Klaus Deininger, Daniel Ayalew, Tekie Alemu

Land Rental in Ethiopia

Marshallian inefficiency or factor market imperfections and tenure insecurity as

binding constraints?

Klaus Deininger, Daniel Ayalew, Tekie Alemu

Motivation

• Land is a key productive asset and source of income– the livelihood of more than 80% of the population

depends on subsistence agriculture

• Scarce land endowment under the threat of degradation & high level of climatic shocks

• Productive use of land & land markets important for poverty alleviation – directly & indirectly

Objectives

• To explore the efficiency of sharecropping in the context of state-controlled land ownership– Marsahallian inefficiency VS costless monitoring and

enforcement of contracts

• To investigate the extent to which land markets bring the ratio of inputs into line for efficient production– focus on the role of transaction costs and

imperfections in other input markets (labor and oxen)

Ethiopian context: framework

• Land policy: constitutional “use rights”, but the policy is still ambiguous– everyone who wants to engage in agriculture

has the right to receive a piece of land for free• this provision may conflict with the desire for

greater tenure insecurity

– rental allowed with restrictions; but sale, exchange or mortgage is prohibited

– responsibility for enacting laws is transferred to regional governments

Ethiopian context: regional policies

• No clear cut position regarding redistribution– upon request of the community (Amhara, Southern)– redistribution of irrigated land (Amhara, Oromia, Southern)– no clear statement about redistribution in Tigray

• Restrictions on rental markets– can rent out only 50% of their holdings (Oromia, Southern)

• A holder can lose his use rights if permanently employed in non-farm activities (e.g. Amhara)- implications on long term rental contracts & off-farm employment

Ethiopian context: final remark

• Legal provisions lack clarity & weak institutions to interpret regulations and adjudicate in case of disputes– 47% of households in our data expect to gain or lose

land due to redistribution– 1997 Amhara redistribution is still fresh

• limit investment incentives and the propensity to rent-out land

• Recently launched ambitious land certification program (6 mn households by late 2005)

Evidence for Ethiopia

• Empirical evidence:– no Marshallian inefficiency in terms of variable input

application (Fafchamps and Pender 2006, Gavian and Ehui 1999; Kassie and Babigumira 2006)

– mixed results regarding the role of rental market in making adjustments towards desired cultivated area

• partial adjustment (Teklu and Lemi 2004, Holden and Ghebru 2006)

• relatively efficient land lease market (Fafchamps and Pender 2006)

Data

• A three years panel data from Amhara highlands (covers 1999, 2001, and 2004 agricultural years)

• 1520 randomly selected households from 12 villages (kebeles) in 6 districts (woredas)

• Detailed household and plot level information– household resource endowments, participation in

different markets, input and crop output at the plot level, etc.

Descriptive statistics

Table 1: Household level descriptive statistics

Variable Total By land market participation status

Leased-out Autarky Leased-in

Household characteristics

Female head dummy (%) 15.37 39.91 11.82 3.29

Head literate (%) 41.31 29.13 41.24 51.14

Age of head in years 48.38 51.61 48.71 45.11

No. of rooms in dwellings 1.95 1.74 1.95 2.13

Owned area in ha 1.08 1.04 1.11 1.06

Number of adult males 1.49 1.04 1.57 1.70

Number of adult females 1.46 1.33 1.51 1.46

No. of dependents (<15 or >60) 2.72 2.26 2.74 3.05

Agricultural production

Net land leased-in 0.01 -0.68 0.00 0.60

Cultivates any land (%) 63.3

Owned cultivated area in ha 0.94 0.36 1.10 1.05

Total cultivated area in ha 1.09 0.36 1.11 1.65

Share of good quality land (%) 38.43 36.67 39.44 37.72

Owns bulls or oxen (%) 67.3 23.7 72.4 91.2

No. of bulls/oxen (for owners) 1.19 0.39 1.21 1.78

Value of other livestock in (‘000) Birr 1.20 0.57 1.20 1.69

Number of observations 4268 872 2301 1095

Percent 100 20 54 26






Head literate (%) 41.31 29.13 41.24 51.14



Owned area in ha 1.08 1.04 1.11 1.06














Percent 100 20 54 26






Head literate (%) 41.31 29.13 41.24 51.14



Owned area in ha 1.08 1.04 1.11 1.06














Percent 100 20 54 26

Table 2: Plot characteristics for owner-cum-sharecroppers

Owner-cum-sharecroppers All Own cult. Shared in

Relation w. partner (TP) TP relative (%) 59.54 59.54 TP neighbor (%) 15.48 15.48 TP member of VA (%) 14.12 14.12 TP unrelated (%) 10.86 10.86 Harvest share received (%) 53.29 53.29 Yield and input intensity Plot size in ha 0.26 0.25 0.28 Crop output/ha (Birr) 2868.60 2973.47 2480.64 Used fertilizer (%) 40.33 41.54 35.84 Fertilizer used per ha (kg) 100.86 106.49 81.50 Male family labor/ha (d) 178.72 189.69 138.15 Fem. family labor/ha (d) 90.53 97.96 63.01 Total family labor/ha (d) 269.25 287.65 201.16 Used hired labor (%) 17.41 16.10 22.28 Hired labor/ha (d) 50.98 52.95 45.73 Pair of oxen days per ha 56.47 56.82 55.21 Source of traction Own pair of oxen (%) 67.34 66.76 69.48 Oxen exchange for labor (%) 10.93 12.05 6.78 Oxen sharing (%) 18.85 18.26 21.01 Gift/support (oxen party) (%) 2.00 1.95 2.17 Oxen rental (%)1 0.70 0.76 0.50 Number of observations 8562 6740 1822 Number of households 606 606 606







Model: Marshallian inefficiency

• Test for the differences in input intensity and yield on owned and sharecropped plots of the same household (Shaban 1987)

• Estimate household fixed effects model of yield and input intensity functions on ownership dummy, plot characteristics and time-varying district dummies

• Test if the coefficient of ownership dummy is statistically different from zero

Model: optimal adjustmentFriction model (Rosett 1959)

• This model accounts for three observations: positive, zero, and negative net land leased-in

• We allow for asymmetric coefficients• Right-hand side variables: owned land, other

input endowments (family labor and oxen), wealth indicators and other household characteristics

• We extend the model to capture farming ability• Measurement of farming ability

– use household fixed effects from a panel Cobb-Douglas production function

Empirical Results

Table 3: Determinants of value of crop output per ha for owner-cum-sharecroppers: Household fixed effects estimates

Log(value of output per

ha) Ownership dummy 0.032 (0.96) Number of years possessed -0.002 (0.96) Good soil quality 0.143*** (4.25) Medium soil quality 0.114*** (3.60) Flat land 0.057 (0.93) Gently sloped land 0.074 (1.19) Irrigated land 0.209*** (3.14) Constant 7.076*** (105.03) Observations 8562 Number of households 606 R-squared 0.05

Absolute value of t statistics in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%.

Table 4: Determinants of input intensity per ha for owner-cum-sharecroppers

Household Fixed Effects Random Effects Tobit: Marginal Effects (dlny/dx)

Variable

Family male labor per ha

(log)

Family labor per ha (log)

Pair of oxen days per ha

Quantity of fertilizer per ha

Ownership dummy 0.126*** 0.155*** 0.031 0.309*** (4.70) (5.76) (0.98) (5.05) Number of years possessed -0.002* -0.002 -0.002* -0.006** (1.73) (1.38) (1.91) (2.33) Good soil quality 0.036 0.064** 0.039 -0.163*** (1.33) (2.39) (1.24) (2.78) Medium soil quality 0.020 0.039 0.047 -0.115** (0.78) (1.51) (1.59) (2.09) Flat land 0.026 0.043 0.043 -0.098 (0.53) (0.87) (0.75) (0.92) Gently sloped land 0.018 0.033 0.052 -0.146 (0.37) (0.67) (0.87) (1.35) Irrigated land 0.213*** 0.213*** 0.126** -0.150 (4.04) (4.02) (1.96) (1.13) Constant 4.373*** 4.659*** (81.85) (86.57) Observations 8562 8562 8562 8562 Number of households 606 606 606 606 R-squared 0.11 0.11

Table 5: Determinants of net land leased-in: Maximum likelihood estimates Simple friction model Ability included Variable Leased-out Leased-in Leased-out Leased-in Owned area (ha) 0.266*** -0.182*** 0.270*** -0.186*** (4.66) (3.46) (4.57) (3.45) Proportion of good soil quality -0.060 -0.008 -0.078 -0.010 (0.96) (0.14) (1.12) (0.17) Number of dependents -0.053*** 0.065*** -0.043** 0.064*** (2.64) (3.46) (2.02) (3.34) Number of adult male -0.081** 0.070** -0.088** 0.070** (2.56) (2.40) (2.55) (2.36) Number of adult female -0.001 -0.043 0.006 -0.046 (0.03) (1.25) (0.19) (1.29) Number of bulls and oxen -0.622*** 0.284*** -0.613*** 0.279*** (9.15) (4.25) (8.83) (4.21) Value of other livestock × 10-3 (Birr) -0.091*** 0.035** -0.082*** 0.035** (2.94) (1.99) (2.61) (1.99) Number of rooms of the household -0.026 0.073*** -0.039 0.079*** (0.90) (2.65) (1.21) (2.75) Age of household head (years) 0.011*** -0.011*** 0.012*** -0.011*** (6.00) (5.20) (5.86) (5.04) Female headed household 0.615*** -0.652*** 0.606*** -0.637*** (7.42) (5.86) (6.84) (5.55) Household head can read and write 0.027 0.001 0.022 -0.002 (0.47) (0.03) (0.35) (0.03) Agricultural ability -0.070 0.124** (1.31) (2.00)




Conclusion

• No evidence for Marshallian inefficiency;• Evidence of huge inefficiencies in terms of

resource adjustments due to the presence of transaction costs (large part of it is likely to be policy induced), but helps to transfer land to more productive farmers; and

• Policy implication: focus on identifying factors that prevent adjustment to desired level of cultivated land (e.g. tenure insecurity)

• Implications on off-farm employment

Model: optimal adjustment Adjustment model (Bliss and Stern 1982):

,, KLOfhKDCAhk (2)

- k is net land leased-in - DCA is desired cultivated area - K is total owned land area - O is oxen power - L is family labor - h is non-decreasing in (DCA – K) and h(0)=0

A first-order Taylor series expansion of equation (2) yields

,''' CKhLfhOfhk LO (3)

Simple friction model Ability included Threshold effect and constant parameters Leased-out Leased-in Leased-out Leased-in Woreda2 -0.269** 0.298*** -0.282** 0.304*** (2.52) (3.25) (2.54) (3.23) Woreda3 -0.272*** -0.190* -0.269** -0.135 (2.62) (1.94) (2.35) (1.28) Woreda4 -0.784*** 0.321*** -0.875*** 0.315** (6.07) (2.68) (6.18) (2.57) Woreda5 -0.982*** 0.238** -1.068*** 0.272** (7.34) (2.08) (7.33) (2.30) Woreda6 -0.462*** -0.074 -0.463*** -0.019 (4.00) (0.69) (3.70) (0.16) Time trend 0.098*** 0.134*** 0.096*** 0.138*** (3.43) (4.49) (3.04) (4.50) Constant -0.608*** 0.380*** -0.709*** 0.367** (3.53) (2.67) (3.75) (2.51) σ 1.011*** 1.045*** (9.50) (9.38) Log likelihood -4659.95 -4666.20 Wald chi2(23) 146.51*** 136.59*** Observations 4268 4119

Table 6: Wald tests of Equality of coefficients on opposite sides of the land lease market Hypothesis tested (using estimated coefficients given in the first two columns of table 5)

Wald statistic chi2(r)

Symmetry of all coefficients (except woreda dummies), r=11 33.87*** Symmetry of subset coefficients (resource endowment: labor, land, oxen, assets), r=8 33.42*** Owned area 4.05** Owned area=-1, r=2 246.55*** Good soil quality 0.77 Dependants 0.24 Male adult 0.08 Female adult 1.08 Oxen 22.25*** Other livestock 2.79* Rooms 1.91 Age 0.11 Female headship 0.11 Literacy 0.17 Equality of intercept terms

Woreda1a 0.91 Woreda2 0.81 Woreda3 9.07*** Woreda4 6.53*** Woreda5 16.94*** Woreda6 9.97***

Klaus Deininger, Daniel Ayalew, Tekie Alemu

Documents

land rental

piece of land

role of rental market

amhara implications

community amhara

amhara highlands

input markets labor

context of state