Journal of Agriculture and Rural Development in the Tropics and Subtropics
Vol. 117 No. 1 (2016) 21–34
urn:nbn:de:hebis:34-2015120349476 ISSN: 2363-6033 (online); 1612-9830 (print) – website: www.jarts.info
An analysis of household farm investment decisions
under varying land tenure arrangements in Ghana
Michael Ayamga , Richard W. N. Yeboah ∗, Sylvester Nsobire Ayambila
Faculty of Agribusiness and Communication Sciences, University for Development Studies, Tamale Ghana
Abstract
Land tenure insecurity is widely perceived as a disincentive for long-term land improvement investment hence the
objective of this paper is to evaluate how tenure (in)security associated with different land use arrangements in Ghana
influenced households’ plot level investment decisions and choices. The paper uses data from the Farmer-Based Or-
ganisations (FBO) survey. The FBO survey collected information from 2,928 households across three ecological zones
of Ghana usingmulti-staged cluster sampling. Probit and Tobit models tested the effects of land tenancy and ownership
arrangements on households’ investment behaviour while controlling other factors. It was found that marginal farm
size was inversely related to tenure insecurity while tenure insecurity correlate positively with value of farm land and
not farm size. Individual ownership and documentation of land significantly reduced the probability of households los-
ing uncultivated lands. Individual land ownership increased both the probability of investing and level of investments
made in land improvement and irrigation probably due to increasing importance households place on land ownership.
Two possible explanations for this finding are: First, that land markets and land relations have changed significantly
over the last two decades with increasing money transaction and fixed agreements propelled by population growth and
increasing value of land. Secondly, inclusion of irrigation investment as a long term investment in land raises the value
of household investment and the time period required to reap the returns on the investments. Households take land
ownership and duration of tenancy into consideration if the resource implications of land investments are relatively
huge and the time dimension for harvesting returns to investments is relatively long.
Keywords: Ghana, Land tenure, soil improvement, investment decision, probit
1 Introduction
Tenure security enables farmers make long-term in-
vestments to improve soils in the expectation that they
will hold their land rights long enough to reap the asso-
ciated benefits. Secure land rights provide incentives for
investing in soil conservation, land improvements and
other productivity-enhancing operations since farmers
are assured of reaping the stream of benefits associated
∗Corresponding author
Richard W. N. Yeboah
UDS, P. O. Box TL 1882, Tamale Ghana
Email: narteyrwn@hotmail.com
Phone: +233(0)277896282
with their investments (Platteau, 1995). In contrast, the
effect of lack of land tenure security is uncertainty in a
farmer’s mind about the value of improvements made to
the land. This uncertainty tends to increase as farming
becomes more commercialised and expected to have a
bearing on how farmers manage their land and thus in-
fluence agricultural productivity (Alemu, 1999; Sjaastad
& Bromley, 1997).
Low agricultural investment and productivity in
Ghana have among other things, been attributed to prob-
lems with security of land rights and gender disparity
with respect to access to and control of productive re-
sources including land (Agbosu, 2000; Tsikata, 2003).
Published online: January 27, 2016 Open access article licensed under a Creative Commons Attribution 4.0 International License CC BY
http://creativecommons.org/licenses/by/4.0
22 M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34
With public agricultural investment at nine percent of
GDP, and trailing the Comprehensive African Agricul-
tural Development Program (CAADP) - Maputo Decla-
ration’s target of 10 percent, it is worrying that Ghana’s
quest to modernise agriculture and maintain growth lev-
els consistent with middle income economies would be
severely compromised if the levels of private and public
investments in agriculture remain low. Though a mul-
tiplicity of factors may account for the low agricultural
investment in Ghana, the land tenure and administration
system has often been cited as a disincentive for land in-
vestment. Aryeetey & Udry (2009) reports that land and
its management in Ghana has often been poorly under-
stood and is generally perceived to induce tenure inse-
curity.
Land tenure and administration system in Ghana is a
combination of customary and statutory processes. Cus-
tomary land tenure systems are usually managed by tra-
ditional rulers, land earth priests, council of elders, and
family or lineage heads. Legislations and interventions
by colonial and early post-independence governments
led to the establishment of the pluralist legal system in
which English common law was grafted onto Ghanaian
communal societies, largely neglecting differences be-
tween the early 19th century capitalist economic struc-
tures and the egalitarian communal institutions of Ghana
(Agbosu et al., 2007). The resulting legal pluralistic
land administration has laid the foundations for con-
flicts and is widely perceived to induce tenure insecurity
which has hindered farm investments and agricultural
growth (Agbosu et al., 2007).
This paper examines smallholder farmers’ land in-
vestment decisions under varying land tenure arrange-
ments using plot level data from 2,928 households
across Ghana. The paper is motivated by the increas-
ing need for knowledge on how indigenous land tenure
arrangements influence the willingness of plot holders
to undertake long-term soil and farm improvement in-
vestments. Exploring how land relations in Ghana af-
fect farm households’ investment behaviour supports
the achievement of the dual goals of identifying entry
points for government to influence smallholder farm in-
vestment and also providing feedback for ongoing land
tenure and administration reform.
2 Land tenure and smallholder farm invest-
ment: the theory and empirical evidence
Resource allocation under different land use arrange-
ments has been a subject of debate in theory especially
on the subject of enforceability of land and labour con-
tracts (Otsuka et al., 1992). Marshall (1890) and Mill
(1848) conclude that share tenancy results in inefficient
resource allocation, arguing that since the share tenant
receives as marginal revenue only a fraction of the value
of his/her marginal product of labour, the tenant’s incen-
tive to supply labour or other inputs at the optimum level
is limited1. On the other hand, Cheung (1969) and John-
son (1950) have contended that, if effort is costlessly
enforceable, sharecropping arrangements can be as effi-
cient as owner-cultivated and fixed-rent tenancy. Ahmed
et al. (2002) reports that relative to land cultivated by
owners and fixed-fee rental land, alternative land tenure
systems such as sharecropped and gifted land prevailing
in Ethiopia were less efficiently cultivated and resulted
in considerable economic losses to farmers due to the
restrictions imposed by landowners on the latter. They
suggested that policies that facilitate individualised land
transactions would result in considerable gains to farm-
ers in particular and the national economy at large.
The growing list of studies examining the relationship
between land tenure security and long-term farm invest-
ment or productivity have done little to quell uncertainty
and controversy over the much hypothesised nexus es-
pecially in Africa. Most of the studies in Africa, includ-
ing those mentioned above have produced mixed results
with the greater majority failing to establish the hypoth-
esised strong links between tenure security, investments
and productive efficiency.
Considering the relationship between land title and
investment, most empirical studies have produced in-
conclusive results. In Ghana, Rwanda and Kenya for ex-
ample, Migot-Adholla et al. (1991) found that increas-
ingly individualised land rights do not appear to have
any effect on agricultural investment and yields. In ar-
eas of Kenya with land registration, no link was found
between land titling and long-term investments to im-
prove land (Barrows & Roth, 1989). In Zimbabwe, Har-
rison (1987) found little variation in the productive per-
formance between small holder farmers with no land ti-
tle and large scale commercial farmers with land titles.
Evidence linking individualised land rights with
higher productive efficiency is weak and inconsistent.
ThoughLaffont &Matoussi (1995) found significant ev-
idence of inefficiency under sharecrop arrangements in
a study in Tunisia, the study failed to provide evidence
1 The tenant receives a predetermined share of the output (e.g.
40%) and also pays or decides on variable input use. The exoge-
nously determined share of the output may under value his labour and
variable input contribution or over compensate the landlord for use of
the land.
M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34 23
of higher efficiency under individualised land holdings.
Ahmed et al. (2002) found significant inefficiency on
sharecropped land but not so on land under fixed-rent
contracts. Unlike in Africa, the evidence on the relation-
ship between titled land rights and productivity has been
more consistent in Asia and Latin America, where some
link is established between yields, farm investments and
tenure security Feder et al. (1987); Salas et al. (1970);
Shaban (1987); Villamizar (1984).
A cursory look at the approaches used in investigating
the tenure security-productivity hypotheses in Ghana re-
veals a certain degree of proclivity to notions of superi-
ority of individualised land rights and land titling as ul-
timate determinants of tenure security. As a result, the
research into the land tenure security-productivity hy-
potheses appears lopsided with emphasis on identifying
analytical and modelling deficiencies as opposed to in-
terrogating issues that border on conceptualisation and
operationalisation of land tenure security, the most sig-
nificant parameter of the hypothesis.
The common response to the failure to observe ex-
pected relationships between tenure security and pro-
ductivity is the attempt to argue that tenure security is
endogenous and that earlier studies lacked the econo-
metric rigour to adequately account for the perceived en-
dogeneity of tenure security (Besley, 1995; Hayes et al.,
1997; Twerefou et al., 2011). Many of the more recent
investigations of the tenure security-productivity hy-
potheses have therefore focused on resolving the issues
of endogeneity in tenure security mostly through multi-
stage econometric modelling (Besley, 1995; Hayes
et al., 1997; Twerefou et al., 2011). The findings of
these studies however have not been radically different
in terms of resolution of the ambiguity surrounding the
relationship between tenure security and expected im-
provements in investments.
Using the same data set as Migot-Adholla et al.
(1991), Besley (1995) assumed that land rights were en-
dogenous with farmer investment aimed at improving
their rights over land. He concluded that better land
rights facilitated investment in Wassa but not in An-
loga, a direct opposite of the findings made by Migot-
Adholla et al. (1991). Twerefou et al. (2011) in their
study of tenure security, investments and the environ-
ment in Ghana, set tenure security as endogenous. They
found that investments in farmlands in Ghana were low,
appeared not to enhance tenure security, and argued that
the reverse causation assumption of tenure security en-
hancing investment seemed non-existent. They further
concluded that tenure security appeared to be an incen-
tive for investment and had a positive and significant
effect on investment though the authors suggested the
results were not robust because the model did not con-
trol for endogeneity. Twerefou et al. (2011) are con-
tradicted by Dzanku (2007) who failed to establish an
overwhelming link between land rights and investment
in irrigation and soil improvements.
The farmer bases his investment decisions on his level
of tenure security and chooses between investments in
capital equipment, which is not lost in the event that
he/she loses his rights to land and long-term soil im-
provement and irrigation-related investments, which are
completely lost in an eviction (Feder et al., 1987). Sev-
eral studies on land tenure security and farm investment
and productivity, including Place & Hazell (1993) in
Ghana, Rwanda and Kenya, and Hayes et al. (1997) in
Gambia have applied the Feder et al. (1987) framework.
The model assumes that the household’s utility is in-
creasing in present value of future income stream (π),
and household characteristics and asset wealth (Hc):
maxIiU[E(πt);H
c
t ] Subject to
E(πt) =
T∑
t=1
δt(ptqtAtE[ωt] − cIi(e)Iit;Hct )
qt = y(st,Kt)
st = s0
1 − er(ϕt,
t∑
ω=1
Iiω)
 (1)
Where πt the value of future income stream (π t) at the
end of the household’s planning (T ), cI i is the unit cost
of conservation investment (Ii) discounted by δt. Hct de-
notes household asset wealth characteristics. The unit
cost of conservation investment is assumed to be de-
creasing in level of farmer experience (cI i(e) < 0). The
expected crop revenues are given by the product of crop
price (pt) and yield (qt), and land area (At), and dichoto-
mous expectation of land tenure in the period t(E[ω t]).
Yield in season t is assumed to be concavely increasing
with soil depth2 (q′(st) > 0) and also a function of other
conditioning factors (Kt) such as weather, pests and soil
fertility.
Unlike Place &Hazell (1993) and Hayes et al. (1997),
this study adds the dimension of investment in the devel-
opment of irrigation structures.
2 Soil depth is a proxy of good soil conditions. Deep and friable
soil is presumed good conditions for crop productivity and yields are
assumed to increase with improving conditions. Soil depth is defined
as depth in soil to which the roots of a plant can readily penetrate.
24 M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34
3 Methodology
3.1 Empirical model
Based on the theory and the large body of empirical
research undertaken on the subject of land tenure and
farm investment across a number of African countries
by Clay et al. (1998), Feder & Feeny (1993) and Hagos
& Holden (2006) the study sets the household as a util-
ity maximizing entity that chooses between short-term
complementary inputs use and long-term land invest-
ments based on the household’s rate of time preference
and specifies an estimable empirical model as:
Ii = f (tenure,wealtht−1,HcT−1, plot,market, crop, Zone)
(2)
where: Ii measures household conservation and irriga-
tion investment in plot i. The survey solicited house-
hold responses on the investments made in soil and wa-
ter conservation and irrigation expressed in Ghana Cedi
(GHS). The tenure variable represents factors that in-
fluence the farmer’s expectation of retaining tenure or
land rights such as whether the plot is owner operated,
rented, temporally transferred (loaned). The duration of
tenure is also included and is expected to improve farm-
ers perceived tenure security. The wealthT−1 variable
denotes household wealth and asset holdings including
relative farm size, livestock holdings, labour and other
resource endowments. HcT−1 represents household de-
mographic characteristics such as age and education of
household head. The variable plot represent farm char-
acteristics such as soil type, drainage, degree of frag-
mentation (ratio of total number of parcels to total farm
size), and access to irrigation. The variablemarket mea-
sures market access variables such as borrowing and ac-
cess to agricultural extension information. The variable
crop3 denotes the type of crop cultivated, either short
duration crops or annuals or perennials that require land
for several seasons. The zone variable controls for loca-
tion fixed effects such as distance to markets, population
density and rainfall.
The household soil conservation and irrigation invest-
ment decision making is assumed to happen at two lev-
els. The household first decides whether to invest or
not to invest and upon deciding to do the former, make
decision on the level of investment. Both decisions are
influenced by factors including those outlined in the em-
pirical model (equation 3.). The level of conservation is
3 Dummies for annual crops (including maize, millet, cassava,
cowpea, yams, among others); perennial cash crops (comprising co-
coa, oil palm and mango); non-perennial cash crops (pineapple, soy-
bean, melons)
given by
IL = X1β1 + ε1 (3)
where IL is the level of household conservation invest-
ment which depends on the vector of X1 explanatory
variables outlined in (2).
The decision to invest or not is given by
ID = X2Ω2 + ν2 (4)
where (X and ID) are observed, whereas I L is observed
only when ID = 1.
The model assumes that ε1andν2 are independent of
X implying that X is exogenous, and ν2∼N(0, 1). Given
such a model, if the error terms in Equations (3) and (4)
are related, they must first be estimated jointly given the
premise that the household chooses whether to invest
and then, having decided positively chooses the level of
conservation investment. This implies there could be
problems of selection bias hence requiring that the two
equations be estimated jointly. The estimation proce-
dure therefore involved testing for the presence of se-
lection bias using the Heckman selection model (Heck-
man, 1990), and examining the likelihood ratio test of
independence.
Sample selection bias was tested using a Heckman
two-step model (Deaton, 1997). Adjustment for the
standard errors in the level of investment model (equa-
tion 3) for heteroscedasticity using Powell’s Censored
Least Absolute Deviations (CLAD) estimator was done.
The CLAD estimator unlike the standard estimators of
the censored regression model is robust to heteroscedas-
ticity, consistent and asymptotically normal for a wide
class of error distributions (Arabmazar & Schmidt,
1981).
The Heckman test for sample selection bias tests the
null hypothesis of H0: ρε1ν2 = 0 and the alternative
hypothesis HA: ρε1ν2 
 0. The measure of correlation
between ε1 and ν2 is the correlation coefficient ρ. If the
study rejects the null hypothesis of ρε1ν2 = 0, then the
decision to invest equation (the sample selection equa-
tion) and the level of investment (outcome equation)
cannot be said to be independent and thus must be es-
timated jointly by the Heckman technique. The Wald’s
test of independence indicates that ρε1ν2 is not signifi-
cantly different from zero hence failure to reject the null
hypothesis H0: ρε1ν2 = 0 i.e. the models revealed no
significant selection bias. As indicated, the significance
of this result is that the sample selection (decision to
invest) and outcome equations (money value of invest-
ments) could be treated as two independent equations
and estimated separately.
M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34 25
The binomial probit and censored regression (Tobit)
models were used to estimate the decision to invest and
level of conservation and irrigation investments, respec-
tively. When the dependent variable is dichotomous (0,
1), the probit and the logit models are preferable but for
continuous dependent variables that are censored at or
below zero the Tobit model is preferable (Anley et al.,
2007). The Tobit model is used to allow for censoring of
households that made zero investment in land improve-
ment.
The standard Probit and Tobit models may be formu-
lated as:
y∗i = x
′
iβ + i
yi =
{
1 if y∗i = 1
0 otherwise
y∗i = x
′
iβ + i
yi =
{
y∗i if y
∗
i > 0
0 if y∗i ≤ 0
(5)
where i = 1, 2, . . . ,N, and i is assumed to be
NID (0, δ2) and independent of xi. This model is a cen-
sored regression model where observations may be cen-
sored from below.
3.2 Study area and data
The study used the Farmer Based Organisation (FBO)
survey data collected by the Institute of Statistical, So-
cial and Economic Research (ISSER) of the Univer-
sity of Ghana in 2008 (ISSER, 2008). The survey was
intended to facilitate the monitoring and evaluation of
the Millennium Challenge Compact signed between the
Government of Ghana and the Millennium Challenge
Corporation (MCC) of the United States of America.
The strength of the data used for this study lies in
the geographical spread of the sample. Land tenure sys-
tems in Ghana have been largely influenced by ethnic-
ity, population pressures and agriculture intensification,
with the land tenure systems in the South differing from
tenure systems prevalent in the Northern parts of Ghana.
Geographically, the study covered six regions; North-
ern, Central, Eastern, Volta, Ashanti and the Greater Ac-
cra Regions located in three ecological zones.
A sample of 2,928 farm households drawn from 23
districts in six regions within the three distinct agro-
ecological zones of Ghana, namely, the Northern Agri-
culture Zone (Northern Region of Ghana), the Afram
Basin (Ashanti and Eastern Regions of Ghana), and the
Southern Horticultural Belt (South-East Coastal Plains
of Volta Region) was used (Figure 1).
The FBO survey collected information on the over-
all living circumstances and farming activities of mem-
bers of FBOs and their respective households. In-depth
household data was collected using two sets of ques-
tionnaires; a household questionnaire and a community
questionnaire: The survey collected information on a
wide range of household attributes including the demo-
graphic, education and health characteristics; migration;
household transfers; information seeking behaviour of
households; household assets and participation in fi-
nancial markets (borrowing, savings and lending be-
haviour); household agriculture activities including land
ownership and transactions and agriculture processing
and, non-farm enterprises of households. Information
was also collected on the location of households, com-
munity facilities and farm sizes using geographic posi-
tion system units (GPS). The community questionnaire
was essentially a market price survey.
The data was collected using multi-stage probabil-
ity sampling, clustered, and stratified with probability
proportional to the size of target population to sample
households. The approach used a two-stage sampling
technique. During the first stage, 27 enumeration ar-
eas (EAs) were selected using systematic sampling with
probability proportional to size method (PPS) for each
district. The EAs were the same as those used during
the 2000 Population and Housing Census by the Ghana
Statistical Service (GSS, 2000). The selected enumera-
tion areas were listed fully to know the total number of
households that served as sampling frame from which
an appropriate sample size was selected.
4 Results
4.1 Descriptive statistics
Table 1 presents description and summary statistics of
variables used in equations 3 & 4.
The standard errors were adjusted to correct for clus-
tering effects. Only 20% of sampled households are
female-headed. Although about 70% of household
heads reported to have attended basic school, only about
20% are able to read and write a simple sentence in En-
glish.
Only 14% of sampled households perceive their
tenure status as insecure. The distribution appears to be
even across observations in terms of the modes of land
acquisition and forms of land rights. Though 30% of
sampled households have land titles, only 20% are re-
ported to exercise complete rights over their land. Close
to 20% of households purchased their land outright
while 30% received land as gifts. Only one percent
26 M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34
Fig. 1: The study area in regions and districts
of land is held in sharecropping arrangements. About
80% of sampled households indicated they could vacate
their lands and still maintain ownership but the period of
time for which they could vacate the plots averaged at 6
months of the production season.
About a third of sampled households made land im-
provements and irrigation related investments. House-
holds invested an average of about GHS 2004 per acre
4 GHS 200 is equivalent to USD95 at the then prevailing exchange
rate of 2.1 Ghana Cedis (GHS)
on land improvement and irrigation. Compared with
average value of farm produce and average non-farm
income, households invest about 29% of farm rev-
enue (excluding livestock) or 60% of non-farm revenue.
Sampled households have owned and used land for over
9 years on the average with over 10% of the farms
enclosing pre-existing physical structures such as farm
houses, fences, and storage yards among others.
M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34 27
Table 1: Descriptive statistics
Variable and Description Estimate Standard Error
Household Characteristics
Sex (dummy 1= male 0= female) 0.8 –
Age (Number of years) 47.1 12.7
Dependency ratio (ages below 15 and above 64/ ages 15 to 64) 1.1 0.9
Basic education (Dummy 1=attended basic school 0=otherwise) 0.7 –
Literacy (dummy, 1= can read and write; 0=otherwise) 0.2 –
Assets and Wealth variables
Livestock Holding (in Tropical Livestock Units) 2.5 7.9
Land Holding (acre) 12.2 11.7
Value of output (GHS value of output per acre) 695.1 3,888.7
Family labour (man hours per season) 270.8 573.5
Non-farm income per household (GHS) 336.5 2,001.1
Land Tenure and Security Variables
Ownership with deed (Dummy, 1= registered land with deed; 0= otherwise) 0.3 –
Gift land (Dummy, 1= received land as gift; 0= otherwise) 0.3 –
Sharecropped land (Dummy, 1= land acquired under sharecropping; 0= otherwise) 0.1 –
Assurance of tenure security (number of years) 0.5 0.7
Years of land ownership (Number of years) 9.1 6.7
Probability of investment in land (1= if household made investment and 0= otherwise) 0.3 –
Intensity of investment land (in GHS) 202.5 846.7
Number of physical structures (farm houses, storage houses) 0.1 0.5
Crop and Location Variables
South (Dummy, 1= southern horticultural belt; 0= otherwise) 0.2 –
Northern agricultural zone (Dummy, 1= northern agricultural zone; 0 =otherwise) 0.4 –
Affram Basin (Dummy, 1= Affram Basin; Otherwise=0) 0.3 –
Cash crops (Dummy, 1= cash crops; 0 = otherwise) 0.1 –
Plot Characteristics
Index of land fragmentation (Number of plots/ acre) 1.6 1.8
Access to irrigation (Dummy, 1= irrigated, 0= land rain-fed) 0.1 0.3
Soil water retention (Number of hours it takes farms to drain) 21.4 70.1
Ratio of zonal to household farm5 0.9 0.0
Market Access and Participation
Receive extension visits (1= farm received extension visit 0=otherwise) 0.5
Distance to major market (Kilometers) 2.3 16.5
Adoption of technologies (Dummy, 1= yes; 0= no) 0.2
Source: Authors’ computations fromMIDA FBO Survey, 2008
5 Refers to the farm size per capita of the household divided by the average farm size per capita in the zone.
28 M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34
Larger dispersions occur in farm revenue, non-farm
income and household labour use. Since farm sizes and
scales of production vary it is expected that larger dis-
persions would occur in farm revenue. Larger families
tend to have more access to family labour. The large
standard deviation of family labour use is in tandem
with the wide variation in family sizes.
The result of the probit model of the decision to invest
in soil and water conservation and irrigation is presented
in Table 2. Endogenous variable such as the number of
physical structures established on the land by the house-
hold were excluded from the model because of lack of
appropriate instruments that would enable us to predict
the variable.
Land tenancy arrangements such as renting and share-
cropping were found to have significant influence on
households’ decision to invest and had the expected
negative signs. While land ownership (without deed)
did not exert a significant influence on the probabil-
ity of households deciding to invest, land documenta-
tion (ownership with deed) was found to exert a posi-
tive effect on the decision to invest and was significant
at the 10 percent level. Households were also found to
be less willing to invest in farm lands they received as
gifts (gifted land) and which they had not formally doc-
umented as well as on sharecropped plots. The assur-
ance of tenure security exerted a significant and positive
effect on the probability of investing in soil and water
conservation and irrigation.
The distance to major markets was found to exert a
negative effect on the decision to invest and was signif-
icant at 1%. Access to irrigation as well as the ratio of
household to zonal land size also had the expected sig-
nificant positive effects on the decision of households to
invest.
The effect of household resource endowment on the
decision to invest was found to be mixed. Household
livestock expressed in Tropical Livestock Units (TLU)6
had no significant influence on this decision. On the
contrary, labour availability (family labour) and value
of output per area (or crop income) did significantly in-
crease the probability of households deciding to invest
in land conservation and irrigation. Households with
high labour would be in the position to implement com-
plementary farm activities required to optimize the ben-
efits from land improvement related investment.
6 Tropical Livestock Units are livestock numbers converted to a
common unit. The Conversion factors are: cattle = 0.7, sheep = 0.1,
goats = 0.1, pigs = 0.2, chicken = 0.01. Factors taken from Har-
vestChoice (2011)
Location of households and crop characteristics such
as cultivation of cash crops were also found to have sig-
nificant effects on the decision to invest. Compared with
the southern horticultural belt (used as the reference cat-
egory), households located in the Affram Basin and the
northern agricultural zone had higher propensities to in-
vest contrary to a priori expectations that the relatively
high level of commercialisation of agricultural products
and the higher population density in the South would
make these households more willing to invest in land
improvements.
Cultivation of vegetables and spices was found to pos-
itively influence the decision to invest in soil and water
conservation and irrigation. This was expected since the
production of vegetables and spices in many instances
is done in the dry or minor seasons and would usu-
ally involve significant investment in irrigation. The
cultivation of perennial cash crops and non-perennial
cash crops had the expected positive and negative co-
efficient signs, respectively, but only the coefficient for
non-perennial cash crops was statistically significant.
Land fragmentation7 had a positive effect on the prob-
ability of investing and was significant at 1%. The ratio
of zonal to household farm size was significant at 10%
and had a positive coefficient. Soil depth was significant
at 10% and positively correlated with the probability
of investing. Drainage (an indicator of water retention)
had a negative coefficient but was not significant. Ac-
cess to agricultural extension as well as the adoption of
improved technology did not significantly affect house-
holds’ probability to invest.
The level of soil conservation and irrigation invest-
ment was measured by the amount (GHS) spent by the
household to improve and conserve soil water or fer-
tility and to facilitate irrigation. Households’ level of
investment in soil conservation and irrigation was hy-
pothesised to be influenced by a number of household
socioeconomic and farm characteristics as well as the
provenance of land (Table 3). Compared to the results of
the probability of investment, differences in the amounts
spent by households in soil conservation and irrigation
were to a large extent explained by differences in types
of land use and ownership arrangements as well as the
duration of land ownership.
7 Land fragmentation ratio measures the ratio of total household
farm size (in acres) to the number of parcels or plots. It is an indica-
tor of whether household land holdings are continuous large tracts or
small scattered plots.
M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34 29
Table 2: Maximum Likelihood Estimates of the Determinants of the Decision to Invest
Variable Coefficient Robust Std. Errors Z
Household Characteristics
Sex of household head 0.4065 0.0971 4.19***
Age of household head –0.0240 0.0091 –2.62**
Age square of household head 0.1754 0.0862 2.03*
Dependency ratio 0.0101 0.0309 0.33
Basic education 0.2133 0.0382 5.58***
No education 0.2884 0.0856 3.37**
Household Assets and Wealth variables
Livestock Holding (in TLU) –0.0343 0.0386 –0.89
Household land holding –0.0156 0.0075 –2.08*
Value of output per area 0.4721 0.0886 5.33***
Family labour 0.0733 0.0342 2.14*
Land Tenure and Security Variables
Gifted land –0.3087 0.06015 –5.13***
Sharecropped land –0.2323 0.0676 –3.44**
Rented land –0.1173 0.0654 –1.79*
Ownership with deed 0.1633 0.0676 2.42*
Ownership without deed –0.0136 0.0944 –0.14
Assurance of tenure security 0.1389 0.0554 2.50*
Years of land ownership –4.3728 3.9281 –1.11
Crop and Location Variables
Affram Basin 0.4404 0.0796 5.53***
Northern agricultural zone 0.7883 0.1149 6.86***
Distance to major market –0.0140 0.0038 –3.65***
Perennial cash crops 0.0198 0.1002 0.20
Non-perennial cash crops –0.1275 0.0625 –2.04*
Vegetables and spices 0.2350 0.0624 3.77***
Plot Characteristics
Ratio of zonal to household farm size 0.1805 0.0877 2.06*
Index of land fragmentation 0.1829 0.0510 3.58***
Access to irrigation 0.9210 0.0710 12.96***
Drainage –0.1371 0.3398 –0.40
Soil depth 25.8536 14.5829 1.77*
Access to Innovation and Technology
Receive extension visits –0.0130 0.0539 –0.24
Adoption of improved technology 0.0771 0.0498 1.55
Constant –1.5085 0.2811 –5.37***
Log likelihood –1866.34
Wald χ2 (30) 365.43
Prob > χ2 0.0000
Pseudo R2 0.1009
Dependent Variable is the decision to invest into soil conservation and irrigation (1 if the household
invested and 0 if household did not invest)
Source: Authors’ computations fromMIDA FBO Survey, 2008
*, **, and *** are levels of significance at 10%, 5%, and 1%, respectively.
30 M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34
Table 3: Determinants of Soil and Water Conservation and Irrigation Investment expenditure
Variable Coefficient Robust Std. Err. t-statistic
Household Characteristics
Sex of household head –102.1851 36.4605 –2.80**
Age of household head 11.6229 6.0322 1.93*
Age of household head square –146.8364 58.7225 –2.50*
Household size 28.5339 3.9584 7.21***
Dependency ratio –90.2414 17.4334 –5.18***
Attained basic education –32.1379 19.5568 –1.64
No formal education –121.8411 40.1405 –3.04**
Household Assets and Wealth variables
Livestock holding (in TLU) –23.8377 21.5921 –1.10
Land holding –7.8167 4.12557 –1.89*
Non-farm income (per capita) 17.5791 10.2533 1.71*
Value of output per area 0.1187 .018828 6.31***
Family labour 19.6628 21.6363 0.91
Land Tenure and Security Variables
Family land –39.4991 44.0785 –0.90
Gifted land –47.0821 45.5502 –1.03
Sharecropped land 6.1550 39.7833 0.15
Rented land 93.5805 43.5905 2.15*
Ownership with deed 150.3289 44.5321 3.38**
Ownership without deed 67.4012 37.4213 1.80*
Duration of land contract 8 17.4348 4.15372 4.20***
Years owning land 9 –7.1505 4.26031 –1.68*
Duration of tenure security 10 281.6785 26.7354 10.54***
Crop and Location 11 Variables
Affram Basin 138.6426 39.2577 3.53***
Northern agricultural zone 326.5296 63.9384 5.11***
Perennial cash crops 61.9845 58.6894 1.06
Non-perennial cash crops –83.3227 29.2898 –2.84**
Vegetables and spices 39.7842 34.7841 1.14
Distance to major market –1.9830 1.54175 –1.29
Plot Characteristics
Ratio of zonal to household farm size 51.8104 50.2338 1.03
Index of land fragmentation –36.7666 16.1320 –2.28*
Access to irrigation 258.0953 33.1152 7.79***
Drainage 213.7851 108.114 1.98*
Market Access and Participation
Receive extension visits 88.5725 27.2893 3.25**
Adopted improved technology 35.7479 24.5838 1.45
Constant –1062.186 206.2269 –5.15***
Log pseudo likelihood –4034.3159
F(38,1352) 11.30
Prob > F 0.0000
Pseudo R2 0.0596
Left-censored observations at INVSWCON_IRR ≤1 893
Uncensored observations 459
Observations 1352
Dependent variable: GHS (₡) invested in soil and water conservation and irrigation.
Source: Authors’ computations fromMIDA FBO Survey, 2008
*, **, and *** are levels of significance at 10%, 5%, and 1%, respectively.
M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34 31
Land ownership (both with deed and without deed)
had significant positive effects on households’ invest-
ment expenditure. Consistent with a priori expectations,
family owned, rented, gifted and sharecropped lands had
negative coefficient but were however not statistically
significant.
The signs of household wealth variables were mixed
in terms of their consistency with a priori expectations.
Livestock holding (in TLU) did not have the expected
positive sign even though the coefficient was not sig-
nificant. Non-farm income and value of output had the
expected signs and was significant at 10% and 1% re-
spectively. Location dummies which were included to
capture site specific effects in investment indicated that
autonomous investment in soil conservation and irriga-
tion was relatively higher in the Affram basin and the
northern agricultural zone. The dummies for these two
locations were significant at 1%.
With respect to plot characteristics, the degree of land
fragmentation and drainage were found to be significant
in explaining the amounts households invested in soil
and water conservation and irrigation. Households in-
vested less on highly fragmented plots and more on irri-
gable plots and well drained plots. With regards to mar-
ket and information access, farmers who received ex-
tension visits were found to invest higher amounts than
those who did not. Adoption of improved technology
had the expected positive sign but did not significantly
influence household investment expenditure.
Some household and demographic factors seemed to
significantly influence the levels of soil conservation and
irrigation investments. Both age and age-squared of
household head had significant effects on the amounts
invested exerting positive effects on the amounts house-
holds invested in soil conservation and irrigation.
Household size was found to have a significant pos-
itive effect on the amounts invested and was significant
at 1%. Although not significant, formal education was
inversely related to the amounts households invested in
soil conservation and irrigation but the formal educa-
tion variable was not significant. As expected however,
8 Duration of land contract refer to the number of years a land
owner has agreed to allow a tenant use land in contract (formal or
informal) arrangement.
9 Refers to the number of years the farmer has owned the land.
Households who have owned land for longer periods are perceived
more secure.
10 Duration of tenure security is the length of time that a given right
is legally valid. The economic dimension requires, in addition, that
the time horizon be sufficiently long to enable the holder to recoup
(see Roth & Haase, 1998)
11 Southern Horticultural belt is used as reference category
not having formal education significantly reduced the
amounts households invested in soil and irrigation.
5 Discussion
Land documentation and duration of tenure security
positively influenced households’ decision to invest and
is an indication that land titling programmes and poli-
cies can facilitate households’ farm investment by en-
hancing tenure security. This finding is an indication
that formalisation of land rights though not a compre-
hensive indicator of land tenure security, contributes sig-
nificantly to enhance land holders’ perception of their
tenure security and their decisions to invest in lands.
Household land relative to zonal farm size was in-
cluded to test for farm size effects. Contrary to the find-
ing by Holden & Yohannes (2002), the study found that
marginal farm size was inversely related with tenure in-
security. Tenure insecurity was rather found to correlate
positively with value of farm land and not farm size as
Alemu (1999) posited. The reason for this observation
is that owning large tracts of land may not be a source
of insecurity if there are no competing claims. However,
as land values rise, interests in land both as a factor of
production and commodity increases. Local elites who
probably were not interested in land re-entered the mar-
ket and may appropriate land belonging to less powerful
holders.
Contrary to expectations that households that owned
more livestock would invest more in land because they
were regarded as wealthy (Holden & Yohannes, 2002),
the reverse held. One possible explanation is the use
of livestock as security against crop failure. Households
may not sell their livestock to invest in land conservation
and irrigation but may do so in the event of crop failure.
The expectation on the relationship between location
dummies was that farmers in the southern horticultural
zone would invest more because land was scarce and
there was the need to maximise output. This expecta-
tion was based on the Boserup (1965) population pres-
sure hypothesis which posits that high population den-
sity (land scarcity) would stimulate investment and use
of improved technology. It was expected that the high
population density of the southern horticultural belt will
lead to higher propensities to invest. The absence of
this population pressure effect could however be due
to abundant non-farm employment opportunities in the
South which significantly increases the opportunity cost
of farm labour and other resources and also offer oppor-
tunities that do not require ownership of land.
32 M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34
The study expected that highly fragmented land (mea-
sured by the land size in acres per number of plots)
would discourage investment as the prospects of com-
mercial scale production would be limited. The findings
however suggest that households’ decision to invest was
positively correlated with land fragmentation. This is
contrary to a priori expectation that highly fragmented
plots, usually common with communal or family held
land, would have a negative effect on the probability of
investing. One possible explanation for this observation
is that households may be using land fragmentation as
some sort of risk spreading activity. Further, land own-
ership overrides fragmentation in investment decision.
The results on the relationship between the monetary
value of investments and tenure security variables were
mostly consistent with a priori expectations. Farmers
who owned land investedmore relative to those who had
acquired land through tenancy. The negative relation-
ship between the amounts invested and the cultivation
of family owned, rented, gifted and sharecropped lands
could be indicative of some degree of Marshallian in-
efficiency (Marshall, 1890) transmitting to households’
willingness to invest and levels of investment expendi-
ture on long-term soil improvement and irrigation.
In relation to the socio-demographic factors included
in the models as control variables, the results showed
that age and age-squared exerted significant positive ef-
fects on the amounts invested. The expectation was
that age and age-squared would have positive and neg-
ative effects respectively on the amounts invested. That
would have implied younger farmers invested more in
soil and water conservation and in irrigation because
they constituted the most productive group and most
likely to be engaged in farming for commercial reasons.
The results however suggest that investment did not vary
much along age lines.
Household size was positively related to the amounts
invested. Large households usually cultivate larger
farms and hence are more likely to invest more. House-
holds with lower number of dependents as opposed to
workers were expected to invest more because relative to
households with higher dependency ratios, they would
have surplus production, and the proceeds could be rein-
vested in farming. The negative sign of the dependency
ratio variable is thus consistent with this expectation.
Higher dependency ratio reduces the amount of invest-
ment in soil conservation and irrigation.
6 Conclusion
This paper explored the relationship between house-
holds’ perception of their tenure security and how these
perceptions influenced plot level farm investment deci-
sions in soil conservation and irrigation. The paper mea-
sured land tenure security both in terms of right entitle-
ments and the duration within which holders perceived
the rights to be secure. The household was modelled as
a utility maximizing entity that choose between short-
term complementary inputs use and long-term land in-
vestments based on the household’s evaluation of its
land tenure security.
Findings of the paper support the long-held view that
better land rights facilitate long-term farmland improve-
ment investments. Tenure security, often posited to in-
fluence household farm investment decisions appears to
hold in the case of Ghana. Individual ownership of land
increased the probability of investing and the level of
investments made in land improvement and irrigation.
This evidence probably points to the increasing impor-
tance households place on land ownership and also un-
derscores the importance of tenure security in facilitat-
ing land titling programmes. Several studies, especially
those conducted in the middle to late twentieth cen-
tury, failed to plausibly establish strong links between
land ownership and investment as this study finds. Two
possible explanations for this particular finding of the
study are: First, there is the possibility that land mar-
kets and land relations have changed significantly over
the last two decades with increasing money transaction
and fixed agreements propelled by population growth
and increasing value of land. Second, the inclusion of
irrigation investment as a long term investment in land
raises the value of household investment and also the
time period required to reap the returns on the invest-
ments.
The finding of this paper seems to reinforce Boserup
(1965) and Ault & Rutman (1979) predictions that land
tenure in Africa evolved in tandem with principles of
economic efficiency and that households would acquire
full rights (i.e. take full rights into consideration) when
the need for such rights arose, propelled either by pop-
ulation growth and land scarcity or by emerging market
opportunities. It could be deduced that households take
land ownership and tenancy arrangements into consider-
ation if the resource implications of land investments are
relatively huge and the time dimensions for harvesting
returns to investments is relatively short.
It is recommended that for productivity to increase
due to intensification, land owners should allow for
long-term lease to provide the needed security that will
M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34 33
encourage huge investment into development of land for
long-term benefits.
References
Agbosu, L., Awumbila, M., Dowuona-Hammond, C. &
Tsikata, D. (2007). Customary and Statutory Land
Tenure and Land Policy in Ghana. Technical Publi-
cation, No. 70, Institute of Statistical Social and Eco-
nomic Research (ISSER), University of Ghana, Ac-
cra.
Agbosu, L. K. (2000). Ghana Land Law: Contradic-
tion between Anglo-American and Customary Con-
ceptions of Tenure Practices. Working Paper, No.
33, Land Tenure Center, An Institute for Research
and Education on Social Structure, Rural Institu-
tion, Resource Use and Development, University of
Wisconsin-Madison, USA. URL http://digital.
library.wisc.edu/1793/21905.
Ahmed, M. M., Gebremedhin, B., Benin, S. & Ehui, S.
(2002). Measurement and sources of technical effi-
ciency of land tenure contracts in Ethiopia. Environ-
ment and Development Economics, 7, 507—527.
Alemu, T. (1999). Land Tenure and Soil Conservation:
Evidence from Ethiopia. Ekonomiska Studier, Gote-
borgs Universitet.
Anley, Y., Bogale, A. & Haile-Gabriel, A. (2007).
Adoption decision and use intensity of soil and wa-
ter conservation measures by smallholder subsistence
farmers in Dedo district, Western Ethiopia. Land
Degradation& Development, 18 (3), 289—302.
Arabmazar, A. & Schmidt, P. (1981). Further Evidence
on the Robustness of the Tobit Estimator to Het-
eroskedasticity. Journal of Econometrics, 17, 253–
258.
Aryeetey, E. & Udry, C. (2009). Towards Efficient and
Equitable Land Tenure Reforms in Ghana-ISSER.
November 2009.
Ault, D. E. & Rutman, G. L. (1979). The Develop-
ment of Individual Rights to Property in Tribal Africa.
Journal of Law & Economics, 22 (1), 163–182.
Barrows, R. & Roth, M. (1989). Land Tenure and In-
vestment in African Agriculture: Theory and Evi-
dence. Land Tenure Center, LTC Paper No. 136, Uni-
versity of Wisconsin-Madison, USA.
Besley, T. (1995). Property Rights and Investment In-
centives: Theory and Evidence from Ghana. The
Journal of Political Economy, 103 (5), 903–937.
Boserup, E. (1965). Conditions of Agricultural Growth:
The Economics of Agrarian Change Under Popula-
tion Pressure. New York, Aldine Publishing Co.
Cheung, S. N. S. (1969). The Theory of Share Tenancy.
Chicago, University of Chicago Press.
Clay, D., Reardon, T. & Kanganiemi, J. (1998). Sus-
tainable intensification in the highland tropics: Rwan-
dan farmers’ in land conservation and soil fertil-
ity. Economic Development and Cultural Change,
46 (2), 351—377.
Deaton, A. (1997). The Analysis of Household Surveys.
A Microeconometric Approach to Development Pol-
icy. Baltimore, The Johns Hopkins University Press
(for the World Bank).
Dzanku, F. M. (2007). Land Rights, Sustainable Nat-
ural Resource Use and Agricultural Productivity in
Ghana. Institute of Statistical, Social and Economic
Research (ISSER) University of Ghana.
Feder, G. & Feeny, D. (1993). The theory of land tenure
and property rights. In: Hoff, K., Braverman, A. &
Stiglitz, J. E. (eds.), The Economics of Rural Orga-
nization: Theory, Practice, and Policy. pp. 240–258,
Oxford University Press, Oxford, UK.
Feder, G., Onchan, T., Chlamwong, Y. & Hongladorom,
C. (1987). Land Policies and Farm Productivity in
Thailand. John Hopkins University Press, Baltimore
MD, 1988.
GSS (2000). Ghana – Population and Housing Census
2000. Ghana Statistical Service (GSS).
Hagos, F. & Holden, S. (2006). Tenure security, re-
source poverty, public programs, and household plot-
level conservation investments in the highlands of
northern Ethiopia. Agricultural Economics, 34, 183–
196.
Harrison, P. (1987). The greening of Africa: break-
ing through in the battle for land and food. Paladin
Grafton Books, London.
HarvestChoice (2011). Total livestock population
(TLU) (2005). International Food Policy Re-
search Institute, Washington, DC. and and Univer-
sity of Minnesota, St. Paul, MN. URL http://
harvestchoice.org/node/4788.
Hayes, J., M. Roth, M. & Zepeda, L. (1997). Tenure Se-
curity, Investment and Productivity in Gambian Agri-
culture: A Generalized Probit Analysis. American
Journal of Agricultural Economics, 79 (2), 369–382.
34 M. Ayamga et al. / J. Agr. Rural Develop. Trop. Subtrop. 117 - 1 (2016) 21–34
Heckman, J. (1990). Varieties of Selection Bias. Amer-
ican Economic Review, 80 (2), 313–318.
Holden, S. & Yohannes, H. (2002). Land Redistribu-
tion, Tenure Insecurity, and Intensity of Production:
A Study of Farm Households in Southern Ethiopia.
Land Economics, 78 (4), 573–590.
ISSER (2008). GLSS5+ and FBO Surveys and Impact
Evaluation of the MiDA Program. Institute of Statis-
tical, Social and Economic Research (ISSER).
Johnson, D. G. (1950). Resource allocation under share
contracts. Journal of Political Economy, 58, 111—
123.
Laffont, J. J. & Matoussi, M. S. (1995). Moral hazard,
Financial Constraints and Sharecropping in El Oulja.
Review of Economic Studies, 62, 381–399.
Marshall, A. (1890). Principles of Economics. Macmil-
lan, London.
Migot-Adholla, S. E., Hazell, P. B. & Place, F. (1991).
Indigenous Land Rights Systems in Sub-Saharan
Africa: A Constraint on Productivity? World Bank
Economic Review, 5, 155–175.
Mill, J. S. (1848). Principles of Political Economy.
Longmans, London.
Otsuka, K., Chuma, H. & Hayami, Y. (1992). Land and
labor contracts in agrarian economies: theories and
facts. Journal of Economic Literature, 30 (4), 1965–
2018.
Place, F. & Hazell, P. (1993). Productivity Effects
of Indigenous Land Tenure Systems in Sub Saharan
Africa. American Journal of Agricultural Economics,
75, 10—19.
Platteau, J.-P. (1995). Reforming Land Rights in Sub-
Saharan Africa: Issues of Efficiency and Equity. UN-
RISD Discussion Paper No. 60. Geneva.
Roth, M. & Haase, D. (1998). Land Tenure Secu-
rity and Agricultural Performance in Southern Africa.
BASIS-CRSP Program, Land Tenure Center, Madi-
son.
Salas, O. A., Knight, F. & Sáenz, C. (1970). Land Ti-
tling in Costa Rica: A Legal and Economic Survey.
USAID and University of Costa Rica Law School.
Shaban, R. A. (1987). Testing between competing mod-
els of share cropping. Journal of Political Economy,
95 (5), 893–920.
Sjaastad, E. & Bromley, W. D. (1997). Indigenous Land
Rights in Sub-Saharan Africa: Appropriation, Secu-
rity and Investment Demand. World Development,
25 (4), 549–562.
Tsikata, D. (2003). Securing Women’s Interests within
Land Tenure Reforms: Recent Debates in Tanzania.
Journal of Agrarian Change, 3 (1 and 2), 149—183.
Twerefou, D. K., Osei-Assibey, E. & Agyire-Tettey, F.
(2011). Land tenure security, investments and the en-
vironment in Ghana. Journal of Development and
Agricultural Economics, 3 (6), 261–273.
Villamizar, F. (1984). IBD Financing in Land Adminis-
tration Programs. Paper presented at the International
Workshop on Land Tenure Administration, Salvador,
Brazil, August 20–24.