The Quality of Skill Acquisition in Young Workers’ First Job

Francis Green – Scott M. Montgomery Abstract. Young workers’ experiences of their first job form an important stage in the transition from school to work, and the quality of that experience is likely to have a strong bearing on future chances in the labour market. We provide evidence on a key aspect of that experience, namely, whether the skills acquired during the first job are firm-specific or in some degree transferable. More than one in six young people acquired only firm-specific skills in their first substantial job. High levels of prior human capital, short training spells or training leading to qualifications tend to raise, while being in casual or temporary jobs tends to lower, the likelihood that transferable skills are acquired.

1. Introduction Young workers’ experiences of their first job form an important stage in the transition from school to work, and the quality of that experience is likely to have a strong bearing on future chances in the labour market. In Britain, and indeed in any country where substantial proportions of young people quit full time education by age 16, early job experiences are especially important if they are to provide routes to skill acquisition alternative to schools or colleges. Access to company training has been found, for example, to have notable benefits for future wages both in Britain and elsewhere.1 Relatively little is known, however, about what skills young people pick up from their early employment experiences. In this paper we provide evidence on a key aspect of that experience, namely, Francis Green, Department of Economics, University of Kent, Canterbury CT2 7NZ, Scott M. Montgomery, University Department of Medicine, Royal Free Hospital School of Medicine. Approved by the Editorial Board on May 13, 1998 LABOUR 12 (3) 473–487 (1998)

JEL J24

© Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.

474

Francis Green – Scott M. Montgomery

whether the skills acquired during the first job are firm-specific or in some degree transferable. The theoretical significance of the dichotomy between firmspecific and transferable skills has been recognised at least since Becker (1964). For employees, transferable skills are preferable because they raise the outside wage and hence, also, the wage they can negotiate with their current employer. For this reason, in a perfectly competitive labour market employers would only fund training for firm-specific skills since only then could they expect to receive a return in the form of a marginal product above wages. Where, however, there are labour market frictions, or where there is asymmetric information about workers’ skills, employers retain an incentive to fund transferable skills acquisition (Katz and Ziderman, 1990; Stevens, 1994). But whoever does the funding, the circumstances are comparatively uncommon in Western capitalism where the individual employee could be content with acquiring only firm-specific skills. Similarly for society as a whole, positive labour mobility implies that the greater the extent to which the labour force’s skills are firm-specific, the quicker those skills are eroded and need replacing. Whether the skills are transferable or firm-specific, then, is an important dimension of skills quality. Despite the theoretical importance of the firm-specific/ transferable dichotomy, empirical work has typically had to be indirect, for want of a direct measure of this aspect of skills quality. A typical assumption is that the length of job tenure is a proxy for the accumulation of skills (sometimes the relation might be modelled explicitly, in a way that allows the stock of specific skills to drop out of any estimating equation). Similarly, inferences about the impact of skill-specificity are derived from wage-tenure profiles, or from prior information about the structure of internal labour markets. For example, it is often assumed that Japanese companies invest proportionately more in specific skills, compared with US companies. However, the basis of this assumption is essentially that job mobility is much lower in Japanese companies (Hashimoto, 1993, 1994; Abe, 1994), rather than any independent evidence that the technical skills used in Japanese companies are inherently firmspecific. Alternatively, it has been argued that there are gender differences in the propensity to acquire general and specific skills, and that these can account for differences in the wage-tenure profiles of men and women (Becker and Lindsay, 1994). Some studies infer patterns of skill-specificity from wage changes resulting from job transitions. For example, Ong and Mar (1992) traced the © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

The Quality of Skill Acquisition in Young Workers’ First Job

475

earnings of semi-conductor workers in their new jobs after being laid off. Those who found jobs in the same industry earned no less than those who went back to the same firm after the lay-off, but those who went into different industries did suffer a loss of earnings. From this the authors concluded that these workers’ skills were largely industry-specific, but not firm-specific. Finally, it is not unusual for the distinction between on-the-job and off-the-job training to be used to infer how transferable are the skills being acquired, even if the connection is theoretically loose. It would, however, be advantageous to be able to investigate the degree of skill-specificity more directly, rather than rely only on indirect inferences. On one hand, staying several years in one job is hardly a guarantee that primarily firm-specific skills are being acquired (as any academic could testify from personal experience). Making inferences from wage profiles is potentially open to a problem of tautology, namely that since it assumes an impact of skills transferability on wages, that cannot then be used to explain wages. Moreover, it is also common for on-the-job training to impart skills potentially useful to other employers, or for off-the-job training to be custom-designed for particular employers. In this paper we utilise a survey to investigate, directly, individuals’ own perception as to how transferable are the skills they acquired in their first job. This provides a subjective measure, which like other subjective measures is open to the potential problem of self mis-perceptions and to responses coloured by social desirability bias. We have no independent test or examination to verify the individuals’ perception. Nevertheless, we find that the responses are plausible in those cases where something is commonly known about the nature of the job. And, in any case, we argue that a direct approach to this issue, even if subjective, provides a useful contrast to the indirect approaches mentioned above. We are concerned in particular with the quality of the first job experience of young people. Given that many young people in Britain leave school before 18, do they make up for missing those years of education by undergoing useful work experiences, useful in the sense that they come out of that first job with something that can be applied elsewhere in the labour market? What factors govern the quality of that experience? In particular, we shall be interested to examine whether the different types of training that are part of that experience contribute to a perceived improvement in transferable skills. After describing the data in Section 2, Section 3 sets out a simple © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

476

Francis Green – Scott M. Montgomery

theoretical framework and methodology for analysing the quality of the skills acquired in the course of a job. The results are reported in Section 4 and Section 5 concludes. 2. The data We utilise survey data drawn from a cohort of residents of Great Britain born between April 5 and 11, 1970, known as BCS70 (Ekinsmyth et al., 1992). The study began as the British Births Survey, which collected information on all births which took place in the target week. The original objective was to collect obstetric and gynaecological information on a representative sample of the British population. Subsequent sweeps of data collection for the entire cohort took place at ages 5, 10 and 16 years. For these later sweeps, the information collected spanned all of the major life domains: education, development, health, demographic, social and economic characteristics. The survey conducted at age 21 years, in early 1992, was restricted to a randomly selected sample of 1,650 cohort members resident in England and Wales.2 The main aim of this survey was to collect information on young people’s transition from education to employment and on literacy and numeracy difficulties in a young population. Information was gathered using interviews conducted by professional interviewers and by self-completion questionnaire. Respondents were questioned in detail about their first job, after leaving full-time education, that lasted at least three months. We excluded holiday jobs and any other jobs undertaken during fulltime education. To gauge the quality of skills acquisition, and in particular the skill-specificity of the jobs, respondents were asked: ‘‘Would you say that the skills you have gained in this job are useful? Which of the statements on this card best describes the skills you have gained in this job?’’ On the card shown were four possible mutually exclusive responses: ‘‘Only useful in this particular job’’, ‘‘Only useful with employers in the same line of work’’, ‘‘Also useful in jobs with quite different lines of work’’ and ‘‘Don’t know’’. In what follows, we refer to the first response category as ‘‘firmspecific’’. Those people who learn only the particular aspects of the organisation they are working for would be included in this group, as well as those who learn, for example, about machinery that is entirely specific to the firm.3 The next two categories we can think of as, respectively, ‘‘industry-specific’’ and ‘‘general’’, and refer to them © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

477

The Quality of Skill Acquisition in Young Workers’ First Job

together as ‘‘transferable’’ skills. As Row 1 of Table 1 indicates, some 18.3% off respondents reported acquiring only firm-specific skills, while 43.0% reported industry-specific skills and the remainder general skills. Only a small proportion (1.4%)

Table 1. Quality of skill acquisition during first job Percentage of employees Only firm-specific Some industry- Some general skills* specific skills* skills* ALL Males Females Occupations: Managers, administrators & professionals Associate professional & technical Clerical & secretarial Craft & related Personal & protective service Sales Plant & machine operatives Other occupations Industries: Agriculture, forestry & fishing Energy & water supplies** Extraction of minerals & ores other than fuels; manufacture of metals, mineral products & chemicals Metal goods, engineering & vehicle industries Other manufacturing industries Construction Distribution, hotels & catering (repairs) Transport & communication Banking, finance, insurance, business services & leasing Other services

18.3 17.3 18.7

43.0 47.3 39.1

38.7 35.4 42.2

3.2

29.0

67.8

8.6 8.8 11.0 31.3 18.8 34.1 12.5

45.7 33.2 52.4 43.8 44.9 47.7 10.5

45.7 58.0 35.6 24.9 36.3 18.2 77.0

33.3 {3} 3.6

44.4 {2} 35.7

22.2 {3} 60.7

17.2

42.5

40.3

29.2 12.6 18.6

44.2 51.6 44.8

26.6 35.8 36.6

12.8 7.4

31.9 39.4

55.3 53.1

23.3

41.3

35.4

Notes: *See text for definitions of firm-specific, industry-specific and general skills. **Absolute numbers. © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

478

Francis Green – Scott M. Montgomery

answered‘‘Don’t know’’ and these cases were excluded from the analyses. Rows 2 and 3 show only a very small difference between males and females in respect of firm-specific skills, but a somewhat greater proportion of males were acquiring industry-specific skills. A tabulation according to broad occupational group is given in subsequent rows of Table 1. The general conclusion is that firmspecific skills development is much more common in the lowerstatus occupations, that is, in the manual occupations apart from craft occupations. Most notably, plant and machine operatives are acquiring the least transferable skills. This might seem to go against the oft-held notion that unskilled workers form part of a flexible workforce, able to switch with little cost from job to job. But there is no real contradiction. These workers are acquiring very few skills during the course of their work, and their general ability, for example, the physical fitness necessary for heavy manual work, is something that is brought to the job in the first place. At the other end of the scale, only a very small proportion (three percent) of managers and professionals perceived themselves to have acquired no more than firm-specific skills, and more than two-thirds thought they had acquired general skills. Analysis of the degree of firm-specificity in more disaggregated occupational groups provides some confirmation that respondents interpreted the question correctly, in that there is a correspondence between a priori notions about which occupations are likely to develop transferable skills and which specific skills. For example, amongst secretaries, personal assistants, typists, and word processor operators, no respondent recorded firm-specific skills; while among those in the Armed Forces some 27% reported firm-specific skills, well above average. Moreover, among manual occupations we find, as should be expected, a relatively high degree of transferability in the skills being developed in the craft occupations. Some further corroboration that the responses are correctly gauging something about skills quality can be gleaned by considering where respondents go to with their skills. Of those reporting only firm-specific skills, some 67.6% will have gone to other jobs or stayed in the same job till age 21, but for those who acquired transferable skills, the figure was 78.4%. By contrast, 20.3% of the firm-specific skills group went into unemployment compared to only 11.5% of those with transferable skills.4 Table 1 also gives a breakdown of skill quality according to industry. Skill specificity is relatively high in ‘‘Other Manufacturing’’ industries. These are mainly the low-paying sectors of © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

The Quality of Skill Acquisition in Young Workers’ First Job

479

manufacturing, such as textiles and apparel. Also high, in terms of firm-specific skills, were agricultural workers. By contrast, in the extraction, metal manufacture and chemical industries, only very few respondents reported that their acquired skills were firm-specific, and most thought that they had acquired general skills. Also providing mostly transferable skills were the Banking and Finance industries. This initial examination of the data suggests, therefore, that there is a link between the nature of the job and the quality of the skills perceived to be acquired by respondents: broadly speaking, the ‘‘better’’ jobs (that is, in the higher status occupations, or in the higher-paying industries) also convey the better quality of skills. 3. Theory and estimation method We now turn to the analysis of what factors determine the quality of the skills gained during the first job. We assume that additional skills are an outcome of the job experience. Let Q* represent a latent quality index, measuring these skills increments. Q* is determined by the inputs to the job experience, just as in a production function. First, prior human capital and other personal attributes denoted by a vector, Xi , are likely to affect how people learn from their job experience. We expect that, other things equal, the greater the human capital the greater the facility with which new skills may be acquired. In part, this expectation is based on a voluminous literature that has shown the complementarity between education and training (Ashton and Green, 1996, chapter 3). Also, one can think of school-based qualifications as a signal indicating the ability to learn. Second, there are a number of observable job-experience attributes, denoted Yi , including whether training was provided, which have a direct bearing on the quality of skills acquired. On one hand, training courses might be expected to add to skills quality. On the other hand, one might expect other job features to be complementary with skills acquisition quality, in terms of their attractiveness to individuals: i.e. ‘‘good’’ jobs offering both transferable skills acquisition and other positive attributes. Third, there is the length of time, ti , in the job itself. A longer time allows scope for more and better skills acquisition. Fourth there will be unobservable job attributes, Zi*, which impinge on the skills acquired. Thus we write: Qi*\f(Xi , Yi , ti , Zi*)

(1) © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

480

Francis Green – Scott M. Montgomery

The unobserved job qualities might be thought of as random error, in some versions of a matching model of the youth labour market. It would seem more appropriate, however, to assume that the individuals concerned have some knowledge of job quality, in terms not only of current wages but also of future prospects. A job-queuing model assumes access to the jobs to be determined by prior human capital attributes, and this is the approach we adopt here. In other words, we assume quite simply that: Zi*\g(Xi )

(2)

The response to our question described above is taken to be the observed variable corresponding to this latent quality index. For this purpose, we are primarily concerned only with whether the acquired skills are in some degree transferable. Define S* as the negative of the latent quality index: Si*\µQi*

(3)

Then, after assuming a linear specification of (1), the model becomes: Si*\a0+a1Xi+a2Yi+a3ti+ui Si\1

for Si*E0

(firm-specific only)

Si\0

for Si*R0

(some transferable)

(4)

In effect, this is a reduced-form model of the firm-specificity of skills. We assume that ui takes the logistic form, and estimate (4) using the standard logit method. To economise on space, we discuss our detailed expectations along with the results. 4. Results Table 2 presents the means of the covariates used in the analysis. After the exclusion of cases with missing values, we were left with 1,184 observations for the analysis. Table 3 gives the results. We first estimated the logit excluding the length of time in the job — see column (1). The coefficient estimates provide some weak confirmation of the expectation that greater human capital acquired prior to the job raises the probability that the first job will convey further transferable skills — either directly because prior skill enables individuals to acquire transferable skills more readily or indirectly because the prior skill enables those individuals to attain jobs with better (unobserved) characteristics. Thus, having a limiting © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

481

The Quality of Skill Acquisition in Young Workers’ First Job Table 2. Variable descriptions a Description

Meanb

Prior Human Capital

ILEA1 ILEA2 ILEA3 ILEA4 ILEA5 INDEP ALEV LIMDI Other Personal Characteristics SEX PARSOC

Score of school qualifications up to age 16 years divided into 5ths (with mean score) Lowest quintile (3.2) Second quintile (15.7) Third quintile (23.0) Fourth quintile (31.8) Top quintile (46.65)

0.20 0.20 0.20 0.20 0.20

Attended private school At least one A-level Disability/chronic illness that limits daily life

0.04 0.12 0.02

1\male Parent in Social Class 1 or 2

0.47 0.22

Human Capital Acquired in Job SHORT_TRAIN Number of training spells of less than 3 days TRAIN_QUALS Training at work designed to lead to qualifications (not YT) – 3 days+ YT_QUALS YT at work designed to lead to qualifications YT YT at work not designed to lead to qualifications TENURE Number of months in first job

28.3

Other Job Characteristics PRIVATE CAS/TEMP FIXED UNEMPLOYMENT

0.86 0.08 0.07 11.9

Workplace Size (number of workers; baseline s\10) 11–499 500+

Job in private sector Casual or temporary work Fixed period job Mean local unemployment rate (%)

1.30 0.18

0.05 0.02

0.59 0.12

Notes: a Except in the cases SHORT-TRAIN, TENURE and UNEMPLOYMENT, all variables are 0/1 dummies. b The base is the 1184 cases with no missing values for any variable used in the analysis.

© Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

482

Francis Green – Scott M. Montgomery

Table 3. Logit estimates of the determinants of skill specificity Prior Human Capital ILEA1 ILEA2 ILEA3 ILEA4 INDEP ALEV LIMDIS Other Personal Characteristics SEX PARSOC Human Capital Acquired in Job SHORT-TRAIN TRAIN-QUALS TRAIN YT-QUALS YT

(1)

(2)

µ0.050 (0.296) 0.547* (0.289) 0.028 (0.296) 0.087 (0.289) µ0.863 (0.567) 0.041 (0.304) 0.825* (0.503)

µ0.029 (0.297) 0.561* (0.290) 0.055 (0.298) 0.130 (0.290) µ0.903 (0.567) 0.026 (0.305) 0.818 (0.505)

0.105 (0.162) µ0.325 (0.218)

0.088 (0.163) µ0.337 (0.218)

µ1.208** (0.236) µ0.437* (0.260) 0.168 (0.343) µ0.436 (0.404) µ0.561 (0.612)

µ1.167** (0.237) µ0.411 (0.261) 0.234 (0.347) µ0.405 (0.406) µ0.578 (0.615) µ0.007* (0.004)

TENURE Other Job Characteristics PRIVATE CAS/TEMP FIXED UNEMPLOYMENT Workplace Size 11–499 500+ CONSTANT µ2 Log likelihood Model Chi squared Likelihood ratio index No. of cases

0.3929 (0.270) 0.5700** (0.267) 0.093 (0.330) 0.073** (0.032) µ0.231 (0.177) 0.397 (0.276) µ2.498** (0.559) 1018.149 88.59 (p\0·00) 0.080 1184

0.388 (0.271) 0.551** (0.268) 0.033 (0.332) 0.072** (0.032) µ0.231 (0.177) 0.431 (0.278) µ2.317** (0.568) 1015.025 91.72 (p\0·00) 0.083 1184

Standard errors in parentheses; * indicates significant at the 10% level, ** at the 5% level. The dependent variable is SI , a dummy variable indicating acquisition of only firm-specific skills. © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

The Quality of Skill Acquisition in Young Workers’ First Job

483

prior health problem lowers the quality of the first job experience. Having been at an independent school attracts a sizeable negative coefficient, suggesting a reduction in the chance of receiving only firm-specific skills, but the standard error is large and the coefficient is not quite significant at the ten percent level. Those in the second lowest quintile of school qualifications are significantly more likely than those in the highest quintile to report acquiring only firmspecific skills. There is, however, no significant effect from the lowest quintile.5 Nor does possession of an A-level have any effect. Only a few cases fall into this category — those who took A-level but chose not to go through to university education. Those who went to university were excluded from our sub-sample because they would not have had any employment before the interview date. Of the other personal characteristics, having a parent in either of the top two classes, attracts a negative coefficient with a p-value of 13.6%. This, too, could be interpreted broadly as a human capital effect. Interestingly the impact of gender on the quality of skills acquisition is small and very insignificant. Turning next to aspects of the job experience itself, we investigated whether training undertaken during the period of the job added to skills quality. This was confirmed in the case of training courses of three days: those who went on one or more such courses were more likely to have acquired transferable skills. Also, the greater the number of shorter training courses, each lasting less than three days, the more likely the skills will be transferable. However, the other two types of training had no significant effect. Participation in a Youth Training programme at work designed to lead to qualifications carries a negative coefficient, but this is significantly different from zero. And those who went on one or more training course of three days or more, not leading to a qualification, were also no more nor less likely to receive transferable skills.6 The last set of variables entered in column (1) consists of job attributes which, while not directly concerned with human capital acquisiton, might be expected to have an indirect effect by altering the quality of the job experience. We expected that temporary jobs would not convey such good skills as permanent ones, if only because employers are less likely to want to develop their temporary or casual workers. This was confirmed, in that having a temporary or casual job significantly raised the probability of only acquiring firmspecific skills. Being in the private sector worked in the same direction but the effect was not quite significantly different from © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

484

Francis Green – Scott M. Montgomery

zero at the ten percent level. Living in a high unemployment area significantly reduced the chances of gaining transferable skills from the job. One interpretation would be that employers have less need in a slack local labour market to offer the prospect of acquiring transferable skills in order to attract labour. Finally, neither workplace size, nor having a fixed period contract, had any notable effect. Column (2) reports the results of a logit estimation including job tenure, ti . Although the production function framework we utilise here suggests that job tenure will have a positive effect on the quality of skills acquisition, job tenure is highly likely to be endogenous. On one hand, employees who find that they are only picking up firm-specific skills will, ceteris paribus, wish to leave for other jobs with better prospects. On the other hand, those with transferable skills are likely to be more attractive to other employers. The two effects work in opposite directions, but either way the coefficient estimate could be picking up a reverse causation. There are no plausible instrumental variables in the data which might be used in a separate equation estimating job tenure but excluded from the logit estimation. Hence, we proceed simply by presenting equations with and without job tenure. In the event, job tenure has a significant positive effect. In light of the above, we interpret this as either that the longer you stay the more likely you are to acquire transferable skills, or that if you are only acquiring firm-specific skills you are more likely to quit. The inclusion of job tenure has no substantial effects on the coefficient estimates for other variables; however, the human capital variables LIMDIS and TRAIN-QUALS no longer quite reach the ten percent significance level.7 5. Conclusion We have argued the case for utilising the direct survey method to measure the extent to which the skills imparted in the course of a job are firm-specific or transferable. Despite the usual qualifications to be made about subjective responses, there is no good case for ignoring individuals’ self perception of the skills they are acquiring; moreover, the indirect methods of inferring skill transferability leave something to be desired. With our data set, we have not been able to cross-check our skills quality variable against wage-tenure profiles, but the relative degrees of specificity at least conform to a © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

The Quality of Skill Acquisition in Young Workers’ First Job

485

priori notions about different occupations. The positive correlation we have found between skills transferability and job tenure casts some doubt on the oft made assumption that those with long job tenure are only acquiring firm-specific skills. There are often important institutional constraints and incentives limiting worker mobility that have nothing to do with the type of technical skills used in jobs. For our sample of young workers, the fact that more than one in six perceive that they acquired only firm-specific skills in their first job (with a mean length of just over two years) is discouraging. For these people, leaving full-time education might at the time have appeared a rational course of action — we have no measure of the wages received in the first job, which could have had a major bearing on the incentive to quit school. But the longer term consequence of quitting education with low qualification levels and then failing to acquire soon the skills necessary to move around in the world of work, are likely to have been detrimental. Nor could we reject on our evidence the hypothesis that Youth Training Scheme participation failed as a source of transferable skills. The evidence suggests, rather, that what is needed is employer training which leads to qualifications and simply more of those short (under three days) training spells. Our data do not allow us to pinpoint just how important are the consequences of that first job experience, as we do not have details of all the subsequent jobs our respondents undertook. No doubt some will have left their first job for another with better prospects for skill acquisition. Moreover, we cannot be sure that the first job experience is of no value to other employers even when the reported skills are firm-specific. Employers have stated in any number of surveys that they place a value on personal attributes like integrity and reliability, qualities which may not be regarded by the individuals themselves as skills. Such attributes can be demonstrated through a job record and can be conveyed to future employers through references, interviews and other recruitment procedures. Nevertheless, our evidence points to the existence in late 1980s Britain of a number of bad jobs for youths, in which no transferable skills are acquired, offering an unattractive route through from the world of school to that of work. This finding highlights the dangers of a low-participation education system, if it is not accompanied by a universal apprenticeship system, or something similar, to fill the skill acquisition needs of young workers. By the same token, recent advances in education participation are thereby to be welcomed. © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

486

Francis Green – Scott M. Montgomery

Notes 1

See Ashton and Green (1996), Chapter 3, for a review. The 21-year-old survey has also been checked for consistency with estimates derived from the Labour Force Survey (Green et al., 1994). 3 We considered whether to allow respondents to say that they had picked up no skills whatsoever from the job, but resolved that, while skills should be considered sufficiently broadly to include for example local knowledge about the workplace, everyone will have acquired at least some firm-specific skills. Hence the item stem was used merely as a device to get respondents to start thinking about skills. Consider those individuals who felt that they had learned nothing in the way of transferable skills. If they nevertheless perceived their acquired local knowledege as some sort of skill, we expected that they would respond in the first category. But if they did not regard their local knowledge as a skill, they would respond in the ‘‘Don’t Know’’ category. There were, in the event, very few in this catregory. Moreover, the feedback from interviewers at the pilot stage was that this question was being undertstood in the ways expected by the researchers. 4 In reporting these differences, we are not proposing a causal model with transferability affecting job transitions, which could not in any case be estimated with the available data. Rather, we suggest simply that since those workers reporting transferable skills acquisition are subsequently making the better job transitions, it remains plausible to see the responses as a measure of skills quality. 5 It is possible that there may be greater reporting bias, associated with a boosting of self-esteem for this group. 6 These findings must be qualified by the possible presence of the familiar problem of endogeneity of training participation. There was no plausible way with the data of securing identification in a model that included a training selectivity equation. If unobserved characteristics leading individuals to acquire transferable skills also predispose those individuals to be selected into training, the coefficients are likely to overstate the impact of training on skills acquisition. 7 We also experimented with runs incorporating industry dummies. The dummies were, however, not significant. 2

References Abe, Y. (1994) ‘‘Specific Capital, Adverse Selection, and Turnover — a Comparison of the United States and Japan’’, Journal of the Japanese and International Economies 8(3): 272–292. Ashton, D. and Green, F. (1996) Education, Training and the Global Economy, Cheltenham: Edward Elgar. Becker, E. and Lindsay, C. M. (1994) ‘‘Sex-differences in Tenure Profiles — Effects of Shared Firm-specific Investment’’, Journal of Labor Economics 12(1): 98–118. Becker, G. S. (1964) Human Capital, New York: National Bureau of Economic Research. Ekinsmyth, C., Bynner, J., Montgomery, S. and Shepherd, P. (1992) An Integrated Approach to the Design and Analysis of the 1970 British Cohort Study and the National Child Development Study, City University, London: SSRU. © Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.

The Quality of Skill Acquisition in Young Workers’ First Job

487

Green, F., Hoskins, M. and Montgomery, S. (1994) The Effects of Training, Further Education and YTS on the Earnings of Young Employees, University of Leicester, Discussion Papers in Economics. Green, F., Hoskins, M. and Montgomery, S. (1996) ‘‘The Effects of Training, Further Education and YTS on the Earnings of Young Employees’’, Oxford Bulletin of Economics and Statistics 58: 471–488. Hashimoto, M. (1993) Education, Training, and Economic Performance: Postwar Japanese Experience, Human Capital Investments and Economic Performance, RAND/EAC/IET, Santa Barbara. Hashimoto, M. (1994) ‘‘Investment in Employee Relations in Japanese Firms’’, Journal of the Japanese and International Economies 9(1): 75–95. Katz, E. and Ziderman, A. (1990) ‘‘Investment in General Training: the Role of Information and Labour Mobility’’, The Economic Journal 100 (September): 1147–1158. Ong, P. M. and Mar, D. (1992) ‘‘Postlayoff Earnmings Among Semiconductor Workers’’, Industrial and Labor Relations Review 45(2): 366–379. Stevens, M. (1994) ‘‘A Theoretical Model of On-the-job Training with Imperfect Competition’’, Oxford Economic Papers 46(4): 537–562.

© Fondazione Giacomo Brodolini and Blackwell Publishers Ltd 1998.