The Colonial Origins of Comparative Development: An Empirical Investigation
2000 · paper · Economics
Uses differences in European settler mortality across former colonies as an instrument to estimate how institutions affect income per capita. It reports large effects of institutions, and that Africa and distance from the equator no longer predict lower income once institutions are controlled for.
Checked: PDF from economics.mit.edu, fetched 7 Nov 2026. Page numbers are those of that PDF.
9 findings
Substance Section IV.B: 'Our 2SLS estimate, 0.94, implies that the 2.24 differences in expropriation risk between these two countries should translate into 206 log point (approximately 7-fold) difference.' The gap of 2.24 matches neither the 2.2 used in Section II.B (7.8 − 5.6) nor Table A2 (7.82 − 5.55 = 2.27), and 0.94 × 2.24 gives about 211 log points, not 206.
- p. 20 Section IV.B, first paragraph
implies that the 2.24 differences in expropriation risk between these two countries should translate into 206 log point
- p. 31 Appendix Table A2, Chile and Nigeria
7.82
- p. 11 Section II.B
5.6, and Chile, which has approxi- mately the 75th percentile of the institutions index, 7.8
- p. 20 Section IV.B, first paragraph
Our 2SLS estimate, 0.94, implies that the 2.24 differences in expropriation risk between these two countries should translate into 206 log point (approximately 7-fold) difference.
Only the authors can say what is right here.
Judge's note: Checked p.20 (2.24 and 206), p.11 (Nigeria 5.6, Chile 7.8; the OLS calculation 0.52 × 2.2 = 1.14 uses a gap of 2.2) and Table A2 on p.31 (Chile 7.82, Nigeria 5.55). The printed 206 is consistent with a gap of 2.2, so 2.24 appears to be the inconsistent figure, but only the authors can say which gap was intended. The '7-fold' description holds under any of these gaps.
Second check: another model agreed
Report a problem with this finding
Substance Appendix Table A1, Democracy in 1900: 'Equal to 1 if country not independent at that date', whereas the text says colonies are assigned 'the lowest possible score' (0 on the 0–10 scale) and Table 1 reports a mean of 0 for mortality quartile (4).
- p. 30 Appendix Table A1, Democracy in 1900
Equal to 1 if country not independent at that date.
- p. 11 Section II.A
we assign these countries the lowest possible score
- p. 10 Table 1, Democracy in 1900, quartile (4)
0
Only the authors can say what is right here.
Judge's note: A1 image (p. 30) reads 'Equal to 1 if country not independent at that date' for Democracy, the same wording as the Constraint-on-executive entry (1–7 scale). The p. 11 text says colonies get 'the lowest possible score'. On the stated 0–10 scale that score is 0. Table 1 (p. 10) shows Democracy in 1900 = 0 for quartile (4), whose 15 countries have mortality ≥ 280 and include colonies such as Angola and Mali (Table A2). Coding colonies as 1 would make that mean at least positive. The A1 coding statement conflicts with the text and with Table 1.
Second check: another model agreed
Report a problem with this finding
Substance Yellow fever paragraph (p. 1393): 'comparable to our baseline estimate of 0.95 reported in Table 4', but Table 4 column (1) reports 0.94.
- p. 26 Section V.A, yellow fever paragraph
comparable to our baseline estimate of 0.95 reported in Table 4
- p. 19 Table 4, Panel A, column (1)
0.94
- p. 20 text on Table 4
The 2SLS coefficient is 0.98 instead of 0.94 as in column (1)
- p. 26 Sec. V.A, yellow fever paragraph
comparable to our baseline estimate of 0.95
Judge's note: The Table 4 crop shows Panel A column (1) = 0.94 (0.16), and no Table 4 column equals 0.95. The text elsewhere (pp. 18, 20, 24 and the Table 8 discussion) uses 0.94 as the baseline. The p. 26 image reads 0.95.
Second check: another model agreed
Report a problem with this finding
Substance Footnote 25: 'British colonies have, on average, an index of institution that is 0.63 points lower'. The first-stage British dummy in Table 5 Panel B col (1) is +0.63, and the footnote itself derives '69 log points higher income', so 'lower' should read 'higher'.
- p. 21 Footnote 25
an index of institution that is 0.63 points lower
- p. 22 Table 5, Panel B, British colonial dummy, column (1)
0.63
- p. 21 footnote 25
British colonies have, on average, an index of institution that is 0.63 points lower
- p. 21 text
the British dummy is positive, and marginally significant
Judge's note: Table 5 Panel B on p.22 shows the British colonial dummy as 0.63 (0.37) with no minus sign. The main text on p.21 says the British dummy 'is positive, and marginally significant'. The footnote's own arithmetic (1.10 × 63 ≈ 69, then −78 + 69 = −9) needs a positive effect on institutions.
Second check: another model agreed
Report a problem with this finding
Reference or label Page 11 cites "Monte- squieu [1784] (1989)", while page 6 has "Charles de Montesquieu [1748] (1989)" and the reference list has "[1748] 1989".
- p. 11 Section II.B, last line
squieu [1784] (1989)
- p. 6 Introduction
Charles de Montesquieu [1748] (1989)
- p. 34 References
Press, [1748] 1989.
- p. 11 Section II.B
Monte- squieu [1784] (1989)
Judge's note: The page 11 image shows [1784]. The page 6 image and the reference entry on page 34 show [1748]. The reference list has only one Montesquieu entry, so 1784 is a transposition.
Second check: another model agreed
Report a problem with this finding
Reference or label The Table 1 note defines quartile (2) as "greater than or equal to 65.4 and less than 78.1" and quartile (3) as "greater than or equal to 78.1 and less than 280". Applied to Appendix Table A2, these cut-offs give 13 and 22 countries, not the printed 18 and 17.
- p. 10 Table 1 notes
(2) greater than or equal to 65.4 and less than 78.1; (3) greater than or equal to 78.1 and less than 280
- p. 10 Table 1, Number of observations row
14 18 17 15
- p. 31 Appendix Table A2, main mortality estimate
78.1
Only the authors can say what is right here.
Judge's note: I classified all 64 countries in A2 (page 31 image and layout text). Under the stated cut-offs: Q1 (<65.4) = 14, Q2 = 13, Q3 = 22 (including the five countries at exactly 78.1: CRI, SLV, HND, PRY, VEN), and Q4 (≥280) = 15. The printed 18 and 17 are reached only if the 78.1 countries go into Q2. With that placement the Q2 mean log GDP is 151.34/18 = 8.41, matching the printed 8.4. The boundaries in the note therefore do not reproduce the published counts. Unrounded mortality values might explain it, but A2 prints 78.1 for these countries.
Second check: another model agreed
Report a problem with this finding
Reference or label The text says "In row 3, we also give output per worker in 1988" and that the expropriation index is "re- ported in the second row". In Table 1, output per worker is row 2 and the expropriation index is row 3.
- p. 10 Section II.A, text
In row 3, we also give output per worker in 1988
- p. 10 Section II.A, text
Our main variable, re- ported in the second row, is an index of protection against expropriation
Judge's note: The page 10 image of Table 1 has the rows in this order: Log GDP per capita, Log output per worker in 1988, Average protection against expropriation risk. No single way of counting rows makes both text pointers correct, so they are swapped.
Second check: another model agreed
Report a problem with this finding
Reference or label The text says "The estimate in column (1), 0.52,", but Table 2 column (1) (whole world) is 0.54; 0.52 is column (2) (base sample).
- p. 11 Section II.B, Nigeria–Chile example
The estimate in column (1), 0.52,
- p. 12 Table 2, columns (1) and (2)
0.54 0.52
- p. 11 Section II.B, magnitude paragraph
The estimate in column (1), 0.52
- p. 12 Table 2, Average protection against expropriation risk
0.54 0.52 0.47 0.43
Judge's note: The Table 2 image shows 0.54 in col (1) and 0.52 in col (2). The worked example uses 0.52 and the base-sample comparison of Nigeria (5.6) and Chile (7.8); 0.52*2.2 = 1.144 matches the printed 1.14. The column pointer should read (2).
Second check: another model agreed
Report a problem with this finding
Reference or label Footnote 29: 'The standard deviation of distance from the equator in the world is 1.89, greater than 1.33 in our sample', yet latitude is defined as 'distance from the equator, scaled between 0 and 1'.
- p. 26 Footnote 29
The standard deviation of distance from the equator in the world is 1.89, greater than 1.33 in our sample.
- p. 12 Section II.B
distance from the equator, scaled between 0 and 1
- p. 30 Appendix Table A1, Latitude
scaled to take values between 0 and 1
Only the authors can say what is right here.
Judge's note: Page images show footnote 29 (p. 26) with 1.89 and 1.33. Section II.B (p. 12) and Appendix Table A1 define latitude on a 0–1 scale. The standard deviation of a variable bounded in [0, 1] cannot exceed 0.5, so these values must be in some other unit (e.g. scaled by 10) or misprinted. The footnote does not state a unit. Only the authors can say which.
Second check: another model agreed
Report a problem with this finding
21 minor (typos and the like; most checked by the judge only)
Substance The text dates the "Beauchamp Committee in 1795" that chose where to send British convicts, a decision that led to sending them to Australia.
- p. 7 Sec. I.A
Beauchamp Committee in 1795
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: the Beauchamp Committee sat in 1785, and the First Fleet reached Australia in 1788. Within the paper, the committee is described as deciding where to send convicts and ending with the decision for Australia; the paper does not date that settlement. The page 7 image shows 1795. Only the authors can confirm the intended year.
Second check: another model agreed
Report a problem with this finding
Substance The text says "Slavery persisted in Brazil until 1886".
- p. 9 Section I.C
Slavery persisted in Brazil until 1886
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: slavery in Brazil was abolished in 1888 (Lei Áurea). The page 9 image shows 1886, so this is not an extraction artefact. Nothing in the files settles the year.
Second check: another model agreed
Report a problem with this finding
Substance Table 1 gives quartile (3) mean log GDP per capita as "7.73", but the 17 quartile-3 countries in Appendix Table A2 average 7.78.
- p. 10 Table 1, Log GDP per capita, quartile (3)
7.73
- p. 10 Appendix Table A2
Log GDP per capita (PPP) in 1995
Only the authors can say what is right here.
Judge's note: Q3 countries per A2 (78.1 < mortality < 280): DZA, BHS, DOM, IDN, JAM, KEN, MAR, NIC, PAN, SEN, SDN, TZA, TTO, HTI, VNM, COG, ZAR. Their log GDP sums to 132.23, a mean of 7.78. The other quartile means recompute to the printed values: Q1 124.45/14 = 8.89 (8.9 printed), Q2 151.34/18 = 8.41 (8.4), Q4 107.98/15 = 7.20 (7.2). A gap of 0.05 is too large for rounding of the A2 entries. A different data vintage is possible, and only the authors can say which figure is right.
Second check: another model agreed
Report a problem with this finding
Reference or label The Table 3 notes define 'European settlements is percent of population that was European or of European descent in 1900', but the variable is reported as a fraction (Table 1 means 0.31, 0.16, 0.005; text: 'fraction of the population').
- p. 18 Table 3 notes
European settlements is percent of population that was European or of European descent in 1900.
- p. 10 Table 1, European settlements in 1900
0.31
- p. 11 Section II.A
gives the fraction of the pop- ulation of European descent in 1900
Judge's note: Page 10 image: the Table 1 values for European settlements in 1900 run from 0.005 to 0.32 (SD 0.4). Page 11 calls it 'the fraction of the population of European descent'. Page 17 also says 'fraction'. Page 18, Table 3 notes, say 'percent'. The unit label disagrees with the scale of the data and with the text, and coefficients such as 3.20 only make sense on a 0–1 scale.
Second check: another model agreed
Report a problem with this finding
Reference or label The text describes the early-institution rows as constraints in 1900, then "the second is an index of democ- racy in 1900", then "In the following row" constraints in the first year of independence. Table 1 places "Constraint on executive in first year of independence" between the 1900 constraint and democracy rows.
- p. 11 Section II.A
The first is a measure of constraints on the executive in 1900 and the second is an index of democ- racy in 1900.
- p. 18 Table 1 rows
Constraint on executive in first year of independence
Judge's note: The page 10 image of Table 1 gives the row order: Constraint on executive in 1900, Constraint on executive in first year of independence, Democracy in 1900. The page 11 text describes them in a different positional order. This is a minor row-pointer mismatch; the data are unaffected.
Second check: another model agreed
Report a problem with this finding
Reference or label The Figure 2 x-axis reads 'Average Expropriation Risk 1985-95' (caption: 'OLS Relationship Between Expropriation Risk and Income'), but the plotted index is protection against expropriation, where higher means more protection.
- p. 13 Figure 2 x-axis label
Average Expropriation Risk 1985-95
Judge's note: Page 13 image: on the axis, USA, CAN and SGP sit near 10 and ZAR and HTI near 4. The Table 2 notes on page 12 say 'a higher score means more protection against expropriation'. The axis label therefore names the opposite of what the scale measures. The fix is to label it as protection against expropriation risk. The report covers Figure 3 too, but I checked only Figure 2.
Second check: another model agreed
Report a problem with this finding
Wording 'Latitude itself is also significant' is said of columns (3)–(6) of Table 2, but latitude is 0.37 (0.51) in column (4) and 0.92 (0.63) in column (6).
- p. 12 Section II.B
Latitude itself is also significant and has the sign found by the previous studies
- p. 12 Table 2, Latitude row
0.89 0.37 1.60 0.92
Only the authors can say what is right here.
Judge's note: Page 12 image, Table 2: the latitude coefficient-to-SE ratios are about 1.8 in col 3, 0.7 in col 4, 2.3 in col 5 and 1.5 in col 6. The sentence comes right after 'in columns (3)–(6), we add latitude as a regressor' and before the continent dummies are introduced, so it may refer only to cols 3 and 5. Even then, col 3 is significant only at about the 10% level. As written, the sentence is broader than the table supports; the authors should say which columns it covers.
Second check: another model agreed
Report a problem with this finding
Wording Text (p. 1391) says Appendix A shows that 'the inclusion of an endogenous variable positively correlated with income or institutions will bias the coefficient on institutions downwards', but Appendix A sets up z as a negative influence (phi < 0, pi < 0, cov(z_i, R_i) < 0).
- p. 24 Section V.A, ethnolinguistic fragmentation paragraph
the inclusion of an endogenous variable positively correlated with income or institutions will bias the coefficient on insti- tutions downwards
- p. 29 Appendix A
cov(zi, Ri) < 0, that is, the factor zi is likely to be negatively correlated with positive influ- ences on income
Judge's note: Read p. 1391 (PDF p. 24) and Appendix A (PDF p. 29). The appendix assumes pi < 0, phi < 0, cov(eta, eps) < 0 and cov(z, R) < 0, i.e. z is 'negatively correlated with positive influences on income'. Flipping the sign of z (z' = -z) flips kappa and cov(z, eps) together, so the bias term kappa*cov(z, eps) is unchanged and the downward-bias result also holds for a positively correlated variable. The text's description therefore does not match how the appendix sets up the case, but the result itself holds for either sign. It is a wording inconsistency, not a wrong result.
Second check: another model agreed
Report a problem with this finding
Wording Appendix A equation (A1) prints z_i = 1/(1 − φπ)·((μ + φπ) + φαR_i + φε_i + η_i); the constant term does not follow from the model.
- p. 29 Appendix A, equation (A1)
((μ + φπ)
Only the authors can say what is right here.
Judge's note: Reworked from the printed system Y = μ0 + αR + πz + ε and z = μ1 + φY + η. Substituting gives z = μ1 + φμ0 + φαR + φπz + φε + η, so z(1 − φπ) = (μ1 + φμ0) + φαR + φε + η. The image shows '(μ + φπ)': it uses an undefined μ and the product φπ, which should be φμ0. The constant does not enter cov(z, ε) or the bias in (A2), so the stated downward-bias result stands.
Second check: another model agreed
Report a problem with this finding
Wording Appendix Table A1 describes Democracy as 'An 11-category scale, from 0 to 10' built from components '(from 1 to 3)', '(from 1 to 2, with a bonus of 1 point…)' and '(from 1 to 4)'; as printed, these components cannot sum to 0.
- p. 30 Appendix Table A1, Democracy
Competitiveness of Political Participation (from 1 to 3); Competitiveness of Executive Recruitment (from 1 to 2, with a bonus of 1 point if there is an election); and Constraints on Chief Executive (from 1 to 4)
Only the authors can say what is right here.
Judge's note: Image of A1 (p. 30) confirms the wording. If each component starts at 1, the minimum total is 3 and the maximum is 10, which conflicts with the stated 0–10 range and with the Table 1 means of 0.19 and 0. The ranges may be meant as point increments starting at 0, but the printed description is ambiguous.
Second check: another model agreed
Report a problem with this finding
Typo The author footnote gives the MIT ZIP code as "Cambridge, MA 02319" in both the Acemoglu and the Johnson addresses.
- p. 2 Author footnote
Cambridge, MA 02319
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: MIT's Cambridge, MA ZIP code is 02139. The crop of the page 2 footnote shows 02319 printed twice, so this is not an extraction artefact. It looks like a digit transposition.
Second check: none (judge only)
Report a problem with this finding
Typo Hayek's first name is printed "Frederich A. von Hayek (1960)" in the text and "von Hayek, Frederich A." in the reference list.
- p. 5 Introduction, right column
Frederich A. von Hayek (1960)
- p. 34 References
von Hayek, Frederich A.
- p. 5
Frederich A. von Hayek (1960) argued
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: the author of The Constitution of Liberty is Friedrich A. von Hayek. The page 5 and page 34 images both show "Frederich".
Second check: none (judge only)
Report a problem with this finding
Typo The text reads "Wil- liiam Easterly and Ross Levine (1997)", with a doubled 'i'; the reference list has "Easterly, William".
- p. 6 Introduction, right column
Wil- liiam Easterly
- p. 6 Introduction, right column
Wil- liiam Easterly and Ross Levine (1997)
Judge's note: The page 6 image shows "Wil-/liiam". The reference list entry spells the name William.
Second check: none (judge only)
Report a problem with this finding
Typo Section V.A cites 'see Weber [1976] or Andersen [1983]'; the reference list and p. 1373 give 'Anderson, Benedict' / 'Benedict Anderson, 1983'.
- p. 24 Section V.A
Andersen [1983]
- p. 31 References
Anderson, Benedict.
- p. 24 Section V.A, left column
see Weber [1976] or Andersen [1983]
- p. 6 Introduction
Benedict Anderson, 1983
Judge's note: The page image of p. 1391 shows 'Andersen [1983]'. The reference list (p. 31) has only 'Anderson, Benedict. Imagined communities. London: Verso, 1983.', and p. 1373 cites 'Benedict Anderson, 1983' for the same point. No 'Andersen' entry exists, so this is a misspelling of the author's name.
Second check: none (judge only)
Report a problem with this finding
Typo Footnote 11 reads "since Political and Risk Services construct these data"; the source is called "Political Risk Services" everywhere else.
- p. 11 Footnote 11
since Political and Risk Services construct these data
Judge's note: The footnote 11 crop on page 11 shows "Political and Risk Services". A text search finds "Political Risk Services" in the main text, the table notes and the reference list.
Second check: none (judge only)
Report a problem with this finding
Typo The parasite is spelled 'Plasmodium falcipo-rum', 'Plas-modium falciporum' and 'P. falciporum' (p.13), and 'falciporum malaria is endemic' (Appendix Table A1, p.30); the species name is falciparum.
- p. 13 Section III.A
Plasmodium falcipo- rum
- p. 30 Appendix Table A1
falciporum malaria is endemic
Judge's note: The page 13 image and the p.30 crop both show 'falciporum'. The text layer has a further 'falcipo-' at line 2771. This is a consistent misspelling of the species name.
Second check: none (judge only)
Report a problem with this finding
Typo 'some simple geographic or climac-tic feature' (p.15) should read 'climatic', and footnote 31's 'reject at that 10-percent level' (p.26) should read 'at the'.
- p. 15 Section III.A, end
geographic or climac- tic feature
- p. 13 Section III.A
Plasmodium falcipo- rum
- p. 26 Footnote 31
reject at that 10-percent level
- p. 26 footnote 31
reject at that 10-percent
Judge's note: The p.15 crop shows 'climac-tic' and the footnote 31 crop shows 'at that 10-percent level'. Both are copy-editing slips. The 'falciporum' part of this cluster is the same defect as C021 and is judged there.
Second check: none (judge only)
Report a problem with this finding
Typo The text has '(Jose A. Najera and Joahim Hempel, 1996)' and the reference list has 'Najera, Jose A. and Hempel, Joahim.'
- p. 14 Section III.A
Joahim Hempel, 1996
- p. 34 References
Hempel, Joahim.
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that the WHO author's given name is spelled 'Joachim'. Both crops show 'Joahim', spelled the same way in the text and the reference list.
Second check: none (judge only)
Report a problem with this finding
Typo Footnote 14 cites 'David McCullogh, 1977', but the reference list entry is 'McCullough, David. The path between the seas ... 1977.'
- p. 15 Footnote 14
David McCullogh, 1977
- p. 34 References
McCullough, David.
- p. 34 References
McCullough, David. The path between the
Judge's note: Footnote 14 (p.15 crop) shows 'McCullogh' and the reference list (p.34 crop) shows 'McCullough'. The two spellings in the article disagree.
Second check: none (judge only)
Report a problem with this finding
Typo '(fraction of the population of European decent)' on p.17 should read 'descent'.
- p. 17 Section IV.A, Panel B description
decent)
- p. 17 Section IV.A, Panel B discussion
European decent)
Judge's note: The p.17 crop shows 'decent).' The paper writes 'European descent' elsewhere, for example on p.11 and in the Table 3 notes.
Second check: none (judge only)
Report a problem with this finding
Typo Section V.A: 'This has little affect on our results.' The noun should be 'effect'.
- p. 21 Section V.A, first column
This has little affect on our results.
Judge's note: Confirmed on the page 21 image, first column, Section V.A.
Second check: none (judge only)