Some findings about the unemployed highly educated persons in Pakistan.

Khan, Shahrukh Rafi ; Ali, Syed Zahid

INTRODUCTION

Two aspects of the problem of the unemployed educated persons are discussed in this essay. Firstly, the magnitude and incidence of the unemployment of such persons are examined. One point that becomes apparent from looking at the second-ary data is that the bulk of the educated unemployed persons have been among those less than thirty years of age. Thus it appears that, at least in the past, most of the highly educated persons eventually got absorbed in the labour force. Secondly, in the light of the above, the important problem that comes to the fore is that of waiting. The results of an analysis of survey data, particularly on this dimension of the unemployment of the educated, have been reported here.

This essay is in two sections. In the first section, the data sources are described and their limitations pointed out. The method used as an organizational device for the presentation of information contained in the survey responses is also mentioned. The second section contains our findings. In this section the magnitude and incidence of the unemployment of the educated are also reviewed. Special emphasis is placed on an analysis of post-graduate unemployment by socio-economic background. We report on our assessment of the extent to which the unemployed are themselves responsible for being unemployed and also on the other causes of the variation in the waiting period or job-search.

DATA AND METHOD

Two sources of data, one secondary and one primary, are utilized. Both are described here in some detail since the quality and reliability of these data determine how seriously the results can be taken.

The secondary data source used is the 1981 Population Census [3;7]. In fact, the 1982-83 Labour Force Survey [8] could have been used, instead of, or along' with, it. However, this option is not taken up for three reasons. Firstly, the Labour Force Surveys do not report unemployment by age and gender disaggregations, as is done in the Population Census. Secondly, the labour force unemployment data collection is on the basis of "current status". This means that those respondents who were unemployed in the reference week show up as such in the data. As opposed to this, the census unemployment data collection is based on "usual status". The latter definition in principle seems more reasonable to us. Thirdly, the population census data are in absolute amounts and therefore much easier to work with than the Labour Force Survey data which are in percentages. This reason is not as trivial as it may seem because one often has to rely on outside sources, such as the population census, to generate base estimates.

As part of its Medium-term Plan (1978-83), the International Institute of Educational Planning (UNESCO), in connection with the concerned Ministries of Education, sponsored 21-country-studies to analyse the relationship of education and unemployment. Pakistan was among those countries. Data were collected, pertaining to various populations including the educated unemployed persons. For this paper, the files of the educated unemployed persons and the educated employees were utilized. Unfortunately, only those with post-graduate degrees were sampled.

The details of the survey sampling are contained in the country report [2, sections 1.4-1.6]. The returns for the employees numbered 2,671, constituting a 60 percent response rate. The sample size of the unemployed was 260. An attained sample of 260 may appear small. However, quite to the contrary, it was large considering that the stock of unemployed post-graduate persons for 1982-83 (the years of the survey) was 1,492. This number was derived from the Labour Force Survey [8, p. 178].

Unfortunately, there was no sampling frame from which to identify unemployed post-graduates. To start with, educational institutions would have been a costly procedure. This would have entailed selection of all the universities for a particular year. Then, a sample for all the enrolled students for that year would have to be traced and the unemployed identified.

The alternative was to rely on the "live files" of the employment exchanges. This is fruitless because while post-graduates do register with employment exchange offices, the latter do not directly get demand notices for the post-graduate or even the graduate level. In view of this, the nature and extent of registration with these offices are likely to be affected.

Field investigators were encouraged to identify the unemployed through contacts established at higher educational institutions. Of those identified, a response of 42 percent was attained. Although a large enough sample was attained, the findings from this procedure can only be suggestive. The large sample and the nature of the information collected did encourage us to proceed with data analysis. The importance of the topic and the fact that little research has been done on it to date in Pakistan lent further encouragement. The data for the employees group were, however, generated from a probability sample using standard stratifying procedure for wider representation.

The method used to analyse the secondary data is straightforward. However, the method used to analyse the primary data was more complex. This is because the questionnaire responses cannot simply be taken at face value. Thus the responses were used only as a means for ascertaining the subjective preferences and attitudes of the unemployed concerning why they were not at work. As a check on these perceptions, more objective information available from outside sources was also used.

This juxtaposition of subjective perceptions and objective information was used to gauge two 'features' that could determine the success of an individual on the job market. These features are market realism and market power.

Market realism means that the expectations of prospective job Candidates is in concordance with the prevailing market conditions with regard to, for example, wage and occupational expectations. Market power can be viewed as resulting from competitive or non-competitive factors. The non-competitive factors relate, for example, to the respondents' socio-economic background, i.e. not only to how "well connected" they are, but also to their regional background and gender. This could actually be viewed as extra market power, but we find our categorization more convenient. The competitive factors relate to the respondents' own abilities and decisions during their academic career. These would include how well they perform, their selection of subjects and how quickly they finish their studies.

FINDINGS

Secondary data were used to identify the magnitude and incidence of unemployment of the educated. Female unemployment exceeded male unemployment for those with education below the matric level but the reverse was true for those with education above the matric level. Urban unemployment in almost all educational, regional and gender categories and age groups exceeded rural unemployment. The most significant finding was that unemployment for the under 30-years age-group significantly exceeded the more than 30-years age-group for all categories. Considering only those with a graduate or higher degree, unemployment was 18.8 percent for those under 19, 10.7 percent for the 20-24 age-group, 5.4 percent for the 25-29 age-group, 2.6 percent for the 30-34 age-group and very close to 1 percent thereafter.

This suggested to us that age was the critical variable (one that is ignored in the Labour Force Surveys) for studying unemployment. In this case, it suggested that our study could profitably concentrate on an analysis of waiting, since after a period of time most people did seem to be getting absorbed into the labour market. Similar conclusions were reached by Psacharopoulos and Sanyal from their research on Egyptian and Philippines data [9, pp. 455-458; 10, pp 32-33]. The rest of our study, therefore, focuses on identifying the incidence of waiting, explaining its causes heuristically and then more technically and on identifying the determinants of the differentials in waiting. Primary data were utilized for this purpose.

Although the relevant sample was that of the educated unemployed persons, the employee group data were used throughout as a control group. No major difference was identifiable between the education level attained by the parents of the respondents for the two groups. Parents' income did seem to be an important variable using simple cross-tabulations. Unemployed females from the lower income group exceeded by a factor of five the females employed from the lower income group. It seemed, therefore, that being female and from a poor background was a double disadvantage on the job market. It was also surprising to find that over a third of the unemployed males had fathers in the upper income category. Raffi et al. [11, p. 19] concluded from similar results based on another data set that those from more prosperous backgrounds are more able to afford a bout of unemployment in the hope of securing a desired job.

Of course, respondents from all income categories may have their waiting period prolonged if their expectations concerning the job market are not realistic. About 95 percent of the unemployed wanted professional, managerial or administrative positions. Judging from the actual occupational distributions for post-graduates reported in the Labour Force Survey [8, p. 161], only about three-fifths of them could be accommodated in such professions. Since the Labour Force Survey does not isolate the effect of age, this is probably an overstatement of what young post-graduates could reasonably expect. Sectoral preferences are similarly unrealistic. Three-fourths of the unemployed showed a preference for government sector jobs. The actual sectoral job distribution reported in the population census [5, pp. 115-117] indicated that for the under 25-years age-group in the Professional, Technical and related Worker category, only 13 percent worked in the government sector. In the corresponding Legislator-cum-Administrator and Manager categories, the government sector's employment share was 38 percent and 23 percent, respectively. Finally, the expected mean salary of the unemployed exceeded the actual mean salary by a considerable margin in two out of three categories for which information was available from the survey data.

The unemployed ranked "meeting needs of a specific future career" and "better employment opportunities" as very low when stating reasons for pursuing higher education. Further, over 50 percent of them were unwilling to move to rural areas or away from their homes for employment.

In studying the competitive and non-competitive factors determining market strength, we discovered once again that the unemployed were not entirely free from responsibility. Judging from their perceptions, the contrary view is also significant. Twenty-eight percent of the employees mentioned that having possessed "contacts" aided their search for a job. This percentage is probably an understatement. Even so, the corresponding unemployed amounted to only 4 percent. The unemployed also ranked the lack of connections as the most significant reason for their unemployment. While the significance of such a view prevailing among the unemployed cannot be ignored, it cannot be wholly accepted at face value either.

In examining the performance of the unemployed versus that of the employees and allowing for the effect of age, we found that among males and females, 15 and 35 percent respectively could be categorized as poor performers, while the corresponding numbers for the employees were 3 and 2 percent respectively. The lack of career-orientation among the unemployed was noted earlier.

Our attempts to test some of the above findings, using regression analysis, met only partial success. The proxies used for market realism were government-sector work preference and expected salary. Competitive market strength variables were performance, field of specialization and age, while parents' income, gender and domicile were the non-competitive market strength variables. Length of waiting was used as the dependent variable for both the employed and the unemployed groups.

One major problem in this analysis was that the waiting period by its very definition was not complete for the unemployed. We proceeded on the assumption that the waiting period, beyond the time captured by the survey, would not differ in a systematic pattern for the explanatory variables chosen for the regression. Surprisingly, the explanatory power of the equation for the unemployed group was considerably better; and [[bar.R].sup.2] of 16 percent compared with the dismal 1 percent for the employed group.

The strongest result that emerged from this exercise was that competitive factors did seem to count more on the job market. For both groups, performance, as expected, was negatively and significantly associated with the length of waiting period. It also had the highest beta coefficient for the employees group, and very close to the highest for the unemployed group. Parents' income, on the other hand, was not even picked up as a significant variable in our step-wise procedure for the unemployed group and it had a very small coefficient and the lowest beta coefficient for the employee group.

CONCLUDING REMARKS

Solutions to the unemployment of the educated can be found on both the demand and supply sides. The demand side solution is a general one related to the strategy of growth. More specific and immediate solutions can therefore be sought on the supply side. One of the specific suggestions that can be derived from our findings is that the information base for career counselling needs to be improved. There appears to have been a marked discordance between expectations of the unemployed and the market reality. While efforts for expanding the role of employment exchanges have met with administrative problems in the past, converting them into more effective information collecting and disseminating organizations may be possible. As a beginning, both government and private sector organizations should be required to provide them wage and vacancy data by occupation, education and field of specializations. This would be invaluable to both career guidance officers in academic institutions and to researchers.

Two policies could be considered to restrict supply. One is a mandatory waiting period after the first college degree which was recommended by Blaug [1, p. 76]. Another is to experiment with differential fees in higher education so that more serious thought is given to its acquisition.

Finally, a perception among the unemployed that unfair practices are responsible for their predicament must be taken seriously, even though our findings suggest that they themselves are not free from blame in this regard. A random audit of application files for positions by a section in the Ombudsman's office, and quick responses to appeals against sifarish would help in curbing demoralization among the educated unemployed persons.

Comments on "Some Findings about the Unemployed Highly Educated Persons in Pakistan"

I will follow the convention in vogue at this conference and cite an adapted version of an old Chinese saying. It goes like this, "If you want to plan for one year plant rice, if you want to plan for five years think of industry, but if you want to plan for a hundred years educate men". I think that this old Chinese adage is as true today as it was centuries ago when it was first coined. On the other hand, the problem of unemployment is the single most serious problem facing the developing countries today. Put the two together in the context of Pakistan today and you have an extremely important and worthwhile research topic that has not been researched before, and for this the authors need to be commended.

I will now break convention by being brief and to the point. It is evident that the paper is still in preliminary form and needs to be revised considerably. I will list a number of points that need to be clarified.

1. The paper deals only with the unemployment of those with post-graduate qualifications and not with the unemployment relating to all levels of education as the title suggests.

2. The definition of unemployment used in the paper is not clearly stated. Are the unemployed defined as those currently seeking work or are they simply those with post-graduate qualification who are without a job? The type of definition used would affect the results, especially as they relate to unemployed females with low-income fathers reported in one of their tables. These high percentages could then also be explained by social and traditional attitude that permit education of females but restrict their entering the job market.

3. The waiting period used in the analysis is by definition open-ended. It is difficult to form any definite conclusions because of this definitional problem. It might be useful to conduct an analysis on the waiting times of those who are currently employed. This, of course, would depend on the availability of the relevant data.

4. The authors have presented a fairly detailed and critical review of the survey data used in the analysis. However, the study would benefit tremendously from the inclusion of a couple of paragraphs on the sampling procedure used. This would help in determining how far the results of the study can be generalized.

5. The regression analysis, in its present form, because of the very low explanatory power of the equations and the general lack of significance of the estimates, contributes very little to the analysis.

I have restricted my comments to some of the basic issues that need to be clarified. The authors have presented a pioneering study on a topic of great national importance. Given the importance of the topic and the fact that it is the first study of its kind and that it attempts to make use of whatever data are available, I think the study needs to be considered seriously and further work in the area should be given priority.

Sohail Jehangir Malik

Pakistan Institute of Development Economics, Islamabad

REFERENCES

[1.] Blaug, M. Education and the Employment Problem in Developing Countries. Geneva: ILO. 1973.

[2.] IIEP (UNESCO) Ministry of Education (Pakistan). "The perceptions of Students, Graduates and Employers". Chapter Six in Higher Education and Employment Opportunities in Pakistan. Paris. 1986. (Forthcoming)

[3.] Pakistan. Statistics Division. Population Census Organization. 1981 Census Report of Baluchistan. Islamabad. December 1984.

[4.] Pakistan. Statistics Division. Population Census Organization. 1981 Census Report of NWFP. Islamabad. December 1984.

[5.] Pakistan. Statistics Division. Population Census Organization. 1981 Census Report for Pakistan. Islamabad. December 1984.

[6.] Pakistan. Statistics Division. Population Census Organization. 1981 Census Report of Punjab. Islamabad. December 1984.

[7.] Pakistan. Statistics Division. Population Census Organization. 1981 Census Report of Sind. Islamabad. December 1984.

[8.] Pakistan. Federal Bureau of Statistics. Statistics Division. Labour Force Survey 1982-83. Karachi. May 1984.

[9.] Psacharopoulos, G., and Bikas C. Sanyal. "Student Expectations and Graduate Market Performance in Egypt". Higher Education. Vol. 11. 1982.

[10.] Psacharopoulos, G., and Bikas C. Sanyal. "Student Expectations and Labour Market Performance: The Case of Philippines". Higher Education. Vol. 10. 1981.

[11.] Raffi, R., Robert E. Klitgaard and W. Eric Gustafson. "Idle Brains: Graduate Unemployment in Karachi". Karachi: Applied Economics Research Centre. March 1977. (Research Report No. 11)

* Dr Khan was Research Economist at the Pakistan Institute of Development Economics, Islamabad, and Mr Ali a graduate student at the McMaster University, Canada, when this paper was written.