Do Financial Linkages Ease the Credit Rationing of Forest Rights Mortgage Loans? Evidence from Farm Households in Fujian Province, China

Abstract

Affected by the small scale of forest farmers’ land and the imperfect development of the forest land transfer market, China’s forest rights mortgage loans have suffered from more serious credit rationing. The application of financial linkage theoretically has the effect of solving credit rationing. However, previous studies on financial linkages have focused on the field of credit lending, whereas the applied studies in the field of mortgage lending are mainly case studies on transaction models, and empirical tests on the application of financial linkages in mortgage lending are lacking. Therefore, to fill this gap, this study analyzed the effect of financial linkage on credit rationing for forestry mortgage loans and the mechanism of action through a study of farmers in Fujian, one of the key collective forestry areas in China, using the PSM method with 785 sample values. The results show the following: (1) Financial linkages have a significant mitigating effect on supply rationing, demand rationing, and the total rationing of forest rights mortgage loans. Compared to non-participation, financial linkages significantly reduced total rationing by 15.2%, with a 5% reduction in supply rationing and a 10.2% reduction in demand rationing. (2) The impact of financial linkage differs significantly among heterogeneous farmers. It has a significant mitigating effect on supply rationing for small-scale farmers, but not for large-scale farmers; meanwhile, it better mitigates demand rationing for large-scale farmers than for small-scale farmers. (3) In the mechanistic test, financial linkages were found to moderate the relationship between the value of collateral and supply rationing for small-scale farmers, and transaction costs play an intermediary role in the relationship between financial linkages and demand-based rationing. According to the study results, in order to promote the development of forest rights mortgage loans, it is necessary to develop different lending strategies for heterogeneous farmers, to further encourage small- and medium-sized farmers to participate in financial linkages, as well as to further reduce the related transaction costs.

1. Introduction

Forest rights mortgages are an important mechanism for realizing the value of ecological products, as well as an important policy in solving the shortage of rural collateral and alleviating rural credit constraints [1]. Since 2004, China has promoted reform of the forest rights system in southern collective forest areas with the goal of clear property rights, including beginning work on forest land titling, enhancing the legal validity and security of forest land and wood property rights to a certain extent through the issuance of forest rights certificates, and beginning work on forest rights mortgage loans based on clear property rights. Mortgage loans cause forest land use rights and forest ownership rights to become effective collateral in the financial market, thereby making it possible to turn forestry assets into forestry capital. Over more than a decade of development, forest rights mortgages have played an important role in opening up rural investment and financing channels, realizing the value of ecological products, easing some farmers’ credit constraints, and promoting income increase. However, forest rights mortgage loans have also encountered greater practical difficulties in their development. Numerous studies have shown that the accessibility of forest rights mortgage loans is low, and most farmers, especially small-scale, suffer from more serious credit rationing [2,3].

Credit rationing is the result of credit decision-making between the fund supplier and the fund demander, which includes supply rationing and demand rationing. Firstly, supply rationing is a result of decision-making by the supplier, i.e., financial institutions. For the sake of profit and risk, financial institutions will put forward strict additional loan conditions, which means that borrowers who fail to meet the conditions are refused loans or have them given up automatically by banks, resulting in an unbalanced phenomenon where supply is less than demand [4,5,6]. Both the actual situation and studies show that banks’ willingness to supply forest rights mortgage loans is poor. Studies by He et al. (2020) [7], Jin (2017) [8], and Yang et al. (2018) [9] showed that banks put forward harsh contract terms for forest rights mortgage loans and impose a large number of additional conditions on borrowers, including credit control on mortgaged forest-land areas, mortgage rates, loan periods, loan quotas, and additional guarantees, which causes farmers, especially small-scale, to suffer from serious supply rationing. Additionally, it was found that the main reasons for this dilemma are the small area of forest land owned by most farmers [3], difficulties in forest land transfer [10], weak security and stability of forest land ownership [11], and other reasons, which lead to small guarantee functions of collateral and the problems of information asymmetry not being solved well [12,13]. Therefore, high transaction costs and risks cannot be avoided and the bank has to implement rationing for the sake of profit and risk. Secondly, demand rationing is caused by self-decision of the demander. The loan demander chooses to automatically withdraw from the loan application due to the cost–benefit principle, considering the high transaction cost and interest cost of the loan, or worrying about the risk of collateral loss, which leads to the phenomenon of self-rationing [14]. Relevant studies have shown that farmers are not willing to participate in forest rights mortgage loans [15]. The high transaction cost and high interest rate are the main reasons for the lack of farmer willingness to participate [16,17]. Among them, farmers’ lack of financial knowledge, the complex transaction process of forest rights mortgage loans, and the generation of additional transaction costs, such as appraisal and guarantee fees, are the main reasons for high transaction costs [18]. At the same time, the high transaction risk and cost make banks charge high interest rates for this loan business [19]. Thus, the high cost of financing makes farmers less willing to participate. The low willingness of banks to supply, coupled with the low willingness of farmers to demand, has combined to create a situation where forest rights mortgages experience more severe credit rationing.

Therefore, the purpose of financial linkages applied in forestry mortgage loans is mainly to solve the credit rationing dilemma faced by farmers. Financial linkages refer to the cooperation between formal financial institutions and third-party rural intermediaries rooted in rural localities, jointly taking advantage of the lower opportunity cost of capital of formal financial institutions and the information and contract enforcement advantages of rural intermediaries to achieve mutually beneficial cooperation between formal financial institutions and rural intermediaries. This facilitates the credit business expansion of formal financial institutions to rural credit markets and solves the problem of credit rationing for farmers [20,21]. In practice, financial linkage has evolved from the direct transaction mode of “formal financial institutions + farmers” into a transaction mode of “formal financial institutions + rural intermediaries + farmers” with the participation of third-party rural intermediaries [22]. According to literature, the main types of financial linkages that appear in forest rights mortgage transactions include: “formal financial institution + forestry cooperative + farmers” [23], “formal financial institution + joint insurance group + farmers” [24], and “formal financial institution + village authority + farmers” [25]. Mi et al. (2013) [20] and Pagura and Kirsten (2006) [21] showed that financial linkages had an informational advantage over formal financial institutions because the rural intermediaries were rooted in local rural markets and had long-term social and economic ties with farmers. Meng and Wang (2022) [22] and Zhou and Li (2020) [26] showed that financial linkages could use the interpersonal trust network of acquaintance society to create reputational incentives to monitor farmers’ repayment and effectively reduce default rates. In addition, McGuire and Conroy (1997) found that financial linkages had the function of organizing loans and providing guarantees to each other, which also helped to reduce transaction risks and costs for banks and improved farmers’ access to credit [27]. Empirical studies by Teng and He (2009) [28], Shen et al. [29], and Puhazhendi and Badatya (2002) [30] showed the role of financial linkages in the application of credit loans in terms of reducing transaction costs and interest costs, facilitating access to loans, and promoting loan amounts.

Literature combing found the following areas to be inadequate in existing studies. First, the existing research on credit rationing of forest rights mortgage loans mainly focuses on the reasons for credit rationing, while the research on financial linkages in the application of forest rights mortgage loans mainly focuses on transaction pattern analysis of financial linkages and case studies. There is a lack of empirical research on the effect of financial linkages in forest mortgage applications. However, at the practical level, an empirical test on the application effect of financial linkages is important guidance for the formulation of future promotion policies addressing this type of loan business. Second, since financial linkages have been applied to a lesser degree in the field of mortgage loans, research on financial linkages’ impact on credit rationing has mainly focused on the field of micro-credit loans. However, with the emergence of forest rights mortgage loans and agricultural land mortgage loans, financial linkages have also been applied to these two types of loans, so it is necessary to continue expanding the exploration of financial linkage application areas. It is also important to further explore the role of financial linkages in credit rationing because the addition of collateral may cause changes in the mechanism of its role. This paper builds on the existing studies to investigate the effect of financial linkage on credit rationing of forest rights mortgage loans and its mechanism of action, as well as the differences in the effects on heterogeneous farmers. Therefore, this study is important to remedy the shortcomings of the original research, to test the application effect of financial linkage and propose future countermeasures that are beneficial to the development of forest rights mortgage loans. Based on the existing studies, this study investigated the impact of financial linkage on credit rationing of forest rights mortgage loans and its mechanism of action, in an attempt to supplement the existing studies.

The research framework used in this paper is as follows. Based on the literature review, we firstly conducted a theoretical analysis on the mechanism of the effect of financial linkages on supply rationing and demand rationing of forest rights mortgage loans and proposed the research hypothesis. Secondly, we analyzed the impact of financial linkage on credit rationing of forest rights mortgage loans and the difference in impact among heterogeneous farmers by using the propensity score matching (PSM) method. We then tested the relevant mechanism, including on the moderating effect of financial linkages and the mediating effect of transaction costs. Finally, we draw conclusions and propose countermeasures based on the results of the analysis.

2. Theoretical Analysis and Research Hypothesis

2.1. Financial Linkages and Supply Rationing of Forest Rights Mortgage Loans

According to the modern credit rationing theory, it is believed that the main reason for supply rationing is information asymmetry, which causes adverse selection and moral hazard problems. Therefore, the purpose of mortgage loans is to solve the problem of information asymmetry. An effective collateral mechanism is designed so that collateral can serve as an information transfer screening mechanism that distinguishes the risk types of borrowing farmers, in order to mitigate adverse selection due to information asymmetry. The value of collateral can also serve as an incentive mechanism to avoid post-loan moral hazard problems of farmers [31,32]. However, in the context of forest rights system reform, the small size of forest land owned by most farmers and the scattered plots mean that the collateral value of most ordinary farmers does not meet the condition requirements for forest rights mortgage loans. Zhang’s (2015) research found that only 17% of sample households’ forest-land area met the minimum requirements for bank loans [3]. Additionally, due to the high asset specificity of forest rights as collateral, the imperfect development of the forest land transfer market and the low degree of marketization make it more difficult to transfer forest land, which makes the evaluation value of forest assets very low at present and also causes farmers with smaller forest-land areas to suffer from more serious supply rationing [7,10].

Financial linkages alleviate the supply rationing of forest rights mortgage loans, mainly because they can produce a complementary or substitution effect for collateral when farmers face low collateral value or high risk of liquidation. The mechanism of financial linkage itself plays the function of information transmission and incentive to ease the problem of adverse selection and moral hazard, thereby alleviating supply rationing. First, financial linkages can take advantage of the information of rural intermediaries to provide information transmission and information screening, thus solving the problem of adverse selection. According to the theory of financial linkages, financial linkages are based on the introduction of third-party rural intermediaries which are embedded in the rural community and are local organizations based on local, kinship, and business relationships, who have lived in the same area as the borrower for a long time and share the same reputable information network of the rural community [20]. Therefore, in the financial linkage model, banks entrust the search, screening, and selection of pre-loan information to rural intermediaries, and these intermediaries rely on their information advantages to establish an information transmission mechanism between farmers and formal financial institutions, which can serve as a supplement or substitute for the information transfer mechanism of collateral and alleviate adverse selection [33]. Second, financial linkages can use reputation incentives as complementary or alternative mechanisms to collateral to mitigate post-loan moral hazards. In the financial linkage model, the borrower and the rural intermediary have formed reputation capital in various economic transactions and social exchanges for a long time. Once the borrower defaults on credit, the rural intermediary can impose credible penalties on the defaulter by terminating other transactions or social exchanges, so as to achieve the purpose of reputation incentives, and promote the decline in farmers’ default rate; therefore, it can be used as a substitute for collateral value incentive mechanisms to alleviate moral hazard [34,35]. By modeling the moral hazard of borrowers, Hong (2011) demonstrated that rural intermediaries can significantly improve borrowers’ financing terms by being able to use mutual monitoring among members [36].

In addition, financial linkages have the advantages of collateral disposal and guarantee function, which can further reduce the transaction cost and risk of the supply side. Rural intermediaries, due to their endogeneity, have more relational transaction advantages than financial institutions. Under the characteristics of the imperfectly developed forest rights transfer market environment, financial institutions can greatly reduce transaction costs and realization risks by entrusting the collateral disposal to rural intermediaries. The case study of “bank + forestry cooperative + farmers” in Sanming, Fujian Province, by Hong and Fu (2018) proves that forestry cooperatives have the advantage of more flexible disposal of collateral than banks, which can reduce banks’ transaction costs and risks [19].

In summary, it is argued that financial linkages can play a mitigating role in the supply rationing of forestry mortgage loans. Based on this, the following hypotheses are proposed.

Hypothesis 1a.

Financial linkages have a significant negative impact on the supply rationing of forest rights mortgage loans.

Financial linkages play a complementary or alternative role to the collateral function through information advantages and reputation incentives, which means that financial linkages have a moderating effect on the relationship between collateral value and supply rationing. For small-scale farmers with low collateral value, the moderating effect of financial linkages is better in facilitating the relief of supply rationing because of the weak collateral function. However, for large-scale farmers with sufficient collateral value, the moderating effect of financial linkages is worse because of the strong collateral function. Therefore, the moderating effect of financial linkages on the relationship between collateral value and supply rationing is considered to mainly occur for small-scale farmers with small forest-land areas. Thus, the following hypothesis is proposed. Meanwhile, the mechanism of action is shown in Figure 1.

Hypothesis 1b.

Financial linkages play a moderating effect in the relationship between collateral value and supply rationing for small-scale farmers.

2.2. Financial Linkages and Demand Rationing of Forest Rights Mortgage Loans

The demand rationing of forest rights mortgages is mainly caused by borrowers’ automatic abandonment of loan applications due to reasons such as perceived high transaction costs, high interest costs, or fear of collateral loss. Financial linkages can alleviate demand rationing by using their information transfer and loan organization advantages to reduce transaction and interest costs. This improves the expected utility of farmers’ loan applications, thus alleviating farmer demand rationing.

Financial linkages can reduce transaction costs. Firstly, rural intermediaries transmit not only information for banks, but also credit information for borrowers. Through this information transmission, farmers participating in financial linkage can more easily learn about forest rights mortgage loan policies, interest preferential policies, loan processes, and various credit information other than forest rights mortgage loans, which saves their information search costs [37]. Additionally, it helps to improve farmers’ financial knowledge and credit experience, thus further reducing relevant transaction costs. Secondly, imperfections in the forest land transfer market and the control of forestry property rights mean that forest property assets as collateral have strong asset specificity and uncertainty in the transaction. As an endogenous organization, rural intermediaries have the advantage of relationship-based transactions compared with financial institutions. Therefore, financial institutions usually delegate the process of collateral asset evaluations, borrower qualification review, and post-loan default disposal to rural intermediaries for small forest rights mortgage loans, which streamlines the transaction process and thus saves transaction costs for both borrowers and lenders [38].

Financial linkages can reduce interest costs. Firstly, according to the above analysis, they reduce the transaction costs of both supply and demand through information advantages, supervision advantages, and loan organization advantages. The expected return of financial institutions will therefore be enhanced, which is conducive to promoting the supply side to reach a deal under the condition of lower interest costs. Secondly, rural intermediaries have the advantages of loan organizations. Through the organization function of rural intermediaries, farmers’ loans can be organized and farmers’ “individual loans” can be changed into “group loans”, thus providing higher collateral value and forming a certain loan scale effect. Organized groups have greater advantages in loan interest rate negotiations than individual farmers. Previous empirical studies have also proved that financial linkages have a certain effect on reducing loan interest rates [19].

In summary, it can be considered that financial linkages can moderately reduce the transaction and interest costs of forest rights mortgage loans, improve the expected income of farmers, and alleviate demand rationing. Based on the above analysis, the mechanism of action is shown in Figure 2, and the following hypotheses are proposed.

Hypothesis 2a.

Financial linkages have a significant negative impact on the demand rationing of forest rights mortgage loans.

Hypothesis 2b.

Transaction costs play a role of intermediary effect in the relationship between financial linkages and demand rationing.

3. Research Design

3.1. Identification Methods of Credit Rationing Types for Forest Rights Mortgage Loans

To empirically examine the impact of financial linkages on credit rationing for forest rights mortgages, we first need to identify the extent and type of credit rationing received by farmers. In the survey, the direct elicitation method (DEM) [39] was used to determine whether farmers were receiving credit rationing for forest rights mortgage loans; the specific discriminatory method and mechanism are shown in Figure 3. The sample households were classified into four categories—lack of demand, obtaining a loan, supply rationing, and demand rationing—according to whether they had demand for forest rights mortgage loans, whether they received loans, and whether they were subject to supply rationing or demand rationing. Among them, supply rationing and demand rationing refer to the rationed type, and lack of demand and obtaining a loan refer to the unrationed type. The identification of supply rationing refers to the identification method adopted by Zhang and Zhang (2014) [5]. In this method, supply rationing includes the situation where farmers who apply for loans have not obtained loans, as well as the situation that farmers who need loans give them up because they believe that the collateral conditions are not met, or that they can only obtain a smaller loan amount or a shorter loan period even if they apply. Notably, supply credit rationing in this study only included full-quantity rationing due to the supply-side failure of borrowers to obtain loans, but not partial quantity rationing due to the failure of the loan amount to meet borrowers’ demand for funds. The main reason is that this study principally examined the mitigation of supply rationing due to the facilitation of loan access by financial linkages. However, the main reason for the poor satisfaction of loan amount in partial quantity rationing involves the low valuation of forest rights due to the underdevelopment of the current forest rights transfer market, which is not fully consistent with the aforementioned theoretical analysis; therefore, it was not within the scope of this study, but could be a potential research avenue in the future. The classification refers to Gu’s (2019) study on the credit rationing of agricultural land mortgages [40].

3.2. Data Sources

The data in this study were obtained from the research data of farmers in the sample areas of the Fujian Province of China as part of the monitoring project of collective forest system reform in 2021. In our research we conducted a questionnaire survey of farmers through a “voluntarily completed, paid questionnaire” format. All respondents were interviewed in an informed manner, with data treated confidentially and used only for scientific research. The sample area included Sanming, Yongan, Youxi, Longyan, Zhangping, and Yongding in western Fujian, Nanping Zhenghe and Jianou in northern Fujian, Putian Xiangyou and Zhangzhou Changtai in southern Fujian, and Ningde Pingnan and Fuan in eastern Fujian. The reason for choosing the sample area is that Fujian Province was the earliest region to start piloting forest rights mortgage loans, was as well as one of the first regions in China to start reforming the forest rights mortgage loan transaction model. Among these regions, forest rights mortgage loans in Sanming and Zhangping of Longyan, Fujian Province, have been more successful. The “Fulin loan” promoted in Sanming is a typical financial linkage model, used as a case study for many other regions. However, because of certain variability in regional development levels, this study adopted a combination of stratified sampling and random sampling. Firstly, in Yongan, Youxi, and Zhangping counties, where the level of forestry development is better and the promotion of forest rights mortgage loans is more typical, seven villages were selected from each county, and 20 samples were taken from each village, making a total of 403 samples (17 samples were lost). Secondly, in the other seven counties where the promotion of forest rights mortgage loans is poor, three villages were selected from each county, and 20 samples were taken from each village, also forming 420 samples, making a total of 823 samples. After eliminating invalid samples and samples with missing data, 785 valid samples were finally obtained, with a valid proportion of 95.38%. The specific distribution of the sample data is shown in

Table 1. Distribution of sample farmers.

Sample Area	Sample Cities	Sample Counties (Cities and Districts)	Number of Samples	Percentage
Western Fujian	Sanming	Yongan	135	17.20%
		Youxi	124	15.80%
	Longyan	Zhangping	125	15.92%
		Yongding	58	7.39%
Northern Fujian	Nanping	Jianou	57	7.26%
		Zhenghe	57	7.26%
Minnan	Putian	Xiangyou	58	7.39%
	Zhangzhou	Changtai	57	7.26%
Min Dong	Ningde	Pingnan	58	7.39%
		Fu An	56	7.13%
Total			785	100%

.

3.3. Methods and Model Construction

To test the effect of financial linkages on the credit rationing of forest rights mortgages (test for Hypothesis 1a and Hypothesis 2a), this study used the propensity score matching (PSM) method proposed by Rosenbaum and Rubin (1985) [41]. Farmers may face the problem of self-selection bias in whether they participate in or receive financial linkages. This is due to the possibility that variables affecting the participation of farmers in financial linkage behavior may also affect credit rationing. For example, farmers with a large forest area and a high degree of forestry specialization may be more likely to participate in rural intermediaries, such as forestry cooperatives, to form financial linkages, and farmers with the above characteristics may also be more likely to obtain loans themselves. Thus, to prove that the financial linkage influences its credit rationing, not the effect of other characteristic variables, we excluded the endogeneity problem due to sample self-selection bias, using the PSM method to solve this problem. According to this method, the sample was divided into a treatment group (the sample participating in financial linkage) and a control group (the sample not participating in financial linkage). The method constructed a hypothetical state contrary to the facts through a counterfactual framework, treated involvement in financial linkages as an intervention, and measured the treatment effect of such an intervention, giving the results obtained by this method stronger credibility [42].

Firstly, the equation of farmers’ participation in financial linkage behavior was constructed. According to the stochastic utility decision model, it was assumed that the utility of farmers’ participation in financial linkage is ${\mathrm{U}}_{1\mathrm{i}}$, and the utility of no participation in financial linkage is ${\mathrm{U}}_{0\mathrm{i}}$. When ${\mathrm{U}}_{1\mathrm{i}}-{\mathrm{U}}_{0\mathrm{i}}>0$, the farmer will choose to participate in financial linkage, and when ${\mathrm{U}}_{1\mathrm{i}}-{\mathrm{U}}_{0\mathrm{i}}\le 0$, the farmer will not participate in financial linkage. Thus, this study established an equation model of farmers’ participation in financial linkage behavior as follows:

$${\mathrm{M}}_{\mathrm{i}}=\mathsf{\phi }\left({\mathrm{X}}_{\mathrm{i}}\right)+{\mathsf{\epsilon }}_{\mathrm{i}}$$

(1)

where ${\mathrm{M}}_{\mathrm{i}}$ denotes whether farmers participate in financial linkage. This variable is a binary variable with a value of 0 or 1. Moreover,${ \mathrm{X}}_{\mathrm{i}}$ is an exogenous explanatory variable (reflected as a control variable in Section 3.4.3) affecting farmers’ participation in financial linkages, and ${\mathsf{\epsilon }}_{\mathrm{i}}$ is the random error term.

Secondly, an equation model is constructed to measure the effect of participating in financial linkages on their credit rationing:

$${\mathrm{Y}}_{\mathrm{ki}}=\mathsf{\varphi }\left(\mathrm{Z}\right)+{\mathsf{\lambda }\mathrm{M}}_{\mathrm{i}}+{\mathsf{ϵ}}_{\mathrm{i}}$$

(2)

where ${\mathrm{Y}}_{\mathrm{ki}}$ is the outcome variable (reflected as a dependent variable in Section 3.4.1) in Equation (2), which represents the type of farmer credit rationing, where $\mathrm{k}$ = 1, 2, 3 denotes total rationing, supply rationing, and demand rationing, respectively. $\mathrm{Z}$ denotes the exogenous explanatory variables that affect farmers’ credit rationing. ${\mathrm{M}}_{\mathrm{i}}$ denotes whether the farmer participates in financial linkages. Additionally, ${\mathsf{ϵ}}_{\mathrm{i}}$ is a random disturbance term.

Thirdly, paired samples were selected by calculating propensity score values. The calculation of the propensity score (PS) was achieved through the model construction of Equation (1), which was performed using a logit model to calculate the conditional probability to derive the propensity score of individual farmer i to participate in financial linkages. Next, sample matching was performed based on the obtained PS values; propensity score matching was then performed. Considering the robustness of the results, three methods of least-nearest-neighbor matching, radius matching, and kernel matching were used to match the samples in this study.

Fourth, the average treatment effect (ATT) was calculated, which indicates the extent to which interventions through participation in financial linkages can produce changes in the credit rationing of farmers’ forest rights mortgages. The calculation of the average treatment effect can be expressed as follows:

$$\begin{gathered} \mathrm{ATT}=\mathrm{E}\left[{\mathrm{Y}}_{1\mathrm{i}}-{\mathrm{Y}}_{0\mathrm{i}}|{\mathrm{rd}}_{\mathrm{i}}=1\right]=\mathrm{E}\left\{\mathrm{E}\left[{\mathrm{Y}}_{1\mathrm{i}}-{\mathrm{Y}}_{0\mathrm{i}}|{\mathrm{rd}}_{\mathrm{i}}=1,{\mathrm{X}}_{\mathrm{i}}=\mathrm{X}\right)]\right\} \\ =\mathrm{E}\left\{\mathrm{E}\left[{\mathrm{Y}}_{1\mathrm{i}}|{\mathrm{rd}}_{\mathrm{i}}=1,{\mathrm{X}}_{\mathrm{i}}=\mathrm{X}\right]\right\}-\mathrm{E}\left\{\mathrm{E}\left[{\mathrm{Y}}_{0\mathrm{i}}|{\mathrm{rd}}_{\mathrm{i}}=0,{\mathrm{X}}_{\mathrm{i}}=\mathrm{X}\right]\right\} \end{gathered}$$

(3)

where ${\mathrm{Y}}_{1\mathrm{i}}$ and ${\mathrm{Y}}_{0\mathrm{i}}$ denote the standardized values of credit rationing received by the same farm household in cases of both participation in financial linkages and non-participation in financial linkages, respectively. ${\mathrm{rd}}_{\mathrm{i}}=1$ indicates that farmer $\mathrm{i}$ belongs to the sample participating in financial linkage and ${\mathrm{rd}}_{\mathrm{i}}=0$ indicates that the farmer belongs to the sample not participating in financial linkage. ${\mathrm{X}}_{\mathrm{i}}=\mathrm{X}$ denotes that the observable characteristic variable ${\mathrm{X}}_{\mathrm{i}}$ is controlled. The resulting $\mathrm{ATT}$ is the difference in credit rationing between farmers involved in financial linkages and those not involved in financial linkages.

3.4. Variable Design

3.4.1. Dependent Variable

This study examined the mitigating effect of financial linkages on the credit rationing of forest rights mortgage loans; thus, the explanatory variables included three dependent variables of total rationing, supply rationing, and demand rationing of forest rights mortgage loans, where the identification of supply rationing and demand rationing was based on the results shown in Figure 3; total rationing included supply rationing and demand rationing.

3.4.2. Treatment Variable

Financial linkages were used as the explanatory variable of PSM, as well as the core explanatory variables. This variable is a 0-1 variable. It was obtained by querying whether farmers were involved in a rural intermediary that provided financial services, which was limited by the field research sample, in which only three types of rural intermediaries were involved: joint household groups, forestry cooperatives, and village authorities. The method of acquisition was mainly through setting the following questions in the questionnaire. Firstly, we asked whether the village had taken out forest rights mortgage loans through the model of “bank + village authority + farmer”. Secondly, the farmers were asked whether they had joined forestry cooperatives and, if so, whether these provided loan guarantees or assistance with loans (such as information transfer and credit certification). Thirdly, farmers were asked whether they had joined the joint household group and whether this group could provide mutual guarantee and other services to assist in the loan. If the above was true, it was included in the financial linkage sample and was assigned a value of 1. Otherwise, it was a non-financial linkage sample, assigned a value of 0.

3.4.3. Control Variables

The control variables here are also known as the covariates in the PSM model. The principle of covariate setting is to use characteristic variables that will influence both the treatment variable and dependent variable. In this study, we refer to the relevant literature on the study of factors influencing financial linkages and credit rationing, and selected four types of control variables, as shown in

Table 2. Description of variables.

Variable Category	Variables	Code	Variable Description
Dependent variable	Total rationing	Y1	Subject to supply rationing or demand rationing; 1 = yes; 0 = no
	Supply rationing	Y2	Subject to complete quantity rationing; 1 = yes; 0 = no
	Demand rationing	Y3	Subject to demand rationing; 1 = yes; 0 = no
Explanatory variable	Financial linkage	FL	Participation in financial linkages; 1 = yes; 0 = no
Control variables: ① Family characteristics	Age	AGE	Age of farmers
	Education	EDU	1 = Elementary school and below; 2 = junior high school; 3 = secondary or high school; 4 = college or bachelor’s degree or above
	Household income	HI	1 = Logarithm of total household income in the previous year
	Main source of income	MSI	Working part-time; 2 = business; 3 = fixed wage income; 4 = agricultural production; 5 = forestry production; 6 = government subsidies
	Fixed assets	FA	Logarithm of the estimated amount of fixed assets
② Forest land and forestry specialization characteristics	Forest-land area	FLA	Total area of family forest land (/hm2)
	Major tree species	MTS	1 = Bamboo forest and other short-term tree species; 0 = timber forest and other non-short-term tree species
	Forest insurance	FI	Participation in forest insurance; 1 = yes; 0 = no
	Forestry planting activities	FPA	Engagement in forestry planting activities; 1 = yes; 0 = no
	Forestry training	FT	Receive forestry-related training; 1 = yes; 0 = no
③ Social capital and lending characteristics	Gift expenses	GE	Logarithm of prior year’s gift expense amount
	Village cadres	VC	Someone in the family is a village cadre; 1 = yes; 0 = no
	Other loan experience	OLE	Existence of other loan experience; 1 = yes; 0 = no
	Distance to financial institution	DIS	Distance from home to nearest financial institution (/km)
④ Regional characteristics	Regional policy	RP	1 = Typical areas for forest rights mortgage promotion; 0 = atypical areas

. The first category was family characteristics variables, including the five indicators of age, education, household income, main source of income, and fixed assets. Family characteristics are the variables that most studies on credit rationing and financial linkages utilize, and the indicators were selected with reference to studies by Duan et al. (2015) [43] and Shen (2018) [44]. The second category was the forest land and forestry specialization characteristics variables, including forest-land area, major tree species, forest insurance, forestry planting activities, and forestry training. Firstly, this study used forest-land area owned by farmers to reflect the value of collateral [45]. Secondly, because it is considered a forest rights mortgage loan, the degree of forestry specialization influenced whether farmers participated in financial linkage as well as the transaction cost when applying for the loan [46,47], based on which of the above indicators were selected. The third category was the social capital and lending characteristic variables, including gift expenditure, village cadres, other loan experience, and distance to financial institutions. These indicators are also frequently used in related studies [43]. The fourth category was the regional characteristics, setting an indicator of regional policy and classified according to whether forest rights mortgages are promoted for the typical region. In this study, Yongan, Youxi, and Zhangping were typical areas for the promotion of forest rights mortgages. The supporting policies for forest rights mortgage loans were more suitable in these three areas than in others and the regional government had guidance and policy support for the transaction mode of financial linkage [19]; thus, it is easier to form financial linkages in these areas, and easier to obtain loans.

3.5. Descriptive Statistics of Variables

3.5.1. Statistics of the Credit Rationing Degree

Table 3. Statistics of credit rationing degree of forest rights mortgage loans.

Farmers Type		Obtaining a Loan	Supply Rationing	Demand Rationing	Lack of Demand	Total
Total Sample	Number	68	56	104	557	785
	Proportion	8.66%	7.13%	13.25%	70.96%	100%
Demand sample	Number	68	56	104	/	228
	Proportion	29.82%	24.56%	45.61%	/	100%
Small-scale farmers	Number	31	49	59	475	614
	Proportion of total sample	5.05%	7.98%	9.61%	77.36%	100%
	Proportion of demand sample	22.30%	35.25%	42.45%	/	100%
Large-scale farmers	Number	37	7	45	82	171
	Proportion of total sample	21.64%	4.09%	26.32%	47.95%	100%
	Proportion of demand sample	41.57%	7.87%	50.56%	/	100%

Note: Small-scale farmers refer to farmers with a forest-land area < 5.33 hm²; large-scale farmers refer to farmers with a forest-land area ≥ 5.33 hm²; demand sample refers to the total sample minus the lack of demand sample.

shows the statistics of the types and degrees of credit rationing for forest rights mortgage loans. Notably, the demand sample refers to the total sample minus the non-demand sample, which indicates the farmers with a demand for forest rights mortgage loans. Therefore, we used two statistical values in the statistics of credit rationing degree: one is the credit rationing degree based on the total sample; the other is the credit rationing degree based on the demand sample. The sample data show that, firstly, the demand ratio and the access ratio of forest rights mortgages were both low. The demand ratio of the total sample was only 29.04%. The demand ratio of small-scale farmers was only 22.64%, while the demand ratio of large-scale farmers is 52.05%. The access ratio of the total sample was only 8.66%, and the access rate of small-scale farmers was 5.05%, while the access rate of large-scale farmers was 21.64%. Heterogeneity varies widely among farmers, with both access and demand ratios much lower for small-scale farmers than for large-scale farmers. Secondly, from the demand sample, the degree of credit rationing for forest rights mortgage loans was high. Total rationing accounted for 70.17% of the demand sample (20.38% of the total sample), demand rationing accounted for 45.61% of the demand sample (13.25% of the total sample), and supply rationing accounted for 29.82% of the demand sample (8.66% of the total sample). Thirdly, heterogeneous farmers exhibited a large variation in credit rationing. From the demand sample, the supply rationing ratio of small-scale farmers was 35.25%, much higher than that of large-scale farmers at 7.87%, while the demand rationing ratio of large-scale farmers was 50.56%, which was slightly higher than that of small-scale farmers at 42.45%.

3.5.2. Descriptive Statistics of Control Variables

The descriptive statistical analysis of the control variables is detailed in

Table 4. Descriptive statistics of control variables.

Variable Category	Variables	Total Sample		Treatment Group (A)		Control Group (B)		Difference (A-B)
		Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Err.
Dependent variable	TR	0.204	0.393	0.149	0.314	0.224	0.414	−0.075 ***	0.031
	SR	0.071	0.257	0.047	0.192	0.080	0.276	−0.033 **	0.021
	DR	0.133	0.324	0.096	0.259	0.146	0.344	−0.050 **	0.026
Explanatory variable	FL	0.265	0.441	1.000	0	0.000	0	1.000	0
Control variables: ① Family characteristics	AGE	53.520	10.966	52.413	9.643	53.919	11.387	−1.505 *	0.886
	EDU	1.871	0.786	1.932	0.816	1.849	0.775	0.083	0.064
	HI	10.953	1.025	11.096	1.085	10.901	0.998	0.195 **	0.083
	MSI	3.178	1.658	3.034	1.573	3.230	1.686	−0.197	0.134
	FA	13.329	4.305	13.761	4.350	13.008	4.289	0.753	0.349
② Forest land and forestry specialization characteristics	FLA	4.365	7.693	6.823	11.063	3.479	5.800	3.344 ***	0.811
	MTS	0.561	0.496	0.451	0.498	0.601	0.490	−0.149 ***	0.039
	FI	0.237	0.425	0.382	0.487	0.185	0.388	0.197 ***	0.034
	FPA	0.724	0.447	0.702	0.458	0.733	0.443	−0.031	0.036
	FT	0.323	0.467	0.389	0.488	0.299	0.458	0.091 **	0.038
③ Social capital and lending characteristics	GE	7.528	2.854	7.631	2.878	7.492	2.848	0.139	0.231
	VC	0.415	0.493	0.442	0.497	0.405	0.491	0.037	0.040
	OLE	0.291	0.454	0.386	0.488	0.256	0.437	0.130 ***	0.036
	DIS	9.329	7.165	9.108	7.668	9.408	6.982	0.300	0.582
④ Regional characteristics	RP	0.417	493	0.563	0.497	0.364	0.481	0.199 ***	0.039
N		785		208		577

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively; “Treatment Group” refers to the financial linkage sample group; “Control Group” refers to the non-financial linkage sample group.

. First, the number of samples involved in financial linkages was 208, accounting for 26.5%, and the number of samples not involved in financial linkages was 577, accounting for 73.5%. By comparing the differences in the values of the variables between farmers participating in financial linkages and those not participating in financial linkages, the following was found. In the variables of forest land and forestry specialization characteristics, there were significant differences between the financial linkage and non-financial linkage groups. Forest-land area owned by farmers in the financial linkage group was significantly higher than that in the non-financial linkage group, and the mean values of forestry training and forest insurance in the financial linkage group were significantly higher than those of the non-financial linkage group. Meanwhile, the mean values of household income, other loan experiences, and regional policy were significantly higher in the financial linkage group than in the non-financial linkage group, whereas the mean values of age in the financial linkage group were lower than in the non-financial linkage group.

4. Empirical Results

4.1. Estimation of Propensity Scores

A logit model was used to estimate PS values and to analyze the factors that influence farmers’ financial linkage behavior. Following the recommendation of Rosenbaum and Rubin (1983) [41] to introduce a higher-order term of ${\mathrm{X}}_{\mathrm{i}}$ in the model in order to further improve the calculation results’ precision, the squared term of the forest-land area (FLA-Square in

Table 5. Logit estimation results of propensity scores.

Variable Type	Variable	Regression Coefficient	Standard Error
Family characteristics	AGE	−0.011	0.010
	EDU	−0.225 *	0.134
	HI	0.116	0.103
	MSI	−0.103 *	0.056
	FA	−0.029	0.019
Forest land characteristics and forestry specialization characteristics	FLA	0.090 ***	0.031
	FLA-Square	−1.188 × 10−3 *	6.57 × 10−4
	MTS	−0.478 ***	0.184
	FI	0.883 ***	0.220
	FPA	−0.492 **	0.209
	FT	0.358 *	0.198
Social capital and lending characteristics	GE	0.014	0.034
	VC	0.032	0.192
	OLE	−0.048 ***	0.015
	DIS	0.305 *	0.208
Regional characteristics	RP	0.733 ***	0.211
N		785
LR chi2		104.67 ***
Pseudo R2		0.203

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

) was chosen in this study; the reason for the choice was to consider that farmer financial linkage behavior will increase with the increase in forest-land area, but when the forest-land area is sufficiently large, farmers no longer need financial linkages. Therefore, the relationship between forest-land area and farmers’ financial linkage behavior was considered as an inverted U-shaped relationship.

Table 5 shows the estimation results of the propensity score. The model estimation results found that among the variables of family characteristics, education and the main source of income passed the significance test, and both had a significantly negative effect on whether foresters participated in the financial linkage behavior of forest rights mortgages. All the variables of forest land characteristics and forestry specialization passed the significance test, among which both forest-land area and forest-land-area-squared passed the significance test, indicating that the relationship between forest-land area and farmers’ financial linkage behavior was an “inverted U-shaped” relationship. Farmers whose main tree species were timber forests and other non-short-term tree species were found to be more likely to participate in financial linkage behavior. Additionally, the variables of forest insurance, forest planting activities, and forestry training passed the significance test, indicating that farmers who participated in forest insurance, received forestry training, and did not directly engage in forestry planting activities but used forest assets as a means of capitalization were more inclined to engage in financial linkage behavior. Among the indicators of lending characteristics, two variables of other loan experience and distance to the financial institution passed the significance test, indicating that foresters with loan experience or whose households are close to financial institutions are more likely to participate in or obtain financial linkages. In addition, the regional characteristics passed the significance test, indicating that the regional market environment and policy boosts were more advantageous for the development of financial linkages.

4.2. Balance Test and Common Support Test

To ensure the matching effect, the balance test and the common support test were performed.

Table 6. Balance test of variables under the least-nearest-neighbor matching method.

Control Variables	Mean Value before Matching			Mean Value after Matching			Deviation Rate after Matching (%)
	Treatment Group	Control Group	p-Value	Treatment Group	Control Group	p-Value
AGE	52.253	53.929	0.064 *	52.408	51.429	0.341	9.3
EDU	1.950	1.853	0.137	1.921	1.948	0.749	−3.3
HI	11.126	10.902	0.008 ***	11.075	11.007	0.510	6.5
MSI	3.000	3.227	0.097 *	2.990	2.890	0.555	6.1
FA	13.757	13.080	0.406	13.664	13.470	0.706	4.0
FLA	6.863	3.432	0.000 ***	5.804	5.044	0.394	8.5
FLA-Square	173.844	44.372	0.000 ***	124.68	85.613	0.318	9.5
MTT	0.444	0.601	0.000 ***	0.456	0.419	0.472	7.4
FI	0.384	0.186	0.000 ***	0.372	0.340	0.523	7.1
FPA	0.707	0.732	0.495	0.702	0.707	0.911	−1.2
FT	0.399	0.296	0.008 **	0.387	0.419	0.533	−6.6
GE	7.668	7.492	0.453	7.641	7.530	0.700	3.9
VC	0.460	0.406	0.189	0.450	0.461	0.838	−2.1
LE	0.394	0.255	0.000 ***	0.377	0.429	0.298	−11.3
DIS	8.751	9.458	0.226	8.936	9.062	0.860	−1.8
RP	0.556	0.365	0.000 ***	0.539	0.550	0.838	−2.1

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

shows the results of the balance test for each variable before and after matching under the least-nearest-neighbor matching method.

The results showed that, for most of the variables before matching, there were significant differences between the treatment group and the control group, while after propensity matching, the differences between the treatment and control groups were significantly reduced, except for two variables; the deviation rates of the remaining variables were reduced to less than 10%. In addition, the p-values of all variables were increased to different degrees and were not significant, indicating no significant difference between the two groups in the key variables; thus, it can be considered that the two groups had basically the same distribution after matching. This indicated that the selection of the variables was reasonable and that the matching process was effective. Therefore, the results of the balance test were good and the choice of model was appropriate.

Figure 4 and Figure 5 show the kernel density plots before and after matching, which are for the common support test. To ensure the matching quality of sample data, the density function plots were further plotted to test the common support domain after matching and obtaining the PS values. Samples of the treatment and control groups after matching were very close to each other in all characteristics, a common value interval existed, and the matching effect was good; therefore, the common support hypothesis was satisfied.

In addition, through matching, 11 samples were lost in the control group, and 7 samples were lost in the treatment group, for a total of 18 lost samples. In the end, a total of 767 samples participated in the matching, with a low sample loss ratio; again, this showed that the common support conditions for the control and treatment groups were satisfactory.

4.3. Results of the Effect of Financial Linkages on Credit Rationing

After the above treatment, the treatment and control group samples were basically the same in the other characteristic variables. The only differences were in whether financial linkages could be obtained. On this basis, the least-nearest-neighbor matching method, radius matching method, and kernel matching method were used to calculate the average treatment effect. The results obtained by the three different methods were basically consistent, which showed that the results had good robustness. To further ensure the robustness of the results, this study adopted the average value of the three methods as the calculation result of financial linkage effects, as shown in

Table 7. Average treatment effects of propensity value matching (total rationing).

Matching Method	Treatment Group	Control Group	ATT	Std. Err.	T-Value
Least-nearest-neighbor matching	0.105	0.257	−0.152 ***	0.045	3.34
Radius matching	0.108	0.258	−0.150 ***	0.033	4.50
Nuclear matching	0.103	0.258	−0.155 ***	0.033	4.67
Average value	0.105	0.258	−0.152	-	-

Note: *** indicate significance at the 1% levels; ATT refers to the average treatment effects.

,

Table 8. Average treatment effects of propensity value matching (supply rationing).

Matching Method	Treatment Group	Control Group	ATT	Std. Err.	T-Value
Least-nearest-neighbor matching	0.042	0.105	−0.063 **	0.031	−2.01
Radius matching	0.041	0.085	−0.044 **	0.022	−2.00
Nuclear matching	0.041	0.085	−0.043 **	0.022	−2.00
Average value	0.041	0.092	−0.050	-	-

Note: ** indicate significance at the 5% levels; ATT refers to the average treatment effects.

and

Table 9. Average treatment effects of propensity value matching (demand rationing).

Matching Method	Treatment Group	Control Group	ATT	Std. Err.	T-Value
Least-nearest-neighbor matching	0.063	0.152	−0.089 **	0.037	−2.41
Radius matching	0.067	0.173	−0.106 ***	0.027	−3.89
Nuclear matching	0.062	0.173	−0.111 ***	0.027	−4.11
Average value	0.064	0.166	−0.102	-	-

Note: **, and *** indicate significance at the 5%, and 1% levels, respectively; ATT refers to the average treatment effects.

. The results show that financial linkages have a significantly negative effect on supply rationing, demand rationing, and total rationing, which means that there is a mitigating effect. Additionally, financial linkages passed the significance test under all three matching methods, indicating that the results were robust. The results show that financial linkages significantly reduce total rationing by 15.2% (144.76% change), including a significant reduction in supply rationing by 5% (54.3% change) and a significant reduction in demand rationing by 10.2% (61.4% change) compared to non-participation in financial linkages. The above results supported Hypothesis 1a and Hypothesis 1b.

4.4. Heterogeneity Test

Heterogeneity analysis was conducted by dividing farmers into two groups—small-scale and large-scale farmers—based on the area of forest land owned by farmers to reflect the value of collateral. The impacts of financial linkages were measured separately for these two groups using the PSM method; the results are shown in

Table 10. Heterogeneous farm subgroup average treatment effects.

Type	Farmers’ Classification	Treatment Group	Control Group	ATT	Std. Err.	T-Value
Supply rationing	Smallholder farmers	0.036	0.116	−0.08 **	0.038	−2.09
	Moderate-scale farmers	0.053	0.018	0.035	0.045	0.77
Demand rationing	Smallholder farmers	0.043	0.109	−0.065 *	0.039	−1.71
	Moderate-scale farmers	0.105	0.281	−0.175 *	0.095	−1.84
Total rationing	Smallholder farmers	0.080	0.225	−0.146 ***	0.034	−4.25
	Moderate-scale farmers	0.169	0.355	−0.185 **	0.091	−2.04

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively; ATT refers to the average treatment effects.

. First, the results of supply rationing show that small-scale farmers who participated in financial linkage exhibited a significant reduction in supply rationing by 8.0% (68.97% change), as opposed to those who did not participate in financial linkages, whereas the influence of financial linkage on supply rationing for large-scale farmers did not pass the significance test. This indicates that the effect of financial linkages on supply rationing differs significantly due to the difference in collateral value, and the effect is more significant for small-scale farmers. Secondly, in terms of the demand rationing results, participation in financial linkages reduced the demand rationing suffered by small-scale farmers by 6.5% (59.63% change) and reduced the demand rationing of large-scale farmers by 17.5% (62.28% change). This indicates that financial linkages have a higher mitigating effect on demand rationing for large-scale farmers than for small-scale farmers. Thirdly, in terms of total rationing, participation in financial linkages reduced the total rationing of small-scale farmers by 14.6% (182.5% change) and reduced the total rationing of large-scale farmers by 18.5% (109.47% change). Although the data showed that the absolute value of financial linkages’ mitigating effect on the total rationing of small-scale farmers was slightly lower than that of large-scale farmers, the change ratio of small-scale farmers was significantly higher than that of large-scale farmers, indicating that the marginal effect on small-scale farmers would be higher.

5. Mechanism Test

5.1. Moderating Effect Test

Based on the heterogeneity analysis, further tests were conducted on the moderating effect of financial linkages in the collateral value and supply rationing relationships of small-scale farmers (tested for Hypothesis 1b). The logit model was used to perform stepwise regressions of forest-land area, financial linkages, and the interaction term of forest-land area and financial linkages; the regression results are shown in

Table 11. Moderating effect test.

Variables	Sample Type: Small-Scale Farmers $\mathbf{Dependent} \mathbf{Variable}: {\mathbf{Y}}_2$
	(1)	(2)	(3)
FLA	−0.022 ** (0.010)	−0.021 ** (0.010)	−0.032 ** (0.014)
FL		−0.068 ** (0.034)	−0.246 ** (0.156)
FLA × FL			−0.004 * (0.003)
Other control variables	Control	Control	Control
N	614	614	614
Pseudo R2	0.153	0.170	0.179

Note: *, ** indicate significance at the 10%, 5% levels.

. The first step was to examine the effect of the forest-land area on supply rationing. The results are shown in Table 11, column (1); the increase in forest-land area has a significant negative impact on supply rationing. In the second step, the financial linkage variable was added based on the first step; the results are shown in Table 11, column (2): both forest-land area and financial linkage have a significant negative effect on supply rationing. In the third step, the interaction term of forest-land area and financial linkage were added based on the second step; the results are shown in Table 11, column (3). The interaction term showed a significantly negative effect in the same direction as the effect of forest-land area on supply rationing. This result indicated that financial linkage had a moderating effect on the relationship between collateral value and the supply rationing of small-scale farmers, and the moderating effect enhances the collateral guarantee function, which further alleviates supply rationing. This result supports Hypothesis 1b. In addition, the same moderating effect test was carried out for large-scale farmers’ samples and total samples; however, it was found that the moderating effect of financial linkage was not significant in these two types of samples (these results are omitted here), which shows that the moderating effect of financial linkages is significant mainly for small-scale farmers with insufficient collateral value. This result is consistent with the results of the previous heterogeneity test.

5.2. Intermediary Effect Test

According to the aforementioned theoretical analysis, financial linkages mainly alleviate the transaction and interest costs of forest rights mortgage loans through information transmission and loan organization advantages, so as to achieve the purpose of alleviating demand rationing. Therefore, the role of transaction and interest costs of financial linkage was analyzed next.

Firstly, the intermediary effect of transaction costs was tested (for Hypothesis 2b). “Other loan experiences” was used as an intermediary variable to indicate the transaction costs of farmers. Farmers with other loan experiences are more familiar with credit knowledge and the process and related credit preferential policies, which can save transaction costs and help alleviate farmers’ demand rationing. At the same time, the participation of financial linkage can give farmers more opportunities to obtain various loans. The logit regression model, constructed to test the intermediary effect of transaction costs, was tested based on the intermediary effect test process proposed by Wen et al. (2005) [48]; the conclusions are shown in

Table 12. Mediation effect test for transaction costs.

Variables	$$\begin{gathered} \left(1\right) \mathbf{Dependent} \\ \mathbf{Variable}: {\mathbf{Y}}_3 \end{gathered}$$	$$\begin{gathered} \left(2\right) \mathbf{Dependent} \\ \mathbf{Variable}: {\mathbf{F}}_{\mathbf{i}} \end{gathered}$$	$$\begin{gathered} \left(3\right) \mathbf{Dependent} \\ \mathbf{Variable}: {\mathbf{Y}}_3 \end{gathered}$$
Financial linkages	−0.322 *** (0.077)	0.178 ** (0.072)	−0.277 *** (0.078)
Mediating variable: loan experience			−0.222 *** (0.076)
Other control variables	Control	Control	Control
N	785	785	785
Pseudo R2	0.167	0.426	0.179

Note: **, and *** indicate significance at the 5% and 1% levels; the variable ${\mathrm{F}}_{\mathrm{I}}$ refers to the variable “Loan experience”.

. In the first step, Table 12, column (1) presents the impact of financial linkage on demand rationing. The results show that financial linkages have a significantly negative impact on demand rationing (DR). In the second step, Table 12, column (2) details the impacts of financial linkages on intermediary variables, i.e., whether there are other loan experiences. The results showed that financial linkages had a positive significant impact on other loan experiences. In the third step, Table 12, column (3) indicates the impact of financial linkages and intermediary variables on demand rationing at the same time. The results suggest that both the financial linkage variables and intermediary variables had a significantly negative impact on demand rationing, which showed that transaction costs play a partial intermediary role in the relationship between financial linkages and the demand rationing of forest rights mortgage loans. This result supports Hypothesis 2b.

Secondly, an analysis of interest cost was performed. The interest cost could only be observed in the sample of obtaining loans, limited by the number of samples; therefore, the inter-group difference test was used for analyses. The sample of 68 households which obtained loans were grouped according to whether they participated in financial linkages or not, with examination of whether there were significant differences in loan interest rates among different groups. The results are shown in

Table 13. Comparison of group differences in interest costs.

Variables	Treatment Group (A)		Control Group (B)		Differences (A–B)
	Number	Average	Number	Average	Average	T-Value
Mortgage Annual Interest Rate	45	0.071	23	0.099	0.025 ***	3.011

Note: *** indicate significance at the 1% levels; “Treatment group” refers to the financial linkage sample group; “Control group” refers to the non-financial linkage sample group.

. There were 45 samples participating in financial linkages and 23 samples not participating in financial linkage. The average loan interest rate of the sample group participating in financial connection was 7.1% per year, while the average loan interest rate of the sample group not participating in financial linkages was 9.9%. There was a significant difference between the two groups, and the average interest rate of the financial linkage group was 2.8% lower than that of the non-financial linkage group. Therefore, it was considered that financial linkage reduces interest costs, thus helping to alleviate demand rationing.

6. Discussion and Implications

6.1. Discussion of Results

The application of financial linkages in micro-credit loans has been widely discussed, and the experience of micro-finance institutions, such as Grameen Rural Bank, demonstrates the importance of making full use of the internal mechanisms of rural financial markets, such as credit relationships within villages. However, relevant studies on its applications in mortgage lending have rarely been mentioned. The application of financial linkage in forest rights mortgage loans occurs in the context of forest rights system reform, which causes high transaction risks due to the relatively small area of forest land allocated to most foresters and the low marketability of forest rights transfer. Additionally, further research on the effect of financial linkage’s application in forest rights mortgage loans and its mechanism of action is very important, which was the main purpose of this study. The key findings of this paper are discussed below.

Firstly, financial linkages showed a significant negative effect on supply rationing, demand rationing, and total rationing of forest rights mortgage loans. It means that there is a significant mitigating effect of financial linkage on credit rationing and it helps to promote the development of forest rights mortgage loans, especially for farmers who have a demand for loans but who suffer from supply rationing due to collateral validity problems or demand rationing because they cannot afford higher financing costs. This result is consistent with the case study results of Chen and Gao (2018) [49] on the application of financial linkages in agricultural land mortgage loans.

Secondly, there were differences in the effects of financial linkage on heterogeneous farmers. The effects of financial linkages on supply rationing were significant only for small-scale farmers, although the effects of financial linkages on demand rationing were greater for large-scale than for small-scale farmers. It indicates that financial linkages mainly alleviate supply rationing due to insufficient collateral, while supply rationing for large-scale farmers was usually caused by other reasons. Moreover, because the amount of loans obtained by large-scale farmers was larger than that of small-scale farmers, the reduction in transaction costs will produce higher marginal benefits for large-scale than for small-scale farmers, thus the alleviation of demand rationing by financial linkages for large-scale farmers has a greater effect than that of small-scale farmers. Previous studies on the application of financial linkages in mortgage lending have mainly focused on cases with small-scale farmers; therefore, there is a lack of research on the differences in heterogeneous farmers. However, the results of this study suggest that financial linkages are equally effective in the demand rationing of forest mortgage loans for moderately sized farmers. This result implies a greater need for different programs and policies for heterogeneous farmers in the future.

Thirdly, financial linkages have a moderating effect on the relationship between collateral value and supply rationing for small-scale farmers. This result is consistent with the finding that financial linkages exhibit a substitution effect on collateral in microcredit lending [34,50]. This suggests that there are mechanisms for financial linkages to act as substitutes or complements to collateral through information advantages and reputational incentives. This implies that the design of future measures should focus more on how to further expand this information advantage and reputation incentive mechanism. However, when we performed propensity score matching calculations, we found that the forest-land area showed an inverted U-shaped relationship with financial linkage participation behavior. This suggests that small-scale farmers are less likely to participate in financial linkages, thus preventing many small-scale farmers from benefiting from financial linkages. Therefore, policies to further support small-scale farmers’ participation in rural intermediation should be developed.

Fourth, transaction costs play an intermediary role in the relationship between financial linkages and the demand credit rationing of forest rights mortgage loans. This suggests that the mechanism of action of financial linkages to mitigate demand rationing through the reduction of transaction costs also holds. This result is consistent with that of Jiang’s (2020) [38] research on agricultural land mortgage loans. High transaction and interest costs are the key reasons for serious demand credit rationing, as well as the most important reasons for the low effective demand of forest rights mortgage loans. Therefore, it is important to reduce the transaction costs, as well as interest costs, of forestry mortgages to promote the development of this business.

6.2. Implications

Firstly, the promotion and application of financial linkages in forest rights mortgages can be further encouraged, with a focus on enhancing the participation rate of small-scale farmers. Financial linkage has a significant alleviating effect on farmers’ credit rationing; therefore, the innovation and promotion of financial linkages should be actively encouraged. It is also important to actively guide small-scale farmers with loan needs to participate in rural intermediary organizations, which can help promote the alleviation of supply-based rationing for small-scale farmers. Meanwhile, small-scale farmers with loan needs should be actively guided to participate in rural intermediary organizations, and innovations in financial linkage models that are more conducive to the absorption of small-scale farmers should be encouraged, in favor of promoting the alleviation of supply rationing for smallholder farmers.

Secondly, differentiated forestry mortgage policies should be developed for heterogeneous farmers. There are differences in the mitigation effects of financial linkages on supply rationing and demand rationing for heterogeneous farmers. Therefore, differentiated policies for heterogeneous farmers are needed. For small-scale farmers, because the loan amount is small, it is necessary to further simplify their transaction processes and reduce their transaction costs in order to effectively improve their loan utility. As for large-scale farmers, some may face insufficient loan amounts or property rights controls; it would be necessary to actively promote the development of the forest rights transfer market and relax the related property rights control, to effectively enhance the value of forest rights collateral.

Thirdly, we should further devote ourselves to reducing transaction costs and risks. Government departments can promote the reduction of transaction costs by improving the forest rights collection and storage system, the forest rights transfer and transaction system, and the loan subsidy system. Banks can reduce transaction costs by providing rural intermediaries with more credit information training services or expanding their network platforms. Rural intermediaries can reduce transaction risks by further standardizing related financial services.

6.3. Limitations and Further Research

There are still limitations in this study and some areas that could be discussed in the future. First, this study only examined the effect of financial linkage participation on credit rationing; it did not further explore whether differences in the depth of financial linkage participation have a differential effect on credit rationing. For example, the degree of cooperation between farmers and rural intermediaries, the extent to which they benefit from rural intermediaries, and the length of cooperation may have a differential effect on their reputation incentives; thus, whether this also has a differential effect on credit rationing should be examined. In addition, it would be worthwhile to further investigate the factors influencing the willingness of rural intermediaries to provide financial linkage services. Why are some rural intermediaries willing to provide financial linkages while others are not, and what are the motivations of rural intermediaries to provide financial linkages? All these questions are worthy of further research.