Table 7 Forward-looking content of satellite imagery data during COVID-19 pandemic.

From: Eye in outer space: satellite imageries of container ports can predict world stock returns

 

h = 1

h = 2

h = 3

h = 4

Panel A: Predictive regression for ∆RWI_ISLt+h

∆GNCt

0.005 (0.031)

−0.311* (−1.911)

0.105 (0.652)

−0.119 (−0.727)

∆GNCt × COVIDt

−0.029 (−0.099)

0.163 (0.579)

0.003 (0.012)

0.072 (0.250)

COVIDt

0.099 (0.354)

0.382 (1.389)

0.306 (1.113)

0.456 (1.613)

∆RWI_ISLt

−0.163 (−1.181)

0.057 (0.422)

0.297** (2.225)

−0.248* (−1.821)

Adjusted R2

2.90%

10.39%

13.55%

11.79%

Panel B: Predictive regression for ∆BDIt+h

∆GNCt

−0.220 (−1.373)

−0.319** (−1.991)

0.192 (1.189)

0.104 (0.621)

∆GNCt × COVIDt

0.182 (0.646)

−0.095 (−0.337)

0.025 (0.087)

0.009 (0.030)

COVIDt

0.391 (1.461)

0.552** (2.042)

0.471* (1.703)

0.513* (1.787)

∆BDIt

0.093 (0.703)

−0.053 (−0.399)

0.148 (1.091)

−0.150 (−1.074)

Adjusted R2

12.24%

13.35%

12.47%

9.14%

Panel C: Predictive regression for ∆AIPt+h

∆GNCt

0.062 (0.362)

−0.125 (−0.747)

−0.036 (−0.209)

−0.244*** (−2.975)

∆GNCt × COVIDt

−0.012 (−0.040)

0.396 (1.341)

0.091 (0.301)

−0.510* (−1.689)

COVIDt

−0.112 (−0.388)

−0.082 (−0.285)

0.008 (0.028)

0.181 (0.579)

∆IPt

0.037 (0.260)

−0.171 (−1.230)

−0.178 (−1.251)

−0.066 (−0.531)

Adjusted R2

0.78%

5.68%

3.20%

9.79%

  1. This table investigates the predictive ability of global ports (GNC) to global container throughput (RWI), the decrease in freight rate (BDI), and the growth rate of averaged real industry production (AIP). To examine the change of predictive relationship around COVID-19, we add a dummy variable that takes the value of one on the COVID-19 period, defined as the time after 2020. The sample period covers from 2017:01 to 2021:11. We report t-statistics based on heteroscedasticity robust standard errors in parentheses. The asterisks ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.
Back to article page