MSc Economic Policy Methods Seminar · MSc Economic Policy Methods Seminar Introduction Aims of...
Transcript of MSc Economic Policy Methods Seminar · MSc Economic Policy Methods Seminar Introduction Aims of...
MSc Economic Policy Methods Seminar
MSc Economic Policy Methods SeminarLecture 4: Assessing three papers
Dr John McDonagh
November 21, 2015
MSc Economic Policy Methods Seminar
Table of contents
Introduction
Paper I
Paper II
Paper III
Student presentations
MSc Economic Policy Methods Seminar
Introduction
Aims of today - general
I Look at three papers from each of your modules.
I Take some lessons on both methodology andpresentation for your own work.
I Remember all three published so would have gonethrough a robust referee process.
I Papers for presentations.
MSc Economic Policy Methods Seminar
Introduction
Three papers todayI Comp and regulation Rosenthal, Leslie. The cost of
regulation in education: do school inspections improveschool quality?. No. 2000/09. Department of Economics,Keele University, 2001.
I Money and banking Barrell, Ray, Tatiana Fic, and JohnFitz Gerald. ”The banking sector and recovery in the EUeconomy.” National Institute Economic Review 216, no. 1(2011): R41-R52.
I Evaluation Banerjee, Abhijit V., Rukmini Banerji, EstherDuflo, Rachel Glennerster, and Stuti Khemani. ”Pitfallsof Participatory Programs: Evidence from a randomizedevaluation in education in India.” American EconomicJournal: Economic Policy(2010): 1-30.
MSc Economic Policy Methods Seminar
Paper I
Rosenthal (2000). The cost of regulation in
education: do school inspections improve school
quality?
MSc Economic Policy Methods Seminar
Paper I
Empirical results
MSc Economic Policy Methods Seminar
Paper I
Establishing exogeneity of visits
MSc Economic Policy Methods Seminar
Paper I
Features on Paper 1
I Simple hypothesis.
I Exogeneity is assessed robustly in first stage.
I Building in dynamic effects.
I Fixed v random effects in panel data.
I Handling limited dependent variables: % achieving A-Cand probability of receiving an inspection are notcontinuous variables.
I Economic v statistical significance. eg althoughstatistically significant how many students did it affect?
I Having a story if results don’t turn out as expected!
MSc Economic Policy Methods Seminar
Paper I
Thoughts on Paper 1
I Surprised fixed v random results effects were similar.
I Omitted variables captured in FE to the extent they areconstant; did thy change? Socio-economic variables;staff student ratio; intake of non-native speakers?
I Formal tests of exogeneity.
I Exams as a measures of success? Ehren and Visscher useinterviews after visit. Compared exams with other pupilswho performed equally well at age 11.
I Useful to see equation; indicate significance(***,**, *);graphs (eg trends); test statistics.
I Exam data has been used in other areas (hedonic pricing).
MSc Economic Policy Methods Seminar
Paper II
Barrell et al, (2011) ”The banking sector and
recovery in the EU economy.”
MSc Economic Policy Methods Seminar
Paper II
First part results
MSc Economic Policy Methods Seminar
Paper II
Sample of second part results
MSc Economic Policy Methods Seminar
Paper II
Features on paper 2
I Large micro datasets.
I Using regressions to build in policy responses, e.g.standard deviation shocks to x and impact on y.
I Squared terms to test for non-linearity.
I Dummy for outlier banks.
I Cyclical and structural are useful ways to think aboutdrivers of macroeconomic variables.
MSc Economic Policy Methods Seminar
Paper II
Thoughts on paper 2
I Time series. Coefficient on the lagged dependent variableis significant. If GDP is also non-stationary what are theproblems?
I Is it reasonable to treat size as exogenous?
I Not using logs can complicate interpretations.
I Use of annexes for tables. Graphs useful.
I Trying to do too much in one paper.
I Other ways to test for size eg, recursive estimates,quantile regressions.
MSc Economic Policy Methods Seminar
Paper III
Paper 3: Banerjee et al (2010). ”Pitfalls of
Participatory Programs: Evidence from a
randomized evaluation in education in India.”
MSc Economic Policy Methods Seminar
Paper III
Paper 3: Methodology
MSc Economic Policy Methods Seminar
Paper III
Paper 3: selected results. Some impact on info
available
MSc Economic Policy Methods Seminar
Paper III
Paper 3: selected results. Little impact on action
MSc Economic Policy Methods Seminar
Paper III
Features on paper 3
I Growing use of experimental economics (especially indevelopment).
I Two advantages: Control and Replicability.
I Power of dummy variables.
I Instrumenting strategies.
I Strategies to avoid ”Cherry picking” results.
MSc Economic Policy Methods Seminar
Paper III
Comments on paper 3
I External validity.
I Subjects may not be representative of relevant decisionmakers.
I Simplicity of laboratory environments may limit transferof results to parallel field. Decision makers in the outsideworld are more sophisticated.
I Technical difficulties in establishing control may also limiteffectiveness.
I Ethical issues.
MSc Economic Policy Methods Seminar
Student presentations
Presentation papers
I Comp and regulation Lyons, Sean. ”Measuring theeffects of mobile number portability on service prices.”Journal of Telecommunications Management 2.4 (2010).
I Money and banking Duffy, David, and Ciara Morley.Standard Variable Rate (SVR) Pass-Through in the IrishMortgage Market: An Updated Assessment. No.RN2015/2/3. Economic and Social Research Institute(ESRI), 2015.
I Evaluation Barnett, Adrian G., Jolieke C. van der Pols,and Annette J. Dobson. ”Regression to the mean: whatit is and how to deal with it.” International Journal ofEpidemiology 34.1 (2005): 215-220.
MSc Economic Policy Methods Seminar
Student presentations
Student presentations 5th December
I 3 Groups.
I Max 5 slides and 30 mins each.
I Briefly set out what the paper is testing and the results.
I Lessons on the empirical strategy and how the paper ispresented.
MSc Economic Policy Methods Seminar
Student presentations
Recap on Key questions - lecture 3
1. Is the research question clearly set out? (Test: can Iexplain to a non-economist?)
2. Is the dataset fully described (including any weaknesses).
3. Are there any glaring econometric issues? An omittedvariable? Is reverse causality ignored etc..
4. Are the results adequately described? Readers shouldn’tbe left wondering “did they test for this, try this?”
5. Is the research question answered? Is the paper put incontext of the literature? Is future research set out?
MSc Economic Policy Methods Seminar
Student presentations
Recap on possible econometric problems - lecture 1
I Some of these are serious and cause bias:
I Reverse casualty.I Omitted variables.I Errors-in-variables.
I Others affect efficiency and can be handled moremechanically:
I Heteroskedasticity.I Autocorrelation.
I Others are a fact of economic life and have to be dealtwith as best we can:
I Multicollinearity (unless extreme)I Outliers in the data.