Exercises

1. Researchquestion

• Whatistheresearchquestionofthepaper?

2. Data

• Whatistheunitofobservation? Whatarethemainvariablesofinterest?

• Create a table with summary statistics similar to Table 1 (only consider the first two

columnsofthattable). Forsimplicity,youcanusetwotables. Oneforallthecountries

inthedataset,andasecondonlyconsideringthecountriesinthebasesample.

3. OLSregression

• ReplicatetheTable2andtheFigure2ofthepaper.

• Why doing a simple OLS estimation is not feasible in this case? In particular, explain

withyourownwordswhatisthesourceofendogeneity.

4. IVestimation:

First-stage

• Explain briefly, using your own words, what is the relevance condition? What is the

relevanceconditioninthisspecificarticle?

• TheTable3,PanelBofthepapershowstherelationshipbetweenEuropeansettlement

andinstitutions,andsettlermortality. Replicatecolumns1-4and9-10ofthistableusing

thebasesamplecountries.

• Replicatecolumns9-10ofpanelAintheTable3ofthepaper(first-stageregression).

• Create a graph that shows the first-stage relationship between settler mortality and

theirmeasureofinstitutions(asinFigure3ofthepaper).

• Use the test for instrument strength presented in class to test the instrument used in

thispaper.

Exclusionrestriction:

• Explain briefly, using your own words, what is the exclusion restriction? What is the

exclusionrestrictioninthisspecificarticle?

• What is your personal opinion about the validity of the exclusion restriction in this

case?

2

5. Results:

• ReplicatethePanelAofTable4ofthepaper,withtheexceptionofcolumns3and4.

• Select one of the coefficients from the previous point. What is its economic interpreta-

tion?

• Discuss(briefly)theeconomicimportanceoftheresultsofthispaper.

SomeNotes

• Statahint: RememberthattodoIVregressionsinStatayoucanusethecommandivregress. I

uploadedthemanualofthecommandtotheMoodlesite.

• A subset of the countries in the sample belong to a subsample called base sample. These

observations have a value of 1 in the variable baseco in the dataset. Unless the respective

columnofthepapersaysWholeworld,youcanusethebasesampleforyourcomputations.

• Seelastpageofthishandoutforadescriptionofthevariablesinthedataset.

• Stata hint: If you want to show the country labels in your plots (as the authors do in the

paper)youcanusetheoption,mlabel(shortname),whereshortnameisthenameofthevariable

thatstoresthecountrylabelsinthedataset.

• ToexportfromStatatableswithmultiplecolumns(astheonesonthearticle)andsummary

statisticsyoucanusethepackageoutreg2ofStata. SeethenotesontheMoodlesite.

3

ListofVariables

• shortname: Countryname(short).

• log_gdp_pc: logGDPpercapita(PPP)in1995(WorldBank).

• log_outworker: logoutputperworkerin1988(Hall&Jones).

• avexpr: Averageprotectionagainstexpropriationrisk,1985-1995.

• setmort: Settlermortality.

• log_setmort: LogEuropeansettlermortality.

• africa: Dummy(= 1)forAfrica.

• asia: Dummy(= 1)forAsia.

• other: Dummy(= 1)ifnotinAfrica,Asia,ortheAmericas.

• euro_1900: Europeansettlers1900.

• democ_1900: Democracyin1900.

• cons_indep: Constraintsonexecutiveinfirstyearafterindependence.

• cons_1900: Constraintsonexecutivein1900.

• lat_abst: Latitudeofcapitalcity(normalized).

• baseco: Dummyforbasesample(= 1ifinbasesample).

4