《计量经济学》课程教学资源（PPT课件讲稿，英文版）ch13 Panel data methods

Panel data methods B+Bx,1+…kti Vit i Economics 20- Prof anderson

Economics 20 - Prof. Anderson 1 Panel Data Methods yit = b0 + b1 xit1 + . . . bk xitk + uit

A True panel vs a Pooled Cross section o Often loosely use the term panel data to refer to any data set that has both a cross sectional dimension and a time-series dimension o More precisely it's only data following the same cross-section units over time o Otherwise it's a pooled cross-section Economics 20- Prof anderson

Economics 20 - Prof. Anderson 2 A True Panel vs. A Pooled Cross Section Often loosely use the term panel data to refer to any data set that has both a cross￾sectional dimension and a time-series dimension More precisely it’s only data following the same cross-section units over time Otherwise it’s a pooled cross-section

Pooled cross sections o We may want to pool cross sections just to get bigger sample sizes o We may want to pool cross sections to investigate the effect of time e We may want to pool cross sections to investigate whether relationships have changed over time Economics 20- Prof anderson

Economics 20 - Prof. Anderson 3 Pooled Cross Sections We may want to pool cross sections just to get bigger sample sizes We may want to pool cross sections to investigate the effect of time We may want to pool cross sections to investigate whether relationships have changed over time

Difference-in-Differences e Say random assignment to treatment and control groups, like in a medical experiment o One can then simply compare the change in outcomes across the treatment and control groups to estimate the treatment effect o For time 1, 2, groups A, B(y2B-y2A) (yLB-yLA, or equivalently (y2B-yLB 72A-yLA, is the difference-in-differences Economics 20- Prof anderson 4

Economics 20 - Prof. Anderson 4 Difference-in-Differences Say random assignment to treatment and control groups, like in a medical experiment One can then simply compare the change in outcomes across the treatment and control groups to estimate the treatment effect For time 1,2, groups A, B (y2,B – y2,A) - (y1,B – y1,A), or equivalently (y2,B – y1,B) - (y2,A – y1,A), is the difference-in-differences

Difference-in-Differences(cont) o Aregression framework using time and treatment dummy variables can calculate this difference-in-difference as well e Consider the model: yit=Bo t B,afterit+B treatment, o+ B,treatment after t o The estimated B3 will be the difference-in- differences in the group means Economics 20- Prof anderson 5

Economics 20 - Prof. Anderson 5 Difference-in-Differences (cont) A regression framework using time and treatment dummy variables can calculate this difference-in-difference as well Consider the model: yit = b0 + b1 treatmentit + b2afterit + b3 treatmentit*afterit+ uit The estimated b3 will be the difference-in￾differences in the group means

Difference-in-Differences(cont) o When don't truly have random assignment the regression form becomes very useful Additional x's can be added to the i regression to control for differences across the treatment and control groups Sometimes referred to as a natural experiment especially when a policy change is being analyzed Economics 20- Prof anderson 6

Economics 20 - Prof. Anderson 6 Difference-in-Differences (cont) When don’t truly have random assignment, the regression form becomes very useful Additional x’s can be added to the regression to control for differences across the treatment and control groups Sometimes referred to as a “natural experiment” especially when a policy change is being analyzed

Two-Period Panel data It's possible to use a panel just like pooled cross-sections. but can do more than that Panel data can be used to address some kinds of omitted variable bias If can think of the omitted variables as being fixed over time, then can model as having a composite error Economics 20- Prof anderson 7

Economics 20 - Prof. Anderson 7 Two-Period Panel Data It’s possible to use a panel just like pooled cross-sections, but can do more than that Panel data can be used to address some kinds of omitted variable bias If can think of the omitted variables as being fixed over time, then can model as having a composite error

Unobserved Fixed Effects o Suppose the population model is yit=Bo+ Cd2+Bxm+…+Bxik+a1+ Here we have added a time-constant component to the error. v,=a, +u o If a, is correlated with the x's, OLS will be biased, since we a is part of the error term o With panel data, we can difference-out the unobserved fixed effect Economics 20- Prof anderson 8

Economics 20 - Prof. Anderson 8 Unobserved Fixed Effects Suppose the population model is yit = b0 + d0d2t + b1 xit1 +…+ bk xitk + ai + uit Here we have added a time-constant component to the error, uit = ai + uit If ai is correlated with the x’s, OLS will be biased, since we ai is part of the error term With panel data, we can difference-out the unobserved fixed effect

First-differences o We can subtract one period from the other to obtain Ayi=+B△xn+…+BAxk+ △ This model has no correlation between the x's and the error term. so no bias Need to be careful about organization of the data to be sure compute correct change Economics 20- Prof anderson 9

Economics 20 - Prof. Anderson 9 First-differences We can subtract one period from the other, to obtain Dyi = d0 + b1Dxi1 +…+ bkDxik + Dui This model has no correlation between the x’s and the error term, so no bias Need to be careful about organization of the data to be sure compute correct change

Differencing w/ Multiple Periods o Can extend this method to more periods o Simply difference adjacent periods e So if 3 periods, then subtract period 1 from period 2, period 2 from period 3 and have 2 observations per individual ◆ Simply estimate by Ols, assuming the△ln are uncorrelated over time Economics 20- Prof anderson 10

Economics 20 - Prof. Anderson 10 Differencing w/ Multiple Periods Can extend this method to more periods Simply difference adjacent periods So if 3 periods, then subtract period 1 from period 2, period 2 from period 3 and have 2 observations per individual Simply estimate by OLS, assuming the Duit are uncorrelated over time