Pre-employment vocational training : a case study in South India

UU'WI 1 + Thia paper is prpared for * staff use. The Views are thoge of the author and not necessarily those of the Bank. THE woRLD BANK June 1976 Pre-Eployment Vooational Training_ A Case Study in FO-uth India This is a study of skilled machine tool workers. It indicates that both informal learning-by-doing on the job and formal on-the- job training appears more effective in training better workers than the courses of the vocational institutes. Report No. 10274 Prepared by William P. Faller, Consultant Population and Human Resources Division Development Economics Department 6bv7lopmnt lolicy Staff 671-19v  wSunmry This study examines the relationship of education, training and performance in specific jobs, with a view to determining which paths of education and training appear to be most effective job preparation. The sample includes 474 skilled tradesmen from a machine tool factory in the south of India. The findings show that schooling is related to job per- formance and may possibly be an important prerequisite for on-the-job learning, and that iworkers who received formal on-the-job training or who acquired skills informally by observation and practice are generally better producers than workers who were trained in government sponsored vocational institutes. Nevertheless, a credential from a vocational institute, though associated with lower levels of productivity, appears to ensure more rapid promotion within a trade. The study also notes that vocational training institutes operate at higher cost levels than in-firm training programs.  Pro-E*rpmnt Vocational Truinings A Case Study in South India ISSUES Several observers of education in less developed countries have expressed skepticism about the effectiveness of government sponsored vo- cational schools and trade training institutes, and rightly so. A number of studies have found that these programs commonly suffer from several weaknesses, including old and inappropriate equipment, instructors who are out of touch with modern industrial methods, insufficient time for supervised work experien-e, students who are more interested in advancing in the education system than acquiring a vocational skill, high dropout and absenteeism rates, graduates who end up in jobs which are unrelated to the training provided by the institute, and high cosl:s and low returns compared to other types of secondary level education (Fortin, 1967; Blaug, 1971; Staley, 1970). This study examines another dimension of the vocational training issue, namely the rpla:ionqhip of training and performance in specific jobs. It explores the relative effects of alternative types of job training on the performance and promotion of tradesmen in a large, modern I am grateful for helpful comments from Martin Carnoy, Stanford University, Eugene Staley, who recently retired from the Ford Foundation in India, and Mark Blaug, Institute of Education, London University.  2 factory in south India. Specifically, the study examines three propositions: 1) in-employment training is generally more effective and less costly than pre-employment training provided in vocational and technical institutes outsidc the employtnent system (Staley, 1970); 2) the level of schooling among workers in certain skilled occupations is surpassing that which is actually required for job performance (Berg, 1970); and 3) there is a ten- dency for employers to give excessive attention to type of training credential when selecting workers for promotion. METHIODS Data on job performance, schooling, training, experience, promotion, and socioeconomic variables were obtained from interviews with 474 turners, millers and grinders, and their 35 supervisors. Statistical analysis in- cluded linear regression estimates of job performance as a function of years of schooling and type of t.raining, holding constant physical capitai, job experience and socioeconomic characteristics. The costs of different training prograins were taken from the records of the company and nearby vocational training institutes. Worker efficiency ratings were used as a measure of job performance.2 Log and exponential curves were tried but did not provide better estimates. 2 For each metal cutting job, standard hours had been estimated for the completion of various tasks. These were based on extensive time and motion studies. A worker's efficiency rating was computed by dividing standard hour estimates by actual hours required to complete the taslc. It is important to note that the result:. of all work had to conform to quality standards; if not, a produced piece would be rejected and pro- duced again. In this instance, a worker would be charged with all the hours that he spent on the rejected piece and the new piece. Hence ef- fiency ratings reflect quality as well as quantity of work.  3 To test the reliability of this measure, ratings for workers were taken for three years and correlated. The lowest correlation was .73.1 Efficiency ratings are subject, of course, to error, but it was assumed that error was randomly distributed, affecting all workers equally. If not, unusual shapes would have been found in the distribution of ratings, or notable differences in efficiency means, between different trades or departments in the factory. The advantage of using efficiency ratings is that they are related specifically to a worker's individual output (quality and quantity), and already take into account extraneous factors affecting worker performance such as machine capacity, flow of materials from co-workers, ease of access to tools and instruments, and hardness of raw materials. Alternative measures, such as hourly wages and incentive pay were considered but re- jected. It was clear that basic wages were sensitive to many considera- tions other than performance, for example, seniority. Moreover, there was only a small wage range for workers in the same occupational category. Incentive pay was unsatisfactory partly because it was calculated as a percentage of wages, and partly because it reflected the extent to which workers were fully engaged in productive work; that is, it depended on the regularity of job assignment. THE FINDINGS Trends in schooling. training and unskilled work experience Schooling and formal training are steadily replacing unskilled work experience, a traditional means of acquiring job know-how. At the A perfect correlation would not be expected. Presumably older workers decline in efficiency while younger ones increase.  4 same time there has been a shortening of the overall period of trade pre- paration (training plus unskilled experience). These trends can be ob- served in Table 1. The rise in-years of-schooling undoubtedly reflects a combination of factors, including the rapid expansion of secondary school opportunities, and an upward drift in the education qualifications demanded by employers. The trend in training is part of a generalAmovement in India from informal master-apprentice schemes to formal training programs with prescribed cur- ricula and standards, set durations, regular training staff, and govern- ment sponsorship. Traininu paths Worker training paths were grouped in the following three mutually exclusive categories: 1) Industrial Training Institutes (ITI), a one and a half year government sponsored vocational program for boys with some secondary schooling, sometimes followed by one year of pre-employment apprentice training; 2) In-Firm Training (IFT), any organized training program operated within the company for employed workers; 3) Picking Up Trade (PUT), a situation in which unskilled workers acquire job know-how simply by observing and imitating co-workers and supervisors. ITI represents a form of pre-employment vocational training, while TFT and PUT involve in-employment skill acquisition. The distribution of workers by training path is given in Table 2. The relative influence of schooling,- training and experience on job performan-ce The full results of the regression equations used to test these re- lationships are found in Appendix I. The total variance (R2) explained is not large, but interest here is mainly the relative contribution (the co- efficients) of the variables included in the equation, not the development  5 Table 1 Schooling. Training and Unskilled Experience for Workers in Different Aze Groups 46 and 25 and older 41-45 36-40 31-35 26-30 under Years of schooling 8.0 8.1 9.5 10.2 10.2 10.3 Years of training 1.8 2.4 2.3 2.3 2.4 2.3 Years of unskilled experience prior to becoming a tradesman 9.0 5.4 5.2 2.6 1.5 0.6 Tr,aining plus unskilled experience prior to becoming a tradesman 10.8 7.8 7.5 4.9 3.9 2.9  6 Table 2 Number of Workers by Training Path Pre-employment vocational training 214 Organized in-firm training 227 Picked up trade informally while working as unskilled laborer 33 Total 474  7 of an accurate model for predicting job performance. Table 3 summarizes the coefficients for schooling, training and trade experience (see Appendix I for the coefficients of other variables). Most workers in the factory had 8 to 12 years of formal schooling, that is, some secondary schooling. From Table 3 it appears that variation in the number of years of secondary schooling ma'kes a difference in mean performance levels. Schooling is signiffcantly and positively related to job performance -- workers with more schooling perform better than workers with less. If the mean years of schooling were increased from 10.2 to 11.2 in the equation, worker performance means would rise by about one percentage point. The relationship of experience and job performance is also signifi- cant, though possibly the negative coefficient is misleading because the distribution of years of experience was not normal in the factory. The bkewed distLibution, which reflects a large percentage of older, less productive workers, probably accounts for the negative coefficient. Were the distribution normal, findings would probably show that the relationship of job performance and experience is not linear and that the shape of the curve changes according to level of experience. At least three levels seem likely. Level one would include the first few years on the job, at which time workers acquire specific factory techniques. At this level the curve rises sharply. A second level would be reached when workers reach a plateau of expercise. Coefficients diminish and there is not a discernable relationship between performance and exper;ence. A final level would begin when workers reach an age of declining physical and mental fitness, or are unable to maintain production levels for other reasons. At this point the curve slowly declines and coefficients become negative. The data  8 Table 3 Job Performance as a Function of Schoolin , Exp.rience and Training Years of schooling 1.02* Years of trade experience -0.71* Months of trade training 0.03 Type of training1 Pre-employment vocational training 0.00* Organized in-firm training 6.21* Picked up trade informally while working as unskilled laborer 7.11* Total variance explained 0.23 *Significant at P c,05 The three training paths were treated statistically'as dummy variables. A worker wias given a value of 1 in the category that repre- sented his training path and a 0 in the other two training categories. Therefore the means used in the equation equal the proportion of workers who had followed a particular Lraining path.  * ~~~~~~~~~~~~~~~~~~~~~~~~~~9 presented in Table 3 do not contradict the hypothesis. The negative co- efficient could reflect the fact that the company had not added to its work force in several years and that workers have peaked in expertisc. In regard to training, the relationship between the duration of training and job performance is ind.--ernable. On the other hand, type of training is significantly related. .ne coefficients indicate that workers with in-firm training are more productive than workers from pre-employmert vocational institutes, even when schooling, trade experience, and socio- economic variables are held constant at their means. The differec&ees are marked. If the proportion of workers with organized in-firm training were increased from 48 to 100 percent in the equation, mean performance levels would rise from 61.5 to 64.8, resulting in a 5.4 percent gain in worker efficiency. The survey ef supervisors adds further support to this finding. Nearly 72 percent reported that workers who had received basic trade train- ing in a firm are the most productive. Job performance as a function of combinations of schooling and training Although in-firm training and informally picking up the trade are positively associated with job performance, it could be that in-firm learning is effective only when workers have had an unusually large number of years of .- ondary schooling. Perhaps workers with more schooling are better equipped educationally to learn on the job. To examine this question, workers were divided into two groups representing different levels of 1 schooling attainment: medium (9-11 years) and high (12 yearsand above). Regression coefficients were estimated for each group treating job performance as a function of type of training, trade experience and socioeconomic variables, The coefficients for cach type of training are given in Table 4 (see Appendix II for the full results). 1 It would have been interesting to compare workers-with low levels of education (less than 9 years). However, there were not enough workers from pre-employment vocational institutes in the factory with this level of educa- tion.  10 Table 4 The Relationshi of Job Performance Trainin Path and Schooling l.evels Path Medium Education HiSh Education Pre-employment vocational 0.00* 0.00* Grganized in-firm 4.85* 4.15* learning Pickcd up trade -0.14 13.44* informally while working as unskilled 1aborer Performance means 61.77 69.75 Variance explained .16 .44 *Sianificant at P < .05  11 The data show that in the medium and high schooling groups workers who acquired skills in a firm perform considerably better than workers who learned the trade in pre-employment vocational institutes. It can be in- ferred, then, that irrespective of the schooling levels represented in the factory, workers with in-employment training are more likely to achieve higher levels of performance than workers with pre-employment training. The coefficients for picking up the trade^informally are also interesting to note -- they are small and negative for workers with medium levels of schooling, yet large and positive for workers with high levels of schooling. These results may mean that workers with more years of schooling are in a better position to learn a trade on the job, and more likely to achieve high rates of production. If this is true, schooling could be an important prerequisite for on-the-job learning. Comparative costs The evidence presente' above certairnly raises serious doubt abou pre-employment vocational training, but before drawing any firm conclusions about the value of this program costs should be examined. Although workers with vocational institute backgrounds are poorer producers, it is conceivable that the cost of their training is lower than the cost of in-employment training. If this were found to be true, it is possible that vocational institutes might be relatively efficient from a cost/effectiveness point of view. Costs were divided into direct costs, including staff and materials, and indirect costs, including depreciation and earnings or production fore- gone while workers are involved in training. In-firm training costs were adjusted downward to take into account the benefits that accrue to the company when trainees are engaged in productive work as apprentices.  12 Private costs, other than foregone earnings, were excluded because reliable data were not available. It was assumed that these costs were negligible and would not substantially alter estimates of the relative cost differences between the three training paths. Table 5 summarizes discounted costs per trainee. The results show that pre-employment vocational training is clearly more expensive than training programs offeied by the firm. The difference is accounted for mainly by the longer duration of pre-employment vocational training, as well as the fact that trainees spend less time in productive work while learning (which benefits the company and therefore lowers in- firm training costs). In brief, pre-employment vocational institutes appear not only to turn out relatively inefficient workers, but also to operate at higher cost than in-employment training programs. Recruitment and promotion Level of schooling and type of training are important considerations in screening applicants for jobs. According to the company's management, applicants with a full secondary education and a vocational institute cre- dential are preferred. It is possible, however, that these criteria are not applied uniformly to all job candidates. For example, workers who migrated from rural areas have distinctly higher levels of schooling than workers who grew up in towns or cities. Furthermore, a large proportion of workers wvth rural backgrounds who attended high school, successfully completed that level (66.7 percent for workers from villages compared to only 45.3 percent for workers from cities, significant at P < .01). These findings may mean that only villagers of exceptional ability (compared to urban dwellers) continue with secondary schooling and eventually seek in- dustrial employment, or it may mean that the company restricts its recruitment  13 Table 5 Costs per Trainee (Rupees) Training Path Discounted at 5% Discounted at 10% Pre-employment vocational institute (ITI-ACT) 7,580 6,824 Organized in-firm training 4,550 4,345 Picked up trade1informally while working 0 0 Since we could not find accurate data, a zero cost was assigned to picking up the trade even though there are undoubtedly depreciation, materials and labor costs incurred when workers practice trade techniques and seek advice from supervisors and co-workers. These costs, however, are probably small compared to the costs of the other two training paths.  14 of workers with rural backgrounds to those who are exceptionally well educated. In this latter case, a relatively high level of schooling may be viewed by employers as some sort of compensation or insurance against a disadvantage or risk associated with employing rural migrants. Schooling and training.are obviously important in getting a job. Do these factors also play a part in determining promotion within metal cutting occupations?I According to the firm's manage.nent the answer is no. Candidates who have been in a wage-grade for a prescribed period and have met disciplinary requirements are considered on a merit-seniority basis. Workers interviewed concurred with this view. To examine the empirical relationship of these variables, wage-grade (there were five) was treated as a function of job performance, seniority, education, type of training, and socioeconomic variables. The results are given in Table 6. Table 6 indicates that years of formal schooling have little bearing on promotion. On the other hand, type of training appears to be extremely 2 important. The coefficient for pre-employment vocational training is positive and statistically significant, and this variable accounts for a large portion of the explained variance. Since this coefficient was computed relative to organized in-firm training, it appears that a pre- employment vocational training credential is a much better predictor of high wage-grade than organized in-firm training. 1~~~~~~~~~~~~~~~~~~~~~~~~ The concern here is promotion within an occupational category, not to a higher level occupation such as a supervisory position. 2It should be noted that workers with pre-employment vocational training do not enter the factory at higher wage-grades than workers who were trained on the job. Once unskilled workers have mastered a trade they start at the same wage-grade as workers with pre-employment training credentials.  15 Table 6 Variables Associated with Promotion Variables Coefficient R2 Job performance .009* .030 Seniority .100* .150 Schooling -.019 .002 Type of training: Pre-employment vocational .504* .110 Picked up trade informally -.045 _ Organized in-firm training .000* - Socioeconomic variables .040 Other variables .040 Total variance explained .372 *Significant at P c .05 Computed relative to learning the trade in an organized training program in the firm.  16 Seniority is also related to wage-grade but explains less variance .han expected, probably because of the number of relatively young workers with vocational certificates who are found in the upper wage-grade levels. Another surprise i8 the small coefficient for job performance, but this result may be due also co the preponderance of workers with vocational certificates (who have lower performance levels) in higher wage categories. Surprisingly, socioeconomic variables, including religion, caste, and residence, are not strongly related to wage-grade. In summary, a worker with a formal pre-employment vocational training certificate has an advantage in the promotion system, even though this type of training is associated with lower levels of productivity. CONCLUSIONS AND IMPLICATIONS It has been noted that the level of schooling and the incidence of organized trade training are increasing while traditional means of acquiring trade skills are declining. The desirability of this trend depends on whether more years of education and standardized training produce better workers, and, if so, whether the improvement in worker performance justifies increased schooling and training costs. According to the data, additional years of secondary schooling add to worker performance, and may improve a worker's ability to learn on the job. Therefore Berg's notion that schooling levels are rising beyond what is required for job performance (Berg, 1970) does not seem to apply yet. It has also been seen that type of training is related . job performance: workers who mastered their trade in a firm are more productive than workers who graduated from a vocational institute. The reasons are not clear, but according to supervisors the quality of training offered by vocational isistituLes is inferior to company based  17 training programs. Moreover, the costs of training in vocational institutes are substantially higher per trainee than the costs of in-firm training .programs. In brief, the trend towards more organized training may be desirable if training takes place in the firm, but not if it takes place in pre-employment vocational institutes. The study shows also that workers trained in vocational institutes are promoted more rapidly within a trade than other workers, even though productivity levels of these formally trained workers are lower. It seems, therefore, that management gives undue weight to formal training credentials in selecting workers for promotion. Since this is a case study it is obvious that its results should be treated with caution. Because the study is not based on a representative sample of the total population of workers with different training back- grounds, its results might not be valid beyond the company studied. It could be argued, for example, that the firm recruits the worst of the vocational institute graduates and therefore the mean performance level of these workers is lower than what would be expected from an average vocational institute graduate. A counter argument would be that the company is a high prestige firm which offers good wages and fringe benefits, and provides excellent job security which is valued highly in India. A guess is that the company ranks among the first choices of newly trained tradesmen, and that therefore the workers in the factory represent the best of the voca- tional institute graduates. Assuming that the findings have some general validity, what are their implications for planning job preparation for skilled trades such as turning, milling and grinding? First, as a general rule, it seems that the responsi- bility for administering trade training should shift gradually to industry,  18 but with government supervision of standards. Government sponsored voca- tional institutes which offer pre-employment trade training could be con- verted to other types of educational institutions. There are obvious ex- ceptions, of course; where direct government participation in training would be required, such as in areas of the country where industry is not sufficiently developed to provide adequate training. Second, when training outside the employment system is necessary, training offered might be more flexible and broAdly based, rather than concentrating on a single trade. Trades could be analyzed to identify common tasks which would suggest a core curriculum that all trainees would follow irrespective of the trade they eventually select. Required training that is not part of the core might be 'Left to employing establishments, assisted by the government, or at least deferred, if given in givernment centers, until trainees are fairly certain of employment. The advantage of this approach is that it does not commit a trainee to a particular trade until he is employed or nearly employed, thereby minimizing the risk of waste that occurs when trainees are employed in trades for which they were not trained. Third, the fact that a number of different training paths were ob- served in the study probably reflects a diversity of resources that firms and other agencies can or are willing to allocate to training, as well as wide differences in the education and environmental backgrounds of trainees. Perhaps an array of training schemes (for example, programs with different degrees of lourmal training and supervised work experience) is necessary to accomodate varying needs of employers and trainees. This argues for the support of multiple, locally tailored programs -- not only the support of one or two standard national schemes.  19 Finally, the support and supervision of multiple approaches to trade training would require a well organized administration that could study local manpower needs, the training potential of local firms, ensure chat national training *tandards are followed, and provide direct training assistance if needed. Perhaps a decentralized form of Staley's concept of an occupational training organization (Staley, 1970), managed by local re- presentatives of firms, unions, schools, training institutes and other relevant government organizations, could serve.this purpose. In sunmcry, the study provides further evidence that in-employment training is more effective and less costly than pre-employment vocational training in preparing workers for certain trades. The continued development of vocational institutes should be viewed with circumspection by education planners, and administrators of international assistance organizations.  Seleeted Bibliograp Alexander-Frutschi, Marian, ed. Occu_ational Education and Training for Development: Account of an International Conference. Stanford International Development Education Center, Stanford University, 1968. Berg, Ivar. Education and-Jobs: The Great Training Robbery. Center for Urban Education, New York, 1970. Blaug, Mark, Layard, P.R.Q. and Woodhall, M. The Causes of Educated Un- Employment in India. London, 1964. Blaug, Mark. The Rate of Return to Investment in Education in Thailanid. Bangkok, 1971. Eninger, M.U. Process and Product of Trade and Industrial 1ligh School Level Vocational Education in the United States, Vols. I and II, Educational Systems Research Institute, Pittsburgh, 1968. Fortin, B. "Education and Training Programmes." Human Resources for Industrial Develo-ment. ILO, Geneva, 1967. Fuller, W.P. Education, Training and Productivity: A Study of Skilled Workers in Two Factories in South India. Stanford International Development Education Center, Stanford University, 1970. A condensed version of this study appeared in Manpower Journal, Vol. VIII, No. 1, April-June, 1972, Institute of Applied Manpower Research, Delhi. Horowitz, M.A. and Herrnstadt, I.L. "The Training and Education of Tool and Die Makers," Proceedings of the Industrial Relations Research Association, Winter, 1967. Lambert, Richard D. Workers, Factories and Social Change in India. Columbia University, New York, 1968. Maton, J. "Experience on the Job and Formal Training as Alternative Means of Skill Acquisition: An Empirical Study," International Labour Review (September, 1969), Geneva. Mincer, J. "On-the-Job Training: Costs, Returns and Some Implications," Journal of Political Economy, Vol. LXX (October, 1962), Part 2. Staley, Eugene, Planning Occupational Education and Traininz for Development. Orient Longmans Ltd., New Delhi, India, 1970.  Appendix I The Rclationship of Job Perfor;-.ance, Schooline, TraininR, Work Experience and Other Variables Variabl!js Training Pre-employment vocational training -- * Formal in-firm training 6.21* Picked up trade informally 7.11 Years of schooling * 1.02 Months of training 0.03 * Years trade experience -0.71 Wage grade 5.57 SES -1.02 job satisfaction -1.20 Job aspirations 0.27 Religion 2/ Catholic -3.78 Protestant -0.78 Muslim 3.58 Caste 3/ Brahmin -2.17 Lingyat -3.16 Kshatriya Vaisya Schedule -3.58 Nqoncaste -3.84 Environment 4 Farm 2.11 City -0.31 Constant 53.90 Total variance explained .23 *Significant at P L .05. Relative to pre-employment vocational training. Relative to Hindu. Relative to Sudra. Relative to town.  Appendix II Relationship of Job Performatice ant Tvpe of Training for Workers with Different Levels of Schooling 9-11 years 12 or more years Variables - _ Pre-employment vocational training 0.0* 0 0 Formal in-firm training 4.85*0 4.15* Picked up trade informally -0.14 13.44* SES -0.81 -3.42 .2 Religion Catholic -1.51 -28.5;* Protestant -0.78 -4.83 Muslim 3.47 4.21 Caste3 Brahmin -3.49 5.53 LingayaL -8.29 11.59* Kshatriya 0.38 7.91 Schedule -1.46 -6.88 Noncaste -5.80 0.10 Childhood residence4 Farm 4.79 -10.80 City -2.22 3.32 Constant 66.24 77.71 Total variance explained .16 .44 Job performance means 61.77 69.75 *Significant at P c .05. 1 Relative co pre-employment vocational training. Relative to Hindu. Relative to Sudra. 4 Relative to town.