Table 1 Grid-level regression estimates (2016–2019)

From: A human-machine collaborative approach measures economic development using satellite imagery

 

(1)

(2)

(3)

(4)

Δln(siScore)

Δln(NL)

\({\mathbb{1}}\) {ΔsiScore > 0}

\({\mathbb{1}}\) {ΔNL > 0}

Proximity to economic and political hubs

  Log distance to NK-China and Russia border

0.089 (0.071)

0.058a (0.028)

0.002 (0.031)

−0.055 (0.043)

  Log distance to Pyongyang

−0.256c (0.069)

0.175 (0.098)

−0.142c (0.040)

0.223 (0.147)

  Log distance to nearest city

−0.105b (0.033)

0.083b (0.032)

−0.034a (0.017)

−0.090b (0.037)

  Log distance to nearest major port

0.132b (0.048)

0.022 (0.024)

0.044b (0.015)

−0.018 (0.054)

Economic Development Zone (EDZ)

  Agriculture development

0.243c (0.041)

0.070 (0.081)

0.236c (0.022)

0.307b (0.118)

  Tourism development

0.297c (0.083)

0.391 (0.281)

0.182c (0.032)

0.643b (0.198)

  Industrial development

−0.058 (0.128)

−0.083 (0.165)

−0.005 (0.130)

0.153a (0.077)

  Export processing

−0.063 (0.108)

−0.378c (0.047)

−0.038 (0.143)

0.199 (0.127)

Mining site of key minerals

  Gold mine

0.083b (0.031)

−0.024 (0.028)

0.005 (0.019)

0.020 (0.038)

  Coal mine

0.074a (0.034)

0.003 (0.024)

0.039a (0.022)

0.117b (0.049)

  Copper mine

0.090 (0.050)

0.076a (0.040)

−0.000 (0.030)

−0.018 (0.049)

  Iron mine

0.116 (0.072)

−0.126a (0.059)

0.113b (0.037)

−0.042 (0.042)

Nuclear-related site

  Nuclear test site

−0.033 (0.087)

−0.098a (0.046)

0.048 (0.047)

−0.011 (0.113)

  Uranium mine

0.375b (0.156)

0.077 (0.072)

0.197c (0.042)

0.110 (0.078)

  Province FE

Yes

Yes

Yes

Yes

  Mean of outcome variable

−0.09

3.10

0.45

0.30

  Observations

32,578

32,578

32,578

32,578

  1. Table 1 reports ordinary least squares (OLS) regression estimates, which were tested using two-sided tests and there was no additional adjustment made for any of the regressions. The outcome variable in columns (1) and (2) is the difference of logarithmized values between 2016 and 2019. Columns (3) and (4) use an indicator for positive change as the outcome variable. Logistic regression gives similar results. All specifications include province fixed effects, log of district population in 2008, and log of district area. Standard errors are clustered at province level and reported in parentheses.
  2. adenotes statistical significance at 0.10,
  3. bat 0.05, and
  4. cat 0.01.