Table 5.
Summary of estimates from robustness models
| Variables of interest | Main model | Sensitivity model: limit comparison countries to West Africa only | Sensitivity model: propensity score weighting of country groups | Falsification model: pseudotreatment groups | Placebo model: pseudooutbreak year | Permutation test: randomized treatment time |
|---|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | (6) | |
| Treatment Group 1 × 2014 | −90.38*** (0.00) | −119.83* (0.09) | −107.89* (0.09) | −30.87 (0.46) | 45.53 (0.16) | −90.38* (0.08) |
| Treatment Group 1 × Post‐2014 | 49.05 (0.37) | −8.99 (0.92) | 61.28 (0.38) | 144.66** (0.04) | −48.03 (0.31) | 49.05 (0.52) |
| Treatment Group 2 × 2014 | −269.50** (0.02) | −235.86* (0.06) | −143.57* (0.09) | −107.75 (0.18) | 574.55 (0.16) | −269.50 (0.74) |
| Treatment Group 2 × Post‐2014 | −1447.67 (0.20) | −1290.18 (0.28) | −168.82 (0.61) | −256.57 (0.12) | 343.88 (0.29) | −1447.67** (0.02) |
| N | 546 | 204 | 546 | 474 | 546 | 546 |
Note. (1) Main model. Treatment Group 1 consists of three countries with epidemic transmission of 2014 Ebola: Guinea, Liberia, and Sierra Leone. Treatment Group 2 consists of three countries with limited cases of 2014 Ebola: Mali, Nigeria, and Senegal. The comparison group consists of 40 sub‐Saharan Africa countries without Ebola cases. (2) Sensitivity model: Treatment groups are the same as in main model (1); comparison group is limited to 11 unaffected countries in the same SSA subregion as Ebola‐affected countries (Benin, Burkina Faso, Cape Verde, Gambia, Ghana, Guinea‐Bissau, Ivory Coast, Mauritania, Niger, São Tomé and Príncipe, and Togo). (3) Sensitivity model: Treatment and comparison countries are the same as in main model (1); observations in each group are weighted using propensity scores. (4) Falsification model: Pseudotreatment Group 1 consists of three countries identified as being at high risk for 2014 Ebola transmission but experiencing no Ebola cases (Gambia, Guinea‐Bissau, and Ivory Coast). Pseudotreatment Group 2 consists of three countries in geographic proximity to Ebola‐affected countries but experiencing no Ebola cases (Benin, Burkina Faso, and Ghana). The comparison group consists of 34 remaining sub‐Saharan Africa countries without Ebola cases. (5) Placebo model: Treatment and comparison countries are same as in main model (1); pseudooutbreak year set to 2011. (6) Permutation test: Treatment and comparison countries are the same as in main model (1); outcome values are randomized within countries over multiple iterations without specifying a single pseudotreatment year. All estimates obtained from linear models with clustering by country and year. All models include controls for country‐specific time trends, U.S. currency value index, partner country gross domestic product per capita relative to U.S., and partner consumer price index relative to U.S. (not shown). U.S. exports in constant 2016 USD (millions). p values in parentheses.
Statistically significant at 10% level.
Statistically significant at 5% level.
Statistically significant at 1% level.