TABLE 3.
The impact of age structure on COVID-19 total cases on August 2020.
| Independent variables | GLM1,2 | GMM3,4,5 | Quantile (25)6,7,8 |
| Median age | 91.91 (10.58) [8.68] {0.00} | 84.47 (12.21) [6.91] {0.00} | 10.33 (4.94) [2.09] {0.03} |
| Quantile (50)6,7,8 | |||
| 36.01 (5.78) [6.22] {0.00} | |||
| Quantile (75)6,7,8 | |||
| 112.7 (15.49) [7.27] {0.00} | |||
| Aged-65_older | 222.6 (34.15) [6.51] {0.00} | 301.7 (49.39) [6.10] {0.00} | 40.46 (15.30) [2.64] {0.00} |
| Quantile (50)6,7,8 | |||
| 120.5 (19.32) [6.23] {0.00} | |||
| Quantile (75)6,7,8 | |||
| 335.31 (53.97) [6.21] {0.00} | |||
| Aged-70_older | 333.38 (52.65) [6.33] {0.00} | 465.6 (78.30) [5.94] {0.00} | 48.05 (24.5) [1.96] {0.05} |
| Quantile (50)6,7,8 | |||
| 171.4 (28.9) [5.92] {0.00} | |||
| Quantile (75)6,7,8 | |||
| 541.9 (83.3) [6.50] {0.00} |
1By Newton-Raphson-Marquardt steps. 2The coefficient covariance is computed using observed Hessian. 3Estimation weighting matrix: HAC (Bartlett kernel, Newey-West fixed bandwidth = 5.0000). 4Standard errors and covariance are computed following the estimation weighting matrix. 5Instrument specification: total-cases-per-million (–1). 6Sparsity methods: Kernel (Epanechnikov) using residuals. 7Bandwidth methods: Hall-Sheather. 8Estimation successfully identifies the unique optimal solution. () shows standard errors, and {} denotes probability values.