Robustness analyses for the long-term effects of the reform on mortality and natural logarithm of income at ages 78–95, cohorts 1890–1917, Sweden: Exponentiated coefficients from stratified Cox partial likelihood models for mortality (except for panel D), adjusted for left-truncation at age 78, and linear fixed-effects regression estimates for ln(income)
| . | Mortality Risk . | Ln(Income) . | Mortality Risk . | Ln(Income) . |
|---|---|---|---|---|
| A. Adding Investment Into Education, Infrastructure, and Poor Relief | B. Adding Pretreatment Number of Residents With Voting Rights per Capita | |||
| Postb × New Health Districtp | 0.932** | 0.0286** | 0.940** | 0.0187† |
| (.002) | (.006) | (.004) | (.061) | |
| Number of individuals | 39,604 | 38,618 | 39,604 | 38,618 |
| Number of deaths | 36,429 | 36,429 | ||
| Parishes of birth | 414 | 414 | 414 | 414 |
| C. Correcting for Survival Bias Using Heckman Two-Stage Procedure | D. Correcting for Survival Bias Following van den Berg and Drepper | |||
| Postb × New Health Districtp | 0.940** | 0.0200* | 0.935** | –– |
| (.006) | (.044) | (.002) | –– | |
| Number of individuals | 39,604 | 38,618 | 39,604 | |
| Number of deaths | 36,429 | 36,429 | ||
| Parishes of birth | 414 | 414 | 414 | |
| E. Excluding Counties of Birth Affected by Emigration to the United States | F. Adding Controls for Coinciding Unfavorable Health and Economic Conditions | |||
| Postb × New Health Districtp | 0.938** | 0.0186† | 0.937** | 0.0220* |
| (.006) | (.081) | (.007) | (.027) | |
| Number of individuals | 36,023 | 35,134 | 39,604 | 38,618 |
| Number of deaths | 33,127 | 36,429 | ||
| Parishes of birth | 385 | 385 | 414 | 414 |
| G. Excluding Counties of Birth Affected by WWI | H. Adding Parish-Specific Controls for Age Structure | |||
| Postb × New Health Districtp | 0.933** | 0.0248* | 0.943** | 0.0208* |
| (.004) | (.018) | (.009) | (.024) | |
| Number of individuals | 35,320 | 34,453 | 39,604 | 38,618 |
| Number of deaths | 32,391 | 36,429 | ||
| Parishes of birth | 409 | 409 | 414 | 414 |
| I. Adding Parish-Specific Controls for Local Labor Market and Migrant Structure | J. Excluding Counties of Residence Affected by Spanish Flu | |||
| Postb × New Health Districtp | 0.930** | 0.0251* | 0.938* | 0.0225* |
| (.004) | (.016) | (.010) | (.043) | |
| Number of individuals | 39,604 | 38,618 | 32,212 | 31,433 |
| Number of deaths | 36,429 | 29,482 | ||
| Parishes of birth | 414 | 414 | 411 | 411 |
| Year of Birth Fixed Effects | Yes | Yes | Yes | Yes |
| Parish of Birth Fixed Effects | Yes | Yes | Yes | Yes |
| . | Mortality Risk . | Ln(Income) . | Mortality Risk . | Ln(Income) . |
|---|---|---|---|---|
| A. Adding Investment Into Education, Infrastructure, and Poor Relief | B. Adding Pretreatment Number of Residents With Voting Rights per Capita | |||
| Postb × New Health Districtp | 0.932** | 0.0286** | 0.940** | 0.0187† |
| (.002) | (.006) | (.004) | (.061) | |
| Number of individuals | 39,604 | 38,618 | 39,604 | 38,618 |
| Number of deaths | 36,429 | 36,429 | ||
| Parishes of birth | 414 | 414 | 414 | 414 |
| C. Correcting for Survival Bias Using Heckman Two-Stage Procedure | D. Correcting for Survival Bias Following van den Berg and Drepper | |||
| Postb × New Health Districtp | 0.940** | 0.0200* | 0.935** | –– |
| (.006) | (.044) | (.002) | –– | |
| Number of individuals | 39,604 | 38,618 | 39,604 | |
| Number of deaths | 36,429 | 36,429 | ||
| Parishes of birth | 414 | 414 | 414 | |
| E. Excluding Counties of Birth Affected by Emigration to the United States | F. Adding Controls for Coinciding Unfavorable Health and Economic Conditions | |||
| Postb × New Health Districtp | 0.938** | 0.0186† | 0.937** | 0.0220* |
| (.006) | (.081) | (.007) | (.027) | |
| Number of individuals | 36,023 | 35,134 | 39,604 | 38,618 |
| Number of deaths | 33,127 | 36,429 | ||
| Parishes of birth | 385 | 385 | 414 | 414 |
| G. Excluding Counties of Birth Affected by WWI | H. Adding Parish-Specific Controls for Age Structure | |||
| Postb × New Health Districtp | 0.933** | 0.0248* | 0.943** | 0.0208* |
| (.004) | (.018) | (.009) | (.024) | |
| Number of individuals | 35,320 | 34,453 | 39,604 | 38,618 |
| Number of deaths | 32,391 | 36,429 | ||
| Parishes of birth | 409 | 409 | 414 | 414 |
| I. Adding Parish-Specific Controls for Local Labor Market and Migrant Structure | J. Excluding Counties of Residence Affected by Spanish Flu | |||
| Postb × New Health Districtp | 0.930** | 0.0251* | 0.938* | 0.0225* |
| (.004) | (.016) | (.010) | (.043) | |
| Number of individuals | 39,604 | 38,618 | 32,212 | 31,433 |
| Number of deaths | 36,429 | 29,482 | ||
| Parishes of birth | 414 | 414 | 411 | 411 |
| Year of Birth Fixed Effects | Yes | Yes | Yes | Yes |
| Parish of Birth Fixed Effects | Yes | Yes | Yes | Yes |
Notes: Panel D shows exponentiated shared gamma-frailty (Weibull baseline distribution) model estimates, adjusted for left truncation at age 78. The indicator Postb× New Health Districtp denotes treated parishes and cohorts, for which the new local health district established in parish of birth p in a year of birth b and remained in place in the post-treatment period. In panel E, Värmland and Halland counties are excluded as the primary emigrant counties, where the countryside experienced mass migration to the United States. In panel G, Norrbotten and Västerbotten counties are excluded because most were affected by World War I. In panel J, Koppaberg, Gävleborg, Västernorrland, Jämtland, Västerbotten, and Norrbotten counties are excluded because most were affected by the Spanish flu. For further details, see online appendix K. Standard errors (not shown) are clustered at the parish-of-birth level. Numbers in parentheses are p values.
†p < .10; *p < .05; **p < .01