1 资料与方法
1.1 数据来源与样本选择
表1 交易数量(观察)的描述性统计Table 1 Descriptive statistics of the number of transactions (observations) |
| Vaccine | Other | Other pool-procurement | PAHO | Self-procurement | UNICEF | Total |
| BCG | 18 | 3 | 176 | 385 | 683 | 1 265 |
| Combination vaccines | 147 | 45 | 1 084 | 2 386 | 2 427 | 6 089 |
| HPV | 21 | 5 | 136 | 258 | 221 | 641 |
| Hepatitis B | 37 | 12 | 313 | 713 | 451 | 1 526 |
| Hib | 15 | 3 | 38 | 192 | 10 | 258 |
| Measles | 2 | 0 | 0 | 74 | 253 | 329 |
| Pneumococcal conjugate | 36 | 4 | 124 | 415 | 596 | 1 175 |
| Polio | 29 | 15 | 387 | 464 | 1 542 | 2 437 |
| Rotavirus | 20 | 15 | 102 | 189 | 438 | 764 |
| Rubella | 1 | 0 | 0 | 12 | 0 | 13 |
| Total | 326 | 102 | 2 360 | 5 088 | 6 621 | 14 497 |
UNICEF, United Nations Children ’ s Fund; PAHO, Pan American Health Organization; BCG, Bacillus Calmette-Guérin vaccine; Combination vaccines, include DTP-containing vaccines and other multivalent products; DTP, diphtheria, tetanus, and pertussis-containing vaccine; HPV, human papillomavirus vaccine; Hib, Haemophilus influenzae type b vaccine; Polio, poliovirus vaccine; Rotavirus, rotavirus vaccine; Rubella, rubella vaccine. |
1.2 变量选取
表2 关键变量汇总统计Table 2 Summary statistics of key variables |
| Items | Observations | Mean | SD | Minimum | Maximum |
| Price per dose (USD) | 14 497 | 7.72 | 16.83 | 0.03 | 240.30 |
| Transaction volume (doses) | 14 497 | 1 578 766.53 | 8 827 643.06 | 1.00 | 5.56×108 |
| HHI (market concentration) | 14 497 | 0.00 | 0.27 | -0.45 | 0.55 |
| Procurement: self-procurement (dummy) | 14 497 | 0.35 | 0.48 | 0.00 | 1.00 |
| Procurement: UNICEF (dummy) | 14 497 | 0.46 | 0.50 | 0.00 | 1.00 |
| Procurement: PAHO (dummy) | 14 497 | 0.16 | 0.37 | 0.00 | 1.00 |
| Procurement: other pool-procurement (dummy) | 14 497 | 0.03 | 0.17 | 0.00 | 1.00 |
| WHO PQ certified (yes=1) | 14 497 | 0.73 | 0.44 | 0.00 | 1.00 |
| GAVI eligible country (yes=1) | 14 497 | 0.37 | 0.48 | 0.00 | 1.00 |
| Income: high (dummy) | 14 497 | 0.26 | 0.44 | 0.00 | 1.00 |
| Income: upper-middle (dummy) | 14 497 | 0.29 | 0.45 | 0.00 | 1.00 |
| Income: lower-middle (dummy) | 14 497 | 0.29 | 0.46 | 0.00 | 1.00 |
| Income: low (dummy) | 14 497 | 0.15 | 0.36 | 0.00 | 1.00 |
The unit of observation is a vaccine procurement transaction. HHI (market concentration) refers to the Herfindahl-Hirschman index, which has been mean-centered (i.e., subtracted by its sample mean, resulting in a mean of 0.00). For binary variables (dummies), the “Mean” represents the proportion of the sample (e.g., a mean of 0.35 indicates 35% of the observations fall into this category). SD, standard deviation; USD, United States dollar; UNICEF, United Nations Children ’ s Fund; PAHO, Pan American Health Organization; WHO PQ, World Health Organization Prequalification; GAVI, Global Alliance for Vaccines and Immunization. |
1.3 实证模型
1.3.1 关键变量的精细化构建
1.3.2 基准计量模型
1.3.3 事件研究模型
1.3.4 异质性分析
2 结果
2.1 基准回归结果
表3 采购机制和疫苗价格的基准回归结果Table 3 Baseline regression results of procurement mechanisms and vaccine prices |
| Items | Model 1 ln(price) | Model 2 ln(price) | Model 3 ln(price) | Model 4 ln(price) | Model 5 ln(price) |
| UNICEF SD, mean (SE) | -0.478 (0.082)*** | -0.487 (0.082)*** | -0.323 (0.074)*** | -0.285 (0.077)*** | -0.265 (0.075)*** |
| PAHO, mean (SE) | -0.528 (0.167)*** | -0.557 (0.168)*** | -0.370 (0.189)* | -0.341 (0.231) | -0.280 (0.236) |
| Other pool-procurement, mean (SE) | -0.027 (0.136) | -0.042 (0.127) | 0.207 (0.136) | 0.486 (0.133)*** | 0.542 (0.136)*** |
| Other, mean (SE) | 0.168 (0.083)** | 0.137 (0.084) | 0.109 (0.086) | 0.005 (0.112) | 0.026 (0.108) |
| ln(volume), mean (SE) | -0.061 (0.008)*** | -0.027 (0.007)*** | -0.021 (0.007)*** | -0.018 (0.007)** | |
| WHO PQ certified, mean (SE) | -0.024 (0.057) | -0.025 (0.056) | -0.065 (0.028)** | -0.037 (0.026) | -0.048 (0.045) |
| Country FE | Yes | Yes | Absorbed | Absorbed | Absorbed |
| Product FE | Yes | Yes | Absorbed | Absorbed | Absorbed |
| Year FE | Yes | Yes | Yes | Absorbed | Absorbed |
| Country-product FE | No | No | Yes | Yes | Yes |
| Product-year FE | No | No | No | Yes | Yes |
| Country-year FE | No | No | No | Yes | Yes |
| Manufacturer FE | No | No | No | No | Yes |
| Observations | 14 269 | 14 269 | 13 081 | 12 675 | 12 439 |
| R-squared | 0.929 | 0.930 | 0.969 | 0.982 | 0.982 |
* P < 0.10, * * P < 0.05, * * * P < 0.01. Robust standard errors, two-way clustered at the country and product levels, are reported in parentheses. The dependent variable is the natural logarithm of the vaccine price per dose. The baseline reference category for procurement mechanisms is self-procurement. Additional controls included but not shown are GAVI status and income group. “Absorbed” indicates that the single-dimensional fixed effects (FE) are perfectly absorbed by the inclusion of higher-dimensional interactive FE. UNICEF SD, United Nations Children ’ s Fund Supply Division; PAHO, Pan American Health Organization; WHO PQ, World Health Organization Prequalification; GAVI, Global Alliance for Vaccines and Immunization; SE, standard error. |
2.2 动态效应与平行趋势
2.3 异质性分析
表4 买方异质性分析Table 4 Buyer heterogeneity analysis |
| Items | Baseline | Buyer size | GAVI eligibility | Income group | |||||||
| Small buyer | Large buyer | GAVI | Non-GAVI | LIC | LMIC | UMIC | HIC | ||||
| Other, mean (SE) | 0.109 (0.086) | 0.040 (0.063) | 0.116 (0.125) | 0.060 (0.099) | 0.124 (0.098) | -0.528 (0.300)* | 0.025 (0.107) | 0.335 (0.221) | 0.063 (0.055) | ||
| Other pool-procurement, mean (SE) | 0.207 (0.136) | 0.343 (0.182)* | 0.304 (0.222) | 0.453 (0.187)** | 0.201 (0.139) | 0.252 (0.176) | 0.203 (0.136) | ||||
| PAHO, mean (SE) | -0.370 (0.189)* | -0.379 (0.276) | -0.340 (0.190)* | -0.134 (0.061)** | -0.616 (0.351)* | -0.268 (0.292) | -0.256 (0.118)** | -0.185 (0.220) | -1.222 (0.267)*** | ||
| UNICEF SD, mean (SE) | -0.323 (0.074)*** | -0.487 (0.130)*** | -0.255 (0.072)*** | -0.118 (0.048)** | -0.418 (0.093)*** | -0.311 (0.289) | -0.288 (0.108)*** | -0.386 (0.086)*** | -0.475 (0.073)*** | ||
| ln(volume), mean (SE) | -0.027 (0.007)*** | -0.052 (0.014)*** | -0.010 (0.009) | -0.002 (0.005) | -0.038 (0.011)*** | 0.002 (0.007) | -0.017 (0.008)** | -0.040 (0.016)** | -0.036 (0.017)** | ||
| n | 13 081 | 5 681 | 6 959 | 4 987 | 8 033 | 1 963 | 3 847 | 3 565 | 3 336 | ||
* P < 0.10, * * P < 0.05, * * * P < 0.01. Robust standard errors, two-way clustered at the country and product levels, are reported in parentheses. The dependent variable is the natural logarithm of the vaccine price per dose. All regression models include country-product and year fixed effects. The baseline reference category for procurement mechanisms is self-procurement. Additional controls included but not shown are WHO PQ certified, GAVI status, and Income group. LIC, low-income country; LMIC, lower-middle-income country; UMIC, upper-middle-income country; HIC, high-income country; UNICEF SD, United Nations Children ’ s Fund Supply Division; PAHO, Pan American Health Organization; WHO PQ, World Health Organization Prequalification; GAVI, Global Alliance for Vaccines and Immunization; SE, standard error. |
表5 卖方异质性分析Table 5 Seller heterogeneity analysis |
| Baseline | HHI interaction | PQ interaction | |
| UNICEF SD, mean (SE) | -0.323 (0.074)*** | -0.317 (0.073)*** | -0.381 (0.104)*** |
| PAHO, mean (SE) | -0.370 (0.189)* | -0.357 (0.189)* | -0.376 (0.179)** |
| Other pool-procurement, mean (SE) | 0.207 (0.136) | 0.202 (0.138) | 0.459 (0.200)** |
| Other, mean (SE) | 0.109 (0.086) | 0.124 (0.092) | 0.067 (0.076) |
| ln(volume), mean (SE) | -0.027 (0.007)*** | -0.027 (0.007)*** | -0.027 (0.007)*** |
| WHO PQ certified, mean (SE) | -0.065 (0.028)** | -0.064 (0.028)** | -0.098 (0.045)** |
| HHI (centered), mean (SE) | -0.038 (0.074) | ||
| UNICEF×HHI, mean (SE) | 0.095 (0.140) | ||
| UNICEF×PQ certified, mean (SE) | 0.069 (0.074) | ||
| Country-product FE | Yes | Yes | Yes |
| Year FE | Yes | Yes | Yes |
| Observations | 13 081 | 13 081 | 13 081 |
| R-squared | 0.969 | 0.969 | 0.969 |
* P < 0.10, * * P < 0.05, * * * P < 0.01. Robust standard errors, two-way clustered at the country and product levels, are reported in parentheses. The dependent variable is the natural logarithm of the vaccine price per dose. The baseline reference category for procurement mechanisms is self-procurement. Additional controls included but not shown are GAVI status and income group. HHI (centered) indicates that the Herfindahl-Hirschman index has been mean-centered prior to forming the interaction term. This allows the main effect coefficient of UNICEF SD to be interpreted directly as the effect at the sample average level of market concentration. UNICEF SD, United Nations Children ’ s Fund Supply Division; PAHO, Pan American Health Organization; WHO PQ, World Health Organization Prequalification; GAVI, Global Alliance for Vaccines and Immunization; FE, fixed effects; SE, standard error. |
