Research ArticleHerd Behavior and Its Effects on Default in Chinese Microcredit

Ya Bu 1 Shouzhe Zhang 2 and Anzhong Huang 1

1School of Economics and Management Jiangsu University of Science and Technology Zhenjiang 212100 China2School of Economics and Management Chuzhou University Chuzhou 239000 China

Correspondence should be addressed to Shouzhe Zhang zszchzueducn

Received 13 May 2021 Accepted 2 August 2021 Published 24 August 2021

Academic Editor Shaohui Wang

Copyright copy 2021 Ya Bu et al (is is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Herd behavior means a mode of behavior will be infected among the individuals in a group and its existence and influence inspecific financial market are uncertain (e paper utilizes the logistic model to test the existence and influence of herd behavior ofborrowers in Chinese microcredit market We find that (1) herd behavior exists in Chinese microcredit market (2) the two mostimportant factors affecting the default of microcredit resulted from herd behavior are the nature of microcredit institutions andthe education level of borrowers and (3) herd behavior will increase the possibility of default but it is relatively small

1 Introduction

Herd behavior means a mode of behavior which will beinfected among the individuals in a group that is to say theindividuals in a group will think and act like most people Asfor the influence of herd behavior in financial marketsJegadeesh and Kim [1] regarded that herd behavior wouldaggravate market volatility undermine market stability andfurther increase the fragility of the financial system So herdbehavior is the trigger of financial crisis At present theacademic community has mainly focused on the three as-pects of influence of herd behavior in the financial market

First is the influence of herd behavior on the stockmarket(is kind of research mainly studies the risk of stock pricefluctuation and stock price collapse which resulted from herdbehavior (eg [2]) Second is the influence of herd behavioron commodity market (is kind of research focuses on theexistence or the influence on the commodity market (eg[3 4]) (ird is the influence of herd behavior on Internetfinance (is kind of research mainly takes the peer-to-peerlending as the research object and regards the herd behaviorwhich will result in higher risk of default (eg [5])

However we find that there are few scholars who havecarried out research studies on the influence of herd be-havior on the credit market In view of this this paper takesChinese microcredit as the research object to explore the

existence of herd behavior of borrowers and its effects ondefault (e reason for choosing microcredit as the mostresearch object is that microcredit is the most important toolassigned by the World Bank to help to construct inclusivefinance and alleviate poverty (erefore studying the impactof herd behavior on microcredit will help developingcountries to achieve poverty alleviation goals

2 Testing the Existence of Herd Behavior

21 Method Because the data availability of the LSV methodis stronger and easier to implement it is widely used inacademic circles to test the existence of herding behavior(ispaper adopts the modified LSV method by Wermers [6] (emethod has three specific dimensions first the herd behavioris measured by the time interval of applying for loanWith thetimes of application increasing the time interval betweenapplications will shorten which is used to demonstrate theexistence of herd behavior (eg [7]) (e second is to testwhether themarket share increases with time which is used todemonstrate the existence of herd behavior (eg Lee) and [5](e third is to use ldquoWhether to Get Subsequent Loanrdquo as thedependent variable to test herd behavior (eg [8])

(e quality of date must be high in measuring the ex-istence of herd behavior no matter through testing timeinterval or testing market share (erefore this paper

HindawiComplexityVolume 2021 Article ID 9229871 5 pageshttpsdoiorg10115520219229871

chooses ldquoWhether to Get Subsequent Loanrdquo as the depen-dent variable and ldquothe Times of Current Loanrdquo as the in-dependent variable (e null hypothesis is that if thedependent variable is significantly and positively influencedby the independent variable there is herd behavior inmicrocredit market otherwise there is no herd behavior

22 Variables and Econometric Model According to thecharacteristics of Chinese microcredit market the paperselects 15 control variables (e detailed descriptions of thenames codes and definitions of the variables are shown inTable 1

According to the selected variables the correspondinglogistic regression model is

lnP

1 minus P1113874 1113875 β0 + β1TCL + β2LA + β3LR + β4LL

+ β5NMI + β6OTD + +β7G + β8A + β9EL

+ β10FL + β11SLT + β12NRT + β13ALA

+ β14PA + β15TNB + β16BD + ε(1)

P 1 means the current loan will get subsequent loanwhile P 0 means no subsequent loan

23 Data Resource (e selected sample period is from thefirst quarter of 2013 to the fourth quarter of 2017(e data ofindependent variables and control variables are provided byChina Association of Microcredit (e preliminary empiricalresults show that the value of sig in the Hoster andLemeshow tests is obviously less than 005 It implies that themodel is not good in fitting the original data as a wholeMoreover many variables do not pass the significance test(erefore in order to ensure the reliability of the empiricalresults the common processing methods are stepwise re-gression and factor analysis (e common factor was formedby analysis and then the same common factor was used asan independent variable for logistic regression analysis

24 Results of the Factor Analysis Firstly the KMO andBartlett Test of Sphericity is carried out on the variables thataffect dependent variables According to the results in Ta-ble 2 the KMO statistic value is 0698 which is greater thanthe standard value of 05 so the factor analysis can beperformed (e approximate chi-square value is 240121092and the sig value is 0001 It can be seen that the sphericalhypothesis is rejected and there is a strong correlationamong variables so it is appropriate to carry out factoranalysis

(e common factor was extracted by principal com-ponent analysis (e test results in Table 3 indicate that 5common factors meet the requirements when setting factorswith extraction feature values greater than 1 At the sametime the results show that the cumulative contribution rate

of these 5 factors reaches 8383 which has good explan-atory power

(e factor loading matrix indicates that all selectedfactors have linear relationship with original variables whilethe rotated factor has a high load which can simplify theexplanation of the factor and is more beneficial to the ex-planation of various factors According to the rotatedcomponent matrix all the common factors (Fi) respec-tively include the variables the first (F1) LL ALA and PAthe second (F2) LA A and FLN the third (F3) LR G andTNB the fourth (F4) TCL NMI EL and BD the fifth (F5)OTD SLT and NRT

25 Results of Logistic Regression Taking the common factoras the independent variable the modified logistic model is

lnP

1 minus P1113874 1113875 β0 + β1F1 + β2F2 + β3F3 + β4F4 + β5F5 + ε

(2)

(e results of comprehensive test of coefficients shownin Table 4 indicate that the likelihood of the model is verylarger than the chi-square value and the p value is 0 whichshows that the equation has stronger significance (esummary of the model shows that the minus2 log likelihood valueis larger than the chi-square critical value in the table aboveand indicates that themodel fits well In addition the p valueof Hosmer and Lemeshow test is significantly greater than005 which indicates that there is no sufficient reason toreject the null hypothesis (ere is a significant differencebetween the predicted and observed values of the modelwhich suggests that the overall fitting effect of the model isgood

(e results in Table 5 indicate that the p values of firstand fifth common factors are greater than 005 and othersare smaller than 005 which suggest that the coefficientsof the model have good significance and their influencesare positive In particular the coefficient of the fourthcommon factor is much higher than other commonfactors which suggests that the TCL NMI EL and ELBDare the reasonable explanatory variables (e significantpositive influence also shows that with the increase ofthe current times of application the subsequent bor-rowing will also increase (erefore it is verified thatthere is obvious herd behavior in Chinese microcreditmarket

3 Effects of Herd Behavior on Default

31 Method Since it is not possible to directly obtain thetime when the borrower defaults and the informationprovided by the Microcredit Alliance related to defaults isonly the number of times the borrower defaults so thelogistic default model is the most appropriate selectionSince the influence of the first and fifth common factors isnot significant we will eliminate the variables contained inthe first and fifth common factors(e revised default modelis as follows

2 Complexity

Table 2 Results of KMO and Bartlett tests

KaiserndashMeyerndashOlkin measure of sampling adequacy 0698

Bartlett test of sphericityApproximate chi square 240121092

Df 149Sig 0001

Table 3 (e interpreted total variance

Common factorsInitial eigenvalue Sum of squares extracted Sum of squares rotated

Total Var () Cumulative sum () Total Var () Cumulative sum () Total Var () Cumulative sum ()1 312 2320 2320 352 2187 2187 305 2106 21062 236 1901 4221 301 1911 4098 265 1730 38363 183 1633 5854 285 1763 5861 214 1648 54834 149 1425 7280 190 1520 7381 168 1524 70075 101 1104 8383 153 1002 8383 121 1376 8383

Table 4 Results of logistic regression analysis

Comprehensive test of coefficients Summary of modelChi square df Sig minus2 log likelihood Cox and snell R square Nagelkerke R square

Step 1Step 513297 5 0001 1743170 0179 0301Bloc 513297 5 0001 Results of Hosmer and Lemeshow testModel 513297 5 0001 10561 7 0202

Table 5 Details of common factor

B SE Wals df Sig Exp (B)

Step 1

FAC1 minus0096 0061 0964 1 0632 0952FAC2 0094 0088 51086 1 0000lowastlowastlowast 2543FAC3 0215 0072 37209 1 0021lowast 1209FAC4 1203 0109 241753 1 0000lowastlowastlowast 6071FAC5 minus0102 0057 1002 1 0749 0970

Constant mdash 0091 233762 1 0000lowastlowastlowast 5437Note lowast lowastlowast and lowastlowastlowast denote the 10 5 and 1 significant level respectively

Table 1 Meanings codes and definition of the variables

Names of variable (code) DefinitionDependent variable Whether to get subsequent loan If there is subsequent loan the value is 1 otherwise 0Independentvariable

(e current times of application(TCL)

(e times of application for microcredit

Control variables

Loan amount (LA) (e amount of current applicationLoan interest (LR) (e interest rate of the microcreditLife of loan (LL) (e life of the microcredit to be granted

Nature of microcreditinstitutions (NMI)

If the microcredit institutions are government led the value is 1 otherwise0

Original times of default (OTD) (e total numbers of failing to repay on timeGender (G) (e borrowers are male the value is 1 female 0Age (A) (e ages of borrowers

Education level (EL) Bachelor degree or above the value is 1 below 0First lending or not (FLN) If it is first loan the value is 1 otherwise 0

(e successful loan times (SLT) (e successful loan times by nowNormal repayment times (NRT) Normal repayment times by nowAccumulated loan amount (ALA) (e accumulated amount of loan by now

(e prepared amount (PA) (e prepared amount of repaymentTotal number of borrowing (TNB) Total number of borrowing by now

Proportion of bad debt (BD) Proportion of bad debt of total loan

Complexity 3

lnY

1 minus Y1113874 1113875 β0 + β1TCL + β2LA + β3LR + β5NMI

+ β6OTD + β7G + β8A

+ β9EL + β10FL + β15TNB + β16BD

(3)

Y 1 means the borrower will default while Y 0means no default

32 Results (e results in Table 6 indicate that the value p issignificantly less than 005 which suggests that the com-prehensive test of model coefficients has passed(e value ofminus2 log likelihood is also greater than the previous chi-squarecritical value so the model fits well as a whole In additionthe p value of Hosmer and Lemeshow tests is also greaterthan 005 which shows that the overall goodness of fit of themodel is high

(e results in Table 7 suggest that the value of p issmaller than 005 and the coefficient is negative whichindicate that the bigger the value of TCL the less thepossibility of default (eoretically the herd behavior ofborrowers in Chinese microcredit market can reduce thepossibility of default However because the absolute value ofits coefficient is very small its actual impact on the possi-bility of default is also very small which suggests that theborrowerrsquos behavior is rational to some extent When mostborrowers can do this the market will automatically elim-inate those borrowers with high risk of default thus reducingthe risk of default in the entire market (erefore herdbehavior in China microcredit market can reduce thepossibility of default

(e borrowerrsquos ability also affects the possibility of de-fault (ere is a positive correlation between BD and thepossibility of default When a borrower has strong invest-ment ability he or she will choose the loan with lower defaultprobability and exclude the target with higher defaultprobability thus reducing the probability of default Anotherindicator of investorrsquos ability TNB is negatively correlatedwith default (is also shows that the more experienced theborrowers are the more they can choose the target with lesspossible default

At the same time EL is positively related to the prob-ability of default which means that the higher the value ofEL the higher the probability of default (is may be due tothe fact that borrowers understand the characteristics ofinvestorsrsquo preference for highly educated so low-educatedborrowers pay more attention to the default (e nature ofmicrocredit institutions has a significant impact on defaults(e government-led microcredit has higher possibility ofdefault while the nongovernment-led microcredit institu-tions are relatively lower

4 Discussion

From the findings we learn that herd behavior exists amongborrowers of Chinese microcredit and that herd behaviorhas an impact on default but it is not significant

Generally the applicants of microcredit are mainly smallbusiness owners and the poor and they have the charac-teristics of large population and residential dispersion so thecost of accessing information about them is higher which isthe most important factor resulting herd behavior In ad-dition conformity psychology trading mechanism as wellas some factors of microcredit itself etc are the factorsresulting herd behavior too However the two most im-portant factors with Chinese characteristics are the nature ofmicrocredit institutions and the education level ofborrowers

(e Chinese government has long implemented thepoverty alleviation policy through free financial subsidiesMeanwhile besides government-led type Chinese micro-credit institutions also include banking P2P Fintechcompanies etc among which only government-ledmicrocredit institutions are public ownership So the bor-rowers whose education levels are lower whose decision-making capacities are limited and prefer to conformitybehavior mistakenly believe that the government-ledmicrocredit is also similar to the free financial subsidies(erefore herd behavior is significant among them Andmore importantly the funds of government-led microcreditcome from the governmentrsquos finance and public ownershipmeans the property rights are not clear So when theborrower defaults the government has no correspondingmeasures even credit rating On the contrary the funds ofother types come from market financing with clear propertyrights So when the borrower defaults the institution willtake various measures or even resort to violence to recoverthe loan (erefore the default cost of the government-ledmicrocredit is very low which leads herd behavior ofborrowers to be more serious

Since the market share of government-led microcredit islimited when we take Chinese microcredit as a whole wefind that herd behavior has no serious impact on default Wesurmise that even if there are serious impacts of herd be-havior on default of government-led microcredit it may bediluted Whether it is the case or not the article does nothave a definite answer In addition this article takesmicrocredit as a whole and it is unable to answer somespecific questions (erefore we believe that future researchstudies can take into consideration the differences of herdbehavior in gender age region etc

In order to utilize microcredit to achieve poverty alle-viation more effectively the paper proposes the following

First utilize the positive role of the herd behavior in themicrocredit market Although the empirical results showthat the herd behavior is rational to some extent and canreduce the overall loan default the impact is very small(erefore it is very important to expand this positive in-fluence which can be practiced from two aspects On the onehand it can reduce the occurrence of irrational herd be-havior of investors on the other hand it can strengthen theleading role of key borrowers in the group

Second build a credit rating system and playing the roleof market supervision Herd behavior in Chinese micro-credit market and its negative impact on default are partlydue to the nonself-discipline of borrowers such as the

4 Complexity

influences of government-led microcredit and educationallevel (erefore the market should establish a credit ratingsystem and use the reputation restraint mechanism topromote self-discipline of borrowers

(ird strengthen government supervision (e opera-tion status of the platform also affects investorsrsquo herdingbehavior If the platform operation mechanism is notcomplete even if investors can make rational herding be-havior it may also cause negative impacts However thedomestic platform operation mechanism is quite differentwhich also leads to the conclusion that the herding behaviorin the auction and loan studied in this paper is not com-pletely consistent with the conclusion of other scholars(erefore strengthening the classified supervision of dif-ferent platforms and the supervision of prominent problemson platforms can strengthen the positive role of rationalinvestor herding behavior

Data Availability

(edata are provided by China Association ofMicro-Credithttpchinamfinet

Conflicts of Interest

(e authors declare that they have no conflicts of interest

Acknowledgments

(e research was one of midterm results of major researchproject of philosophy and social science in universities ofJiangsu Province ldquoResearch on the Optimization of theRegulatory PATH of TinTech Innovationsrdquo (2019SJZDA060)humanities and social science project prepared for universitiesand colleges by Anhui Education Bureaus called ldquoResearcheson the route of promoting the industrialization of ecologicalagriculture in Anhui Province by Fintech (SK2018A0418)rdquo aswell as the Humanities and Social Sciences Project of theMinistry of Education of the PRC ldquoResearch on the For-mation Mechanism of Microfinance Poverty Alleviation andInnovation of Jiangsu Practice Modelrdquo (20YJA790028)

References

[1] N Jegadeesh and W Kim ldquoDo analysts herd An analysis ofrecommendations and market reactionsrdquo Review of FinancialStudies vol 23 no 2 pp 901ndash937 2009

[2] X Nianxing Y Shangyao and Y Zhihong ldquoHerd behavior ofinstitutional investors and risk of stock price collapserdquoManagement World vol 7 pp 31ndash43 2013

[3] S Ke and J Huang ldquoResearch on the herd behavior of Chinesereal estate market based on non-linear test model CSADrdquoManagement Reviews vol 24 no 9 pp 19ndash25 2012

[4] X Liu G Filler and M Odening ldquoTesting for speculativebubbles in agricultural commodity prices a regime switchingapproachrdquo Agricultural Finance Review vol 73 no 1pp 179ndash200 2013

[5] X Lin X Li and Z Zheng Evaluating borrowerrsquos default risk inpeer-to-peer lending evidence from a lending platform inChinardquo Applied Economics vol 49 no 35 pp 1ndash8 2016

[6] R Wermers ldquoMutual fund herding and the impact on stockpricesrdquo3e Journal of Finance vol 54 no 2 pp 581ndash622 1999

[7] B Luo and Z Lin ldquoA decision tree model for herd behavior andempirical evidence from the online P2P lending marketrdquo In-formation Systems and E-Business Management vol 11 no 1pp 141ndash160 2013

[8] M Herzenstein U M Dholakia and R L Andrews ldquoStrategicherding behavior in peer-to-peer loan auctionsrdquo Journal ofInteractive Marketing vol 25 no 1 pp 27ndash36 2011

Table 6 Results of logistic regression analysis

Comprehensive test of coefficients Summary of modelChi square df Sig minus2 log likelihood Cox and snell R square Nagelkerke R square

Step 1Step 257438 12 0001 1550232 0120 0203Bloc 257438 12 0001 Results of Hosmer and Lemeshow testsModel 257438 12 0001 14584 8 0068

Table 7 Variables in the model

B SE Wals df Sig Exp (B)

Step 1

TCL 0005 0001 24250 1 0001lowastlowastlowast 1000NMI 0254 0000 33542 1 0001lowastlowastlowast 1076EL 0107 0094 15209 1 0019lowastlowast 0968TNB 0034 0086 13095 1 0023lowastlowast 0953

Note the paper only lists the factors whose influence is significant

Complexity 5

Table 2 Results of KMO and Bartlett tests

KaiserndashMeyerndashOlkin measure of sampling adequacy 0698

Bartlett test of sphericityApproximate chi square 240121092

Df 149Sig 0001

Table 3 (e interpreted total variance

Common factorsInitial eigenvalue Sum of squares extracted Sum of squares rotated

Total Var () Cumulative sum () Total Var () Cumulative sum () Total Var () Cumulative sum ()1 312 2320 2320 352 2187 2187 305 2106 21062 236 1901 4221 301 1911 4098 265 1730 38363 183 1633 5854 285 1763 5861 214 1648 54834 149 1425 7280 190 1520 7381 168 1524 70075 101 1104 8383 153 1002 8383 121 1376 8383

Table 4 Results of logistic regression analysis

Comprehensive test of coefficients Summary of modelChi square df Sig minus2 log likelihood Cox and snell R square Nagelkerke R square

Step 1Step 513297 5 0001 1743170 0179 0301Bloc 513297 5 0001 Results of Hosmer and Lemeshow testModel 513297 5 0001 10561 7 0202

Table 5 Details of common factor

B SE Wals df Sig Exp (B)

Step 1

FAC1 minus0096 0061 0964 1 0632 0952FAC2 0094 0088 51086 1 0000lowastlowastlowast 2543FAC3 0215 0072 37209 1 0021lowast 1209FAC4 1203 0109 241753 1 0000lowastlowastlowast 6071FAC5 minus0102 0057 1002 1 0749 0970

Constant mdash 0091 233762 1 0000lowastlowastlowast 5437Note lowast lowastlowast and lowastlowastlowast denote the 10 5 and 1 significant level respectively

Table 1 Meanings codes and definition of the variables

Names of variable (code) DefinitionDependent variable Whether to get subsequent loan If there is subsequent loan the value is 1 otherwise 0Independentvariable

(e current times of application(TCL)

(e times of application for microcredit

Control variables

Loan amount (LA) (e amount of current applicationLoan interest (LR) (e interest rate of the microcreditLife of loan (LL) (e life of the microcredit to be granted

Nature of microcreditinstitutions (NMI)

If the microcredit institutions are government led the value is 1 otherwise0

Original times of default (OTD) (e total numbers of failing to repay on timeGender (G) (e borrowers are male the value is 1 female 0Age (A) (e ages of borrowers

Education level (EL) Bachelor degree or above the value is 1 below 0First lending or not (FLN) If it is first loan the value is 1 otherwise 0

(e successful loan times (SLT) (e successful loan times by nowNormal repayment times (NRT) Normal repayment times by nowAccumulated loan amount (ALA) (e accumulated amount of loan by now

(e prepared amount (PA) (e prepared amount of repaymentTotal number of borrowing (TNB) Total number of borrowing by now

Proportion of bad debt (BD) Proportion of bad debt of total loan

Complexity 3

lnY

1 minus Y1113874 1113875 β0 + β1TCL + β2LA + β3LR + β5NMI

+ β6OTD + β7G + β8A

+ β9EL + β10FL + β15TNB + β16BD

(3)

Y 1 means the borrower will default while Y 0means no default

32 Results (e results in Table 6 indicate that the value p issignificantly less than 005 which suggests that the com-prehensive test of model coefficients has passed(e value ofminus2 log likelihood is also greater than the previous chi-squarecritical value so the model fits well as a whole In additionthe p value of Hosmer and Lemeshow tests is also greaterthan 005 which shows that the overall goodness of fit of themodel is high

(e results in Table 7 suggest that the value of p issmaller than 005 and the coefficient is negative whichindicate that the bigger the value of TCL the less thepossibility of default (eoretically the herd behavior ofborrowers in Chinese microcredit market can reduce thepossibility of default However because the absolute value ofits coefficient is very small its actual impact on the possi-bility of default is also very small which suggests that theborrowerrsquos behavior is rational to some extent When mostborrowers can do this the market will automatically elim-inate those borrowers with high risk of default thus reducingthe risk of default in the entire market (erefore herdbehavior in China microcredit market can reduce thepossibility of default

(e borrowerrsquos ability also affects the possibility of de-fault (ere is a positive correlation between BD and thepossibility of default When a borrower has strong invest-ment ability he or she will choose the loan with lower defaultprobability and exclude the target with higher defaultprobability thus reducing the probability of default Anotherindicator of investorrsquos ability TNB is negatively correlatedwith default (is also shows that the more experienced theborrowers are the more they can choose the target with lesspossible default

At the same time EL is positively related to the prob-ability of default which means that the higher the value ofEL the higher the probability of default (is may be due tothe fact that borrowers understand the characteristics ofinvestorsrsquo preference for highly educated so low-educatedborrowers pay more attention to the default (e nature ofmicrocredit institutions has a significant impact on defaults(e government-led microcredit has higher possibility ofdefault while the nongovernment-led microcredit institu-tions are relatively lower

4 Discussion

From the findings we learn that herd behavior exists amongborrowers of Chinese microcredit and that herd behaviorhas an impact on default but it is not significant

Generally the applicants of microcredit are mainly smallbusiness owners and the poor and they have the charac-teristics of large population and residential dispersion so thecost of accessing information about them is higher which isthe most important factor resulting herd behavior In ad-dition conformity psychology trading mechanism as wellas some factors of microcredit itself etc are the factorsresulting herd behavior too However the two most im-portant factors with Chinese characteristics are the nature ofmicrocredit institutions and the education level ofborrowers

(e Chinese government has long implemented thepoverty alleviation policy through free financial subsidiesMeanwhile besides government-led type Chinese micro-credit institutions also include banking P2P Fintechcompanies etc among which only government-ledmicrocredit institutions are public ownership So the bor-rowers whose education levels are lower whose decision-making capacities are limited and prefer to conformitybehavior mistakenly believe that the government-ledmicrocredit is also similar to the free financial subsidies(erefore herd behavior is significant among them Andmore importantly the funds of government-led microcreditcome from the governmentrsquos finance and public ownershipmeans the property rights are not clear So when theborrower defaults the government has no correspondingmeasures even credit rating On the contrary the funds ofother types come from market financing with clear propertyrights So when the borrower defaults the institution willtake various measures or even resort to violence to recoverthe loan (erefore the default cost of the government-ledmicrocredit is very low which leads herd behavior ofborrowers to be more serious

Since the market share of government-led microcredit islimited when we take Chinese microcredit as a whole wefind that herd behavior has no serious impact on default Wesurmise that even if there are serious impacts of herd be-havior on default of government-led microcredit it may bediluted Whether it is the case or not the article does nothave a definite answer In addition this article takesmicrocredit as a whole and it is unable to answer somespecific questions (erefore we believe that future researchstudies can take into consideration the differences of herdbehavior in gender age region etc

In order to utilize microcredit to achieve poverty alle-viation more effectively the paper proposes the following

First utilize the positive role of the herd behavior in themicrocredit market Although the empirical results showthat the herd behavior is rational to some extent and canreduce the overall loan default the impact is very small(erefore it is very important to expand this positive in-fluence which can be practiced from two aspects On the onehand it can reduce the occurrence of irrational herd be-havior of investors on the other hand it can strengthen theleading role of key borrowers in the group

Second build a credit rating system and playing the roleof market supervision Herd behavior in Chinese micro-credit market and its negative impact on default are partlydue to the nonself-discipline of borrowers such as the

4 Complexity

influences of government-led microcredit and educationallevel (erefore the market should establish a credit ratingsystem and use the reputation restraint mechanism topromote self-discipline of borrowers

(ird strengthen government supervision (e opera-tion status of the platform also affects investorsrsquo herdingbehavior If the platform operation mechanism is notcomplete even if investors can make rational herding be-havior it may also cause negative impacts However thedomestic platform operation mechanism is quite differentwhich also leads to the conclusion that the herding behaviorin the auction and loan studied in this paper is not com-pletely consistent with the conclusion of other scholars(erefore strengthening the classified supervision of dif-ferent platforms and the supervision of prominent problemson platforms can strengthen the positive role of rationalinvestor herding behavior

Data Availability

(edata are provided by China Association ofMicro-Credithttpchinamfinet

Conflicts of Interest

(e authors declare that they have no conflicts of interest

Acknowledgments

(e research was one of midterm results of major researchproject of philosophy and social science in universities ofJiangsu Province ldquoResearch on the Optimization of theRegulatory PATH of TinTech Innovationsrdquo (2019SJZDA060)humanities and social science project prepared for universitiesand colleges by Anhui Education Bureaus called ldquoResearcheson the route of promoting the industrialization of ecologicalagriculture in Anhui Province by Fintech (SK2018A0418)rdquo aswell as the Humanities and Social Sciences Project of theMinistry of Education of the PRC ldquoResearch on the For-mation Mechanism of Microfinance Poverty Alleviation andInnovation of Jiangsu Practice Modelrdquo (20YJA790028)

References

[1] N Jegadeesh and W Kim ldquoDo analysts herd An analysis ofrecommendations and market reactionsrdquo Review of FinancialStudies vol 23 no 2 pp 901ndash937 2009

[2] X Nianxing Y Shangyao and Y Zhihong ldquoHerd behavior ofinstitutional investors and risk of stock price collapserdquoManagement World vol 7 pp 31ndash43 2013

[3] S Ke and J Huang ldquoResearch on the herd behavior of Chinesereal estate market based on non-linear test model CSADrdquoManagement Reviews vol 24 no 9 pp 19ndash25 2012

[4] X Liu G Filler and M Odening ldquoTesting for speculativebubbles in agricultural commodity prices a regime switchingapproachrdquo Agricultural Finance Review vol 73 no 1pp 179ndash200 2013

[5] X Lin X Li and Z Zheng Evaluating borrowerrsquos default risk inpeer-to-peer lending evidence from a lending platform inChinardquo Applied Economics vol 49 no 35 pp 1ndash8 2016

[6] R Wermers ldquoMutual fund herding and the impact on stockpricesrdquo3e Journal of Finance vol 54 no 2 pp 581ndash622 1999

[7] B Luo and Z Lin ldquoA decision tree model for herd behavior andempirical evidence from the online P2P lending marketrdquo In-formation Systems and E-Business Management vol 11 no 1pp 141ndash160 2013

[8] M Herzenstein U M Dholakia and R L Andrews ldquoStrategicherding behavior in peer-to-peer loan auctionsrdquo Journal ofInteractive Marketing vol 25 no 1 pp 27ndash36 2011

Table 6 Results of logistic regression analysis

Comprehensive test of coefficients Summary of modelChi square df Sig minus2 log likelihood Cox and snell R square Nagelkerke R square

Step 1Step 257438 12 0001 1550232 0120 0203Bloc 257438 12 0001 Results of Hosmer and Lemeshow testsModel 257438 12 0001 14584 8 0068

Table 7 Variables in the model

B SE Wals df Sig Exp (B)

Step 1

TCL 0005 0001 24250 1 0001lowastlowastlowast 1000NMI 0254 0000 33542 1 0001lowastlowastlowast 1076EL 0107 0094 15209 1 0019lowastlowast 0968TNB 0034 0086 13095 1 0023lowastlowast 0953

Note the paper only lists the factors whose influence is significant

Complexity 5

lnY

1 minus Y1113874 1113875 β0 + β1TCL + β2LA + β3LR + β5NMI

+ β6OTD + β7G + β8A

+ β9EL + β10FL + β15TNB + β16BD

(3)

Y 1 means the borrower will default while Y 0means no default

32 Results (e results in Table 6 indicate that the value p issignificantly less than 005 which suggests that the com-prehensive test of model coefficients has passed(e value ofminus2 log likelihood is also greater than the previous chi-squarecritical value so the model fits well as a whole In additionthe p value of Hosmer and Lemeshow tests is also greaterthan 005 which shows that the overall goodness of fit of themodel is high

(e results in Table 7 suggest that the value of p issmaller than 005 and the coefficient is negative whichindicate that the bigger the value of TCL the less thepossibility of default (eoretically the herd behavior ofborrowers in Chinese microcredit market can reduce thepossibility of default However because the absolute value ofits coefficient is very small its actual impact on the possi-bility of default is also very small which suggests that theborrowerrsquos behavior is rational to some extent When mostborrowers can do this the market will automatically elim-inate those borrowers with high risk of default thus reducingthe risk of default in the entire market (erefore herdbehavior in China microcredit market can reduce thepossibility of default

(e borrowerrsquos ability also affects the possibility of de-fault (ere is a positive correlation between BD and thepossibility of default When a borrower has strong invest-ment ability he or she will choose the loan with lower defaultprobability and exclude the target with higher defaultprobability thus reducing the probability of default Anotherindicator of investorrsquos ability TNB is negatively correlatedwith default (is also shows that the more experienced theborrowers are the more they can choose the target with lesspossible default

At the same time EL is positively related to the prob-ability of default which means that the higher the value ofEL the higher the probability of default (is may be due tothe fact that borrowers understand the characteristics ofinvestorsrsquo preference for highly educated so low-educatedborrowers pay more attention to the default (e nature ofmicrocredit institutions has a significant impact on defaults(e government-led microcredit has higher possibility ofdefault while the nongovernment-led microcredit institu-tions are relatively lower

4 Discussion

From the findings we learn that herd behavior exists amongborrowers of Chinese microcredit and that herd behaviorhas an impact on default but it is not significant

Generally the applicants of microcredit are mainly smallbusiness owners and the poor and they have the charac-teristics of large population and residential dispersion so thecost of accessing information about them is higher which isthe most important factor resulting herd behavior In ad-dition conformity psychology trading mechanism as wellas some factors of microcredit itself etc are the factorsresulting herd behavior too However the two most im-portant factors with Chinese characteristics are the nature ofmicrocredit institutions and the education level ofborrowers

(e Chinese government has long implemented thepoverty alleviation policy through free financial subsidiesMeanwhile besides government-led type Chinese micro-credit institutions also include banking P2P Fintechcompanies etc among which only government-ledmicrocredit institutions are public ownership So the bor-rowers whose education levels are lower whose decision-making capacities are limited and prefer to conformitybehavior mistakenly believe that the government-ledmicrocredit is also similar to the free financial subsidies(erefore herd behavior is significant among them Andmore importantly the funds of government-led microcreditcome from the governmentrsquos finance and public ownershipmeans the property rights are not clear So when theborrower defaults the government has no correspondingmeasures even credit rating On the contrary the funds ofother types come from market financing with clear propertyrights So when the borrower defaults the institution willtake various measures or even resort to violence to recoverthe loan (erefore the default cost of the government-ledmicrocredit is very low which leads herd behavior ofborrowers to be more serious

Since the market share of government-led microcredit islimited when we take Chinese microcredit as a whole wefind that herd behavior has no serious impact on default Wesurmise that even if there are serious impacts of herd be-havior on default of government-led microcredit it may bediluted Whether it is the case or not the article does nothave a definite answer In addition this article takesmicrocredit as a whole and it is unable to answer somespecific questions (erefore we believe that future researchstudies can take into consideration the differences of herdbehavior in gender age region etc

In order to utilize microcredit to achieve poverty alle-viation more effectively the paper proposes the following

First utilize the positive role of the herd behavior in themicrocredit market Although the empirical results showthat the herd behavior is rational to some extent and canreduce the overall loan default the impact is very small(erefore it is very important to expand this positive in-fluence which can be practiced from two aspects On the onehand it can reduce the occurrence of irrational herd be-havior of investors on the other hand it can strengthen theleading role of key borrowers in the group

Second build a credit rating system and playing the roleof market supervision Herd behavior in Chinese micro-credit market and its negative impact on default are partlydue to the nonself-discipline of borrowers such as the

4 Complexity

influences of government-led microcredit and educationallevel (erefore the market should establish a credit ratingsystem and use the reputation restraint mechanism topromote self-discipline of borrowers

(ird strengthen government supervision (e opera-tion status of the platform also affects investorsrsquo herdingbehavior If the platform operation mechanism is notcomplete even if investors can make rational herding be-havior it may also cause negative impacts However thedomestic platform operation mechanism is quite differentwhich also leads to the conclusion that the herding behaviorin the auction and loan studied in this paper is not com-pletely consistent with the conclusion of other scholars(erefore strengthening the classified supervision of dif-ferent platforms and the supervision of prominent problemson platforms can strengthen the positive role of rationalinvestor herding behavior

Data Availability

(edata are provided by China Association ofMicro-Credithttpchinamfinet

Conflicts of Interest

(e authors declare that they have no conflicts of interest

Acknowledgments

(e research was one of midterm results of major researchproject of philosophy and social science in universities ofJiangsu Province ldquoResearch on the Optimization of theRegulatory PATH of TinTech Innovationsrdquo (2019SJZDA060)humanities and social science project prepared for universitiesand colleges by Anhui Education Bureaus called ldquoResearcheson the route of promoting the industrialization of ecologicalagriculture in Anhui Province by Fintech (SK2018A0418)rdquo aswell as the Humanities and Social Sciences Project of theMinistry of Education of the PRC ldquoResearch on the For-mation Mechanism of Microfinance Poverty Alleviation andInnovation of Jiangsu Practice Modelrdquo (20YJA790028)

References

[1] N Jegadeesh and W Kim ldquoDo analysts herd An analysis ofrecommendations and market reactionsrdquo Review of FinancialStudies vol 23 no 2 pp 901ndash937 2009

[2] X Nianxing Y Shangyao and Y Zhihong ldquoHerd behavior ofinstitutional investors and risk of stock price collapserdquoManagement World vol 7 pp 31ndash43 2013

[3] S Ke and J Huang ldquoResearch on the herd behavior of Chinesereal estate market based on non-linear test model CSADrdquoManagement Reviews vol 24 no 9 pp 19ndash25 2012

[4] X Liu G Filler and M Odening ldquoTesting for speculativebubbles in agricultural commodity prices a regime switchingapproachrdquo Agricultural Finance Review vol 73 no 1pp 179ndash200 2013

[5] X Lin X Li and Z Zheng Evaluating borrowerrsquos default risk inpeer-to-peer lending evidence from a lending platform inChinardquo Applied Economics vol 49 no 35 pp 1ndash8 2016

[6] R Wermers ldquoMutual fund herding and the impact on stockpricesrdquo3e Journal of Finance vol 54 no 2 pp 581ndash622 1999

[7] B Luo and Z Lin ldquoA decision tree model for herd behavior andempirical evidence from the online P2P lending marketrdquo In-formation Systems and E-Business Management vol 11 no 1pp 141ndash160 2013

[8] M Herzenstein U M Dholakia and R L Andrews ldquoStrategicherding behavior in peer-to-peer loan auctionsrdquo Journal ofInteractive Marketing vol 25 no 1 pp 27ndash36 2011

Table 6 Results of logistic regression analysis

Comprehensive test of coefficients Summary of modelChi square df Sig minus2 log likelihood Cox and snell R square Nagelkerke R square

Step 1Step 257438 12 0001 1550232 0120 0203Bloc 257438 12 0001 Results of Hosmer and Lemeshow testsModel 257438 12 0001 14584 8 0068

Table 7 Variables in the model

B SE Wals df Sig Exp (B)

Step 1

TCL 0005 0001 24250 1 0001lowastlowastlowast 1000NMI 0254 0000 33542 1 0001lowastlowastlowast 1076EL 0107 0094 15209 1 0019lowastlowast 0968TNB 0034 0086 13095 1 0023lowastlowast 0953

Note the paper only lists the factors whose influence is significant

Complexity 5

influences of government-led microcredit and educationallevel (erefore the market should establish a credit ratingsystem and use the reputation restraint mechanism topromote self-discipline of borrowers

(ird strengthen government supervision (e opera-tion status of the platform also affects investorsrsquo herdingbehavior If the platform operation mechanism is notcomplete even if investors can make rational herding be-havior it may also cause negative impacts However thedomestic platform operation mechanism is quite differentwhich also leads to the conclusion that the herding behaviorin the auction and loan studied in this paper is not com-pletely consistent with the conclusion of other scholars(erefore strengthening the classified supervision of dif-ferent platforms and the supervision of prominent problemson platforms can strengthen the positive role of rationalinvestor herding behavior

Data Availability

(edata are provided by China Association ofMicro-Credithttpchinamfinet

Conflicts of Interest

(e authors declare that they have no conflicts of interest

Acknowledgments

(e research was one of midterm results of major researchproject of philosophy and social science in universities ofJiangsu Province ldquoResearch on the Optimization of theRegulatory PATH of TinTech Innovationsrdquo (2019SJZDA060)humanities and social science project prepared for universitiesand colleges by Anhui Education Bureaus called ldquoResearcheson the route of promoting the industrialization of ecologicalagriculture in Anhui Province by Fintech (SK2018A0418)rdquo aswell as the Humanities and Social Sciences Project of theMinistry of Education of the PRC ldquoResearch on the For-mation Mechanism of Microfinance Poverty Alleviation andInnovation of Jiangsu Practice Modelrdquo (20YJA790028)

References

[1] N Jegadeesh and W Kim ldquoDo analysts herd An analysis ofrecommendations and market reactionsrdquo Review of FinancialStudies vol 23 no 2 pp 901ndash937 2009

[2] X Nianxing Y Shangyao and Y Zhihong ldquoHerd behavior ofinstitutional investors and risk of stock price collapserdquoManagement World vol 7 pp 31ndash43 2013

[3] S Ke and J Huang ldquoResearch on the herd behavior of Chinesereal estate market based on non-linear test model CSADrdquoManagement Reviews vol 24 no 9 pp 19ndash25 2012

[4] X Liu G Filler and M Odening ldquoTesting for speculativebubbles in agricultural commodity prices a regime switchingapproachrdquo Agricultural Finance Review vol 73 no 1pp 179ndash200 2013

[5] X Lin X Li and Z Zheng Evaluating borrowerrsquos default risk inpeer-to-peer lending evidence from a lending platform inChinardquo Applied Economics vol 49 no 35 pp 1ndash8 2016

[6] R Wermers ldquoMutual fund herding and the impact on stockpricesrdquo3e Journal of Finance vol 54 no 2 pp 581ndash622 1999

[7] B Luo and Z Lin ldquoA decision tree model for herd behavior andempirical evidence from the online P2P lending marketrdquo In-formation Systems and E-Business Management vol 11 no 1pp 141ndash160 2013

[8] M Herzenstein U M Dholakia and R L Andrews ldquoStrategicherding behavior in peer-to-peer loan auctionsrdquo Journal ofInteractive Marketing vol 25 no 1 pp 27ndash36 2011

Table 6 Results of logistic regression analysis

Comprehensive test of coefficients Summary of modelChi square df Sig minus2 log likelihood Cox and snell R square Nagelkerke R square

Step 1Step 257438 12 0001 1550232 0120 0203Bloc 257438 12 0001 Results of Hosmer and Lemeshow testsModel 257438 12 0001 14584 8 0068

Table 7 Variables in the model

B SE Wals df Sig Exp (B)

Step 1

TCL 0005 0001 24250 1 0001lowastlowastlowast 1000NMI 0254 0000 33542 1 0001lowastlowastlowast 1076EL 0107 0094 15209 1 0019lowastlowast 0968TNB 0034 0086 13095 1 0023lowastlowast 0953

Note the paper only lists the factors whose influence is significant

Complexity 5