# Table 3 Maize model specifications

 Area harvesteda Hm = δ0 + δ1Pmt + δ2Spt +δ3RLt + δ4Tt+ εt  1(a) Hm = θ0 + θ1ΔPmt + θ2 ∆ Pst+θ3∆RLt − λ (Yt − α0 − α1Pmt − α2Pst − α3RLt − α4Tt) + νt   1(b) Yield Ym= Rpt + IRmt + SDit + $${T}_t^L$$ Production MP = Hm ∗ Ym Beginning stocks Bs= $${E}_{t-1}^s$$ Supply Xm= MP + Bs + Im Ending stocks Es = Bs + MP − Pw + Af Per capita consumption Cm=Sw- Pw+Gd+S05+S11-Tt Human consumption HConsm = Cm ∗ Pop Domestic use DUSEM=FEEDm+SEEDm+HConsm Demand Dm=EXPOm+DUSEM + Es Model closure Xm = Dm (market clearing price)
1. Variable names
2. Hm = Planned maize acreage proxied by area harvested in thousand hectares
3. Pmt= Deflated maize producer price (ETB/ton). It is obtained by dividing the nominal maize producer price by CPI indexed at the 2010 price
4. Spt= Real producer prices for competing crop (sorghum) (ETB/ton)
5. RLt= Rainfall amount (mm) prior to sowing period
6. Tt = a time trend variable used as a proxy for technological progress in maize farming
7. Ym = Maize yield (ton/ha)
8. Rp = Rainfall amount for production season. It includes average rainfall (mm) for the months of June, July, August, and September
9. IRm = Irrigated maize area (ratio)
10. SDi = Maize planted with improved seed (ratio)
11. $${T}_t^L$$ = Linear trend to capture the overtime effects of maize technological improvement on yield
12. MP= Maize production
13. Bs = Beginning stock
14. $${E}_{t-1}^s$$= Lagged ending stock
15. Xm = Total maize supply
16. Im = Maize import
17. Pw= Real wholesale maize price (ETB/ton)
18. Af= Wheat food aid quantity (tons)
19. Cm= Per capita maize consumption
20. Sw=Real wholesale sorghum price (ETB/ton)
21. Gd=Real per capita GDP
22. S05= a shift variable for the period of soaring food prices in the domestic grain market. It takes 1 for the period since 2005 and 0 otherwise
23. S11=a shift variable for the export ban; 1 for the period since 2011 and 0 otherwise
24. Pop= Ethiopia population
25. HConsm = Human maize consumption
26. DUSEM = Total domestic use of maize
27. FEEDm = Feed use of maize
28. SEEDm = Seed use of maize
29. Dm = Total maize demand
30. EXPOm= Maize export
31. aThe equation for maize supply response was estimated following the two-stage procedure proposed by Alemu et al. (2003). First, a static long-run equilibrium regression is given by Eq. 1(a), was estimated. Second, a dynamic error correction model as specified in Eq. 1(b) was estimated by including the lagged residual from Eq. 1(a) (of course, the residual from Eq. 1(a) should be stationary). The planned maize acreage is proxied by area harvested. It has been common practice to proxy an acreage decision by area harvested because of the lack of data on area planted (Meyer and Kirsten 2010). Area harvested is preferred to output because farmers have more control on the former than on the latter. This is because agricultural output is subject to fluctuations, which are beyond the control of farmers.