Wage differences between internal and external candidates
The Authors
Wolter Hassink, Utrecht School of Economics, Utrecht University, Utrecht, The Netherlands
Giovanni Russo, Utrecht School of Economics, Utrecht University, Utrecht, The Netherlands
Acknowledgements
The authors are grateful to the Dutch Ministry of Social Affairs and Employment for giving us access to the AVO-data. Two anonymous referees of this journal are gratefully acknowledged for their constructive comments.
Abstract
Purpose – The purpose of this paper is to investigate three hypotheses for the existence of a wage premium between incumbents and employees who are hired from other employers in the external labour market.
Design/methodology/approach – The paper presents estimates of wage equations for a sample of externally hired workers and internally promoted employees. It uses an employer-employee matched data set of Dutch firms from all economic sectors (1998). It controls for various observed characteristics of the firm, the worker and the job.
Findings – The estimates reject the hypothesis that firms rely more on observable characteristics for wage formation of external candidates. Nor do the estimates favor the prediction that there is a wage premium due to the option value of risky employees. Finally, employees who are recruited internally have on average a 15 percent higher wage (net of tenure) than comparable employees who are hired from other employers.
Research limitations/implications – It was found that there was a limited possibility of identifying risky employees.
Practical implications – Firms do not reward risky employees; the incumbents seem to be of better quality than the external hirees.
Originality/value – Here the focus is on hirees who were previously employed elsewhere. Usually, a broader definition of external hiring is used.
Article Type:
Research paper
Keyword(s):
Internal labour market; Pay; Recruitment; Risk management; Promotion; The Netherlands.
Journal:
International Journal of Manpower
Volume:
29
Number:
8
Year:
2008
pp:
715-730
Copyright ©
Emerald Group Publishing Limited
ISSN:
0143-7720
1. Introduction
Sorting of employees within labor markets is a complex phenomenon as employees may change jobs both between and within firms
Where employees are not fully attached to jobs, the permeability of the ILM to the labor market at large implies that promoted internal (external) candidates have been competing with external (internal) candidates for the vacant position. This phenomenon warrants the investigation of possible differences in wage between the promoted internal candidate and the external candidates hired, after controlling for characteristics of both the job and the employee. We will refer to this difference as the wage premium.
Since the applicants' productivity is not directly observable to the firm, the candidate with the highest expected productivity would be hired
First, firms may be uncertain about the general productivity of an external applicant. We test for this possibility by relating the observable characteristics of both types of applicants to the wage. In case of uncertainty, observable characteristics have more influence on the wages of external candidates than on that of internal candidates.
Second, we concentrate on the wage level that is related to general productivity. We investigate whether there is a difference in the wage between both sources of hiring that may be the result of differences in the observed characteristics. More specifically, we focus on the wage distribution of external hirees and we investigate where the internal hirees can be located given their observed characteristics.
The third explanation is partly related to the theory of Lazear (1998). It focuses on the selection of the external hirees upon entrance. When we assume that firms are able to dismiss easily external hirees during their probationary period
In this study we will use a Dutch employer-employee matched data set. It is derived from administrative personnel records in a broad range of firms from all economic sectors. By comparing the employment records in two years (1997 and 1998) we were able to detect employees' movements into and out of the firm. In the same way we were able to identify internal transitions using the definition above. According to the definition we apply in this paper, employees will experience an internal transition (or movement) when they change departments or job assignments within a certain period.
The structure of the paper is the following. Section 2 discusses the various theories of how uncertainty about productivity may affect the wage earned by internal candidates upon a promotion and external candidates upon hiring. Section 3 deals with the data set. Section 4 gives the statistical model. Section 5 presents the empirical findings. Section 6 offers concluding remarks.
2. Theoretical framework
The match between an employee and a firm may be considered an experience good. That is the value of the match (for both parties, the employee and the firm) is not known before the match actually occurs. The surplus to be split (into wages and profits) is then uncertain a priori. After the match has been formed, both parties can observe the value of the match. Should it to be too low, both the firm and the employee have an interest in searching for better matches. The search for a better match may take place within the firm, through the mechanism of internal mobility, exploiting the possibilities offered by the ILM. In this case the allocative process takes place in isolation. However, if ILMs show a certain degree of permeability, then the process of internal mobility is further complicated by its interaction with external recruitment. In addition to the comparison among internal candidate employees, firms must now extend the comparison to external candidates. It is then interesting to investigate the extent of the advantage (if any) retained by internal candidate employees compared with external candidates.
We assume that wages are not fully determined by job characteristics so that employees in the same job may still earn different wages. Furthermore, we assume that employees' productivity Y has two dimensions: Equation 1 a general one Y(m), that may be fruitfully applied to any firm, and a match-specific one Y(s), that depends on the value of the match between the firm and the employee. Y(s) can only be useful to one firm (or to a selected group of firms). The employees receive the marginal general productivity, which is equal to their outside option. In addition, the firm and the employee share the value of the match-specific productivity. We assume that the external hirees were previously employed with another employer. The main difference between the external and internal candidates is that incumbents have revealed match-specific productivity over their tenure, which is not the case for the outsiders. Consequently, the firm has a better impression about the insider's productivity and this can lead to wage differences between incumbents and otherwise equal external hirees. We consider three cases for the wage difference.
For the first case, we concentrate on the uncertainty about the general productivity of the incumbent and the external applicant
For the second case, a systematic difference in observable characteristics between internal and external candidates differ leads to a wage difference. Thus, we consider the difference: Equation 3 where X is a vector of observable characteristics of the employee (X includes X w as well as further characteristics of the job and the firm; X contains no tenure component). β ex is the vector of parameters that valuate the observed characteristics of the external hirees with respect to their wage.
Finally, we consider the third case for the wage difference between incumbents and external candidates. We concentrate on the match-specific productivity, ceteris paribus on the general productivity. Because of the lack of records on past performance, firms face uncertainty about the newly-hired employees' ability (Y(s)). From the perspective of the employer, external candidates can be risky with respect to the match-specific productivity component. A feature of these risky employees is that at the moment of hiring, they can have a higher variance of the match-specific productivity (conditional on Y(m) and the vector of observables X): Equation 4 Lazear (1998) notes that risky external candidates (of equal general productivity Y(m)) may have an option value, when it is possible for the firm to dismiss some of these risky employees who turn out to have a low match-specific productivity
We identify the potential risky employees by the predictions of Lazear (1998). He shows that young employees will be preferred to old employees when they have the same expected output. In addition, new firms in growing industries prefer younger, riskier employees and old firms in declining industries prefer older employees. We stretch the argument in Lazear (1998) to the comparison between internal and external candidates, the latter type of candidate being riskier than the former ones. The higher uncertainty in the assessment of external candidate's productivity gives rise to the possibility that the hired candidate turns out to be more productive than expected at the time of hiring. The presence of a firm-specific component of the match and the higher variance in the productivity distribution of externally hired employees generates an option value. The option value arises because firms can let go external hirees who turn out to have a very bad match with the firm during their probationary period. In contrast, incumbents do not have such a probationary period. The option value is shared between the firm and the external hiree (which depends on their bargaining position), so that risky external hirees may receive a wage premium. In case of a wage premium for risky employees, it may be included in their outside option
The wage premium that is associated with the option value of external employees becomes: Equation 5 The first term of equation (4) measures the wage that can be attributed to the general productivity of the incumbent. It gives the incumbent's expected wage at the moment of hiring, which is considered net of tenure (which captures the value of the match-specific productivity). The second term of equation (4) gives the expected wage of comparable external hirees, who have equal general productivity, but no match-specific productivity. The wage premium is negative when the firm prefers risky external applicants.
3. Data
We make use of an employer-employee matched data set, derived from the Working Conditions Survey 1999 (hereafter abbreviated as AVO99) of the Dutch Ministry of Social Affairs and Employment (Venema and Faas, 1999). We have access to data of 1,851 Dutch organizations, which we will denote as firms. The data set is derived from administrative personnel records in a broad range of Dutch firms from all economic sectors. There is no restriction on the size of the firms investigated. The data are gathered by means of a two-stage sample. In the first stage, a sample of firms is drawn, where the population of firms is stratified towards the economic sector. The sampling probability depends positively on firm size. In the second stage, within each firm a sample of employees is drawn. The sampling fraction of employees is negatively related to firm size
In our sample, the employees may be observed in October 1997 and October 1998. In what follows, we will refer to both moments as the years 1997 and 1998. Employees who stayed with the firm, as well as employees who had experienced an internal movement are observed twice, both in 1997 and in 1998. Employees who moved in (or out of) the firm during the year are observed at one of these moments only. Externally-hired employees (job quitters) are observed in 1998 (1997) only, of these employees we know their previous labor market status (the labor market status they have moved to).
The data set contains administrative information on age, various wage components, the weekly number of hours worked, gender, education (9 levels), the level of the job (ten levels) and its complexity (eight levels), economic sector (14 sectors), and firm size (number of employees). Since we will use these variables for estimation, information on some firms cannot be used for analysis as for these firms no information was gathered about the complexity of the job, education, and previous labor market status of the external hirees. The net sample we use contains information on 1,838 firms and 44,957 employees who are employed in these firms.
Table I gives information on the various worker flows in our sample. Central to our analysis is the information on mobility within a firm. A worker is defined to have an internal transition (also denoted as internal mobility) when the worker changed in job assignment or changed department within the firm in the period October 1997 – October 1998. In this respect, the definition applied is the same as the one used by the Dutch Labor Demand Panel from the OSA (Organization for Strategic Labor Market Research; Tilburg) that was applied in Hassink (1996). All remaining employees who were employed with the firm both in 1997 and in 1998 are defined as stayers. The rate of employees with an internal transition during the year is 3.7 percent (relatively to all employees who were employed with the firm in 1997); 18.7 percent of the employees were hired externally; the remaining 77.6 percent are employees who stay with their job during the year (see Table I). When we distinguish the external hirees by previous labor market status, hiring from other employers is the largest category (61.7 percent of the external hirees).
4. Empirical strategy
This section discusses the empirical approach to test for three explanations about the wage difference.
Uncertainty of general productivity
First, the measurement of the uncertainty effect is based on equation (1). It boils down to a comparison between the estimated parameters on the observable employee characteristics of the wage equations of both sources of hiring (internal and external candidates). Thus, we estimate a wage equation for a sample of hirees that contains both the external hirees (who were previously employed with another firm) and the incumbents: Equation 6 where subscripts i and j refer to the i-th firm and the j-th employee, respectively. β and γ are vectors of parameters. X is a vector of observables. d ex is a dummy variable for externally-hired employees. ɛ is a stochastic error term. There is indication of an uncertainty effect when the impact of the employee-specific variables on the wage is larger for the external hirees than for the incumbents: Equation 7 The empirical analysis will present estimates of equation (5).
General productivity of internal applicants
Next, we consider the general productivity of the internal applicants (equation (2)). We determine counterfactually the wage of the internal hirees, by relating their observables (excluded tenure) to the value of the observed characteristics of the external hirees. Consequently, we get a measure of the incumbents' wage that can be attributed to their observable characteristics
The wage difference between both sources of hiring may be due to a difference in observable characteristics: Equation 11 Thus, the incumbents' counterfactual wage is on average higher than that of the external candidates. The wage difference is based on an equal valuation of the internal and external candidates, but the difference in wage can be due to different observed characteristics of both candidates. We will present estimates of equation (10).
Option value of risky employees
Finally, we consider the wage premium that is associated with the option value of risky external candidates (equation (4)). We estimate a wage equation for the incumbents. The first component, E(w ic |Y(m), X, T=0), is estimated by the average of the fitted value of the dependent variable at T=0. It gives the incumbent's wage net of the value of the match-specific productivity, which is captured by tenure. For the second component of equation (4), E(w ex |Y(m), X, we need to have comparable external hirees (in terms of Y(m) and X). We compute this term for the incumbents, using the valuation of the general productivity of the external candidates. It gives an estimate of the incumbents' wage that they receive when the firm has no information about the value of the match-specific productivity at all.
Thus, we estimate separately a wage equation for the internal candidates: Equation 12 With the estimated parameters of equation (11), we determine the internal candidates' value of the log hourly wage that we may predict given their observed characteristics X ij ic and T ij ic : Equation 13 as well as their predicted log wage, given the observed characteristics (X ij ic ), but net of tenure (T ij ic ): Equation 14 The tenure component in equation (12) reflects the fact that the match is an experience good. When the match-specific components captured by the returns to tenure are not known ex-ante, tenure effects are more likely to reflect the internal selection process (fast tracks and mummy tracks). In other words, we will purge the wage of internal candidates from the effects of the internal allocative process, whose effects are to be taken back to the firm learning its employees' productivity.
Equation (13) corresponds to the extrapolated log hourly wage of the internal candidate promoted to the new position. This predicted wage does not contain an option value because internal promotees are not entitled to an option value. Incumbents cannot be dismissed if the quality of the match in the new position is below expectation. In contrast, the predicted wage of equation (9) is that wage of an internal candidate when the firm is fully unaware of his match-specific productivity. So, the difference of equations (13) and (9) gives the wage premium of a risky employee: Equation 15 When the wage premium (14) is negative, risky external applicants (who can have a valuable match with the employer) receive a higher wage upon hiring.
5. Estimates
We first contrast the wage distribution of the internally mobile employees against the wage distribution of the employees who did not experience internal mobility (the stayers) or who were hired from the external labor market. The percentiles of the logarithm of the hourly wage (the wage actually earned by the employees in 1998) of these three distributions are compared in Table II. Furthermore, it shows the wage distribution of the stayers and the employees with internal mobility in 1997.
Table II indicates that the wages of external hirees tend to be positioned in the lower part of the wage distribution. In addition, it is clear that the wage distribution of the employees who experience internal mobility has improved compared with the wage distribution of the stayers between 1997 and 1998. The wage increase experienced by the employees with internal mobility must have been above the wage growth experienced by the group of the stayers. However, one can infer that the increase is slightly lower for the employees in the higher percentiles. For the employees with internal mobility the 10th percentile of the hourly wage increased by 15 percent, whereas for employees in the 90th percentile, the hourly wage increased by 13 percent. For the stayers the corresponding increase at both centiles is 5 percent. The right-hand side panel of Table II further shows that the percentiles of the employees with internal mobility are above the percentiles of employees who were hired from other employers. This could be due, however, to tenure effects of the promotees.
Table III gives the descriptive statistics of the variables. For some independent variables, internal candidates and external hirees have different characteristics. Incumbents are on average two years older than applicants who are hired from other employers.
Uncertainty of general productivity
We concentrate on the wage difference between incumbents and external hirees. For the first explanation (uncertain general productivity), we estimated equation (5). See Table IV. Various variables of the interaction terms with the dummy for external hiree are statistically significant: the interaction terms of job level (dummies), job complexity (dummies) and education (dummies) are each jointly significant at the 5-percent level. Remarkably, a comparison of the marginal effects of the statistically significant dummies between incumbents and external hirees (equation (6)) indicates that the wage responds more to the observable characteristics of the incumbents. It implies that employee-specific characteristics are more important to estimate productivity for own employees than for external applicants. This outcome contradicts however the hypothesis that firms rely more on the observed characteristics for the external candidates.
General productivity of internal employees
For the second explanation about the wage difference, we determined the fitted (log) wages of equation (8) and (9). See Table V. Interestingly, a comparison of the first column (4.58) and the second column (4.73) indicates that given the observable characteristics, employees who have had an internal movement have on average a higher wage (net of tenure) than employees who were hired from other employers. The difference in wage of 15 percent reflects the fact that internal applicants have better observable qualities than employees who are hired from other employers
Option value of risky employees
Next, we concentrate on the third explanation. We computed the predicted log wage of the internal candidates using the estimated coefficients of equation (12), in which the tenure effect is included (based on equation (11)). This is reported in the third column of Table V. Finally, the fourth column gives the predicted log wage, net of tenure (equation (13)). The difference between the predicted log wage of the fourth column (4.72) and the second column (4.73) is the wage premium (equation (14)). It appears that the wage premium is nought. Thus, we can conclude that for all external hirees (previously employed with another employer) there is no indication that they receive a wage premium because they are risky.
We focus on specific types of employees who can be characterized as a risky employee, following Lazear's predictions. He argues that there may be a different preference for risky employees in fast growing and declining industries. In addition, young employees may be characterized as risky employees. For all of the other selections that are presented in Table V, we find a wage premium equal to zero. First, we split the sample according to employment change (increasing and decreasing employment). In addition, we identify the firms that are structurally growing above 5 percent annually
6. Conclusion
The permeability of the internal labor market begs the question of whether, and if so to what extent, external hirees are different from employees experiencing internal mobility. At the end of the day, employers posses less specific information on external employees' ability and may treat the external candidate differently in response to it.
Using a large administrative data set of Dutch employees we find that employees' mobility appears to be characterized by the following empirical regularities: Employee turnover into firms generally takes place at the lower end of the wage distribution. Despite the higher level of turnover at lower wages, there appears to be consistent mobility into firms along the entire wage distribution.
We investigated whether uncertainty about the hiree's productivity leads to a wage premium for internal candidate employees relatively to external hirees who were hired from other employers (after controlling for job level, job complexity and employee characteristics). Our estimates reject this hypothesis however. They indicate that the response of the wage to observable characteristics is higher for incumbents than for external hirees.
Interestingly, we find that when we control for the observable characteristics, employees who are recruited internally have on average a 15 percent higher wage (net of tenure) than comparable employees who are hired from other employers.
Since externally hired employees have a higher variance of the match-specific productivity, the ensuing option value could reasonably be expected to accrue to external candidates in their wage. In other words, “risky” external employees receive a higher wage than “safe” internal candidates of equal (general) productivity. To comply with the equal productivity constraint we focus on external hirees who were previously employed elsewhere. We find that, on average, internal and external candidates with equal observable characteristics earn the same wage; therefore, it appears that the option value accruing to external (risky) candidates is not rewarded. This result holds true also on various sub-samples of risky employees.
Equation 1
Equation 2
Equation 3
Equation 4
Equation 5
Equation 6
Equation 7
Equation 8
Equation 9
Equation 10
Equation 11
Equation 12
Equation 13
Equation 14
Equation 15
Table IInternal mobility and different sources of external hiring; number of employees
Table IIPercentiles of the logarithm of the hourly function wage (by type of worker flow; 1998)
Table IIIDescriptive statistics 1998
Table IVEquation (5); dependent variable: log (hourly wage in 1998)a
Table VPredicted (log) hourly function wage, 1998
References
Abowd, J.M., Kramarz, F., Margolis, D. (1999), "High wage workers and high wage firms", Econometrica, Vol. 67 No.2, pp.251-334.
Agrawal, A., Knoeber, C.R., Tsoulouhas, T. (2006), "Are outsiders handicapped in CEO successions?", Journal of Corporate Finance, Vol. 12 No.3, pp.619-44.
Baker, G., Gibbs, M., Holmstrom, B. (1994a), "The internal economics of the firm: evidence from personnel data", Quarterly Journal of Economics, Vol. 109 No.4, pp.881-919.
Baker, G., Gibbs, M., Holmstrom, B. (1994b), "The wage policy of a firm", Quarterly Journal of Economics, Vol. 109 No.4, pp.921-55.
Bollinger, C.R., Hotchkiss, J.L. (2003), "The upside potential of hiring risky workers: evidence from the baseball industry", Journal of Labor Economics, Vol. 21 No.4, pp.923-44.
Burgess, S.M., Lane, J.I., Stevens, D. (1998), "Hiring risky workers: some evidence", Journal of Economics and Management Strategy, Vol. 7 No.4, pp.669-76.
Chan, W. (1996), "External recruitment versus internal promotion", Journal of Labor Economics, Vol. 14 No.4, pp.555-70.
Devereux, P.J., Hart, R.A. (2006), "Real wage cyclicality of job stayers, within-company job movers, and between-company job movers", Industrial and Labor Relations Review, Vol. 60 No.1, pp.105-19.
Doeringer, P.B., Piore, M.J. (1971), Internal Labor Markets and Manpower Analysis, D.C. Heath & Company, Lexington, MA, .
Gibbons, R., Waldman, M. (1999a), "Careers in organizations", in Ashenfelter, O., Card, D. (Eds),Handbook of Labor Economics, Vol. Vol. 3b pp.2373-437.
Gibbons, R., Waldman, M. (1999b), "A theory of wage and promotion dynamics inside firms", Quarterly Journal of Economics, Vol. 114 No.4, pp.1321-58.
Hassink, W.H.J. (1996), "An empirical note on job turnover and internal mobility of workers", Economics Letters, Vol. 51 No.3, pp.338-44.
Hendricks, W., DeBrock, L., Koenker, R. (2003), "Uncertainty, hiring, and subsequent performance: the NFL draft", Journal of Labor Economics, Vol. 21 No.4, pp.857-86.
Lazear, E.P. (1998), "Hiring risky workers", Internal Labour Markets, Incentives and Employment, Macmillan Press Ltd, Basingstoke, pp.143-58.
Lazear, E.P., Oyer, P. (2004), "Internal and external labor markets: a personnel economics approach", Labour Economics, Vol. 11 No.5, pp.527-54.
Treble, J., VanGameren, E., Bridges, S., Barmby, T. (2001), "The internal economics of the firm: further evidence from personnel data", Labour Economics, Vol. 8 No.5, pp.531-52.
Venema, P.M., Faas, A. (1999), “Arbeidsvoorwaardenontwikkelingen in 1998. Een onderzoek van de Arbeidsinspectie naar gerealiseerde loonontwikkelingen, extra uitkeringen en afdrachtverminderingen”, Dutch Ministry of Social Affairs and Employment, The Hague, .
About the authors
Wolter Hassink is affiliated as associate professor of economics with the Utrecht University (Utrecht School of Economics) and is a research fellow of the IZA. He received his PhD from the Free University of Amsterdam in 1996. He is the corresponding author and can be contacted at: w.hassink@econ.uu.nl
Giovanni Russo is professor in Personnel Economics at Trieste University. Before he has worked at Utrecht University (as assistant and associate professor). He received his PhD from the Free University of Amsterdam in 1996. His research interests are in labor economics and the psychology of labor market behavior.