H✳HowToLiveBetter
All long reads
LONG READ / Long reads

Is Marriage Worth It: Breaking One Muddled Ledger into Five Clear Accounts

Separate health, time, money, exit costs and relationship quality, then work through a personal checklist.

Related chapter · Is it worth dating and getting married?

Unofficial English translation of docs/结婚划不划算.md. In case of discrepancies the Chinese original takes precedence.

The reader is asking one bundled question: whether marriage is, on balance, a loss or a gain; whether the partner will provide emotional value, share the housework, earn money — or whether marrying is just a task performed for one's elders. As a single question it cannot be answered, because it adds together several things that cannot be converted into one another. This article takes it apart. Each account uses only numbers that can be found in official statistics or meta-analyses; wherever a figure cannot be computed, that is said plainly; at the end there is a worksheet you can fill in yourself. The article draws no conclusion about "whether to marry."

1. First, break the question apart

"Is marriage worth it" contains at least five accounts, each on a different measure:

  1. Money: direct spending (彩礼 (caili — bride price), the wedding, the marital home) and the legal rules on property. The former has no official statistics; the latter has statutory text.
  2. Time: housework, caregiving, and unpaid labor such as giving way in one's own work for the partner's sake. The measure is minutes per day.
  3. Health: the difference in mortality between married and unmarried people. The measure is relative risk, and it is correlation only.
  4. Legal risk: how property is divided, how long an exit takes, which obligations can be compensated. The measure is statute text.
  5. Relationship quality: emotional value, conflict, support. The measure is effect size, which cannot be converted into money or time.

The fifth is the hardest to quantify, and the closest to what the reader is actually asking.

2. The data and sources behind each account

Registration data: marriage and divorce rates are two crude rates

The Ministry of Civil Affairs' 《2024年民政事业发展统计公报》 (2024 Statistical Communiqué on the Development of Civil Affairs): in 2024, 6.106 million couples registered a marriage nationwide in accordance with the law, down 20.5% from the previous year, for a marriage rate of 4.3‰; 3.513 million couples completed divorce proceedings in accordance with the law, of which 2.622 million were registered divorces handled by civil affairs departments and 891,000 were divorces by court judgment or court mediation, for a divorce rate of 2.5‰. A note to the communiqué spells out the formula: the number of couples marrying (or divorcing) in the year, divided by the average total population of the year, multiplied by 1,000‰.

Both numbers take the entire population as their denominator, which is why they are called crude rates. They answer "how many couples per thousand people registered a marriage or a divorce this year"; they do not answer "will this marriage eventually end in divorce." The latter question needs cohort data, and I did not find it in the official texts.

The health account: married people have lower mortality, but this is correlation, not causation

Three meta-analyses give a consistent order of magnitude:

  • Manzoli et al. (2007) pooled 53 comparisons and more than 250,000 older adults. For all-cause mortality, the RR of the married relative to the unmarried (widowed, divorced or separated, never married) was 0.88 (95% CI 0.85–0.91); widowed 1.11, divorced or separated 1.16, never married 1.11. The result did not vary by sex, study quality, or European/North American region; the most conservative recalculation gave an RR of 0.94.
  • Roelfs et al. (2011) pooled 95 studies and more than 500 million people. The HR for all-cause mortality of the never married relative to the married was 1.24 (1.19–1.30). The effect diminished with age, and the higher the study quality, the smaller the effect.
  • Wang et al. (2020) pooled 21 studies and more than 7.89 million people. Being unmarried was associated with all-cause, cancer, and cardiovascular mortality in both sexes, with stronger cardiovascular and all-cause associations in men; among the never married, men had a 31% higher risk of death from stroke and 9% higher all-cause mortality than women.

All three are observational studies. People who are healthy, well off, and socially connected are both more likely to marry and more likely to live long — this is the selection effect. Manzoli found signs of publication bias. These numbers should be read as "mortality is somewhat lower among the married," not as "marriage will make you 12% less likely to die." The samples are mostly European and North American; for China they can serve only as an order-of-magnitude reference.

The time account: the gender gap in unpaid labor is an upper-bound reference

The National Bureau of Statistics' Third National Time Use Survey (May 2024): among those who took part in unpaid labor, the daily average was 2 hours 45 minutes — 1 hour 52 minutes for men, 3 hours 29 minutes for women; the participation rate was 75.6%, men 67.5%, women 83.9%. By activity, the population average for housework was 1 hour 17 minutes per day (1 hour 59 minutes among participants, 28 minutes less than in 2018; participation rate 64.9%); accompanying and caring for family members averaged 30 minutes (1 hour 46 minutes among participants; participation rate 28.4%).

The second survey, in 2018: the population average for housework was 1 hour 26 minutes — 45 minutes for men, 2 hours 6 minutes for women — with participation rates of 40.4% for men and 75.6% for women. Accompanying and caring for children day to day averaged 36 minutes — 17 minutes for men, 53 minutes for women — with participation rates of 12.3% for men and 25.1% for women.

Both surveys break results down only by sex, urban or rural residence, and age — not by marital status — so there is no official number for "how much more housework per day after marriage." What can be used is the gender gap: in 2024, unpaid labor differed between men and women by 1 hour 37 minutes per day, which can serve as an upper bound for the gap that may open up when the division of labor is never talked through.

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

There are no official statistics on bride price, weddings, or the marital home, so this article states no numbers. What can be stated are the rules in the Marriage and Family book of the 《民法典》 (Civil Code), in force since January 1, 2021:

  • Article 1,062: Wages, bonuses, and remuneration for labor services acquired during the marriage; income from production, business operations, and investment; income from intellectual property; and property acquired by inheritance or as a gift (except where a will or a gift contract clearly specifies it belongs to one party alone) are community property of the husband and wife, who have equal rights to dispose of it.
  • Article 1,063: One spouse's premarital property; compensation or damages received for personal injury; property specified in a will or a gift contract as belonging to one party alone; and articles of daily use for one spouse's exclusive use are that spouse's separate property.
  • Article 1,065: A man and a woman may agree that premarital and postmarital property be owned separately, jointly, or in part separately and in part jointly. The agreement must be in writing and is legally binding on both parties. Where there is no agreement or the agreement is unclear, the preceding two articles apply.
  • Article 1,088: Where one spouse bears the heavier obligations — raising children, caring for elderly family members, assisting the other spouse's work — that spouse has the right to request compensation from the other at the time of divorce. The specifics are settled by agreement between the two; failing agreement, the court decides.

These rules work in both directions: the higher-earning side looks at Articles 1,062 and 1,065; the side carrying more of the unpaid labor looks at Article 1,088. An agreement only has to be in writing; notarization is not a statutory requirement.

Exit costs: the cooling-off period and the conditions for a court case

  • Article 1,076: Where both spouses agree to divorce, they sign a written divorce agreement and apply in person at the marriage registration office. The agreement must record their agreed positions on the raising of children and on the handling of property and debts.
  • Article 1,077: Within thirty days of the day the registration office receives the application, either spouse may withdraw it if no longer willing to divorce. Within thirty days after that period ends, both spouses must apply in person for the divorce certificate; if they do not apply, the application is deemed withdrawn.
  • Article 1,079: A spouse who wants a divorce may go through mediation or file suit directly. The court is to mediate first, and is to grant the divorce where the relationship has indeed broken down and mediation has failed. Where mediation fails, divorce shall be granted in cases of bigamy or cohabitation with another person; domestic violence or the mistreatment or abandonment of family members; gambling, drug use, or other vices that persist despite repeated correction; or separation for two full years because the relationship has broken down. Where the court has denied a divorce and the parties have then been separated for one full year and one party sues again, the divorce shall be granted.

Of the 3.513 million divorces in 2024, 891,000 went through court judgment or court mediation. The cooling-off period applies only to registered divorce.

Relationship quality: an account worth more attention than "married or not"

Robles et al. (2014) pooled 126 studies and more than 72,000 people: the higher the marital quality, the better the health, with effect sizes r between 0.07 and 0.21 — including r = 0.11 for mortality risk and r = −0.13 for cardiovascular reactivity during marital conflict. These effect sizes are small, but comparable in magnitude to the association between health and health behaviors such as diet. Some outcomes showed publication bias. Studies that explicitly tested for sex differences found essentially no difference between men and women. Most of the designs cannot support causal inference. Dhindsa et al. (2020) likewise note in their cardiovascular review that marital dissatisfaction and marital quality have a significant effect on cardiovascular risk.

This article did not find a meta-analysis that directly compares mortality in a "low-quality marriage" with the unmarried. "A bad marriage is worse than not marrying" is currently an inference, not a numerical conclusion. The statement that holds up is this: the effect of "married or not" carries a selection effect; "good quality or bad" is another independent variable. A decision made on status alone, without looking at quality, leaves the second account out of the calculation.

3. How to run the numbers yourself: a fill-in worksheet

Fill in each account by its own measure, and do not add numbers across measures. Where a cell cannot be filled, write "don't know" — that is information in itself.

What the partner can provide (by measure)

  • Money: what the partner's income, debts, and premarital property are; whether marital property follows the default rules or a written agreement, and what the agreement says.
  • Time: how much unpaid labor the partner does per day now (national average among participants: men 1 hour 52 minutes, women 3 hours 29 minutes); how it would be divided after marriage; how it would be divided when there are children or elderly to care for; whose work can give way.
  • Relationship quality: how many conflicts there have been in the past six months and how they ended; what the partner did when you were ill, unemployed, or low; what you did in the same situations.
  • Emotional value: whether your energy increases or drains when you are with the partner. This takes months of time together to judge; a list of conditions will not show it.

What you will put in (by measure)

  • Money: how much will be spent on the marriage itself (bride price, wedding, marital home), and where that money would go instead if you did not marry.
  • Time: how much unpaid labor you expect to add per day, and how much 人情往来 (renqing — reciprocal visits and gifts between families) per week.
  • Health: whether your daily routine, diet, and exercise will get better or worse because of the marriage. This line is closer to you than any RR in a meta-analysis.
  • Exit cost: registered divorce takes a 30-day cooling-off period plus a further 30 days to apply for the certificate; going to court requires meeting the conditions of Article 1,079. Whether you can bear that timeline is a question only you can answer.

How the law protects you (applies to both sides)

  • Is there documentation for premarital property (home purchase contract, bank statements, gift contract, will), and can it prove the property is "premarital" or "clearly designated for one party alone"?
  • Should there be a written property agreement, and would it cover premarital property, marital acquisitions, or both?
  • After marriage, who will bear more of the unpaid labor; does that person know about the right to compensation under Article 1,088; are both of you willing to agree on a compensation method in advance?
  • Debts: of money borrowed in one person's name during the marriage, which counts as joint debt, and does each of you know about it?

Marrying for the elders (fill this column separately)

  • What the elders gain: peace of mind, 面子 (mianzi — social face), an end to the pressure to marry, someone to provide for them in old age.
  • Which of these gains belong to the elders, and which are actually your own.
  • The costs in the three columns above are borne entirely by you; the elders' gains do not automatically turn into your health or your relationship quality.
  • Written out separately, whether this account weighs heavy or light is for the person filling in the sheet to judge.

4. Common mistakes

Mixing up divorce-rate measures. The crude divorce rate (divorcing couples per thousand people; 2.5‰ in 2024) and the divorce-to-marriage ratio (divorces in the year divided by marriages in the year; 3.513 million / 6.106 million ≈ 57.5%) are two different things. The numerator and the denominator of the ratio are not the same people: most of those divorcing this year married years ago, and a falling number of marriages mechanically pushes the ratio up. It does not mean "half of all marriages end in divorce," and the communiqué contains no such indicator.

Treating correlation as causation. That the married have mortality about 12% lower is the pooled result of meta-analyses, but health, income, and social life each independently affect both "can one marry" and "can one live long." The account can be computed, but it must be discounted, and it cannot stand in for a judgment about this particular relationship.

Treating marriage as a task. A task-driven marriage looks only at "status" and not at "quality," while the effect of relationship quality on health is a separate account, independent of marital status, and is usually omitted from task-driven decisions.

Taking national averages for your own numbers. The gender gap in the time use surveys is a national average; the property rules are default rules that can be changed in writing. National numbers are for magnitude and for a starting point in negotiation, not for conclusions.

Where there is no official number, do not fill the gap with unofficial ones. Bride price, weddings, and the cost of raising children have no official statistics; this article leaves them blank. A reader's own numbers, filled in by the reader, are more accurate than any institution's estimate.

Sources

This essay is reproduced from the open-source HowToLiveBetter project, preserving the original wording and verification notes. Refer to the original repository for updates.

View source revision · cc91bd6Imported: 2026-09-29

Entry links point to the matching topic in this site’s imported edition; source numbering may differ from the current entry number.

Explore another topic

01 / LONG READ

Home Emergency Supplies Checklist: What to Buy, Where to Put It, and How Often to Check

What to keep at home, where to put it, when to check it, and what you can skip.

Read the essay
02 / LONG READ

A stranger has an accident on the street: walk away or stop

Understand the costs and boundaries of walking away, calling for help and intervening.

Read the essay