Effect of land tenure security on agricultural productivity among small scale farmers in Kenya: a conditional mixed processes analysis

Abstract

Security of land tenure among rural households is an integral part in achieving high agricultural productivity. However, there is limited information on the nexus between land tenure security, credit access and agricultural productivity. This study investigates the effects of land tenure security on maize productivity through credit access. Using pre-tested semi-structured questionnaires, cross-sectional primary data were collected from randomly selected 366 small-scale maize farmers in Trans-Mara East and Trans-Mara West sub counties of Narok County, Kenya. Descriptive statistics and conditional mixed process (CMP) model were used to analyse the data. Results show that about 46.99% of the interviewed households had land title deeds. Household heads with land title deeds were more educated, had larger land and farm sizes, had more productive parcels of land and obtained higher gross margins as compared to those without land title deeds. Moreover, the majority of household heads with secure land tenure were married, had community leadership position and acquired land through purchase. The CMP results indicate that, land tenure security was significantly influenced by sex, marital status, education level of the household head as well as presence of land disputes, soil fertility, and purchase of land. The study confirmed the relationship between land tenure security, credit access and maize productivity. Land tenure security significantly and positively influenced credit access, while farmers with access to credit increased maize productivity by 2001.902 kg/ha. The findings call for efforts by stakeholders to enact policies that could promote cost-friendly access to secure land tenure through the acquisition of land title deeds. Additionally, there is need for the government to promote the development and strengthening of both land and credit markets that would stimulate agricultural productivity.

Keywords:

• Endogeneity
• land registration
• land rights
• maize productivity
• title deed

STATEMENT OF PUBLIC INTEREST

Land tenure security is critical in the development of any nation. Improvement in land tenure security through possession of a title deed could improve credit access for small-scale rural farmers. The credit accessed could be used to finance agricultural investments hence increase productivity. However, there is inadequate literature, especially in developing countries, on the relationship between land tenure security, access to credit and agricultural productivity. In order to fill this knowledge gap, this study examines the effect of land tenure security on agricultural productivity through credit access in Kenya. The findings confirm the existence of a relationship between land tenure security, access to credit and agricultural productivity. Land tenure security has a positive effect on access to credit, underscoring the vital role of title deed to be used as a collateral. Empirical results also show that, credit access increases maize productivity.

1. Introduction

The United Nations Sustainable Development Goals (SGDs) (1 and 2) of 2015 envision that, by 2030, the world will be free of hunger and any form of food insecurities and malnutrition (United Nations, 2018). However, 6 years since their adoption by member states, approximately 768 million people globally are still food insecure, with more than three-quarters coming from Africa, Asia, and Latin America (FAO, IFAD, UNICEF, WFP and WHO, 2021). This can be attributed to the mismatch between the focus of agricultural reforms and the intended recipients who are the rural dwellers. Agricultural sector growth in sub-Saharan Africa (SSA) is estimated to be eleven times more effective in reducing poverty than growth in other sectors, hence underscoring the critical role of the sector (Nagarajan et al., 2019). In Kenya, the agricultural sector contributes about 26% of the Gross Domestic Product (GDP) and it employs more than 40% of the total population and more than 70% of the rural population (FAO, 2022). The sector is envisioned to increase the country’s annual growth rate by 10% by 2030 (Government of Kenya, 2021).

Maize farming sub-sector plays a vital role in the overall sector contribution. Maize (Zea mays) is one of the major cereal crops in the world, ranked third after rice and wheat (Mekureyaw, 2017). It is the most widely cultivated cereal grain in Africa and is considered a staple food (Nagarajan et al., 2019). In Kenya, it is considered as both a food security and the main staple food crop. It accounts for approximately 40% of the cultivated area with an estimated annual average production of 3 million MT and a national average productivity of 2 MT per hectare (Government of Kenya, 2021). Food and Agriculture Organization of the United Nations (2022) suggests that about 75% of the maize production in Kenya is from small-scale farmers. However, approximately 20% of the produce is sold to the market, while the rest is consumed at the household level hence underpinning the importance of the crop. Low maize productivity remains a challenge that curtails growth of the sector and the GDP in general (Government of Kenya, 2021). The increase in the Kenyan population currently standing at about 47.6 million and the corresponding limited land area of approximately 582.650 km² are some of the reasons for the increased pressure on productive resources such as agricultural land. This makes it a potential source of land tenure insecurity in Kenya (Government of Kenya, 2021). Land tenure insecurity is one of the key contributors to conflict, low land investments, and low maize productivity (Mekureyaw, 2017).

Access to secure land tenure is considered as one of the key drivers to rural development and reducing poverty (Higgins et al., 2018). Policymakers all over the world have prioritized securing land tenure of their citizens (Alban Singirankabo & Willem Ertsen, 2020). In this study, land tenure security is measured by possession of a land title deed. Possession of a land title deed has been chanted as one of the main ways of ensuring that rights to one’s land are formally recognized and protected by the state. The International Federation of Surveyors defines land titling or registration as the official process of recording lawfully recognized rights to land (Barrows & Roth, 1990). This is usually accompanied by a formal and legal document given to the owner(s) as proof of ownership and registration. The document indicates the rights held on a particular piece of land as provided by law.

Before independence, Kenya operated with a mixture of private and customary land laws, which were viewed as impediments to increase agricultural production (Kieyah & Nyaga, 2010). Post-independence, Swynnerton Plan was published to intensify changes in land tenure systems from customary to private through land registration (Swynnerton, 1954). To ensure access to secure land rights, the government enacted the Registered Land Act of 2012, which reclassified land into three categories: private land, public land, and trust land. The 2010 Kenyan constitution recognizes three types of tenure systems, namely private, communal or customary and public or state (Djurfeldt, 2020). The private or freehold land tenure system consists of land rights held by an individual. This category of land was previously occupied by the white settlers and converted from community land to private land. The community land tenure system also referred to as group ranches is defined as rights over this land held by a particular community as an ancestral property. Public or state or trust land tenure system refers to rights over the land that is held by the state in trust. This land includes land occupied by schools, hospitals, forests and game reserves national parks. Due to the increasing population, diminishing livelihoods and pressure by the political class to formalize and privatize, the community land has been sub divided into plots held by private persons (Willy & Chiuri, 2010) hence the need to secure these rights to land. Land ownership in Narok county is categorized into three (3); community land, trust land, and private land. Private land forms the largest category, with more than 50% of the households possessing land title deeds (Bambio & Agha, 2018).

Under the constitution of 2010, Kenya enacted the Land Act of 2012 to provide guidance on land registration. Today, approximately 30% of Kenya’s total land area is registered with over 5 million title deeds (Jeppesen & Hassan, 2022). These various legislations aim at promoting the security of land tenure, which can be used as collateral for credit access from financial institutions. The credit obtained could be used to invest in the adoption of modern agricultural technologies, hence increasing productivity (Kieyah & Nyaga, 2010). This study defines access to credit as whether a farmer has the ability to apply for credit in a formal credit source. According to the Chandio et al. (2018), there are 15 national commercial bank branches, 8 microfinance institutions and 15 saving and credit cooperative societies (SACCOs) in Narok county. Credit facilities are one of the major financial services offered by these institutions.

Traditionally, land tenure security and agricultural productivity have been investigated separately, hence the linkage between the two is relatively a new subject area (Atwood, 1990; Chigbu et al., 2019). Previous study by Hanstad (1998) argues that the possession of individual and secure land tenure is key to increasing agricultural production, poverty reduction, and economic growth. Land tenure security also positively affects access to credit and agricultural investments (Besley, 1995; H. G. Jacoby et al., 2002; Sossou & Mbaye, 2018). Moreover, Bambio and Agha (2018), Nkomoki et al. (2018), and Paltasingh (2018) argue that security of land tenure, whether land title or customary significantly increases the adoption rate of agricultural technologies, encourages enterprise diversification and use of complementary inputs as well as promoting agricultural investments. Land tenure insecurity demotivates households from investing in the land for fear of eviction, as evidenced by Deininger et al. (2009), Linkow (2016), and Lovo (2016). Other studies by Akudugu (2011), Lambin and Meyfroidt (2011), and Holden and Ghebru (2016) suggest that land tenure security increases land conservation practices and agricultural productivity.

However, S. E. Migot-Adholla et al. (1994) and Pinckney and Kimuyu (1994) found no significant effect of land titling on agricultural productivity due to the underdevelopment of credit markets. Land tenure security has no significant effect on agricultural land investments, access to credit, and crop yield (H. Jacoby et al., 2007; Matchaya, 2010; S. Migot-Adholla et al., 1991; Place & Hazell, 1993). Much of the research concur that tenure security stimulates investments; however, the reverse is also true, as found in research by Brasselle et al. (2002) and Moreda (2018).

Although, previous studies have attempted to investigate the effect of land tenure security on agricultural productivity, there is still limited literature linking land tenure security, access to credit, and agricultural productivity in Kenya and other SSA countries. The available literature has resulted in inconclusive findings, while the results vary by crop enterprise, country, and even the measurement of land tenure security variable. Most of the literature on this subject (Bangwayo-Skeete et al., 2010; Field et al., 2006; Kohansal et al., 2008; Newman et al., 2015; Sheuya & Burra, 2016) have studied only two components that is either land tenure security and agricultural productivity or land tenure security and credit access and productivity, thus breaking the conditional link between land tenure security, access to credit and agricultural. Analysing the linkage between land tenure security and agricultural investments without considering whether access to credit is caused by land tenure security may result in erroneous findings (Joel & Bergaly, 2020; Rashid, 2021). This paper therefore attempts to bridge this knowledge gap and contribute to the body of knowledge by analysing the effect of land tenure security on maize productivity through credit access in Kenya. This paper also uses an innovative econometric model (Conditional Mixed Process) to address the endogeneity problem that exists in this study. Additionally, the research was carried out in an area that is prone to land-related conflicts, hence providing a unique feature to this study. The rest of the paper is organized as follows: section 2 provides the theoretical framework, while section 3 presents the methodology that includes a brief on the study area, sampling procedure and analytical method used in the study. Section 4 provides the descriptive statistics, model results and discussions on the effect of land tenure security on agricultural productivity. Section 5 provides the conclusion and policy recommendations based on the findings. Section 7 presents the study's limitations and recommendations for further research.

2. Theoretical framework

The study is based on the property rights theory and the Cobb-Douglas theory in explaining the hypothesized nexus between land tenure security, credit access and maize productivity.

2.1. Theory of property rights

This theory describes the relationship between property rights and land tenure security. Property rights are institutions (formal or informal) that define or provide a scope of privileges granted to individuals for certain resources or assets such as land (Coase, 1960). These rights include rights to use, obtain income from and exclude others from using the assets such as land. In a wider view, property rights can be described as “bundle of rights” including the right to access and withdrawal, exclusion, management and alienation rights (Schlager & Ostrom, 1992). Formal property rights are legal and are allocated or recognized or supported by government authorities and enforced by the laws of the country. The rights are categorized into 4, namely, private, common, state and open access property rights. Private rights assign individual rights to physical objects and are viewed to be absolute and relative since it ensures transferability of rights to better use of the asset by others (Furubotn & Richter, 2008).

The classical theory of property rights argues that sustainable management of natural resources is only possible with individual property rights since it ensures tenure security (Chigbu et al., 2019). Demsetz (1967) and Hardin (1968) note that individual property rights are rational when making decisions relating to the private ownership system. Radin (1982), Arnold (2002) states that, individual property rights are considered as an innovation towards the agricultural sector structural transformation. Thus, land tenure security, which provides individual rights over the land, grants the owner the rights to use, transfer, mortgage and exclude other people from using the land. The theory of property rights holds that security of land tenure is the foundation of economic growth (Chigbu et al., 2019).

2.2. Cobb Douglas theory

The theory was estimated by Cobb and Douglas (1928) and has been used extensively to estimate productivity and production. It is a form production function representing the relationship between output and inputs in a particular firm or farm. It takes the following form;

(1) $Q=A(t)L^{\beta 1}{K}^{\beta 2}$(1)

In linear form, it takes the form as expressed in equation 2(2) $LnQ=\alpha +\delta T+{\beta }_{1}L+{\beta }_{2}K+\epsilon$(2) ;

(2) $LnQ=\alpha +\delta T+{\beta }_{1}L+{\beta }_{2}K+\epsilon$(2)

where $Q$, $L$ and $K$ are vectors of output, labour and capital respectively, $\delta$ is the time variable, $\alpha$ is a constant, $\beta 1$,$\beta 2$ are elasticities and $\epsilon$ is the error term. This theory is key in explaining the production behaviour of firms. Maize productivity is determined by the labour, capital and other factors such as age, farm size and input use.

2.3. Heterogeneity analysis

Previous studies on the relationship between land tenure security, credit access and agricultural productivity reveal a mixture of findings. A recent study in Benin found that, an enhanced security of land tenure in the initial stages of land formalization has a positive effect on the decision to invest and hence increase in agricultural productivity (Akudugu, 2011). Additionally, Mekureyaw (2017) argues that improved land rights may increase land-related investments while at the same time investments could increase land tenure security. Improved security of land tenure increases transparency, hence giving confidence to financial institutions while accessing credit facilities. The accessed credit could be used in agricultural investment hence increase in agricultural productivity (Barrows & Roth, 1990; Chigbu et al., 2019). However, research in Kenya show that land tenure security did not have a significant effect on yields. The study questioned the rationale for the government's ambitious land registration and titling program (Delville, 2010). Moreover, Field et al. (2006) noted that, there was a negative relationship between land tenure security and agricultural productivity.

3. Methodology

3.1. Study area and sampling procedure

The study is based on cross-sectional data collected in September 2021 from 366 randomly selected small-scale maize farmers in Narok county, Kenya. According to the Chandio et al. (2018), Narok county is one of the 47 counties in Kenya and it borders the Republic of Tanzania to the south, Bomet, Migori, Nyamira, and Kisii counties to the west, and Nakuru and Kajiado counties to the north and east, respectively. Narok county, which is home to about 1,057,873 persons, is divided into six (6) sub-counties (Narok North, Narok South, Trans Mara East, Trans Mara West, Narok East, and Narok West) with Narok town being the county headquarters (Chigbu et al., 2019). The county is mainly occupied by the Maasai community, one of the plain nilotes of Kenya (Wambua et al., 2018).

The county is divided into four (4) agro-climatic zones; humid, sub-humid, semi-humid, arid, and semi-arid. Temperatures in Narok county range from 12° to 28° (Bambio & Agha, 2018). The county has two rainfall seasons; long rains are experienced in the months of February to June, while the short rains are between August and November. The estimated annual average rainfall is between 500 mm and 2500 mm, while crop farming, pastoralism, tourism, and trade are the main economic activities in the county. Maize, beans, and sugarcane are the main crops grown in the county. Other horticultural crops such as tomatoes, potatoes, and cabbages are mainly grown under irrigation. Maize is considered both a food crop and a cash crop in Narok county since it can be consumed and also sold to earn income (Bambio & Agha, 2018).

A multistage sampling procedure was employed to select the study respondents. First, Narok County was purposively selected because it is one of the highest maize producers in Kenya, ranked at position five after Trans Nzoia, Uasin Gishu, Bungoma, and Kakamega counties (Government of Kenya, 2021). Additionally, the predominant community (Maasai) has suffered land-related conflicts even before independence. Besley (1995) and Kariuki et al. (2016) argue that, due to the Maasai community’s nomadic pastoralist way of life, historically, they were economically, socially, culturally, and politically marginalized due to their inability to fit into the western development pathways advocated by the colonial government hence loss of their ancestral land. The second stage involved the selection of Transmara East and Transmara West sub-counties since they are the highest maize producers in the county (Bambio & Agha, 2018). Based on proportionate to size of small maize farmers, 179 and 187 small-scale farmers were sampled in Transmara East and Transmara West, respectively. The third stage involved the choice of four wards (Mogondo, Ololmasani, Angata Barikoi, and Kilgoris Central) in the two sub-counties of Narok County because they were the largest maize producers in their respective sub counties. Based on the number of small-scale maize farmers in the selected wards, 88 respondents were selected in Mogondo, while 92, 64 and 122 respondents were selected in Ololmasani, Angata Barikoi, and Kilgoris Central, respectively. The last stage was a systematic random sampling of small-scale maize producers from a list provided by the county agricultural office. The choice of the study area offers diversity with respect to climatic conditions, socio-economic, institutional, and land-related characteristics. Before the data collection exercise began, a research permit was obtained from the National Commission for Science Technology and Innovation (NACOSTI). The data were collected using a semi-structured questionnaire, with households being the unit of analysis. The data were later coded and analysed using STATA 15 computer software to obtain descriptive and econometric results (Newman et al., 2015)

3.2. Econometric model and estimation

This study analyses the relationship between land tenure security, credit access and maize productivity. This article defines land tenure secure farmer as one whose name is registered in the title deed of the land they farm. On the other hand, access to credit is defined as the ability of farmers to apply for credit. Maize productivity, the dependent variable, refers to the quantity of output produced using given units of inputs such as labour and land (Holden & Ghebru, 2016). When using all the inputs, it is referred to as total productivity, while when using part of the inputs, it is called partial productivity. Due to the problems of data inadequacy caused by poor record-keeping, this paper defines productivity as maize output in kilograms per unit of land in hectares in the 2020 production as used by Dlamini and Masuku (2011), Joel and Bergaly (2020), and Sarma and Rahman (2020). Therefore, land tenure security denoted as LTS is a dummy variable, 1 indicating a farmer who is land tenure secure while 0 otherwise, credit access denoted as CRT is also a dummy variable, 1 representing a farmer who has the ability to take credit and 0 otherwise and productivity denoted as PDT is a continuous variable.

As conceptualized in this study and used by Feder et al. (1988) and Place (2009), there are several important linkages between land tenure security, credit access and maize productivity. Firstly, possession of a title deed on the land increases the ability of the land to be used as collateral to acquire credit facilities. Therefore, the acquired credit facility could be used to purchase complementary inputs such as fertilizer, certified seeds, pesticide, and hired labour to increase maize productivity. Furthermore, credit is expected to reduce liquidity challenges, hence motivating the farmer to make investments in the farm through the adoption of modern technologies resulting in an increase in maize productivity (Deininger et al., 2009; Feder et al., 1988; Holden & Ghebru, 2016; Kariuki et al., 2016). Investments made on the land as a result of the credit acquired may increase the land value and hence its improvement in security during tenure. Thus, access to credit leads to land tenure security and vice versa. This scenario, therefore, results in a reverse causality, which is a source of endogeneity (Brasselle et al., 2002; Joel & Bergaly, 2020).

Additionally, farmers with well-structured project proposals may have a better ability to access credit than those in possession of title deeds. In most of the SSA countries, older farmers may have a higher ability to get credit than younger ones due to their perceived higher repayment ability. This therefore presents a sample selection bias problem. Furthermore, there are unobservable factors that may affect maize productivity (Di Falco et al., 2011; Rashid, 2021). Better motivated farmers and the ability of the farmer to better manage the farm may have an influence on farm performance yet such factors are difficult to measure hence they are omitted in this study. These omitted variables are a source of endogeneity in this study.

Empirically, the relationship between land tenure security, credit access, and productivity can be estimated as follows;

(3) $LT{S}_i={\alpha }_0+{\alpha }_1{X}_{i1}+{\epsilon }_{1i}$(3)

(4) $CR{T}_i={\beta }_0+{\beta }_i{X}_{i2}+\delta LT{S}_i+{\epsilon }_{2i}$(4)

(5) $PD{T}_i={\gamma }_0+{\gamma }_i{X}_{i3}+\eta CR{T}_i+{\epsilon }_{3i}$(5)

Where $X_{i1}$ denotes a vector of explanatory variables influencing land tenure security. These variables are sex, marital status, age, education level of the household head, household size, method of land acquisition, land dispute, fertility of land, land topography, number of years the household has stayed on the land, total land size, whether the household head has any community leadership position, access to the parcel of land, and whether the land is concentrated at one place as described in Table 1. $X_{i2}$ represents a vector of independent variables that affect access to credit. As explained in Table 1, these variables include sex, age, education level of the household head, possession of a title deed, total land size, group membership, total household income, access to market, and maize gross margin. $X_{i3}$ on the other hand, denotes a vector of variables effecting maize productivity such as the household head’s sex, education level, age, ownership on an ox, size of the parcel of land under maize, possession of a title deed, number of contacts with an extension agent, group membership, access to credit, use of hybrid seeds, fertilizer, pesticides, and herbicides. The parameters $\delta$ and $\eta$ represent the predicted value of land tenure security on credit access and the predicted value of credit access on productivity, respectively. ${\epsilon }_{1i}$${\epsilon }_{2i}$, and ${\epsilon }_{3i}$ are the error terms, while $\alpha$$\beta$$\gamma$$\delta$$\eta$ and represent the parameters to be estimated in the equations.

Table 1. Definition and descriptive statistics of continous variables used in the CMP model

Variable name	Description and measurement of variables	Land tenure security status		Significance
		Insecure	Secure
Continuous variables		Mean		t-statistic
Age	Age of the household head in years	42.82 (12.43)	44.55 (11.28)	−1.38
Education level	Years of schooling of the household head	8.51 (3.69)	10.56 (4.14)	−5.00***
Household size	Number of people in the household	4.55 (1.73)	4.53 (1.78)	0.09
Land stay	Number of years household had stayed on the land	16.27 (11.80)	16.93 (11.59)	−0.54
Land size	Total land size in Ha	1.49 (1.38)	1.83 (1.58)	−2.22**
Farm size	Size of land under maize cultivation in Ha	0.77 (0.87)	1.00 (0.86)	−2.59***
Parcel access	Walking time from the homestead to the parcel in minutes	3.53 (2.98)	3.86 (3.08)	−2.26**
Market access	Walking time from the homestead to the nearest market in minutes	39.10 (30.71)	38.11 (29.70)	0.31
Extension contacts	Number of contacts the respondent had with an extension agent	1.07 (0.84)	1.60 (1.61)	−3.98***
Maize productivity	Maize productivity (Kgs/Ha)	2305.19 (1211.95)	2686.36 (1262.72)	−2.94***
Household income	Total household income in KES	215418(179,213.60)	303,047.10 (222,227.30)	−4.17***
Maize gross margin	Maize gross margins in KES	3236285(2,230,325.00)	4,957,516(2,438,796.00)	−7.05***

**, *** represents significance level at 10% and5%respectively

Equation 3(3) $LT{S}_i={\alpha }_0+{\alpha }_1{X}_{i1}+{\epsilon }_{1i}$(3) estimates the determinants of land tenure security in Kenya using a binary probit. Equation 4(4) $CR{T}_i={\beta }_0+{\beta }_i{X}_{i2}+\delta LT{S}_i+{\epsilon }_{2i}$(4) also has a binary dependent variable, and it examines the determinants of credit access with a predicted value from the first equation. The third equation (equation 5(5) $PD{T}_i={\gamma }_0+{\gamma }_i{X}_{i3}+\eta CR{T}_i+{\epsilon }_{3i}$(5) ) analyzes the effect of credit access on maize productivity and uses the predicted variable derived from equation 4(4) $CR{T}_i={\beta }_0+{\beta }_i{X}_{i2}+\delta LT{S}_i+{\epsilon }_{2i}$(4) . Equations 3(3) $LT{S}_i={\alpha }_0+{\alpha }_1{X}_{i1}+{\epsilon }_{1i}$(3) , 4(4) $CR{T}_i={\beta }_0+{\beta }_i{X}_{i2}+\delta LT{S}_i+{\epsilon }_{2i}$(4) , and 5(5) $PD{T}_i={\gamma }_0+{\gamma }_i{X}_{i3}+\eta CR{T}_i+{\epsilon }_{3i}$(5) are recursive, that is, equations 3(3) $LT{S}_i={\alpha }_0+{\alpha }_1{X}_{i1}+{\epsilon }_{1i}$(3) and 4(4) $CR{T}_i={\beta }_0+{\beta }_i{X}_{i2}+\delta LT{S}_i+{\epsilon }_{2i}$(4) have binary dependent variables, while equation 5(5) $PD{T}_i={\gamma }_0+{\gamma }_i{X}_{i3}+\eta CR{T}_i+{\epsilon }_{3i}$(5) has continuous dependent variables. Since there are two equations with an endogenous form of the first equation that is inserted into the second equation as an exogenous variable, a multi-equation model may be appropriate to deal with the possible correlation between the equations. Additionally, the insertion of the predicted variable helps in dealing with the problem of endogeneity. However, the predicted variables can still cause bias. This paper, therefore, uses the conditional mixed process (CMP) model developed by Roodman (2007, 2011) and used by Joel and Bergaly (2020), Dlamini and Masuku (2011), and Rashid (2021). The model helps in accounting for the following: adapting to a system of apparently independent variables (Seemingly Unrelated Regression) (SUR), instrumental variables, system of simultaneous equations, and the differences in the nature of the dependent variables. It also takes care of the cross relationships that may exist between the equations of the model. Following the CMP format, equations 3(3) $LT{S}_i={\alpha }_0+{\alpha }_1{X}_{i1}+{\epsilon }_{1i}$(3) , 4(4) $CR{T}_i={\beta }_0+{\beta }_i{X}_{i2}+\delta LT{S}_i+{\epsilon }_{2i}$(4) , and 5(5) $PD{T}_i={\gamma }_0+{\gamma }_i{X}_{i3}+\eta CR{T}_i+{\epsilon }_{3i}$(5) are expressed as in equation 6(6) $y_1^{\ast }={\theta }_1+{\epsilon }_1$(6) , 7(7) $y_2^{\ast }={\theta }_2+{\epsilon }_2$(7) , and 8(8) $y_3^{\ast }={\theta }_3+{\epsilon }_3$(8) ;

(6) $y_1^{\ast }={\theta }_1+{\epsilon }_1$(6)

(7) $y_2^{\ast }={\theta }_2+{\epsilon }_2$(7)

(8) $y_3^{\ast }={\theta }_3+{\epsilon }_3$(8)

where ${\theta }_1={\beta }_{1}X,{\theta }_2={\beta }_{2}X+\delta y_1,{\theta }_3={\beta }_{3}X+\eta y_2$

(9) $y=g(y^{\ast })=(1\left\{y_1^{\ast }>\left.0\right\},y_2^{\ast },1\left\{y_3^{\ast }>\left.0\right\})\right.\right.$(9)

$\epsilon =({\epsilon }_1,{\epsilon }_2,{\epsilon }_3{)}^{\prime }N(0,\sum )$ and $\sum =\left[\begin{array}{ccc}1& {\rho }_{12}& {\rho }_{13}\\ {\rho }_{12}& 1& {\rho }_{23}\\ {\rho }_{13}& {\rho }_{23}& 1\end{array}\right]$

where $y_1^{\ast }$,$y_2^{\ast }$ and $y_3^{\ast }$ represents latent factors of land tenure security, credit access, and maize productivity, respectively. $X$ represents a vector of explanatory variables influencing land tenure security, credit access, and maize productivity, and ${\rho }_{12}$ represents the correlation between the residual errors of land tenure security and credit access, ${\rho }_{13}$ represents the correlation between the residual terms of land tenure security and maize productivity and ${\rho }_{23}$ represents the correlation between credit access and maize productivity.

With the assumption that $y_i=(0,y_{i2},0{)}^{\prime }$ it is observable, then the corresponding log-likelihood function can be expressed in equation 8(8) $y_3^{\ast }={\theta }_3+{\epsilon }_3$(8) ;

(10) $\begin{array}{rl}& L_i({\beta }_1,{\beta }_2,{\beta }_3,\delta ,\eta ,\sum ;y_i\left|x_i)\right.\\ & ={\int }_{-\mathrm{\infty }}^{-{\theta }_1}\hspace{0.17em}{\int }_{-\mathrm{\infty }}^{-{\theta }_2}\hspace{0.17em}{\int }_{-\mathrm{\infty }}^{-{\theta }_3}\hspace{0.17em}{\phi }_j\left\{({\epsilon }_1,y_{i2}-{\theta }_{12},{\epsilon }_3{)}^{\prime };\left.\sum \right\}\right.d{\epsilon }_{1}d{\epsilon }_{2}d{\epsilon }_3\end{array}$(10)

The choice of the independent variables was informed by studies by Dlamini and Masuku (2011), Joel and Bergaly (2020), and Rashid (2021). Endogeneity was tested using the Durbin-Wu-Hausman test (Durbin, 1954; Hausman & Taylor, 1978)

4. Results and discussion

4.1. Descriptive statistics

The descriptive statistics in Table 1 show that on average, land tenure secure household heads were significantly more educated, with relatively larger total land and farm sizes, higher maize productivity, and household income compared to land tenure insecure household heads. Additionally, the distance from the homestead to the parcel and maize gross margins were significantly higher for secure land tenure households as compared to land tenure insecure households. Land tenure secure households also had significantly more contacts with extension service providers as compared to land tenure insecure households.

According to the results in Table 2, the majority of the land tenure secure household heads were married, acquired land through purchase, and were holders of a community leadership position. Additionally, their land was fertile, and they experienced fewer land disputes in the last 5 years. The majority of the households with secure land tenure also had the ability to access credit from formal financial institutions, used hybrid seeds, herbicides, and owned oxen to provide labour on their farms. However, fewer used inorganic fertilizers as compared to those with insecure land tenure, indicating the disincentive for long-term investments for insecure land tenure holders.

Table 2. Definition and descriptive statistics of categorical variables used in the CMP model

Variable name	Description and measurement of variables	Land tenure security		Significance
		Insecure	Secure
Categorical variables		Percentage		${\mathit{\chi }}^{\mathbf{2}}$
Title deed possession	% of respondents with a land title deed	53.01	46.99
Sex	% of male household heads	51.55	56.98	1.08
Marital status	% of married household heads	59.79	70.35	4.45**
Community leadership	% of household heads who are community leaders	12.37	19.19	3.22*
Land concentration	% respondents with land concentrated in one area	98.45	97.09	0.79
Land acquisition	% of respondents with land acquired through purchase	16.49	29.07	8.30***
Land dispute	% respondents with a land dispute	12.89	7.56	2.78*
Land fertility	% of respondents with high land fertility	59.70	88.95	46.90***
Land topography	% of respondents with land that is hilly	32.99	40.70	2.33
Use of hybrid seeds	% of respondents who used hybrid seeds	90.72	95.35	2.96*
Use of inorganic fertilizer	% of respondents who used fertilizer	60.00	56.52	0.43
Use of herbicides	% of respondents who used herbicides	10.27	11.18	0.07
Credit access	% of respondents with the ability to access credit	39.69	51.74	5.34**
Group membership	% of respondents who are members of at least one group	73.20	76.16	0.42
Oxen ownership	% of respondents owning a donkey	13.34	22.67	5.37**

*, **, *** represents significance level at 10%, 5%, and 1% respectively

4.2. Effects of land tenure security on maize productivity through credit access

Table 3 presents the results of the effect of land tenure security on agricultural productivity through access to credit. The results are in three stages: Stage 1 provides the determinants of land tenure security. Stage 2 presents the effect of land tenure security on credit access, while stage 3 shows the effect of credit access on agricultural productivity. The model fit results, such as the likelihood-ratio test (−3335.930), indicate that it is statistically significant at 1% level thus the model can correctly explain the relationship between the land tenure security and maize productivity. The Durbin-Wu-Hausman test (4.932) for endogeneity was found to be significant at 10% level, indicating the presence of endogeneity in the analysis and thus justifying the use of the CMP model.

Table 3. Conditional mixed processes model results

Variable	Model 1	Model 2	Model 3
	Land tenure security	Credit access	Productivity
	Coefficients	Coefficients	Coefficients
Title deed possession		0.849***(0.235)
Credit access			2001.902***(150.962)
Socio-economic characteristics
Sex	0.485** (0.207)	−0.079 (0.137)	11.923 (198.930)
Marital status	0.403** (0.204)
Age	0.008 (0.008)	−0.008 (0.006)	8.266 (8.440)
Education level	0.078*** (0.021)	0.048** (0.0190)	46.780* (28.449)
Household size	−0.037 (0.036)
Household income		−2.422 (2.251)
Land related characteristics
Land dispute	−0.471** (0.235)
Land acquisition	0.360** (0.155)
Land fertility	0.897*** (0.170)
Land topography	0.167 (0.139)
Land stay	0.012 (0.008)
Land size	0.048 (0.051)	0.065* (0.038)
Parcel access	0.029 (0.021)
Land concentration	−0.586 (0.467)
Plot size			−222.303** (94.189)
Institutional characteristics
Community leadership	0.040 (0.187)
Group membership		0.022 (0.145)	1.493 (187.966)
Market access		−0.002 (0.001)
Oxen ownership			230.030 (169.391)
Extension contacts			76.988 (47.833)
Productivity related factors
Maize gross margin		1.711*** (2.181)
Use of hybrid seeds			476.887** (217.960)
Use of inorganic fertilizer			−228.483** (117.082)
Use of herbicides			−5.582 (198.918)
Diagnostic and model fit
Durbin-Wu-Hausman test	4.932*
Number of observations	366.000
LR chi2(35)	246.110
Prob>chi2	0.0000
Log-likelihood	−3335.930

*, **, *** represents significance level at 10%, 5%, and 1% respectively

4.2.1. Determinants of land tenure security

According to the results in Table 3 model 1, male household heads are more likely to be land tenure secure as compared to female household heads. It is relatively easier for a man to get a title deed than a woman because they are considered superior to women and hence they can easily initiate the land registration process to obtain a title deed for their land. Similarly, Joel and Bergaly (2020) suggest that in traditional societies, women are taken as fragile and can be easily swayed to give the land to another person, thus denying them the security of land tenure is considered a way of protecting the land from grabbing.

Marriage increases the probability of having secure land tenure. Probably, obtaining a title deed by married people is often considered a means of securing the future of their children. Higgins et al. (2018) argue that for fear of disputed inheritance and lengthy legal battles in case parents die, married people tend to secure their land through registration. Education level was also found to influence land tenure security. More educated household heads are likely to be more land tenure secure than the less educated household heads. Education helps one to acquire information on land, financial markets and land policies on registration hence making them more likely to acquire a title deed. Similarly, educated landholders can easily understand the need for the acquisition of a title deed (Kusiluka & Chiwambo, 2018). They can also easily marshal all the available resources towards securing their land (Rashid, 2021).

Land disputes reduce the likelihood of securing the land. Disputed parcels of land do not qualify for land adjudication until the dispute is fully resolved. Similarly, land with disputes attracts low prices in the land market, hence less likely to secure it (Ngango & Hong, 2021). Land purchase positively and significantly influences land tenure security. Acquisition of land through purchase increases the probability of securing land. The land purchasing procedure in Kenya is a legal process that is protected by law, thus purchasing land is deemed as a legal way of protecting one’s land from illegal exploration. In areas with developed land markets, the purchase of land guarantees security for the land since the process is detailed and involves legal procedures (Rashid, 2021). Land fertility was also found to be influencing land tenure security. Households with fertile land are more likely to secure it due to the perceived value attached to it. Coulibaly (2021) notes that land fertility increases its value in the land market thus becoming a target for land grabbers, hence more likely to secure it.

4.2.2. Effect of land tenure security on credit access

The results in Table 3 model 2 show that, possession of a title deed (land tenure security), years of schooling, total land size, and maize gross margins significantly affected credit access. Land tenure security positively affects the ability to access credit. This implies that possession of a land title deed enables holders to use it as collateral to access credit from formal financial institutions. Land tenure security provides an incentive for the development of land markets, hence enabling farmers to use the land as collateral to access credit facilities (Higgins et al., 2018; Joel & Bergaly, 2020; Rashid, 2021). However, Delville (2010) argues that corruption and lack of transparency in public administration may render the title deed worthless to be used as collateral. In addition, land tenure security may not spur credit access in areas where formal lending is underdeveloped (Field et al., 2006; Fort, 2007).

More educated household heads increase the probability of having the ability to access credit. Education exposes an individual to information on credit availability and how to access it. Temesgen et al. (2018) suggest that educated persons have a higher degree level of literacy in matters of finance, available technologies, innovations, risk evaluation strategies, production, and marketing, which enables them to easily respond to market signals as compared to uneducated persons. Moreover, education improves credit availability awareness, which helps in financial decision-making (Mitra et al., 2018; Patriciah & Wario, 2016).

Land size also positively and significantly increases the likelihood of having the ability to access credit facilities. Land is usually taken as a measure of wealth in most communities; hence, households with more land are deemed to be wealthier and thus can easily access credit due to their better repayment ability. Large land size may act as a sign of high social status and hence can facilitate access to credit even in informal credit sources (Chandio et al., 2018; Saqib et al., 2016).

High maize gross margins also positively and significantly increase the probability of having the ability to access credit. Formal lenders could use income obtained from maize production as collateral to access credit; hence, more gross margin may translate to a higher ability to access credit. Joel and Bergaly (2020) argue that revenues from the sale of crop production can be used to pay off loans granted to the farmer in case they default.

4.2.3. Effect of credit access on maize productivity

According to the results in Table 3 model 3, land tenure security, credit access, years of schooling of the household head, and use of hybrid seeds significantly and positively affect maize productivity, whereas parcel size and use of inorganic fertilizer significantly and negatively affect maize productivity. Being the main explanatory variable, access to credit increases farmers’ maize productivity by about 2001.902 Kg/Ha. Agricultural credit facilities can be used to purchase complementary inputs such as fertilizer and hybrid seeds, which in turn can increase maize productivity in the short run. In the long run (Rashid, 2021; Traore, 2012) argue that farmers can use credit to make investments in soil and water conservation methods. Additionally, credit can help in new technology adoption and facilitating the performance of other production factors such as land hence increasing productivity (Ali et al., 2014; Chandio et al., 2018; Sekyi et al., 2017; Zhao & Barry, 2014).

Maize productivity reduces by approximately 222.203 kg/ha, with an increase in farm size by a hectare. Farmers with smaller farm sizes do not enjoy economies of scale hence lower productivity as compared to those with larger farm sizes. Similarly, Akudugu (2016), 1998) and Sekyi et al. (2017) argue that relatively smaller farms sometimes tend to intensify production to maintain their household welfare, whereas their counterparts pursue more extensive methods. This finding, however, contradicts Akudugu (2011) and Melesse and Awel (2020), who suggest that land is often deemed to be a measure of wealth, and thus wealthier farmers use better complementary inputs such as fertilizer and hence greater productivity.

An increase in the years of schooling of the household head by 1 year increases maize productivity by approximately 46.780 kg/ha. Education allows for better adoption of intensive agricultural technology and innovation by farmers, hence improved productivity (Joel & Bergaly, 2020). The use of hybrid seeds also influenced maize productivity. The use of hybrid seeds increases maize productivity by approximately 476.887 kg/ha. Hybrid seeds possess the requisite genetic capability and can adapt to the local conditions, which could lead to high productivity. Hybrid seeds are manufactured with characteristics such as pests and disease resistance, high yielding, and early maturing which may increase the production per unit area of land (Coulibaly, 2021; Dembele, 2011). On the other hand, use of inorganic fertilizer reduces maize productivity by about 228.483 Kg/Ha. Excessive use of inorganic fertilizer alters soil pH, hence inhibiting the absorption of minerals necessary for crop growth and hence reducing productivity. Similarly, Vigani et al. (2015) note that inorganic fertilizer can only result in increased productivity in a controlled biophysical soil environment.

5. Conclusion and recommendations

The study examined the nexus between land tenure security, credit access, and maize productivity as a contribution toward achieving the SDGs, especially goal 2 of zero hunger. The study was underpinned on the assumption that land tenure security facilitates access to credit, which can be used to undertake short- and long-term investments, hence increases in maize productivity. Specifically, the study investigated the determinants of land tenure security, the effects of land tenure security on credit access, and finally, the effects of credit access on maize productivity using the conditional mixed process model.

Descriptive results show that the majority (53.01%) of the sampled farmers had no title deeds. Additionally, the majority of land tenure secure households acquired credit (51,74%), and their land productivity was higher (2686.36 Kg/Ha) as compared to land insecure households. The Durbin W. Hausman's results indicate the presence of endogeneity, hence justifying the use of a conditional mixed process. The CMP results show that land tenure security was significantly influenced by sex, marital status, education level of the household head, as well as the presence of land disputes, soil fertility, and land acquisition through purchase. The study confirmed that there is indeed a relationship between land tenure security, credit access and maize productivity. Land tenure security significantly and positively influenced credit access, while there was also a positive and significant effect of credit access on maize productivity. The results of this study are consistent with findings by Joel and Bergaly (2020) and Rashid (2021) who found that there was a relationship between land tenure security, credit access and agricultural productivity. The findings of this study underscore the need for governments and other stakeholders to develop policies that would facilitate access to land title deeds by the rural population as a way to increase agricultural productivity. Furthermore, agricultural credit could be made accessible to the majority of small-scale farmers to enable them to undertake short- and long-term investments that would increase their productivity.

6. Limitation of the study and further research

The study was aimed at analysing the relationship between land tenure security, credit access and maize productivity for improved livelihoods of small-scale farmers. The study only focused on measuring land tenure security using possession of a land title deed. Since land tenure security has different dimensions, future studies should explore the use of different measures of land tenure security such as ownership and control of land rights. The study also used maize as food security crop, however, in the future, maize crop enterprise can be used as a cash crop to inform on the contribution of land tenure security on income. Furthermore, other crop enterprises such as rice and beans may also be used.

Acknowledgements

The authors express their gratitude to the German Academic Exchange Service (DAAD) through the African Economic Research Consortium (AERC) for funding this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was funded by the German Academic Exchange Service (DAAD) through the African Economic Research Consortium (AERC) under grant No. AE/SP/17-049.

Notes on contributors

Joseph Jabu Mbudzya

Mr. Joseph Jabu Mbudyza is a part-time lecturer in the Department of Agricultural Economics and Agribusiness Management, Egerton University, Kenya. He is a Ph.D. candidate in Agricultural Economics at Egerton University, Kenya, specializing in land economics. He is the holder of a Masters of Science degree in Agriculture and Applied Economics from Egerton University, Kenya, and the University of Pretoria, South Africa. Joseph is also a social entrepreneur and founder of an Agricultural Radio Show dubbed “Kilimo Bora” loosely translated as Good Agriculture to educate farmers on different agricultural skills. His research interests include land economics, impact evaluation, food security, natural resource management, land governance, agricultural marketing, and gender studies. He is also a private consultant in the areas of poverty, food security, smallholder agriculture, gender, land tenure security, land governance, and agri-journalism.