Econometrics of Networks

Learn about recent econometric methods to analyze network data. Networks play an increasingly dominant role in many social, business, and economic environments. Moreover, network data becomes increasingly important and available due to the rise of online social media and digitization.

The course will combine online lectures with hands-on empirical and programming exercises.

Upon successful completion of the course, students will:
- become acquainted with different statistical methodologies for analysing networks while learning how to see these different methodologies complementing each other.
- learn to model network problem situations mathematically, and adapt the methods learned to new situations at hand.
- be able to recognise, understand, and analyse societal and business problems in which networks are central.
- learn how networks affect supply and demand in markets, how this leads to market failures, and how government policies can address these.

1. Examples of Networks and Data
2. Network Statistics, Visualization and Graphs
- Elements of Graph Theory
- Graphs and Matrices
- Bipartite Graphs
- Core-periphery Networks and Nested Split Graphs
- Network Statistics: Average path length, clustering and assortativity
- Centrality in Networks: Degree, eigenvector, Katz-Bonacich centrality and Google's Page Rank
- Network Visualization: Force-directed, circular and layered layout
3. Econometrics of Interactions in Networks
- Spatial Autoregressive (SAR) Model
- Linear Quadratic Utility
- Endogeneity of the Spatial Lag
- Two-Stage Least Squares (2SLS)
- Maximum Likelihood Estimation (MLE)
- Identification Issues
- Correlated Effects, Sorting and Selection
- Endogenous Link Formation
- Multiple Spatial Weight Matrices
- Spatial Panel Data

4. Econometrics of Network Formation
- Exponential Random Graph Model (ERGM)
- Conditional Edge-Independence
- Erdös-Rényi Random Graph
- Logistic Regression
- Unobservable Characteristics (beta-model)
- Tetrad Logit Estimator
- Random Utility Model
- Maximum Likelihood Estimation (MLE)
- Markov Chain Monte Carlo
- Gibbs Sampling
- Metropolis Hastings Algorithm
- Stochastic Block Model (SBM)
- Temporal ERGM
5. Joint Estimation of Outcomes and Network Formation
5.1. Coevolution of Networks and Behaviour: An application to R&D collaboration networks
- Structural Model: Utility and the potential game
- Estimation
- Computational Problem and the Exchange Algorithm
- Double Metropolis-Hastings (DMH) Algorithm
- Unobserved Heterogeneity
- Empirical Illustration: R&D collaborations
5.2. Network Formation with Multiple Activities: An application to team production and co-authorship networks
- Bipartite Network, Production Function, and Utility
- Equilibrium Characterization and Line Graphs
- Estimation with Endogenous Matching
- Empirical Illustration: Co-authorship networks
6. Spatial Modelling Approach for Dynamic Network Formation and Interactions
- Spatial Dynamic Panel Data (SDPD) Model
- A General Dynamic Network Formation Model
- Combining SDPD with the Network Formation Model: Joint likelihood function
- An Empirical Application to Peer Effects in Academic Performance
7. Big Data Meets Networks
- The Digital Layer: How innovative firms relate on the Web
- Automated Robot for Generic Universal Scraping (ARGUS)
- Input, Interface and Output of ARGUS
- Sectoral Hyperlink Network
- Hyperlink Types

Contact hours listed under a course description may vary due to the combination of lecture-based and independent work required for each course. CEA's recommended credits are based on the contact hours assigned by Vrije Universiteit Amsterdam (VU Amsterdam): 15 contact hours equals 1 U.S. credit