You are on page 1of 44

A Modern Approach

E-mail: wanglimq@gmail.com Mobile phone 136 111!"6#


Economics A Morden Approach Prof. Wang Li 1

Welcome to Econometrics
Prof. Wang Li

The Features of this course


%he emphasis is on appl&ing econometrics. We pa& more attention to the Econometrics applications' (ow to ma)e regression models in sol*ing act+al economic pro,lems- anal&.ing phenomena +sing E*iews. We pa& less attention to the deri*ation or proof of an& form+la. (owe*er it is not necessaril& mean omitting the essential deri*ations and proofs.

Economics A Morden Approach Prof. Wang Li $

The Features of this course


%his is an intermediate le*el- Ph.D. co+rse in Applied Econometrics. Text Book: /effre& M.Wooldridge' 0ntrod+ctor& Econometrics A Modern Approach 1o+rth Edition A s&stematic approach ass+mptions are introd+ced onl& as the& are needed to o,tain a concl+sion. 0n-te2t q+estions with answers in Appendi2 1 %he end-of-chapter pro,lems and comp+ter e2ercises are oriented toward empirical wor).3 Economics A Morden Approach Prof. Wang Li

The Features of this course

Fundamental Review:

4 5asic Mathematical %ools Appendi2 A 4 1+ndamentals of Pro,a,ilit& Appendi2 5 4 Mathematical 6tatistics Appendi2 7 4 6+mmar& of Matri2 Alge,ra Appendi2 8 9 :o+ can cop& the electronic *ersion of them.
Probability Population Inferential Statistics Samples
3

Chapter One
7hapter <+tline--0ntrod+ctor& 7hapter

%he ;at+re of Econometrics and Economic 8ata

4 7hapter 1 disc+sses the scope of econometrics and raises general iss+es that res+lt from the application of econometric methods. 4 6ection 1.3 e2amines the )inds of data sets that are +sed in ,+siness- economics- and other social sciences. 4 6ection 1.3 pro*ides an int+iti*e disc+ssion of the diffic+lties associated with the inference of ca+salit& in the social.
Economics A Morden Approach Prof. Wang Li

Lecture Outline

What is econometrics What can econometrics do for +s 8ata t&pes' the inp+t %he q+estion of ca+salit& 6teps in empirical anal&sis What will ,e co*ered in this co+rse
Economics A Morden Approach Prof. Wang Li 6

What is econometrics
8efinition Econometrics is ,ased +pon the de*elopment of statistical methods for estimating economic relationships- testing economic theories- and e*al+ating and implementingpolic&.

Economics A Morden Approach Prof. Wang Li

What is econometrics
0t has ,een defined generall& as the q+antitati*e anal&sis of act+al economic phenomena ,ased on the conc+rrent de*elopment of theor& and o,ser*ationrelated ,& appropriate methods of inference. Econometrics is not >+st economics statistics- nor economic theor&- nor application of mathematics to economics. 0t is the +nification of the three.
Economics A Morden Approach Prof. Wang Li =

What is econometrics
the application of mathematical methods to represent economic theories and anal&.e pro,lems posed in economics

Economics

a topic in applied statistics that concerns the collectionprocessing- compilationdissemination- and anal&sis of economic data.

Mathematical Economics

Economic Statistics

Econometrics

Mathematics

Mathematical Statistics

%his term is closel& related to the term ?statistical theor&the st+d& of statistics from a mathematical standpoint " Economics A Morden Approach Prof. Wang Li

Statistics

Why study Econometrics?


@are in economics to ha*e e2perimental data ;eed to +se non-e2perimental- or o,ser*ational data to ma)e inferences 5eca+se of non-e2perimental nat+re of economic relationships some standard ass+mptions made in statistics are often *iolated and an econometrician has to find a wa& to fi2 statistical methods to fit the sit+ation.
Economics A Morden Approach Prof. Wang Li 1!

Econometrics s! Math! Statistics "#enerally$


9 Econometrics 9 Mathematical 6tatistics 4 8eals with 4 8eals with controlled pro,lematic E2perimental 'ata none2perimental data 4 E&perimental 'ata: 4 %one&perimental 8ata collected in a 'ata: <,ser*ational controlled 8ata- o,ser*ations of en*ironment agents in the real world 4 @esearcher is an 4 @esearcher is a acti*e collector in a passi*e collector of controlled- artificial data from the real en*ironment world Economics A Morden Approach 11
Prof. Wang Li

(o) is econometrics different from economics?


A ma>or distinction is the econometricians concern with dist+r,ance term. Economic specification of cons+mption f+nction' C = f( Y ) where C is cons+mption and Y is income Econometric specification' C=f(Y)+u, where u is a dist+r,ance AerrorB term. Witho+t the dist+r,ance term the relationship is e2act or deterministicC with the dist+r,ance term it is stochastic . Economics A Morden Approach 1$

Prof. Wang Li

What can econometrics do for us?

%est economic h&potheses and theories

E2plain ,eha*iors that are of economic nat+re

A formal economic model can ,e tested

Polic& e*al+ation

An empirical anal&sis +ses data to estimate a relationship %heor& ma& ,e am,ig+o+s as to the effect of some polic& change 4 can +se econometrics to e*al+ate the program
Economics A Morden Approach Prof. Wang Li 13

What can econometrics do for us?


<*erall- econometrics is +sed to'

Estimate parameters of economic relationships %est economic h&potheses and theories E*al+ate go*ernment and ,+siness polic& Predict and forecast o+t of sample

Economics A Morden Approach Prof. Wang Li

13

Types of 'ata

%here e2ist a *ariet& of t&pes of economic data' - 7ross-6ectional 8ata - %ime 6eries 8ata - Pooled 7ross 6ection 8ata - Panel 8ata Each data t&pe has ad*antages and disad*antages.

Economics A Morden Approach Prof. Wang Li

Cross-Section data
A cross-sectional data set consists of a sample of indi*id+als at a gi*en point in time. 4 (ere indi*id+als are considered in ,roader sense. Each o,ser*ation is a new indi*id+al- firm- cit&- co+ntr&- etc. with information at a point in time. 9 E2amples' (o+sehold spending this 7hristmas ;ational Dnemplo&ment this &ear

Economics A Morden Approach Prof. Wang Li 16

Cross-Section data
<,ser*ation Wage E2perience n+m,er AD68Ehr A&earsB B 1 3.1! $ $ 3 $6 3.$3 6.!! 3. !
Economics A Morden Approach Prof. Wang Li

Ed+cation A&earsB 11 1$ = 13
1#

$$ 33

Cross-Section data

7ross-sectional data str+ct+re often ma)es random sampling ass+mption reasona,le.


4 %he ordering of the data does not matter for econometric anal&sis is a )e& feat+re of this t&pe of data sets 4 Widel& +sed in economics- especiall& in fields li)e la,or economics- health economics- +r,an economics- etc.
Economics A Morden Approach Prof. Wang Li 1=

Cross-Section data
Fenerall& the entire pop+lation cannot ,e polled- so a 7ross-6ectional data set is ass+med to ,e a @andom 6ample. (owe*er- a sample is not random if' 1B 5ias occ+rs $B A sample selection pro,lem occ+rs Asome categories of respondents are more li)el& to respond than othersB 3B 6ample si.e is too small -an& st+d& with less than 30-40 o,ser*ations

Economics A Morden Approach Prof. Wang Li 1"

Cross-Section data

8ata ma& also incl+de a 8+mm& Garia,le to indicate if a gi*en o,ser*ation has a gi*en trait Amale- ed+cated- emplo&edetc.B 8+mm& *aria,les will ,e co*ered in chapter #

Economics A Morden Approach Prof. Wang Li

$!

Time Series data

A %ime 6eries data consists of o,ser*ations on a *aria,le or se*eral *aria,les o*er time. 4 %his data ha*e separate o,ser*ations for each time period e.g. stoc) prices 4 <rder is important' data are chronologicall& arranged- we can not change the order of o,ser*ations 4 8ata has specific freq+enc&' ann+alq+arterl&- monthl&- wee)l&- dail&
Economics A Morden Approach Prof. Wang Li $1

Time Series data


<,ser*ation n+m,er 1 $ 3 33 :ear F8P Amlrd.D68B 1"#= 1.3" 1!#" 1"=! $!1! 1.=6 1."= #.=3 Pop+lation Amln. peopleB $.36 $.6# $.=" 3.3
$$

Economics A Morden Approach Prof. Wang Li

Time Series data


8iffic+lties' 9 We can not ass+me that o,ser*ations are independent Ali)e in random samplesB. 4 trending' this terms o,ser*ation depends somewhat on last terms o,ser*ation. e.g. 0ncome- weight- spending- happiness. 4 freq+ent time series data can e2hi,it seasonal patterns Ae.g. ice cream sales fall in winterB. 9 We need special modifications to econometric theor& to anal&.e these data.

Economics A Morden Approach Prof. Wang Li $3

*ooled Cross-Section 'ata


Pooled 7ross-6ection 8ata 6ome data sets ha*e cross-section H time series feat+res. 9 6+ppose- we cond+ct a famil& income s+r*e& for $ &ears' 4 A random sample of !! families is ,eing s+r*e&ed in $!!". 4 Another random sample of !! families Adifferent !! familiesIB is ,eing s+r*e&ed in $!1!.

Economics A Morden Approach Prof. Wang Li $3

*ooled Cross-Section 'ata


%he same o,ser*ation sho+ld not ,e followed o*er different &ears 0f we pool these two independent samples together- we will get a sample of 1!!! random o,ser*ations JK pooled cross section sample. Anal&sis is similar to cross sectional datawith the additional consideration of str+ct+ral changes d+e to time 4 0t is a relati*el& new concept +sef+l for anal&.ing polic& effects.
Economics A Morden Approach Prof. Wang Li $

*anel 'ata

Panel data is a sets of time-series o,ser*ations for a gro+p of indi*id+als. 4 also called longit+dinal data. 4 preferred ordering is ,& gro+ping indi*id+als. i.e. first indi*id+al o*er time followed ,& second indi*id+al o*er time.

Economics A Morden Approach Prof. Wang Li

$6

*anel 'ata
6+ppose- we st+d& co+ntries A, B, , Z for &ears 2008-2010. 9 %he same cross sectional +nit is o,ser*ed o*er a gi*en time period. 9 We can change the order of cross-sectional +nits' B, A,..Z 9 We can not change the order of &ears for each co+ntr&' 2008, 2009, 2010. 6pecial techniq+es are +sed to anal&.e panel data.

Economics A Morden Approach Prof. Wang Li $#

<,ser*. n+m,er 1 $ 3 3 6 # = "

7o+ntr& A A A 5 5 5 7 7 7

:ear $!!= $!!" $!1! $!!= $!!" $!1! $!!= $!!" $!1!

F8P 3.3 3.#= 3.1$ 13.3 1 .6# 16.3$ 6.33 6.=3 #.$#

Pop+lat. $.33 $.3 $.6# =.#" ".3$ 1!.1$ .3$ .#3 ."$
$=

Economics A Morden Approach Prof. Wang Li

*anel 'ata

Ad*antages' 4 a,le to wor) with a larger data set 4 a,le to control for +no,ser*ed characteristics 4 a,le to st+d& the effect of lags 8isad*antages' 4 statistical pro,lems of cross-sectional data 4 statistical pro,lems of time series data 4 more diffic+lt to wor) with ;otes' Panel data is not co*ered in this co+rse- ,+t can ,e +sed in the pro>ect report if e2tra reading is done
Economics A Morden Approach Prof. Wang Li $"

The +uestion of Causality

6impl& esta,lishing a relationship ,etween *aria,les is rarel& s+fficient. <ne goal of econometric anal&sis is to e2amine the ca+salit& of two *aria,les. 4 A simple plotting of two *aria,les or calc+lation of correlation will onl& see if the two *aria,les mo*e together 4 0t cant show ca+sation to ,e ca+sal.

Economics A Morden Approach Prof. Wang Li

3!

Ceteris *aribus

7eteris Pari,+s means Lother Arele*antB factors ,eing eq+alM. 7eteris pari,+s' all other rele*ant factors remain constant. change in : d+e to the change in N N has ca+sal effect on :' ONO: 0n most cases- h&potheses in the social sciences are ceteris pari,+s in nat+re. An& econometric model m+st ha*e ,ehind reason of ca+sation
Economics A Morden Approach Prof. Wang Li 31

E&ample: ,eturns to Education

A model of h+man capital in*estment implies getting more ed+cation sho+ld lead to higher earnings 0n the simplest case- this implies an eq+ation li)e Earnings = 0 + 1education + u 7an it ,e considered ca+salP 6ome things are still +no,ser*ed- Ae.g. innate a,ilit&- famil& ,ac)gro+nd- etc.B.
Economics A Morden Approach Prof. Wang Li 3$

E&ample "continued$

0f we s+ppose that all other factors are fi2ed. %he estimate of coefficient 1- is the ret+rn to ed+cation. 1 ' if ed+cation increase ,& one &ear- how m+ch more earning will one gain. 0n other words- the marginal effect of x on y is constant and eq+al to 1 . While the error term- u- incl+des other factors affecting earnings- which can ,e pro,lematic. 0n the real world- rarel& can all *aria,les ,e fi2ed. Economics A Morden Approach 33
Prof. Wang Li

Steps in Empirical Econometric -nalysis

An empirical anal&sis +ses data to test a theor& or to estimate a relationship (ow do we go in str+ct+ring an empirical AeconometricB anal&sisP 6tep 1' f+nctional form 6tep $' 8ata collection 6tep 3' Model estimation- inferenceprediction
Economics A Morden Approach Prof. Wang Li 33

Steps in Empirical Econometric -nalysis


1ormal economic modeling is sometimes the starting point for empirical anal&sis. 5asic economic +nderstanding is s+fficient. Qe& point' 1orm+late the q+estion of interest caref+ll& 4 1ormal economic model 4 0nformal economic reasoning.

Economics A Morden Approach Prof. Wang Li

Steps in Empirical Econometric -nalysis


7ollect needed data for the q+estion of interest. 6+r*e&- R+estionnaire- 6tatistical :ear,oo) Dsing )nowledge a,o+t model specification to transform an economic model into an econometric model that is read& for estimation and prediction. 4 1+nctional form 4 Garia,les that cannot o,ser*ed 4 error term
Economics A Morden Approach Prof. Wang Li 36

Steps in Empirical Econometric -nalysis


Model estimation- inference- prediction 6pecif& h&pothesis of interest in terms of the +n)nown parameters Dse econometric methods to estimate the parameters and formall& test the h&pothesi.es of interest . f+nctional forms can ,e modified and *aria,les added or deleted as statistical tests are done
Economics A Morden Approach Prof. Wang Li 3#

Steps in Empirical Econometric -nalysis


0nterpreting &o+r res+lts Anal&.e the res+lts &o+ o,tained Pro*ide eno+gh information for a reader to decide whether the& find &o+r res+lts con*incing or not

Economics A Morden Approach Prof. Wang Li

3=

What Will be Co ered in This Course


Part 1 @egression Anal&sis with 7ross-6ectional 8ata 7hapter $ chapter " Part $ @egression Anal&sis with %ime 6eries 8ata 7hapter 1! chapter 1$ Part 3 7arr&ing o+t an Empirical Pro>ect 7hapter 1"

Economics A Morden Approach Prof. Wang Li 3"

What .ou are in this

re/uired to do course

%o practice the end-of-chapter comp+ter e2ercises after class As m+ch as possi,le- the more the ,etter. After Part 1 finished- perform a gro+p presentation Aanal&sis of cross-sectional data B After Part $ finished- perform another gro+p presentation Aanal&sis of time seriesB At the end of this semester or no later than the ,eginning of ne2t semester hand in &o+r paper report and Economics E*iews worfile &o+r own 3! A Morden Approachof Prof. Wang Li Empirical Pro>ect

Tentati e Schedule
6ingle-*aria,le regression Wooldridge- chapters $ M+lti*ariate regression Wooldridge- chapters 3- 3- - 6- and # 8ata Pro,lems Wooldridge- chapters = 1irst time gro+p presentations %ime series regression Wooldridge- chapters 1!- 11- and 1$ 6econd time gro+p presentations

Economics A Morden Approach Prof. Wang Li

31

Final *aper -ssi#nment


The basic purpose of this course is to prepare students to carry out their o)n econometric study!

Econometrics is ,est learned ,& act+all& doing an econometric st+d&. <nl& then will the ?+ninitiated? learn the power as well as the pitfalls of econometrics. 6t+dents will ,e as)ed to form+late an original econometric model- collect data rele*ant to the model- +se econometric techniq+es to estimate the model- and interpret the res+lts of the estimation.
Economics A Morden Approach Prof. Wang Li 3$

%o Final E&amination
Carryin# Out an Empirical *ro0ect

7hoosing a %opic

7ollecting data ,& &o+r self Estimating a Model 0nterpreting &o+r res+lts

(and in &o+r Pro>ect paper report and E*iews worfile

Economics A Morden Approach Prof. Wang Li

33

1radin#
The re ised final e aluation:

7lass participation R+i. 1 Fro+p presentation 1inal e2am in class room part AR+i. $B 1inal e2am Ata)e home partB

S 1!S $ S 3!S 3!S

Economics A Morden Approach Prof. Wang Li

33

You might also like