UCL School of Management

Research seminar

Chul Kim, University of Maryland


Chul Kim, University of Maryland


Friday, 28 October 2016
15:00 – 16:30

UCL School of Management is delighted to welcome Chul Kim, University of Maryland, to host a research seminar on ‘Modeling dynamics in crowdfunding.’


We investigate forward-looking behavior in investment decisions and dynamic network effects in crowdfunding. Individuals decide what to invest in and how much to invest by trading off early investments not to lose a chance to participate against delayed investments to reduce the uncertainties. In addition, individuals refer to their connected network’s decisions, while anticipating the effect of their own decisions on connected network’s decisions. 

From the methodological viewpoint, we propose a dynamic structural choice model by combining a linear-quadratic approach from the dynamic programming literature and multivariate choice model in order to deal with the multiple-discrete/continuous investment decisions: what to invest in and how much to invest. We derive a closed form likelihood function by analytically solving the optimal conditions of the continuous investment amount decisions to reduce the computational burden, which has been a hurdle to trace dynamics of multiple-discrete/continuous choice outcomes. We explicitly capture the inter-temporal characteristics of individual network and allow sequential two-way interactions between a focal individual and his/her connected network. Moreover, we suggest an estimation algorithm based on the Bayesian IJC method not only to estimate individual-level parameters but also to handle various types of states variables (continuous, degenerate continuous, and discrete state) simultaneously.

Empirically, the proposed model is applied to individual-level investment and network data directly collected from Sellaband, a crowdfunding platform for musicians. We find strong evidence of forward-looking behavior in investment decisions and the presence of dynamic network effects. Our proposed model accommodates various dynamic investment patterns of each crowdfunding project such as stagnation, gradual increase, and accelerated investments. Using counterfactuals, we examine various strategies for accelerating investments in a stagnating project. We obtain the largest possible amount as the goal for the fundraising project which guarantees the success of the project and therefore raises the most funding for the project. Also, we suggest more effective conditions for the investment-matching promotional campaign. Our results reveal that the effect of earlier promotions lasts longer than the later promotions because the early-stage network characteristics are more malleable. As a result, we find that earlier promotions garner more investments than later promotions. Finally, our findings show that individuals who have a larger set of co-investors are more influential than others in stimulating the participation of other investors.

Open to
PhD Programme
Last updated Tuesday, 25 October 2016