Content
Citation(s) & Footnote(s)
Book(s)No access
Source argument
Dills and Miron argue that state prohibition had little consistent effect on cirrhosis and federal prohibition a more modest effect than simple before/after comparisons imply. Their interpretation depends on an indirect outcome and observational model choices.
Evidence map
Mortality is an imperfect proxy for alcohol consumption
Cirrhosis mortality is an indirect outcome with demographic, measurement and lag limitations. The authors' average within-country correlation of .75 differs from Figure1's cross-country R-squared .5657. Neither statistic is a causal estimate of Prohibition or whiskey consumption.
Printed pp.1–8; PDF3–10; Tables1–5/PDF29–33; Figure1/PDF38.
Declines preceded national Prohibition
Much of the plotted mortality decline precedes January1920. The absence of an abrupt repeal jump is a weak test because outcome lags are uncertain. Descriptive timing does not establish which policy or social change caused the decline.
Printed pp.9–13; PDF11–15; Figures3–7/PDF41–43.
State prohibition episodes require date and scope qualification
State restrictions differed in coverage and exceptions. Table6 lists passage years, not independently verified effective or enforcement dates. Repeated Alabama and New Hampshire entries reflect repeal and reintroduction, not duplicate records. Historical examples derive from cited secondary accounts.
Printed pp.11–12; PDF13–14; Table6/PDF34.
State regressions are sensitive to period and specification
The state-panel models use fixed effects, year indicators, lagged mortality and population weights. State prohibition coefficients of .238, .307 and -.089 differ across periods. Policy endogeneity and omitted state-time factors limit causal conclusions; these regressions were not independently reproduced.
Printed pp.13–16; PDF15–18; Table7/PDF35.
Federal estimates depend on controls and model design
The authors compare a federal prohibition indicator with real enforcement expenditure and vary tax controls, including interactions. The 1917–1919 indicator does not separately identify influenza, wartime restrictions, immigration or stockpiling effects. Such explanations remain hypotheses.
Printed pp.16–20; PDF18–22; Tables8–9/PDF36–37.
The headline estimate hides a wide specification range
The authors summarize a 10–20 percent cirrhosis reduction, but the tabulated simulated decreases span -20.8 to30.9 percent, with a reported average12.6 percent. Negative decreases denote estimated increases. This is neither a 10–20 percent reduction in whiskey drinking nor a measure of all policy harms or benefits.
Printed pp.20–20; PDF22; Tables8–9/PDF36–37.
Table values qualify the prose's generalizations
Table8 lagged-mortality coefficients are .887–.891, whereas Table9 gives .725–.744; the prose's general .87–.88 description does not cover both. Table9's 1917–1919 t-statistics include1.07,1.24 and1.69, so its blanket significance claim should not be repeated as conventional five-percent significance.
Printed pp.19–20; PDF21–22; Tables8–9/PDF36–37.
Price and evasion explanations remain hypotheses
Nominal illicit prices cannot be equated with real prices without inflation and expenditure-weight information. Low heavy-drinker price responsiveness, evasion, illicit marginal cost and forbidden-fruit effects are proposed explanations, not mechanisms established by this study. Referenced works remain discovery leads, not reviewed sources.
Printed pp.21–23; PDF23–25; referencesPDF26–28.
Researcher synthesis
Use this source to explain why the effects of Prohibition cannot be summarized by a single mortality curve or policy-success label. Distinguish all-alcohol outcomes from whiskey-specific history. Causal interpretations remain attributed, and dated medical propositions are not endorsed as current advice.
Assessment
Full source reading is complete; underlying data and models have not been independently reproduced. Different national scales and missing series limit visual comparisons. Federal/state legal dates need original authority before timeline use. The headline estimate must retain the full model range and table/prose inconsistencies.
Completion checklist
Sources
Dills, Angela K., and Jeffrey A. Miron. “Alcohol Prohibition and Cirrhosis.” NBER Working Paper 9681, May 2003. https://doi.org/10.3386/w9681.
No access
