Democracy Does Cause Growth

• p. 38 Appendix A6, FH components paragraph

  GDP per capita by 6.65% (p−value 0.159) in the long run

• p. 63 Table A8, column 6

  6.70

• p. 38 A6

  increases GDP per capita by 6.65% (p−value 0.159) in the long run

• p. 38 A6

  increases it by 10.25% (p−value 0.049)

Only the authors can say what is right here.

Judge's note: I checked both page images. The text and the table describe the same within specifications (political rights, civil liberties, both together), and no alternative specification is named. The differences are small, but both the point estimates and the p-values differ, so this is not rounding; the text may come from an earlier run. Recomputing from the rounded table coefficients gives 6.73 and 10.96, close to the table values. Only the authors can say which set is final.

Second check: another model agreed

• p. 48

  Papaioannou and Siourounis (2008), Cheibub, Gandhi, and Vreeland (2010)

Judge's note: The Figure A1 note's '(2008)' matches the reference list entry 'Papaioannou, Elias and Gregorios Siourounis (2008)', so the quoted text is correct. The inconsistent year is the '(2006)' in Appendix A1 on p.32 (body and fn 32), which I confirmed on the page image. No 2006 entry exists in the reference list.

Second check: another model agreed

• p. 53 Table 5 notes

  Panel B presents the corresponding first stage estimates and the excluded instruments F statistic.

Judge's note: On the full page image (p.53), the 'Exc. Instruments F-stat.' row (119.1, 33.2, ...) is the last row of Panel A, above the 'Panel B: First-stage regression' heading. Panel B has only the four democracy-wave coefficients. The note points to the wrong panel. The data are not affected.

Second check: another model agreed

• p. 4 Introduction, last paragraph

  The figure plots (log) GDP per capita in countries that democratize relative to continuing nondemocracies

• p. 42 Figure 1 notes

  We normalize the average log GDP per capita in a country to zero.

Only the authors can say what is right here.

Judge's note: I compared p.4 with the Figure 1 page (p.42). The figure note mentions no comparison group and no removal of year effects, so the two descriptions of what is plotted differ. Only the authors can say whether the series is net of continuing nondemocracies.

Second check: another model agreed

• p. 5 Introduction, second paragraph

  The lack of a significant downward trend before year 0 in this figure

• p. 5 Introduction, second paragraph

  there is a significant and clearly-visible increase in GDP per capita

• p. 43 Figure 2

  FIGURE 2: GDP PER CAPITA IN A DEMOCRATIZATION IN REWEIGHTED DATA.

Only the authors can say what is right here.

Judge's note: Figure 2 (p.43) has a single line and its notes give no inference. The text attributes significance to 'this figure', which cannot show it. The inference presumably comes from Figure 3 or the tables, which this sentence does not cite.

Second check: another model agreed

• p. 9 Section 3

  and infant mortality data, all from the World Devel-

• p. 23 Section 7

  log child mortality

Only the authors can say what is right here.

Judge's note: I checked p.9 and Table 1 (p.49); Table 6 text is at line 3102. Infant mortality and child mortality are different indicators, and the data section never mentions child mortality. Only the authors can say which series was used.

Second check: another model agreed

• p. 10 Section 4.1

  different from zero at the 1.1% confidence level

Judge's note: The page image shows the phrase. Table 2 column 1 prints the long-run effect p-value as [0.011]. A 1.1% significance level corresponds to a 98.9% confidence level, so the term is misused.

Second check: another model agreed

• p. 12 Section 4.1, AR2 discussion

  recall the significance of the third and the fourth lags in columns 3 and 4

• p. 50 Table 2, cols 3-4, third and fourth lag rows

  -0.026

• p. 50 Table 2, cols 3-4

  log GDP third lag -0.026 (0.028)

• p. 12

  recall the significance of the third and the fourth lags in columns

Only the authors can say what is right here.

Judge's note: I read Table 2 on the upright page 50 image. Column 3: third lag -0.026 (0.028), fourth lag -0.043 (0.017). Column 4: third lag -0.021 (0.028), fourth lag -0.039 (0.034). Of the four coefficients the sentence points to, only one has |t| above 2. No joint test of lags 3-4 is reported that would support another reading. The text on p.11 (PDF) about column 4 says the additional lags 5-8 are not significant, which is a separate statement and does not contradict this one, so reviewer note 5 adds nothing. Only the authors can say what they meant, so this goes to them as a query.

Second check: another model agreed

• p. 13 Section 4.1, first paragraph

  they are similar to the within estimates, though slightly larger

• p. 50 Table 2, cols 4 and 12

  34.73

Only the authors can say what is right here.

Judge's note: I checked the values in Table 2 (upright page 50). Column 12 is about 58% above column 4 on the long-run effect and about 92% above it on the coefficient, which is hard to call 'slightly larger'. The example the text gives (column 11) does support the claim.

Second check: another model agreed

• p. 14 Section 4.2, last paragraph

  without countries with a standarized residual above 1.96 or below -1.96

• p. 60 Table A5 notes and countries row

  Column 2 removes countries with a a standardized

• p. 60 Table A5 notes

  with a a standardized

• p. 60 Table A5 notes

  Column 2 removes countries with a a standardized residual

Judge's note: Table A5 (p.60) shows 175 countries in every column, so country-year observations were dropped, not whole countries, which matches A3 on p.35 ('we remove points'). The doubled 'a a' appears in the Table A5 notes and twice in the Table A9 notes (p.64), alongside "Cooks'".

Second check: another model agreed

• p. 22 Section 6.2, HHK paragraph

  significant at 5% or less in all other columns, except in column 5

• p. 53 Table 5, Panel C long-run p-value row

  [0.070]

• p. 53 Table 5 Panel C

  [0.425]

Only the authors can say what is right here.

Judge's note: On p.22, the sentence before the claim treats "the coefficient estimate and the long-run effect" together (7% and 13% for column 3). That matches the column 3 long-run p [0.130], so the significance claim appears to cover both quantities. On p.53, Panel C long-run p-values are col4 0.070, col5 0.253, col6 0.011, col7 0.047, col8 0.080, col9 0.062. Only columns 6 and 7 are at or below 5%. The coefficient z-ratios are col4 2.36, col8 1.95 and col9 2.09. The authors should say which quantity the 5% claim refers to.

Second check: another model agreed

• p. 26 Section 9 Conclusion, third paragraph

  these effects are not large enough to lead to negative effects of democracy for low human capital countries

• p. 55 Table 7, Panel B, columns 5-7

  -0.119

• p. 55 Table 7, Panel B, columns 5-7, long-run effect

  -14.59

Judge's note: Table 7 Panel B on p.55, columns 5-7: main effects -0.119, -0.484, -0.474; long-run effects -3.65, -14.59, -14.14 with p 0.855, 0.443, 0.435. Section 8 (text-layer line 1075) states the claim with "significant" ("any significant negative effect"), and the Conclusion on p.26 drops that qualifier. This is a wording point only.

Second check: another model agreed

• p. 33 Appendix A1 item 5 and footnote 35

  We have 174 observations for 16 countries that are only covered by CGV and BMR

Only the authors can say what is right here.

Judge's note: I counted the countries in footnote 35 on the p.33 image: Antigua and Barbuda, Barbados, Germany, Iceland, Luxembourg, Kuwait, Maldives, Malta, Nauru, Syria, Tonga, Vietnam, Yemen, Samoa, Zimbabwe, giving 15. The footnote may not be meant to be exhaustive, so the authors should say which country is missing or whether the number is wrong.

Second check: another model agreed

• p. 36 A4

  long-run effect on democracy

Judge's note: The image shows the sentence as quoted. The quantity under discussion is the long-run effect of democracy on GDP, so 'on democracy' misdescribes it. The fix is evident.

Second check: another model agreed

• p. 55 Table 7 notes

  We report the estimated persistence of the GDP process and the p−value for this being less than 1.

Judge's note: On the full page image (p.55), Panels A, B and C contain only Democracy, Interaction, Long-run effect, p-value, Observations and Countries (Panel B also has the F-stat and Hansen rows). There is no persistence row, so the note sentence looks like boilerplate that does not describe this table.

Second check: another model agreed

• p. 64 Table A9 notes

  Column 2 removes countries with a standardized residual estimated above 1.96

• p. 64 Table A9 notes

  with a a standardized residual

• p. 64 Table A9 notes

  with a a Cooks’ distance

Judge's note: Table A9 (p.64): column 1 has 6309 observations and 174 countries, column 2 has 6015 and 174. The appendix text (main.txt line 1608) describes removing observations, not countries, so the note misdescribes what is dropped. The Table A5 note has the same 'removes countries' wording while its text says 'points' and the country count stays at 175. The 'a a' part of this cluster duplicates C066 and is not counted again here.

Second check: another model agreed

• p. 3 Footnote 3

  we will sometimes described this as

• p. 3 footnote 3

  we will sometimes described this

• p. 21 footnote 24

  We do not not present

• p. 17 Section 5

  bootstraping

Judge's note: I checked every item in the crop images. They all appear as quoted. The fixes are 'describe', 'do not', 'bootstrapping', 'Soviet bloc', 'Papaioannou' and 'democratic'. These are cosmetic.

Second check: none (judge only)

• p. 7 Section 2

  Tabellini and Giavazzi (2005)

• p. 29 References

  Giavazzi, Francesco and Guido Tabellini (2005)

• p. 12 footnote 12

  Arellano and Alvarez (2003)

Judge's note: p.7 and the fn 12 crop show the reversed order. The reference list (p.27, p.29) has 'Alvarez, Javier and Manuel Arellano (2003)' and 'Giavazzi, Francesco and Guido Tabellini (2005)'. Footnote 32 and the text at line 451 use the correct order.

Second check: none (judge only)

• p. 7 Footnote 5

  and Kudamatsu (2012) investigates

Judge's note: Seen on the p.7 page image. The verb should be 'investigate'.

Second check: none (judge only)

• p. 8 Section 3

  Cheibub, Ghandi and Vreeland

Judge's note: I checked the crop image. The reference entry (line 1198) and p.15 use 'Gandhi'.

Second check: none (judge only)

• p. 12 Footnote 12

  Arellano and Alvarez (2003)

• p. 12 footnote 12

  As shown in Arellano and Alvarez (2003) and Hayakawa (2008)

• p. 27 References

  Alvarez, Javier and Manuel Arellano (2003)

Judge's note: The footnote on the page 12 image reverses the author order. The reference entry on p.27 and the text on p.11 (PDF) use Alvarez and Arellano. It is the same work, so this is a minor citation inconsistency.

Second check: none (judge only)

• p. 13 Footnote 14

  In particular, as shown by Hahn, Hausman and Kuersteiner (2002).

Judge's note: I confirmed on the page 13 image that this stands as a complete sentence with no main clause.

Second check: none (judge only)

• p. 14 Section 4.2

  standarized residual

• p. 14 Section 4.2

  these Soviets block countries

• p. 33 Appendix A1 item 3

  democractic

• p. 34 Appendix A2

  country and year fixed affects

Judge's note: I saw all four on the page images and crops.

Second check: none (judge only)

• p. 15 Section 4.2

  Pappaioannou and

Judge's note: The crop shows 'Pappaioannou'. Everywhere else, including the reference list, the name is 'Papaioannou'.

Second check: none (judge only)

• p. 16 Section 5

  Hirano, Imbens and Rider’s (2003)

Judge's note: I checked the crop and the reference entry.

Second check: none (judge only)

• p. 17 Section 5

  obtained by bootstraping

Judge's note: I confirmed the misspelling in the crop.

Second check: none (judge only)

• p. 19 Section 6.1

  Latin American and the Caribbean

Judge's note: Checked the crop of p.19: the text reads "Latin American and the Caribbean". This is a simple misspelling of the region name.

Second check: none (judge only)

• p. 20 Equation (4), second line

  Dct = Pq j=1 πqZct−q

Judge's note: The page 20 image shows the first-stage line with the sum over j=1..q and the summand π_q Z_{ct−q}. The GDP-lag sum on the same line correctly uses φ_j y_{ct−j}. The intended form is clear from the text ("instrumented by the lags of Z_ct"), so this is an index typo with an evident fix.

Second check: none (judge only)

• p. 21 Footnote 24

  We do not not present results

• p. 21

  with efficient weights. 24We do not not present results fro

Judge's note: The crop of p.21 shows the duplicated "not not" in footnote 24.

Second check: code check

• p. 23 Section 7

  (using the same estimator as in Section 5. The variables

• p. 23 Section 7, paragraph after eq. (5)

  (using the same estimator as in Section 5.

Judge's note: The crop of p.23 confirms that no closing parenthesis follows "Section 5".

Second check: none (judge only)

• p. 27 References

  Aghion, Phillipe, Alberto Alesina, and Francisco Trebbi (2008)

• p. 27 References

  Aghion, Phillipe

• p. 29 References

  Economic Letters

• p. 31 References

  Cambridge Universiy Press

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: that the correct forms are Philippe, Economics Letters and Thousand Oaks. "Universiy" is an evident misspelling of "University". All four strings were checked on the crops from pp.27, 29, 30 and 31.

Second check: none (judge only)

• p. 31 References

  Reynal-Querol, Marta.(2005 “Does

Judge's note: The crop of p.31 confirms that "(2005" is never closed.

Second check: none (judge only)

• p. 42 Figure 1 title

  GDP PER CAPITA ARROUND A DEMOCRATIZATION

• p. 42 Figure 1 title

  FIGURE 1: GDP PER CAPITA ARROUND A DEMOCRATIZATION.

• p. 42 Figure 1 title

  ARROUND

• p. 48 Figure A1 notes

  geograohic

Judge's note: Each item is verified on its image. 'AD-DITIONAL' in the Table A11 title is just end-of-line hyphenation, not a typo, so I excluded it. 'fixed affects' is handled separately in C046.

Second check: none (judge only)

• p. 34 A2 Specification Tests

  country and year fixed affects

Judge's note: The page image shows 'fixed affects' in the first paragraph of A2. It is a plain misspelling of 'effects'.

Second check: none (judge only)

• p. 39 Appendix A7

  which is close or to our baseline within estimate

Judge's note: The image shows 'close or to'. The word 'or' is extraneous.

Second check: none (judge only)

• p. 48 Figure A1 panel title

  Middle East and North of Africa

Judge's note: The p.48 image shows 'North of Africa'. The main text (p.~20, line 818 of the extract) uses 'Middle East and North Africa'. The difference is cosmetic.

Second check: none (judge only)

• p. 56 Table A1 notes

  because of the availability of sufficient GDP data

• p. 57 Table A2 notes

  because of the availability of sufficient GDP

• p. 56 Table A1 notes

  not available because of the availability of sufficient GDP data

• p. 57 Table A2 notes

  because of the availability of sufficient GDP data

Judge's note: Both crops confirm the wording in the notes to Table A1 (p.56) and Table A2 (p.57). A negating word was evidently dropped, so the sentence says the opposite of what is meant, though the intent is clear.

Second check: none (judge only)

• p. 60 Table A5 notes

  with a a standardized

• p. 64 Table A9 notes

  countries with a a Cooks’ distance

Judge's note: Seen on the Table A5 crop (p.60) and the full page image of Table A9 (p.64), where 'a a' appears twice. Table A9 also has 'a estimated' and 'Cooks’'.

Second check: none (judge only)