Nowcasting the Japanese Official Labor Statistics Using Density Ratio Estimation with Private Employment Agency's Transaction Data

Policymakers, business managers, and recruitment professionals must be aware of labor market trends to aid their decision-making. For this purpose, a range of official statistics is provided in many countries, offering insights into unemployment rates, workforce numbers, average wages, and so on, typically with a lag of two to three months. However, some indicators are difficult to grasp promptly and lag by more than a year. A prime example is the 'pressure on wages due to supply and demand' in the external labor market. The prompt publication of the indicator would be useful for decision-making by policymakers and those involved in corporate management and recruitment. Job changes in the external labor market occur through various channels: job advertisements, public and private employment agencies, and employee referrals. In this study, we focused on transaction data held by private employment agencies because they hold such data in a form that can be used in real-time. However, transactional data held by private employment agencies is only a biased subset to capture the external labor market as a whole. Therefore, to correct the bias, we proposed applying the concept of density ratio estimation and covariate shift and demonstrated that this nowcasting is feasible under realistic conditions in the Japanese case. This means that economic indicators, which are important for capturing economic trends, were previously unusable in decision-making due to the high difficulty of capturing them and the challenge of their timely publication, can now be utilized in decision-making situations.