****"Iceland on the Rocks: The Mass Political Economy of Sovereign Debt Resettlement"
****By K. Amber Curtis, Joseph Jupille, and David Leblang


*******************************
*For Publication
*DL 08March2013
********************************
set mem 500m
use "FINAL 2011 DATA + 2010 KNOWL Qs 11.18.11.dta"


***GENERATE DV 
tab votechoice11
clonevar vc = votechoice11
recode vc -1=.
tab vc

***TABLE 2: Models with Party ID

*Model 1: borrowing costs (ccdebt)
estsimp logit vc ccdebt11 ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11, cluster(zipcode) nolog
drop b1-b10
est store A
*Model 2: fiscal retrenchment (unemployed + public sector employee)
estsimp logit vc unempl publicmo11 ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11, cluster(zipcode) nolog
drop b1-b11
est store B
*Model 3: investment assets
estsimp logit vc investments ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11, cluster(zipcode) nolog
drop b1-b10
est store C
*Model 4: saturated
estsimp logit vc ccdebt11 unempl publicmo11 investments ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11, cluster(zipcode) nolog

est store D
setx mean
plotfds, changex(p25 p75) cont(ccdebt ftindex id11 educ11) disc( unempl investments progressive independence socdemalliance leftgreen male)  sortorder(ccdebt11 unempl publicmo11 investments ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11) clevel(.90) label
drop b1-b13
***TABLE 3: Models with PM Approval

*Model 5: borrowing costs
estsimp logit vc ccdebt11 ftindex11 id11 pmapprov11 educ11 male11, cluster(zipcode) nolog
drop b1-b7
est store E
*Model 6: fiscal retrenchment (unemployed + public sector employee)
estsimp logit vc unempl publicmo11 ftindex11 id11 pmapprov11 educ11 male11, cluster(zipcode) nolog
drop b1-b8
est store F
*Model 7: investment assets
estsimp logit vc investments ftindex11 id11 pmapprov11 educ11 male11, cluster(zipcode) nolog
drop b1-b7
est store G
*Model 8: saturated PM Approval
estsimp logit vc ccdebt11 unempl publicmo11 investments ftindex11 id11 pmapprov11 educ11 male11, cluster(zipcode) nolog
drop b1-b7
est store H
**NB that only investments and unemployment are robust to alt specification of cues by PM Approval

**20120413jj

**public sector employee is not sig even when we drop unemployment

estsimp logit vc publicmo11 ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11, cluster(zipcode) nolog
drop b1-b10
est store A1
**Is Education picking up cosmopolitanism? Here are a few checks.

*Model 9: model 4, w/ cosmo, w/o ed. Check for effects on coefficients/SEs of cosmopolitanism variables
estsimp logit vc ccdebt11 unempl publicmo11 investments ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 male11, cluster(zipcode) nolog
drop b1-b12
est store A2
**When we re-run model 4 without education (model 9), we get the following. Re: Borrowing Costs: ccdebt now stat sig at .10. Picking up some of the effects of education. Public sector employee becomes stat sig for the first time when we drop Education, coefficient 40% larger with same standard error. Note that it is positive: this is opposite of Tomz's Fiscal Retrenchment hypothesis.

** When we re-run model 4 without education (model 9) Re cosmo variables: ftindex increases by a factor of ten!

*Model 10: model 4, but w/o cosmo, w/education. Contrast Model 9.
estsimp logit vc ccdebt11 unempl publicmo11 investments progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11, cluster(zipcode) nolog
drop b1-b11
est store A3
**It’s what you’d expect: education matters a lot.

**Model 11: just drop ftindex11 and see what coeff on education looks like Baseline is Model 4
estsimp logit vc ccdebt11 unempl publicmo11 investments id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11, cluster(zipcode) nolog
drop b1-b12
est store A4
*Table Four: Interactions between knowledge/sophistociation and material well-being
*Model 9: borrowing costs (ccdebt)
logit vc ccdebt11 ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11 knowledge11, cluster(zipcode) nolog
est store I
*Model 12: saturated
logit vc c.knowledge11##c.ccdebt11 c.knowledge11##c.unempl  c.knowledge11##c.investments publicmo11 ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11, cluster(zipcode) nolog
est store J


label var investments11 "Has Investments"
label var publicmo11 "Public Sector Money"
label var ftindex11 "Feeling Thermometer Index"
label var id11 "Icelandic Identity Index"
label var ccdebt11 "Credit Card Debt Index"

esttab A B C D , se star(* 0.10 ** 0.05) label onecell, using table2.rtf, replace
esttab E F G H , se  star(* 0.10 ** 0.05) label onecell, using table3.rtf, replace
esttab I J , se  star(* 0.10 ** 0.05) label onecell, using table3.rtf, replace

** For Appendix
** Results referenced in footnote #45
esttab A1 A2 A3 A4 , se star(* 0.10 ** 0.05) label onecell, using AppendixTable1.rtf, replace
 
** Results referenced in footnote #46
*Selection Model
heckprob vc ccdebt11 unempl publicmo11 investments ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11, sel(turnout=ccdebt11 unempl publicmo11 investments ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11) cluster(zipcode) nolog
est store A5
*Add spoiled ballets to "no" votes
clonevar vc2 = votechoice11
recode vc2 -1=0
logit vc2 ccdebt11 unempl publicmo11 investments ftindex11 id11 progressive11 independence11 socdemalliance11 leftgreenmovnt11 educ11 male11, cluster(zipcode) nolog
est store A6