on equilibrium transport
and matching models in economics
June 21-25, 2021
This intensive course, part of the 'math+econ+code' series, is focused on the computation of competitive equilibrium with substitutes, which is at the core of surge pricing engines and allocation mechanisms. It will investigate diverse applications such as surge pricing, matching platforms and network congestion. It provides a bridge between theory, empirics and computation and will introduce tools from economics, mathematical and computer science. Students are expected to write their own code (in Python), in the spirit of cooking lessons. Mathematical concepts (such as lattice programming, supermodularity, discrete convexity, Perron-Froebenius theory, etc.) will be taught on a needs basis while studying various economic models. The same is true of computational methods (such as tatonnement algorithms, asynchronous parallel computation, mathematical programming under equilibrium constraints, etc.). Hence there are no prerequisite other than the equivalent of a first-year graduate sequence in econ, applied mathematics or other quantitative disciplines.
This course is very demanding from students, but the learning rewards are high.
• The course will be taught online over five consecutive days, June 21-25, 2021, 8am - 12noon US Eastern time / 2pm-6pm Paris time.
• Additionally, five advanced lectures will be taught online on a monthly basis afterwards, July - November 2021. Precise days of the lectures will be announced later.
Applications are now closed
Day 1: competitive equilibrium
with gross substitutes
Monday 5/21, 4h
Walrasian equilibrium and gross substitutes. Hedonic pricing. Jacobi algorithm.
Day 2: matching models with transferable utility
Tuesday 5/22, 4h
Optimization and equilibrium formulation. Duality. Computation by descent methods. Sinkhorn's algorithm.
Day 3: matching models with imperfectly transferable utility
Wednesday 5/23, 4h
Distance-to-frontier function, matching function equilibrium. Matching models with taxes. Collective models with public consumption.
Day 4: matching models with non-transferable utility
Thursday 5/24, 4h
Gale and Shapley’s deferred acceptance algorithm. Adachi’s algorithm and Tarski's fixed point theorem. Aggregate stable matchings.
Day 5: one-to-many matching models
Friday 5/25, 4h
Kelso and Crawford’s deferred acceptance algorithm. Hatfield and Milgrom’s model of matching with contracts.
+ 5 advanced lectures
July to November, 2h each
Lectures on related advanced mathematical / computational topics to complement the masterclasses and solidify the theoretical knowledge.
Advanced lecture 1
Z-, P- and M-matrices, diagonally dominant matrices, Stieltjes matrices, and Perron-Froebenius theory. M-maps.
Advanced lecture 2
Lattices, submodularity and Topkis’ theorem. Veinott's order. Monotone comparative statics. Strategic complementarities. Uniform Gross Substitutes.
Advanced lecture 3
Discrete convex analysis, L / L# / M / M# convexity.
Advanced lecture 4
Equilibrium pricing on networks. Dynamic programming and Bellman-Ford’s algorithm. The min-cut-max-flow theorem. The Ford-Fulkerson algorithm
Advanced lecture 5
Principles of parallelization; computing on GCP and AWS; introduction to TensorFlow and Keras.