Is Job Search Conducted Randomly in Japanese Labor Market? - Efficiency of Matching Technology and Route Selection by Searchers in Market - (in Japanese)

1. Research Theme It has been pointed out that the existence or rise in unemployment rate caused by 'mismatch' in labor market is one of the major factors for recent continuous high unemployment rate in our country. Such 'mismatch' has often been regarded as coming from technical problems, which are basically independent from business cycles. Many countermeasures have been taken by now, without any remarkable results. This paper thus discusses the importance of technical issues as a background of such 'mismatch', examines appropriateness of countermeasures taken by now, and seeks a direction of effective policies in the future. 2. Method of Empirical Analysis Most of labor statistics in our country examines behavior of workers, and very limited surveys deal with company's vacancy-postings and fillings. Therefore, we have little opportunities to obtain consistent data from both demand-side and supply-side of labor markets. To solve this problem, we merge establishment survey sheets and inflow-worker survey sheets of 'Employment Trend Survey', which enables us to figure out both the condition of vacancy-filling for each establishment and the route-selection pattern for each job-applicant (route here corresponds to the device job-applicants have used, such as public jobcenters, classifieds, etc.). By using these results, we have examined whether the route-selection behaviors by establishments and by job-seekers are random or not, and clarified several technical issues related to estimation. In addition, based on the fact that route-selection behaviors are not random, we have analyzed the effect of agents' route-selection behaviors on aggregate matching speed by using matching data of public jobcenters, which are one of the major routes in Japan. In particular, in addition to the average attributes of establishments and job-seekers, we have added several variables as a proxy for the extent of predictability in route-selection behavior, and estimated aggregate matching function (AMF). 3. Empirical Results To summarize, each empirical result suggests the following implication on current measures related to 'mismatch' problems. 1) As for the analysis of establishments, we cannot yet evaluate that establishments have come to utilize private placement services** as a complementary service, together with other traditional services (** private placement services have been more usual after regulatory reform in job-placement services in the late 90's). Some structural changes have occurred in the usage pattern of vacancy-filling routes, and the characteristics of establishments which used public jobcenters and classifieds have come to be more prominent. However, relation between 'mismatch' countermeasures*** and role-sharing among routes has not yet been apparent (*** such as introduction of private job-placement services into market, as well as computerization of database, etc). 2) As for job-applicants' analysis, their route selection behavior is clearly not random, while it does not show any particular trend through the 90's, thus the policy effects are not clear. 3) With regards to the estimation results of AMF function, the more accurate predictability of route-selection behavior by establishments, the faster it would be for them to get referrals at public jobcenters. On the other hand, such fact doesn't necessarily lead to the speed of filling vacancy for establishments. About predictability of job-seekers' attributes, the more accurate estimation would be, the faster it would be for them to get to referrals, and the higher probability for job-applicants to reach successful matching. This, in other words, means that the more difficult it would be to predict job-seekers' attributes who visit public-jobcenters, the less efficient their performance would be. 4. Conclusion Empirical results show that it is effective for the enhancement of technological efficiency of matching to improve predictability of route-selection behavior by both job-applicants and vacancy-posters. In other words, to maintain ex-ante likelihood with which establishments / job-applicants would choose a particular route would lead to efficient matching. We could expect that such policies would direct enhancement of technological matching efficiency from the most difficult part of matching process, which are suited to neither computerization nor systematic operation (for example, face-to-face counseling). At the same time, we need to pay attention to the fact that adverse selection might occur as a result of 'selection and concentration' policy (that is, job-seekers do not select best-suited routes to their own attributes). We thus need to consider the most appropriate policies for 'concentration' above mentioned, paying attention to the fact that there exists trade-off relationship between them.