************************************************************************************************** *** This do file creates the replication results for *** The Double-Edged Sword of Learning from Disasters: Mortality in the Tohoku Tsunami */ *** Thomas Pluemper (Vienna U of Economics and Business, Alejandro Quiroz Flores(U of Essex) and Eric Neumayer (LSE) */ *** */ *** Published in: Global Environmental Change, 44, 2017, pp. 49-56 */ ************************************************************************************************** ************************************************************************************************** /* Note: You have to change "local DIR" to the directory you copy the original stata files contained */ /* in the zip file and then run the do file. /* You must have Stata version 12 or higher installed */ ************************************************************************************************** */ version 11 drop _all clear matrix clear mata *********************************************************************************** local DIR = "d:\Research\Disasters\" /*change relative path to the directory where the files are located */ cd "`DIR'" *********************************************************************************** use "Article for GEC (learning from disasters).dta", replace set level 90 ** Preliminary analysis * Check for non-linear hazard effects nbreg toh_tot_dead c.toh_max_ht##c.toh_max_ht c.toh_med_dist##c.toh_med_dist hist_tot_dead1933 hist_maxht_1933 ln_pop, cluster(nam) estat ic * Check for interaction of hazard effects nbreg toh_tot_dead c.toh_max_ht##c.toh_med_dist hist_tot_dead1933 hist_maxht_1933 ln_pop, cluster(nam) estat ic ** Baseline model nbreg toh_tot_dead toh_max_ht toh_med_dist hist_tot_dead1933 hist_maxht_1933 ln_pop, cluster(nam) estat ic outreg2 using table1, excel replace noaster * Summary stats su toh_tot_dead toh_max_ht toh_med_dist hist_tot_dead1933 hist_maxht_1933 ln_pop emply_pc finc_strgth_indx_2012 ed_pc old_pc if e(sample) * Correlation between the two hazard measures corr toh_max_ht toh_med_dist if e(sample) * Correlation between the two learning measures corr hist_tot_dead1933 hist_maxht_1933 if e(sample) * Total number of fatalities in sample capture dropvars toh_tot_dead_total egen toh_tot_dead_total=sum(toh_tot_dead) if e(sample) di toh_tot_dead_total * Total number of historical fatalities in sample capture dropvars hist_tot_dead1933_total egen hist_tot_dead1933_total=sum(hist_tot_dead1933) if e(sample) di hist_tot_dead1933_total * Substantive analysis su toh_max_ht if e(sample), detail margins, at((p1) toh_max_ht) margins, at((p50) toh_max_ht) margins, at((p75) toh_max_ht) margins, at((p95) toh_max_ht) su toh_med_dist if e(sample), detail margins, at((p99) toh_med_dist) margins, at((p50) toh_med_dist) margins, at((p25) toh_med_dist) margins, at((p5) toh_med_dist) su ln_pop if e(sample), detail margins, at((p50) ln_pop) margins, at((p75) ln_pop) margins, at((p95) ln_pop) su hist_tot_dead1933 if e(sample) & hist_tot_dead1933>0, detail margins, at(hist_tot_dead1933=0) margins, at(hist_tot_dead1933=69) margins, at(hist_tot_dead1933=957) margins, at(hist_tot_dead1933=1678) su hist_maxht_1933 if e(sample), detail margins, at((p1) hist_maxht_1933) margins, at((p50) hist_maxht_1933) margins, at((p75) hist_maxht_1933) margins, at((p95) hist_maxht_1933) list mun hist_tot_dead1933 if e(sample) & hist_tot_dead1933>0 list hist_maxht_1933 if e(sample) & hist_maxht_1933<1 * Baseline model with other socio-economic determinants nbreg toh_tot_dead toh_max_ht toh_med_dist hist_tot_dead1933 hist_maxht_1933 ln_pop emply_pc finc_strgth_indx_2012 ed_pc old_pc , cluster(nam) estat ic outreg2 using table1, excel append noaster * Baseline model with other socio-economic determinants plus district fixed effect nbreg toh_tot_dead toh_max_ht toh_med_dist hist_tot_dead1933 hist_maxht_1933 ln_pop emply_pc finc_strgth_indx_2012 ed_pc old_pc i.nam, cluster(nam) estat ic outreg2 using table1, excel append noaster ** Robustness analysis * Robustness tests: run-up height above 2.87 metres nbreg toh_tot_dead toh_max_ht toh_med_dist hist_tot_dead1933 hist_maxht_1933 ln_pop emply_pc finc_strgth_indx_2012 ed_pc old_pc i.nam if toh_max_ht>=2.87, cluster(nam) outreg2 using table2, excel replace noaster * Robustness tests: drop, separately, each of the three highest values of historical deaths * Drop Iwate-ken Miyako-shi with 957 historical deaths nbreg toh_tot_dead toh_max_ht toh_med_dist hist_tot_dead1933 hist_maxht_1933 ln_pop emply_pc finc_strgth_indx_2012 ed_pc old_pc i.nam if mun!="Iwate-ken Miyako-shi", cluster(nam) estat ic outreg2 using table2, excel append noaster * Drop Iwate-ken Ofunato-shi with 344 historical deaths nbreg toh_tot_dead toh_max_ht toh_med_dist hist_tot_dead1933 hist_maxht_1933 ln_pop emply_pc finc_strgth_indx_2012 ed_pc old_pc i.nam if mun!="Iwate-ken Ofunato-shi", cluster(nam) estat ic outreg2 using table2, excel append noaster * Drop Iwate-ken Fudai-mura with 1678 historical deaths nbreg toh_tot_dead toh_max_ht toh_med_dist hist_tot_dead1933 hist_maxht_1933 ln_pop emply_pc finc_strgth_indx_2012 ed_pc old_pc i.nam if mun!="Iwate-ken Fudai-mura", cluster(nam) estat ic outreg2 using table2, excel append noaster