사회과학연구(SSK)지원사업_기후변화연구소_5

 "Nowcasting Economic and Socio Variables with Disasters"

 김현학(경제학과 교수 / 국민대학교)

Abstract
In this paper, we utilize the remote sensing data to estimate the employment of Bangladesh.
Since the labor force survey in Bangladesh had been done occasionally, total employment
was measured statistically based on scarce survey data. To exploit the real-time availability
and regionally subdivided characteristic of remote sensing data, we construct the annual
employment by job quality by 64 districts using interpolation. With interpolated employment
data by job quality which shows how many employments are in good quality such as
employer or regular-paid job, we construct the forecasting model for annual employment
data with mixed frequency control variables in quarterly frequency, as well as remote sensing
data. Then we extend the model for estimating quarterly employment status using monthly
series which includes the remote sensing data such as nighttime light and enhanced
vegetation index. Since there is no true observation in our target variable, there is no way to
verify the accuracy of the forecasting model. Instead, we extend our approach to do out-ofsample
forecast since we have the real-time data up to the first half of 2022. In result, remote
sensing data may not capture the overall trend of employment because control variables such
as industrial production dominates. However, remote sensing data shows idiosyncratic
movement by districts. Therefore, our result is helpful for policymaker to implement labor
policy for specific districts by seeing the trend of remote sensing data in real-time.