The family system emphasized the transmission of family common-property along patrilineal lines in imperial China, and only sons were responsible for parental old-age security and the household labor capacity. The daughters were merely temporary members of birth family, when they got married, they were legally cut off from birth family, physically and financially, and joined their husband family. Since they normally had no responsibilities left for the birth family when married out, having a daughter was usually considered to be a pure economic burden, parents and brothers preferred to take advantage of their unpaid housework to reap a high return. In this paper, we argue that an unmarried sister was a family public good provider since she was a temporary member, and birth family had economic motivations to free ride on her unpaid housework to improve brother’s educational outcome. We use the discrete-time event history analysis method with individual fixed effects based on a unique historical population panel data, the China Multi-Generational Panel Dataset-Liaoning (CMGPD-LN), and empirically investigate the relationship between unmarried sisters and brother educational outcome in imperial Chinese family to prove our argument. We find that an unmarried sister increased the probability of her brother to obtain an official student title, and this effect was mainly driven by younger unmarried sister since women married much younger than men. Our argument implies that unmarried sisters should increase brother outcomes on multiple aspects. We use brother’s fertility outcome as another outcome variable, and find that an unmarried sister also increased the probability of her brother giving to another birth. Furthermore, we investigate the effects among male siblings since brother’s age and status may cause different results, and we find that an older brother had a similar effect, while a younger brother was not. This paper has two major contributions. First, by using a unique historical panel data and employing fixed effects model, this paper is able to investigate and focus on the time-varying household effects on individual outcomes, as well as controlling the time-invariant effects within families. In contrary, since most modern census data are cross-sections, previous studies hardly identified the relationships between specific factors and final outcomes without knowing the time-varying effects. Second, this paper provides an alternative explanation to the sibling effects from the perspective of free-riding behavior, while existing mechanism cannot explain our findings since women were not allowed to be educated or take a job, an unmarried sister cannot bring financial support for her brother’s educational expenditure.