Assembling the evidence…
We exploit differences in European mortality rates to estimate the effect of institutions on economic performance. Europeans adopted very different colonization policies in different colonies, with different associated institutions. In places where Europeans faced high mortality rates, they could not settle and were more likely to set up extractive institutions. These institutions persisted to the present. Exploiting differences in European mortality rates as an instrument for current institutions, we estimate large effects of institutions on income per capita. Once the effect of institutions is controlled for, countries in Africa or those closer to the equator do not have lower incomes. (JEL O11, P16, P51)
Strengths
The settler mortality instrument is motivated by a clear three-premise theory, and each link in the causal chain (mortality → settlements → early institutions → current institutions → income) is empirically documented in Table 3. This makes the identification strategy more transparent and credible than prior approaches using ethnolinguistic fragmentation or latitude as instruments.
↳ Sections I and IV; Tables 3–4
The paper tests robustness across subsamples (excluding Neo-Europes, excluding Africa), alternative outcome measures (output per worker), and a wide range of controls including colonial identity, legal origin, religion, geography, natural resources, soil quality, ethnolinguistic fragmentation, and current disease environment. The main coefficient remains stable throughout, with remarkably little change across specifications.
↳ Section V; Tables 4–5
By providing the first credible IV-based estimate of the causal effect of institutions on income per capita, the paper fills a specific gap in the development economics literature where prior work relied on correlational evidence or instruments with questionable validity. The estimated 2SLS coefficient of 0.94 implies substantial but plausible effects of institutional differences.
↳ Section IV; Table 4 column 1
Limitations
The core identifying assumption — that settler mortality affects current income only through institutions — cannot be fully verified. The overidentification tests use European migration and early institutions as additional instruments, but these may themselves face validity concerns. The paper relies on the argument that malaria and yellow fever affected Europeans differentially from indigenous populations, which is plausible but not conclusive.
↳ Section III.A; Section V
The main institutions variable (protection against expropriation risk, 1985–1995) is a commercial index from Political Risk Services, measured with acknowledged error and potentially endogenous to income expectations. While the paper argues this biases OLS downward and reports alternative measures in working paper appendices, the reliance on a single subjective index for the main specifications introduces measurement uncertainty.
↳ Table 1; Section II.A; Footnote 11
The paper establishes that institutions matter for development but does not identify which specific institutional features are most important, how reforms should be sequenced, or over what time horizon changes would operate. The gap between 'institutions matter' and actionable policy implementation remains substantial.
↳ Section IV.B; Section VI
This paper demonstrates consistent strength across all quality dimensions, with particular distinction in its theoretical contribution and methodological design. The settler mortality instrument is well-motivated by a clear three-premise theory, and the empirical tracing of each causal link (Tables 3–4) sets a high standard for macro-historical identification strategies. The robustness analysis across subsamples, outcome measures, and an unusually wide range of controls (Section V, Table 5) reinforces the stability of the main findings. The principal constraint is the inherent untestability of the exclusion restriction — whether settler mortality affects income only through institutions — which all evaluative perspectives identified as the central vulnerability. The paper addresses this threat directly and honestly, particularly through the differential mortality argument in Section III.A, but it cannot be fully resolved within this research design. The slight abstract overstatement regarding the Africa and geography results (the Asia dummy remains significant in some IV specifications, Table 4 columns 7–8) is a minor imprecision in an otherwise disciplined exposition.
Exemplary4.4/5.0
The paper introduces settler mortality as a theoretically grounded instrument for institutional quality, enabling the first credible causal estimate of the institutions-income relationship in a literature previously limited to cross-country correlations. The three-premise theory is empirically traced through each link (mortality → settlements → early institutions → current institutions → income), representing a substantial advance over contemporaneous work.
settler mortality rates were a major determinant of settlements; settlements were a major determinant of early institutions
The IV design is well-suited to the causal question, with the first stage documented, the exclusion restriction argued against the main threat (current disease environment), and robustness demonstrated across subsamples, alternative outcome measures, and extensive controls (colonial identity, legal origin, geography, religion, ethnolinguistic fragmentation, current disease burden). The main limitation is that the exclusion restriction is argued rather than provable, and weak-instrument F-statistics are not reported, though both are era-appropriate for 2001.
The argument structure is exceptionally well-organized, moving from the three-premise theory through OLS motivation, instrument construction, IV results, and robustness in a logical sequence. Language is disciplined throughout, with OLS results described as correlations and the exclusion restriction acknowledged as an assumption rather than established fact.
The paper engages substantively with the prior literature on institutions, geography, and colonial origins, positioning its IV approach against the instruments used by Mauro (1995) and Hall & Jones (1999) and explaining why those are potentially invalid. Limitations are acknowledged honestly, including the weaker within-Africa relationship and the untestable exclusion restriction, though engagement with measurement concerns about the Political Risk Services index is less thorough.
No concerns3 of 4 checks
The paper is internally consistent. Numbers reported in text align with tables, the first-stage R² of 0.27 matches Table 3, OLS estimates in Table 2 are consistent with Panel C of Table 4, and the descriptive statistics in Table 1 cohere with sample descriptions throughout. A minor imprecision in the abstract's claim about the Africa dummy is noted but is not material.
No image or data integrity concerns. Conflicts of interest are not disclosed, but this was not standard practice in economics in 2001. Ethics approval is not applicable (secondary data analysis of historical records). Pre-registration was not a norm. Data stated as available from the authors, which is era-appropriate.
Flags: 1 declared / 5 total
Reference integrity check could not run: no references section detected in extracted text.
No retraction notice found in Retraction Watch.
Sources: Retraction Watch ✓PubPeer (coming soon)
High3.9/5.0
The paper addresses the central question in development economics — why some countries are rich and others poor — with direct relevance to development economists, international financial institutions, and policymakers designing institutional reform programs. The finding that institutions rather than geography drive income differences has clear implications for development policy priorities.
The paper provides causal identification of the institutions-income relationship using cross-country variation across 64 former colonies, but does not specify which institutional reforms are most effective, how to implement them, or over what time horizon. Substantial additional work is required to translate the macro-level finding into actionable policy guidance.
Results are robust across subsamples excluding Neo-Europes and Africa, with alternative outcome measures and extensive controls, but the instrument is inherently limited to former European colonies and the within-Africa relationship is acknowledged to be weaker. Generalizability to non-colonial contexts is limited by design.
if we limit the sample to African countries only, the first-stage relationship ... becomes considerably weaker
The paper fills a specific identification gap in an existing literature (Knack & Keefer, Hall & Jones, Engerman & Sokoloff) and provides the methodological template for subsequent IV-based studies of institutions and development. It directly catalyzed a large follow-on research program including debates about the exclusion restriction and extensions to other institutional dimensions.
Nabu’s assessment, alongside the field’s view.
Evaluation Metadata
AI-powered, human-governed. Something look off? Contact us to request a review.