Climate Change and Adaptation: The Case of Nigerian Agriculture

Abstract

The present research offers an economic assessment of climate change impacts on the four major crop families characterizing Nigerian agriculture. The evaluation is performed by shocking land productivity in a computable general equilibrium model tailored to replicate Nigerian economic development up to 2050. The detail of land uses in the model has been increased by differentiating land types per agro-ecological zones. Uncertainty about future climate is captured, using, as inputs, yield changes computed by a crop model under ten general circulation models runs. Climate change turns out to be negative for Nigeria in the medium term, with production losses and increase in crop prices, higher food dependency on foreign imports, and GDP losses in all the simulations after 2025. In a second part of the paper, a cost effectiveness analysis of adaptation in Nigerian agriculture is conducted. The adaptation practices considered are a mix of cheaper “soft measures” and more costly “hard” irrigation expansion. The main result is that the cost effectiveness of the whole package depends crucially on the possibility of implementing adaptation by exploiting low-cost opportunities which show a benefit-cost ratio larger than one in all the climate regimes.

Appendix

1.1 The Core of ICES Model

The Intertemporal Computable Equilibrium System (ICES) model¹⁴ is multi-regional CGE model of the world economy, built upon the static GTAP-E model (Burniaux and Truong 2002), which in turn is an extension of the basic GTAP model (Hertel 1997). Industries are “typically” modelled through a representative cost-minimizing firm, taking input prices as given. In turn, output prices are given by average production costs. The production functions are specified via a series of nested CES functions. Peculiar to ICES is the “isolation” in the production tree of energy factors which are taken out from the set of intermediate inputs and are inserted as primary production factors in a nested level of substitution with capital. The following figure shows the production structure of the model.

Fig. 6

ICES nested production function

At the top of Fig. 6, production stems from the combination of intermediate inputs (QF) and a value added composite including all primary factors and energy (QVAEN). Perfect complementarity is assumed between value added and intermediates. This implies the adoption a Leontief production function. For sector i in region r final supply (output) results from the following constrained production cost minimization problem for the producer:

$$\begin{aligned}&\min PVAEN_{i,r} QVAEN_{i,r} +PF_{i,r} QF_{i,r} \\&s.t. \,Y_{i,r} =\min \left[ QVAEN_{i,r} ,QF_{i,r} \right] \end{aligned}$$

where PVAEN and PF are prices of the related production factors.

The second nested-level in Fig. 6 represents, on the left hand side, the value added plus energy composite (QVAEN). This composite stems from a CES function that combines four primary factors: land (QLAND), natural resources (QFE), labour (QFE) and the capital-energy bundle (QKE) using $\sigma$VAE as elasticity of substitution. Primary factor demand on its turn derives from the first order conditions of the following constrained cost minimization problem for the representative firm:

$$\begin{aligned}&\min P_{i,r}^{Land} LAND_{i,r} +P_{i,r}^{NR} NR_{i,r} +P_{i,r}^L L_{i,r} +P_{i,r}^{KE} KE_{i,r} \\&s.t.\, QVAEN_{i,r,t} =\left. {\left( {LAND_{i,r}^{\frac{\sigma _{VAE} -1}{\sigma _{VAE} }} +NR_{i,r}^{\frac{\sigma _{VAE} -1}{\sigma _{VAE} }} +L_{i,r}^{\frac{\sigma _{VAE} -1}{\sigma _{VAE} }} +KE_{i,r}^{\frac{\sigma _{VAE} -1}{\sigma _{VAE} }} } \right. } \right) ^{\frac{\sigma _{VAE} }{\sigma _{VAE} -1}} \end{aligned}$$

In the third nested-level, the QLAND bundle combines the AEZ-specific land types and the KE bundle combines capital with a set of different energy inputs. This is a peculiarity of GTAP-E and ICES model. In fact, energy inputs are not part of the intermediates, but are combined to capital in a specific composite.

Furthermore, Energy is produced using Electric and Non Electric commodities in the Fourth nested-level, while the Non Electric commodity is produced using Coal and Otherfuel commodities. At the basic level of the production tree, there are Gas, Oil, Petroleum Products and Biofuels.

Notice that domestic and foreign inputs are not perfect substitutes, according to the so-called “Armington assumption”, which accounts for—amongst others—product heterogeneity. In general, inputs grouped together are more easily substitutable among themselves than with other elements outside the nest. For example, imports can more easily be substituted in terms of foreign production source, rather than between domestic production and one specific foreign country of origin. Analogously, composite energy inputs are more substitutable with capital than with other factors.

A representative consumer in each region receives income, defined as the service value of national primary factors (natural resources, land, labour, capital). Capital and labour are perfectly mobile domestically but immobile internationally. Land and natural resources, on the other hand, are industry-specific. This income is then used to finance three classes of expenditure: aggregate household consumption, public consumption and savings as depicted in the Fig. 7.

Fig. 7

ICES nested tree structure for final demand

Thus, the upper level represented in Fig. 7, mathematically translates into a Cobb-Douglas utility constrained maximization problem:

$$\begin{aligned} \max U= & {} C\mathop \prod \limits _i U_i^{B_i } \\ subject \,to\, X= & {} \mathop \sum \limits _i E_i \left( {P_i ,U_i } \right) \end{aligned}$$

where $U_{i}$ are the per capita utility from private consumption, per capita utility from government consumption, and per capita real savings; C is a scaling factor and $B_{i}$ are distribution parameters. X describes the budget constraint which must meet the sum of three types of expenditures $E_{i}$. $P_{i}$ is the expenditure-share-weighted index of commodity group price indices.

At the second level, per capita utility from private consumption is derived from the aggregation of per capita private consumption of individual commodities. This is done using the Hanoch’s constant difference elasticity (CDE) demand system (Hanoch 1975).

$$\begin{aligned} 1=\mathop \sum \limits _i B_i U^{\Upsilon _i R_i }\left( {\frac{P_i }{X}} \right) ^{\Upsilon _i } \end{aligned}$$

where U denotes utility, $P_{i}$ the price of commodity i, X the expenditure, $B_{i}$ are distribution parameters, $Y_{i}$ substitution parameters, and $R_{i}$ expansion parameters.

1.2 Endogenous Dynamics

ICES model is a recursive dynamic model. This means it presents a sequence of static equilibria which are inter-temporally connected by the process of capital accumulation. Capital growth is standard along exogenous growth theory models and follows:

$$\begin{aligned} Ke_r =I_r +\left( {1-\delta } \right) Kb_r \end{aligned}$$

where Ke $_{r}$ is the “end of period” capital stock, Kb $_{r}$ is the “beginning of period” capital stock, $\delta$ is capital depreciation and $I_{r}$ is endogenous investment. Once the model is solved at a given step t, the value of Ke $_{r}$ is stored in an external file and used as the “beginning of period” capital stock of the subsequent step $t+1$.

The hearth of the model’s dynamics is the endogenous determination of investment demand $I_{r}$. Sources of world investments are savings from households. Regional households save a given share of their income which is firstly “pooled” by a “world bank” and then redistributed back to each region following:

$$\begin{aligned} I_r =\varphi _r RGDP_r e^{\left[ \left( {\rho _r \left( {R_r^E -R^{w}} \right) } \right. \right] } \end{aligned}$$

where RGDP is real GDP, $\rho _{r}$ and $\varphi _{r}$ are given parameters, $R_{r}^{E}$ and $R^{W}$ are the expected rate of return to capital in region r and the world rate of return to capital respectively. According to the previous equation, each region demands investment as long as its real GDP rises or its expected rate of return is higher than the world rate of return $R^{W}$. Investment demand is negatively correlated to $R^{W}$ which on its turn is determined by the general equilibrium condition requiring equalization between global savings and investments. The parameter $\rho _{r}$ reflects the flexibility of capital movement related to changes in the current rate of return. If $\rho _{r}$ has a small value then it will reduce the effect of the growth of the current rate of return when compared with the growth of the global rate of return; basically it can be assumed to reflect policy restrictions. $R_{r}^{E}$ needs a particular comment: ICES does not generate endogenously the expected rate of return to capital according to a fully rational expectation generation process of a forward looking agent; more simply it is assumed that the expected rate of return to capital coincides with the current observed rate of return to capital.

The world investment supply (savings) must match world investment demand, but this is not necessarily so at the regional level. Indeed a region can run a foreign debt or credit position as long as Sr $\ne$ Ir. This will be reflected in disequilibrium in the trade balance.

In ICES model, also the stock of natural resource has an endogenous dynamics. As explained in Hertel et al. (2008), initial calibration values of these variables in the original GTAP database are not obtained from official statistics, but are indirectly estimated to make the model consistent with industry supply elasticity values from the literature. Then to represent in ICES availability of additional resources due to new discoveries, the price of natural resources has been fixed exogenously, making it variable over time in line with exogenous projections, while allowing the model to compute endogenously the corresponding stock levels.

1.3 Exogenous Dynamics

Capital and natural resources are not the only factors expected to vary over time. Population stock, labour stock, labour and land productivity change over time because of natural or technological evolutionary processes. These processes have been also taken into account in the baseline. This has been done by updating exogenously year by year the initial calibration data of all the above mentioned variables according to their expected rates of change.

1.4 GTAP Database

The model and database are calibrated for year 2004, which constitutes also the beginning year for simulations. ICES model relies on GTAP 7 Data Base (Narayanan and Walmsley 2008) with world coverage: countries are aggregated in 113 macro-regions, and all economic sectors, grouped in 57 sectors.

Furthermore, the database and the model account for main GHG emissions $\hbox {CO}_{2}, \hbox {CH}_{4}$ and $\hbox {N}_{2}\hbox {O}$. Following Burniaux and Truong (2002), $\hbox {CO}_{2}$ emissions are calculated proportionally to energy combustion. Data relative to energy volumes are also included in the GTAP 7 Data Base. Emissions of other greenhouse gases, namely methane and nitrous oxide, are also included in ICES. Data relative with emissions of these gases have been calculated starting from the GTAP $\hbox {non-CO}_{2}$ emissions database (Lee 2003).