Inspired by the recent increase in the turnover among analysts, this paper examines the efficiency of matching between analysts and brokerage houses in turnover, and the possibility of adverse selection between current and prospective brokerage houses due to asymmetric information on the analyst forecasting ability. We apply the implication of adverse selection provided by Akerlof (1970) and Wilson (1979) to our analysis of the turnover ratio or job matching among analysts, current and new brokerages. By nature, the current house tends to have hidden information about the predictive accuracy of the analyst additionally as well as publicly-known information such as recent EPS forecasting errors and the contents of her past reports. On the other hand, a prospective employer has to evaluate the analysts' forecasting ability only based on their performance with the public information. Consequently, employers have to make the hiring decision based on limited information about the analysts' actual ability to make accurate forecasts. Several studies such as Jackofsky (1984), Greenwald (1986), and Kim (2012) have already addressed the adverse selection issues in the context of analysts' job separation or multi-period labor market theoretically. In addition to the body of work on this topic, our study proposes a simple model that can link the turnover ratio to the change in analysts' EPS forecast error, and then tests it. For the model, we assume that there are two different types of analysts: low-type analysts with higher EPS forecast error and higher variance; and high-type analysts with lower EPS forecast error and lower variance. Either type is tempted to accept an offer from a new brokerage house whenever the offer is better than what they are currently receiving. Note that, however, high-type analysts are more likely to reject the offer in the end since the current employer will most definitely suggest a better counter-offer to keep them. On the other hand, low-type analysts may end up moving to another house as such counteroffer is much less likely to be made by their current employers. Based on this observation, we conjecture that as long information asymmetry exists between the prospective employer and the analysts on their true forecast ability, low-type analysts are more likely to move to another brokerage house than high-type ones do when the low-type analysts achieve high performance temporarily. For the test, we examine whether a group of analysts who move to a new brokerage house (“movers”) make more EPS forecast errors at the new house than they do at the previous workplace, compared to the other controlling group of analysts who do not change their workplace (“stayers”). We also consider the efficient matching effect to account for the case in which movers perform better at their new workplace compared to how they did in their previous one, because the matching efficiency can certainly help “the movers” to commit lower EPS forecast errors at new house than they do at the old one, compared to their counterparts. In this study, we adopt two variables to indicate the degree of information asymmetry between current and prospective employers: the years of experience of the analysts, and the number of analysts that cover one stock. The greater the two variables become, the better the predictive accuracy can be expected, which means less EPS forecast errors, allowing new employers the access to more information about the ability of the turnover analysts. In particular, this allows us to test not only the presence of the adverse selection problem, but also whether the extent of information asymmetry is different across different segments of analysts' labor market. We construct data set on the Korean analysts' turnover based on their reports from year 2004 to 2010. This paper applies the difference-in- difference method, which estimates the change in forecast errors of movers - experimental group, relative to stayers - the control group between before and after the turnover (job separation). For the overall sample analysis, we find that the EPS forecast error of movers does not in crease significantly after the separation relative to the “stayers,” implying that the adverse selection effect is negligible or limited if there is any. As we segment the analyst turnover market according to the experienced year and the number of competing analysts within the same company, the relative change in the EPS forecast error of movers turns out to vary depending on the criteria. Concretely, the group of movers, who have 3-year or less experience as analysts and also companies that are covered by 20 or fewer analysts, tend to have higher EPS forecast errors after separation, compared to the stayers' group. Otherwise, the change in forecast errors is not different across movers and stayers. These findings imply that adverse selection exists only in some limited analyst labor market segments. The portion of the aforementioned analyst group (with3-years or less experience and covering stocks by 20 or fewer analysts) accounts for only 9.7% of the total analysts. This implies that the adverse selection issue is not too significant in the overall analyst turnover market. From this perspective, our results indicate that efficient matching has occurred despite sharply increasing wages and the possible adverse selection problem.