2. Recently, Thailand emerged to be one of the leading producers of biofuels in Asia (Zhou & Thomson, 2009). It is the result of serious effort to reduce oil import dependency (Russell & Frymier, 2012).

3. National Alternative Energy Development Plan (2004- 2011) To promote the biofuel production To reduce its dependency on imported oil Alternative Energy Development Plan (2008- 2022) Short term plan (2008-2011) 3.0 million liters per day Medium term plan (2012-2016) 6.2 million liters per day Long term plan (2017-2022) 9.0 million liters per day New 10-year Alternative Energy Development Plan (2012-2021) Targeted consumption of bioethanol unchanged -9.0 million liters per day by 2021 Develop several strategic plans focusing on supply and demand sides BioethanolBiodiesel Unpredictable weather Lack of suitable land Limited palm growing area Feedstock supply readiness

5. StudiesBiofuels Policies Saunders et al. (2009) McPhail and Babcock (2012) Miljkovic, Shaik and Braun (2012) U.S Renewable Fuel Standard (RFS) mandates Banse et al. (2008) Kretschmer, Narita and Peterson (2009) Kim et al. (2013) EU biofuels policies Qiu et al. (2010)China’s current bioethanol program StudiesVariables of energy and agricultural commodities Kanamura (2009) Ciaian and Kancs (2011a) Ciaian and Kancs (2011b) Investigate the traditional oil-agricultural commodities relationship by accounting for the upsurge of biofuels. Natanelov, McKenzie and Huylenbroeck (2013) Cha and Bae (2011) Gardebroek and Hernandez (2013) Wu and Li (2013) Du and McPhail (2012) Study the linkages of oil, bioethanol and corn. Vacha et al. (2013) Kristoufek, Janda and Zilberman (2012) Examine the effect of changes in bioethanol prices on a group of energy and agricultural commodity variables. StudiesLand-use framework Monteiro, Altman and Lahiri (2012) Bryngelsson and Lindgren (2013) Ge, Lei and Tokunaga (2014)  Employed the land-use framework to measure impact of production of bioethanol on food prices.  Effect of expanding bioethanol production has induced more land competition between bioethanol and food crops (Ge, Lei & Tokunaga, 2014).

6. Figure 2.1: Market Equilibrium of Biofuel Source: Adopted from Kristoufek et al.(2012). Based on McPhail and Babcock (2012), Demands for corn are modelled as below: D ce = a dce + b dce P c + e dce (1) D co = a dco + b dco P c + e dco (2) Supply of corn is formed as follow: S c = a sc + b sc P c + e sc (3) Based on McPhail and Babcock (2012), Demands for corn are modelled as below: D ce = a dce + b dce P c + e dce (1) D co = a dco + b dco P c + e dco (2) Supply of corn is formed as follow: S c = a sc + b sc P c + e sc (3) Figure 2.2: The Decisions of Producer and Consumer Source: Developed by the researchers P BT P BR PBPB R E T E2E2 S(P B, P F ) D(P B, P G ) BRBR BTBT B E1E1 Monteiro (2009)

7. Unit Root tests (ADF & PP) Auto Regressive Distributive Lag (ARDL) Granger Causality test (Wald test) Source: Adopted from Hirshleifer, Glazer and Hirshleifer (2005). S ƩS” f ƩS’ f P Q 0 Q’Q” ^ P’ P” Q KH G

8.  To avoid any model misspecification bias, we employ lin-log model which followed the study of Fanaei, Khansari and Maskooki (2008).  From Eq. 12, when FS t,cost of bioethanol production, ETH t ; vice versa.  From Eq. 13, when ETH t, demand for feedstock, FS t ; vice versa. Where = Production of bioethanol (Million liter per month) = Farm gate price of sugarcane (Bath per ton) = Export price of cane molasses (Bath per kilogram) = Farm gate price of cassava (Bath per kilogram) ε = Residual of the model

9.  Used to detect unit root or non-stationary in a time series. H 0 : = 0 (has unit root) H 1 : < 0 (no unit root) Reject H 0 if computed absolute value of > critical values. Otherwise, do not reject H 0 ADF Solve serial correlation in error term by adding lagged variable of dependent variable (Gujarati & Porter, 2009, pp.758). PP Solve serial correlation in error term without adding any lagged variable. Able to solve heteroscedasticity in error terms (Phillips & Perron, 1988).

10.  A general dynamic specification model which includes the lags of the endogenous variable and the lagged of exogenous variables to estimate the short-run effects directly and the long- run equilibrium relationship indirectly (Royfaizal, 2009).  The short-run parameters are represented by and. The long-run parameters are denoted as and. represents, and respectively. Do not required pretesting for unit root on series, can be mixture of integration order Cointegrating relationship can be determined efficiently even in small sample size Different variable to have different optimal lags Only involved one reduced form equation to test long run relationship Advantages (Sari & Soytas, 2009; Duasa, 2007)

11.  Use to determine the direction of causality among variables in a model once the ARDL cointegration test had identified the variables having a long-run relationship (Ozturk & Acaravci, 2010).

13. Table 2: Bound testing and Diagnostic test Result

16. 5: