Introduction
Research Context
Since 1986, Vietnam has transitioned from a centrally planned economy to a market-oriented one, achieving significant economic growth and poverty reduction Despite these advancements, disparities between rural and urban areas persist, prompting the government to prioritize rural and agricultural development for sustainable economic growth and political stability Microfinance has emerged as a vital tool for poverty alleviation, with rural credit programs designed to enhance financial access for poor households The government has implemented various credit initiatives to support rural development, yet challenges like credit rationing remain, limiting loan accessibility for vulnerable borrowers This paper aims to investigate the factors influencing credit rationing, particularly focusing on partial credit rationing, to provide insights for policymakers to enhance the effectiveness of government credit programs for the poor.
Research Problem
The rural credit market plays a crucial role in promoting economic development in rural areas, enhancing living standards, and aiding in poverty alleviation A key function of this market is to meet the credit demands of households, providing essential financial support for their growth and stability.
The impact of rural credit on the welfare of rural areas largely hinges on the effectiveness of the rural credit market A significant issue affecting this market is credit rationing, where lenders either deny loans or provide amounts below what borrowers request, even when borrowers are willing to pay higher interest rates to mitigate default risks This practice can severely limit households' ability to meet their credit needs, leading to inefficiencies within the local credit market Consequently, widespread credit rationing can undermine the overall performance and benefits of rural credit systems.
As highlighted by Stiglitz and Weiss (1981), asymmetric information plays a crucial role in credit rationing, particularly in rural credit markets where numerous impoverished households exist The difficulty in assessing the creditworthiness of these borrowers creates uncertainty for lenders regarding loan repayment probabilities To mitigate this issue, lenders seek various types of information about borrowers, including household dependency ratios, household size, land value, and social standing, to evaluate their repayment capacity and inform their lending decisions.
This study investigates the factors influencing lenders' decisions regarding credit rationing, particularly focusing on partial credit rationing across 12 provinces in Ha The impact of various information types on the likelihood and extent of credit rationing for borrowers is explored, with findings varying based on the time period and geographic context of each study.
Tay, Nghe An, Khanh Hoa, Lam Dong, Phu Tho, Quang Nam, Long An, Dac Lac, Dac Nong, Lao Cai, Dien Bien, Lai Chau period 2006-2008.
Research Objectives
The general objective of this study is to examining the relationship between partial credit rationing and its determinants in rural area of Vietnam in period 2006-2008.
The general objective could be archived by meeting following sub-objectives:
Identify key factors that affects the household’s credit accessibility
Identify key factors that affects the degree of partial credit rationing
Identify key determinants on the probability of partial credit rationing
Suggest policy implication to reduce partial credit rationing practice
Research Questions
The research aims at answering the following questions:
What are the key determinants on the credit accessibility of rural households?
What are the key determinants on the partial credit ration probability of rural households?
What are the key determinants on the degree of partial credit rationing of rural households?
Scope of Study
This study examines partial credit rationing at the household level across 12 provinces in Vietnam, including Ha Tay, Nghe An, Khanh Hoa, Lam Dong, Phu Tho, Quang Nam, Long An, Dac Lac, Dac Nong, Lao Cai, Dien Bien, and Lai Chau, during the period from 2006 to 2008 Utilizing data from the Vietnam Access to Resources Household Survey (VARHS) 2008, the research highlights how lenders limit loan amounts, preventing borrowers from fully meeting their credit needs.
Literature Review
Rural credit
Rural Credit is the financial support provided to farmers for their agricultural and related rural activities, accounting for approximately 90 percent of rural finance activities (Pham, T., 2010).
2.1.2 Characteristics of rural credit market
Rural credit market has some distinct characteristics:
High transaction costs in the rural credit market stem from various factors, including the geographical dispersion of local users, the diverse segments within the farming community, and the small value of loans Additional expenses arise from lost time, travel costs, and non-interest charges associated with obtaining and repaying loans, as well as making deposits Furthermore, the underdeveloped infrastructure for transport and communication contributes to elevated information and marketing costs.
The rural credit market is characterized by high risk due to several factors, including vulnerability to adverse climate conditions, low returns on agricultural investments, and the necessity for household consumption Additionally, the concentration of economic activities in small geographic areas leads to a chain effect, exacerbating income variability for farmers as prices fluctuate This situation increases the likelihood of loan defaults, as acceptable collateral is often scarce, and property rights related to mortgaged land can be uncertain and difficult to enforce Compounding these challenges is a weak legal system and ineffective enforcement mechanisms.
Rural credit providers can be categorized into three sectors – formal, semi-formal and informal credit suppliers.
The banking sector in Vietnam comprises various institutions, including commercial banks, foreign bank branches, joint-stock banks, and joint venture banks Notably, the Vietnam Bank for Agriculture and Rural Development (VBARD) serves as a primary source of credit for rural households, while the Bank of Social Policy (BSP), a government-owned non-profit entity, focuses on providing financial support to impoverished communities, ethnic minorities, and households benefiting from social policies.
In rural areas, the provision of loans through socio-political unions is closely linked to government priority programs, bank consignment services, and the activities of various unions such as the PCF, Women’s Union, and Farmers’ Association These loans typically feature low interest rates, smaller amounts, and short-term repayment periods, making them accessible to local communities.
Informal credit sources play a crucial role in rural areas, primarily due to the underdeveloped formal credit market These sources include mutual lending among friends and neighbors, rotating savings and credit associations, specialized moneylenders like pawnbrokers, and traders Loans from the informal sector typically exhibit significant variation in both interest rates and loan amounts.
Asymmetric Information and Credit Rationing
Asymmetric information frequently occurs in rural credit markets, where one party possesses more information than the other during transactions In this context, borrowers often have a clearer understanding of their creditworthiness than lenders, as crucial details like income for repayment and intended loan use are primarily known to the borrowers themselves.
Two main problems related to information asymmetry:
1 Adverse selection- immoral behavior that takes advantage of asymmetric information before a transaction For example, credit borrowers know their project is high risk and high return, so they may readily accept a high level of interest rate that the lenders offer them.
2 Moral hazard - immoral behavior that takes advantage of asymmetric information after a transaction For instance, when the loan has been disbursed to borrowers, lenders may have difficulty in monitoring on how their lending money is used, and the borrowers may use the loan for purposes other than the one they stated in the loan contract The purposes may earn more return to the borrowers, but riskier for the lenders.
Expected return to lender r* Quoted interest rate
2.2.2 Problems of lenders in context of asymmetric information
To comprehend the behavior of credit lenders in the context of asymmetric information, it is essential to first grasp the expected return function of these lenders Jafee and Stiglitz (1990) illustrate that a bank's expected return is influenced by the quoted interest rate, which can be graphically depicted as a concave curve.
Figure 1 – Credit Supplier Expected Return
Credit lenders reach the highest level of expected return when it charges the loan at interst rate r *
When interest rates deviate from the optimal level of r*, lenders experience a decline in expected returns, making them hesitant to adjust their rates.
The assumption of a concave curve for credit lenders' expected return is crucial to Jafee & Stiglitz's (1990) analysis of credit market behavior This concept raises an important question: why does an increase in interest rates lead to a decrease in expected returns for lenders? The answer lies in the imperfect information problem prevalent in the credit market, specifically stemming from adverse selection and adverse incentive effects.
The adverse selection effect can cause the expected return curve for credit suppliers to take on a concave shape, indicating that as interest rates exceed the optimal level, the expected returns start to decline.
As interest rates rise, lenders face an adverse shift in their lending portfolios, leading to an increased risk of default Safe borrowers, who typically seek credit for low-risk, low-return projects, are unable to afford higher interest rates and exit the market In contrast, more risky borrowers seeking funds for high-risk, high-return projects become prevalent in the lending portfolio This shift results in a higher likelihood of default among these borrowers, ultimately diminishing the expected returns for lenders.
Adverse incentive, also known as moral hazard, can influence the expected return curve for credit suppliers, making it concave After securing loans, borrowers may alter their behavior by engaging in riskier projects to achieve higher returns that offset elevated interest rates, deviating from the terms of their lending contracts This shift not only increases the risk associated with the lending portfolio but also diminishes lenders' anticipated returns While lenders attempt to mitigate this risk through monitoring practices, these efforts are often costly and imperfect.
Due to information asymmetry, credit suppliers are hesitant to adjust their interest rates away from the optimal level (r*) This cautious behavior results in a scenario where, despite high demand for credit potentially driving up interest rates according to supply and demand principles, the market equilibrium interest rate remains anchored at the lenders' optimal level Consequently, some borrowers willing to pay higher interest rates to meet their credit needs may find themselves unable to secure the necessary credit, leading to loan application rejections In essence, these borrowers face credit rationing by lenders.
Figure 2 - Rationing in Credit Market
In the realm of lending characterized by asymmetric information, credit suppliers encounter three critical challenges: identifying high-risk borrowers to impose credit constraints (screening), monitoring loans to prevent misuse (incentives), and ensuring repayment from borrowers who have the ability to pay (enforcement) To address these challenges, lenders commonly employ two screening mechanisms: indirect and direct screening methods.
Credit suppliers may charge high interest to cover the risk of default on borrowers However this type screening may lead to the issue of moral hazard and adverse selection.
Credit lenders may give threat of cutting off credit or contractual terms in other exchanges to monitoring the loan use and enforce repayment of borrowers.
In the realm of asymmetric information, credit lenders can implement direct screening mechanisms to evaluate loan approval by assessing clients' repayment probabilities To mitigate the risk of default, credit suppliers can utilize three key strategies.
Lenders must gather and assess essential client information, including income, education level, and age, to effectively evaluate loan risks This screening process allows lenders to identify and limit borrowers who lack sufficient information, ensuring a more accurate risk assessment for loan approval.
Credit suppliers can require borrowers to establish connections with related markets, such as input and output markets, to ensure that loans are utilized appropriately Additionally, they may restrict lending to specific geographic areas or kinship groups, focusing on individuals within a certain region or those they engage in trade with.
Lenders often require borrowers to use collateral, such as land, livestock, or other assets, to mitigate the risk of default If the collateral provided is deemed insufficient to secure the loan, borrowers may be considered unqualified for loan approval.
Credit Rationing
Credit rationing occurs when there is a higher demand for loans than what lenders are willing to supply, often due to interest rates being set below the market-clearing level This situation arises when borrowers are unable to obtain the loans they seek, even if they are prepared to pay higher interest rates than those offered by lenders.
There could be various types of credit rationing depend on how the term - “excess demand for loan” is defined.
Excess demand occurs when a borrower is offered a smaller loan than requested at a specific interest rate, necessitating the acceptance of a higher rate to secure a larger loan This phenomenon is classified as interest rate or price rationing (Jafee, D & Stiglitz, J.E., 1990).
Excess loan demand can occur when individuals struggle to secure loan approvals at interest rates that align with their perceived risk of default This situation exemplifies divergent views rationing, as outlined by Jafee and Stiglitz (1990).
Redlining is a form of credit rationing where lenders deny credit to borrowers based on risk classification, particularly when the lender cannot achieve the necessary return at any interest rate Additionally, loans that may be feasible at a certain required return, influenced by the deposit rate, can become unviable if the required return increases.
Pure credit rationing occurs due to imperfect information, leading to discrimination among borrowers with identical loan terms In this scenario, one group is approved for loans while another is not Jafee and Stiglitz (1990) assert that variations in credit availability, rather than interest rate fluctuations, significantly influence borrowing levels.
According to Barham, Boucher and Cater (1996), Buchenrieder (1996), Heidhues and Schrieder
(1998), Zeller (1993), the case of being credit rationed can fall into three situations:
Figure 3 - Identify Case of being Credit Rationed
Many individuals in need of a loan hesitate to approach credit lenders due to concerns about their creditworthiness, fearing that their applications will be denied.
Individuals seeking credit often face challenges when their loan applications are completely denied by lenders, resulting in an inability to obtain the requested funds.
Loan applications from credit borrowers are often accepted, but the approved loan amount is typically less than requested According to Petric (2003), farm households experience credit rationing by formal lenders, meaning they are unable to borrow sufficient funds to cover essential inputs, investments, and necessary consumption expenses.
2.3.3 Impact of Credit Rationing in Rural Area
Numerous studies have examined the effects of credit rationing on the welfare of rural areas, revealing that it adversely impacts the efficiency of credit functions This inefficiency hinders economic performance and ultimately undermines social welfare in these communities.
Firstly, efficient credit market can improve the productivities in rural area Pham and Izumida
A study by Ali and Deininger (2012) revealed that eliminating credit constraints could significantly boost agricultural productivity in high-potential rural areas of Ethiopia, with an estimated increase of 11% This highlights the crucial role of credit in enhancing household production and agricultural output.
A well-functioning rural credit market plays a crucial role in alleviating poverty and enhancing rural household income Research by Józwiak (2001) indicates that farmers who have access to borrowing typically experience higher income growth and increased family labor utilization Additionally, Krandker and Faruqe (2003) provide evidence that access to credit significantly improves farm welfare.
Research by Li et al (2013) revealed that credit rationing leads to a significant decline in net income by 15.7% and a reduction in consumption expenditure by 18.2% for households in rural China Similarly, Feder et al (1990) demonstrated the detrimental effects of credit rationing on farm profitability.
Empirical Studies
Understanding the behavior of lenders and borrowers, particularly regarding credit rationing and accessibility, requires an analysis of credit demand and supply forces This section reviews previous studies on the determinants of credit demand and supply, establishing an empirical framework for the factors contributing to partial credit rationing.
Age significantly influences credit demand, with younger individuals typically seeking more credit due to their active engagement in business activities that require funding (Mpuga, 2004) Conversely, older individuals often rely on their savings However, research by Tang et al (2010) indicates that older farmers, benefiting from extensive social networks and social capital, are more likely to secure loans than their younger counterparts This finding is supported by Okurut et al (2005) in their study conducted in Uganda.
Credit demand varies by borrower gender, with women in rural areas often viewed as primarily responsible for household tasks rather than engaging in market-oriented activities Consequently, their need for credit is generally less than that of men (Nwaru, 2011).
Education significantly influences credit demand, with research by Tang et al (2010) showing that individuals with higher education are more inclined to borrow, particularly from formal credit sectors However, this trend may change at the four-year university level, where those with advanced education often prefer to depend on their high income rather than seeking credit, as noted by Chen and Chiivakul (2008).
The labor structure within a household significantly influences its credit demand Research indicates that a higher number of adults in a household is typically associated with an increased loan amount (Barslund & Tarp, 2008) Additionally, Pham and Izumida (2002) suggest that households with more adults often turn to credit markets to fund their production expansion needs Furthermore, households with a higher dependency ratio tend to seek more credit to alleviate the financial pressures associated with supporting a larger number of dependents (Pham & Izumida, 2002; Okurut et al., 2005).
Research indicates that household assets, such as livestock and farmland, positively influence credit demand, as these resources require working capital for maintenance and growth (Pham & Izumida, 2002; Hussain & Khan, 2011) However, a contrasting study by Duflo et al (2008) suggests that households with a larger number of livestock tend to demand less credit, as they are often in a stronger economic position and less reliant on borrowing.
Other factors such as household shock (illness), or social position (household having social responsibility in community) were also found to have positive effect on credit demand (Zeller, 1994).
Empirical studies identify key determinants of credit rationing, including demographic characteristics such as individual traits, borrowers' skills, credit history, the reputation of the household head, dependency ratio, gender, education, and collateral (Petric, 2003; Bester, 1987; Diamond, 1989; Pham & Izumida, 2002; Craigwell, 1992; McKee, 1989) Additionally, the purpose of borrowing and the size of the loan influence the likelihood of households facing credit rationing (Pham & Izumida, 2002) Furthermore, political and social networks play a significant role in shaping credit rationing behavior (Ali & Deininger, 2012) Factors such as land holdings, livestock and durable goods possession, along with the level of village infrastructure, are also crucial in determining whether a household experiences credit rationing (Chaudhuri & Cherical, 2012).
Collateral is essential in the rural credit market, serving as a signaling mechanism that helps identify low-risk borrowers who are willing to secure their loans with substantial collateral This practice reduces the risk of moral hazard, as higher levels of collateralization encourage investments in safer projects Conversely, a lack of collateral can lead to rationing of credit (Bester, 1987) Typically, land, livestock, and other assets are associated with credit constraints, with land being the most traditional form of collateral in this market, often receiving significant attention in research (Petric, 2003; Barslund).
Research has consistently identified land as a crucial factor influencing the likelihood of credit rationing Petric (2003) noted that households farming more rented land are more prone to being credit rationed, as rented land does not serve as acceptable collateral for loans Additionally, livestock is recognized as a valuable form of collateral in the rural credit market (Okurut, 2005; Barslund & Tarp, 2008).
Research has consistently shown that human capital, particularly education and farming experience, significantly influences credit constraints (Petric, 2003; Chaudhur & Cherical, 2012; Pham & Izumida, 2002; Vuong et al, 2012; Barslund & Tarp, 2008; Zeller, 1994; Ping, Heidhues & Zeller, 2010) Higher education levels are associated with increased productivity and societal respect, leading to enhanced loan repayment capabilities and improved creditworthiness.
Research indicates that gender plays a crucial role in credit access, with Petric (2003) highlighting that households with more women often face greater credit constraints due to lower income-generating capabilities and repayment ability Conversely, Chaudhur and Cherical (2012) found that female-headed households may have a higher likelihood of loan approval Additionally, larger family sizes tend to decrease the chances of securing loans from banks.
Research indicates that in Ethiopia, both young and elderly individuals face greater challenges in accessing credit compared to middle-aged individuals (Kereta, 2007) While Chaudhur and Cherical (2012) found a positive correlation between age and the likelihood of credit approval, Pham and Izumida (2002) reported a negative relationship However, neither study demonstrated a significant impact of age on lenders' rationing decisions.
Research indicates that the dependency ratio, which reflects the number of dependents relative to household size, is a crucial factor influencing credit lenders' decisions A higher dependency ratio can result in increased credit rationing.
2002) The interpretation was that households with large number of dependents are normally poor as the more dependents means the more economic burden for the households.
Lenders often consider household reputation when making rationing decisions, as evidenced by Pham and Izumida (2002), who found that households with low reputations are more likely to face credit rationing Additionally, Diamond (1989) argues that reputation influences the interest rates set between lenders and borrowers Furthermore, Ali and Deininger (2012) highlight the impact of political and social networks on credit rationing behavior.
Methodology
Data Source and Features
This research will utilize the Vietnam Access to Resources Household Survey (VARHS 2008) as the primary data source to analyze the factors influencing rural credit rationing in Vietnam Initiated in 2002, VARHS has conducted surveys involving approximately 1,000 households across four provinces.
The VARHS survey, conducted in Ha Tay, Phu Tho, Quang Nam, and Long An, is repeated every two years with an expanding sample size In 2006, the survey covered 12 provinces with 2,324 households, and in 2008, it included 12 provinces with 3,223 households The next round of the survey is scheduled for 2010 as part of this ongoing project.
The Vietnamese Government has supported the Vietnam Access to Resources Household Survey (VARHS) since its inception in 2002 to investigate rural households' access to essential resources like land, credit, science and technology, market information, and other materials critical for economic and livelihood development Initially, VARHS surveyed approximately 1,000 households across four provinces: Ha Tay, Phu Tho, Quang Nam, and Long An The survey has been conducted biennially, expanding its scope over the years By 2006, VARHS included 12 provinces and 2,324 households, further increasing the sample size to 3,223 households in 2008, covering provinces such as Nghe An, Khanh Hoa, Lam Dong, Dac Lac, Dac Nong, Lao Cai, Dien Bien, and Lai Chau.
Figure 4 - Survey Site Mapping for VARHS 2008
The VARHS 2008 survey collected comprehensive household information categorized into four key areas: human capital, which includes demographic details such as age, education, labor, and employment; social capital, encompassing group memberships, social networks, political connections, and levels of trust and cooperation; liquidity and assets, detailing housing, land ownership, income, expenditures, savings, insurance, and credit situations; and production, which covers agricultural output, livestock, aquaculture, agricultural services, market access, irrigation, and the impact of production-related disasters.
For the purpose of the research, the data from VARHS is selectively extracted as follow:
Cover page Q1 Is the household classified as rural or urban Dummy Variable Rural/Urban
Q2 What is the ethnicity of this household Identify Ethnicity variable 1A - Roster 1 Q2 What is the relationship of [NAME] with the Identify Household head
To assess the demographic and socio-economic status of the household, key variables must be identified First, determine the gender of the household head, followed by the year of birth to ascertain their age and the number of adult members, which will help calculate the dependency ratio Next, evaluate the highest level of education attained by the household head Additionally, measure the total area occupied by the household in square meters, including all living spaces, and confirm whether the household owns the dwelling Lastly, verify the possession of a red book for any land owned by the household to further understand their asset status.
Q3 Whose name(s) are in the red book? Identify household's asset variable
2 - Land 2 Q3 What is the total area of the plot? Identify Land's are variable
5H - Household income Q10 Total Net Income Identify Household's income
7C - Saving Q3 What was the money value of this saving/asset 12 months ago ? Identify household's asset variable
8A - Credit 1 Q1 Has your household borrowed money or goods
(including seeds or fertilizer) from any source
Identify Credit demand of household (have demand or not)
Q3 Which member(s) of the household applied for the loan?
Identify borrowers and their characteristics that may affect to credit rationing possibility (age, education, …)
In Q4, the individual applied for a specific amount of credit, but it is essential to identify cases of partial credit rationing where the amount received was less than the requested credit In Q5, it is necessary to determine the actual amount received by the individual, whether in cash or cash equivalents.
Identify Partial credit rationing cases (amount of credit receive is less than the amount Q9 From which institution or individual was the loan obtained?
Identify type of credit institution that household access to
The purpose of the loan was clearly stated, and the household had to offer assets as collateral, which raises questions about the type and total value of these assets Additionally, if a guarantor was involved, understanding their relationship to the household member is crucial The creditworthiness of the individual responsible for the loan is highlighted by any instances of missed payments, while the frequency of loan rejections since July 1, 2006, indicates potential credit rationing It's essential to identify the main reasons behind these rejections, as this information may determine whether further regression analysis is needed to explore additional variables.
Does any member of your household hold any office or other positions of public responsibility in the Commune, or higher levels of
10A - Groups1 Q1 Are you a member of any groups, organizations, or associations? Identify social capital variable
Issue of Data Bias (Sample Selection Problem)
Statistically, the sample collection design should ensure the objective representation of the larger population’s characteristics; otherwise the inferences extracted from the sample may not valid for the population.
In certain situations, the sample may not accurately represent the population due to selection biases, which occur when a sample is systematically chosen based on specific criteria rather than through random selection of the dependent variable.
The VARHS-2008 data set presents selection issues due to its dependent variable, which measures credit rationing cases that were not randomly selected from the overall population Specifically, partial credit rationing households are only identifiable among observations that reported having credit, leaving those without credit unidentifiable Consequently, the population is effectively divided into two distinct groups—households with credit and those without—resulting in a sample that is representative of only one group rather than the entire population.
(991 observations) No Credit Access (Credit Balance = 0)
(872 observations) No Condition Access to Credit
Observed Explanatory Variables Un-Observed
Figure 5 - Sample Distribution Source: Author Calculation from VARHS 2008
The study focused on 991 households that received credit, collecting data on interest rates, credit institutions, loan sizes, loan purposes, collateral, and creditworthiness In contrast, information from the 1,146 households that did not receive credit was not included Consequently, when conducting the regression analysis to assess the impact of factors like interest rates and loan sizes on partial credit rationing, only the 991 households were considered, excluding the remaining 1,146 This selective analysis risks bias in conclusions drawn about the entire population of 2,137 households, as the findings are based solely on the subset that received credit (Heckman, 1979).
Heckman Two-Stages Model
This section discusses Heckman’s (1979) theory on the sample selection problem and his proposed solution, particularly applicable when the dependent variable in the outcome equation is continuous, while the selection equation is dichotomous The model will be used to analyze the factors influencing the degree of partial credit rationing.
3.3.1 Sample Selection Bias vs Omitted Variables Bias
Heckman (1979) asserts that using a sample in a regression model that only represents specific groups, rather than the entire population, can introduce biases in the estimation results This sample selection issue may result in erroneous inferences about the broader population.
In detail the logic between sample selection and bias estimation result can be demonstrated as follow:
Assume the Degree of Partial Credit Rationing Function as:
Where � 1� is the continuous variable, defined as the difference between loan size received and loan size applied of households (i.e loan size apply – loan size received) If the value of
Households that received partial credit during the years 2006 to 2008, as indicated by the VARHS 2008 dataset, experienced more extreme financial conditions The analysis shows that a larger value of the variable is associated with a higher degree of credit rationing for these households.
� 1� is a vector of observed variables relating to the i th household’s characteristics such as credit institutions, loan size, loan purposes etc.
Assume the Credit Access Function as:
� ∗ is the amount of credit that the household got in the year 2006 and 2008 If the loan amount larger than 0 (� 2� >0), then � 2� =1, and � 2� =0 otherwise.
� 2� is a vector of variables that supposed to determine the loan size of household, such ashousehold size, house size, household shock, land value, livestock value, etc.
That is the two error term – ε and u are both normally distributed with zero means, constant variances (σ 2 ε, σ 2 ), and they are correlated with the correlation coefficient as ρεu.
The population regression function for the equation (1) can be presented as:
While the regression function for the subsample of available data, i.e the data of � 1 � and � 1� that are only observed when � ∗ >0
As h(ε, u) is a bivariate normal density, using well known results
Where � and Φ are, respectively, the density and distribution function for a standard normal variable, and
The inverse of Mill's ratio represents the relationship between the probability density function and the cumulative distribution function of a distribution This concept allows for the development of a comprehensive statistical model for normal population disturbances Consequently, the conditional regression function for selected samples can be articulated effectively.
Compare with equation (1), the problem of regression under sample selection condition can be treated as omitted variable; specifically the regression has ignored the effect of � ��
Omitted variable biases can lead to an underestimation of the population variance of the coefficients for the repressors, resulting in an inflated significance level in the estimation process.
Sample selection bias is mathematically addressed as an omitted variable problem To tackle this issue, Heckman suggested deriving the inverse Mills ratio from equation (2) and incorporating it into equation (1) to account for the omitted regressor.
To extract the inverse Mill ratio, Heckman conducted the estimation as follow:
)), we have the reduced form of � � :
And the function can be expressed shortly as:
So, if one knew � � and hence � � , one can enter � � as a regressor in the equation:
� 1 � = �(� 1� | � 1� , � ∗ > 0) + � in which � is an error term.
However, as one does not know � 2 , the value of � � is also unknown Heckman (1979) suggested that:
To estimate the value of \( \beta_2 \) using probit regression, one can leverage the known values of \( \beta_1 \) and \( \beta_2 \) in the equation \( y^* = \beta_1 + \beta_2 \cdot X + \epsilon \), where \( y^* = 1 \) if \( y^* > 0 \) and \( y^* = 0 \) otherwise By utilizing the estimator \( \hat{\beta}_2 \), it becomes possible to further estimate \( \hat{y} \) and subsequently compute \( \hat{p} \), the estimator of the probability.
Secondly, estimate � 1 and � �� by OLS method basing on the data of � 1� , � 1� , and �̂ 3.3.3 Application to study & Model Specification
Bivariate Probit with Sample Selection Model
When both dependent variables in the outcome and selection equations are dichotomous, the bivariate probit with sample selection model is the suitable choice This model addresses the issue of non-random samples, similar to Heckman’s sample selection model However, unlike Heckman’s approach, which employs one probit regression for the selection equation and one OLS regression for the outcome equation, the bivariate probit with sample selection utilizes probit models for both equations.
2001) The model is applied to examine the determinants on probability of partial credit rationing.
The model of bivariate probit with sample selection can be illustrated as follow:
The latent variable \( \ast \) represents the difference between the loan amount applied for and the loan amount received, indicating the degree of partial credit rationing experienced by borrowers The vector \( \mathbf{X} \) includes explanatory variables related to \( \ast \), while \( \mathbf{\beta} \) denotes the vector of explanatory parameters Additionally, \( \epsilon \) represents the error term in this context.
That means cases of partial credit ration (� 1 � = 1) are identified if the value of loan size applied
– loan size received larger than 0
Where � ∗ is the latent variables represents for loan amount that credit borrowers got �is the vector of explanatory variable for � ∗ , � is vector of explanatory parameters, and � is error term.
That means cases of credit access (� 2 � = 1) are identified if loan amount borrowers got larger than 0
Assume: � 1� , and � 2� are independent and identical standard normally distributed, i.e.
� 1� , � 2� ~ �� (0,0,1,1)� , and ����(� 1� , � 2� ) = �, their joint probability distribution function
Then, their joint cumulative distribution function (cdf) is
In case of � 1 and � 2 are independent – that is ����(� 1� , � 2� ) = 0, the estimation of
� 1 can be done by employing conventional probit model for ��(� 1 � = 1|� 1� ).
However, in case of sample selection, there is likely to exist a relationship between two error terms, i.e ����(� 1� , � 2� ) = � ≠ 0; and the estimation of � 1 cannot simply base on
The probabilities for three types of observations in a sample are as follow:
In which �(∙) is the univariate standard normal cumulative distribution function, and �(∙ , ∙ , �) denotes the bivariate standard normal cumulative distribution function with correlation �
Given the error terms are independent and identically distributed as a standard bivariate normal with correlation �, i.e � 1� , � 2� ~ φ(0, 0,1,1, �), the probability of ��(� 1 � = 1|� 1� ,
� 2 � = 1) can be stated as follow:
In order to estimate the vector of model parameters is � = (� 1 , � 2 , �), the Maximum likelihood method is applied, in which the following sample log-likelihood derived from (6) is maximized:
3.4.2 Application to study & Model Specification
� 1� takes value of 1 or 0, � 1� = 1 if the household i th is identified as partially credit rationed, otherwise � 1 � = 0
� 2 � takes value 1 or 0, � 2 � = 1 if the household i th is identified as having accessed to credit, otherwise � 2 � = 0
� 1 is a vector of explanatory variables for the probability of credit access, including:
Ethnicity, household size, head age, head gender, head education, number adult, dependency ratio, house size, land value, livestock value, income, shock, social position.
The probability of receiving partial credit is influenced by various explanatory variables, including household size, the age and gender of the household head, and their level of education Additionally, factors such as the dependency ratio, house size, land value, livestock value, income, and the size and purpose of the loan—whether for production, consumption, or investment—play a crucial role Other important considerations include the value of collateral, the social position of the household, and the credit institution involved.
Multicollinearity Test
Multicollinearity poses a significant issue in regression analysis, as it can lead to biased results due to high correlations among explanatory variables This may result in inflated variances, unstable parameter estimates, and biased coefficient signs (Menard, 2002) Therefore, conducting a multicollinearity test is crucial and should be prioritized as an initial step in multiple regression analysis (Mansfield & Helms, 1982).
This study addresses the detection of multicollinearity through a correlation matrix, which illustrates the correlation values among independent variables Correlation values range from 0 to 1, with a value exceeding 0.8 indicating potential multicollinearity To resolve this issue, it is recommended to eliminate one of the two highly correlated variables (Grewal, Cole & Baumgartner, 2004).
Ethnicity Number Adults Head Age Head Gender
Head Education Dependency Degree of Partial Credit
Loan Purposes Collateral Value Loan Size Demand Credit Institution
Livestock Value Household Size Income Household Shock
Analytical Framework
Ethnicity – dummy variable that takes value 1 in case of household i th is Kinh ethnicity and 0 otherwise
Household Size – variable records the number of members in a household
Household Head Age – variable records the age of households’ head
Household Head Gender – dummy variable identifies the households’ head gender The variable takes value of 1 if household head is male and 0 otherwise.
Household Head Education – variable records the education level of households’ head, ranging from lowest level 1 to the highest level 6
Number of Adult – variable records the number of adult member in a household The adult member is identified if household member is older than 18.
Dependency Ratio – ratio of number of dependent member over household size The dependent member is identified if household member age is younger than 18 or older than 70.
House Size – variable records the total house area that the household owns in square meter for the year 2006.
Total Land Value – variable records the total value of land that the household owns The value was the price that the household willing to sell at.
Total Live Stock value – variable records the total value of livestock that the household raised The value was the price the household willing to sell at.
Income – variable records the total net income of the household.
Loan Size Apply – variable records the loan amount that the household requested.
When applying for a loan, households typically state one of four main purposes: production purposes, which encompass loans for agricultural activities such as rice cultivation, livestock, forestry, and non-farm activities; investment purposes, including loans for purchasing or building homes, acquiring land, or buying other assets; consumption purposes, which cover loans for significant life events like weddings and funerals, educational expenses, healthcare costs, and general consumption; and other purposes, which may involve loan repayments or miscellaneous needs.
Total Collateral Value – variable record for the value of the collateral
A credit institution is defined as a dummy variable that distinguishes between formal and informal credit sources Formal institutions include social policy banks, agricultural and rural development banks, state-owned banks, local authorities, private banks, and various credit associations such as farmer unions, veterans unions, women's unions, and people’s credit funds In contrast, informal credit sources encompass private traders, money lenders, friends or relatives, informal credit schemes, and other unregulated lending options.
1 if the credit institution that the household apply the loan to is formal, and 0 if informal.
The social position is represented as a dummy variable that indicates whether a household has any members employed by the government If at least one member of the household works for the government, this variable is assigned a value of 1; otherwise, it takes a value of 0.
Household shock – variable that indicates whether household suffered any shock (such as economic, production, illness, etc.)
Hypothesis
3.7.1 Hypothesis for the probability and degree of partial credit rationing
The age of a household head can significantly influence the likelihood of being partially credit rationed Research by Chaudhur and Cherical (2012) indicates a positive correlation between age and credit approval chances, suggesting that older individuals are often viewed as more mature and trustworthy by lenders Conversely, Pham and Izumida (2002) found a negative relationship, indicating that some lenders may perceive younger borrowers as having greater income-generating potential to repay debts This dual perspective highlights the complex role age plays in credit assessment.
The education level of a household head is inversely related to the likelihood of experiencing partial credit rationing, as noted in studies by Petric (2003), Chaudhuri & Cherical (2012), and Pham & Izumida (2002) This suggests that households led by well-educated individuals are less likely to face restrictions from credit lenders.
Household size can significantly influence the likelihood of partial credit rationing On one hand, larger households may face higher costs, which could adversely affect their ability to repay loans (Arene, 1992) Conversely, larger families may benefit from increased labor availability, potentially enhancing their income-generating capacity (Vuong et al., 2012) Thus, the relationship between household size and loan repayment ability is complex and multifaceted.
The dependency ratio, which reflects the number of dependents in a household, is anticipated to have a positive correlation with economic burdens As the number of dependents increases, households face greater financial responsibilities, making it less likely for them to fully repay their loans (Pham & Izumida, 2002).
Household head gender: household head with male gender are expected to have reversed relationship with the likelihood of being credit rationed (Petric, 2003)
Household reputation and social status significantly influence the likelihood of credit rationing, with a negative correlation observed between the two A strong reputation and esteemed position within the community enhance perceived creditworthiness, making loan approval more likely (Pham & Izumida, 2002; Ali & Deininger, 2012).
Total house size, asset value, land value, and livestock value are negatively correlated, indicating that higher values enhance the quality of collateral required by credit institutions to mitigate credit default risk These factors are crucial for lenders as they help address asymmetric information issues, allowing for better loan monitoring and retrieval Research in the rural credit market of India has confirmed the significant impact of these factors on credit rationing probability, highlighting their importance in effective lending practices (Chaudhuri & Cherical, 2012; Barslund & Tarp, 2008; Ping, Heidhues & Zeller, 2010; Aguilera, 1990; Ali & Deininger, 2012; Okurut, 2005).
Income: the higher income, the less likely being credit rationed as higher income household’s cash flow is more stable, promisingly credit payback (Ping, Heidhues & Zeller, 2010)
Collateral value: the higher the value of collateral, the less probability of credit rationed.
As collateral is a tool of credit lender to enforce credit borrowers to pay back their debt. Higher collateral value means higher level of enforcement (Bester, 1987).
The size of the loan applied for is positively correlated with the likelihood of experiencing credit rationing As the loan amount increases, the risk of non-repayment also rises, making lenders more cautious Additionally, constraints in credit supply can further restrict the total amount of credit available for disbursement.
Loans intended for production and investment are likely to reduce the chances of credit rationing, as these loans can lead to income generation, enabling households to maintain positive cash flow for debt repayment Conversely, loans taken for consumption purposes may increase the likelihood of credit rationing, since such loans are typically expended without guaranteeing sufficient cash flow for repayment.
Credit institutions: Formal Credit sectors are supposed to credit rationed than informal credit sector due to their strictness in loan approval procedures, monitors and loan enforcement.
3.7.2 Hypothesis for the probability of access to credit
The age and education level of a household head positively influence the likelihood of accessing credit Older and more educated individuals tend to have a greater demand for credit to grow their businesses, as noted by Tang et al (2010) Additionally, these mature and well-educated individuals are often perceived as more creditworthy by lenders, as supported by Barslund & Tarp (2008), Zeller (1994), and Kereta (2007).
The number of dependents in a household can influence credit market dynamics in both positive and negative ways On the demand side, households with more dependents face increased economic burdens, making them more likely to seek credit to meet their liquidity needs (Pham & Izumida, 2002; Okurut et al., 2005) Conversely, on the supply side, lenders may be hesitant to extend credit to households with a high dependent ratio, as these families often struggle with poverty and have lower repayment capabilities (Pham & Izumida, 2002).
The number of adults and household size positively influences the likelihood of credit assessment, as larger households typically have a greater demand for financial resources to support activities like business ventures, education, and farming (Barslund & Tarp, 2008) Additionally, with more income-generating members, families enhance their creditworthiness in the eyes of lenders.
Households identified as Kinh in Vietnam tend to have easier access to credit compared to other ethnic groups This advantage is attributed to the Kinh community's prevalence and shared economic and business practices, which facilitate their interactions within the financial system (Vuong et al., 2012).
Male household heads are typically expected to borrow more due to their role as the primary financial providers in the family, engaging in various activities that require funding In contrast, women often face credit rationing, as they are viewed primarily as responsible for domestic duties rather than income-generating activities, leading to a perception of lower credit repayment ability compared to men (Petric, 2003).
Reputation and social status significantly influence household access to credit, as individuals with higher social standing tend to demand more credit and experience lower rates of credit rationing.
Households experiencing shocks, such as natural disasters, economic downturns, illness, or unemployment, often seek loans in the credit market to meet their liquidity needs As a result, these households are more likely to access credit compared to those not facing such challenges (Zeller, 1994).
Model Specification
No Variables Variable Denote Credit Access (Yes) Partial Credit Ration
Degree of Partial Credit Ration
5 Head Gender (Male) head_gend1 + - -
6 Number of Adult num_adult +
9 Total Land Value total_landvalue + - -
10 Total Livestock Value total_livestock_value + - -
12 Loan Size Apply loansize_apply + +
13 Loan Purpose - Production loan_purpose_prod - -
14 Loan Purpose - Investment loan_purpose_invest - -
15 Loan Purpose - Consumption loan_purpose_consump + +
17 Credit Institution (Formal) credit_inst_dummy + +
18 Social Position (Yes) social_position1 + - -
Results and Discussion
Characteristics of Borrowers by Credit Rationing
This section generally examines the correlation between the dependent variable and explanatory variables to provide an intuitive relationship between them.
The under-explored dependent variables in this study are "Credit Access" and "Partial Credit Ration." Credit access indicates whether households obtained credit between 2006 and 2008, with a positive credit balance signifying access and a zero balance indicating a lack of access Meanwhile, Partial Credit Ration describes instances where households received loans that were less than the amounts they requested.
Overview of Credit Rationing Situation
Credit Access & Partial Credit Ration
Figure 7 - Credit Access & Credit Ration Source: Author’s calculation from VARHS 2008
A recent analysis reveals that 56% of households maintain a positive credit balance, while 44% have no credit balance at all Among households with credit access, 11.24% experience partial rationing Notably, there are no instances of partial rationing reported for those without credit access.
Credit Access and Household Head’s Age
Figure 8 - Credit Access & Household Head Age Source: Author’s calculation from VARHS 2008
The age distribution of household heads reveals that the most significant segment is the 41-50 years age group, comprising 30.9% of the sample Following this, 23.9% belong to the 31-40 age group, while 19.2% are aged 51-60 Additionally, 17.8% of household heads are over 60 years old, and the youngest group, those under 30, accounts for 8.3% of the total sample.
The analysis of credit access reveals that the highest percentage of individuals obtaining credit is within the 41-50 age group, at 62.3% This is followed by the 31-40 age group, which has a credit access rate of 47.8%.
R at e o f G et tin g C re di t b y A g e G ro up ( % ) 20 4 0 6 0 0
51-60 age group with 38.7.2% and age group older than 60 with 34.6%, the lowest one is age group younger than 30 (16.6%).
Younger individuals under 30 are less likely to take out credit, while older adults tend to utilize it more frequently Those in the middle age bracket of 31 to 50 years are the most active in accessing credit, but this trend declines as household heads age, particularly for those over 51 years old.
Figure 9 - Household Head Age & Credit Ration Source: Author’s calculation from VARHS 2008
The relationship between head age and the rate of partial ration remains unclear, with the rate consistently hovering around 10% across different age groups Specifically, the rate is 10.3% for individuals under 30, 9.02% for those aged 31-40, peaking at 12.9% for the 41-50 age group, followed closely by 12.7% for ages 51-60, and dropping to 9.22% for those over 60.
Credit Access and Household Head Education Level
Credit Access & Head's Education Level
Structure Education Level Got Credit (%)
Rate of Credit Access by Education Level (%)
Figure 10 - Credit Access & Household Head Education Level Source: Author’s calculation from VARHS 2008
Education level is decoded as following:
Level 1 – No diploma, level 2- Short-term vocational training, level 3 - Long-term vocational training, level 4 - Professional high school, level 5 - Junior college diploma, and level 6 - Bachelor degree Most of the household head education is low educated with education level 1 takes 93.2% proportion of the sample.
The highest credit access rate is observed among individuals with education level 5, reaching 60%, indicating a strong demand for credit among diploma holders This group is often favored by lenders due to their well-developed productive skills and training In contrast, those with education level 6 experience the lowest credit access rate at 23.5%, suggesting potential barriers to credit availability for this demographic.
Individuals with a university education or higher tend to rely less on credit compared to those with lower education levels, where credit reliance remains relatively consistent, hovering around 45%.
Head Education Level & Partial Ration
Figure 11 - Household Head Education Level & Credit Ration Source: Author’s calculation from VARHS 2008
The relationship between education and partial rationing remains ambiguous Notably, the highest rates of partial rationing, calculated as the percentage of partial ration observations relative to total observations at each education level, occur at education level 4, with a rate of 25.9%.
The participation rates in the credit market reveal that levels 5 and 3 have a 0% rate of partial rationing, indicating that their low participation may skew the true reflection of the situation In contrast, education levels 1 and 2 show low partial ration rates of 11% and 10%, respectively.
Credit Access and Loan Purposes
Rice Other crop production (including inputs)
Animal husbandry, forestry, and fishery are essential components of sustainable agriculture, contributing to food security and economic stability Engaging in non-farm activities can diversify income sources, while repaying other loans is crucial for maintaining financial health Investing in housing, land, and additional assets supports long-term wealth accumulation Furthermore, managing expenses related to weddings, funerals, education, and health is vital for family well-being Overall, effective financial planning encompasses general consumption and other specified needs, ensuring a balanced approach to both personal and economic growth.
Rate of Credit Access by Loan Purposes (%)
Figure 12 - Credit Access & Loan Purposes Source: Author’s calculation from VARHS 2008
The allocation of credit varies among different loan purposes The survey conducted in VARHS
In 2008, a significant majority of households utilized credit for agricultural production, with animal husbandry accounting for the largest share at 38.9% of the total examined households Crop production, excluding rice, followed at 15.7%, while rice cultivation represented 11.2% In contrast, credit access for general consumption and expenses such as weddings and funerals was minimal, comprising only 1.42%, 0.755%, and 1.04%, respectively Overall, the data indicates that the majority of credit was directed towards productive purposes, with only a small fraction allocated for non-productive uses.
Other crop production (including inputs) 15.1
Rate of Partial Ration by Loan Purposes (%)
Figure 13 - Loan Purposes & Credit Ration Source: Author’s calculation from VARHS 2008
The relationship between partial rationing and loan purposes reveals a significant disparity, with a substantial portion of partially rationed households borrowing for non-income generating reasons, such as weddings and funerals (37.5%), repaying other loans (36.4%), health expenses (22.2%), education costs (19.7%), and general consumption (13.3%) Conversely, borrowers seeking funds for productive purposes experience lower rates of credit rationing, averaging around 11% Specifically, credit rationing rates are 10% and 14.8% for fishery and non-farm activities, 5.04% for rice cultivation, 15.1% for other crop production, 9.95% for animal husbandry, and 0% for forestry Overall, loans intended for productive uses have a significantly higher likelihood of avoiding credit rationing compared to those for non-productive purposes.
Bank of Agriculture and Rural Development 27.3
Other State-owned commercial Bank 1.51
People's Credit Funds 1.98 Other credit associations 661
Private Money Lender (interest-free loans) 2.55
Informal credit scheme (including Roscas) 378
Percent Credit Access by Credit Institution 20 30
Figure 14 - Credit Access & Credit Institutions Source: Author’s calculation from VARHS 2008
In the examined market, the formal credit sector dominates, accounting for approximately 75% of borrowing households Notably, 59% of surveyed households obtained credit from two major state-owned banks, with the Social Policy Bank providing 31.5% and the Bank of Agriculture and Rural Development contributing 27.3% Other formal credit organizations, such as the Farmers Union and Women's Union, hold smaller market shares of 3.87% and 6.7%, respectively The remaining formal organizations, including private banks and various credit associations, have minimal market presence, with shares ranging from 0.567% to 1.98%.
Determinants of Credit Accessibility
Table 3 - Determinants of Credit Accessibility
Selection Equation: Dependent Variable - Credit Access = 1
Household Head Education Level -0.045 Household Head Gender (Male =1) 0.007
Total Land Value (VND1000) 0.000 * Total Livestock Value (VND1000) 0.000 Household Income (VND1000) 0.000 Household Shock (Shock=1) 0.152 ***
At least 1 Member working for Government=1 0.272 ***
Source: Author calculation from VARHS 2008 Note: *, **, *** denote 10%, 5%, and 1% level of significance, respectively
The first stage of Heckman two stages regression reveals results of the selection equation, ie - the probit estimation of the credit access probability.
Ethnicity significantly influences the likelihood of household credit access, with Kinh households demonstrating a higher probability of obtaining credit in the studied regions, confirmed at a 1% significance level This finding aligns with expectations regarding the relationship between ethnicity and credit access, yet it contradicts the results presented by Vuong et al (2012).
A larger household size significantly increases the likelihood of accessing credit, as indicated by a positive correlation found at the 5% level This suggests that households with more members have a higher probability of obtaining credit.
The age of the household head has a significant negative impact on credit accessibility, with a 5% correlation This suggests that as the age of the household head increases, the likelihood of accessing credit decreases This finding contrasts with the research by Chen and Chiivakkul (2008), which indicated that older individuals were preferred by credit lenders due to their assumed higher repayment ability However, Tang et al (2010) suggest that older adults may require less credit because they tend to rely more on their savings.
The study reveals a significant negative correlation between dependency ratio and credit accessibility, with a 1% confidence level This indicates that households with a higher number of dependents relative to working adults are less likely to access credit Credit providers may hesitate to lend to these households, as they often have limited financial resources and lower repayment capabilities (Pham & Izumida, 2002).
The total land value positively correlates with credit access for households, aligning with expectations, although this relationship is only weakly confirmed at a 10% confidence level This suggests that a higher fixed asset value, particularly land, increases the likelihood of obtaining credit These findings are consistent with previous research by Pham and Izumida (2002), Zeller (1994), and Aghion and Morduch (2004).
The findings indicate a significant result at the 1% confidence level regarding household shocks The positive coefficient suggests that households experiencing economic shocks—such as unemployment, poor harvests, disasters, or personal issues like illness—are more likely to depend on credit as a financial resource to navigate these unforeseen challenges (Zeller, 1994).
Social position significantly influences access to credit, as households with members employed in government or local authorities have a higher likelihood of obtaining loans This finding aligns with previous research by Zeller (1994) and Pham and Izumida (2002).
There is insufficient evidence to establish a significant relationship between factors such as the number of adult members, the education level of the household head, the gender of the household head, and the value of livestock on credit accessibility.
Determinants of Partial Credit Rationing Probability
Table 4 - Determinants of Partial Credit Rationing Probability
Probit model with sample selection Number of Observations 2390
Outcome equation: Dependent Variable - Loan Size Apply & Loan Size Received Differences
Household Size -0.015 Household Head 's Age -0.003 Dependency Ratio -0.092 Household Head Education Level 0.022 Household Head Gender (Male =1) -0.087
House Size -0.003 * Total Land Value (VND1000) 0.000 Total Livestock Value (VND1000) 0.000 Household Income (VND1000) 0.000
Loan Size Applied 0.000 * Loan Purposes (Production =1) -0.239 Loan Purposes (Consumption =1) 0.082 Loan Purposes (Investment =1) -0.366 *
At least 1 Member working for Government=1 -0.324 *
Selection Equation: Dependent Variable - Credit Access = 1
Household Head Education Level -0.045 Household Head Gender (Male =1) 0.007
Total Land Value (VND1000) 0.000 * Total Livestock Value (VND1000) 0.000 Household Income (VND1000) 0.000 Household Shock (Shock=1) 0.152 ***
At least 1 Member working for Government=1 0.272 ***
LR test of indep eqns (Chi2) 11.79 ***
Source: Author calculation from VARHS 2008 Note: *, **, *** denote 10%, 5%, and 1% level of significance, respectively
This article examines various factors influencing a household's likelihood of being credit rationed It highlights not only the determinants of credit accessibility but also specific variables related to credit rationing, including characteristics of credit institutions and loan contracts The regression analysis reveals statistically significant results associated with factors such as household size, loan amount requested, social status, investment loans, and the type of credit institution involved.
Households that own larger homes are less likely to experience partial credit rationing compared to those residing in urban areas This relationship is statistically significant at the 10% level, aligning with the findings and assumptions presented by Ali and Deininger.
Secondly, loan size requested has a positive relationship with probability of partial credit ration.
It shows a weakly statistical significant level at 10% This result is in consistent with presumption as well as the evidence found by Pham and Izumida (2002).
The regression analysis indicates that a household's social position significantly influences the likelihood of being credit rationed, particularly at a 10% confidence level Specifically, households with at least one member employed in government, public office, or local authorities are less likely to receive the full loan amount requested This finding aligns with previous research by Pham and Izumida (2002) and Ali and Deininger (2012), supporting the initial hypothesis.
A study found that loans intended for investment in properties, such as purchasing or constructing houses and acquiring land, are negatively correlated with the likelihood of experiencing partial credit rationing, with results significant at the 10% confidence level This indicates that borrowers seeking loans for investment purposes are more likely to receive the full amount requested, supporting the conclusions drawn by Pham and Izumida (2002).
Formal credit institutions exhibited a significant positive correlation with credit rationing in the studied regions, indicating that state-owned banks, private banks, and unions are more likely to deny borrowers the full amount of their loan requests This trend suggests that these institutions maintain stringent loan approval criteria.
The log-likelihood test for independence between the outcome equations and the selection equation does not reject the null hypothesis at a 1% confidence level, suggesting that the two equations are independent This implies that neglecting the selection equation may lead to biased results due to potential sample selection issues.
Wald test also fails to reject the null hypothesis of all factor simultaneously equal 0 at 10% confident level.
Determinants of Partial Credit Rationing Degree
Statistical analysis indicates that the degree of partial credit rationing is significantly influenced by household characteristics, including household size, head age, and dependency ratio, which are statistically significant at confidence levels of 10% and 5%.
Household size is considered an ambiguous factor, potentially influencing loan constraints in both positive and negative ways Regression analysis indicates a positive correlation between household size and the degree of partial credit rationing, suggesting that larger households experience greater limitations on loan amounts This finding aligns with the results of Arene (1992).
Head age is an ambiguous factor influencing the extent of partial credit rationing, with findings indicating a negative correlation; older household heads tend to have a lower demand for unmet loans This observation aligns with Kereta (2007), which noted that both younger and older individuals in Ethiopia experience greater credit constraints compared to those of middle age This trend may be attributed to the perception of higher creditworthiness among older household heads by lenders.
It also reveals that dependency ratio negative correlated with magnitude of partial credit ration.
The study reveals that households with a higher number of dependent members tend to receive less credit than they request, aligning with the findings of Pham and Izumida (2002).
The regression analysis reveals that collateral significantly influences the degree of partial credit rationing, with house size and collateral value showing a strong negative correlation with the amount of loan credit rationed at 1% and 10% significance levels, respectively Additionally, while livestock value also negatively impacts partial credit rationing, it lacks statistical significance These findings align with the initial hypotheses and corroborate previous research, including studies by Petric.
Research indicates that household income has a negative correlation with the level of partial credit rationing, although this relationship is not statistically significant This suggests that lenders are less likely to impose restrictions on the loan amounts requested by higher-income households These findings align with the results presented by Ping, Heidhues, and Zeller (2010).
Table 5 - Determinants of Partial Credit Rationing Degree
Heckman selection model two-step estimates Number of Observations 2390
Outcome equation: Dependent Variable - Loan Size Apply & Loan Size Received Differences
Household Size 225.595 * Household Head 's Age -46.018 **
Household Head Education Level -232.529 Household Head Gender (Male =1) -258.764
Total Land Value (VND1000) 0.001 Total Livestock Value (VND1000) -0.006 Household Income (VND1000) -0.009
Loan Purposes (Production =1) 173.144 Loan Purposes (Consumption =1) 1736.836 * Loan Purposes (Investment =1) -1620.190
At least 1 Member working for Government=1 110.808
Selection Equation: Dependent Variable - Credit Access = 1
Household Head Education Level -0.048 Household Head Gender (Male =1) 0.002
Total Land Value (VND1000) 0.000 * Total Livestock Value (VND1000) 0.000 Household Income (VND1000) 0.000 Household Shock (Shock=1) 0.186 ***
At least 1 Member working for Government=1 0.276 ***
Source: Author calculation from VARHS 2008 Note: *, **, *** denote 10%, 5%, and 1% level of significance, respectively
Research indicates a strong positive correlation between the size of the loan requested by households and the extent of credit rationing they experience, with findings being statistically significant at the 1% level This suggests that as borrowers seek larger loans, the disparity between their requested loan amount and the actual amount received also increases This outcome aligns with the findings of Pham and Izumida (2002).
Research indicates that borrowers seeking loans for consumption purposes experience a greater disparity between the amount requested and the amount received, with a significant positive correlation at the 1% level, aligning with Diagne's (1999) findings Conversely, loans intended for productivity purposes show a positive relationship with partial rationing, although this correlation lacks statistical significance.
Wald test also fails to reject the null hypothesis of all factor simultaneously equal 0 at 1% confident level (Chi 2 = 0.0000).
The significant level of Mill ratio is 1%, indicating that there is a potential sample selection bias unless the selection equation is not taken into account.
Multicollinearity Test
The correlation matrix indicates that none of the explanatory variables exhibit a correlation value exceeding 0.7 Since the threshold for identifying multicollinearity is a correlation value above 0.8, multicollinearity is deemed an insignificant concern for the three regression models discussed.