRF + RNIR({1+taxF+feeW})NIR. \end{aligned}$$

(16)

In Eq. (16) RNRF and RNIR refer to rental rates of the rain-fed land, NRF, and irrigated land, NIR, respectively. The fee rates are on fertilizer use (taxF) and on water usage (feeW).

At a lower level, land aggregate is a CES composite of the irrigated and rain-fed land types:

$$\begin{aligned} N_A =AN_A \left[ {\theta _{NRF,A} NRF_A^{-\rho _{NA}} +({1-\theta _{NRF,A} })NIR_A^{-\rho NA} } \right] ^{-1/\rho _{NA}}. \end{aligned}$$

(17)

The optimal choice of the farmer towards utilization of irrigated versus rain-fed land is given from the optimizing conditions and is subject to the taxation (fees) instruments:

$$\begin{aligned} \frac{NRF_A }{NIR_A }=\left[ {\frac{\theta _{NRF,A} RNIR}{({1-\theta _{NRF,A} })RNRF}} \right] ^{1/({1+\rho _{NA} })}. \end{aligned}$$

(18)

We assume that the amount of water usage in irrigation is given by a Leontieff coefficient on the irrigated land:

$$\begin{aligned} IRW^D=\omega NIR. \end{aligned}$$

(19)

Likewise, fertilizer usage is modeled as a fixed ratio of the aggregate land:

$$\begin{aligned} FRT^D=\phi ({NIR + NRF}). \end{aligned}$$

(20)

The water and fertilizer usage are to be affected by fee/subsidy instruments (taxF and taxW) as introduced above.

In the land markets, the rental rates of the irrigated and rain-fed land types are determined endogenousy

$$\begin{aligned} NRF_A&= NRF^{SUP}\end{aligned}$$

(21)

$$\begin{aligned} NIR_A&= NIR^{SUP}. \end{aligned}$$

(22)

We specify a dualistic structure in the labor markets where rural and urban labor are differentiated. Rural labor market wages are fully flexible and the low productivity problem is revealed in low wages (rural poverty). Urban labor market is subject to real wage fixity and an endogenous unemployment mechanism is generated.

Within intertemporal dynamics, rural labor migrates into urban centers via a simple Harris–Todaro framework with migrants responding to expected urban wage rate and rural wage differences.

The amount of rural labor migrating to the urban labor market is determined through:

$$\begin{aligned} LMIG=\mu \frac{({EWU-W_{AG} })}{W_{AG} }LSUP_{AG}. \end{aligned}$$

(23)

Here $W_{AG}$ is the rural labor wage rate (flexible), and EWU is the expected urban wage rate. $\mu$ is an elasticity parameter used to control the responsiveness of the migration decision in response to the wage differentials. The expected urban wage rate is a weighted average of the (fixed) real urban wage rate and the sectoral employment levels in the urban sectors:

$$\begin{aligned} EWU=W_{URB} \sum \limits _{i\in Non-Ag} {\left( {\frac{L_i^D }{LSUP_{URB}}}\right) }. \end{aligned}$$

(24)

Given the migrated labor and supplies of both types of labor, urban labor market is cleared through quantity adjustments via unemployment:

$$\begin{aligned} UNEMP=LSUP_{URB} -\sum \limits _{i\in noAg} {LD_i }. \end{aligned}$$

(25)

Rural labor market wages are flexible:

$$\begin{aligned} LSUP_{AGR} =L_A^D. \end{aligned}$$

(26)

Likewise, given the aggregate physical capital stock supply in each period, the capital market equilibrium, $\sum \nolimits _{\mathrm{i}} {\text{ K }_\mathrm{i}} =\overline{K} ^S$ implies an equilibrium profit rate $r$ for the economy. Consequently, physical capital is mobile across sectors. It is the difference in sectoral profit rates that leads to the sectoral allocation of aggregate investments in within-period dynamics of the model.

1.1 Environmental pollution and instruments of abatement

The model accommodates two types of environmental pollution: gaseous emissions (in terms of CO$_{2}$ equivalents and PM10) and waste generation.

Waste is thought to be in “solid” and “water” discharge form and is generated from

• 1. urban waste (to be formulated as a ratio of urban consumption);

• 2. waste from industrial processes, and

• 3. waste from water usage in agricultural production.

On the other hand, three basic sources of $CO_{2}$ and PM10 emissions are distinguished in the model: (1) due to industrial processes, (2) due to (primary and secondary) energy usage, and (3) due to energy use of households. Total air emissions is the sum over from all these sources:

$$\begin{aligned} CO_2 EM_i =\sum \limits _j {CO_i EM_{j,i}^{INM} +CO_2 EM_i^{ENG} +CO_2 EM_i^{IND} }. \end{aligned}$$

(27)

Depending on the source of emission, we assume different allocation mechanisms of carbon dioxide. Following Gunther et al. (1992), the emissions from industrial processes is regarded to depend on the level of industrial activity, therefore is hypothesized proportional to gross output:

$$\begin{aligned} CO_2 EM_i^{IND} =\overline{\delta }_i XS_i. \end{aligned}$$

(28)

Total emissions due to energy usage, TOTCO $_{2}$ ENG are generated from two sources: sectoral emissions due to combustion of primary energy fuels (coal and petroleum and gas) and sectoral emissions due to combustion of secondary energy fuels (refined petroleum):

$$\begin{aligned} TOTCO_{2}ENG=\sum \limits _i {\left[ {\sum \limits _j {\left( {CO_2 EM_{j,i}^{INM} +CO_2 EM_{j,i}^{ENG} }\right) } } \right] }. \end{aligned}$$

(29)

Under both sources, the mechanism of emission is dependent on the level of pollutant-emitting inputs (energy input at primary and at secondary levels) in each sector:

$$\begin{aligned} CO_2 EM_{j,i}^{ENG}&=\varpi _{\text{ j,i }} \;ID_{j,i} \quad \quad j=CO,\;PG\end{aligned}$$

(30)

$$\begin{aligned} CO_2 EM_{j,i}^{ING}&=\overline{\varepsilon }_{\text{ j,i }} \;ID_{j,I} \quad \quad j=RP. \end{aligned}$$

(31)

Total emission of CO$_{2}$ in the use of energy by households is given by:

$$\begin{aligned} TOTCO_2 HH=\sum \limits _i {\overline{\psi }_i CD_i }. \end{aligned}$$

(32)

Here, $\overline{\psi }_i$ is the coefficient of emissions of CO$_{2}$ in private consumption ($CD_{i}$) of the basic fuels coal (CO) and refined petroleum (RP) by households.

Pollutant tax/fee can serve as one of the instruments and is thought to be introduced at per tons of carbon dioxide emitted, on production, on intermediate input usage and on consumption respectively. The revenues are directly added to the revenue pool of the government budget.

PM10 emissions and instruments of environmental policy with respect to PM10 are modeled in the same manner.

1.2 Income generation and demand

Private sector is aggregated into one household. Household income comprises returns to labor input, net of social security taxes, and land rental income. As was introduced under policy scenario 2 household income is further accentuated by transfers from the green wage fund.

The net profit transfer of the enterprise income to private household is mainly composed of returns to capital as a factor of production:

$$\begin{aligned} EtrHH&=({1-t_{Corp} })\sum \limits _{\mathrm{i}} {\text{ rK }_\mathrm{i} } -EERPtrROW-NFI^G+GtrEE \nonumber \\&\quad \times r^DDomDebt^G-r^FeForDebt^E+eForBOR^E. \end{aligned}$$

(33)

A constant proportion of the total profit income is distributed to the rest of the world to represent the net factor income of foreigners in Turkey. In Eq. 33, GtrEE is the net transfers of the government to private enterprises, $r^{D}$ DomDebt $^{G}$ is the interest income of the enterprises (banking sector) out of government domestic debt and $r^{F}$ ForDebt $^{E}$ is the interest payments of the private enterprises for their already accumulated foreign debt. As $e$ represents the exchange rate variable, ForBOR $^{E}$ is the new foreign borrowing of the private sector in foreign exchange terms.

Private household saves a constant fraction, $s^{p}$ of its income. The residual aggregate private consumption then is distributed into sectoral components through exogenous (and calibrated) shares:

$$\begin{aligned} CD_i =cles_i \cdot \frac{PRIVCON}{PC_i } \end{aligned}$$

(34)

where PC $_{i}$ is the composite price of product $i$ which consists of the unit prices of domestic and foreign commodities, united under the imperfect substitution assumption through an Armington specification. Aggregate public consumption is distributed into sectoral production commodities in the same manner with policy-driven public consumption shares. It is assumed that aggregate public consumption is a constant fraction of aggregate public income.

Government revenues are composed of direct taxes on wage and profit incomes and profit income from state economic enterprises. The income flow of the public sector is further augmented by indirect taxes on domestic output and foreign trade (net of subsidies), sales taxes and environmental taxes.

The set of environmental tax/fee instruments are tabulated in Table 3.

Table 3 Tax instruments used in the model

The model follows the fiscal budget constraints closely. Current fiscal policy stance of the government is explicitly recognized as specific targets of primary (non-interest) budget balance. We regard the government transfer items to the households, to the enterprises and to the social security system as fixed ratios to government revenues net of interest payments. Then, under a pre-determined primary surplus/GDP ratio, public investment demand is settled as a residual variable out of the public fiscal accounts.

The public sector borrowing requirement, PSBR then, is either financed by domestic borrowing, $\Delta \,DomDebt^{G}$ or by foreign borrowing $\Delta \, eForDebt^{G}$.

1.3 General equilibrium

The overall model is brought into equilibrium through endogenous adjustments of product prices to clear the commodity markets and balance of payments accounts. With real wages being fixed in each period, equilibrium in the urban labor market is sustained through adjustments of urban employment.

Given the market equilibrium conditions, the following ought to be satisfied for each commodity $i$:

$$\begin{aligned} CC_i = CD_i + GD_i + IDP_i + IDG_i + INT_i \end{aligned}$$

(35)

that is, the aggregate absorption (domestic supply minus net exports) of each commodity is demanded either for private or public consumption purposes, private or public investment purposes or as an intermediate good.

The model’s closure rule for the savings-investment balance necessitates:

$$\begin{aligned} PSAV + GSAV + e CAdef = PINV + GINV. \end{aligned}$$

(36)

The CAdef in the equation above determines the current account balance in foreign exchange terms and equals to the export revenues, the remittances and private and public foreign borrowing on the revenue side and the import bill, profit transfers abroad and interest payments on the accumulated private and public debt stocks on the expenditures side:

$$\begin{aligned} CAdef&=\sum {P_i^W } E_i +ROWtrHH+ForBor^E+ForBor^G \nonumber \\&\quad -\left[ \sum {P_i^W M_i +(trrow\sum {({1-t_{Corp} })rK_i}})/e\right. \nonumber \\&\quad \left. +r^F\;ForDebt^E+r^FForDebt^G \right] . \end{aligned}$$

(37)

The private and public components of the external capital inflows are regarded exogenous in foreign exchange units. The additional endogenous variable that closes the Walrasian system is the private investments, PINV. Finally, the exchange rate $e$, serves as the numeriare of the system.

1.4 Dynamics

The model updates the annual values of the exogenously specified variables and the policy variables in an attempt to characterize the 2011–2030 growth trajectory of the Turkish economy. In-between periods, first we update the capital stocks with new investment expenditures net of depreciation. Labor endowments are increased by the respective population growth rates. Similarly, technical factor productivity rates are specified in a Hicks-neutral manner, and are introduced exogenously. Urban nominal wage rate is updated by the price level index which is endogenous to the system.

Finally, at this stage we account for the evolution of debt stocks. First, government’s foreign borrowing is taken as a ratio to aggregate PSBR:

$$\begin{aligned} e\;ForBor^G= ({gfborrat})PSBR. \end{aligned}$$

(38)

Thus, government domestic borrowing becomes:

$$\begin{aligned} DomBor = ({1 - gfborrat}) PSBR. \end{aligned}$$

(39)

Having determined the equations for both foreign and domestic borrowing by the government, we establish the accumulation of the domestic and foreign debt stocks of the public sector:

$$\begin{aligned} DomDebt_{t+1}&= DomDebt_t + DomBor_t\end{aligned}$$

(40)

$$\begin{aligned} ForDebt^G_{t+1}&= ForDebt^G_t + ForBor^G_t. \end{aligned}$$

(41)

Similarly, private foreign debt builds up as:

$$\begin{aligned} ForDebt^P_{t+1} = ForDebt^P_t + ForBor^E_t. \end{aligned}$$

(42)

TFP increase is one of the drivers of growth; various assumptions held in greening scenarios are detailed in the main text. In the reference scenario, TFP growth is specified as:

$$\begin{aligned} AX_{t+1}^i =({1+tfpGR_i })AX_t^i. \end{aligned}$$

(43)

Capital and labor growth follows standard specification as:

$$\begin{aligned} \overline{K} _{t+1}^S&=({1-dprt})K_t^S +\sum \limits _i {({IDP_i +IDG_i })} \end{aligned}$$

(44)

$$\begin{aligned} \overline{L} _{t+1}^S&=({1+popgr_t })\overline{L} _t^S. \end{aligned}$$

(45)

Appendix 2: description of the CGE model calibration and base path (2011–2030)

1.1 Data

The model is built-around a multi-sectoral SAM of the Turkish economy based on the Turkish Statistical Institute (TurkStat) 2002 Input Output (I/O) Data. The I/O data is re-arranged accordingly to give a structural portrayal of intermediate flows at the intersection of commodities row and activities column in the 12-sector 2010 macro-SAM. Table 4 provides the sectoral input-output flows of the macro SAM in correspondence with the TurkStat I/O data.

Table 4 Sectoral aggregation over TURKSTAT 2002 I/O Data

1.2 The base path 2011–2030

All alternative policy scenarios analyzed in this report are to be portrayed with respect to a base-path reference scenario. Having calibrated the parameter values, we construct a benchmark growth path for the Turkish economy for the period of 2011–2030, under the following assumptions:

• Constant technology (calibrated parameters in the production functions remain fixed).

• Exogenously determined foreign capital inflows.

• Exogenous real interest rates.

• Endogenous real exchange rate under the constraint of the current account balance.

• Constant nominal wage rate for urban labor.

• Fiscal policy in accordance with the announced policy rule of targeted primary surplus. Domestic interest rates (net costs of domestic debt servicing) are reduced over to 5 % by 2015 onwards from their base values of 8 % in 2010. The ratio of primary (non-interest) surplus is initially set at 0.04 as a ratio to the GDP over 2011–2015. As a result of reduced interest costs on public domestic debt then, it is gradually reduced to 0.0 by 2020 and is kept at that level over the rest of the base path.

• No specific introduction of environmental policy action/taxation/quota.

Furthermore, population growth rate is set at 1 % for rural labor until 2020, then to be decreased to 0.7 % per annum. Urban labor force is assumed to increase by 0.5 % per annum. Migration elasticity parameter, $\mu$, is taken as 0.02 to match historical data on migration as reported in Çakmak et al. (2008).

Hicks-neutral productivity growth is assumed at an exogenous rate of 0.5 % for agriculture and 0.8 % for the non-agriculture sectors. In some of the scenarios below, we have implemented submodels to create endogeneity of TFP growth in response to health and environmental benefits.

The total available irrigated a land is assumed to expand by 0.5 % per annum. Rate of depreciation for physical capital stock is set at 0.20.