AØKK08207U Dynamic Programming - Theory, Computation, and Empirical Applications

Volume 2014/2015
Education
MSc in Economics
Content

The overall purpose of the course is to provide a fundamental understanding of dynamic programming (DP) models and their empirical application. The DP framework has been extensively used in economic modeling because it is sufficiently rich to model almost any problem involving sequential decision making over time and under uncertainty. Prominent examples are saving/consumption decisions, retirement behavior, investment, labor supply/demand, housing decisions. The course will first introduce participants to theoretical concepts, and then focus on empirical applications covering both discrete and continuous decision problems as well as the estimation of dynamic games. 
 

Learning Outcome

The purpose of the lectures and the exercise classes is that the student should

  • Acquire knowledge about stochastic dynamic programming and the involved computational hurdles (Curse of dimensionality, high dimensional integration, multiplicity of solutions, etc.)
  • Acquire knowledge about solution methods for dynamic structural models of sequential decision making under uncertainty of both finite and infinite horizons (backward recursion, value function iterations, policy iterations, and Mathematical Programming with Equilibrium Constraints (MPEC), Carroll's method of endogenous gridpoints (EGM)).
  • Acquire knowledge about estimation methods (full solution methods: Mathematical Programming with Equilibrium Constraints (MPEC) and Nested Fixed Point Algorithm (maximum likelihood, minimum distance, indirect inference, GMM  and simulation versions of these); Non-full solution methods: CPP-estimator, Nested Pseudo likelihood (policy iteration estimators); GMM using Euler equations).
  • Acquire knowledge about numerical techniques to evaluate integrals (Quadrature methods and  Monte Carlo integration) involved in evaluating expectations future states of the world and to integrate unobservable out of the sample criterion used in estimation (e.g. the likelihood function).
  • Acquire knowledge about the numerical approximation and interpolation techniques required to approximate value functions over continuous state variables (splines, orthogonal polynomials, neural net).
  • Acquire knowledge about a variety of dynamic structural models
  • Acquire knowledge about how evaluate policy initiatives by means of counter factual simulations
  • Get hands on experiences with solving and/or estimating relatively simple models (cake eating, stochastic growth, consumption/savings, investment, labor demand/supply and simple dynamic games).

 

The purpose of the term paper is to make students combine many of the simplified building blocks we covered in the computer exercises. By combining these building blocks, students should be able to solve and estimate more sophisticated model.In particular the students should be able to

  •  Solve and estimate dynamic games or single agent models and test hypotheses using solution and estimation methods discussed in the course. Ideally students should be able to replicate the results from an already published paper and thereby get hands on experience with the involved techniques.
  • Investigate the consequences policy proposals by means of counterfactual simulations program the estimators applied in the paper using MATLAB (or GAUSS, FOTRAN and C).
  • Present the analysis in a short and focused term paper.

 

Hence, after completing the course, the student should have acquired the tools necessary to understand papers and undertake empirical analysis on a (simple) dynamic structural model and to present the analysis in a short and focused paper. The acquired skills in dynamic programming theory and practice provide a strong background that enable students to do empirical analyses at a high level suitable for a Master or even a PhD thesis.

  • Jérome Adda and Russell Cooper: “Dynamic Economics: Quantitative Methods and Applications” MIT Press 2003, ISBN: 978-0-262-01201-0
  • Kenneth Judd: “Numerical Methods in Economics” MIT Press 1998, ISBN: 978-0-262-10071-7
  • 15-20 papers: Ranging from classic seminal contributions to recent state of the art work from the research frontier.
Pre-requisites are third year micro- and macroeconomics and quantitative methods 1-3 as well as Advanced Microeconometrics. While the latter is essential as this course provides the necessary computational skills (MATLAB Programming) and knowledge about estimation techniques, it is not a formal requirement.
2 x2 hours of lecture per week for 10½ week and 2 hours of exercise for 12 weeks and some weeks to write the paper.

The lectures focus on theory whereas the class provides hands on knowledge of solution and estimation of the models. Ideally, the whole process of estimating a dynamic structural model empirically is learned by writing a term paper that has to be handed in at the end of the semester.

Time and room:
Time and room for the lectures and exercise classes: Please press the link under "Se skema" (See schedule). 15F means Spring (Forår) 2015

The springsemester is available partly in English at this link:
https:/​/​skema.ku.dk/​KU1415/​reporting/​textspreadsheet?objectclass=module&idtype=id&identifier=43809&t=SWSCUST+module+textspreadsheet&days=1-5&weeks=27-52&periods=1-68&template=SWSCUST+module+textspreadsheet

The first lecture in the springsemester will be the 2.th of February 2015 at 8.00 o´clock (AM). The exercise classes begins in week 6.
  • Category
  • Hours
  • Class Instruction
  • 24
  • Exam
  • 0,42
  • Lectures
  • 42
  • Preparation
  • 160
  • Total
  • 226,42
Credit
7,5 ECTS
Type of assessment
Written assignment
Oral examination, 25 min under invigilation
A written termal paper and an oral exam without preparation.
Exam registration requirements
Project description and term paper should be handed in before deadline.
Aid

All aids can be used to the project assignment.

The student can only take the project assignment in to the oral examination otherwice it is a closed book examination.

 

 

Marking scale
7-point grading scale
Censorship form
External censorship
100 % censorship
Exam period
For the Spring semester 2015: From 26 May to 28 June 2015 The projct description has to be uploaded not later than monday the 6. of April 2015. The deadline for the assignment will be informed by the teacher in the begin of the semester. The oral exam wil be in the week number 22 of 2015. More information is available at https://intranet.ku.dk/economics_ma/examination/Pages/default.aspx
Re-exam
As ordinary with the same assignment.
Criteria for exam assesment

The student must in a satisfactory way demonstrate that he/she has mastered the learning outcome of the course.