Differences-in-Differences
“Difference-in-Differences Estimation” Imbens/Wooldridge
Differences-in-Differences (using R)
“Difference-in-Differences Estimation” Imbens/Wooldridge
Fuzzy Differences-in-Differences
6 Oct 2015 In many applications of the differences-in-differences (DID) method ... †University of Warwick
10th Warwick Economics PhD Conference Extended Programme
9 Jun 2022 differences can be explained in part by the varying constraints faced by policymakers in a given country. Discussant: Qianyang Zhang.
Lecture 3: Differences-in-Differences
This will lead to biased OLS estimates. Waldinger (Warwick). 5 / 55. Page 6. 2 Ways of Estimating
THE CREATION OF DIFFERENCE
18 Oct 2017 CHAPTER 3. THE CREATION OF DIFFERENCE. For the Rede Lecture at the University of Cambridge in 1875 after his.
Lecture Notes 6: Dynamic Equations Part A: First-Order Difference
23 Sept 2017 We may write “difference equation” even when considering a recurrence relation. University of Warwick EC9A0 Maths for Economists. Peter J.
Academic Research Careers in Scotland
There were marked differences in the gender composition of contract research staff by subject Chapter 3 describes the characteristics of.
7. Piaget Kohlberg
and Others on Moral Development
Contested Knowledge: A Critical Review of the Concept of
This according to Virtanen
![Lecture Notes 6: Dynamic Equations Part A: First-Order Difference Lecture Notes 6: Dynamic Equations Part A: First-Order Difference](https://pdfprof.com/Listes/16/26902-16diffEqSlidesA.pdf.pdf.jpg)
Lecture Notes 6: Dynamic Equations
Part A: First-Order Dierence Equations
in One VariablePeter J. Hammond
latest revision 2017 September 23rd, plus minor correction typeset fromdiffEqSlides18A.tex University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 1 of 54Lecture Outline
Introduction: Dierence vs. Dierential Equations
First-Order Dierence Equations
First-Order Linear Dierence Equations: IntroductionGeneral First-Order Linear Equation
Particular, General, and Complementary Solutions
Explicit Solution as a Sum
Constant and Undetermined Coecients
Stationary States and Stability for Linear First-Order Equations Local Stability of Nonlinear First-Order Equations University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 2 of 54Walking as a Simple Dierence Equation
What is the dierence
between dierence and dierential equations?It is relatively common to indicate by:
a subscript a d iscretetime function lik em7!xm; parentheses a continuou stime function l iket7!x(t).Walking on two feet can be modelled as a
discrete time p rocess with time domain T=f0;1;2;:::g=Z+=f0g [N that counts the number of completed steps.Aftermsteps, the respective positions`;r2R2
of the left and right feet on the ground can be described by the two functionsT3m7!(`m;rm). University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 3 of 54Walking as a More Complicated Dierence Equation
Athletics rules limit a walking step to be no longer than a stride. So a walking process that starts with the left foot might be described by the two coupled equations m=( (rm1) ifmis odd m1ifmis evenandrm=( (`m1) ifmis even r m1ifmis odd form= 0;1;2;:::. Or, if the length and direction of each pace are aected by the length and direction of its predecessor, by m=( (rm1;`m2) ifmis odd m1ifmis even andrm=( (`m1;rm2) ifmis even r m1ifmis odd form= 0;1;2;:::. University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 4 of 54Walking as a Dierential Equation
Newtonian physics implies that a walker's centre of mass must be a continuous function of time , described by a 3-vector valued mappingR+3t7!(x(t);y(t);z(t))2R3.The time domain is thereforeT:=R+.
The same will be true for the position of, for instance, the extreme end of the walker's left big toe.Newtonian physics requires that the
acceleration3-vecto r
described by the second derivative d2dt2(x(t);y(t);z(t))2R3 should be well dened for allt. The biology of survival requires it to be bounded. Actually, the motion becomes seriously uncomfortable unless the acceleration (or deceleration) is continuous | as my driving instructor taught me more than 50 years ago! University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 5 of 54Lecture Outline
Introduction: Dierence vs. Dierential Equations
First-Order Dierence Equations
First-Order Linear Dierence Equations: IntroductionGeneral First-Order Linear Equation
Particular, General, and Complementary Solutions
Explicit Solution as a Sum
Constant and Undetermined Coecients
Stationary States and Stability for Linear First-Order Equations Local Stability of Nonlinear First-Order Equations University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 6 of 54Basic Denition
LetT=Z+3t7!xt2Xdescribe a discrete time process,
withX=R(orX=Rm) as thestate space . Its dierence at time tis dened as xt:=xt+1xtA standard
rst-o rderdierence equation tak esthe fo rm x t+1xt= xt=dt(xt) where eachdt:X!X, or equivalently,TX3(t;x)7!dt(x)
University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 7 of 54Equivalent Recurrence Relations
Obviously, the dierence equationxt+1xt= xt=dt(xt)
is equivalent to the recurrence relation xt+1=rt(xt) whereTX3(t;x)7!rt(x) =x+dt(x), or equivalently,dt(x) =rt(x)x.Thus dierence equations and recurrence relations
are entirely equivalentWe follow standard mathematical practice
in using the notation for recurrence relations, even when discussing dierence equations.We may write \dierence equation"
even when considering a recurrence relation. University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 8 of 54Existence of Solutions
Example
Consider the dierence equationxt=px
t11 withx0= 5.Evidentlyx1=p51 = 2, thenx2=p21 = 1,
and nextx3=p11 = 0, leavingx4=p01 undened as a real number.The domain of (t;x)7!px1 is limited to
D:=Z+[1;1).Generally, consider a mappingD3(t;x)7!rt(x) whose domain is restricted to a subsetDZ+X. For the dierence equationxt+1=rt(xt) to have a solution one must ensure that (t;x)2D=)(t+ 1;rt(xt))2Dfor allt2Z+ University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 9 of 54Lecture Outline
Introduction: Dierence vs. Dierential Equations
First-Order Dierence Equations
First-Order Linear Dierence Equations: IntroductionGeneral First-Order Linear Equation
Particular, General, and Complementary Solutions
Explicit Solution as a Sum
Constant and Undetermined Coecients
Stationary States and Stability for Linear First-Order Equations Local Stability of Nonlinear First-Order Equations University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 10 of 54Application: Wealth Accumulation in Discrete Time
Consider a consumer who, in discrete timet= 0;1;2;:::: I starts each periodt with an amountwtof accumulated wealth; I receives incomeyt; I spends an amountet; I earns interest on the residual wealthwt+ytetat the ratert. The process of wealth accumulation is then described by any of the equivalent equations w t+1= (1 +rt)(wt+ytet) =t(wtxt) =t(wt+st) where, at each timet, I t:= 1 +rtis theinterest facto r; I xt=etytdenotesnet exp enditure; I st=ytet=xtdenotesnet saving . University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 11 of 54Compound Interest
Dene the
comp oundinterest facto r R t:=Y t1 k=0(1 +rk) =Y t1 k=0k with the convention that the product of zero terms equals 1 | just as the sum of zero terms equals 0. This compound interest factor is the unique solution to the recurrence relationRt+1= (1 +rt)Rt that satises the initial conditionR0= 1.In the special case whenrt=r(allt),
it reduces toRt= (1 +r)t=t. University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 12 of 54Present Discounted Value (PDV)
We transform the dierence equationwt+1=t(wtxt)
by using the compound interest factorRt=Qt1 k=0k in order to discount both future wealth and expenditure.To do so, dene new variables!t;t
for the present discounted values (PDVs) of, respectively: 1. w ealthwtat timetas!t:= (1=Rt)wt; 2. net exp enditurextat timetast:= (1=Rt)xt.With these new variables,
the wealth equationwt+1=t(wtxt) becomes R t+1!t+1=tRt(!tt)ButRt+1=tRt, so eliminating this common factor
reduces the equation to!t+1=!tt, with the evident solution!t=!0Pt1 k=0kfork= 1;2;:::. University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 13 of 54Lecture Outline
Introduction: Dierence vs. Dierential Equations
First-Order Dierence Equations
First-Order Linear Dierence Equations: IntroductionGeneral First-Order Linear Equation
Particular, General, and Complementary Solutions
Explicit Solution as a Sum
Constant and Undetermined Coecients
Stationary States and Stability for Linear First-Order Equations Local Stability of Nonlinear First-Order Equations University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 14 of 54General First-Order Linear Equation
The general rst-order linear dierence equation
can be written in the form x tatxt1=ftfort= 1;2;:::;T for non-zero constantsat2Rand a forcing termN3t7!ft2R. When this equation holds fort= 1;2;:::;T, whereT6, this equation can be written in the following matrix form: 0 BBBBBBB@a11 0:::0 0 0
0a21:::0 0 0
0 0a3:::0 0 0
0 0 0:::aT11 0
0 0 0:::0aT11
CCCCCCCA0
BBBBBBBBB@x
0 x 1 x 2... x T2 x T1 x T1 CCCCCCCCCA=0
BBBBBBB@f
1 f 2 f 3... f T1 f T1 CCCCCCCA
University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 15 of 54Matrix Form
The matrix form of the dierence equation isCx=f, where:1.Cis theT(T+ 1)co ecientmatrix whose elements a re
c st=8 :asift=s1 ift=s+ 1
0 otherwise
fors= 1;2;:::;Tandt= 0;1;2;:::;T;2.xis theT+ 1-dimensional column vector (xt)Tt=0of endogenous unknowns, to be determined;
3.fis theT-dimensional column vector (ft)Tt=1of exogenous shocks.
University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 16 of 54Partitioned Matrix Form
The matrix equationCx=fcan be written in partitioned form asU eTxT1
x T =f where:1.Uis an upper triangularTTmatrix;
2.eT= (0;0;0;:::;0;1)>is theTth column vector
of the canonical basis of the vector spaceRT;3.xT1denotes the column vector which is the transpose
of the rowT-vector (x0;x1;x2;:::;xT2;xT1).In fact the matrixUsatises
(U;eT) = (diag(a1;a2;:::;aT);eT) + (0T1;ITT) Hence there areTindependent equations inT+ 1 unknowns, leaving one degree of freedom in the solution. University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 17 of 54An Initial Condition
Consider the dierence equationxtatxt1=ft,
orCx=fin matrix form. An initial condition sp eciesan exogenous value x0 for the valuex0at time 0.This removes the only degree of freedom
in the system ofTequations inT+ 1 unknowns.Consider the special case whenat= 1 for allt2N.
The obvious unique solution ofxtxt1=ft
is then that eachxtis thefo rwardsum x t= x0+X t s=1fs of the initial state x0, and of thetexogenously specied succeeding dierencesfs(s= 1;2;:::;t). University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 18 of 54A Terminal Condition
Alternatively, a
terminal condition for the dierence equationxtxt1=ft species an exogenous value xTfor the valuexT at the terminal time T.It leads to a unique solution as a
backw ardsum x t= xTX Tt1 s=0fTs of the exogenously specied I terminal state xT; I preceding backward dierencesfTs (s= 0;1;:::;Tt1). University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 19 of 54Lecture Outline
Introduction: Dierence vs. Dierential Equations
First-Order Dierence Equations
First-Order Linear Dierence Equations: IntroductionGeneral First-Order Linear Equation
Particular, General, and Complementary Solutions
Explicit Solution as a Sum
Constant and Undetermined Coecients
Stationary States and Stability for Linear First-Order Equations Local Stability of Nonlinear First-Order Equations University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 20 of 54Particular and General Solutions
We are interested in solving the systemCx=f
ofTequations inT+ 1 unknowns, whereCis aT(T+ 1) matrix. When the rank ofCisT, there is one degree of freedom.The associated
homoge neousequation Cx=0 will have a one-dimensional space of solutionsxHt=xHt(2R).Given any
pa rticularsolution xPtsatisfyingCxP=f for the particular time seriesfof forcing terms, the general solution xGtmust also satisfyCxG=f.Simple subtraction leads toC(xGxP) =0, soxGxP=xH
for some solutionxHof the homogeneous equationCx=0.Soxsolves the equationCx=f
i there exists a scalar2Rsuch thatx=xP+xH, which leads to the formulaxG=xP+xHfor the general solution. University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 21 of 54Complementary Solutions
Consider again the general rst-order linear equation which takes the inhomogeneous fo rmxtatxt1=ft.The associated
hom ogeneousequation tak esthe fo rm x tatxt1= 0 (for allt2N) with a zero right-hand side.The associated
c omplementarysolutions make up the one-dimensional linear subspaceL of solutions to this homogeneous equation. The spaceLconsists of functionsZ+3t7!xt2Rsatisfying x t=x0Y t s=1as(for allt2N) wherex0is an arbitrary scaling constant. University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 22 of 54From Particular to General Solutions
Consider again the
inhomogeneous equation x tatxt1=ft for a general RHSft.The associated
hom ogeneousequation tak esthe fo rm x tatxt1= 0LetxPtdenote apa rticularsolution ,
andxGtany alternativegeneral solution , of the inhomogeneous equation. Our assumptions imply that, for eacht= 1;2;:::, one has xPtatxPt1=ft
xGtatxGt1=ft Subtracting the second equation from the rst implies that xGtxPtat(xGt1xPt1) = 0
This shows thatxHt:=xGtxPtsolves the homogeneous equation. University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 23 of 54Characterizing the General Solution
Theorem
Consider the inhomogeneous equationxtatxt1=ft
with fo rcingterm ft.Its general solutionxGtis the sumxPt+xHtof
I any particular solutionxPtof the inhomogeneous equation; Ithe general complementary solutionxHtof the corresponding homogeneous equationxtatxt1= 0.University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 24 of 54
Linearity in the Forcing Term, I
Theorem
Suppose thatxPtandyPtare particular solutions
of the two respective dierence equations x tatxt1=dtandytatyt1=etThen, for any scalarsand,
the equationztatzt1=dt+ethas as a particular solution the corresponding linear combinationzPt:=xPt+yPt.Proof.Routine algebra.
University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 25 of 54Linearity in the Forcing Term, II
Consider any equation of the formxtatxt1=ft
whereftis a linear combinationPn k=1kfktofnforcing termshfktink=1.The theorem implies that a particular solution
is the corresponding linear combinationPn k=1kxPktof particular solutionshxPktink=1to the respectivenequationsxtatxt1=fkt(k= 1;2;:::;n). University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 26 of 54Lecture Outline
Introduction: Dierence vs. Dierential Equations
First-Order Dierence Equations
First-Order Linear Dierence Equations: IntroductionGeneral First-Order Linear Equation
Particular, General, and Complementary Solutions
Explicit Solution as a Sum
Constant and Undetermined Coecients
Stationary States and Stability for Linear First-Order Equations Local Stability of Nonlinear First-Order Equations University of Warwick, EC9A0 Maths for Economists Peter J. Hammond 27 of 54Solving the General Linear Equation
Consider a rst-order linear dierence equation
x t+1=atxt+ft for a p rocessT3t7!xt2R, where eachat6= 0quotesdbs_dbs29.pdfusesText_35[PDF] Exemples de différenciations pédagogiques en classe
[PDF] L 'IMPARFAIT ET LE PASSÉ COMPOSÉ
[PDF] Différencier et ? comprendre Différencier et ? comprendre
[PDF] La composition des différents laits - Nutrition Enfants Aquitaine
[PDF] 1 QUELQUES RÉFLEXIONS SUR L 'AMOUR L 'amour, mais de quoi
[PDF] les differents types de capteurs - Technologue pro
[PDF] 28 Les coffrages de chantier - Infociments
[PDF] Télécharger le fichier PDF Les coffrages de chantier - Infociments
[PDF] 1 Les différents types de recherche - Chamilo
[PDF] TYPES OF SECURITIES 1 Shares A share is an equity security Its
[PDF] Les Niveaux d
[PDF] Types de forces - Pages - Construction
[PDF] Cycle terminal - mediaeduscoleducationfr - Ministère de l
[PDF] PATHOLOGIES OCULAIRES COURANTES