Microsoft Word CVHon Tom tat luan an EN 280222 MINISTRY OF EDUCATION AND TRAINING CAN THO UNIVERSITY o0o SUMMARY OF DOCTORAL DISSERTATION Major Finance – Banking Major code 62340201 CAO VAN HON IMPACT[.]
MINISTRY OF EDUCATION AND TRAINING CAN THO UNIVERSITY -o0o - SUMMARY OF DOCTORAL DISSERTATION Major: Finance – Banking Major code: 62340201 CAO VAN HON IMPACTS OF CREDIT RATIONING ON THE AMOUNT OF CAPITAL ALLOCATED TO INPUTS AND ON RICE YIELD OF FARMERS IN THE MEKONG RIVER DELTA Can Tho, 2022 THE RESEARCH WAS COMPLETED AT CANTHO UNIVERSITY Supervisor 1: Assoc Prof Le Khuong Ninh Supervisor 2: The dissertation will be defended before the council of the University at:…… …… hour to … hour, date month year 202 Discussant 1: Discussant 2: i LIST OF PAPERS RELATED TO THE DISSERTATION Domestic magazine Cao Van Hon and Le Khuong Ninh (2019) Impact of credit rationing on rice yield of farmers in the Mekong River Delta Asian journal of Economic and Banking, No 160, pages 4456 Cao Van Hon and Le Khuong Ninh (2020) Impact of credit rationing on capital allocated to inputs used by rice farmers in the Mekong River Delta, Vietnam Journal of Economics of Development, Vol.22 No.1, pp 47-60 Proceedings of the international conference Cao Van Hon and Le Khuong Ninh (2019) Impact of credit rationing on rice yield of farmers in the Mekong River Delta International conference on Business, Economics & Finance held at Can Tho University, December 2019 Cao Van Hon and Le Khuong Ninh (2020) Geographical distance and access to bank credit of rice farmers in Vietnam” International conference on Business, Economics & Finance held at Can Tho University, December 2020 ii Chapter INTRODUCTION 1.1 IMPORTANCE OF THE RESEARCH PROBLEM The Mekong River Delta is a key rice production area of Vietnam Each year, it contributes over 50 per cent of the total rice output and account for about 90 per cent of the exported rice (Statistical Yearbook, 2018) That achievement is the result of the Mekong River Delta being accreted by a large amount of alluvium from the Mekong River pouring into Vietnam through two tributaries of Tien and Hau rivers The Mekong River Delta is also favored by nature in terms of climate with harmonious rain and sun, helping irrigate and wash away harmful pathogens on crops In general, the Mekong River Delta has enough natural conditions to increase rice productivity and output However, the reality shows that most of the farmers only focus on production without paying attention to preparing enough capital to buy factors (or agricultural inputs) and develop output markets to ensure good prices for products while these markets fluctuate very unpredictably Therefore, rice farmers in the Mekong River Delta face many challenges because of their small production scale They often face the problem of fluctuating output prices, while the cooperation among the relevant agents (i.e., the state, scientists, businesses, and farmers) is very weak As a result, rice farmers remain poor because of low and precarious income, especially when the effects of climate change become more and more obvious Due to low income, farmers cannot afford to buy inputs following production requirements in terms of quantity, quality and time, leading to low productivity and the income of farmers Therefore, farmers want to borrow but are often denied access to credit due to asymmetric information and limited liability, which result in risk for credit institutions Consequently, only some rice farmers get enough credit while others are given just a proportion of their requests or completely rejected despite being willing to pay higher interest rates Then, credit rationing emerges as described by Stiglitz and Weiss (1981), among others Due to credit rationing, a number of rice farmers not have enough capital to acquire inputs for production so as to achieve maximum rice yield They may, then, contemplate two options, i.e., using less of all inputs (the scale effect) or less of the inputs that are not much vital to rice yield (the substitution effect) Reducing the number of inputs will not prevent diseases and ensure the essential nutritional needs of rice, so the yield is low and the quality of rice seeds is not satisfactory In other words, the above phenomenon lowers productivity and product quality as compared to the case of no credit rationing as demonstrated by the theories (Feder, 1985; Lee and Chambers, 1986; Carter, 1988; Fare et al., 1990; Carter and Niebe, 1990; Blancard et al., 2006; Ciaian et al., 2012) The above analysis shows that capital is very important for agricultural production, but farmers are credit-rationed due to asymmetric information and transaction costs, which affects their production Therefore, this is a topic that needs to be explored to serve as a scientific and practical basis for sustainable agricultural development policies However, according to our knowledge there have not been any in-depth studies on the relationship between credit rationing and the amount of capital allocated to inputs and rice yield of farmers in the country in general and in the Mekong River Delta in particular Therefore, the topic “Impacts of credit rationing on the amount of capital allocated to inputs and rice yield of farmers in the Mekong River Delta” was selected by the author 1.2 RESEARCH OBJECTIVES 1.2.1 General objective The purpose of this dissertation is to estimate the impact of credit rationing on the amount of capital allocated to inputs and on rice yield of farmers in the Mekong River Delta to propose solutions to help the farmers better use capital for production and improve rice yield 1.2.2 Specific objectives To meet the abovementioned general objective, the dissertation has the following specific objectives: (i) Assessing the current situation of credit rationing, the status quo of input use, and rice yield of farmers in the Mekong River Delta (ii) Estimating the impact of credit rationing on the amount of capital allocated to inputs used by rice farmers in the Mekong River Delta (iii) Estimating the impact of credit rationing on rice yield of farmers in the Mekong River Delta (iv) Proposing policy implications to help rice farmers in the Mekong River Delta to reduce credit rationing through rationalizing capital used to buy inputs and improve rice yield 1.3 OBJECT AND SCOPE OF THE RESEARCH 1.3.1 Research object The research object of the dissertation is the amount of capital allocated to inputs and rice yield of farmers in the Mekong River Delta under the influence of credit rationing by credit institutions 1.3.2 Research scope The thesis studies the effect of credit rationing on the amount of capital allocated to inputs and rice yield of farmers in the Mekong River Delta, including the provinces/cities of An Giang, Bac Lieu, Ca Mau, Can Tho, Hau Giang, Kien Giang, Soc Trang, Tra Vinh and Vinh Long at two time points of 2015 and 2018 to verify the science and practicality of the research problem that the thesis is interested in 1.4 CONTRIBUTION OF THE THESIS 1.4.1 Theoretical aspect The dissertation builds a theoretical basis on the impact of credit rationing on the amount of capital allocated to inputs and the impact of credit rationing on rice yield of farmers This theoretical basis is built mainly on the principles of Microeconomics and Microfinance with an application the credit sector, especially agricultural and rural credit 1.4.2 Practical aspect The results of estimating the effect of credit rationing on the amount of capital allocated to inputs and on rice yield of farmers in the Mekong River Delta are a scientific ground with a practical value This result helps scientists and policy makers to devise policies for sustainable agriculture and rural development In addition, the nonlinearity in the effect of credit rationing on the amount of capital allocated to inputs and on rice yield of farmers in the Mekong River Delta is also a new contribution of high scientific value This helps scientists as well as policy makers guide farmers to use the optimal amount of inputs to improve productivity and income while protecting people's health and the natural environment Chapter LITERATURE REVIEW Chapter THEORETICAL BACKGROUND AND RESEARCH METHODOLOGY 3.1 THEORETICAL BACKGROUND 3.1.3 Theoretical background of the impact of credit rationing on the amount of capital allocated to inputs Credit rationing leads to insufficient capital to buy inputs for production, so rice farmers must contemplate how to allocate the available capital to inputs so as to minimize this adverse effect To model that behaviour, the thesis is based on the arguments developed by Debertin (2012) Firstly, let consider a rice farmer who aims to minimize production cost due to credit rationing imposed by the credit institution This farmer’s production function is y f ( M , N ) , with y being rice output and M, N being inputs Then, the farmer’s minimum cost of production is: MinMPM NPN (3.1) M ,N given a constraint of y0 f ( M, N) , where PM and PN are the price of M and N , respectively To minimize the cost, the following Largangian expression can be used: MPM NPN y f ( M , N ) (3.2) The conditions for minimizing the cost read: f PM PM f M M M f PN PN f N N N y0 f ( M , N ) (3.3) Therefore: PM f M (3.4) PN f N (3.5) Dividing (4) by (3) gives: f PM M PN fN or fM f N PM PN (3.6) In Expression (3.6), f M is the marginal productivity of input M and fM / PM is the marginal productivity of one dong invested in input M Similarly, fN / PN the marginal productivity of one dong invested in input N According to a principle of microeconomics, f M / PM f N / PN means production cost is minimized given output y , so profit is maximized If credit markets are perfect, the source of financing is irrelevant or rice farmers have full access to credit (Foltz, 2004) In other words, rice farmers get sufficient capital to acquire inputs in order to produce the output that minimizes production cost and maximizes profit conforming to Expression (3.6) However, because rural credit markets are virtually imperfect due to information asymmetry and transaction cost, leading to advserse selection and moral hazard and causing risk for the credit institution (Stiglitz Weiss, 1981) As a result, it rations the amount of credit granted to rice farmers, so the latter does not have enough capital to buy the amount of inputs that satisfies Expression (3.6) Then, the scale effect emerges, affecting the scale of input use but not the relative input intensities – a phenomenon called symmetric credit rationing In concrete, the scale effect corresponds to the case in which farmers reduce both M and N , so rice yield definitely plunges Besides the scale effect, there also exists the substitution effect that affects both the level of input use and their relative intensities since more credit rationed inputs will be substituted by less ones (asymmetric credit rationing) In both cases, due to credit rationing rice farmers use an amount of inputs deviating from what is supposed to be the most efficient (i.e., maximizing profit) Moreover, the impact of credit rationing on the amount of capital allocated to inputs used by rice farmers are nonlinear because the marginal productivity varies according to the level of inputs applied Therefore, to estimate the impact of different degrees of credit rationing, in addition to identifying the treatment effect of using credit using the propensity scores, this paper also estimates the treatment effect of heterogenous intensities of credit rationing facing rice farmers 3.1.4 Theoretical background of the effect of credit rationing on rice yield of farmers To build the theoretical framework of the impact of credit rationing on rice yield of farmers, the thesis is based on the arguments developed by Blancard et al (2006), and Ciaian et al (2012) To that, a representative profit-maximising farmer with a possibility for input credit rationing is taken into consider ation This farmer’s production function is y f (M, N) , with y being rice output and M, N being inputs and farm’s profits function are: pf ( M , N ) PM M PN N (3.7) where, p are output price and PM , PN are input price M, N Due to credit rationed, farmer have C amount of money to buy inputs The value of C constrains the amount of inputs that farmers can use for production through the inequality: PM M PN N C (3.8) where α and δ are dummy variables which distinguish farmer with credit rationing between inputs ( M and N ) If and , it implies a symmetric farm’s credit rationing for both inputs M and N If a farmer only credit rationed with one input, it implies an asymmetry in credit rationing, i.e., and (credit rationing affects only M) or and (credit rationing affects only N ) Farmer maximises profits subject to credit rationing (3.8) according to LaGrangean: pf ( M , N ) PM M PN N (PM M PN N C ) (3.9) The optimal conditions for a credit rationing farm are as follows: f p PM pf M PM PM M M M f p PN pf N PN PN N N N P M M PN N C So: pf M (1 ) PM (3.10) pfN (1 )PN (3.11) From equations (3.10) and (3.11), pfM pfN is the marginal value product of both inputs and N If without credit rationing ( ), farmers will have enough production capital (with non-binding credit C ) and the marginal value product of both inputs M and N is equal to the input price itself ( pfM PM and pfN PN ) This result shows that without credit rationings, farmers M will maximize profits according to the common principle in microeconomics that marginal revenue is equal to marginal cost ( MR MC ) However, due to credit rationing, farmers not have enough capital to produce at the above optimal level, so rice yield will be negatively affected If a farmer is credit rationed both inputs ( , and ) from equations (3.10) and (3.11), follows that the marginal value product of both inputs is higher than the price of inputs in equilibrium if a farm is symmetrically credit rationed: pfM PM and pfN PN If a farmer is credit rationed only input M ( 1, 0) and , the marginal value product M is higher than the input price itself (pfM PM) and the marginal value product N is equal to the input price itself ( pfN PN ) Similarly, If a farmer is credit rationed only input N ( 0; 1) and , pfM PM and pfN PN Thus, if credit rationed of both inputs is the marginal value product of both inputs is higher than the price of inputs in equilibrium Then, farmers could potentially increase its profits by increasing input use but they cannot so because of credit rationing N EAS/ E EAS NB B ND D NF ES F I A NA III MA MB ES MF II MD EAS/ M EAS E Source: Ciaian et al (2012) Figure 3.1 Optimization of credit-rationed farmers On the basis of the theory presented previously, to make it easier to visualize, in Figure 3.1 the vertical axis shows the quantity of input N and the horizontal axis shows the quantity of input M used in production Without credit rationing ( ), farmers have enough money C to buy the amount of inputs that satisfies expressions (3.10) and (3.11) Then, the optimal output corresponds to the point D where the relative marginal value product of inputs M and N is equal to their relative market prices pf M P M pf N PN The equilibrium D is determined by the tangency between isoquant I and isocost EE The output level given by isoquant I represents the optimal farmers’output for the given input and output prices The amount of credit the farmer receives in this case has no effect on the farmer's output or rice yield Asymmetric credit rationing Consider the case of a farmer is credit rationed for one input Specifically, farmer is credit rationed as to input M ( and ) or input N ( and ) Due to the credit ration, farmers only have the amount of money C (C C ) to buy inputs Then, the amount C1 with the rationing in the expression (3.8) makes ( 0) The binding credit C affects one of the farm’s equilibrium conditions (3.10) and (3.11), depending on which input is credit rationed In Figure 3.1 credit rationing shifts of the isocost curve from EE to EASEAS , and isoquant from I to II This scale effect of an asymmetric credit rationing shifts the equilibrium from D to F , which is the tangent point between isocost EASEAS and isoquant II Isoquant II is below I implying a lower output with than without the binding (asymmetric) credit rationing At the same time, expressions (3.10) and (3.11) imply that: pf i ( C ) pf j ( C ) K Pi Pj K pf i ( C ) pf j ( C ) (1 1 ) AS Pi Pj , AS where input i is assumed to be credit rationed and the K -index is not credit rationed As a result, asymmetric credit rationing changes the relative marginal value product of inputs by the shadow price of credit rationed input In response to this, a farm substitutes the credit rationed input for a credit un-rationed input because pf i (C ) pf j (C ) K pf i (C1 ) pf j (C1 ) AS In Figure 3.1, the isocost curve rotates from EASEAS to EAS/ EAS/ The rotation of the isocost curve is determined by the adjustment of the relative input prices by shadow price from pM pN to K (1 1 ) pM pN , with AS PM P (1 i ) M PN K PN AS determined by the credit rationing the substitution effect C1 , The rotation takes place until point B, which is by way to move point F to point B This result is called In summary, an asymmetric credit rationing reduces the equilibrium output, decreases the credit rationed input, and may increase or decrease the credit un-rationed inputs Farmers expand the use of the credit non-rationed inputs, if the substitution effect dominates the scale effect In 3.2 RESEARCH METHODS 3.2.1 Data collection method To achieve the set objectives, the thesis uses both primary and secondary data For secondary data, the thesis uses data from relevant agencies such as the General Statistics Office, Provincial People’s Committees, Provincial Statistics Departments, the State Bank branches in the Mekong River Delta provinces, and related research at home and abroad The empirical model specified previously requires data on the determinants of access of rice farmers to credit and variables capturing the amount of capital allocated to inputs used by them The data used in this paper were collected through direct interviews with heads of 1,017 rice farmers in 2015 and 1,065 in 2018 randomly selected out of provinces and cities in the MRD (Table 3.1) through the questionairs modified after several pilot surveys Table 3.1 Survey sample structure 2015 Local Number of proportion farmers (per cent) An Giang 99 09,73 Bac Lieu 76 07,47 Ca Mau 85 08,36 Can Tho 110 10,82 Hau Giang 100 09,83 Kien Giang 91 08,95 Soc Trang 188 18,49 Tra Vinh 100 09,83 Vinh Long 168 16,52 Total 1.017 100,00 2018 Number of farmers 200 117 100 126 118 145 92 70 97 1.065 proportion (per cent) 18,78 10,99 09,39 11,83 11,08 13,62 08,64 06,57 09,11 100,00 Source: Research and Design 3.2.3 METHODOLOGIES It is hard to estimate the impact of credit rationing on the amount of capital allocated to inputs and rice yield of farmers due to the selection bias, implying the assignment to treatment (i.e., having a non-full access to credit) is non-random and depends the farmer’s traits This paper addresses this problem by using a relatively large size data set of 1,017 rice farmers in 2015 and 1065 rice farmer in 2018, which allows us to employ a semi-parametric propensity score matching (PSM) estimator PSM is commonly used in empirical studies (e.g., Rosenbaum and Rubin, 1983; Bento et al., 2007; Roberts and Key, 2008; Briggeman et al., 2009; Pufahl and Weiss, 2009; Katchova, 2010; Ciaian and Kancs, 2012) due to its ability to control the selection bias by constructing the counterfactual The advantage of the PSM estimation method is control the sampling bias 3.2.2.1 Impact of credit rationing on the amount of capital allocated to inputs Based on the abovementioned arguments, this paper specifies an empirical model to estimate the impact of pertinent factors on credit rationing facing rice farmers in the MRD as follows: creditrationing i 1landi income i residence i age i educationi genderi dis tan ce i socialpositioni exp erience i i (3.12) In Model (3.12), the dependent variable ( creditrationingi ) is constructed based on the ratio of the amount of credit granted to the farmer and the amount of credit he has applied for ( borrowratei ) If borrowratei , there is no credit rationing, so ( creditrationingi ) has a value of If borrowrate i , there is credit rationing, so ( creditrationingi ) has a value of The independent variable ( landi ) is the farmer's area of rice production, measured in 1,000m2 The variable ( incomei ) is the average income of household, measured in million VND/year The variable ( resedencei ) is the time the farmer has lived in the locality, measured in years The variable ( agei ) is the age of the household head, measured in years The variable ( educationi ) is the schooling of household head, measured in years The variable ( genderi ) is the gender of household head, if the head of household is male, the variable ( genderi ) takes the value 1, otherwise ( genderi ) takes the value The variable ( distan cei ) is the distance from the farmer's house to the nearest credit institution, measured in km The variable ( socialpositioni ) is a dummy variable, if the head of household has relatives, acquaintances working in state agencies or political organizations and vice versa, the variable ( socialpositioni ) takes the value 1, otherwise the variable ( socialpositioni ) takes the value The variable ( exp erience i ) is the number of years engaging in farmer's rice production, measured in years Model (3.12) will be estimated using Probit estimator to identify the propensity score Based on propensy score matching identified, the thesis use the kerrnel comparision method to identify the impact of credit rationing on the amount of capital allocated to inputs used and rice yield of farmers To provide a more precise picture of the impact of credit rationing on the amount of capital allocated to inputs used and rice yield of farmers, this paper uses the method conducted by Ciaian and Kans (2012) to divide the sample into categories with descending degrees of credit rationing Specifically, the first category includes rice farmers with borrowratei 0,2 , category with 0,2 borrowrate 0,4 , category with 0,4 borrowrate 0,6 , etc, Then compare category with category 1, category with category 2, etc 3.2.2.2 Method to estimate the effect of credit rationing on the amount of capital of farmers PSM is employed to determine the difference between the treatment and the control, which is called the average treatment effect on the treated (ATT), after controlling for differences 10 among them For a given rice farmer who gets a full access to credit, the observed mean amount of capital allocated to an input is E(Y0 D 0) and the unobserved (hypothetical) mean amount of capital allocated to an input is E(Y1 D 0) Similarly, for a given rice farmer who does not get a full access to credit, the observed mean outcome is E(Y1 D 1) and the unobserved (hypothetical) mean outcome that a rice farmer who does not get a full access to credit would have realized had they indeed has a full access to credit E(Y0 D 1) , where E () is the expectation operator in each of the expressions Following Rosenbaum and Rubin (1983), the parameter of interest in this paper is the ATT ATT E (Y1 Y0 D 1) E (Y1 D 1) E (Y0 D 1) The central interest of impact evaluation of this paper is not on E (Y0 D 0) but E(Y0 D 1) For that purpose, PSM uses balancing scores to extract the observed mean outcome of the farmers who not get a full access to credit and are most similar in observed traits to the farmers who get a full access to credit, i.e., it uses E (Y0 D 0) to estimate the counterfactual E(Y0 D 1) In order for the true parameter to be estimated, it is required that: E (Y0 D 1) E (Y0 D ) , which ensures that the ATT is free from self-selection bias The amount of capital allocated to inputs used by rice farmers, includes the amount of capital used to buy seed (seedi), fertilizer (fertilizeri), pesticide (pesticidei) and to hire labour (hiredlabouri) It is expected that seedi is not affected by credit rationing because this input is vital to rice yield, so the amount of seed used hardly varies according to the degree of credit rationing Different from seedi, fertilizeri is expected to be negatively influenced by credit rationing, implying that rice farmers will use less of this input due to the scale and substitution effects as credit is rationed As to the substition effect, when credit is rationed, rice farmers will use more of family labour to take care of the crop instead of fertilizers For persticide, the primary goal of rice farmers when using it is to control pests, so pesticidei is expected not to be influenced by the degree of credit rationing facing the farmer When the amount of fertilizers used decreases because of credit rationing, rice farmers use less of hired labours to fertilize Thus, hiredlabouri has an inverse relationship with the degree of credit rationing facing the farmer 3.2.2.3 Method to estimate the imfact of credit rationing on rice yield of farmers To build a theoretical model of the effect of credit rationings on rice yield of farmers, the thesis is based on the arguments developed by Blancard et al (2006) and Ciaian et al (2012) To that, analyze a household with the goal of maximizing profits in a credit-rationed conditions Then, for a given rice farmer who does not get a full access to credit, the observed rice yield is E(S1 D 1) and the assumed rice yield (unobserved) is E(S0 D 1) , Similarly, for a household with no credit rationings (control group), the observed rice yield is E(S0 D 0) and the assumed rice yield (unobserved) is E(S1 D 0) , where E () is the expectation, according to Rosenbaum and Rubin (1983), ATT is: ATT E(S1 S0 D 1) E(S1 D 1) E(S0 D 1) 11 The central interest of impact evaluation of this paper is not on E ( S D 0) but E(S0 D 1) For that purpose, PSM uses balancing scores to extract the observed mean outcome of the farmers who not get a full access to credit and are most similar in observed traits to the farmers who get a full access to credit, i.e., it uses, PSM uses E(S0 D 0) to estimate the counterfactual E(S0 D 1) , In order for the true parameter to be estimated, it is required that: E (S0 D 1) E (S0 D 0) which ensures that the ATT is free from self-selection bias As mentioned in the previous section, using the estimation results from the Probit regression, a probability for each farmer has a credit rationing (propensity score) is calculated Based on the propensity score, for each farmer in the treated group, a control farmer will be built using the kernel comparison tool (Kernel) This allows to compare each treated observation only with controls having similar values of observable traits Chapter STATUS OF RICE PRODUCTION PRODUCTIVITY AND CREDIT FOR FARMERS IN MEKONG RIVER DELTA 12 Chapter IMPACT OF CREDIT RATIONING ON CAPITAL ALLOCATED TO INPUTS AND ON RICE YIELD OF FARMERS IN MEKONG RIVER DELTA 5.1 DESCRIPTION OF RICE FARMERS 5.1.1 Characteristics of farmers 5.1.2 Credit status The average size of loans from credit institutions to a farmer is 13,46 million per year in 2015 and 43,56 million per year in 2018 In addition, the average number of loans of farmers in 2015 was 2.06 times/household/year and in 2018 was 1.38 times/household/year Transaction costs often go hand in hand with the number of loans, so a small number of loans (a lot of money/time) will help farmers save costs It creates incentives for farmers to access official capital and raise the awareness of capital use Although official credit is available, this type of credit does not seem to fully meet the needs of farmers’ production, so farmers still continue to buy agricultural goods on credit with a relatively large amount of money Specifically, the amount of money purchased on credit of farmers in 2015 was 30.28 million VND/year and in 2018 was 33.19 million VND/year Selling agricultural materials on credit (or lending in kind) to farmers brings risks to agents, but because of saving transaction costs, high selling prices and minimizing competitive pressure, This policy is commonly applied by material agents to farmers, albeit selectively (Le Khuong Ninh and Cao Van Hon, 2013) The average number of delays was 0.04 times in 2015 and 0.14 times in 2018 This information also implies that farmers are very less likely to miss their loan repayments, as formal credit is the source of financing are favored by farmers, due to low interest rates, abundant capital and sometimes accompanied by many incentives In 2015, as much as 241 farmers (23,70 per cent) were borrow the full the amount of credit they requested from credit institutions and 23,29 per cent (248 farmers) in 2018 Therefore, 734 famers (72,17 per cent) in 2015 and 711 famers in 2018 (66,76 per cent) were able to borrow only apart of the amount of credit they requested from credit institutions and 42 famers (4,13 per cent) in 2015 and 106 famers (9,95 per cent) in 2018 were denied loans 13 Table 5.1 Characteristics of rice famers in the MRD 2015 (N = 1.017) Criteria Age of household head (years) Number of people per household Schooling of household head (years) Residence in the locality (years) Area of agricultural land (m2) Distance to the nearest credit institution (km2) Income (million VND/person/year) Income from rice production (million VND/person/year) Rice yield (ton/ha) Formal loan (million dong/year) Number of loans at credit institutions Number of missed appointments when paying debt Amount of purchase on credit (million VND/year) Seed cost (million VND/ha) Fertilizer cost (million VND/ha) Pesticides Cost (million VND/ha) hired labor Cost (million VND/ha) Other costs (million VND/ha) 51,14 4,59 6,35 47,39 15,67 10,77 46,00 11,11 1,39 3,33 12,99 14,81 7,17 40,41 20,00 1,00 0,00 2,00 1,30 0,50 5,50 80,00 9,00 16,00 80,00 150,00 40,00 470,00 50,89 4,20 6,31 47,08 19,58 6,72 76,03 10,88 1,23 3,30 12,86 13,76 4,37 57,65 20,00 1.00 0,00 2,00 1,00 1,00 4,00 78,00 8,00 16,00 78,00 108,00 24,00 465,41 t-test (2018 sv 2015) –0,120 –6,272*** –0,129 –0,469 6,269*** –15,380*** 13,418*** 26,84 29,89 2,02 403,50 53,92 48,93 3,2 300,00 15,345*** 7,76 13,46 2,06 0,04 28,89 1,66 6,26 5,33 5,90 0,59 0,23 35,14 2,27 0,25 33,06 0,975 2,07 2,25 2,15 0,22 2,01 0,00 0,00 0,00 0,00 0,68 1,61 1,13 1,29 1,30 12,74 500,00 10,00 3,00 324,00 4,12 12,08 13,44 12,71 0,00 7,02 43,56 1,38 0,14 33,19 1,91 6,37 4,88 6,77 0,76 2,51 49,78 1,58 0,54 45,57 0,69 2,26 2,71 3,46 0,04 2,77 0,00 0,00 0,00 0,00 0,60 1,00 0,90 2,05 2,20 12,50 370,00 9,00 4,00 405,00 4,14 10,83 12,00 11,53 0,00 –2,908*** 19,694*** –7,435*** 5,730*** 2,501** 9,223*** 1,305 –4,266*** 7,487*** 12,365*** Mean SD Min 2018 (N = 1.065) Max Mean SD Min Max Source: The authors’ survey (2015 and 2018) 14 Table 5.2 Situation of credit rationing for farmers 2015 Number of Percentage Criteria observations of total (farmers) ( per cent) Non-rationed 241 23,70 Rationed 776 76,30 2.1 Can’t borrow 42 4,13 2.2 Borrow a part 734 72,17 Total 1.017 100,00 2018 Number of Percentage observations of total (farmers) ( per cent) 248 23,29 817 76,71 106 09,95 711 66,76 1.065 100,00 Source: The authors’ survey (2015) 5.1.4 Actual situation of capital allocation for inputs The main production costs of rice farmers are concentrated in seeds, fertilizers, pesticides and labor costs In fact, rice production activities of farmers also incur a number of other costs, but the proportion is insignificant in the total costs Table 5.1 shows that the average seed cost of farmers was 1.66 million VND/ha in 2015 and 1.91 million VND/ha in 2018 Average fertilizer cost was 6.26 million VND/ha in 2015 and 6.37 million VND/ha in 2018 Similarly, the average cost of pesticides used by farmers was 5.33 million VND/ha in 2015 and 4.88 million VND/ha in 2018 The average cost of hired labor used by farmers 6.77 million VND/ha in 2018 and 5.90 million VND/ha in 2015 5.1.5 Rice yield It is worth noting that the average rice yield of farmers has a difference of 7.76 tons/ha in 2015 and 7.02 tons/ha in 2018 This result may be due to the fact that rice yield in the Mekong River Delta has reached a threshold, so investments in production through inputs will not bring about practical results To overcome this disadvantage, the improvement of rice grain quality (such as the rational implementation of production of ST25 rice – the world's best rice variety) aims to take advantage of market opportunities (that is, quality rice can be sold at a good price), thereby increasing income for rice farmers Table 5.5 rice yield of farmers in the Mekong River Delta 2015 2018 TT Area (ha) Number of Number of Yield (ton/ha) Yield (ton/ha) farmers farmers ≤ 0,5 228 7,60 94 7,40 > 0,5–1,0 264 7,76 193 7,19 > 1,0–2,0 345 7,99 336 7,08 > 2,0–3,0 107 7,50 183 7.09 > 3,0–4,0 38 7,21 117 6,88 > 4,0–5,0 15 8,41 70 6,39 > 5,0 28 7,02 72 6,49 Source: The authors’ survey (2015 and 2018) 15 Table 5.5 shows that, when farmers have less than hectares of land, their rice yields are high (from 7.50 to 7.99 tons/ha in 2015 and from 7.08 to 7.40 tons/ha in 2018) Then the land area increases, the yield tends to decrease The lowest yield was in 2015 when farmers had to hectares of rice land (7.21 tons/ha) and in 2018 the lowest when farmers had to hectares (6,39 tons/ha) When a farmer’s rice land area is hectares or more, the yield tends to increase in 2018 and decrease in 2015 (it was 7.02 tons/ha in 2015 and 6.49 tons/hectare in 2018 ha) 5.2 IMPACT OF CREDIT RATIONING ON THE AMOUNT OF CAPITAL ALLOCATED TO INPUTS USED BY RICE FARMERS Several factors affect the access of rice farmers to credit The results shown in Table 5.6 are based on a Probit model, which identify the factors that affect the likelihood of a rice farmer getting access to credit Rice farmers with a lot of agricultural land are less credit rationed as land i has a negative coefficient at a significance level of per cent and per cent in 2015 and in 2018 Given the fact that incomei has a negative coefficient at a significance level of per cent and per cent in 2015 and in 2018, credit rationing is less likely to occur with rice farmers of high income Identically, educationi also has a negative coefficient at a significance level of per cent in 2015 and per cent in 2018, divulging that it is easier for better educated rice farmers to borrow from credit insttutions as compared to others Meanwhile, the positive coefficient at a significance level of 10 per cent of genderi in 2015, implies that credit rationing is more possible to appear with male rice farmers than with female ones As mentioned, the geograhical distance of the rice farmer from the nearest credit institution is a proxy for the degree of information asymmetry and transaction cost Table 5.6 shows that the further away a rice farmer is located from a credit institution, the more probable credit rationing occurs because distancei has a positive coefficient at a significance level of per cent both years 2015 and 2018 Other variables such as residencei , agei , postioni and experiencei have coefficients that are not statistically significant, so there is no conclusion about the effect of the duration that a farmer has resided in the locality, age and social position of household heads on the likelihood of credit rationing 5.3 Results of comparison the amount of capital allocated to inputs between rationed and non-credit-rationed farmers According to Table 5.7, the amount of capital allocated to fertilizer and hired labour is affected by credit rationing while that allocated to seed and pesticide is not, as expected Specifically, both fertilizeri and hirelabouri have negative coefficients at the same significance level of per cent in 2015 and in 2018, which implies that when facing credit rationing, farmers tend to use less of those inputs This finding is consistent with the theoretical background reviewed and identical to those of Lee and Champer (1986) and Blancard et al (2006) The coefficients of seed i and pesticidei are negative but not statistically significant in both years, 2015 and 2018, divulging that the amount of capital allocated to seed and pesticide is not influenced by credit rationing because farmers use these inputs rigidly for a fear of bad harvests that will certainly deprive them of income 16 Table 5.6 Determinants of credit rationing facing rice farmers Dependent variable: creditrationingi (1 if there is credit rationing and if otherwise) 2015 Variable Interpretation of variables Estimated coefficient 2018 Estimated coefficient Z-value Z-value C Constant 1,3963*** 4,16 1,0002*** 2,94 landi Area of land (1.000m2) –0,0072** –2,52 –0,0115*** –2,90 incomei Income (VND million) –0,0045** –4,51 –0,0028*** –3,56 residencei Residence in the locality (years) 0,0046 0,94 0,0006 0,12 agei Age of household head –0,0104 –1,57 0,0001 0,01 educationi Formal schooling of household head (years) –0,0424*** –2,96 –0,0315** –2,30 genderi Gender of household head (male = 1) 0,2486* 1,74 –0,1642 –1,06 distancei Distance to the nearest credit institution (km) 0,0248*** 3,83 0,0965*** 7,01 positioni Position of household head (yes = 1) –0,1292 0,38 0,0491 0,54 –0,0078 –1,52 –0,0062 –1,24 experiencei Number of years engaging in rice production Number of observations (N) 1.017 1.065 Significance level 0,000 0,000 –525,92407 –518,5192 Log likelihood Notes: (*), (**) and (***) designate statistical significance at the 10 per cent, per cent and per cent, respectively Source: The authors’ survey (2015 and 2018) 17 Table 5.7 Results of comparing the amount of capital allocated to inputs between rationed and non-credit-rationed farmers 2015 (N = 1.017) Estimated coefficient t-value 2018 (N = 1.065) Inputs Estimated t-value coefficient seedi –0,005 –1,070 –0,006 –1,432 fertilizeri –0,072*** –3,557 –0,083*** –7,028 pesticidei –0,015 –0,721 –0,025 –1,019 hiredlabouri –0,074*** –4,456 –0,156*** –5,140 Notes: (*), (**) and (***) designate statistical significance at the 10 per cent, per cent and per cent, respectively Source: The authors’ survey (2015 and 2018) 5.2.3 Results of comparing the amount of capital allocated to inputs between lowcredit constrained farmers and heavily credit-constrained farmers Table 5.9 shows that Fertilizer and hired labour have positive coefficients with varying degrees of significance for all comparisons at both the time points of 2015 and 2018 The amount of capital allocated to fertilizer increases as credit rationing drops but the magnitude of the increase is uneven since the amount of fertilizer applied depends on the growth stage of rice plants and the farmer’s own judgement All the comparisions for hired labour have positive coefficients, which means that when rice farmers face less severe credit rationing, they use more of hired labour to take care of their crop This reflects the reality of rice production in the MRD that is basically labour intensive Table 5.9 Results of comparing the amount of capital allocated to inputs between lowcredit rationed farmers and hight credit-constrained farmers Unit: million VND/1,000 m2 2015 (N = 1.017) 2018 (N = 1.065) Categories compared Seed Fertilizer Pesct icide Hired labour Seed (2) vs (1) 0,008 0,076*** 0,034 0,078** (3) vs (2) 0,007 0,091*** 0,026 (4) vs (3) 0,005 0,104*** (5) vs (4) 0,009 0,106** Fertilizer Pesctic ide Hired labour 0,006 0,078** 0,015 0,074** 0,92*** 0,003 0,103** 0,017 0,092** 0,013 0,089*** 0,014 0,171*** 0,016 0,131** 0,056 0,122** 0,009 0,120** 0,001 0,128** (6) vs (5) 0,021 0,061** 0,029 0,106** 0,015 0,167** 0,084 0,120*** Notes: (*), (**), (***) designate statistical significance at the 10 per cent, per cent and per cent, respectively Source: The authors’ survey (2015 and 2018) Thus, the estimated results in Table 5.9 show that fertilizer and hired labor are factors affected by credit rationing, seeds and pesticides cannot be concluded to be affected by credit rationing However, according to the theoretical basis just presented, non-credit- rationed inputs 18 can be used instead of credit-rationed inputs to ensure constant productivity The results show that credit-rationed factors are not substitutes Instead, these factors are complementary as farmers improve their credit rationings This result is similar to the classic study results of Ferder et al (1990) 5.3 IMPACTS OF CREDIT RATIONINGS ON ON RICE YIELD 5.3.1 Results of comparing of rice yield between rationed and non-credit-rationed farmers The results of comparing the average productivity between credit-rationed and non-creditrationed farmers were statistically significant at per cent for both 2015 and 2018 The average difference (ATT) is –0.100 in 2015 (equivalent to 11.49 per cent compared to the average rice yield of farmers in this year) and -0.084 in 2018 (11.35 per cent) showing that rice yield of creditrationed farmers decreases by 100kg/1,000m2 in 2015 and 84kg/1,000m2 in 2018 compared to noncredit-rationed farmers The coefficient estimates are negative in both 2015 and 2018, implying that credit remains an important factor for rice farmers in the Mekong River Delta This is consistent with the reality of rice production of farmers, with the theoretical background rviewed and identical to those of Ciaian et al (2012) for countries with countries transition economy in Central Europe Table 5.10 Results of comparing of rice yield between low-credit rationed and high-credit rationed farmers Unit: million VND/1,000 m2 Classify 2015 (N = 1.017) Coefficient Value t –0,100*** –6,201 2018 (N = 1.065) Coefficient Value t –0,084*** –5,440 Credit rationing vs no credit rationing Comparative pair (group) (2) sv (1) 0,109*** 3,217 0,093*** 3,508 (3) sv (2) 0,102*** 3,666 0,078*** 4,608 (4) sv (3) 0,110*** 3,071 0,086*** 2,954 (5) sv (4) 0,092*** 2,465 0,090** 2,040 (6) sv (5) 0,089*** 2,508 0,079** 2,215 Notes: (*), (**), (***) designate statistical significance at the 10 per cent, per cent and per cent, respectively Source: The authors’ survey (2015 and 2018) 5.3.2 Results of comparing of rice yield between low-credit rationed and high-credit rationed farmers Subdividing farmers with credit rationings into several groups to test the nonlinear effects of credit rationings on rice yield of farmer The results (Table 5.10) show all comparison pairs’ results are statistically significant The highest levels of yields variation are found between group and group in 2015-110 kg/1,000m2, and between group and group in 2018-93 kg/1,000m2 19 Chapter CONCLUSIONS AND POLICY IMPLICATIONS 6.1 Conclusion Credit rationing prevails for rice farmers in the MRD due to asymmetric information and limited liability Credit rationing affects the amount of capital allocated to inputs used in rice production of farmers through the scale and substitution effects The above behavior of the farmers will affect the rice yield because it can cause the deficiency and/or imbalance between the nutrients needed for the growth of the rice plant On the basis of an overview of theoretical and empirical research at home and abroad, the thesis has built a theoretical basis on the influence of credit rationings on the amount of capital allocated to inputs and rice yield of farmers Through data collected from nine provinces/cities in the Mekong River Delta, the thesis uses the propensity score method PSM (Propensity score matching) to estimate the effect of credit rationing on the amount of capital allocated to inputs and rice yield of farmers According to this method, the first uses Probit regression to determine the factors affecting credit rationing for farmers Then, the comparative method is uses to examine the difference between credit rationing farmers and non credit rationing farmers in terms of capital allocation for inputs and rice yield In addition, the thesis also divides credit rationing farmers into several groups with decreasing credit rationing to examine the nonlinear effect of credit rationing on the amount of capital allocated to inputs and rice yield Research results show that the majority of surveyed farmers in the area have credit rationings, specifically, in 2015 there were 76.30 per cent and in 2018 there were 76.71 per cent of farmers with credit rationings Besides the lack of capital, farmers also have difficulty in choosing inputs such as seeds, fertilizers and pesticides In addition, farmers are very interested in the labor factor because most of the young workers tend to go to urban areas to earn a living, make it even more difficult for the already difficult farmers in rice production Therefore, the cost of rice production of farmers is relatively high At the same time, the average rice yield of farmers reached 7.76 tons/ha in 2015 and 7.02 tons/ha in 2018 However, at different levels of land ownership, farmers have different rice productivity different Research results show that, farmers with production land area from to have high yield (> tons/ha) and when farmers have more than of productive land, the rice yield of lower farmers (< tons/ha) in 2018 Probit regression estimation results show that there are five factors affecting credit rationing to farmers in 2015 Specifically, they are land area, average income, level of education of household head, gender of household head, and distance to the nearest credit institution In 2018, there are four factors affecting credit rationing to farmers, namely land area, average income, level of education of farmers head, and distance to the nearest credit institution The estimation results of these two periods are relatively consistent Estimated results on the effects of credit rationing on the amount of capital allocated to inputs in rice production of farmers in the Mekong River Delta in 2015 and 2018 show that credit 20 rationing impacted the amount of capital allocated to fertilizer and hired labor The amount of capital allocated to fertilizer and hired labor of Credit-rationed farmers was decreased compared to that of non-credit-rationed farmers in 2015 and in 2018 In addition, when considering the effects of various degrees of credit rationing, all comparison pairs’ results are statistically significant for fertilizer and hired labor This means that the amount of capital allocated to fertilizers and hired labor increased while that seeds and pesticides remained unchanged Estimated results on the effects of credit rationing on rice yield of farmers in the Mekong River Delta in 2015 and 2018 show that credit rationing impacted rice yield: rice yield of creditrationed farmers decreases by 100kg/1,000m2 in 2015 and 84kg/1,000m2 in 2018 compared to noncredit-rationed farmers In addition, when considering the effects of various degrees of credit rationing, all comparison pairs’ results are statistically significant The highest levels of yields variation are found between group and group in 2015-110 kg/1,000m2, and between group and group in 2018-93 kg/1,000m2 6.2 POLICY IMPLICATIONS According to the analysis results, rice farmers in the Mekong River Delta lack information on inputs such as seeds, fertilizers and pesticides Therefore, farmers here use inputs that are not optimal in production The consequence of this problem is that farmers have low income In addition, the estimated results from the empirical research model show that farmers are rationed by factors such as income, distance from household to the nearest credit institution, production scale and education of the household head Credit rationing causes farmers to reduce the use of inputs (including fertilizers and hired labor) and rice yield Therefore, the thesis proposes some policy implications to reduce credit rationings, increase rice yield and rationally use inputs to help farmers promote production efficiency and have sufficient resources capital for the purchase of inputs 6.2.1 Reducing credit rationing for farmers The analysis results show that commercial banks are interested in the income, information and production land area of farmers which help banks control information about farmers to reduce risks Therefore, the role of state management agencies, commercial banks and farmers is very important Firstly, in order to easily access credit, farmers need to improve their income Therefore, farmers need to contract production with companies or link production chains to reduce income loss at intermediaries (rice storks and traders) For state management agencies (Government) it is necessary to develop the rice market and use consignment warehouses for farmers to rent Only then will the income of farmers be stable In addition, in order for farmers to reduce income loss when selling rice, the government should encourage enterprises to build processing factories at the raw material areas to help farmers deal directly with businesses In addition, the policy of stabilizing input material prices helps farmers reduce costs and ease production planning At that time, farmers can improve rice yield and increase income 21 Secondly, to improve credit restrictions, farmers need to join production groups or cooperatives (cooperatives) The research results show that farmers with large land areas are less likely to have credit rationings, but most farmers in the Mekong River Delta have small land areas Therefore, producing in groups or participating in cooperatives will help farmers accumulate the land area of other farmers for easy loans In addition, the cooperative will find inputs and outputs for farmers to reduce intermediary costs Commercial banks can lend in groups to secure collateral Each household has a small area of production land, but combining many farmers will get a large area However, not everyone in the group has the need to use capital, so those who need it will have enough capital for production This approach may not be agreed by the group members because the household does not need to borrow money, which will be risky Therefore, the bond between members as close relatives is necessary because there is a necessary trust in each other In addition, unsecured credit in agriculture should be expanded and closely monitored by state agencies In practice, unsecured credit only applies to poor farmers and the relatively small amount of credit is provided by the social policy bank These small loans are not enough to meet the needs of agricultural development Therefore, expanding unsecured credit both in terms of loan object and value is a good condition for farmers to use low-interest loans In addition, supervision by state agencies is also very important for farmers to access credit Currently, the state has a policy of unsecured lending to farmers (Decree 116/2018/ND-CP amending and supplementing a number of articles of Decree No 55/2015/ND-CP dated June 9th 2015 of the Government on policies to serve agricultural and rural development) up to 100 million VND However, for many reasons, it is difficult (almost impossible) for farmers to access this unsecured capital Therefore, the supervision of the competent authorities is a necessary requirement for commercial banks so that farmers have the opportunity to access credit Thirdly, to reduce credit constraints, farmers and commercial banks must find ways to reduce asymmetric information This helps commercial banks properly assess the creditworthiness of farmers and minimize risks Therefore, commercial banks need to expand their transaction points to the locality close to farmers and recruit members of the household as employees On the one hand this reduces the distance to farmers, on the other hand, reduces asymmetric information between farmers and commercial banks If it is not possible to build a network of activities in the locality, commercial banks can cooperate with material agents to lend farmers to buy agricultural materials Specifically, farmers borrow money from banks but only receive agricultural materials from local shops In addition, a commercial bank can set up a material dealer as a correspondent bank At that time, the bank lends to the agent, the agent lends to the household Currently, the rural credit market in the Mekong River Delta is underdeveloped, so the state supports commercial banks in this area For example, the state helps find the location to build the headquarters and subsidizes interest rates On the other hand, the state needs to invest in transport and communication infrastructure to reduce asymmetric information Convenient transportation infrastructure makes it easy for bank employees and farmers to travel At that time, the appraisal and customer care work is easier due to the reduction of asymmetric 22 ... uses data from relevant agencies such as the General Statistics Office, Provincial People’s Committees, Provincial Statistics Departments, the State Bank branches in the Mekong River Delta provinces,... underdeveloped, so the state supports commercial banks in this area For example, the state helps find the location to build the headquarters and subsidizes interest rates On the other hand, the state needs... ) pf j (C ) K pf i (C1 ) pf j (C1 ) AS In Figure 3.1, the isocost curve rotates from EASEAS to EAS/ EAS/ The rotation of the isocost curve is determined by the adjustment of the relative input