Book contents
- Frontmatter
- Contents
- List of figures
- List of tables
- List of contributors
- 1 The enigma of simplicity
- PART I The importance of simplicity
- PART II Simplicity in theory and practice
- 8 Simplicity, information, Kolmogorov complexity and prediction
- 9 Simplicity and statistical inference
- 10 Rissanen's theorem and econometric time series
- 11 Parametric versus non-parametric inference: statistical models and simplicity
- 12 The role of simplicity in an econometric model selection process
- 13 Simplicity in a behavioural, non-parametric context
- 14 Keep it sophisticatedly simple
- 15 Communication, complexity and coordination in games
- 16 The simplicity of an earnings frontier
- 17 Simplicity: views of some Nobel laureates in economic science
- Index
12 - The role of simplicity in an econometric model selection process
Published online by Cambridge University Press: 22 September 2009
- Frontmatter
- Contents
- List of figures
- List of tables
- List of contributors
- 1 The enigma of simplicity
- PART I The importance of simplicity
- PART II Simplicity in theory and practice
- 8 Simplicity, information, Kolmogorov complexity and prediction
- 9 Simplicity and statistical inference
- 10 Rissanen's theorem and econometric time series
- 11 Parametric versus non-parametric inference: statistical models and simplicity
- 12 The role of simplicity in an econometric model selection process
- 13 Simplicity in a behavioural, non-parametric context
- 14 Keep it sophisticatedly simple
- 15 Communication, complexity and coordination in games
- 16 The simplicity of an earnings frontier
- 17 Simplicity: views of some Nobel laureates in economic science
- Index
Summary
Introduction
Most of the model selection strategies found in the econometric literature are based on the use of statistics defined in terms of a mix of fit and parsimony. The aim of this chapter is to analyse the form in which the different model selection criteria combine these two elements and to demonstrate the need to define a trade-off between one of them in function of the other, given that they move in opposite directions.
We will first comment on a line developed within the philosophy of science which justifies the use of the evaluation of alternative theories following a bi-polar procedure. Reference will also be made to a number of relevant contributions formulated in the same direction within economics.
We will then propose a general expression, of which the majority of the model selection criteria developed in econometrics are particular cases. This general expression explicitly combines the fit and parsimony indicators. We will also study the form adopted by the parsimony factor that corresponds to each one of the criteria.
Finally, we will propose a framework based on the use of the mean square error of prediction (MSEP) and study the conditions that the parsimony factor must comply with.
- Type
- Chapter
- Information
- Simplicity, Inference and ModellingKeeping it Sophisticatedly Simple, pp. 207 - 226Publisher: Cambridge University PressPrint publication year: 2002