China relaxed its strict One-Child Policy to universally allow couples to have two children in 2016. Although the new policy suggests an improvement in welfare for couples, as they now have more freedom to achieve their desired fertility levels, it has the drawback of possibly increasing gender inequality both in the labor market and within the household. This paper starts with a difference-in-difference method to show that the new policy increased the gender wage gap between women and men and negatively affected the intrahousehold bargaining power of women. Motivated by this empirical pattern, I then build and estimate a dynamic collective household model to quantify the welfare impact of the new policy on both genders using a novel machine learning method and indirect inference. The results suggest that the welfare cost of the Two-Child Policy for women is equivalent to 6.00% of lifetime consumption, while the welfare benefit of the policy for men is equivalent to 7.23% of lifetime consumption. Policy experiments suggest that implementing anti-discrimination laws for women in the labor market significantly improves women’s welfare while providing public childcare subsidies is most effective in stimulating fertility in the post-policy era.

This paper proposes a novel method to identify the One-Child Policy’s impact on couples’ childbearing using self-reported survey measures. We use couples’ pre-policy ideal number of children together with the answers in the post-policy period to back out the counterfactual number of children without the One-Child Policy. Findings indicate a significant average re- duction of 0.2714 children per couple in 2014 due to the policy. Variations in policy effects are explored across educational, urban/rural, and occupational groups, with highly educated urban women in government jobs experiencing the most pronounced impact. Sub-region analysis suggests significant policy stringency differences among provinces.

Small businesses/entrepreneurs increasingly rely on crowdfunding platforms to raise funds for their entrepreneurial projects. While such a financing strategy enables entrepreneurs and funders to interact with fewer geographic constraints, we find empirical evidence consistent with strong local biases among funders in online crowdfunding markets. What drives these biases? Our results suggest that in addition to funders’ local preferences, information frictions play a more important role in driving the local biases, preventing high-quality entrepreneurial projects from being identified and funded by investors outside their regions. Exploring the role of platform-design features, we find that, with the presence of strong local biases, not reveal- ing locational information of projects could be welfare improving. Business strategies could be tailored for different project categories to help mitigate non-local funders’ informational disadvantages and foster local entrepreneurship.

Product developers nowadays are able to automate product and service design in a large scale without consumer preference data. However, such design may not meet the needs of heterogeneous consumers with dynamically evolving preferences. Leveraging a unique unmanned photo gallery setting where consumers choose templates to self-take photos, we study how the design templates can be automated efficiently in large scale, while incorporating heterogeneous consumer preference. Our approach can improve the product design when facing “cold-start” problem by leveraging external data of user generated content (UGC) from external websites. We first estimate a “deep” choice model using deep neural network, and then we collect a set of external photos and remove human subjects to create templates that are similar to the internal templates offered by the firm. We then leverage both internal and external data to generate new templates. We compare templates generated using generative adversarial network (GAN) without incorporating consumer preferences and advanced generative models, conditional-DCGAN, with preference information. We show how incorporating preference information can improve the quality of generated templates.