Love Match, Marriage Distance, and Marriage Payment: Evidence from Rural China

Abstract

The excessive marriage payment indicated by brideprice/dowry imposes a huge family financial burden, especially on rural households. Although some determinants have been examined, the relationship between matching types of couples and marriage payment is rarely known. The goal of this study is to analyze this relationship and the role of marriage distance in it. We adopt Logit, Tobit, and SUR models with the database across 70 years in rural China from a nearly national representative sample. The results show that love-match couples are 10.7% and 10.3% less likely to pay brideprice and dowry than that of parental matchmaking, respectively, and they pay less amount of brideprice and dowry. There is an inverted U-shaped relationship between marriage distance and brideprice/dowry, and love match has the largest correlation with marriage payment behaviors for couples with long marriage distance.

1. Introduction

Brideprice/dowry is increasingly attracting the attention of scholars and policymakers because of its pervasive prevalence in developing countries and the significant property transfers and bargaining it entails [1,2]. They can be several times larger than the annual income of rural households, which can be substantial enough to affect the welfare of women and a society’s distribution of wealth [3,4,5,6,7,8]. Thus, China’s No. 1 Central Document in recent years explicitly states that excessive brideprice has an adverse effect on a healthy social environment due to its huge economic burden imposed on rural households. The high brideprice makes the phenomenon of “poverty due to marriage” occur in rural China frequently, which is also one of the most important reasons for the decline of the marriage rate in China. This poses great challenges to the sustainable development of Chinese society and the rational change of population structure.

According to Becker’s marriage market theory framework [9,10,11], the means of matching could be one of the determinants of marriage payments. There are two ways of matching couples: love match and arranged marriages. Love match means a couple become married because of romantic love; however, arranged marriage means they become married due to parents choosing their marital partners [12]. In ancient China, most people got married based on parents’ order or matchmaker’s word (Fumu Zhiming, Meishuo Zhiyan). The couple did not even meet or knew little about each other before becoming married. Love match at that time was even not a moral or well-bred behavior. Love match and arranged marriage have a new meaning and performance in contemporary China. With the establishment of the People’s Republic of China, the proportion of love match gradually increased, especially after the Reform and Opening up and the implementation of the Marriage Law. However, there are still a considerable number of couples becoming married based on parents’ consideration of potential marriage partners’ economic situation or the pressure of social customs.

The relationship between match types of couples and marriage payments in other countries has been paid high attention. The co-existence of arranged marriage and dowry in South Asia has been well noted by many scholars [3,13,14,15,16]. Singh (1997) used data from 10 villages in India and found that the financial burden of arranging the daughters’ marriage on their parents has increased [17]. Some studies found that cash transfer at the time of marriage can potentially serve as a screening instrument to distinguish grooms of different qualities because some characteristics related to the quality of the groom are unobservable in ‘arranged’ marriage, such as personality and ability [12]. Diamond-Smith et al. (2020) conducted in-depth qualitative interviews in Nepal and found that husbands with semi-arranged (couples often talk or meet before marriage) have growing ambivalence about dowry [18].

Studies focusing on this issue in China are rare. To the best of our knowledge, there are only two empirical studies touching up on this topic in China. Wang studied the internal correlation between matching types and brideprice paid by the grooms’ family using case studies in rural areas of Sichuan and Henan provinces [19]. The survey shows that the love-match marriage model dominated by the children correlates with the low costs of brideprice, vice versa. The ontological value of intergenerational responsibility lays behind the two different marriage models in rural area. The other research found that the increasing popularity of love match plays a role in solving the marriage squeeze based on the case study method in a village in Shaanxi province [20].

There are some potential gaps to be narrowed in previous studies. Firstly, few studies in China have empirically tested the relationship between love match and marriage payment. Secondly, most previous studies ignore the substitution effect of wedding house investment and wedding reception expenditure on brideprice or dowry, which is not in line with reality. Thirdly, the relationship of love match on marriage payment may vary with the marriage distance, which measures the distance between the couple’s birthplaces. There are few studies focusing on the role of marriage distance in this relationship.

Therefore, the goal of this study is to estimate the correlation between matching types and marriage payment behaviors in rural China. Specifically, we first analyze the trends of brideprice and dowry in rural China from 1950 to 2015 based on field survey data with a nearly national representative sample. We then evaluate the impact of love match on brideprice and dowry. Finally, we explore the heterogeneity of love match on brideprice and dowry with different marriage distances.

Our work contributes to literature in the following ways. Firstly, our research adopts data covering nearly 70 years with a nearly nationally representative sample which will enrich literature by analyzing long trends of marriage payment in rural China. Secondly, to our best knowledge, this is the first study to assess the correlation between love match and brideprice/dowry based on relatively rigorous empirical means. Thirdly, it also investigates the heterogeneity of love match on brideprice/dowry between different marriage distances. The results have implications to further deduce burden due to marriage payment in rural China, even for other developing countries.

The rest of this paper is organized as follows. Section 2 presents conceptual framework and literature review. Section 3 introduces the methods, including sampling and data collection, definition of variables, and model specification. Section 4 shows descriptive analysis. Section 5 presents and discusses the results of empirical analysis. The conclusion and implications are in Section 6.

2. Conceptual Framework and Literature Review

Individuals or households maximize their expected utility when becoming married, given their own human capital and family background [21]. We simply categorize the expected utility of marriage into two types: emotional benefit including expected emotional support, such as harmony of the marriage felt by the groom/bride, and non-emotional benefit, which contains expected economic gains, net marriage payment and risk prevention [12,22]. There are many different combinations of these two kinds of benefits on the indifference curve of maximum utility. However, there are preference differences in the combination of these two kinds of benefits between different decision makers with the same utility [23]. It indicates that the marginal rate of substitution between emotional benefit and non-emotional benefit is different between a love match and an arranged marriage. If we define the U as the expected utility of a marriage union, we have:

U = non-emotional benefit + emotional benefit = expected economic output + net marriage payments + risk prevention + emotional benefit

Hence,

Net marriage payments = U − (expected economic output + risk prevention + emotional benefit)

Further,

Brideprice/Dowry = U − (expected economic output + emotional benefit + risk prevention) − Other marriage payments

Why do couples and their parents have different preferences? Firstly, marriage directly affects the welfare of their parents rather than simply forming a new family of two individuals. Many “goods” produced by the couple, including their labor market income, household goods and services, children, and elder care, can be sharable and beneficial to their parents [22,24,25]. Elderly care is mainly provided by adult children, especially in rural China. When parents are the primary decision makers, they will take the non-emotional benefit as the principal factor to make sure they can enjoy the parental goods [22,26]. The emotional attraction is not a parental good, which will be undervalued by parents. The love-match males care more about the attractiveness of their wife and the harmonious of their marriage than those of arranged marriage [22,26,27]. Secondly, marital search is costly and primary decision makers will bear most of the search cost. It is difficult for parents to estimate the emotional output, so they can only use quantifiable economic criteria as the screening criteria to reduce the search cost [12]. Therefore, different preferences of decision makers lead to different factors driving the maximization of utility function.

Expected economic output of a marriage can be regarded as “goods” produced by the couple. Based on the evidence from numerous studies, spouse human capital determines household productivity. Human capital, such as education, age, health, and training, is an important drive of income growth and determines individual productivity [28]. Let $h_m$ ≥ 0 and $h_f$ ≥ 0 denote the human capital of young man and woman, respectively. $h_m$ and $h_f$ jointly determine the total $f$ ($h_m,h_f$), which reflects both the couple’s household production output and joint income.

Marriage payment is the outlay of money and/or goods as part of the marriage process. It includes brideprice, dowry and other expenses, such as wedding reception and house or property investments [1,29,30,31,32]. These kinds of marriage payment are mutually substituted. Brideprice and dowry can be affected by the housing investment in rural China [33]. In Vietnam, houses, wedding receptions, and expenditure of wedding ceremonies are also other forms of marriage payment to supplement brideprice or dowry [31]. For example, the bride’s family will charge less for brideprice if the groom’s family builds a good wedding house, or the groom’s family will not claim dowry if the bride’s family decides to share the cost of the wedding reception.

Emotional benefit of a marriage can be interpreted as love or attraction between husband and wife and can be regarded as a function of love match [22,27]. It seems difficult to be accurately and quantifiably measured, compared with economic benefit and is often unpredictable based on commonly observed characteristics. Some previous studies, especially in modern western societies, used “whether a marriage ends up in divorce” as a natural measure of marital quality [34,35]. However, the extremely low divorce rate in rural China makes this measure less useful. Some other studies used harmony within a couple to measure emotional support in marriage [22,24,35]. Regardless of many measurements of marital quality, there are ample evidences that love romance is strongly associated with marital quality. Huang et al. (2017) found that couples who marry in a love match have a higher degree of couple harmony, fewer conflicts, and higher emotional support among urban couples compared with those are parental matchmaking. They also found supporting and robust evidence for this conclusion using six provinces sample of urban and rural couples in China [22]. Those who married for love are significantly happier and have higher marital satisfaction than those who have arranged marriage [36]. Data showed significant fitness benefits of mating with partners of an individual’s own choice, highlighting elevated behavioral compatibility between partners with free mate choice [37].

Marriage distance influences the bargaining of marriage payment. This comes from the two functions of marriage payment: insurance against the risk of marital breakdown and compensation for the loss of labor in original family [8]. Mixed results were found in previous studies. Those who leave the original intermarriage circle will face greater marital risks, including family abuse, social, cultural, and economic difficulties [38,39,40]. However, other studies found that a female migrant might marry a man distant from her natal family in exchange for a more desirable location and an improvement of her economic well-being. The man has achieved the goal of marriage and augmented his household’s labor supply at a cheap price [41,42].

This seems to make sense logically. A couple from the same area, within the same intermarriage circle, with similar personal characteristics and living habits, are more likely to have less conflict and stay married. As marriage distance increases, the similarity diminishes and the risk of divorce increases. However, more couples will tolerate each other to maintain the situation of marriage because brides are completely separated from their original living environment with a long marriage distance, which decreases the risk of divorce. Logically, wives’ returns to their original families after marriage are more negatively correlated with marriage distance. A bride with a smaller marriage distance is still able to provide economical and emotional support to her natal family after becoming married, such as helping her parents do domestic work and giving goods or money to her parents. On the contrary, the further a bride marries, the lower her family’s expectation of reward and help after her marriage.

The number of siblings has an impact on marriage payment. Under the traditional culture in China, parents have the responsibility and obligation to give all their children economic resources to promote their marriage [43,44]. Resource dilution theory holds the belief that the number of children determines the number of resources available to each child in a family since the resources in a family are limited [15,43,44]. If males’ family has more children, females’ family will change its pricing strategy and try to become more resources for her daughter and son-in-law to avoid being diluted by males’ siblings. Roy (2015) found that parents appear to compensate their daughters by giving daughters higher dowries because of their son’s exploitation of family resources [15]. Wei and Jiang (2017) found that there is intragenerational dilution exploitation between brothers using the data collected from 241 villages of nine provinces in China in 2014 [45]. Some case studies also found that intragenerational exploitation existed, which refers to the competition between brothers within the family or the exploitation of brothers over sisters [2,45,46,47].

Based on above framework and literature review, groom $m$ and bride $f$ have,

$$Bridepric{e}_m=f\left(Expected economic outpu{t}_m, Emotional benefi{t}_m,Risk Preventio{n}_m,OtherM{P}_m\right)$$

$$Dowr{y}_f=f\left(Expected economic outpu{t}_f, Emotional benefi{t}_f,Risk Preventio{n}_f,OtherM{P}_f\right)$$

where,

$$Expected economic outpu{t}_i=f\left(h_m,h_f\right)=f\left(ag{e}_m, schoolin{g}_m, sibiling{s}_m,ag{e}_f, schoolin{g}_f, sibiling{s}_f\dots \mathrm{.}\right)$$

$$Emotional benefi{t}_i=f\left(love match\right)$$

$$Risk preventio{n}_i=f\left(marriage\_distance\right)$$

Hence,

$$Bridepric{e}_m=f\left(Love match,marriage distance, h_m,h_f,OtherM{P}_m\right)+\epsilon$$

$$Dowr{y}_f=f\left(Love match,marriage distance, h_m,h_f,OtherM{P}_f\right)+\epsilon$$

Based on the conceptual framework, we proposed the following hypotheses:

Hypothesis 1 (H1).

Love-match couples are less likely to pay brideprice and dowry when they become married than those of parental matchmaking.

Hypothesis 2 (H2).

Love-match couples pay smaller amount of brideprice and dowry when they become married than those of parental matchmaking.

Hypothesis 3 (H3).

There is a heterogeneous effect of love match on brideprice and dowry with different marriage distances.

3. Data and Methods

3.1. Sampling and Data Collection

This paper uses the data from China Rural Development Survey (CRDS) 2016 collected by the authors in 2016. The survey covers 2000 households in 100 villages of 25 counties across 5 provinces in China based on a nationally representative sample using a multi-stage stratified cluster sampling. The sampling procedures are as follows.

First, each sample province was randomly selected from China’s major agroecological zones. They were Hebei, Jilin, Shaanxi, Jiangsu and Sichuan. Second, we randomly selected five sample counties from each sample province. All the counties in each province were listed by the enumeration team in descending order of per capita gross value of industrial output (GVIO), which is often more reliable than net per capita income, is a good predictor of development potential and standard of living [48]. Then, five counties were randomly selected from the list in each selected province. After that, the team chose sample townships from each selected county, with one “more well-off” group and one “poorer” group. Following the same procedure, we chose two villages in each selected township. Finally, the survey teams used the roster of villagers to randomly choose 20 households in each village and acquire their name, contact number, and address. If households randomly selected were absent, the survey teams contacted the household by telephone to make an appointment for an interview.

The survey team collected detailed information about each family member spanning three generations. The definition of three generations infers to the paper “Intergenerational Transmission of Education: The Case of Rural China” and published in China Economic Review [49]. This information includes age, gender, siblings and years of schooling. The team also conducted a detailed investigation into the marital histories of family members, including marital status, year of first marriage, whether the marriage is a love match, value of brideprice (for groom’s family), or dowry (for bride’s family), marriage distance, wedding house investments, and wedding reception expenditure.

In this study, we included only couples whose current marriages are their first marriage, because marriage payments were significantly less common among those who experienced marital dissolution. Finally, we had a sample of 5128 couples married between 1950 and 2015.

3.2. Definition of Variables

3.2.1. Marriage Payment Behavior

We use two types of variables to measure marriage payment behaviors: brideprice /dowry and value of brideprice/dowry. Although at the practical level, there is indeed a mutual compensation situation between the brideprice and dowry. However, considering that after brideprice is given to the brides’ family, the brides’ family can also choose not to pay any dowry. Therefore, we choose to treat the brideprice and dowry equivalently. In the following, we also use the seemingly unrelated regression (SUR) method to jointly estimate to deal with this bias. If paying brideprice/dowry when a couple becomes married, brideprice/dowry equals 1 or 0 otherwise. We use continuous variables to measure value of brideprice/dowry.

We collected the nominal value of brideprice/dowry in the survey. To remove the effect of changes in prices during this period, we use the consumer price index to express the number of investments in other years. All brideprice/dowry prior to 1985 are converted using the general retail price index, which was first calculated in 1950 and is the standard index for conversions. Brideprice/dowry from 1985 onwards are converted using the rural CPI, a more accurate reflection of rural prices that was first introduced in 1985 [5,6]. We also create a logarithmic transformation of the value of brideprice/dowry.

3.2.2. Love Match

The key variable in this study is love match, which indicates whether the couple become married due to love or not. In the survey, we collected information on how an individual had got to know his or her spouse initially. The interviewee was asked to choose from “introduced by parents or relatives”, “introduced by friends”, “introduced by matchmaker” or “love match”. We clearly told the interviewee that “love match” refers to a model in which he/she got married with romantic love [19,22,24], including parents, relatives or friends introducing if they do not play a crucial role in their decision to become married. It means there is minimal intervention in the process of marriage and love, and the man and woman develop their relationship freely [26,36,37,50]. The value of love match equals 1 if the couple got married based on romantic love or 0 if they got married mainly due to non-emotional factors.

There are some other variables. Marriage distance measures the distance between the couple’s birthplaces. We chose the birthplaces which may reflect the cultural or community exposure of their natal families because we wanted to measure the effect of birthplace as an indicator of parental influence. Marriage age includes the ages of groom and bride when they got married. Education of groom and bride, as a proxy of human capital, are included in the models. Siblings of husband and wife are measured by the numbers of siblings to consider competition among siblings for resources in original family [45]. Considering the substitution effect of wedding house investment and wedding reception expenditure, we also consider them as control variables in model specification.

3.3. Model Specification

In this study, we focused on two types of dependent variables: brideprice/dowry and value of brideprice/dowry. Due to the different properties of dependent variables, we have different specifications.

3.3.1. Estimation of Brideprice/Dowry

We used logistic regression model to predict the likelihood of paying brideprice/dowry considering brideprice/dowry is a binary variable. The specification is as follows:

$$\mathrm{Brideprice}/{\mathrm{Dowry}}_i=\alpha +\beta LM+\gamma Control+\delta D+\epsilon$$

(1)

Here $\mathrm{Brideprice}/{\mathrm{Dowry}}_i$. represents the likelihood of paying brideprice/dowry, namely brideprice, dowry. $LM$ indicates love match. $Control$ represents all control variables, including marriage distance, marriage age, education, siblings, wedding house investment, and wedding reception expenditure. Square of marriage distance was added to examine the U-shaped (or inverted U-shaped) relationship between marriage distance and brideprice/dowry.

In order to control the trend of brideprice/dowry, marriage year was used in the models. Marriage year was divided into 11 cohorts: one for every five years, except those married between 1950 and 1965, who numbered few enough to fit into one category. Village dummies was added in models to identify the influence of economic development and regional differences on Brideprice/Dowry. α is the constant. β would be the impact of love match on marriage in the absence of omitted variable bias. ε is the error term that accounts for other factors. γ and δ are the vectors of impacts of control variable on Brideprice/Dowry. To analyze the heterogeneity effect, the sample was divided into three subgroups based on the quantile of marriage distance. Subgroups were estimated and their coefficients were compared. The definitions and descriptive statistics of these variables are shown in

Table 1. Descriptive statistics of variables.

Variable	Definition	Obs.	Mean	Std.	Min	Max
Brideprice	Groom’s family paid brideprice (1 = Yes, 0 = No)	5128	0.643	0.479	0	1
Value of brideprice	Value of brideprice (Yuan, in 1985 real value)	5128	1304.625	2329.694	0	14,054.814
Ln value of brideprice	Ln value of brideprice	5128	2.705	5.576	−4.605	9.551
Dowry	Bride’s family paid dowry (1 = Yes, 0 = No)	5128	0.560	0.496	0	1
Value of dowry	Values of dowry (Yuan, in 1985 real value)	5128	783.827	1701.472	0	18,739.752
Ln value of dowry	Ln value of dowry	5128	1.540	5.557	−4.605	9.838
Value of wedding house investment	Value of wedding house investment (Yuan, in 1985 real value)	5128	2569.054	7788.112	0	62,676.277
Value of wedding reception expenditure	Value of wedding reception expenditure (Yuan, in 1985 real value)	5128	1023.672	1336.640	0	8198.641
Love match	Couple is love match to marry (1 = Yes, 0 = No)	5128	0.231	0.422	0	1
Marriage distance	Distance between the spouses’ pre-marital residence (100 km)	5128	0.932	2.961	0	30.810
Square of marriage distance	Ln Marriage distance2	5128	10.403	12.607	0	132.547
Marriage age of groom	Groom’s age at marriage (Years)	5128	23.76	3.815	16	45
Marriage age of bride	Bride’s age at marriage (Years)	5128	22.08	3.404	15	40
Education of groom	Groom’s schooling year (Years)	5128	8.305	2.700	0	19
Education of bride	Bride’s schooling year (Years)	5128	7.573	2.920	0	22
Siblings of groom	Number of groom’s siblings	5128	2.610	1.558	0	8
Siblings of bride	Number of bride’s siblings	5128	2.784	1.563	0	8

Data Source: China Rural Development Survey.

.

3.3.2. Estimation of Value of Brideprice/Dowry

In this section, the dependent variable “value of brideprice/dowry” is different from Section 3.2.1. The value of brideprice/dowry is continuous variable and includes some observations with a positive probability of 0. Considering the distribution characteristics of the dependent variables, we first used Tobit regressions as a benchmark to examine the relationship between love match and the value of brideprice/dowry [51]. The specification for the Tobit regressions is:

$$\mathrm{Brideprice}/{\mathrm{Dowry}}_i=\alpha +\beta LM+\gamma Control+\delta D+\epsilon$$

(2)

where $Brideprice/Dowr{y}_i$. represents value of brideprice/dowry. $LM$, $Control$, $D$, α, β, γ, δ, and ε have the same definition as those in Equation (1).

However, it is reasonable to assume that brideprice payment behavior of groom’s family is not independent of dowry payment behavior of bride’s family. In other words, brideprice and dowry are jointly determined. The behavior of paying brideprice/dowry can therefore be indicated by Equations (3) and (4):

$${\mathrm{Brideprice}}_i=\alpha +\beta LM+\gamma Control+\delta D+\epsilon$$

(3)

$$Dowr{y}_i=\alpha +\beta LM+\gamma Control+\delta D+\epsilon$$

(4)

In this case, Equations (3) and (4) may seem unrelated but there are correlations between their errors term, which would lead to inefficient estimates if regressions were conducted separately. We adopted Zellner’s seemingly unrelated regression model (SUR) for Equations (3) and (4) jointly as a robustness check to obtain efficient estimates through generalized least squares (GLS) [51].

4. Descriptive Analysis

4.1. Trends of Brideprice and Dowry in Rural China

There was an overall upward trend of paying brideprice since the 1950s in rural China. It increased from 28 percent in the 1950s to 66 percent in 2016 with an average increase of 0.5 percentage points a year (Figure 1). However, there are differences across periods. Specifically, in the early years of the founding of the People’s Republic of China (1950–1964), the share of paying brideprice was quite low and less than 30 percent of males paid brideprice when becoming married. It rose sharply from the 1960s to the 1990s (1965–1990). In the mid-1990s (1995), the prevalence of paying brideprice reached its peak with an average annual growth rate of 1 percent. This is probably because the deepening of the “Reform and Opening Up” has broken the traditional circle of intermarriage in rural areas. In addition, the reform of the household registration system has been promoted and the government has not encouraged brideprice.

The share of paying dowry correspondingly increased since the 1950s. It rose from 28 percent in the 1950s to 57 percent in 2010s with annual growth of 0.48 percentage points. It also varied during this period. Specifically, in the early years of the founding of the People’s Republic of China (1950–1964), the share of paying dowry was quite low, and less than 30 percent of females paid dowry when becoming married. At the beginning of the “Reform and Opening up”, there was a significant increase in the number of women who left the countryside and engaged in off-farm employment. Due to the rapid increase of women’s disposable income, they were more likely to pay dowry when becoming married. From the 1970s to 1990s, the share of paying dowry raised dramatically from 40 percent to 60 percent. The share of dowry payments peaked in the 1980s.With the deepening of “Reform and Opening Up”, women’s consciousness has been liberated continuously, and the status of rural women in China has also been rising. The importance of dowry as a guarantee of family status after marriage has also been declining. Similar to brideprice, the prevalence of dowry has not increased since the beginning of the 21st century and even declined slightly.

The dowry is less popular compared with brideprice according to our data, which is consistent with the conclusion of previous studies [4,6,52]. Brides generally join a groom’s family due to the Chinese tradition of patrilocal residence, so the brideprice is relatively common, and dowry is not a necessary condition. The prevalence of brideprice payments reached its peak at just over 70 per cent in the mid-1990s, while the prevalence rate of dowries has stayed steadily around 60 per cent since 1990.

Our research uncovered some interesting findings about trends in the value of brideprice and dowry. We also introduced the per capita net income of rural residents in Figure 2 to become a more intuitive sense of the value of brideprice and dowry. The data on brideprice, dowry and per capita net income of rural residents are real figures, given a price-index adjustment.

From Figure 2, we can see that in the 20 years after the founding of the People’s Republic of China, the value of the brideprice has been at a relatively low level, only 60–80 yuan, which is only about 50% of the per capita income of rural residents. By the 1970s, the value of the brideprice increased slightly. Since the growth rate of rural residents’ per capita income was small during this period, the value of the brideprice has been basically the same as the annual per capita net income of rural residents. After entering the 1980s, the “Reform and Opening up” led to the rapid development of the century. During this period, the value of brideprice also increased dramatically. This linear upward trend still exists until after the 21st century. Since the 1980s, the absolute value and growth rate of the value of brideprice have exceeded the per capita net income of rural residents. By the 2010s, the value of brideprice had doubled the annual per capita net income of rural residents.

The change in the value of dowry is basically similar to that of brideprice but slightly different. Similarly, from the 1950s to the 1970s, the value of dowry was always low, only one-third to half of the per capita net income of rural residents. After entering the 1980s, the value of dowry has increased rapidly, and the growth rate has exceeded the growth rate of net income in the same period. After 1990, the absolute value exceeded the per capita net income of rural residents, but the growth rate slowed down in the mid-1990s. After entering the 21st century, the dowry value has resumed its rapid growth trend. Although the growth rate is not as fast as the per capita net income of rural residents, its absolute value has been higher than the per capita net income of rural residents by about 15%.

The value of dowry in rural China has always been lower than brideprice. However, the difference of them between the 1950s and 1980s is not that big. After entering the 1990s, due to the sharp increase in the value of brideprice, the difference between the value of brideprice and dowry is becoming even bigger. This is in line with findings in Brown’s 2009 survey of Shaanxi and other provinces [2,6].

Since few studies comprehensively describe marriage payment based on nationally representative data, there is a lack of evidence for horizontal comparison with our study. However, based on the data and conclusions of existing studies, we have found that our findings on marriage payment behaviors in sample areas are relatively consistent with other studies.

4.2. Love Match, Brideprice and Dowry

Table 2. Difference in brideprice and dowry payment behaviors, by love match.

Marriage Payment	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.549	0.671	0.122	0.000 ***
Dowry	0.470	0.586	0.117	0.000 ***
Value of brideprice	1663.992	1196.624	−467.367	0.000 ***
Value of dowry	1060.494	700.680	−359.813	0.000 ***

Note: Mean value are reported. *** p < 0.01. Data source: China Rural Development Survey.

shows the differences in brideprice and dowry payment behaviors between love match and arranged marriage. The share of paying brideprice/dowry for love match couples is significantly lower than that of arranged marriages, which is compliance with our assumptions. However, the value of brideprice and dowry present a phenomenon contrary to our theoretical hypothesis. After considering the cohort of marriage year, our hypothesis that love-match couples pay less was confirmed (Appendix A 

Table A1. Brideprice and dowry paying behaviors in different cohorts of marriage age, by love match.

Marriage Payments (Marriage Cohort = 1950–1964)	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.417	0.278	−0.139	0.311
Dowry	0.167	0.292	0.125	0.358
Value of brideprice	96.674	61.344	−35.330	0.559
Value of dowry	4.211	48.594	44.384	0.293
Marriage Payments (Marriage Cohort = 1965–1969)	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.333	0.386	0.053	0.658
Dowry	0.222	0.353	0.131	0.265
Value of brideprice	67.381	84.231	16.850	0.766
Value of dowry	52.706	61.599	8.893	0.813
Marriage Payments (Marriage Cohort = 1970–1974)	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.571	0.541	−0.030	0.828
Dowry	0.500	0.415	−0.085	0.533
Value of brideprice	172.937	151.517	−21.420	0.775
Value of dowry	68.163	79.515	11.352	0.816
Marriage Payments (Marriage Cohort = 1975–1979)	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.393	0.594	0.201	0.040 **
Dowry	0.214	0.488	0.274	0.005 ***
Value of brideprice	132.867	175.273	42.406	0.415
Value of dowry	30.567	116.103	85.536	0.025 **
Marriage Payments (Marriage Cohort = 1980–1984)	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.356	0.662	0.306	0.000 ***
Dowry	0.444	0.621	0.177	0.022 **
Value of brideprice	151.951	353.717	201.766	0.023 **
Value of dowry	308.530	266.237	−42.293	0.634
Marriage Payments (Marriage Cohort = 1985–1989)	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.468	0.700	0.233	0.000 ***
Dowry	0.377	0.585	0.209	0.001 ***
Value of brideprice	654.922	599.257	−55.665	0.682
Value of dowry	401.798	388.630	−13.168	0.886
Marriage Payments (Marriage Cohort = 1990–1994)	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.504	0.742	0.238	0.000 ***
Dowry	0.471	0.652	0.181	0.000 ***
Value of brideprice	546.324	895.395	349.071	0.006 ***
Value of dowry	519.240	725.888	206.648	0.082 *
Marriage Payments (Marriage Cohort = 1995–1999)	Love Match		Dif.= (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.580	0.750	0.170	0.000 ***
Dowry	0.497	0.642	0.145	0.001 ***
Value of brideprice	1056.118	1222.139	166.021	0.268
Value of dowry	830.743	718.359	−112.384	0.416
Marriage Payments (Marriage Cohort = 2000–2004)	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.563	0.721	0.157	0.000 ***
Dowry	0.494	0.649	0.156	0.001 ***
Value of brideprice	1515.629	2069.136	553.507	0.019 **
Value of dowry	1032.428	1077.706	45.278	0.786
Marriage Payments (Marriage Cohort = 2005–2009)	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.612	0.754	0.142	0.000 ***
Dowry	0.510	0.671	0.162	0.000 ***
Value of brideprice	2176.225	2618.061	441.836	0.068 *
Value of dowry	1214.685	1435.606	220.921	0.188
Marriage Payments (Marriage Cohort = 2010–2016)	Love Match		Dif. = (2) − (1)	t-Test (p-Value)
	Yes (1)	No (2)
Brideprice	0.568	0.753	0.185	0.000 ***
Dowry	0.486	0.667	0.180	0.000 ***
Value of brideprice	2953.139	3867.454	914.315	0.003 ***
Value of dowry	1826.071	2054.675	228.604	0.363

Note: Mean value are reported. *** p < 0.01, ** p < 0.05, * p < 0.1. Data source: China Rural Development Survey collected by the Center for Chinese Agricultural Policy, CAS.

).

The difference in marriage payment behavior between love-match and non-love-match couples changed during 1950s–2015. It is not significant in 1950s–1970s (1st–3rd cohort), but the brideprice/dowry value in arranged marriage is significantly higher than that in love match after entering the 1980s (4th cohort).

5. Empirical Results and Discussion

5.1. Love Match and the Likelihoods of Paying Brideprice/Dowry

The empirical results of Logit model (

Table 3. The effect of love match on the likelihood of paying brideprice and dowry in rural China during1950 to 2016.

Variables	Brideprice		Dowry
	Logit	SUR	Logit	SUR
	(1)	(2)	(3)	(4)
Love match	−0.103 ***	−0.107 ***	−0.099 ***	−0.103 ***
	(−6.103)	(−6.682)	(−5.571)	(−6.218)
Ln Marriage distance	−0.039 ***	−0.041 ***	−0.049 ***	−0.049 ***
	(−6.712)	(−7.773)	(−8.125)	(−8.844)
Square of ln marriage distance	−0.004 ***	−0.004 ***	−0.006 ***	−0.006 ***
	(−4.331)	(−5.142)	(−5.755)	(−6.387)
Marriage age of groom	−0.013 ***	−0.014 ***	−0.014 ***	−0.014 ***
	(−5.785)	(−6.192)	(−5.655)	(−5.840)
Marriage age of bride	−0.006 **	−0.006 **	−0.001	−0.001
	(−2.196)	(−2.284)	(−0.477)	(−0.507)
Education of groom	0.002	0.002	0.003	0.003
	(0.614)	(0.792)	(1.072)	(1.173)
Education of bride	−0.001	−0.001	0.003	0.003
	(−0.340)	(−0.223)	(1.213)	(1.281)
Siblings of groom	−0.001	−0.003	0.000	−0.001
	(−0.205)	(−0.510)	(0.059)	(−0.207)
Siblings of bride	0.007	0.005	−0.001	−0.002
	(1.380)	(1.179)	(−0.259)	(−0.496)
Wedding house investment of groom	0.000	0.000	0.000 *	0.000 *
	(1.240)	(1.544)	(1.711)	(1.919)
Wedding house investment of bride	−0.000	−0.000	−0.000	−0.000
	(−1.327)	(−1.534)	(−0.556)	(−0.683)
Wedding reception expenditure of groom	0.000 ***	0.000 ***	0.000 ***	0.000 ***
	(7.328)	(9.086)	(6.446)	(7.765)
Wedding reception expenditure of bride	0.000	0.000 **	0.000 ***	0.000 ***
	(1.485)	(1.972)	(2.957)	(4.666)
Cohort of marriage year	Yes	Yes	Yes	Yes
Village dummies	Yes	Yes	Yes	Yes
N	5128	5128	5128	5128
Log Lik.	−2697.521		−2892.468

Note: a. t-statistics in parentheses in Logit model and z-statistics in parentheses in SUR model, *** p < 0.01, ** p < 0.05, * p < 0.1; b. Marginal effects are reported. Data source: China Rural Development Survey.

, column 1 and 3) are basically consistent with those of SUR model (Table 3, column 2 and 4), indicating that our results are robust. For brevity, we only present the results of SUR models.

Love match is significantly correlated with the likelihood of paying brideprice, which supports Hypothesis 1. We find that the groom’s family is 10.7 percent less likely to pay brideprice if the couple is love-match (p < 0.01) (Table 3, row 1, column 2), which is consistent with the findings of descriptive analysis. There is a significant inverted U-shaped relationship between marriage distance and the likelihood of paying brideprice (p < 0.01) (Table 3, rows 2 and 3, column 2). Marriage age of groom and bride are negatively correlated with the likelihood of paying brideprice. Every 1-year increase of the marriage age of groom and bride reduces the likelihood of paying brideprice by 1.4% and 0.6%, respectively. Wedding reception expenditures of the couple are both positively related to the likelihood of paying brideprice, but the magnitudes of coefficient are negligible (p < 0.01) (Table 3, row 12 and 13, column 2).

The relationship of love match and the likelihood of paying dowry is similar to that of brideprice. The bride’s family is 10.3 percent less likely to pay dowry if the couple is love matched (p < 0.01) (Table 3, row 1, column 4). The significant inverted U-shaped relationship between marriage distance and likelihood of paying dowry also exists. The probability of paying dowry decreases by 1.4 percent for each additional year of groom’s marriage age (p < 0.01) (Table 3, row 4, column 4). Groom’s wedding house investment is also slightly related to the likelihood of paying dowry (p < 0.1) (Table 3, row 10, column 4). Wedding reception expenditure by spouses is significantly and positively related to the likelihood of paying dowry (p < 0.01) (Table 3, row 12 and 13, column 4).

5.2. Love Match and the Value of Brideprice/Dowry

Both Tobit regression and SUR results are presented in

Table 4. The effect of love match on the value of brideprice and dowry in rural China during 1950 to 2016.

Variables	Ln Value of Brideprice		Ln Value of Dowry
	Tobit	SUR	Tobit	SUR
	(1)	(2)	(3)	(4)
Love match	−1.899 ***	−1.275 ***	−2.005 ***	−1.151 **
	(−5.756)	(−7.095)	(−5.395)	(−6.342)
Ln Marriage distance	−0.752 ***	−0.481 ***	−1.033 ***	−0.565 ***
	(−6.226)	(−8.047)	(−7.723)	(−9.349)
Square of ln marriage distance	−0.088 ***	−0.054 ***	−0.130 ***	−0.068 ***
	(−4.386)	(−5.252)	(−5.688)	(−6.622)
Marriage age of groom	−0.275 ***	−0.167 ***	−0.320 ***	−0.161 ***
	(−6.014)	(−6.425)	(−6.131)	(−6.127)
Marriage age of bride	−0.123 **	−0.085 ***	−0.022	−0.030
	(−2.351)	(−2.944)	(−0.387)	(−1.041)
Education of groom	0.052	0.030	0.078	0.043
	(1.106)	(1.019)	(1.431)	(1.471)
Education of bride	0.011	0.001	0.089	0.053 *
	(0.245)	(0.180)	(1.627)	(1.795)
Siblings of groom	−0.076	−0.050	−0.047	−0.007
	(−0.838)	(−0.871)	(−0.464)	(−0.125)
Siblings of bride	0.078	0.048	−0.080	−0.025
	(0.852)	(0.865)	(−0.747)	(−0.435)
Wedding house investment of groom	0.000 **	0.000 **	0.000 ***	0.000 ***
	(2.058)	(2.464)	(2.636)	(3.133)
Wedding house investment of bride	−0.000	−0.000 *	−0.000	−0.000
	(−0.953)	(−1.639)	(−0.218)	(−0.494)
Wedding reception expenditure of groom	0.001 ***	0.000 ***	0.001 ***	0.000 ***
	(8.631)	(11.181)	(7.269)	(9.502)
Wedding reception expenditure of bride	0.000	0.000 **	0.001 ***	0.000 ***
	(1.627)	(2.532)	(3.782)	(6.241)
Cohort of marriage year	Yes	Yes	Yes	Yes
Village dummies	Yes	Yes	Yes	Yes
N	5128	5128	5128	5128

Note: a. z-statistics in parentheses in Tobit model and SUR model, *** p < 0.01, ** p < 0.05, * p < 0.1; b. Marginal effects are reported. Data source: China Rural Development Survey.

. The p-values of the Breusch-Pagan test in the SUR models are less than 0.01, which indicates the SUR model should be used. The models perform well and produce results that are mostly consistent. The effects of most control variables are also as expected. For brevity, we report the econometric findings based on the results of SUR models.

According to our results, the value of brideprice provided by the couples who are parental matchmaking is 1.27 times that of love-match couples (p < 0.01) (Table 4, row 1, column 2), which fully supports Hypothesis 2. The inverted U-shaped relationship between the marriage distance and brideprice value still exists (p < 0.01) (Table 4, row 2 and 3, column 2). The value of brideprice decreases by 16.7 and 8.5 percent for each additional year of groom’s and bride’s marriage age, respectively (p < 0.01) (Table 4, row 4 and 5, column 2). Both wedding house expenditure and wedding reception expenditure are significantly related to the value of brideprice (p < 0.01) (Table 4, row 10–13, column 2).

Similar results are found for the value of dowry. Couples who are parental matchmaking become a value of dowry which is 1.15 times higher than those who are love-match (p < 0.01) (Table 4, row 1, column 4). An inverted U-shaped relationship is also found between love match and dowry value (p < 0.01) (Table 4, row 2 and 3, column 4). The dowry value decreases by 16.1 percent for every 1-year increase of the groom’s marriage age (p < 0.01) (Table 4, row 4, column 4). Unlike the value of brideprice, marriage age of bride is not significantly associated with the value of dowry. The value of dowry increased by 5.3 percent for each additional year of the bride’s schooling (p < 0.1) (Table 4, row 7, column 4). The groom’s family’s wedding house investment and the bride’s and groom’s wedding reception expenditure are both significantly associated with the value of dowry (p < 0.01) (Table 4, row10, 12 and 13, column 4).

These results further support the view of case studies made by Wang (2019) that “love match mode dominated by children” has a high affinity with “low brideprice”, while “parental matchmaking mode dominated by parents” is closely related to “high brideprice” [19].

5.3. Robustness Check

To verify the robustness of our results, we randomly selected a subsample of 2500 observations to run the previous regressions for robustness tests. Due to space limitation, we only report the regression results of key variables. In

Table 5. Regression results of robustness test.

	Brideprice		Dowry		Ln Value of Brideprice		Ln Value of Dowry
Variables	Logit	SUR	Logit	SUR	Tobit	SUR	Tobit	SUR
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Love match	−0.140 ***	−0.147 ***	−0.136 ***	−0.141 ***	−2.595 ***	−1.761 ***	−2.694 ***	−1.527 ***
	(−6.001)	(−6.319)	(−5.439)	(−5.902)	(−5.582)	(−6.714)	(−4.986)	(−5.834)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Cohort of marriage year	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Village dummies	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	2500	2500	2500	2500	2500	2500	2500	2500

Note: a. t-statistics in parentheses in Logit model and z-statistics in parentheses in Tobit and SUR model, *** p < 0.01; b. Marginal effects are reported. Data source: China Rural Development Survey.

, we can see that love match maintain a robust and significant negative impact on brideprice and dowry, indicating that the results are robust.

In order to prevent the conclusions of our study from being unstable at the cohort and regional level, we chose two provinces (Jiangsu and Sichuan) and two cohorts (before 80s and after 80s) to analyze, respectively. Jiangsu province, as a coastal province in the east, has a higher economic development level and a more liberal social atmosphere and is less likely to be influenced by China’s traditional gender norms and patriarchal traditions. However, Sichuan province, located in inland areas, is on the contrary. It turns out that our results are still robust in these two provinces and cohorts (Limited by space is not shown in the article, if interested can contact the author.). It indicates that there is negative correlation between love match and marriage payment even considering the prevalence and changes of modernization and traditional culture in rural China.

5.4. Role of Marriage Distance

To measure the role of marriage distance in the relationship between love match and marriage payment behavior, we divided the sample into three groups based on the quantile of marriage distance: short-, mid- and long-distance. The results are presented in

Table 6. Heterogeneous effect of love match on the likelihood of paying brideprice and dowry.

	Brideprice								Dowry
	Logit				SUR				Logit				SUR
	Near	Mid	Far		Near	Mid	Far		Near	Mid	Far		Near	Mid	Far
Love match	−0.088 ***	−0.090 ***	−0.139 ***		−0.096 ***	−0.092 ***	−0.140 ***		−0.063 **	−0.105 ***	−0.149 ***		−0.071 **	−0.106 ***	−0.151 ***
	(−3.206)	(−2.841)	(−5.282)		(−3.518)	(−3.092)	(−5.432)		(−2.141)	(−3.161)	(−5.604)		(−2.464)	(−3.330)	(−5.771)
Control variables	Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes
Cohort of marriage year	Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes
Village dummies	Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes
Observations	1709	1710	1709		1709	1710	1709		1709	1710	1709		1709	1710	1709
Log Lik.	−802.823	−822.147	−942.146						−889.423	−923.816	−965.517

Note: a. Robust t-statistics in parentheses, *** p < 0.01, ** p < 0.05; b. Marginal effects are reported; c. Cluster at household level. Data source: China Rural Development Survey.

and

Table 7. Heterogeneous effect of love match on the value of brideprice and dowry.

	Brideprice								Dowry
	Tobit				SUR				Tobit				SUR
	Near	Mid	Far		Near	Mid	Far		Near	Mid	Far		Near	Mid	Far
Love match	−1.670 ***	−1.312 **	−2.837 ***		−1.216 ***	−0.964 ***	−1.666 ***		−1.323 ***	−1.893 ***	−3.330 ***		−0.812 ***	−1.111 ***	−1.706 ***
	(−3.185)	(−2.252)	(−5.169)		(−4.143)	(−2.917)	(−5.509)		(−2.194)	(−2.915)	(−5.477)		(−2.651)	(−3.239)	(−5.782)
Control variables	Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes
Cohort of marriage year	Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes
Village dummies	Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes		Yes	Yes	Yes
Observations	1709	1710	1709		1709	1710	1709		1709	1710	1709		1709	1710	1709

Note: a. Robust z-statistics in parentheses, *** p < 0.01, ** p < 0.05; b. Marginal effects are reported; c. Cluster at household level. Data source: China Rural Development Survey.

.

The negative effect between love match and paying brideprice is the largest among the couples with long marriage distance (p < 0.01) (Table 6, row 1, columns 6 and 12), which provides support for Hypothesis 3. Although we find a significant negative correlation between love match and the likelihood of paying a brideprice in all three groups (p < 0.01) (Table 6, row 1, columns 4–6 and 10–12), the coefficient of love match is 0.140 for long-distance group, which is much larger than those in short and middle marriage distance groups (p < 0.01) (Table 6, row 1, column 4 and 5).

The negative effect of love match on the likelihood of paying dowry is also the largest for the couples with long marriage distance. Specifically, the likelihood of paying dowry decrease 7.1 percent and 10.6 percent for couples who are love-match and with short and middle marriage distance, respectively (p < 0.05) (Table 6, row 1, column 10 and 11). However, the couples who are love-match and with long marriage distance is 15.1 percent less likely to pay dowry than those with long marriage distance and arranged marriages (p < 0.01) (Table 6, row 1, column12).

The results show that love match has the largest negative effect on the value of brideprice for couples with long marriage distance. For the couples with long marriage distance and arranged marriage spent 1.66 times more than those with love match and long marriage distance (p < 0.01) (Table 7, row 1, column 6). In short marriage distance group, the couples with arranged marriages spent 1.21 times more than those with love match (p < 0.01) (Table 7, row 1, column 4). However, in middle marriage distance group, the couples who are arranged marriage only spent 96.4 percent more than those with love match (p < 0.01) (Table 7, row 1, column 5).

The effect of love match on the value of dowry is the largest for couples with long marriage distance. The coefficients of love match on the value of dowry for the couples with short, middle, and long marriage distance are −0.812, −1.111, and −1.706 (p < 0.01) (Table 7, row 1, column 10–11), respectively. It indicates that the love-match couples with short, middle, and long marriage distance spent 81.2, 111.1 and 170.6 percent less than those who are arranged marriage and with the same marriage distance, respectively.

5.5. Discussions

Why is love match negatively correlated with paying brideprice/dowry? Based on the conceptual framework we mentioned before, children have different priorities from their parents when it comes to marriage, due to preference differences of both sides. Firstly, parental matchmaking is more concerned with economic factors, so the brideprice or dowry of couples with parental matchmaking are relatively high. Couples who are love-match care more about emotional benefits, and the support brought by the emotional connection between the couples’ substitute for the economic function of marriage payment to some extent. Secondly, from the perspective of risk prevention function of marriage payment, a major point of brideprice is to secure the material future of the bride’s family. The couples of being love are more likely to build a stronger foundation of trust, so they do not have to pay an extra premium or provide a guarantee to address the uncertainty of their future lives. Thus, love match is associated more relatively with less likelihoods and smaller amount of brideprice/dowry, whereas parental matchmaking is just the opposite.

Marrying an individual living in a long distance will make it more difficult to support their natal family [4,41,53]. An increased marriage distance therefore increases the pressure on the brideprice and dowry, but only to a certain extent of marriage distance. It is more common for women from poorer and more remote areas to migrate long distances to wealthier ones for marriage. This “marrying up” improves the economic status and well-being of these women, which helps to explain why those with a long marriage distance have a weaker bargaining power [42,47,54,55]. Some people in rural China who are suffering from severe marriage squeeze therefore choose to marry women from far away in order to reduce their marriage payment burden [19,45,56,57]. This explains why empirical studies have drawn opposite conclusions. There is a significant inverted U-shaped relationship between marriage distance and marriage payment behaviors. When the marriage distance is divided into three groups (short-, mid-, and long-distance), the effect of love match is most obvious in the long-distance group. The greatest reduction in the likelihoods and amount of paying bidirectional cash transfers could be found in this group, such as brideprice and dowry.

6. Conclusions and Implications

This study examines the correlation of love match (versus parental matchmaking) and marriage payment behaviors by adopting Logit, Tobit, and SUR models based on data from 1950s to 2010s in rural China. We also analyze the role of marriage distance in this relationship. Some robust and interesting conclusions are yielded.

Love match is associated with less likelihood of paying brideprice and dowry and smaller amount of brideprice and dowry. There is a significant inverted U-shaped relationship between marriage distance and the likelihood of paying brideprice and dowry and the value of brideprice and dowry. For short-distance marriage, an increase in the marriage distance also increases the pressure on bidirectional cash transfers. Love match, however, prevent the increased likelihood of paying brideprice and dowry and decreases the value of brideprice and dowry when a long marriage distance is involved. The results indicate that the cohort of marriage year and region are two important factors in marriage payments in rural China.

Our study has some implications in the village governance of recent rural China. “Poverty caused by marriage” has been common in rural China, and the Chinese government is determined to address the problem of the sky-high brideprice [58,59]. However, it is difficult to achieve this goal by administrative decree due to complex cultural customs. Our study indicates that love match could be a helpful way and the government is suggested to strengthen the efforts on publicity and advocate love-match marriage. However, we cannot force two people to fall in love to solve the problem of high level of marriage payments, for love is one of the basic rights of human beings. However previous studies have shown that the vulnerable group is more likely to rely on parental matchmaking, such as low-educated rural people, and they should be paid more attention to reduce the phenomenon of “poverty caused by marriage”.

Although, to the best of our knowledge, this is the first study in China to systematically explore the relationship of love match and marriage payment behaviors, we are aware of the shortcomings of this study. First, due to data limitation, it can only analyze the correlation between love match and marriage payment but not their causal relationship. Advanced methods should be applied to yield robust results in the future when the appropriate data is available. Second, studying the effect of dowry on the well-being of a wife or the whole household is also very important, which will be the primary topic in our next study.