Methodology
4.4 Ricardian Model Specification
As noted by Kumar (2009) crop growth and the behaviour of the producers of agricultural goods would be influenced by changes in climate because such changes should be considered as a change in input structure. Kumar (2009) additionally describes a production function F which considers k purchased inputs and l climate inputs relate to the output. Letting Pi and Yi be the output
price and quantity of the ith good, respectively, Xij the quantity of the jth
purchased input used in the production of the ith good, and qj the price of the jth
purchased input, the profit-maximizing behaviour of the producer can be expressed as:
(1)
which is then subject to a production function:
j
87 Yi F (Xi1, Xi2, …, Xik, E1, E2, …, El) (2)
This specification is different from the conventional one since the environmental/climate inputs (variables E in the above equation) are included. Although there is no market for climate inputs, profits, input demands and output supply can be theoretically represented as functions of measured market inputs and climate variables (Kumar, 2009). However, it is difficult to obtain the functional relationship between output and changes in climate inputs based on an associated econometric analysis, hence researchers often partition the production function expressed in equation (2). To measure the supply shifts in the case of agriculture, researchers first estimate yield changes and then introduce them into economic models. While scholars commonly use such neutral technology change assumptions in the literature on climate change impacts, Kumar (2009) argued that it is not necessary to make such an assumption. Thus, equation (2) becomes:
Yi F1 (Xi1, Xi2, …, Xik)* F2( E1, E2, …, El ) (3)
Kumar (2009) asserts that such partitioning can indicate fairly complex technical relationships among market inputs as described by econometrically related production relationships and among climate inputs as described by crop simulation models. In order to assess the economic and welfare implications, researchers often integrate the crop responses to climate parameters, estimated using crop simulation models, with either a partial or general equilibrium framework (e.g. Lobell et al., 2005; Lobell and Field, 2007; Wang et al., 2008;
88 Chen et al. 2010; Lobell et al. (2011); Muller et al. (2011); Auffhammer et al., 2012).
According to Kumar (2009) the Ricardian approach, on the other hand, integrates the climate response curves of numerous crops to arrive at the overall crop response curve with regard to different crops have different climatic requirements. In a clairvoyant farmer scenario, the farmer would willingly shift from one crop to another rather than suffer the losses from not shifting over. It is clear that the transition between crops would involve costs. Thus, to take into account the costs and benefits of adaptation, the relevant dependent variable should be net revenue or land values (that is, capitalized net revenues) and not yields. Therefore, the Ricardian approach estimates a variant of equation (2). Kumar (2009) explained that climate change impacts can be measured as changes in net revenue or land value as shown below.
Consider a crop with the aggregate demand Yi and let the production function be
as shown in equation (2). There will be a cost function (obtained through cost minimization) associated with Q (which expresses the set of prices of the inputs used in the production), E and Yi, given by equation (4):
Ci = Ci (Yi, Q , E ) (4)
where, Ci is the cost of production of good i. Excluding ‘land’ out of the vector
of inputs X and taking its annual rent as
ql
, the profit maximization equation can be written as:89 where, Li is the amount of land used for producing Yi. Under perfect competition
for land, the rent of land can be expressed as:
(6)
If ‘i’ is the best use for the land, given the environment E and factor prices Q, the observed market rent on the land will be equal to net profits from the production of good ‘i’. Land value, which is the present value of the stream of revenue over time, can be defined as:
(7)
The Ricardian approach investigates the relationship between land rent (equation 6) or land value (equation 7) and the independent variables, P, Q, and E.
Under the assumption that market prices will remain unchanged given environmental changes, then the welfare value of a change in the environment can be written as:
W(EA – EB) = [PYB - Ci(Yi, Q, EB )] - [PYA - Ci(Yi, Q, EA)] (8)
Substituting equation (6) in equation (8), then:
W(EA – EB) = (
qlB
LEB -qlA
LEA) (9)where
qlA
andqlB
are land prices under different environmental conditions. Alternatively, the present value of this welfare change can be given by:W(EA – EB) e -t dt = (VlB - VlA) (10) Li [Pi Yi - Ci (Yi, Q , E )]
ql
= 0 0 Vl = ql
e -t dt 90 The definitions of the Ricardian estimate of the value of environmental changes are expressed in equations (9) and (10). It can be assumed that output prices are unchanged under changed climate conditions, the change in aggregate land values or the change in the present value of net revenues then captures exactly the value of the change in the climate. This variant of the Ricardian approach can be applied due to the non-existence of well functioning land markets in developing countries (Dinar et al., 1998; Kumar, 2009).
As suggested by Kumar (2009) the empirical strategy of the Ricardian study is to estimate a functional relationship between land value, or net revenue, and climate variables using cross-sectional data while controlling for variables that could cause variability in the dependent variable. Variability in the dependent variable caused by factors other than climate can be controlled through:
(a) soil characteristics (soil quality could differ significantly across the cross- section could lead to variability in the farm-level net revenue);
(b) the level of technology penetration (wide spread across the cross-sectional units in terms of mechanisation, and penetration of growing innovations leading to variability in the dependent variable);
(c) the extent of development (different opportunity cost of land and market access and alternative livelihood opportunities across the cross-sectional units could be contributable to the variability in the farm-level net revenue).
91 NR = (Tj, T2j, Rj, R2j, TjRj, SOIL, BULLOCKS,TRACTORS, CULTIVATORS,
POPDEN, LITPOP, IRR) (11)
where, NR represents farm-level net revenue per hectare and T and R represent temperature and rainfall respectively. According to Meldelsohn and Dinar (2003); Gbetibouo and Hassan (2004) and Kumar (2009), it is noteworthy that a quadratic functional specification along with climate interaction terms should be adopted in each study. As noted by Kumar (2009), the climate coefficients have not significantly changed when they include the prices of major cereal crops in the model specification. However, no evidence exists from previous studies about the influence of input prices. It is therefore assumed their cross-sectional variation is not significant (Kumar, 2009).
Other explanatory variables that may be included are soil; farm households own bullocks, tractors and cultivators; population density; literate population and irrigated land where soil represents soil quality, farm households own bullocks, tractors and cultivators represent the extent of mechanization, whereas population density, literate population and irrigated land represent the extent of development.
To examine how changing climate variables including temperature and precipitation are impacting the farmers’ income in NE Thailand, the Ricardian approach is selected. As there is a variety of crops cultivated in the region, such as rice, cassava, sugarcane, maize, para rubber, fruits and vegetables. The Ricardian approach may be successfully applied to the case of NE Thailand as
92 the geographical distribution of the crops under study seem to be highly correlated with variation in climate patterns across the region. In addition, the Ricardian analysis has been applied at very small geographic scales (Salvo et al, 2013). It is therefore practical to employ the Ricardian model to NE Thailand. As highlighted by Timmins (2006), the Ricardian model can use readily available data on land values or net revenues from agricultural production, therefore eliminating the need for costly field studies or the collection of expensive panel data over long period of time. These agricultural data are readily accessible in Thailand, which enables this study to carry out the Ricardian analysis for NE Thailand. Moreover, the Ricardian method can provide a useful starting point for policy interventions. According to Stage (2010), a Ricardian study can help identify the production patterns that farmers are likely to switch to, given the anticipated changes in climate; policy makers and analysts can use these projections to identify policy measures that can make it easier for farmers to switch to these new production patterns.