What is a regression model?

Regression versus correlation

Simple regression

Some further terminology

Simple linear regression in EViews -- estimation of an optimal hedge ratio

The assumptions underlying the classical linear regression model Properties of the OLS estimator

Precision and standard errors

An introduction to statistical inference

27

27

28

28

37

2.6

2.7

2.8

2.9

v

40

43

44

46

51

vi

Contents

2.10 A special type of hypothesis test: the t-ratio

2.11 An example of the use of a simple t-test to test a theory in ﬁnance: can US mutual funds beat the market?

2.12 Can UK unit trust managers beat the market?

2.13 The overreaction hypothesis and the UK stock market

2.14 The exact signiﬁcance level

2.15 Hypothesis testing in EViews -- example 1: hedging revisited 2.16 Estimation and hypothesis testing in EViews -- example 2: the CAPM

Appendix: Mathematical derivations of CLRM results

65

67

69

71

74

75

77

81

3 Further development and analysis of the classical linear

regression model

3.1 Generalising the simple model to multiple linear regression 3.2 The constant term

3.3 How are the parameters (the elements of the β vector) calculated in the generalised case?

3.4 Testing multiple hypotheses: the F -test

3.5 Sample EViews output for multiple hypothesis tests

3.6 Multiple regression in EViews using an APT-style model

3.7 Data mining and the true size of the test

3.8 Goodness of ﬁt statistics

3.9 Hedonic pricing models

3.10 Tests of non-nested hypotheses

Appendix 3.1: Mathematical derivations of CLRM results

Appendix 3.2: A brief introduction to factor models and principal components analysis

120

4 Classical linear regression model assumptions and

diagnostic tests

4.1 Introduction

4.2 Statistical distributions for diagnostic tests

4.3 Assumption 1: E (u t ) = 0

4.4 Assumption 2: var(u t ) = σ 2 < ∞

4.5 Assumption 3: cov(u i , u j ) = 0 for i = j

4.6 Assumption 4: the xt are non-stochastic

4.7 Assumption 5: the disturbances are normally distributed

4.8 Multicollinearity

4.9 Adopting the wrong functional form

4.10 Omission of an important variable

4.11 Inclusion of an irrelevant variable

129

129

130

131

132

139

160

161

170

174

178

179

88

88

89

91

93

99

99

105

106

112

115

117

Contents

vii

4.12 Parameter stability tests

4.13 A strategy for constructing econometric models and a discussion of model-building philosophies

4.14 Determinants of sovereign credit ratings

191

194

5

5.1

5.2

5.3

5.4

5.5

5.6

5.7

5.8

5.9

5.10

5.11

5.12

5.13

Univariate time series modelling and forecasting

Introduction

Some notation and concepts

Moving average processes

Autoregressive processes

The partial autocorrelation function

ARMA processes

Building ARMA models: the Box--Jenkins approach

Constructing ARMA models in EViews

Examples of time series modelling in ﬁnance

Exponential smoothing

Forecasting in econometrics

Forecasting using ARMA models in EViews

Estimating exponential smoothing models using EViews

206

206

207

211

215

222

223

230

234

239

241

243

256

258

Multivariate models

Motivations

Simultaneous equations bias

So how can simultaneous equations models be validly estimated? Can the original coefﬁcients be retrieved from the π s ?

Simultaneous equations in ﬁnance

A deﬁnition of exogeneity

Triangular systems

Estimation procedures for simultaneous equations systems

An application of a simultaneous equations approach to

modelling bid--ask spreads and trading activity

Simultaneous equations modelling using EViews

Vector autoregressive models

Does the VAR include contemporaneous terms?

Block signiﬁcance and causality tests

VARs with exogenous variables

Impulse responses and variance decompositions

VAR model example: the interaction between property returns and the macroeconomy

VAR estimation in EViews...