27
Introduction Advanced Econometrics - HEC Lausanne Christophe Hurlin University of OrlØans October 2013 Christophe Hurlin (University of OrlØans) Advanced Econometrics - HEC Lausanne October 2013 1 / 27

Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

  • Upload
    others

  • View
    87

  • Download
    2

Embed Size (px)

Citation preview

Page 1: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

IntroductionAdvanced Econometrics - HEC Lausanne

Christophe Hurlin

University of Orléans

October 2013

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 1 / 27

Page 2: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Instructor Christophe Hurlin

Contact [email protected]

Teaching assistant Sara Cavalli

Personal websitehttps://hec.unil.ch/docs/index.php/churlin/cours/512/session_1461Personal websitehttp://www.univ-orleans.fr/deg/masters/ESA/CH/churlin_E.htm

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 2 / 27

Page 3: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

What is econometrics? Not an easy question!

Main objective of this course is to de�ne this term!

Econometrics can be de�ned as the statistical analysis of economic(�nancial) phenomena.

"Econometrics is the quantitative analysis of actual economicphenomena based on the concurrent development of theory andobservation, related by appropriate methods of inference", P. A.Samuelson, T. C. Koopmans, and J. R. N. Stone (1954)

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 3 / 27

Page 4: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

Econometrics is fundamentally based on four elements:

1 A sample of data

2 An econometric model

3 An estimation method

4 Some inference methods

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 4 / 27

Page 5: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

Question: Why using a sample?

Let us assume that we want to study a characteristic / property xof the individuals of a population.

The individuals (unit) of the population are not necessarily somepersons: it can be �rms, assets, countries, time index etc..

The characteristic x may be quantitative (salary, weight, total asset,GDP etc.) or qualitative (social status, genre etc.)

The characteristic x may be stochastic or deterministic (weight,size etc..).

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 5 / 27

Page 6: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

Adam and Eve, Titian (1490-1576)

In the case of a population oftwo individuals, inference,econometrics, etc (and thiscourse.).. are useless.

Let us imagine that Adamweighs 80 kg and Eve 50 kg...

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 6 / 27

Page 7: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

When the population is largeor in�nite, sampling is theonly mean to study the weight

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 7 / 27

Page 8: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

De�nition (Population)A population can be de�ned as including all people or items with thecharacteristic one wishes to understand.

1 In most of cases, it is impossible to observe the entire statisticalpopulation, due to cost constraints, time constraints, constraints ofgeographical accessibility.

2 A researcher would instead observe a statistical sample from thepopulation in order to attempt to learn something about thepopulation as a whole.

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 8 / 27

Page 9: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

In most of cases, the sample is random:

De�nition (Probability sampling)A probability sampling is a sampling method in which every unit in thepopulation has a chance (greater than zero) of being selected in thesample.

Consequence: a sample is a collection of random variables even thecharacteristic x is deterministic.

sample: fX1,X2, ...,XNg

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 9 / 27

Page 10: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

Example (random sample)

Let us consider a population of four persons and denote by ex the weight(assumed to be non stochastic) of the individual with:

exA = 80 exB = 50 exC = 40 exD = 90Consider a random sample of N = 2 individuals denoted by8<: X1|{z}

Weight of the �rst indi. selected in the sample

,X2

9=;So we can obtain a realisation

fx1, x2g = f50, 80g or fx1, x2g = f90, 40g or fx1, x2g = f90, 90g etc.

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 10 / 27

Page 11: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

FactWhatever the assumption made on the characteristic X (deterministic orstochastic) the result of the probability sampling is a random sample, i.e.a collection of random variables X1, X2, ..,XN .

FactGiven the sampling probability method used, we can assume that theserandom variables are independent and identically distributed (i.i.d.).

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 11 / 27

Page 12: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

FactIn general, in economics and �nance, only one realisation of the sampleis available: this is your data set!

fx1, x2, .., xNg

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 12 / 27

Page 13: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

The challenge of econometrics and mathematical statistic is to drawconclusions about a population (or the true DGP) after observing only onerealisation fx1, ..xNg of a random sample (your data set..).

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 13 / 27

Page 14: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

In econometrics, data come from one of the two sources: experiments andnon experimental observations

1 Experimental data are based on (randomized controlled)experiments designed to evaluate a treatment or policy or toinvestigate a causal e¤ect.

2 Data obtained outside an experimental setting are calledobservational data (issued from survey, administrative records etc...)

All of this lecture is devoted to methods for handling real-worldobservational data

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 14 / 27

Page 15: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

Whether the data is experimental or observational, data sets can be mainlydistinguished in three types:

1 Cross-sectional data

2 Time series data

3 Panel data

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 15 / 27

Page 16: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

Cross-sectional data:

Data for di¤erent entities: workers, households, �rms, cities,countries, and so forth.

No time dimension (even if date of data collection varies somewhatacross units, it is ignored).

Order of data does not matter!

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 16 / 27

Page 17: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

Time series data:

Data for a single entity (person, �rm, country) collected at multipletime periods. Repeated observations of the same variables (GDP,prices).

Order of data is important!

Observations are typically not independent over time;

In this case the notion of population corresponds to the DataGenerating Process (DGP).

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 17 / 27

Page 18: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

Panel data or longitudinal data:

Data for multiple entities (individuals, �rms, countries) in whichoutcomes and characteristics of each entity are observed at multiplepoints in time.

Combine cross-sectional and time series issues.

Present several advantages with respect to cross-sectional and timeseries data (depending on the question of interest!).

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 18 / 27

Page 19: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

De�nition (Econometric model)An econometric model speci�es the statistical relationship that is believedto hold between the various economic quantities pertaining to a particulareconomic phenomenon under study.

An econometric model can be derived from a deterministic economicmodel by allowing for uncertainty, or from an economic model whichitself is stochastic.

However, it is also possible to use econometric models that are nottied to any speci�c economic theory.

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 19 / 27

Page 20: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

We can distinguish:

1 Parametric model: the relationship (joint probability distribution)between the dependent variable /vector Y and the explicativevariables X is fully characterised by a set of parameters θ

Y = f (X ; θ) + ε

where link function f (.) is assumed to be known.

2 Non parametric and semi-parametric models: the link functioncan not be described using a �nite number of parameters. The linkfunction is assumed to be unknown and has to be estimated.

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 20 / 27

Page 21: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

The general approach of econometrics is the following:

1 Step 1: Model speci�cation

2 Step 2: Estimation of the parameters

3 Step 3: Validation1 Signi�cance tests;2 Speci�cation tests;3 Backtesting (forecasting performances);4 Etc.

4 Step 4: Use of the model (forecasting, feedback on the building ofthe model, etc.)

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 21 / 27

Page 22: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

Objectives of the course

The objective of this course are mainly related to the steps 2 (Estimation)and 3 (Validation)

More speci�cally:

1 To provide a global understanding of modern econometric methods;

2 To give a critical assessment of the presented methods;

3 To constitute an introduction and a basis for the more speci�ceconometric courses of the Masters.

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 22 / 27

Page 23: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

References

Amemiya T. (1985), Advanced Econometrics. Harvard University Press.

Greene W. (2007), Econometric Analysis, sixth edition, Pearson - PrenticeHil (recommended)

Johnson J., Econometric Methods, 3rd edition, MacGraw-Hill

Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (aspecial thank)

Ruud P., (2000) An introduction to Classical Econometric Theory, OxfordUniversity Press.

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 23 / 27

Page 24: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Introduction

Course outline

Chapter 1: Estimation theory

Chapter 2: Maximum Likelihood Estimation (MLE)

Chapter 3: The multiple linear regression model: the Ordinary LeastSquares (OLS) estimator

Chapter 4: Inference and statistical hypothesis testing

Chapter 5: The Generalized Least Squares (GLS) estimator

Chapter 6: Endogeneity, error-in-variables and the Instrumental Variables(IV) estimator

Chapter 7: The Generalized Method of Moments (GMM)

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 24 / 27

Page 25: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

End of the general introduction

Christophe Hurlin (University of Orléans)

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 25 / 27

Page 26: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Course rules

Your grade will be determined based on the following criteria:

Final Exam (F) (compulsory)

Mid-term (MT) (optional)

Retake exam (RE) (compulsory if necessary).

Two cases:

1 Without retake exam, the �nal grade is given by:

GRADE = 0.7 � F + 0.3 �max(MT ,F )

2 With a retake exam, the �nal grade is given by

GRADE = 1 � RE

In other words, if you need to redo the exam, then the grade will be simplybased on the make-up exam grade - the mid-term exam no longer counts.

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 26 / 27

Page 27: Advanced Econometrics - HEC Lausanne Christophe Hurlin · Pelgrin, F. (2010), Lecture notes Advanced Econometrics, HEC Lausanne (a special thank) Ruud P., (2000) An introduction to

Course rules

1 The �nal exam and the retake exams cover the entire course(including the exercises). The mid-term exam covers the partsindicated by the instructor.

2 All exams (mid-term, �nal, and retake exams) are closed book.

3 All type of calculator is authorized for all the exam.

4 The duration of the �nal exam and the retake exam is 180 minutes.The midterm is 120 minutes.

5 Careful and clear justi�cation of your answers will be rewarded. Thesolution approach has to be clear to the grader. In particular,(numerical) results without analytical derivations receive no grade.

Christophe Hurlin (University of Orléans) Advanced Econometrics - HEC Lausanne October 2013 27 / 27