Skip to content
Life Guide Situations Guides

    Type a keyword to search
    English

    Is marriage worth it: breaking one muddled account into five clear ones

    Many people want to ask one vague question: does marriage cost you or pay off? That question has several different things packed into it: whether your partner can give emotional support, whether they will share the housework, whether they can earn money, and whether you are marrying just to satisfy your elders. These cannot be converted into one another, and lumped together they cannot be answered, so this article takes them apart and counts each one separately. Each account uses only numbers from official statistics or meta-analyses (studies pooled together into one calculation), and where something cannot be calculated, it says so. At the end there is a checklist you can fill in yourself. The article does not give a verdict on whether you should marry.

    1. First, take the question apart

    "Is marriage worth it" holds at least five separate accounts. They have different units and different ways of counting, so they cannot be added together.

    1. Money: first, the betrothal gift (彩礼, bride price), the wedding and the home you actually have to pay for; second, who the law says owns which property. There are no official statistics for the first; the second has written provisions.
    2. Time: housework, caring for family, giving way for your partner's career. This is work nobody pays for, and the unit is minutes per day.
    3. Health: how much the death rate differs between married and unmarried people. This account can only show that two things occur together; it cannot show that marriage produced the gap.
    4. Legal risk: how property is divided, how long it takes to get out, which contributions you can later claim compensation for, and who can make decisions for you when you are seriously ill. For this account, you read what the legal text says.
    5. Relationship quality: whether your partner gives you emotional support, and how much the two of you fight. This can only be measured by the strength of the associations found in research. It cannot be turned into money or into time.

    The fifth account is the hardest to turn into a number, but it is the closest to what most people are really asking.

    2. The data and sources for each account

    Registration data: the marriage rate and divorce rate are two crude rates

    The figures come from the Ministry of Civil Affairs' 2024 Statistical Bulletin on the Development of Civil Affairs (《2024 年民政事业发展统计公报》). In 2024, 6.106 million (610.6万) couples registered a marriage under the law nationwide, 20.5% fewer than the year before, a marriage rate of 4.3 per thousand. 3.513 million (351.3万) couples completed divorce procedures under the law, a divorce rate of 2.5 per thousand. Of those 3.513 million (351.3万), 2.622 million (262.2万) registered their divorce with the civil affairs authorities, and 891,000 (89.1万) were divorced by a court judgment or mediation. The bulletin's notes explain how the two rates are calculated: take the number of couples marrying (or divorcing) that year, divide by the year's average total population, and multiply by 1000.

    The denominator of both rates is the whole population, which is why they are called crude rates. "Crude" means no breakdown of any kind has been applied. They answer how many couples out of every thousand people registered a marriage or divorce this year; they cannot answer whether a given marriage will eventually break up. To answer that, you need data that follows the same group of people over a long time. This article could not find such data in an official source.

    The health account: married people have a lower death rate, but this is correlation, not causation

    Three meta-analyses have pooled many studies and give results of a consistent size. "All-cause mortality" below means all deaths added together, regardless of cause.

    • Manzoli et al. (2007) pooled 53 comparisons covering more than 250,000 older people. Here "not in a marriage" covers three kinds of people: widowed, divorced or separated, and never married. Compared with these three groups, married people had an all-cause mortality RR of 0.88, about 12% lower. RR is the ratio of the death risk in two groups; below 1 means the married group is lower. The 95% CI of 0.85-0.91 is the plausible range for this number. Looking at the three groups separately, the widowed were at 1.11, the divorced or separated at 1.16, and the never married at 1.11, which is about 10%, 16% and 10% higher. The results held when men and women were analyzed separately, when better and worse studies were analyzed separately, and when Europe and the Americas were analyzed separately. Recalculated with the most conservative method, the RR was 0.94, only about 6% lower.
    • Roelfs et al. (2011) pooled 95 papers covering more than 500 million (5亿) people. Compared with married people, people who had never married had an all-cause mortality HR of 1.24, about 24% higher. Like RR, HR is a ratio; above 1 means higher risk. 1.19-1.30 is the plausible range for this number. The older the age, the smaller the gap. The more rigorous the study, the smaller the gap it found.
    • Wang et al. (2020) pooled 21 studies covering more than 7.89 million (789万) people. Not being in a marriage was linked with higher all-cause, cancer and cardiovascular mortality, in both men and women. The links with cardiovascular and all-cause mortality were stronger in men. Comparing never-married men with never-married women, the risk of dying from stroke was 31% higher and all-cause mortality was 9% higher.

    All three only followed and recorded groups of people; none divided people into comparison groups. People who are healthy, earn more and have more social contacts are more likely to get married in the first place, and also more likely to live longer in the first place. This is called selection effect. The Manzoli paper also found signs of publication bias. Publication bias means that studies with attractive results are more likely to get published, so the pooled result leans optimistic. So these numbers can only be read as "death rates are somewhat lower among married people"; they cannot be read as "getting married means 12% fewer deaths for you." The samples in these studies are mostly European and American, so applied to China they are only a rough guide to the size of the effect.

    The time account: the gender gap in unpaid work is an upper-limit reference

    The figures come from the National Bureau of Statistics' Third National Time Use Survey, conducted in May 2024. First, three terms. Unpaid work is the part of work that nobody pays for. "Participants" are the people who actually did the activity that day. "Participation rate" is the share of people who did it that day. The results: participants in unpaid work spent an average of 2 hours 45 minutes a day, of which men spent 1 hour 52 minutes and women 3 hours 29 minutes. The participation rate was 75.6%, with men at 67.5% and women at 83.9%. Looking at housework alone, spread across all residents it comes to 1 hour 17 minutes a day. Counting only people who actually did it that day, it is 1 hour 59 minutes, 28 minutes less than in 2018, with a participation rate of 64.9%. Looking at caring for family members alone, spread across all residents it is 30 minutes. Counting only people who actually did it that day, it is 1 hour 46 minutes, with a participation rate of 28.4%.

    The 2018 Second Survey found this: housework spread across all residents was 1 hour 26 minutes, of which men did 45 minutes and women 2 hours 6 minutes. The participation rate was 40.4% for men and 75.6% for women. Caring for children, spread across all residents, was 36 minutes, with men at 17 minutes and women at 53 minutes. The participation rate was 12.3% for men and 25.1% for women.

    Both surveys broke results down only by sex, urban or rural, and age, not by whether people were married. So there is no official number for "how much more housework you do after marriage." What can be used is the gap between men and women. In 2024 the gender gap in unpaid work was 1 hour 37 minutes. If the division of housework is not worked out, that is about the most the gap between two people could grow.

    The money account: default rules, written agreements and compensation for housework

    There are no official statistics for betrothal gifts, weddings or the home, so this article gives no figures for them. What can be stated clearly are the rules set by the marriage and family part of China's Civil Code (民法典), which has been in force since January 1, 2021.

    • Article 1062: from the day of marriage to the day of divorce, four kinds of property acquired in that time count as the couple's joint property. First, wages, bonuses and payment for services. Second, income from business and investment. Third, income from intellectual property. Fourth, property that is inherited or received as a gift. The fourth has an exception: if a will or gift contract says it goes to one person only, it is not joint property. Both partners have equal rights to handle joint property.
    • Article 1063: four kinds of property are personal and do not go into the joint property at divorce. First, property owned before the marriage. Second, compensation or damages received for bodily injury. Third, property that a will or gift contract says belongs to one spouse only. Fourth, daily-life items used only by one spouse.
    • Article 1065: the two of you may agree on who owns property from before and during the marriage. You can agree that each keeps their own, that everything is shared, or that some is separate and some is shared. The agreement must be in writing, and once written it legally binds both of you. If there is no agreement, or the agreement is unclear, the two articles above apply.
    • Article 1088: if one spouse put in more than the other by raising children, caring for older relatives or helping the other spouse's work, that spouse can ask the other for compensation at divorce. How much is first agreed between the two; if they cannot agree, the court decides.

    These rules work for both sides. The higher earner should focus on Articles 1062 and 1065. The partner who does more at home should focus on Article 1088. A property agreement only has to be in writing; the law does not require notarization.

    The cost of getting out: the cooling-off period and conditions for a lawsuit

    • Article 1076: if both partners want a divorce, they first sign a written divorce agreement, then apply in person at the marriage registration office. The agreement must state who will raise the children and how property and debts are handled. All of this must be what the two have agreed.
    • Article 1077: for 30 days counted from the day the registration office receives the application, either partner who changes their mind can withdraw it. These 30 days are the well-known cooling-off period. In the 30 days after the cooling-off period ends, the two must apply in person to receive the divorce certificate; if they do not, the application is treated as withdrawn.
    • Article 1079: if one wants a divorce and the other does not agree, they may first go through mediation, or sue directly in court. Once in court, the court also mediates first. If the court finds the relationship has truly broken down and mediation will not help, it shall grant the divorce. In these situations, if mediation fails, the court shall grant the divorce: bigamy or living with someone else, domestic violence or abuse or abandonment, gambling, drug use or other bad habits that persist despite correction, and separation for two full years because of the breakdown of the relationship. There is one more situation: if the first lawsuit was refused and the two then live apart for another full year, a new lawsuit shall be granted.

    Of the 3.513 million (351.3万) divorces in 2024, 891,000 (89.1万) went through a court judgment or mediation. The cooling-off period applies only to registered divorces; it does not apply to court cases.

    Legal status: a spouse is a close relative and can make decisions for you when you are seriously ill

    Once you are married, your partner is legally your close relative. Of the close relatives listed in Article 1045 of the Civil Code, the spouse comes first. This status matters in the following situations.

    • Surgery: Article 1219 provides that before surgery, special examinations or special treatment, the risks must be explained to the patient and the patient's explicit consent obtained. If the patient is unconscious and cannot give it, or it is not appropriate to tell the patient, the doctor explains to a close relative, who gives consent. While the patient is conscious, the patient signs; a spouse cannot sign in their place.
    • Emergency rescue: Article 1220 provides that if a patient's life is in danger and neither the patient's nor a close relative's opinion can be obtained, treatment may begin immediately with approval of the hospital's head. If family is not there, the law allows the hospital to save the patient first.
    • Long-term inability to decide for yourself: when an adult loses the capacity for civil conduct, Article 28 puts the spouse first in the order of guardians, and parents and children second. If a guardian was designated in writing beforehand, Article 33 makes the designated person the guardian.
    • Inheritance: the first-order heirs in Article 1127 are the spouse, children and parents. If there is no will, the spouse shares the estate with your children and parents.

    This benefit comes with two discounts. First, Article 1219 says only "close relatives" and does not rank them. Parents and adult children are also close relatives, and the law does not say whose view wins when several disagree. Second, an unmarried partner can make up most of it too. The way is to complete, while both of you are clear-headed, a written power of attorney, a consensual guardianship agreement (意定监护) and a will; see §10.18 (power of attorney, consensual guardianship and wills). That tip is written for same-sex couples, but these three documents work the same for an unmarried opposite-sex couple. Whether a consensually appointed guardian can sign a surgery consent form directly, though, varies from hospital to hospital. With marriage, you get all of these statuses at once when you register, without arranging them one by one, and with less risk of a hospital or relatives picking at your documents.

    Relationship quality: an account more worth counting than "whether you are married"

    Robles et al. (2014) pooled 126 studies covering more than 72,000 people. The conclusion: the higher the marital quality, the better the health. To measure how closely two things move together, they used a number called the effect size r. The closer r is to 0, the less the two things are related. The larger the absolute value of r, the more closely they move together. A minus sign in front of r means that when one is high, the other is low. The r values for these associations ranged from 0.07 to 0.21. For death risk, r = 0.11. For cardiovascular reactions during conflict between the two partners, r = -0.13. These are all small values. Still, the link between health behaviors such as eating well and health is of the same size. Some results showed signs of publication bias (attractive results are more likely to be published). Some studies specifically tested for differences between men and women and mostly found none. Most of these studies only followed and recorded people without comparison groups, so they cannot establish cause. A cardiovascular review by Dhindsa et al. (2020) also points out that marital satisfaction and marital quality clearly affect cardiovascular risk.

    This article did not find a meta-analysis that directly compares the death rate of "poor-quality marriage" with "never married." So "a bad marriage is worse than no marriage" is for now only an inference, with no numbers behind it. What does hold up is this: the difference between "married or not" has selection effect mixed into it, and "whether the marriage is good" is a separate matter. Looking only at whether people are married, and not at how things are going, leaves out the marital quality account entirely.

    3. How to count it for yourself: a fill-in checklist

    Fill in each account on its own. Do not add numbers from different accounts together. For any box you cannot fill in, write "don't know." "Don't know" is itself useful information.

    What your partner can provide (fill in one by one)

    • Money: how much your partner earns, how much debt they have, what property they had before the marriage. Whether property after marriage follows the law's default rules or needs a written agreement. If you write an agreement, what it will say.
    • Time: how much unpaid housework and care your partner does each day now. The national average can serve as a comparison: among people who actually did it that day, 1 hour 52 minutes for men and 3 hours 29 minutes for women. How these tasks will be divided after marriage. How they will be divided when there are children, or when older relatives at home need care. Whose job can be put second of the two.
    • Relationship quality: how many conflicts the two of you had in the past six months and how each one ended. What your partner did when you were ill, out of work or low in mood. And the reverse: what you did when your partner met these things.
    • Emotional value: when you are with your partner, do you feel you have more energy or that you are being drained? This takes a few months of being together to see, and a checklist of qualifications cannot show it.

    What you have to put in (fill in one by one)

    • Money: how much you will spend on the marriage itself, including the betrothal gift, the wedding and the home. Where that money would go if you did not marry.
    • Time: estimate how much more housework and care you will do each day. Then estimate how much more time each week you will spend on social obligations.
    • Health: whether your routine, eating and exercise will get better or worse because of marriage. This is closer to your own situation than any of the numbers calculated by meta-analyses.
    • Cost of getting out: a registered divorce means waiting out the 30-day cooling-off period, then collecting the certificate within the following 30 days. A court case means meeting one of the situations in Article 1079. Whether you could bear that period.

    How to protect yourself legally (applies to both sides)

    • Whether you have documents for your pre-marriage property, such as a home purchase contract, bank statements, a gift contract or a will. Whether those documents can prove that it was "owned before the marriage" or "stated to belong to one side only."
    • Whether to sign a written property agreement. Whether the agreement covers only pre-marriage property, only what is earned after marriage, or both.
    • After marriage, who does more housework and care. Whether the one who does more knows they can ask for compensation under Article 1088. Whether the two of you are willing to agree on how compensation works in advance.
    • Debts: which debts borrowed in one person's name after marriage would count as the couple's joint debts. Whether the other partner knew about the loan.
    • If you are seriously ill or unconscious, who you want to sign and decide for you. Once married, your spouse is a close relative, but your parents are too, and opinions may differ. If you want to name one person, you can arrange consensual guardianship in writing in advance. If you are not marrying, you need to arrange all of these yourself.

    Fill in this section separately if you are marrying for your elders

    • What the elders get: peace of mind, saving face, no more nagging to marry, someone to care for them in old age later.
    • Which of these are really benefits to the elders, and which are actually benefits to you.
    • The costs in the three sections above are all borne by you. The benefits the elders get do not automatically turn into your health, and do not automatically turn into the quality of your relationship.
    • With it written out separately like this, whether this account is heavy or light is for the person filling it in to judge.

    4. Common mistakes

    Mixing up the two divorce rates. One is the crude divorce rate, which counts how many couples per thousand people divorce; in 2024 it was 2.5 per thousand. The other is the "divorce-to-marriage ratio," which divides the number of divorces that year by the number of marriages that year: 351.3 / 610.6 ≈ 57.5%. The numerator and denominator of the divorce-to-marriage ratio are not the same group of people; most of those divorcing this year married many years ago. When fewer people marry, this ratio jumps up by itself, so it cannot show that "half of all marriages end in divorce." The Ministry of Civil Affairs bulletin does not contain this indicator either.

    Treating "occurs together" as "because of." Married people's death rate is about 12% lower; that is the result of pooling many studies. But being healthy, earning more and having more social contacts all affect two things at once: whether a person can get married, and whether a person lives long. This account can be counted, but it should be read with a discount, and it cannot take the place of judging the specific relationship in front of you.

    Treating marriage as a task to complete. Treated as a task, you look only at the state of "married or not" and not at "how it is going." But the effect of relationship quality on health is independent of whether you are married, and a task-style decision usually leaves this account out entirely.

    Treating the national average as your own number. The gender gap in the time use survey is a national average. Property rules are the default settings the law provides, and a written agreement can change them. National figures can give you a rough size and a starting point for negotiation, not a conclusion.

    Where there are no official figures, do not fill in with unofficial ones. There are no official statistics on betrothal gifts, weddings or the cost of raising a child, so this article leaves them blank. Filling in your own real numbers is more accurate than quoting any institution's estimate.

    Sources

    Translated from the Chinese original. Chinese original