Optimal number of job changes

Hübler, Olaf Article Optimal number of job changes Zeitschrift für Wirtschaftsund Sozialwissenschaften (ZWS) - Vierteljahresschrift der Gesellschaft für Wirtschaftsund Sozialwissenschaften, Verein für Socialpolitik Provided in Cooperation with: Duncker & Humblot, Berlin Suggested Citation: Hübler, Olaf (1989) : Optimal number of job changes, Zeitschrift für Wirtschaftsund Sozialwissenschaften (ZWS) - Vierteljahresschrift der Gesellschaft für Wirtschaftsund Sozialwissenschaften, Verein für Socialpolitik, ISSN 0342-1783, Duncker & Humblot, Berlin, Vol. 109, Iss. 1, pp. 75-92, https://doi.org/10.3790/schm.109.1.75 This Version is available at: https://hdl.handle.net/10419/291701 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Zeitschrift für Wirtschaftsu. Sozialwissenschaften (ZWS) 109 (1989), S. 75 - 92 Duncker & Humblot, Berlin 41 Optimal number of job changes By Olaf Hubler* This paper estimates the individual optimal number of job changes which is derived from an earnings function where the working history affects the earnings in a direct and an indirect way. In comparison between actual and optimal number of job changes it is found that workers with a high degree of on-the-job training and unexperienced persons have underoptimal mobility behavior. 1. Introduction Empirical studies of individual labor turnover are usually based on probit models.1 The dependent variable is a dummy separating movers and stayers. The following or a similar decision rule to quit is used: A worker i compares his expected utilities (earnings streams) in the present job (V (PJ¿)) and in an alternative job (V (A J*)), and accepts the offer of the alternative job only if the utility associated with the latter is at least as great as in the first job plus the costs associated with mobility (0*): (1) V(AJi) > V(PJi) + Ci. Although this outcome may be what intuitively appealing, it is not necessarily valid. Implicitly, decision rule (1) assumes that an accepted offer is acceptable forever. The worker is allowed to move only once; the possibility of further search for higher-paying jobs is ignored. But this is the normal behavior as Thurow points out: "The sensible search strategy is to accept the first job offered but to keep on looking. Whenever a better job is found at a higher wage rate the sensible searcher quits the first job and takes the second."2 With respect to this behavior, problems will not arise if (2) V(Ji{1)) < V(J/2)) - Ci(1'2) < V(J/3)) - C/2,3) < ..< V(Ji{L)) - Ci (L-1'L), where V (J/1*) is the expected utility of individual i in job 1 and C/z" 1( l) are mobility costs if i moves from job (I - 1) to job 1. In some cases, however, the * I am grateful for helpful comments by John T. Addison, Knut Gerlach and Gerhard Kocklauner. 1 E.g. Borjas / Rosen (1980), Bartel (1982), Robinson / Tomes (1982), Blau / Kahn (1983), Antel (1985), Shah (1985), Osberg / Mazany / Apostle / Clairmont (1986). 2 Thurow (1983), 194. ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 76 Olaf Hübler isolated decision between two jobs according to rule (1) may be suboptimal. It is possible to get another wage offer after a job change that would have been preferred before changing but which is not sufficiently high to induce a second change. Furthermore, we have to consider situations in which an offer is accepted, although equation (1) does not hold. Such a decision might be appropriate if a job change induces more attractive offers later on. Examples are migration into regions with larger labor markets and higher probabilities to obtain better jobs than in the present region; moves to jobs where the contact with firms offering more attractive jobs is improved; and job changes which signal the worker's flexibility and productivity and thus may induce promotions of further offers. We focus upon the latter consideration, one that has been neglected in the economic literature on search and labor turnover. Even studies of optimal search3 appear unaware of the problem. Assuming perfect information, all potential offers can be included within the decision rule. But in reality, there exists uncertainty about the prospective offers. The purpose of this paper is to derive the optimal number of job moves under optimizing behavior. Our basic hypothesis is that the decision to quit the current job depends on the worker's employment history, especially on the past number of job changes. If the actual number (NOM) is smaller than the optimal number (ONOM), a worker should intensify his search and possibly accept an offer even if equation (1) does not hold. He or she should reject a utility increasing offer (equation (1)) if NOM exceeds ONOM. The paper is organized as follows. In section 2, we give arguments that employment history measured by NOM affects earnings in direct and indirect ways. We present a basic earnings model from which alternative optimal mobility functions are derived. The empirical investigation - data, specification, estimation problems, results and interpretation - follows in section 3. 2. Some theoretical aspects and the model 2.1 Effects of turnover on earnings Some arguments suggest that earnings depend on working history as measured by the number of jobs, although most empirical investigations of earnings functions neglect this aspect.4 First, some economists interpret mobility costs as investment human capital.5 Individuals with different amounts of investment have to be compensated according to their costs. Second, the 3 Morgan / Manning (1985). 4 Exceptions are Borjas / Mincer (1978), Bartel (1980), Mincer / Jovanovic (1981), Hubler (1984). 5 See, for example, Fein (1965). ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 Optimal number of job changes 77 number of jobs is a screening device for firms. In general, employees are risk-adverse. Risk-lovers are scarce. Workers who change jobs, though returns are uncertain, signal such a characteristic. Third, movers possess better information about their abilities and suitable jobs in the sense of jobshopping and job-matching6 than stayers. Fourth, the number of jobs is a proxy for the offers. We can assume that the most able employees are attracted away from their jobs earlier than other workers.7 These arguments support the hypothesis of a positive correlation between earnings and the number of jobs. However, a negative dependency is also possible. According to implicit contract theory, movers have inferior reputations. A job change may signify the violation of a long-term implicit contract. Firms will be prepared to hire workers with a poor reputation only if they can pay them lower wages than other employees. If we incorporate working history, z, measured by the number of job moves into a conventional linear model of earnings functions as an independent variable, then the idea of an optimal z under maximizing behavior is useless. The greater (smaller) z, the greater (smaller) are earnings. Substituting the linear term fizz by non-linear terms (e.g. c0 z + Ci z2) we obtain a nontrivial z. Our theoretical considerations suggest that this approach makes sense. We expect opposite signs on c0 and Ci; for example, growth of information by job-shopping decreases with the number of jobs.8 But this is not the whole story of the effects of working history or job moves on earnings. Our hypothesis is as follows: The returns to earnings determinants vary with z. The term CiZ2 is merely a proxy for the unspecified indirect earnings effects of z. In other words, the employers evaluate identical personal characteristics differently, and this evaluation is based on working history. For their part, employees are in a better position to signal their abilities the longer and the more diversified is their working history, while firms interpret employment history as a good signal for workers' prospective behavior. Following the idea of varying returns to earnings determinants, we have to introduce the so-called indirect effects of z on earnings by systematically varying parameters. Assume that the total individual coefficients can be divided into three components (a constant element for all individuals, a zdependent part and a stochastic component) (3) fa = (a0 + axZi+ £i) • fij and fiiz = (a0 + ai z{ + e¿) • , 6 Johnson (1978), for example, presents a model of job shopping. 7 Lazear (1986) makes this assumption, s See Miller (1984). ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 78 Olaf Hübler where i = 1, n; j = 1, k; a0 and ax are unknown parameters, £ is a (nx 1) vector of disturbances, E (e¿) = 0 and E (s¿) = o\. Not all earnings effects are observable and some of them are small. These variables are summarized in an error term i¿, where E (ui) = 0 and E (u\) = (al) and E (i¿¿ £j) = 0. Accordingly, our earnings functions may be written (4) y = (diag (a0 + ax zx + eu ..., a0 + ax zn + £„)) • (X0 + fizz + u), where y ~ (nx 1), X ~ (nx/c), u ~ (nx 1), ft ~ (kx 1), z ~ (nx 1), ~ (lxl), a0 ~ (lxl), ai ~ (lxl), Zi ~ (lxl), e¿ ~ (lxl), i = 1, n. A more flexible approach is given by substituting (3) by (3a) fiij = (a0 + ay Zi + £i) • ft and fiiz = (a0 + ai2 z¿ + • & , where j = 1, ..., /c. Consequently, more realistic cases may be considered, where z affects y indirectly via some but not all earnings components (some aij or ai2 are zero). The decision rule for job changes in conventional models is not only based on earnings. Mobility costs (C) have also to be considered. We assume that working history also affects C. The more jobs sampled, the higher will be the total mobility costs. But the workers learn to manage job changes, so that marginal costs may be expected to fall with an increasing number of job moves. In particular, pre-move information decreases the marginal costs of search.9 We complete our model with the following mobility cost function constructed in a similar fashion to the earnings function (5) C = (diag (b0 + bi ¿i + , • . b0 + 6i zn + en)) -(Wy+ yzz + u), where C is a (nxl) vector of mobility costs, b0, bi and yz are unknown parameters, Wis a (nxl) matrix of I costs determinants, y is a (Zxl) vector of unknown parameters, f¿ is an error term, and u is a (nxl) vector of disturbances. Furthermore, we assume E (e¿) = 0, E (e¿) = a|, E (u¿) = 0, E (u\) = o\ and E (u e') = 0. Equation (5) can be modified in the same way as the earnings function if we substitute (3) by (3 a) (5a) Yiy = (b0 + by, z{ + e¿) • yy and yiz = (b0 + blz z> + £<) • yz, where j' = 1,..., I. 9 See Herzog / Hofler / Schlottmann (1985), 374. ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 Optimal number of job changes 79 2.2 Optimal number of job moves Equations (4) and (5) are the basis for deriving an optimal working history function. But it is not useful to calculate the expression 3 (y - C)/3 z. We have instead to consider expected net earnings. Let us assume that after the job change there are no further mobility costs incurred and that all individuals work until age 65. Moreover, suppose that earnings increase uniformly with the factor exp (gt), where g is a constant rate, that may be positive or negative. This rate is a composite of the usual rate of wage growth, the conventional discount rate, the death and layoff risk, and of the probability of interrupting the working career. This definition allows us to calculate the net returns of a given working history10 65-AGEi (6) V(Zi) = | y(Zi) exp (gt) dt - C (Zi) o = (y (Zi) (exp (g (65-AGEO) - 1) / g) - C (Zi) = : y (Zi) • F (AGEi) - C (zt) i= 1, ...,n, where AGE, is the age of the i-th person and t ist a time variable. If we calculate (7) (3 (y' F/d z) - (3 (C' ¿)/3 z) = 0, where F = (F (AGEi), ..., F(AGE,))' =: (Fu ..., Fn)', 6 = (1, ..., 1)', the solution for all zt is (8) z = — (blYz diag (1, . . 1)„ - a1fizdiag(F1,...,Fn))-1. 2 ((aQp2F-boYzi) + (Mdiag (Fu . . Fn)) X P - b,W y) + (a! (diag (Fu . .Fn)) u + pz (fll (diag (Fu . . , Fn)) e-^u-y.e). This model is nonlinear. In special cases, we obtain a linear specification (Pz = 0 or ai = 0 or au = 0, if we use (3a) instead of (3)). The first case excludes the direct effects of z on y. The second case neglects all indirect effects. In the third case, the interaction of z with itself vanishes. We then have (8a) z = a0 + ai F + (diag (Fu . . Fn)) Xa2 + Wa3 + v = : Xa+v, 10 We follow Weiss (1984) who has presented a quit decision rule for conventional models. ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 80 Olaf Hübler where r in X and W indicates that only a part of X = (XT: •) and W = (Wr: •) interact with z, in other words, that some aij and are zero. a0 and ai are unknown parameters, a2 and a3 are (kxl) and (Ix 1) vectors of unknown coefficients, and vis a (nx 1) vector of disturbances. As we have emphasized that the term of z2 is only a proxy for unspecified indirect earnings effects of z, the assumption aiz = 0 entails little loss of information if the model is well specified. Another way of obtaining a linear model is to use a first-order Taylor expansion as an approximation. In this case, F and all variables of matrix X and W which interact with z are included as exogenous variables in the optimal working history function. The crucial point of (8) and (8 a) is its assumption of a uniform rate of wage growth. We argue that g depends on z, too. For reasons suggested earlier, the actual returns of earnings determinants are a function of z. Therefore, it is sensible to expand this hypothesis to prospective returns. Accordingly, we have to replace (7) by (9) 3 V(Zi)/dZi = {[{dyildzi)gi-(dgi/dzi)yi}lg2 i}- (exp (9i (65-AGEj) -1)4- (Vi / 9i) (d9i/d z{) • (65-AGEj) exp (g{ (65-AGEO) - (9 C4 / 8 = 0 . The resulting optimal working history function from (9) is nonlinear. An approximate linear approach is again obtained by using a first-order Taylor expansion. This means that compared with the Taylor-expansion of (8) we have nearly the same specification, although AGE is now substituted for F. We propose an alternative approach. Take (8 a) and complete this function by a function of F* which depends on z (10) Fi = ((exp (kzi (65-AGEj)) - 1) / kz{) x? , where gi = kziy k = const., and x* is an error term assuming E (xf) = 1. We approximate (10) by (11) In Fi = d0 + di Zi + d2 ZjAGEj + d3 In z{ + x , where d; are unknown parameters (j = 0, 1, 2, 3) and x is a disturbance term. We now have to estimate a nonlinear, two-equation model of (8a) and (11). ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 Optimal number of job changes 3. Empirical investigations 3.1 Data 81 The information on individual working histories used in the empirical analysis is based on a 10 percent random sample of all employed persons in the state of Bremen. In November 1981, a questionnaire was sent to 26,453 employees, the purpose of which was to obtain data on schooling, job change, industry, type of work, working conditions, job stability, earnings, and personal characteristics. The sample size used in this study consists of 4,657 employees. Only those persons are included which where not unemployed far at least ten years. The purpose of this restriction is that we want to exclude dismissals. We have no information whether an individual job move is voluntary or not. But most quitters change jobs directly without intervening unemployment, while most layoffs are unemployed between jobs. We expect in accordance with Antel and Mincer11 that the latter have a different search behavior than quitters, while Borjas / Rosen12 argue that the decomposing in quits and layoffs is artifical since workers who know that a layoff is about to occur may quit and firms who know that workers are about to quit may lay them off. All the variables used in the empirical analysis are described in Table 1 but some of them has to be explained a little bit more. In our investigation the most important variable 2: is measured by the answer to the question: "How many firms did you work for during your life?" (NOF = number of firms). Earnings are expressed by the natural logarithm of monthly gross income (In Y). Social background (SB) is proxied by father's occupation status. From seven categories (1 - unskilled worker, 2 - skilled worker, 3 - farmer, 4 - white collar worker, 5 - civil servant, 6 - self-employed, 7 - manager) an ordinal scale 1 - 7 is constructed where the average income of unskilled workers is the lowest and that of the managers is the highest. In one question the employees are asked: "What is your degree of management tasks?" (1 - no, 2 - small, 3 - middle, 4 - high). The answer is interpreted as the individual position in the hierarchy of the firm (HIER). Another variable measured by an ordinal scale is the degree of on-the-job training (DOJT). The answer to the question "On your job what degree of training is necessary that an average new person is fully trained and qualified? (1 - no training, 2 - longer training period, 3 - vocation requiring an apprenticeship, 4 - vocation requiring an university education) is used as a proxy for special human capital. 11 Antel (1985); Mincer (1986). 12 Borjas / Rosen (1980). ZWS 109 (1989) 1 6 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 82 Olaf Hübler 3.2 Specification and expected signs We have to specify the earnings and the mobility costs function. In the human capital tradition, the logarithm of earnings is explained by schooling (S), experience (EX) and its square (EXSQ). In more recent studies,13 the exogenous variables are supplemented by the number of years of tenure in the current firm (TEN) and square of tenure (TENSQ), or TEN and the previous experience (PEX = EX - TEN) are substituted for EX.14 We follow the latter approach. Notable explanations of earnings-tenure profiles are the firm-specific human capital, agency, self-selection, implicit contract and segmented labor market hypotheses.15 We add selected job and personal characteristics such as firm size (SIZE), measured by the number of employees, position in the hierarchy of the firm (HIER), working time (TIME), sex (MEN; 1 - man, 0 - otherwise), social background (SB). The direct earnings determinants (X) and the resulting sign expectations of the partial derivatives in the earnings functions are as follows (12) X = / (NOF, S, TEN, TENSQ, PEX, PEXSQ, SIZE, HIER, TIME, MEN, SB) . ? + + - + - (+) + - + + Several recent empirical studies of the determinants of wages have come to the conclusion that a positive and significant correlation between SIZE and wages exists, but apparently this positive effect of firm size is statistically significant only for firms with more than 100 employees.16 For firms with fewer than 100 employees, there is no consistent relationship between firm size and wages. The positive sign of HIER can be explained by the degree of responsibility17 or by the rank-order tournament theory.18 If working time is expanded the earnings also increase. Therefore, considering the definition of TIME (see Table 1) the expected sign of this variable is negative. Pay differences between men and women may be attributed to sex segregation by firm or unobserved characteristics which differ between male and female. Nepotism is one reason that we have to expect a positive correlation between SB and earnings. 13 See, for example Hashimoto / Raisian (1985). 14 See Mincer / Jovanovic (1981); Holmlund (1984). 15 See Arai (1982); Cornfield (1982); Hashimoto / Raisian (1985). 16 Weiss / Landau (1984) present theoretical arguments and empirical evidence for this relationship. 17 See, for example, Lydall (1968). 18 Lazear / Rosen ( 1981). ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 Optimal number of job changes 89 f 1 if DIFF < 0 Dl = \ 0 otherwise f 0 if - 0,5 < DIFF < + 0,5 D2 = <1 ^ 1 otherwise and apply some maximum-likelihood estimates of probit models. The results are presented in Table 3. From columns (1) and (2) we may conclude that persons with limited previous experience, high degree of onthe-job-training, inferior social background, and whose spouse is not working have too low a propensity to move, or that employees with opposite characteristics change too frequently. To decide which interpretation is correct, we must examine the estimates given in column (3) and (4). It emerges that young men with completed high school, a high degree of on-the-job training, limited previous experience, with an inferior family background do not behave rationally under the net earnings maximization assumption. Some of these characteristics describe employees who have recently begun their working career. Table 3: Maximum-likelihood estimates of probit models to determine systematic factors of underoptimal mobility behavior (asymptotic | 11 -ratios in parentheses) exogenous variables* endogenous variables D1 (1) (2) (3) (4) NOF 3.9816 (23.66) -0.0445(1.57) AGE -0.0085 (1.35) -0.0168(6.41) MEN -0.1006 (0.63) 0.1609(2.54) S -0.0061 (0.42) 0.0297 (0.92) 0.0235 (1.64) 0.0288 (2.32) TEN 0.0010 (0.37) -0.0022 (0.83) PEX 0.0232 (10.72) -0.0074(3.60) SIZE -0.0292 (1.40) 0.0095 (0.46) DOJT -0.1156 (3.18) 0.0735 (2.08) SWO 0.1746 (2.93) -0.2193 (3.76) SB 0.0283 (2.89) -0.0243 (2.53) const. 0.3107 (1.90) -9.8641 (16.82) -0.4538(2.83) 0.0767 (0.45) (-2) log likelihood ratio 184.44 2313.55 47.06 66.40 D2 * See Table 1. ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 90 Olaf Hübler From the results of Table 3 we would emphasize two major points. First, labor turnover is important in the beginning of the working history to inquire into potential job opportunities and one's abilities in order to improve job matching. And, as we previously mentioned, employees with a low degree of knowledge and with less favourable family background can improve their economic chances by turnover. But our results show that in both cases the "solution" will not sufficiently be used. In short, the mobility of these persons is underoptimal. Second, persons with a high degree of onthe-job training have a tendency to indulge in underoptimal mobility behavior. It is rational that they do not change their jobs as frequently as persons with a low degree of on-the-job training. Specific human capital which cannot be used in other firms results from on-the-job training. But we suppose that the general human capitel effects of on-the-job training which induce greater potential job opportunities are underrated, because the information on the current firm is better than on other firms. Summary Under maximizing behavior, an optimal mobility function is derived from an earnings function and a mobility cost function, where the coefficients are variable depending on working history. Two approaches are distinguished: one with uniform and one with variable rates of growth. The most important determinants of optimal number of jobs are the earnings growth factor, previous experience, firm size, social background, and schooling. A comparison between optimal and actual number of job changes and the ML estimator of a probit model reveals that high skilled workers with limited experience and poor family background have a propensity to quit that is too low. Zusammenfassung Ausgehend von einkommensmaximierendem Verhalten wird aus einer Einkommensund einer Mobilitätskostenfunktion eine Funktion des langfristig optimalen Arbeitsplatzwechsels abgeleitet. Unterschieden werden zwei Ansätze mit variablen Koeffizienten, wobei einmal von einer konstanten und zum anderen von einer individuell variierenden Einkommenswachstumsrate ausgegangen wird. Die wichtigsten Bestimmungsfaktoren des optimalen Arbeitsplatzwechsels sind: Einkommenswachstumsfaktor, bisherige Berufserfahrung, Firmengröße, Schichtzugehörigkeit und Schulausbildung. Ein Vergleich zwischen der optimalen und der tatsächlichen Anzahl an Arbeitsplatzwechseln sowie die ML-Schätzung eines Probitansatzes zeigen, daß gut ausgebildete Arbeitskräfte mit geringer Berufserfahrung und solche, die aus unteren sozialen Schichten kommen, eine zu geringe Mobilitätsneigung besitzen. ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 Optimal number of job changes References 91 Antel, J. J. (1985), Costly Employment Contract Renegotiation and the Labor Mobility of Young Men. American Economic Review 75, 976 - 991. Arai, K. (1982), Theories of the Seniority-Based Wage System. Hitotsubashi Journal of Economics 23, 53 - 67. Battel, A. P. (1980), Earnings Growth on the Job and between Jobs. Economic Inquiry 18, 123 - 137. — (1982), Wages, Nonwage Job Characteristics, and Labor Mobility. Industrial and Labor Relations Review 35, 578 - 589. Blau, F. D. / Kahn, L. W. (1983), Unionism, Seniority, and Turnover. Industrial Relations 22, 362 - 373. Borjas, G. J. (1981), Job Mobility and Earnings over the Life Cycle. Industrial and Labor Relations Review 34, 365 - 376. Borjas, G. J. / Mincer, J. (1978), The Distribution of Earnings Profiles in Longitudinal Data, in: Z. Griliches / W. Krelle / H.-J. Krupp / O. Kyn (eds.), Income Distribution and Economic Inequality. New York, 175 - 197. Borjas, G. J. / Rosen, S. (1980), Income Prospects and Job Mobility of Younger Men, Research in Labor Economics 3, 159 - 193. Breusch, T. S. / Pagan, A. R. (1979), A Simple Test for Heteroscedasticity and Random Coefficient Variation. Econometrica 47, 1287 - 1294. Chirinko, R. S. (1982), An Empirical Investigation of the Returns to Job Search. American Economic Review 72, 498 - 501. Cornfield, D. B. (1982), Seniority, Human Capital, and Layoffs: A Case Study. Industrial Relations 21, 352 - 364. Fein, R. (1965), Education Patterns in South Migration. Southern Economic Journal 32, Suppl., 106 - 124. Hashimoto, M. (1981), Firm-Specific Human Capital as a Shared Investment. American Economic Review 71, 475 - 482. Hashimoto, M. / Raisian, J. (1985), Employment Tenure and Earnings Profiles in Japan and the United States. American Economic Review 75,721-735. Herzog, H. W. / Hofler, R. A. / Schlottmann, A. M. (1985), Life on the Frontier: Migrant Information, Earnings and Past Mobility. Review of Economics and Statistics 67,373 - 382. Holmlund, B. (1984), Labor Mobility. Stockholm. Hübler, O. (1984), Zur empirischen Überprüfung alternativer Theorien der Verteilung von Arbeitseinkommen, in: L. Bellmann / K. Gerlach / O. Hübler, Lohnstruktur in der Bundesrepublik Deutschland. Frankfurt / New York, 17 - 189. Johnson, W. R. (1978), A Theory of Job Shopping. Quarterly Journal of Economics 92, 261 - 277. Jorgenson, D. W. (1961), Multiple Regression Analysis on a Poisson Process. Journal of the American Statistical Association 56,235-245. Judge, G. G. / Griffiths, W. E. / Hill, R. C. / Lütkepohl, H. / Lee, T.-C. (1985), The Theory and Practice of Econometrics. New York. ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08 92 Olaf Hübler Lazear, E. P. (1986), Raids and Offer-Matching. Research in Labor Economics 8, Part A, 141 - 165. Lazear, E. P. / Rosen, S. (1981), Rank-Order Tournaments as Optimum Labor Contracts. Journal of Political Economy 89, 841 - 864. Leighton, L. / Mincer, J. (1982), Labor Turnover and Youth Unemployment, in: R. B. Freeman / D. A. Wise (eds.), The Youth Labor Market Problem: Its Nature, Causes and Consequences. Chicago, 235 - 275. Lydall, H. F. (1968), The Structure of Earnings. London. Maddala, G. S. (1983), Limited-Dependent and Qualitative Variables in Econometrics. Cambridge. Miller, R. A. (1984), Job Matching and Occupational Choice. Journal of Political Economy 92, 1086 - 1120. Mincer, J. (1986), Wage Changes in Job Changes. Research in Labor Economics 8, Part A, 171 - 197. Mincer, J. / Jovanovic, B. (1981), Labor Mobility and Wages, in: S. Rosen (ed.), Studies in Labor Markets. Chicago and London, 21-63. Morgan, P. / Manning, R. (1985), Optimal Search. Econometrica 53, 923 - 944. Osberg, L. / Mazany, R. L. / Apostle, R. / Clairmont, D. (1986), Job Mobility, Wage Determination and Market Segmentation in the Presence of Sample Selection Bias. Canadian Journal of Economics 19, 319 - 346. Robinson, C. / Tomes, N. (1982), Self-Selection and Interprovincial Migration in Canada. Canadian Journal of Economics 15, 474 - 502. Shah, A. (1985), Are Wage Incentives and Unionism Important Determinants of Job Tenure? Oxford Economic Papers 37, 643 - 658. Thurow, L. C. (1983), Dangerous Currents: The State of Economics. Oxford. Weiss, A. (1984), Determinants of Quit Behavior. Journal of Labor Economics 2,371387. Weiss, A. / Landau, H. J. (1984), Wages, Hiring Standards, and Firm Size. Journal of Labor Economics 2, 477 - 499. ZWS 109 (1989) 1 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.109.1.75 | Generated on 2023-04-04 12:12:08