Predicting the Demand for Rides - An Application to NYC Taxi Data
I consider the problem of modeling and prediction of spatially correlated demand. I propose an algorithm for inferring unmet taxi demand and explore different methods for predicting met and unmet demand. I compare the performance of linear time series models and neural networks in forecasting spatially correlated time series.
Spatial Equilibrium and Search Frictions - an Application to the NYC Taxi Market
This paper uses a dynamic spatial equilibrium model to analyze the effect of matching frictions and pricing policy on the spatial allocation of taxicabs and the aggregate number of taxi-passenger meetings. A spatial equilibrium model, in which meetings are frictionless but aggregate matching frictions can arise endogenously for certain parameter values, is calibrated using data on more than 45 million taxi rides in New York. It is shown how the set of equilibria changes for different pricing rules and different levels of aggregate market tightness, defined as the ratio of total supply to total demand. Finally, a novel data-driven algorithm for inferring unobserved demand from the data is proposed, and is applied to analyze how the relationship between demand and supply in a market with frictions compares to the frictionless equilibrium outcome.
Estimation Of Peer Effects In Endogenous Social Networks: Control Function Approach
Joint with Hyungsik Roger Moon
We propose a method of estimating the linear-in-means model of peer effects in which the peer group, de ned by a social network, is endogenous in the outcome equation for peer effects. Endogeneity is due to unobservable individual characteristics that in influence both link formation in the network and the outcome of interest. We use estimates of the unobserved individual effects as a control function to control for the endogeneity of the social network matrix in the outcome equation for peer effects. We leave the functional form of the control function unspecified and treat it as unknown. To estimate the model, we use a sieve semiparametric approach, and we establish asymptotics of the semiparametric estimator.
Revision requested at The Review of Economics and Statistics
This paper proposes a new double-question survey method that elicits information about how individuals' subjective belief valuations are compared and related to their price expectations. An individual respondent is presented with two sets of questions, one that asks about his/her belief regarding the value of an asset (whether it is over- or under-valued), and another regarding his/her expectations of the future price of that asset. Responses to these two questions are then used to measure the extent to which prices are likely to move towards or away from the subjectively perceived fundamental values. Using a theoretical asset pricing model with heterogenous agents we show that there exists a negative relationship between the agents expectations of price changes and their asset valuation. Double question surveys on equity, gold and house prices provide evidence in support of such relationships, particularly in the case of house price expectations. The effects of demographic factors, such as sex, age, education, ethnicity, and income are also investigated. It is shown that for house price expectations such demographic factors cease to be statistically significant once we condition on the respondents' location and their asset valuation indicator. The results of the double-question surveys are then used to construct leading bubble and crash indicators, and their potential value is illustrated in the context of a dynamic panel regression of realized house price changes across a number of key Metropolitan Statistical Areas in the US.
A Bayesian Comparison of Models of Network Formation
A prominent feature of real-world social networks is a high level of clustering.
I review different approaches to modeling network formation and clustering
and I apply Bayesian model selection to evaluate the models. Preliminary
results confirm that models that treat links as pairwise independent do not
generate the levels of clustering observed in the data. Models that include
unobserved heterogeneity perform slightly better than models with only observable