UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The underemployment of B.C. college graduates Cram, Daniel William 1997

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1997-0518.pdf [ 3.8MB ]
JSON: 831-1.0087771.json
JSON-LD: 831-1.0087771-ld.json
RDF/XML (Pretty): 831-1.0087771-rdf.xml
RDF/JSON: 831-1.0087771-rdf.json
Turtle: 831-1.0087771-turtle.txt
N-Triples: 831-1.0087771-rdf-ntriples.txt
Original Record: 831-1.0087771-source.json
Full Text

Full Text

T H E U N D E R E M P L O Y M E N T OF B.C. C O L L E G E G R A D U A T E S by D A N I E L WILLIAM C R A M B.A. (Hon.), The University of Alberta, 1987 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF T H E REQUIREMENTS FOR T H E D E G R E E OF MASTER OF ARTS in T H E FACULTY OF G R A D U A T E STUDIES (Department of Anthropology and Sociology) We accept the thesis as conforming to the required standard T H E UNIVERSITY OF BRITISH COLUMBIA September 1997 © Daniel William Cram, 1997 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is. understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of "^OQwO^L The University of British Columbia Vancouver, Canada Date DE-6 (2/88) University of British Columbia Abstract T H E U N D E R E i W L O Y M E N T OF B.C. C O L L E G E GRADUATES by Daniel W. Cram Chairperson of the Supervisory Committee: Dr. Neil Guppy Department of Sociology Underemployment is a much discussed but little researched topic. The thesis begins with a broad discussion of the theory and methodology underlying the recent research on underemployment. It then proceeds to a quantitative analysis of underemployment using data from the 1995 follow-up of B.C. college leavers from vocational, technical and two-year academic university transfer programs. The study finds that, overall, one third of B.C. college leavers were employed in jobs that did not require the level of education that they had attained. As expected, there were significant differences by field of study and subsequent occupation. The rate of underemployment among students from academic programs was eight times the rate of underemployment for students from vocational programs and twice that of students from career/technical programs. Additionally, almost a third of all college leavers were employed in Sales and Service occupations and roughly two-thirds of those were underemployed. Labour market segmentation theory provides the most useful theoretical explanation for these findings. The markedly uneven rates of underemployment experienced by college leavers in the core and peripheral sectors support the labour market segmentation perspective. In conclusion, underemployment is a useful, though limited construct. Such a measure should only be used in conjunction with other measures of employment outcomes like unemployment, salary and full/part-time employment status. ii T A B L E OF C O N T E N T S Abstract ii TABLE OF CONTENTS iii LIST OF TABLES iv LIST OF FIGURES v Chapter 1 Introduction and overview 1 Introduction 1 Overview 1 Chapter 2Education, employment and underemployment 3 Introduction 3 The Link Between Education, Employment and Underemployment 4 Underemployment: Measuring Educational and Occupational Skills 16 Chapter 3 Methods 22 Introduction 22 Sample Design 22 Data 24 The Dependent Variable - Underemployment 27 Structural Independent Variables 31 Personal Independent Variables 33 Research Question 36 Analyses 36 Chapter 4 Results 38 Bivariate Analyses of Underemployment and the Independent Variables 38 Underemployment and Non-Market Characteristics 44 Discriminant Analysis - Underemployment 59 Chapter 5 Conclusions 72 BIBLIOGRAPHY 80 APPENDIX A 84 APPENDIX B 89 iii LIST OF TABLES Table 1: National Occupation Classification Skill Level Criteria 29 Table 2: Underemployment - Number of Employed B.C. College Graduates by College Program Type and Occupational Skill Level 31 Table 3: Occupation by Employment Status 40 Table 4: Industry Sector by Employment Status 41 Table 5: Occupation by Industry - Rate of Underemployment and Percent of All Employed in Labour Market Segment 43 Table 6: Field of Study by Employment Status 45 Table 7: General Program Type by Employment Status 46 Table 8: Top Five Labour Market Destinations of College Leavers by General Program Type 48 Table 9: Credential Completion by General Program Type and Employment Status 50 Table 10: Previous Post-Secondary Experience by Employment Status 51 Table 11: High School Completion by Employment Status 52 Table 12: Gender by Employment Status 53 Table 13: Visible Minority Status by Employment Status 54 Table 14: Disability Status by General Program Type by Employment Status 55 Table 15: Full-Time and Part-Time Employment by Industry Sector by Employment Status 56 Table 16: Average Monthly Salary by Full/Part Time Employment by Industry Sector by Employment Status 58 Table 17: Average Monthly Salary by Full/Part Time Employment by Gender by Employment Status 59 Table 18: Discriminant Analysis - Employment Status by Occupation, Industry, General Program Area, Field of Study and Personal Characteristics 63 Table 19: Job to Training Relatedness by General Program Area 69 iv LIST O F FIGURES Figure 1 : Discriminant Analysis - Total Structure Correlations by Standardi2ed Function Coefficients 67 v Chapter 1 Introduction and overview Introduction In this thesis, I explore the extent to which students from B.C. colleges experience underemployment, or occupational-skill mismatch, after leaving college. Specifically, I examine whether structural characteristics like occupation and industry, or personal characteristics like (i.e.) gender, disability status, visible minority status and field of study are more important in determining underemployment. In addition, I explore briefly the effects of underemployment on earnings and the coincidence of underemployment and part-time employment. I expand the research on education and employment by studying vocational, technical and two-year academic university transfer programs in community colleges; a poorly studied segment of higher education. I draw from a broad theoretical perspective to inform this study, a perspective necessary to consider the impact of both the structural and personal characteristics on underemployment. Overview Chapter Two begins with an overview of the recent literature linking education and employment. The unique contributions of human capital theory, credentialism and labour market segmentation theory are integrated to create a more complete understanding of underemployment. I follow the suggestion by Guppy "to consider competing theoretical explanations about the processes underlying this link" (1992:87). To that end, I will examine each theoretical perspective for its relative ability to explain the conditions of underemployment. Paralleling the theoretical discussion of underemployment is a discussion of the methodological issues surrounding the measurement of skills acquired through education 1 and the skills required on the job. Several alternative methods for measuring education and job skill mismatch are examined. As well, I explore the limitations inherent in each method. Chapter Three describes the data base used in this study and outlines the operationalization of the dependent variable: underemployment. Labour market segmentation theory informs the operationalization of the structural independent variables, occupation and industry, while the personal independent variables, demographic characteristics and educational experience, reflect a wider array of research on job-skill mismatch, credentialism and labour market segmentation theory. The specific research question examined in this study is also discussed in relation to the structural and personal independent variables: are structural or personal characteristics more important in determining underemployment? Chapter Four presents the detailed bivariate and multivariate analyses while Chapter Five summarizes and discusses the results in relation to the original research question and the related literature. Chapter Five concludes with a discussion of the policy implications arising from this research and a caution on subsequent use of the underemployment construct. 2 Chapter 2 Education, employment and underemployment Introduction If you asked Canadians if there was a link between formal education and the Canadian economy, most would likely agree that a strong link existed. If you asked them to describe the relationship between the two, most would likely characterize it at an individual level and say that you need a good education to get a good job or conversely, without an education you are out of luck in the job market. Where did we get these ideas? How has the association between education and employment come to dominate our perspective of higher education? And specifically, how have we come to determine that higher levels of education lead to better jobs? For the most part, we believe in the relationship between education and employment because, at the most basic level, it makes sense. Without basic literacy, numeracy, social and problem-solving skills, the vast majority of occupations are unattainable. Only the simplest of manual tasks do not require at least one of these skills. Further, since these basic skills are taught, over a relatively fixed period of time, by our formal primary and secondary education systems, the connection between education and employment is easily made. Our common sense perception of the relationship between education and employment is strengthened by the prevailing economic tenet: higher education is an investment. Human capital theory was formalized in the early 1960s to account for that portion of economic growth, at the national level, which could not easily be explained in terms of the returns to increased investments in traditional capital, i.e., manufacturing plants and machinery. The theory posits that expenditures on education are actually a form of capital investment; an 3 investment in people with a return that can be measured as a marked improvement in salary (Schultz 1961). When an individual invests in education, they invest in the quality components such as skill, knowledge and similar attributes that affect particular human capabilities to do productive work. In so far as expenditures to enhance such capabilities also increase the value productivity of human effort (labour), they will yield a positive rate of return (Schultz 1961:8). It is no wonder that individuals and governments invested heavily in education during the sixties and seventies when the rationale appeared so sound: increased levels of skill acquisition through formal schooling equate with higher job skill requirements and responsibilities. Both individuals and nations could profit from higher levels of education. It came as a surprise then, when higher levels of education did not perpetually translate into desirable employment outcomes. By the early to mid seventies, following a period of phenomenal growth in the post-secondary education system, the production of college graduates exceeded the demand for college trained individuals in the labour force (Berg 1970; Freeman 1975). Overeducation and underemployment, two terms used synonymously for a mismatch between the level of education attained and the level of education required to perform the job (Smith 1986), were the new reality for many college trained graduates. Since that time, critics and proponents of human capital theory have monitored the extent of underemployment and documented the diminishing returns to higher education. The Link Between Education, Employment and Underemployment There is a long-standing debate over the various functions of higher education. The link between education and employment reflects only one view of the primary role for education: that of occupational training. An alternate perspective stresses the social and cultural value of 4 education in creating responsible citizenry. This latter view does not preclude the notion of occupational relevance, but recognizes the less tangible, critical and innovative functions provided by higher education. In an examination of the relationship between occupational structures and higher education, Ulrich Teichler noted that: Graduates should not merely be prepared to take over given tasks. They should also be prepared to reconsider and reshape the job tasks. Students in higher education might acquire tools and learn rules. They also have to be capable and motivated to question established professional practices and to cope with undetermined work tasks (1990:976). This debate over the various roles of higher education has historical significance in B.C.. Two B.C. government policy documents, separated by less than a decade, reflect the two competing perspectives on the role of higher education. In 1988, the B.C. government acted on the recommendations of the report Access to Advanced Education and Job Training in British Columbia and greatly expanded access to university level education across the province. The rationale behind this expansion was based both on improving the equity of access to higher education and on broader educational goals as evidenced in the following statement: If our province is to continue to develop a healthy, civilized society in an increasingly complex and interdependent world, more people must participate successfully in advanced education and job training (MAEJT 1988:4). By 1995, the government had adopted a much narrower view of the role for higher education in B.C.. 1 The report of the B.C. Labour Force Development Board, Training for 1 The extent of the B.C. governments shift away from wider educational goals to narrower employability issues was reflected even in a name change for the ministry responsible for post-secondary education. The name was changed from the Ministry ojAdvanced Education andjob Training to the Ministry of Skills, Training and Labour. Following the amalgamation of the ministries of education and the MSTL, and the splitting off of the labour ministry, the name was subsequendy changed to the Ministry of Education, Skills and Training. 5 What? highlights the ascendancy of the occupational training role for higher education. At the outset of the report, the board clearly states its "bias toward the learning system's role in preparing students and workers for the world of work" (LFBD 1995:1). This remarkable shift in philosophy is further reflected in the board's recommendation that the government: challenge colleges and universities to review the balance between academic and applied programs within their institutions, with the objective of re-allocating resources from academic to applied, in order to provide additional programs to address the skills gap (LFBD 1995:44). These widely divergent education policies highlight the public nature of the debate over the link between education and employment. The policy debate was, and is currently, fueled by reduced federal and provincial funding for post-secondary education. The recent research on underemployment and overeducation mirrors the concern over the current state of the relationship between the systems of formal education and the labour market. In this section, I discuss three theoretical perspectives on the link between education and employment and focus specifically on the research into mismatched underemployment. I begin with a brief summary of the theoretical positions as they relate to each other followed by a discussion of the two extreme positions: credentialism and human capital theory. I then examine the specific research on mismatched underemployment and conclude the section with a discussion of the moderate theoretical perspective: labour market segmentation theory. To explain the larger debate over the relationship between education and employment, Hunter (1988) placed the research on this topic along a simple continuum. At the one end, human capital theorists view the relationship between education and employment as technical-functional where "schooling is the major means by which individuals acquire the mental skills and capacities for self-direction necessary for successful future performance in an occupation" 6 (Hunter 1988:753). This perspective drove the expansion of the post-secondary education system in the U.S., and later in Canada, after World War II and reflects neoclassical economic theory. At the other pole, the credentialist theorists contend that, due to preferential hiring practices, the transmission of skills has become largely spurious to the relationship between education and employment. At best, educational certificates function as signals to employers, indicating which potential employees are most suitable for jobs requiring responsibility and autonomy (Hunter 1988). A third perspective, falling somewhere between the poles, has recently emerged. Labour market segmentation theorists argue that the relationship between education and employment, as well as broader issues of inequity and the distribution of power in society, "are in large measure a function of the way work is organized in modern society" (Clairemont, Apostle & Kreckel 1983:246). Under this perspective, the structure of the labour market plays a much larger role in determining individual employment outcomes. Unlike the neoclassical economic theories, labour market segmentation theory postulates that there are distinct market segments with unequal returns to similar levels of education, barriers to movement between market segments rather than open mobility, differing levels of employee control over conditions in the workplace, and fixed structures which determine career paths within and between firms in the same segments (Clairemont et. al. 1983; Hunter 1988). Credentialism The credentialist view of the education / employment relationship arose from some of the earliest sociological critiques of the human capital theory. Ivar Berg, in his pivotal book, Education and Jobs, The Great Training Robbery (1970), was among the first to point out the weakening link between higher education and good jobs. His research is a foreboding tale of increasing underemployment for university graduate, dissatisfaction with work due to 7 underuuiization of skills and, most telling, the displacement of those without educational credentials. Berg was bitterly disheartened by his perception that employers were inflating job requirements in a bid to absorb the glut of, supposedly superior, university graduates. Berg and other credentialists (V. Burris (1983), Shokey (1989), Smith (1986) and Anisef, Ashbury, and Turrittin (1992)), argued that, given the relatively stable economic returns to higher education and despite increasing levels of underemployment, education must be providing some value to employers beyond an increase in worker productivity. Blaug summarized the credentialist notion that education credentials are being used as "surrogates for qualities which employers regard as important, predicting a certain level of job performance without however making any direct contribution to it" (1982). And as long as employers rely on educational credentials to allocate the preferred jobs, the trend to overeducation will continue (V. Burris 1983, B. Burris 1983). Halaby rejects the concept of credentialism outright claiming it is both "controversial and arguable" (1994:47). He argues that if credentials are being used as a signal of ability and productivity by employers then students are making the right choice by investing in higher education. The students' investment is as much in valuable signaling information as it is in wider educational skills (Halaby 1994). Hunter has a more moderate view on the role of credentials. In a 1993 study on the formal education, skills and earnings of Canadians, Hunter found a significant positive relationship between years of schooling, productivity and earnings apart from the effects of credentials. He concluded that neither human capital theory, credentialism nor labour market segmentation could alone adequately explain the differentials in earnings (Hunter 1993). 8 Human Capital Theory During the seventies and eighties, the human capital theorists accepted and closely monitored the decline in economic returns to higher education. Although the supply of college graduates was increasing, the growth in professional and managerial occupations, the traditional destination of the college trained, was slowing. In 1970-71, one third of male college graduates in the U.S. and two third of female graduates could not find work in their fields of study compared with roughly 10% a decade before (Freeman 1975). Although the effect of the slowdown was felt most by new entrants to the labour force, overall, in the five years between 1969 and 1973, the real and relative salaries of college graduates dropped 11-25% (Freeman 1975:289). Unlike the credentialists, however, Freeman did not see the decline as a failure of human capital theory. On the contrary, he countered, the large drop in college salaries was "the classic price system response to manpower surpluses" (Freeman 1975:289). Freeman further demonstrated that the market had responded in a rational way to the surplus in graduates with a commensurate drop in college enrollments.2 Throughout the next two decades, Freeman explored overeducation and championed the economics of education with a series of carefully reasoned articles on trends in the demand for education, salary differentials, occupational shifts, and labour market supply and demand for college graduates (see Freeman 1975; 1977; 1986). In an overview of the research on education and employment written in 1986, Freeman concluded that, although a broadly defined human capital view of education is not a panacea for labour market growth, it has become firmly embedded in the research on education and employment (1986:382). 2 In 1986, Smith extended Freeman's salary data and reported that, relative to high school graduates, U.S. college graduates had gained back whatever advantage they had lost despite a continuing trend towards overeducation (Smith 1986). Smith believed that the declines reported in Freeman's earlier study were a temporary product of the U.S. recession. 9 Occupational Skill Mismatch Although the economists focused on the economic returns to education, sociologists focused on other causes and consequences of underemployment and the extent to which known labour market inequities were reflected in patterns of underemployment. Apart from credentialism, the most popular and enduring topic of research on underemployment among sociologists is occupational-skill mismatch or the underutilization of skills (Clogg and Sullivan 1983; Clogg and Shokey 1984; Livingstone 1987; Redpath 1990; Krahn and Lowe 1990). Beyond the simple concern with earnings differentials, this broad view explores the extent, as well as the structural and personal characteristics associated with skill mismatch. As part of a broader project to develop a series of underemployment indicators, Clogg and colleagues (Clogg and Sullivan (1983), Clogg and Shockey (1984) and Clogg, Sullivan and Mutchler (1986)), examined the divergent U.S. trends in mismatched underemployment between blacks and non-blacks and women and men. Overall, in the decade between 1969 and 1980, the rate of mismatched underemployment almost doubled from 7.8 to 14.2 percent (Clogg and Sullivan 1983:129; Clogg and Shokey 1985:241). However, alone among five indicators of underemployment,3 the rates of mismatched underemployment were consistently lower for blacks and women than they were for non-blacks and men. Neither Clogg and Sullivan (1983) nor Clogg and Shockey (1984) commented on this peculiar finding beyond stating that the relative rates of mismatched underemployment between men/non-blacks and women/blacks had remained stable over time. In a critique of Clogg and Sullivan (1983), Tipps and Gordon (1985), argued that the relatively lower rates of mismatched underemployment for 3 Clogg and Sullivan include unemployment, involuntary part-rime work, education-job skill mismatch, working poverty and discouraged workers (those no longer seeking work) under their broad view of underemployment (Clogg and Sullivan 1983). 10 women and blacks were misleading largely because males and non-blacks had greater access to higher levels of education. Clogg and Shokey also examined the rates of mismatch underemployment among different occupational groups. Between 1969 and 1980, the rates of underemployment for managers, craft, sales, clerical and other service workers showed absolute growth of between 4-9%.4 Clogg and Shokey conclude that mismatched underemployment has increased for "virtually every age, sex, colour, occupation, cohort, and schooling group examined" (Clogg and Shokey 1984:254). More recendy, Redpath (1990) and Krahn & Lowe (1990) used the same mid-eighties panel study of Canadian university graduates to examine the relationship between field of study, personal and structural labour market characteristics and underemployment. Redpath found that clerical/sales and service occupations accounted for almost two thirds (60%) of the underemployed university graduates in her study (Redpath 1990:262). Krahn and Lowe, reporting in a slighdy different manner, found that, two years after graduation, over 20% of all university graduates surveyed were working in clerical/sales and service occupations in consumer and service industries (1990:21). Redpath also examined the unique effects of sex on underemployment but found it was not significant compared with the effects for field of study and industry (1990:175-184). She concluded that the effect of gender on job-skill match occurred when women select their field of study. "Women tend to enter careers such as teaching, traditionally dominated by women 4 This finding supports a remarkably accurate prediction made in a 1967 Census monograph cited by Berg in 1971. In that monograph, the large projected increase in college graduates would ".. .have to be absorbed in managerial, sales, clerical and some craftsmen occupations" (Berg 1970;66). 11 and this increases their probability of being matched" (Redpath 1990:148). In Redpath's sample of university graduates, women were over-represented in the education faculty where the subsequent rates of underemployment were low (1990:172). Labour Market Segmentation Labour market segmentation (hereafter referred to as LMS) theorists are not solely concerned with the relationship between education and employment but their perspective helps to both locate underemployment and understand its variable implications depending on market location. In the broadest sense, LMS theory is the sociologists' response to a strictly neoclassical economic view of the labour market. In their book Work, Industry and Canadian Society, Krahn and Lowe describe the different focal points of the human capital and LMS perspectives. Basically the segmentation approach recognizes long-standing patterns of inequality within the labour market and tries to account for how such power differences are maintained and perpetuated. The human capital model, with its emphasis on individual differences, mostly overlooks such structural inequities (Krahn and Lowe 1993:142). Central to the LMS perspective is dual labour market theory. In research on labour market segmentation in Canada, Krahn and Gartrell (1983) summarized dual labour market theory as follows. Dual market theory divides the market for labour into two basic segments; a primary and a secondary labour market. One of two key tenets of dual market theory is that the working conditions and employment outcomes are markedly different between the primary and the secondary labour markets. The primary labour market serves the core industrial sector and is characterized by full-time work with good pay and benefits, opportunities for further training, upward mobility and job. Conversely, the secondary labour market, serving the peripheral industrial sector, is characterized by temporary, part-time work with poor pay, little opportunity 12 for training or advancement and little or no job security. The second tenet of dual market theory is that subsequent movement between and within, labour market segments is restricted by long-standing structural barriers. The effect of these barriers is to limit the movement and minimize the choices available to people seeking work or advancement (Krahn and Gartrell 1983). Additionally, LMS theorists note that certain types of individuals are concentrated in different market segments; women, visible minorities, youth and the disabled are over-represented in the poorer segments (Clairemont, et. al. 1983; Krahn and Lowe 1993). The research done to operationalization market segments provides several alternative models from which to choose. Industries, occupations, firms, and workers' characteristics, have all been used as the basis for typologies of market segmentation (Clairemont, et. al. 1983). Industrial and firm typologies are important to LMS research because they define segments according to characteristics of the core and peripheral industrial economic sectors or product markets. Occupations and worker traits define segments according to the characteristics of the market for labour (Clairemont, et. al. 1983; Hachen 1988). In 1988, Myles, Picot and Wannel examined the industrial and occupational restructuring that occurred in Canada between 1981 and 1986. They were specifically interested in determining the effect on wages due to the marked growth in service industries. To do so, Myles and colleagues used an industry typology that highlighted the differential rates in growth and wages among five service sectors as well as natural resources, manufacturing and construction. Contrary to the widely held belief that the growth in service industries and occupations was the main cause of the decline in wages, they concluded that workers' age had the single largest effect. Specifically, youth had experienced the largest declines in wages during this time independent of the more moderate effects of industrial restructuring (Myles, et. al. 1988). 13 The Economic Council of Canada, used an innovative industry market segmentation perspective in their report, Good Jobs, Bad Jobs; Employment in the Service Economy (Economic Council of Canada 1990). Like Myles, they proposed a typology that highlighted the increasing importance of service industries to the Canadian economy. In their model, the primary and secondary sectors are collapsed into a single goods producing sector while the tertiary sector is expanded into three service subsectors: dynamic, traditional and non-market services. Service industries are grouped into subsectors according to: their recent and expected growth; the quality of employment outcomes within the industry; the impact of technological change; and the public or private nature of their revenue (Economic Council of Canada 1990:2). The Dynamic service industries represent that portion of the private revenue services where employees can expect relatively good jobs: higher skill levels, better pay and job security. The dynamic industries both contribute to and benefit from the success of the goods producing sector. And like the Canadian industries in the goods producing sector, dynamic service industries have expanded into international markets. The Traditional service industries "represent the old "Main Street" variety of services" (Economic Council of Canada 1990:2). These industries typically operate in local rather than international markets. The impact of improvements in technology, if any, have tended to reduce the skill levels required in these industries. Jobs in the Traditional service sector are the bad jobs in the Good Jobs, Bad Jobs economy. The Canadian Non-Market service industries are recognizable as public sector industries. These industries are distinct from both the dynamic and traditional sectors by the presence of almost universal unionization. While these industries function outside competitive markets, there is increasing "attention now being paid to such matters as productivity growth and technology change" (Economic Council of Canada 1990:2). Compared to jobs in the 14 traditional service sector, jobs in the non-market service industries have good employment outcomes. The E C C concluded that growing segmentation in the labour market has resulted in fewer good job opportunities for Canadians: We have also found evidence of segmentation in terms or earnings, skill content, job stability, and the location of employment. Two quite distinct "growth poles" account for virtually all of the employment expansion in the 1980s: one includes highly skilled, well compensated, stable jobs, while the other consists of nonstandard jobs with relatively low levels of compensation and stability (ECC 1990:17). Krahn and Lowe (1990) and Boylan (1992) both adopted a cross-classification of occupation and industry as a mechanism to identify market segments. Krahn and Lowe used this market segmentation perspective to provide a more comprehensive view of the labour market for Canadian high-school and university graduates. They found that 80% of female high school graduates found work in clerical, sales and services occupations in service industries compared to 53% of male high school graduates. A much smaller, though still significant, proportion of university graduates were working in the same labour market segments: 25% of females and 20% of males (Krahn and Lowe 1990:19). Krahn and Lowe also found that faculty of study was a significant factor in determining the subsequent labour market segment for university graduates. Specifically, 41% of graduates from Arts faculties were employed in clerical, sales and service occupations in service industries: a proportion to two to six times as high as graduates from other faculties. Boylan used a cross classification of occupation and industry as a tool to illustrate the structure of external labour markets. Unlike internal labour markets, external markets lack specific administrative controls which provide structural opportunities for upward mobility. 15 However, Boylan argued that the structure of external markets, although diffuse, exhibited both barriers to mobility between employers as well as advantages. Some market segments confer advantage when moving between employers; advantages like transferable skills, educational skills and credentials, opportunities to signal one's abilities to other employers, and networks of social contacts. Boylan found that occupation was much more important than industry in conferring external advantages to mobility. Professional, technical and managerial positions as well as sales positions all provided for improved mobility between market segments. Notably, Boylan concluded that "The position one occupies affects one's current income, prospects for moves to other employers, and future income growth net of personal characteristics" (Boylan 1992:120). Underemployment: Measuring Educational and Occupational Skills Measuring Educational Skills A source of debate among those studying overeducation is generated by disagreements over how to operationalize mismatched underemployment. At the heart of the debate is the question of how to measure skill; both the skills learned through education and the skills required on the job. Compared with the debate over measures of occupational skill requirements (see below), there is a dearth of discussion surrounding the measurement of the skills gained through formal education. Aside from the credentialists, who maintain that the skill transmission function of formal education is largely unimportant, few researchers address the kinds of skills provided by higher education. Most often, researchers start with the assumption that skill transmission is a function of higher education and then move straight to operationalization. While I am uncomfortable with such a sweeping assumption, a challenge of this tenet is beyond the scope 16 of this thesis. I will, however, address a minor point in operationalizing educational attainment: the assumption of the homogeneity of skills. Research studies that use total years of schooling as a measure of acquired educational skills "ignore the diversity of qualitatively distinct types of skills that are generated by differences in the curricular content of schooling" (Hallaby 1994:49). In this treatment of educational attainment, the skills gained from one baccalaureate university degree to another, from one university to another or from one degree holder to another are perfectly equivalent. This assumption of skill homogeneity is particularly harmful to research on skill mismatch: a university graduate employed in an occupation that requires university level skills would be deemed adequately employed regardless of the type of degree they hold. One alternative to accepting the assumption of the homogeneity of skills is to account for type of schooling. Anisef, Ashbury and Turrittin (1992) and Livingstone (1987) attempt to account for the type of post-secondary education when operationalizing educational attainment. Anisef and colleagues (1992) compared the differential occupational status attainment of the community college (technical and vocational training) and university degree trained graduates but failed to control for differences due to years of schooling (Guppy 1992). Livingstone, in a study of underemployment in Ontario, used a self-reported measure of educational attainments that allowed for a distinction between community college and university level studies. However, he collapsed the measure in his subsequent discussions of underemployment (Livingstone 1987). Hallaby in his 1994 critique of this methodological assumption proposed an alternative measure for research on overeducation. He suggested that individuals be asked whether they thought their skills were being adequately used in their current position. While this perspective seems laudable, it is balanced by the research of B. Burris (1983) who found that clerical workers, regardless of their level of education, felt their skills were underutilized. 17 Measuring Occupational Skills When considering how to measure the skills required on the job, there are at least two competing methods: (1) the occupation specific mean educational attainment of workers and; (2) an independent measure of entry level job-skill requirements. The mismatched underemployment measure of the Labour Utilization Framework, or LUF, figures prominently in the debate over how to measure the skill requirements of occupations (Clogg and Sulivan 1983; Clogg and Shockey 1984, 1985). The L U F measure of occupational mismatch is based on the mean years of schooling within occupational groups. If an individual's educational attainment (measured in years of schooling) exceeds one standard deviation above the mean for their occupational group then they are considered underemployed. There are two primary criticisms of the LUF measure of underemployment. First, this solely statistical construct introduces an artificial component into the measure of underemployment. In each occupational category approximately 16% of the workers will, as an artifact of the method, be defined as overeducated (V. Burris 1983:457). By their own admission, Clogg and colleagues created this measure "...without recourse to...(c) formal models based on economic or sociological assumptions" (Clogg, et al. 1986: 379). While this treatment may be statistically sound, it is substantively questionable. By using the standard deviation as a cut off, Clogg and colleagues necessarily assume that overeducation is a universal condition of all occupations. Second, the LUF method of measuring underemployment is susceptible to differences between employer recruitment practices and job skill requirements (Redpath 1990). If employers choose to inflate the job requirements, so as to attract more highly trained applicants, this measure would be unable to account for the resulting discrepancy without using some independent measure of job skills. 18 An alternative method of measuring job skills examines the skill requirements needed to perform the job at an entry level (Berg 1970; V. Burris 1983; Redpath 1990),. Entry level skill requirements are expressed as aggregates of years of formal schooling. These aggregate scores are in turn based on the General Education Development score attached to the Canadian Classification and Dictionary of Occupational titles.5 Redpath summarized the rationale underlying the G.E.D. score, "higher levels of general skills are usually acquired through higher levels of schooling" (Redpath 1990:55). While the G.E.D. score is purportedly an independent measure of job skill requirements it is also subject to several problems. First, and most obviously, the G.E.D. score, like the mean educational attainment score, is only a proxy of job skill requirements. The validity of this measure hinges on the assumption that years of education actually equate with an increase in the attainment of skills and subsequently, that these skills are used on the job. Hunter also noted this weakness and chose instead to use "factor-analytically derived composites based on forty-three different aspects of the task requirements of about 6,500 occupational categories..." (1993: 26). He identified eight factors that differentiated the task content of jobs; cognitive complexity, routine activity, responsibility, verbal activity, gross motor skills, fine motor skills, persuasive skills, and creativity (Hunter 1988, 1986). These eight dimensions of job-skills suggest that the G.E.D. alone is at best an extremely over-simplified measure of entry-level job skills. Second, using entry level requirements at the occupational group level assumes a uniformity of skill requirements across jobs within an occupation whereas the reality is more likely to reflect a diversity of requirements (Halaby 1994). The actual skill requirements of positions are formed as much by the individuals who occupy them as by the requirements of the job. Skill use on the job becomes an amalgam 5 The G.E.D. score is also attached to the U.S. Dictionary of Occupational Tides (D.O.T.) (Redpath, 1990). 19 of the skills of the worker and the requirements of the position. Third, by expressing entry level skill requirements as aggregate years of formal schooling, the G.E.D. method ignores the effect of informal skill acquisition. Individuals acquire skills through means other than formal schooling like ".. .work experience, travel, self study, hobbies, and leisure activities..." (Redpath 1990:56). Finally, the scale variability of the G.E.D. is restricted, particularly in relation to higher levels of education. Of the six scale points, only the upper two reflect a level of education at the post-secondary level (Redpath 1990:55). Four reasons influenced my decision to use entry level skill requirements, rather than average educational attainment, to measure job-skill mismatch. First, data on mean educational attainment are not as readily available as data on entry level skill requirements. The benchmark scores of mean occupational educational attainment, first used by Clogg and colleagues, were created from the 1970 U.S. census data (Clogg, et. al. 1986). Given that I am dealing with a Canadian data set, it would be necessary first to determine benchmark mean educational attainment scores from Canadian census data before comparing it to the occupational attainment of the B.C. college leavers in this study. In addition, since the B.C. college leavers data set codes occupations using the National Occupation Classification schema, I would be restricted to using N O C coded census data which has only been used since 1991. The necessity of using relatively recent census data to establish mean educational attainment benchmarks would, in turn, mask any tendency over the past two or three decades towards overeducation in the Canadian labour market (V. Burris 1983). Second, as noted earlier, using the mean educational attainment score in measuring mismatch underemployment creates an universally artificial condition of underemployment in each occupational group (see page 18). Since I am contending that the level of underemployment varies by market segment, as defined by 20 occupation and industry, I do not wish to start with a statistically imposed assumption that underemployment is universal across all market segments. Third, a measure of job-skill requirements based on entry level skills is particularly appropriate to a study of college leavers. While mean educational attainment may be an appropriate base for gauging underemployment in the entire work force, entry-level skill measures are more appropriate for examining the underemployment of those seeking entry-level employment. The college leavers in this study are far more likely than the population as a whole to be seeking entry-level employment. Fourth, the new National Occupational Classification schema includes a widely accessible and replicable G.E.D.-like measure of entry-level skill requirements. The skill level criteria of the N O C groups occupations by the amount of training or education required to perform the job at an entry level. This criterion was included in the new schema specifically to address the issue of education to occupation relatedness. The new N O C is an improvement over the older Standard Occupation Classification to which measures of educational attainment could only be measured through reference to the Dictionary of Occupational Titles. 21 Chapter 3 Methods Introduction This chapter describes the sample design, data, and methodology used in this thesis and the main research questions. First, I provide an overview of the sample design and the population of B.C. college leavers. I follow this with a discussion of the strengths and weaknesses of using data from the annual follow-up of former B.C. College and Institute students and describe the variables drawn from this data set. I define and operationalize the dependent variable, underemployment, making use of the Skill Level criteria in the National Occupational Classification system. Additionally, I describe two structural independent variables, industry and occupation, and seven personal independent variables. My research questions are drawn from the literature review in Chapter Two. Sample Design The college system in B.C. is a little studied segment of higher education. The small amount of study that does take place is policy based and initiated by government rather than by sources of independent research. While I am not suggesting that these studies are completely self-serving, neither are they free from political bias favouring the position of the government in power. In this thesis, I attempt to examine the public policy issue of education and employment at a step removed from the larger political debate. At the same time, I realize the necessity of interpreting the results in relation to the local and wider debate on education and employment. B.C. colleges and institutes form a distinct part of the post-secondary system in B.C.. The college system was implemented in the early sixties to improve access to post-secondary 22 education for the geographically dispersed population of the province. For two decades, the colleges provided vocational and career/technical training in addition to first and second year university level studies. The system was expanded again in the late eighties and early nineties to further improve access to degree level education. By the mid nineties, after a period of association with the provincial universities, several colleges (subsequently called university colleges) outside the major metropolitan areas were given degree granting status. This study takes place just following the start-up of the university colleges but before significant numbers of degree level students were graduating. While the population of college leavers studied in this thesis is quite diverse, including both graduates and near-graduates, from academic and applied programs, it is nonetheless a cohesive population. Apart from articulation agreements with the universities, the college and institute system operates independently from either the university or the private training systems. Compared to the university system, the college system emphasizes; open rather than restricted access; and teaching over research. At the same time, the college sector provides both academic education and applied vocational training. Compared to both the private training institutions and universities, the college system has far less autonomy with regards to their operations, planning, programming and funding. I believe that the post-college experiences of all college leavers are important to understanding the college sector. While there may be differences in outcomes between graduates and near-graduates, academic and applied students, it is preferable to demonstrate those differences and explain why they occur rather than exclude groups beforehand without demonstrating first, that differences in outcomes do exist. 23 Data Each year, B.C. colleges and institutes collaborate to census survey their former academic, career/technical and vocational students. The survey is designed to collect information on employment outcomes, further education, and students' perceptions of their college experience. Population selection criteria are established by the Outcomes Working Group6 (OWG) but implemented independently at each institution. The criteria for survey selection are as follows: • Academic students (students in programs leading to baccalaureate degrees) must have completed at least 12 credits at the college. On average, those academic students submitted for survey in 1995 completed 44.2 credits (std. dev. = 28.0). • Career / Technical students (students in 2-year diploma programs with an applied or occupational orientation) must have completed at least half of their program. The implementation of this cut-off is determined at the institutional level and is not universally associated with credits. Typically, diploma level studies are well defined, structured programs where transition from year to year is easily measured. • Students in Vocational programs (applied or occupationally oriented programs whose duration is typically a year or less) must have completed their studies to be eligible for survey. These cohort guidelines are intentionally designed to capture both graduates as well as those who have completed a significant amount of education at the college level. At the 24 instimtional-program level, where much of this information is used, the comments of both graduates and non-completers are equally valued. However, in this examination of employment outcomes, where a credential may be valued or essential for employment, I examine the intervening effect of completion status on other independent variables. All students were surveyed by phone in the spring of 1995, nine to 12 months after leaving the college. For applied programs, the twelve month survey window reflects the colleges' desire to examine the short-term training-to-work transitions. Because applied programs are typically oriented to specific occupations, a relatively quick transition from education to related work is expected. Conversely, the lack of a positive transition to work within the nine month period is perceived as a negative outcome. Additionally, since the survey also serves the program level evaluation process, the shorter survey window ensures that the results are recent enough to be relevant to the current program. For academic programs, the twelve month window was deemed sufficient to examine the short-term transitions of students from college to university. Of the 27,549(a) students surveyed, 899(b) were excluded from the cohort because they were in programs other than academic, career/technical or vocational. In addition, 1,011(c) students were deemed ineligible for survey during the pre-screening process of the phone survey, 743 refused to respond and 4,995 could not be reached. In all, 19,901 (d) students responded to the survey. Excluding those ineligible for survey, the total response rate for the 6 Established in 1987, the Outcomes Working Group is a working committee of representatives from colleges, institutes and the Ministry of Education and Skills Training. The working group administers all aspects of the project including survey design, survey-contractor selection, reporting, data access and dissemination. 25 survey was 77.6%7. An additional 144 responses from students in Basic Skills programs were excluded from the final analysis. Basic Skills programs offer training at the primary or secondary school level, outside the scope of this analysis. For this thesis, only those students who were employed at the time of survey were included in the analysis. Of those students responding to the survey, 17,361 (87%) were in the labour force and eligible for employment: 90% were employed either full or part-time while 11% were unemployed. Of those employed, 15,285 had provided enough employment information to code their occupation, using the National Occupational Classification (NOC) system, and the industry in which they were employed, using the Standard Industrial Classification (SIC). The base data set for this analysis is comprised of those 15,285 employed respondents with coded occupations and industries. The use of a population data set creates some problems when interpreting sampling statistics and calculating significance. I have chosen to include tests of significance on the loosely held assumption that the population of graduates surveyed could be considered a sample of college leaver populations. Additionally, although the response rate to the survey is high, it is not perfect, therefore the use of sample statistics is warranted to support generalizing to the entire class of college leavers in 1995. Finally, sample statistics are included as loose guidelines providing some assurance of methodological rigor. In the next three sections, I will discuss the dependent variable (underemployment) and nine independent variables: two structural independent variables (occupation and industry) and seven personal independent variables (general program type, field of study, program completion 7 The total survey response rate is calculated as follows: (d) # responding / (# surveyed - (# ineligible programs + # ineligible 26 status, previous post-secondary experience, high school completion, gender, visible minority status, and disability status). In addition, I will discuss two alternative measures of underemployment (salary and part-time employment) as they relate to the dependent variable. The Dependent Variable - Underemployment The dependent variable, underemployment, is operationalized by comparing the level of education attained to the level of education required on the job. As a result, the dependent variable is actually created from two separate variables. In order to identify a leaver as either underemployed or adequately employed we need to know; a) the amount of training a college graduate has completed; and b) the level of education required to perform their job. College leavers are deemed underemployed when the level of training they have received at college exceeds the level of formal training that their occupation requires. In this section, I discuss level of education attained and level of education required by occupation separately and as they are used together to create the dependent variable; underemployment. Level of Education Attained The level of education a college leaver has attained is measured by Program type, one of several program grouping schema used by the B.C. Outcomes Working Group. Each program of studies is classified by the OWG according to; the expected duration of the program when taken full time; the level of education required to enter the program; and the type of credential granted upon completion (MoEST 1996:D-5,6,). Programs granting certificates may require high school completion to enter and are classified, by duration, as either 0-6 months certificate or 6-12 months certificate program types. for other reasons)). The actual calculation is as follows: (19,901/(27,549 - (899 + 1,011)) (MoEST 1996:D4-D8). 27 Diploma programs range in duration from 12 to 36 months (typically 24 months) and require high school completion to enter. Diploma programs are classified as 12-36 month diploma. These first three program types are almost exclusively applied or occupationally oriented. The objective of these programs is to provide the skills necessary for employment in specific occupations. B.C. colleges also offer general academic, university level studies. There are two university level program types; lower division degree and upper division degree. The lower division degree is comprised of university level courses at the first and second year level and requires high school completion to enter. Typically, lower division programs do not offer a credential at completion, rather they lead to upper division degree level studies. Upper division degree programs are comprised of courses at the third and fourth year university level and they require the completion (or the equivalent) of lower division degree level studies to enter.8 Upon completion of upper division degree programs, students are granted baccalaureate degrees. In the degree granting university-colleges, lower division degree programs lead directly to upper division degree level studies, just as they would at a traditional university. At non-degree granting colleges, lower division degree level programs are designed to transfer, through articulation agreements, to degree granting institutions. Level of Education Required by Occupation The level of education required by an occupation is measured here using the skill level criteria of the National Occupational Classification system (NOC). The N O C code is based on two general grouping criteria. A skill type criteria, which groups occupations by the type of work 28 performed and a skill level criteria, which groups occupations by the amount of training or education required to perform the job at an entry level (Statistics Canada 1991:iv-v). Table 1 outlines the type and level of education required to perform jobs at different N O C skill levels.9 Table 1: National Occupation Classification Skill Level Criteria N O C S k i l l E d u c a t i o n level L e v e l required to perform job Sample Occupa t ions Level A • a university degree (bachelor's, Human Resources Professionals master's or post-graduate). Business Service Professionals Pharmacists, Dietitians, Nutritionists Level B • 2-3 years of post secondary Technical Occupations in Dental education at the college level Health • 2-4 four years of apprenticeship Sales and Service Supervisors training Machinists • 3-4 years of high school supplemented with two or more years of on-the-job training, training courses or work experience Level C • 1-4 years of high school Child and Home support workers • up to two years of on-the-job Longshore workers training, training courses or work Material handlers. experience Level D • up to two years of high school and Primary production labourers short work demonstration or on- Food counter attendants the-job teaming Elemental sales occupations Management • not assigned Administrative Mangers in Business Occupations and Finance Mangers in Health, Education, Social and Community Services Managers in Manufacturing Utilities 8 In some upper division degree programs, students may enter the program after completing related diploma level studies rather than lower division degree courses. 9 Occupations were coded by Statistics Canada using responses to questions about job title and main job duties. 29 Occupations with significant management responsibilities "are not assigned to a skill level category because factors other than education and training (e.g., previous work experience, capital) are often more significant determinants for employment" (Statistics Canada 1991:iii). Underemployment In this thesis, underemployment is measured by comparing the level of education attained (program type) to the subsequent level of education required by the occupation (skill level). A condition of underemployment exists when the level of education attained exceeds the level of education required by the occupation. Table 2 demonstrates how underemployment has been determined at each intersection of program type and skill level. For example, students from 0-6 month, certificate programs, employed in jobs at skill level D (n=25) have been classified as underemployed: they have completed training beyond that required by their current jobs. 0-6 month, certificate students employed in occupations at skill level C and above are adequately employed since their jobs require levels of training the same as or greater than the training programs they have recently completed. Conversely, 162 former students from Upper division, degree programs found employment in occupations at skill level A. They have been classified as adequately employed since their occupations require a university degree; the level of education which they have attained. Other students from Upper division, degree programs, employed at skill levels B and below, are deemed underemployed since their training exceeds that required by their occupation (Table 2). Although management occupations are not assigned a N O C skill level, I have classified former students employed in these jobs as adequately employed. Management occupations are 30 commonly viewed as desirable positions with diverse job duties and significant personal and organizational responsibilities. These positions require a broad, rather than a specific, skill set and may be reasonably viewed as adequate employment rather than underemployment. Table 2: Underemployment - Number of Employed B.C. College Graduates by College Program Type and Occupational Skill Level NOC Occupation Skill Level Criterion Mngmt Not Level A Level B 2-3 Level C Level D Total, College Program Type Classified Univ Degree Yrs Post Sec 1-4 Yrs Sec 1-2 Yrs Sec All Levels Adequacy Employed 0-6 months, certificate 3 2 21 41 0 67 6-12 months, certificate 208 266 2539 2024 403 5440 12- 36 months, diploma 320 1046 1313 881 182 3742 Lower division, degree 402 510 1348 2557 892 5709 Upper division, degree 27 161 63 60 16 327 Underemployed Total, all program types 960 1985 5284 5563 1493 15285 Structural Independent Variables The structural independent variables, occupation and industry, are drawn from the research on labour market segmentation. As discussed in Chapter Two, Boylan (1992) used occupation and industry to operationalize market segments in his exploration of external markets. Based on the LMS expectation that individuals can expect unequal returns from similar amounts of education depending on the market segment they occupy, I expect that the rates of underemployment will vary significantly by occupation and industry. In this section, I first discuss occupation and industry separately and then as they combine, by cross-classification, to create an enhanced operationalization of market segments. 31 Occupational Group Occupation is coded using the skill type criteria of the National Occupational Classification system (NOC). All occupations are categorized under one of 10 general occupational groups. Each of these occupational groups, or skill types, is defined by ... ... the type of work performed, although other factors related to skill type are also reflected in the NOC. One of these factors is similarity with respect to the educational field of study required for entry into an occupation. Another factor is the industry of employment where experience within an internal job ladder or within an industry is usually a pre-requisite for entry (Statistics Canada 1991:iii). Industry Sector I have defined industry sectors following the schema used by Myles, Pico and Wannell in their 1988 study of changing wage distributions in Canada. This industry grouping method was developed to examine the effects of industrial restructuring on wages, particularly among Canadian youth. The industry model developed by Myles and colleagues was used by Krahn and Lowe (1990) in their examination of Canadian university and high school graduates and by Redpath (1990) in her doctoral dissertation on the mis-matched underemployment of university graduates. This model is appropriate to this thesis. Eighty percent of the B.C. college leavers surveyed annually find employment in service industries (MoEST 1993:23).10 Occupational Group by Industry Sector I have operationalized labour market segments for B.C. college graduates using a cross-tabulation of the industry sector classification and the N O C occupational type. The resulting matrix functions like a simple map of the labour market for B.C. college grads with an industry 1 0 An alternate industry grouping model, based on the Economic Council of Canada's "Good Job, Bad Jobs; Employment in the Service Economy" was excluded from this thesis. The ECC model describes three service industry subsectors; Dynamic, Traditional and Non-Market. The service industries were grouped into subsectors according to their recent and expected growth, the quality of employment outcomes within the industry, the impact on technological change, and the public or private nature of their revenue (Economic Council of Canada 1990:2). The model was rejected because it masked and failed 32 longitude and an occupational latitude. This map allows us to pinpoint the occurrences of underemployment. Krahn and Lowe used this technique to identify the specific labour market destinations of a sample population of Canadian high-school and university graduates (Krahn and Lowe 1990:15). Personal Independent Variables The seven personal independent variables are drawn from the research on mismatched underemployment and labour market segmentation. In this section I discuss how each of these variables relates to the research in Chapter Two as well as their expected relationship to underemployment. General Program Type and Field of Study General program type and field of study are included as measures of the type of education attained. Both Redpath (1990) and Anisef, Ashbury and Turrittin (1992) found that the type of educational program significantly influenced subsequent occupational outcomes. Given these findings, I also expect that the rate of underemployment will vary significantly by general program type and field of study. General program type reflects the type of skills taught in a program. Academic programs are typically classroom based, university level, baccalaureate studies, "...the majority of academic programs are traditional arts, humanities, natural sciences and social sciences, [although] this program area can also include professional degrees, e.g. Nursing and Education" (MoEST 1996:D-3). Career/technical programs are typically more applied in nature, but provide a to account for marked differences in the levels of underemployment within the Dynamic service subsector, contrary to the conclusions of the ECC analysis. 33 combination of academic (credit may be transferable to university) and job-specific skills training (i.e. computer programming, electronics technician, dental hygienist). Vocational programs are occupationally oriented and focus primarily on job-specific skills training. Fields of study are grouped using twelve topical program groupings of the CIP code. This code has recently been adopted by the Outcomes working group as the standard program taxonomy as a replacement for the more general program area grouping. Program Completion Program completion status is included as an independent variable because the population of college leavers contains both graduates and non-graduates. It is also expected to have an interaction effect with program type. I expect that those who have not completed their program will be over-represented among the underemployed. Program completion status was measured by the response to the question "When you left [name of institution], had you completed the requirements for a credential such as a degree, diploma, or certificate?" Previous post-secondary experience, and high school completion were similarly measured by responses to the questions: "Did you take any post-secondary education before attending [name of institution]?" and "Before attending [name of institution], did you complete secondary (high) school?" Gender, Visible Minority Status and Disability Status These three personal ascribed characteristics are drawn directly from the research on underemployment and labour market segmentation. The research on mismatch underemployment, done by Clogg and colleagues (see Chapter Two), suggest that those groups that typically experience inequities in the labour market (i.e. women, visible minorities and the 34 disabled) have lower rates of underemployment than their counterparts (i.e. male, anglo-saxon, able bodied). However, Redpath's research on underemployment found that gender did not have a statistically significant impact on underemployment. LMS research suggests that these traditional equity groups are over-represented in the less advantaged market segments; the peripheral sector. In as much as underemployment is a disadvantageous market outcome, I expect that the equity groups will be over-represented among the underemployed. Gender was determined from college student records systems during the survey population selection. Visible minority status was self-reported by response to the question "Are you, because of your race or colour, in a visible minority group in Canada?" Disability status was also self-reported by the response to the question "Do you have a long-term condition or health problem that limits, or which you feel is perceived by others as limiting, the kind or amount of activity you can do in the workplace?" Salary and Part-time Employment Salary and part-time employment figure prominently in human capital, labour market segmentation and underemployment research. However, I have chosen to treat them as alternate or parallel measures of underemployment rather than causal or dependent variables. I employ this treatment based on the work of Clogg and colleagues (see Chapter Two) in establishing salary and part-time employment status as two of seven measures of underemployment. I do, however, conjecture that the mismatch underemployed will have lower salaries than the adequately employed and that those employed part-time will have higher rates of underemployment than those employed full-time. 35 A standardized monthly salary was calculated from the survey question "What was your gross salary or wage from your [main] job, before deductions?". Respondents were allowed to respond in the manner most convenient to them (i.e. hourly, daily, weekly, bi-monthly, monthly, annually or other). Additional information on weekly hours worked was used where necessary. Research Question My central research question is quite simple: Are the structural characteristics of the labour market more important in determining the underemployment of college leavers than personal characteristics? I believe that the British Columbia's predominandy service economy is markedly segmented and will exhibit the characteristics associated with the dual market perspective. In addition, the research of Redpath (1990) and Krahn and Lowe (1990) suggest, that the two structural independent variables, occupation and industry, are key to understanding underemployment. To that end, I believe the structural variables, occupation and industry, will prove to be more important than the personal variables in determining underemployment. Analyses In the following chapter, I use two types of analyses to examine the relationship between the dependent variable, underemployment, and the nine independent variables. First, I cross-tabulate each independent variable with the dependent variable, underemployment, and examine the relative rates of underemployment. Second, I use multivariate discriminant analysis to measure the relative effects of each independent variable while controlling for the effects of all other variables. I chose multivariate discriminant analyses for two reasons. First, it uses a nominal, rather than an interval level, dependent variable. Second, a two stage discriminant and 36 classification analysis provides; 1) measures of the unique and combined strength of the independent variables to discriminate between the nominal groups of the dependent variable and; 2) a subsequent, easily interpretable, test of the ability of the discriminant factor to correcdy classify cases based only on the information provided by the independent variables. 37 Chapter 4 Results Bivariate Analyses of Underemployment and the Independent Variables Overall, of those students who graduated or left B.C. colleges in 1994 and had not since returned to the same college in the following nine to twelve months, 33% were employed in jobs that did not require the level of education that they had attained. Occupation A brief examination of occupational differences is all that is needed to demonstrate that the overall rate of underemployment is a poor indicator of the conditions within occupational labour markets. Rather than clustering around the average, the rates of underemployment tend to the extremes across occupational groups. At the outer limits, occupations unique to processing and utilities have a rate of underemployment (64.2%), over one hundred times the rate found in Art, Culture, Recreation and Sport (0.6%). Although these two groups account for only 7.3% of all jobs, they are by no means unusual. Of nine occupational groups (excluding Sr. Management occupations), three have rates of underemployment over 60% while three have rates below 2% (Table 3). Significandy, almost a third (31.4%) of all college leavers were employed in Sales and Service occupations and 60.5% of those were underemployed. This single occupational group accounts for more than half (57.6%) of all underemployment. A further 17.7% of college leavers found jobs in Business, Finance and Administration occupations and over a third (37.5%) of those were underemployed. Together, these two occupational groups account for over three quarters of all underemployment. 38 The jobs of the adequately employed were more evenly distributed across occupational groups with no more than one-fifth accounted for by any one group. Sales and Service occupations provide 18.8% of all adequate employment while Business, Finance and Administration and Trades, Transport and Equipment related occupations provide 16.7% and 16.8% respectively. Industry Unlike the pattern among occupational groups, the rates of underemployment within industrial sectors range loosely around the average (33.1%). However, as with occupational groups, a single industry sector dominates. Almost a third (31.2%) of all jobs for college leavers were found in the Consumer Services industries and over half (54%) of those resulted in underemployment. Conversely, the Education, Health and Welfare industries account for almost a quarter (23.1%) of all jobs and the lowest sectoral rate of underemployment at 13.1%. The Goods production industries employ a fifth (19.1%) of college leavers with over a quarter (28.0%) of those deemed underemployed. The Business service industries account for 12.5% of all entry-level employment for college leavers with less than a quarter (23.4%) of those positions rated as underemployment. The rate of underemployment in Distribution service industries is relatively high at 38.3%; however, this sector accounts for only 8.7% of all jobs. Finally, Public Administration industries provide employment for 5.5% of college leavers with 29.1% of those underemployed (Table 4). 39 Table 3: Occupation by Employment Status E m p l o y m e n t Status * O c c u p a t i o n U n d e r -employed (Row %) Adequate ly E m p l o y e d (Row %) T o t a l E m p l o y e d (Column %) Management Occupations 100.0 0.5 C70) (70) Business, Finance and Administration 37.5 62.5 17.7 (1012) (1688) (2700) Natural and Applied Sciences and Related Occs. 0.9 99.1 8.3 (12) (1259) (1271) Health Occupations 7.7 92.3 10.5 (123) (1481) (1604) Occs. in Social Sciences, Education, Gov. and Religion 1.6 98.4 7.6 (19) (1143) (1162) Occs. in Art, Culture, Recreation and Sport 0.6 99.4 4.1 (4) (617) (621) Sales and Service 60.5 39.5 31.4 (2903) (1898) (4801) Trades, Transport and Equip, operators and related occs. 19.3 80.7 14.1 (414) (1734) (2148) Occs. unique to primary industry 60.4 39.6 2.7 (253) (166) (419) Occs. unique to processing and utilities 64.2 35.8 3.2 (314) (175) (489) Total 33.1 66.9 100.0 N (5054) (10231) (15285) Differences between occupational groups are statistically significant (p<.01, Chi-square test). 40 Table 4: Industry Sector by Employment Status E m p l o y m e n t Status * U n d e r - Adequa te ly T o t a l Industry Sector employed E m p l o y e d E m p l o y e d (Row %) (Row %) (Column %) Goods Production 28.0 72.0 19.1 (817) (2103) (2920) Distribution Services 38.3 61.7 8.7 (509) (819) (1328) Consumer Services 54.0 46.0 31.2 (2576) (2191) (4767) Business Services 23.4 76.6 12.5 (446) (1461) (1907) Educ / Health / Welfare 13.1 86.9 23.1 (463) (3066) (3529) Public Administration 29.1 70.9 5.5 (243) (591) (834) Total 33.1 70.9 100.0 (5054) (10231) (15285) Differences between industrial sectors are statistically significant 'p<.01, Chi-square test). Labour Market Segments: Industry and Occupation Table 5 shows the proportion of all college leavers employed in each market segment and the accompanying rates of underemployment. For the purpose of this thesis, a market segment is defined by the intersection of an occupational group and an industry sector. The five largest market segments employ over half (51.4%) of all college leavers. The rates of underemployment in these five market segments range from some of the lowest to some of the highest. The single largest labour market segment for college leavers, Sales and Service occupations in the Consumer service industries, employs over one fifth (22.2%) of all leavers; and almost two-thirds (65.2%) of those are underemployed. Conversely, Health occupations in Education, Health and Welfare industries employ almost one in ten college leavers (9.7%) and only 6.9% of those are underemployed. The fifth largest labour market segment is also in the Education, Health and Welfare industry sector. Occupations in the Social 41 Sciences, Education, Government and Religion, in this industry sector, employ roughly 1 in 20 (6%) college leavers while the rate of underemployment in this market segment is only 1.6%. Only one non-service market segment ranks among the top five labour market destinations for college leavers. Occupations unique to primary industry in Goods producing industries employed 7.5% of all college leavers. This market segment is relatively strong compared with other service sector markets with an underemployment rate of 15.5%. This finding supports the notion that primary industries can still provide good employment opportunities relative to the service sector. 42 Table 5: Occupation by Industry - Rate of Underemployment and Percent of All Employed in Labour Market Segment Industry Sector O c c u p a t i o n Educ / Goods Distrib. Consumer Business Health / Public Production. Services Services Services Welfare Admin. Total Senior Management % underemployed in segment % of all employed coll. leavers Business, Finance and Administration % underemployed in segment % of all employed coll. leavers Natural and Applied Sciences and Related Occupations % underemployed in segment % of all employed coll. leavers Health Occupations % underemployed in segment % of all employed coll. leavers Occupations in Social Sciences, Education, Government and Religion % underemployed in segment % of all employed coll. leavers Occupations in Art, Culture, Recreation and Sport % underemployed in segment % of all employed coll. leavers Sales and Service Occupations % underemployed in segment % of all employed coll. leavers Trades, Transport and Equipment Operators and Related Occupations % underemployed in segment % of all employed coll. leavers Occupations Unique to Primary Industry % underemployed in segment % of all employed coll. leavers Occupations Unique to Processing and Utilities % underemployed in segment % of all employed coll. leavers Total, All Occupations % underemployed in segment % of all employed coll. leavers 0.1 36.4 2.2 0.5 2.8 1.9 0.3 0.0 0.1 0.0 0.4 51.0 1.0 15.5 7.5 55.4 2.0 63.2 2.8 28.0 19.1 0.1 52.7 2.4 0.7 0.9 0.0 0.0 0.0 0.0 0.0 0.3 61.3 1.9 28.0 2.9 33.3 0.0 60.0 0.1 38.3 8.7 0.1 52.0 2.5 2.2 0.3 50.0 0.2 3.3 0.4 0.4 1.8 65.2 22.2 15.2 3.0 73.8 0.4 75.0 0.2 54.0 31.2 0.1 31.2 6.5 1.3 3.0 0.0 0.0 1.1 0.6 0.0 0.5 49.0 1.4 46.2 0.3 66.7 0.0 80.0 0.0 23.4 12.5 0.1 29.2 2.2 0.0 0.3 6.9 9.7 1.6 6.0 0.0 0.5 38.4 4.0 6.7 0.1 90.9 0.1 40.0 0.0 13.1 23.1 0.1 31.4 1.9 1.4 1.0 4.5 0.1 1.5 0.4 3.5 0.6 69.1 0.9 57.9 0.2 71.4 0.2 100.0 0.0 29.1 5.5 0.5 37.5 17.7 0.9 8.3 7.7 10.5 1.6 7.6 0.6 4.1 60.5 31.4 19.3 14.1 60.4 2.7 64.2 3.2 33.1 (15285) 43 Underemployment and Non-Market Characteristics The research examined in Chapter Two provides both structural and personal explanations of underemployment. I have briefly examined the relationship between underemployment and the structural elements of the labour market; occupation and industry. In this next section I examine the relationship between underemployment and several personal characteristics. Field of Study The level of underemployment across fields of study ranges from 6% for individuals from Health programs to 61.9% for students from the Liberal Arts, Sciences and Humanities. Of the 12 fields of study, seven have rates of underemployment less than 18%, four have rates between 23% and 40% and one has a rate over 60% (see Table 6). In addition to having a high rate of underemployment, students from Liberal Arts, Sciences and Humanities programs account for one third of all those surveyed. Combined with the high rate of underemployment, students from the Liberal Arts, Sciences and Humanities account for two-thirds (62.9%) of all those underemployed. The rate of underemployment for students from Liberal Arts, Sciences and Humanities programs is possibly biased by the survey methodology. The survey is designed to capture the perceptions of students after they have left a BC college. Since Arts and Sciences programs at BC colleges typically require transfer to a university after the second year, many of these students are being surveyed while they are continuing their education while in other program areas, students are surveyed after they have completed their education. Additionally, it might be reasonable to assume that students who are continuing their studies may at the same time be 44 employed in occupations that are not career-related: their jobs are simply a means to pay the bills. Table 6: Field of Study by Employment Status E m p l o y m e n t Status * F i e l d o f Study U n d e r -employed R o w % Adequate ly E m p l o y e d R o w % T o t a l E m p l o y e d R o w % Agriculture & Natural Resources 15.6 84.4 100.0 (41) (222) (263) Construction, Mechanics and Metal Trades 9.4 90.6 100.0 (169) (1625) (1794) Transportation 16.1 83.9 100.0 (37) (193) (230) Engineering, Electrical and Electronics 13.3 86.7 100.0 (106) (689) (795) Business management 29.0 71.0 100.0 (750) (1840) (2590) Liberal Arts, Sciences and Humanities 61.9 38.1 100.0 (3183) (1960) (5143) Communications 23.3 76.7 100.0 (28) (92) (120) Fine/Visual and Performing Arts 39.2 60.8 100.0 (185) (287) (472) Recreation, Hospitality and Service 14.9 85.1 100.0 (74) (423) (497) Computer Information Services 17.2 82.8 100.0 (43) (207) (250) Education 23.3 76.7 100.0 (162) (532) (694) Legal and Social 31.4 68.6 100.0 (161) (352) (513) Health related 6.0 94.0 100.0 (115) (1809) (1924) Total, All Fields of Study 33.1 66.9 100.0 (N) (5054) (10231) (15285) * Differences between fields of study are statistically significant (<.01, Chi-square test). 45 Certainly, the rates of continuing education are very different for students from academic programs and career-related programs. Thirty-six percent of Liberal Arts, Sciences and Humanities students were continuing their studies at the time of survey compared to only 6.1% of students from all other program areas. However, this distinction does not seem to account for the high rates of underemployment among Liberal Arts, Sciences and Humanities students. Even when we exclude those students who are continuing their studies, the underemployment rate for students from Liberal Arts, Sciences and Humanities programs drops by only three percent to 58.2%. In addition, these students still account for over half (54.1%) of all those who were underemployed and not continuing their education. General Program Type Academic, Career and Vocational programs have significantly different rates of underemployment. Students from Academic programs are eight times as likely to be underemployed (59.5%) as students from Vocational programs (7.4%) and twice as likely as Career students (28.3%) (see Table 7). Table 7: General Program Type by Employment Status E m p l o y m e n t Status * F i e l d o f Study U n d e r - Adequa te ly T o t a l employed E m p l o y e d E m p l o y e d R o w % R o w % R o w % Academic 59.5 40.5 100.0 (3577) (2438) (6015) Career/Tech 28.3 71.7 100.0 (1072) (2715) (3787) Vocational 7.4 92.6 100.0 (405) (5078) (5483) Total 33.1 66.9 100.0 (5054) (10231) (15285) 46 Given that Career and Vocational programs are targeted at specific occupational groups and that rates of underemployment differ across labour market segments, it would be reasonable to assume that the different rates of underemployment across general program types are due, in part, to the divergent labour market destinations of program graduates. An examination of general program type by labour market segment proves the relationship is not that simple. Table 8 shows the proportion of leavers from each general program area employed in the top five labour market segments. Although the top five market segments do not exhaust the destinations of college leavers, they are reasonably representative. The top five market segments account for over half of all employment for college leavers (51.4%) and the rates of underemployment range from a low of 1.6% to a high of 65.2%. Over one third (36%) of Academic students are employed in Sales and Service occupations in Consumer service industries, whereas only 12% of Career students and 14% of Vocational students are similarly employed. While the labour market destinations of Career and Vocational students are much more diverse than those of Academic students this difference alone does not account for the higher rate of underemployment for Academic students. Even within market segments, Academic students consistently report rates of underemployment two to eight times as high as the rates for Career and Vocational students. 47 Table 8: Top Five Labour Market Destinations of College Leavers by General Program Type L a b o u r M a r k e t Segment Genera l P rog ram T y p e A c a d e m i c Career V o c a t i o n a l Sales and Service Occupations in Consumer Services Industries % employed in segment (col %) % in segment underemployed N employed in segment Health Occupations in Education, Health and Welfare Industries % employed in segment (col %) % in segment underemployed N employed in segment Trades, Transportation and Equipment Op. Occupations in Goods Producing Industries % employed in segment (col %) % in segment underemployed N employed in segment Business, Finance and Administration Occupations in Business Services Industries % employed in segment (col %) % in segment underemployed N employed in segment Social Sciences, Education, Government and Religion in Education, Health and Welfare Industries % employed in segment (col %) % in segment underemployed N employed in segment 35.7 80.8 (2576) 3.1 37.8 (188) 4.5 42.4 (269) 6.1 58.4 (365) 7.8 3.2 (468) 12.4 67.4 (470) 16.6 4.9 (627) 3.1 19.7 (117) 8.6 29.7 (327) 3.9 0.0 14.2 20.7 (779) 12.2 0.0 (669) 13.9 5.4 (762) 5.5 0.0 (303) 5.6 0.0 (309) Program Completion An exclusive focus on program completers fails to account for the labour market outcomes of a significant portion of students who never complete their studies. As discussed in Chapter Three, this survey includes responses from both program graduates and non-completers. Over a third (37.7%) of those included in this analysis had not completed the requirements for a credential when they left the college. 48 In a market that both values credentials and looks unfavourably on unfinished tasks, those who left before completing a credential are disadvantaged in finding adequate employment. Over half (56%) of those who had not completed a credential when they left college were underemployed compared with only 18.9% of those who graduated (Table 9). It is important to remember that this analysis includes responses from students who were continuing their education at the time of survey and had not yet completed a credential. Again this is more likely to be true of students from Academic programs as compared with students from Career or Vocational programs. With the exception of some programs in the degree granting university colleges, Academic programs in B.C. colleges are designed to transfer to university after the second year of college academic studies. If we examine the employment outcomes of only those students who have completed a credential, the rates of underemployment for Academic students drop significantly but remain higher than the rates of underemployment for both Career and Vocational graduates. 49 Table 9: Credential Completion by General Program Type and Employment Status Comple t ed the requirements for a credent ia l Genera l P r o g r a m T y p e N o Y e s U n d e r -employed R o w % Adequate ly E m p l o y e d R o w % U n d e r -employed R o w % Adequa te ly E m p l o y e d R o w % t> Academic 62.9 37.1 47.8 52.2 (2930) (1730) (644) (703) Career 43.4 56.6 25.2 74.8 (282) (368) (789) (2345) Vocational 9.3 90.7 7.2 92.8 (44) (430) (361) (4646) Total 56.3 43.7 18.9 81.1 (3256) (2528) (1794) (7694) *the difference between general program type and credential completion is statistically significant (<.01, Chi-square test) Previous Post-secondary Experience Forty-two percent of those included in this analysis had previous post-secondary experience. The level of previous education among those surveyed includes: some experience but without a credential (21.1%), certificate (9.7%), diploma (6.6%) and university degree (5.2%). A small portion (1.3%) of those surveyed had completed more than one post-secondary credential prior to their most recent college studies. The rate of underemployment for those without any previous post-secondary experience (40.3%) is twice the rate of those with previous experience (22.7%) (see Table 10). While previous post-secondary experience does seem to reduce the likelihood of underemployment, this result might also be interpreted as contributing to the underestimation of underemployment. If the measure of education attained, used in the operationalization of 50 underemployment, were to include all educational experiences and not simply the most recent college experience, then we could expect the overall rates of underemployment to rise. Table 10: Previous Post-Secondary Experience by Employment Status E m p l o y m e n t Status * U n d e r - Adequa te ly T o t a l Previous post-secondary experience employed E m p l o y e d E m p l o y e d no 40.4 59.6 100.0 (3600) (5303) (8903) yes 22.8 77.2 100.0 (1454) (4925) (6379) Total 33.1 66.9 100.0 (5054) (10228) (15285) * the difference in rates of underemployment are statistically significant (<.01, Chi-square test) between those with previous post-secondary experience and those without. At least two factors contributed to my decision to exclude previous post-secondary experience from the operationalization of underemployment. First, the B.C. Students' Follow-Up survey examines students' post college experiences relative to their most recent program of studies. From the analysis so far, it appears that the type of education is as important as the duration of the education in determining the extent of underemployment. In the absence of more detailed information about previous field of study, it would be difficult to sum the effects of previous education with the students most recent college experiences. Second, it is difficult to view the results of education as a strictly linear, cumulative process (i.e. are two, two-year diplomas equal to one four year baccalaureate degree?). If they are cumulative, then the levels of underemployment are much higher than those shown here. 51 However, such a treatment is contrary to common sense. Under a strict cumulative model, students who, over their lifetime, pursued a gradual course of upgrading from vocational to degree level studies would eventually be viewed as underemployed even if they ended up in a job requiring their university level training. High school completion High school completion is a pre-requisite for most vocational programs and all diploma level and academic programs offered in the B.C. college system. Only 6.9% of all those included in this analysis had not completed a high school diploma. However, the lack of a high school diploma does not seem to contribute to underemployment. On the contrary, those who had not completed high school were significantly less likely to be underemployed than those with a high school diploma (Table 11). Table 11: High School Completion by Employment Status C o m p l e t e d h i g h schoo l E m p l o y m e n t Status * U n d e r - Adequate ly employed E m p l o y e d T o t a l E m p l o y e d no 16.8 83.2 100.0 (178) (879) (1057) yes 34.3 65.7 100.0 (4875) (9350) (14225) Total 33.1 66.9 100.0 (5053) (10229) (15282) * the difference in rates of underemployment are statistically significant (<-01, Chi-square test) between those who completed high school and those who did not. This relationship is not unusual given the manner in which underemployment is operationalized and the limited range of educational and occupational options available to those 52 without a high-school diploma. Individuals who do not complete high school are seldom eligible for anything other than short-term, non-technical vocational programs like janitorial or food services. There are fewer opportunities for graduates of these programs to be underemployed simply because they often have only a minimum of post-secondary experience and are employed at all but the lowest occupational levels. These programs constitute some of the lowest levels of post-secondary educational attainment and are targeted at occupations whose educational requirements are equally low. Gender Overall, there is a small, but statistically significant, difference between the rates of underemployment for women and men. The rate of underemployment for women is about 3% higher than the rate of underemployment for men (Table 12): women comprise 53.9% of those included in this study and 55.8% of all college leavers who are underemployed. The small, but significant difference between females and males is indicative of the weakness of mismatch underemployment as a measure of structural inequities between traditionally disadvantaged groups in the labour force. Clogg, Sullivan and Mutchler, (1986) reported that the rate of Table 12: Gender by Employment Status E m p l o y m e n t Status * U n d e r - Adequa te ly T o t a l Gender employed E m p l o y e d E m p l o y e d Male 31.7 68.3 100.0 (2224) (4781) (7005) Female 34.5 65.5 100.0 (2811) (5326) (8173) Total 33.3 66.7 100.0 (5035) (10107) (15142) * the difference between females and males is statistically significant (<.01, Chi-square test) 53 mismatch underemployment, alone among five other rates of underemployment, was lower for women than for men. Redpath found that the rate of underemployment for female university graduates was slightly, but not significantly, higher than the rates of underemployment for males (1991). She also found that the differences between men and women remained non-significant when controlling for faculty of study. Visible Minority Status Visible minority status has a small impact on underemployment. Visible minorities comprise 15.9% of the study population and 17.6% of all those who are underemployed (Table 13). As with gender, the rate of mismatched underemployment may mask the labour force inequities traditionally associated with visible minority status. Clogg, et. al, (1986) noted that the rates of mismatched underemployment, from a 1980 U.S. population survey, were lower for blacks and Hispanics than for non-blacks although they did not discuss why this might be the case. Table 13: Visible Minority Status by Employment Status I n a v is ib le minor i ty group i n Canada by virtue o f race or colour E m p l o y m e n t Status * U n d e r - Adequa te ly employed E m p l o y e d T o t a l E m p l o y e d no 32.3 67.7 100.0 (4162) (8715) (12877) yes 37.0 63.0 100.0 (892) (1516) (2408) Total 33.1 66.9 100.0 (5054) (10231) (15285) * the difference between those in a visible minority group and those not in a visible minority group is statistically significant (<-01, Chi-square test) 54 Disability Status Those with disabilities make up about 4% of the study population. The impact of their disabilities on underemployment is unexpectedly negative. Those with a disability are less likely to report underemployment (26.6%) than those without a disability (33.2%) (Table 14). Again, as with high school completion, the lower overall rate of underemployment for those with disabilities is likely related to a higher proportion of these students enrolled in vocational programs. Forty-seven percent of students with disabilities were enrolled in vocational programs compared to 36% of students without disabilities. However, even for those students enrolled in Academic and Career programs, the subsequent rate of underemployment for the disabled was lower than for students without disabilities. Table 14: Disability Status by General Program Type by Employment Status D i s a b i l i t y Status N o Y e s U n d e r - Adequate ly U n d e r - Adequa te ly Genera l P r o g r a m T y p e employed E m p l o y e d employed E m p l o y e d Academic 59.5 40.5 56.9 43.1 (3448) (2344) (116) (88) Career 28.6 71.4 21.7 78.3 1041 2603 28 101 Vocational 7.3 92.7 7.6 92.4 (380) (4792) (22) (268) Total 33.3 66.7 26.6 73.4 (4869) (9739) (166) (457) * the difference between the disabled and the able-bodied within program areas is statistically significant (<.01, Chi-square test) 55 Part-Time Employment Overall, college leavers in part-time jobs are more likely to be underemployed than those with full-time work (47.7% vs 30.1% respectively). However, the co-incidence of part-time employment and underemployment varies significantly by industrial sector. The rate of underemployment for those employed part-time in Consumer Services industries (72.7%) is over four times as high as the rate of underemployment for part-time workers in Education, Health and Welfare industries (16.3%) (Table 15). Table 15: Full-Time and Part-Time Employment by Industry Sector by Employment Status F u l l / P a r t t ime employment Indus t r ia l Sector Pa r t -T ime F u l l - T i m e U n d e r -employed R o w % Adequate ly E m p l o y e d R o w % U n d e r -employed R o w % Adequa te ly E m p l o y e d R o w % Goods Production 43.0 57.0 27.4 72.6 (64) (85) (748) (1986) Distribution Services 58.4 41.6 36.2 63.8 (73) (52) (428) (753) Consumer Services 72.7 27.3 48.0 52.0 (853) (320) (1706) (1849) Business Services 42.3 57.7 20.8 79.2 (94) (128) (347) (1321) Educ / Health / Welfare 16.3 83.7 11.8 88.2 (143) (732) (307) (2293) Public Administration 31.7 68.3 28.9 71.1 (26) (56) (216) (531) Total 47.7 53.3 30.1 69.9 (1253) (1373) (3752) (8733) * the difference between the rates of underemployment within industrial sector and full / part time employment status are statistically significant (<.01, Chi-square test) 56 The Labour Utilization Framework (LUF), developed by Clogg and Sullivan (1983), recognizes involuntary part-time employment as one of several measures of underemployment. They found that the 1980 rate of part time employment for women in the U.S. was 1.3 times the rate for men. Among employed BC college leavers, women are twice as likely as men to be employed part-time (22.9% vs. 11.1%) but only marginally more likely to be skill mismatched (see above under Gender). This finding supports the earlier conclusion that mismatch underemployment is insensitive to traditional gender inequities in the labour market. Furthermore, if we view part-time employment as another form of underemployment then the co-incidence of part-time and mismatch underemployment may be viewed as an extreme form of underemployment. Under this scenario, 6.1% of men are doubly underemployed while 10.4% of women experience both part-time and mismatch underemployment. Salary Underemployment is associated with lower salaries for both full and part-time workers. Overall, underemployed, full-time workers make 79% of what their adequately employed counterparts earn. The impact of part-time work exacerbates the problem. On average, an underemployed, part-time worker makes 67% of an adequately employed part-time salary (Table 16). Additionally, there is an interaction effect between underemployment, full part time status and industrial sector. The average monthly salary for full-time, underemployed workers in consumer services is 75% of the average salary for the adequately employed. On the other hand, average salaries for underemployed, full time workers in the Goods producing and Public Administration sectors are close to 95% of what the adequately employed earn. In addition to a smaller salary gap, these two industrial sectors pay the highest average salaries. 57 Table 16: Average Monthly Salary by Full/Part Time Employment by Industry Sector by Employment Status F u l l / P a r t t ime employment Indus t r ia l Sector Part-time F u l l - t i m e U n d e r -employed Adequate ly E m p l o y e d U n d e r -employed Adequa te ly E m p l o y e d Goods Production $1,144 $1,316 $2,533 $2,676 (55) (66) (669) (1738) Distribution Services $900 $1,044 $2,005 $2,452 (61) (43) (376) (665) Consumer Services $731 $837 $1,407 $1,876 (724) (270) (1502) (1586) Business Services $817 $1,235 $1,837 $2,277 (78) (100) (299) (1109) Educ / Health / Welfare $1,069 $1,391 $2,037 $2,370 (124) (643) (267) (2027) Public Administration $1,196 $1,101 $2,431 $2,547 (24) (51) (194) (486) Total $818 $1,221 $1,852 $2,342 (1066) (1173) (3307) (7611) There is also an interaction effect on salary between full/part time employment status, mismatch and gender. Women who are underemployed in full-time jobs are paid 75% of what their male counterparts earn. Conversely, women who are employed part-time in adequate jobs earn 106% of their male counterparts (Table 17). This unusual finding is likely due to the over-representation of women in Health occupations which have greater salary equity and more favourable part-time working conditions. 58 Table 17: Average Monthly Salary by Full/Part Time Employment by Gender by Employment Status F u l l / P a r t t ime employment Part-t ime F u l l - t i m e U n d e r - Adequate ly U n d e r - Adequa te ly Gender employed E m p l o y e d employed E m p l o y e d Male $886 $1,169 $2,120 $2,557 (370) (297) (1612) (3803) Female $784 $1,239 $1,592 $2,121 (693) (858) (1681) (3716) Total $819 $1,221 $1,851 $2,342 (1063) (1155) (3293) (7519) Discriminant Analysis - Underemployment I examined the combined relationship of ten variables to underemployment using discriminant analysis. Discriminant analysis is a multivariate technique, similar to multiple regression, which uses a nominal grouping (dependent) variable and interval or ratio level discriminating (independent) variables11. The technique uses the combined means and variances of the discriminating variables to distinguish between two or more categories of a grouping variable. A mathematically derived function equation combines the effect of each discriminating variable, while controlling for the effect of all other variables, into a single function score for each case. For each characteristic (independent variable) being used to distinguish between the groups, a coefficient is calculated. The products of all coefficient score pairs are summed to 1 While the discriminant analyses requires interval level independent variables, a violation of this assumption is only critical to the tests of significance. When the sample is sufficiendy large or the data set is a population sample, as is the case here, the test of significance is less important and therefore the violation of this assumption can be safely ignored. Additionally, if the percentage of correct classifications is high than the violation of the assumptions was not harmful (Klecka 1980:62). 59 produce the function score. The coefficients are derived to maximize the distance between the mean function scores for each category of the grouping variable (Klecka 1980:18-23). The utility of the discriminant analyses function is twofold. First, an overall canonical correlation coefficient can be derived from the function providing ... a measure of association which summarizes the degree of relatedness between the groups and the discriminant function. A value of zero denotes no relationship at all while large numbers (always positive) represent increasing degrees of association with 1.0 being the maximum (Klecka 1980:36). The canonical correlation is interpreted in much the same manner as a Pearson product coefficient. The squared correlation coefficient can also be interpreted as the proportion of the total variance between the group means explained by the function (Klecka 1980:36-37). Second, standardised discriminant function coefficients and total structure coefficients provide a mechanism to examine the relative and total contribution of each distinguishing variable to the function equation. The standardized coefficients describe the relative importance of each variable to the total discriminant function equation while controlling for the effects of all other variables. The interpretation of standardized function coefficients is analogous to the interpretation of standardized regression coefficients (Klecka 1980:66). The total structure coefficients are simple bivariate correlations between the discriminating variables and the total function score. They are useful in determining the extent to which variables share discriminating information with the function equation (Klecka 1980:31-32). A careful examination of both coefficients can assist us in deciphering which variables contribute the most to distinguishing between the adequately employed and the underemployed. I included ten variables already examined in the bivariate analysis, in the discriminant analysis. A dichotomous dummy variable was created for each category of the industry, 60 occupation, field of study, and general program type variables. For the industry referent group I excluded the education/health/welfare industry group. For the occupation referent group I excluded health occupations. Likewise, for the field of study referent group I excluded Health fields of study. Finally, for the general program referent I excluded the Vocational general program type. In each case, the referent groups had among the lowest rates of underemployment. Individuals employed in management occupations were excluded from the analysis (n=70). Occupations at the senior management level are coded by definition as adequate employment and thus lacked any variability. The final analysis included thirty-three dummy and interval level independent variables12. Salary and part-time employment status were also excluded from the analysis. While both of these variables contribute to distinguishing between the adequately employed and the underemployed, neither can be reasonably considered causally prior to underemployment. Differences in salary are more appropriately viewed as an outcome of underemployment while part-time work status is a related form of marginal employment or an alternate operationalization of underemployment. Interpretation of Standardised Discriminant Function Coefficients For the discriminant analysis, 15,215 cases were processed and 214 were excluded because they were missing one or more discriminating variables. The underemployed group contained 5,011 cases while the adequately employed group contained 9,990 cases. A single discriminant function was generated from the thirty-three grouping variables.13 Overall, the 1 2 A correlation matrix of the grouping variables can be found in Appendix A. The absence of perfect colinearity among the grouping variables warranted the inclusion of all thirty three independent variables in the discriminant analysis.. 1 3 The number of possible functions is always the lesser of either 1) the number of groups minus one or 2) the number of discriminating variables (Klecka 1980:34). Since this is a two-group discriminant analysis, only a single function is derived. 61 mirty-three combined variables discriminate moderately well between the adequately employed and the underemployed (canonical corr. = 0.6820, chi-square = 9376.62, df= 33, p=.000). Based on a review of the standardized function coefficients, General Program Area and Occupation type are the key discriminating characteristics of the function equation (Table 18).14 Relative to students from Vocational programs, students from Academic and Career programs are significantly more likely to be underemployed. The Academic (coeff = 0.661) and Career (coeff. = 0.509) program area dummy variables rank first and third in their total relative contribution to the discriminant function. Seven of the eight occupational dummy variables rank among the top ten discriminating variables in the function equation and the eighth occupational dummy ranks eleventh. Relative to occupations in Health, Sales and Service occupations are associated with higher rates of underemployment and rank second only to the Academic program dummy in terms of contribution to the discriminant function equation. Of the eight occupational types (and relative to Health occupations), five are more likely to result in underemployment. Three occupational types, Natural & Applied Sciences, Social Sciences/Education/Government & Religion, and Art/Culture/Recreation & Sport, are more likely to result in adequate employment. The standardized discriminant function coefficients are interpreted in relation to the mean function score for each group (the group centroid). In this case, the mean function score for the underemployed group is positive (mean= 1.31) while the group centroid for the adequately employed is negative (-0.65). 62 Table 18: Discriminant Analysis - Employment Status by Occupation, Industry, General Program Area, Field of Study and Personal Characteristics Standardized Di sc r iminan t T o t a l Independent Var iables F u n c t i o n Structure Coefficients Coefficients General Program Area (Dummy Variables) Gen Prog - Academic 0.661 0.664 Gen Prog - Career 0.509 -0.088 Industry Sector (Dummy Variables) Ind - Busn Srv Ind -0.047 -0.116 Ind - Consmr Srv Ind 0.100 0.440 Ind - Distr Srv Ind 0.097 0.054 Ind - Goods Prod Ind -0.040 -0.079 Ind - Pub Adm Srv Ind 0.000 -0.028 Occupational Type (Dummy Variables) Occ - Art/Cul/Rec -0.257 -0.212 Occ - Bus/Fin/Adm 0.313 0.062 Occ - Nat/Appl Sciences -0.196 -0.306 Occ - Sales/Services 0.600 0.578 Occ - Soc/Ed/Gov/Rel -0.237 -0.284 Occ - Trd/Trans/Equip 0.157 -0.174 Occ - Unq. to Primary Ind 0.265 0.141 Occ - Unq. to Proc/Util 0.313 0.176 Field of Study (Dummy Variables) Prg - Arts/Sci/Humanities 0.264 0.640 Prg - Agri/Nat Resources 0.037 -0.073 Prg - Busn Mng 0.014 -0.058 Prg - Cns/Mech/Mtl 0.099 -0.270 Prg - Communication 0.069 -0.028 Prg - Comp Inf Systems 0.040 -0.065 Prg - Education 0.102 -0.068 Prg - Eng/Elec 0.050 -0.145 Prg - Fine/Vis/Perf 0.102 0.031 Prg - Leg/Social 0.140 -0.007 Prg - Rec/Hosp Services -0.043 -0.105 Prg - Transportation 0.033 -0.065 Personal Characteristics Taken previous post-secondary =1 -0.074 -0.271 Visible Minority (minority = 1) 0.030 0.055 Complete Credential (completion = 1) -0.056 -0.567 Complete High School (completion = 1) 0.029 0.138 Disability Status (disabled = 1) -0.008 -0.040 Gender (Female = 1) 0.040 0.042 63 Relative to general program area and occupation, the contribution of industry sectors to the discriminant function is moderate to weak. Students employed in two industry sectors, Distribution Services (coeff. = 0.097) and Consumer Services (coeff. = 0.100) are more likely to be underemployed than students in the Education/Health and Social Services industries. The function coefficients for both of these industry sector dummies are roughly one sixth as large as the Academic program area and Sales and Service occupation coefficients. While Business Service (coeff. = -0.047) and Goods Producing (coeff. = -0.040) industries are statistically more likely to result in adequate employment, their substantive significance is negligible relative to the more prominent discriminating characteristics. The Public administration industry dummy has no independent discriminating effect in the function equation (coeff. = 0.000). With the exception of Liberal Arts, Sciences and Humanities (coeff. = 0.264), the contributions of the Field of Study dummy variables to the discriminant function range from moderate to negligible. Relative to Health programs, all but one field of study (Recreation and Hospitality Services - coeff. = -0.043) are more likely to lead to underemployment (see Table 17). After accounting for the unique effects of general program area, industry, occupation and field of study, the personal characteristics play a weak role in discriminating between the adequately employed and underemployed. Students with previous post-secondary experience (coeff. = -0.074) and those who completed the requirements for a credential (coeff. = -0.056) are more likely to be adequately employed. The unique effects of these two characteristics are very weak relative to the key variables in the function equation. Women (coeff. = 0.040), visible minorities (coeff. = 0.030) and high school completers (coeff. = 0.029) are also slightly more likely to be underemployed when compared to their opposites but the discriminating power of 64 these characteristics, while significant, is small. Finally, consistent with the bivariate analysis, individuals with a disability (coeff. = -0.008) are marginally more likely to be adequately employed. As discussed earlier, this is probably due, in part, to the large proportion of those with disabilities in vocational rather than academic or career programs. Classification Matrix The ability of the discriminant function to distinguish between the adequately employed is further supported by the results of a classification procedure. The extent to which the classification procedure can correctly predict group membership is an indirect measure of the utility of the discriminant function. Klecka states that "the proportion of cases correcdy classified indicates the accuracy of the procedure and indirectly confirms the degree of group separation" (1980:49). Based on the single discriminant function, the classification procedure correctly classified 84% of all cases (see Appendix B). An Alternate Interpretation of the Discriminant Analysis - Total Structure Coefficients The simple bivariate correlations between the discriminant function score and the independent variables, (total structure coefficients), provide an alternative method of interpreting the discriminant analysis. Unlike the standardized coefficients, the total structure coefficients do not take into account the effects of the other independent variables. The simple relationship between the discriminant function score and the independent variables help identify the kind of information being used to discriminate between the two groups. We can "name" a function on the basis of the [total] structure coefficients by noting the variables having the highest coefficients. If those variables seem to be measuring a similar characteristic, we could name the function after that characteristic (Klecka 1980:31). 65 Table 17 lists the total structure coefficient alongside the associated standardized coefficient for each independent variable. For this alternate interpretation of underemployment, I have chosen to examine only those variables that have a total or standardized coefficient larger than 0.200. While the 0.200 cut-off is somewhat arbitrary, a separate discriminant function, using only the fourteen variables with standardized or total structure coefficients larger than 0.200, supports this treatment. The reduced discriminant function yields a canonical correlation of .6693 (chi-square= 9030.129, df=14, p>.0000), less than 2% smaller than the original discriminant function which contained all thirty three independent variables. 66 Discriminant Analysis Total Structure Coefficients by Standardized Coefficients -0.600 -0.400 -0.200 Occ-Art/Cul/Rec 0.800 0.600 0.400 -0.200 0.000 Si IS Underemployed •Prg - Arts/Sci/Hum •jen Prog Academic •Occ - Sales/Srv • Ind - Consm Srv • Occ-Soc/Ed/Gov/Rel 4 Occ-Nat/Appl Sci Adequately Employed -0.200 • Taken -0.400 -• -0.600 Occ - Proc/Util ^Occ - Prim Ind • Occ - Busn/Fin Standardised Coefficients o.;;oo 0.400 0.600 Gen Prog Career 0.800 • Post-Sec Prg-Cns/Mech/Md Completed Cred A visual representation of the relative importance of each discriminant characteristic is provide in Figure 1. The characteristics deemed of lesser importance (with standardized and total structure coefficients less than 0.200) are enclosed in the box surrounding the center-point of the graph. The remaining variables have been retained for analysis because of their importance to either the discriminant function equation (standardized coefficients greater than 0.200) or the kind of the information being used to discriminate between the adequately employed and the underemployed (total structure coefficients greater than 0.200). The General Program Area variables (Gen Prog - Academic, Gen Prog - Career and the reference variable Gen Prog Vocational) and the two Field of Study variables (Prg -Arts/Sci/Humanities and Prg - Cns/Mech/Mtl) may represent an applied / non-applied program dimension. To the extent that a student is from an applied program versus an academic program, the greater the likelihood that individual will be adequately employed. The bivariate analysis of General Program Area earlier in this chapter supports this conclusion although its not certain that General Program area is in fact measuring an 'applied' skills dimension. Certainly, the B.C. system funding mechanism, by which the programs are defined as either Academic, Career or Vocational, includes an applied program dimension. Students from applied programs gain specific job skills and credentials that make them eligible for jobs in markets that are closed to others without the same training. This direct relationship between educational programs and jobs may account for the higher levels of adequate employment. Self-reported measures of training-to-job relatedness support the idea of a more direct link between applied program and subsequent occupation (Table 19). Students from the Applied Career (56.4%) and Vocational (59.7%) program areas report significantly higher levels of training relatedness than students from Academic programs (34.9%). This 68 relationship holds true even when we exclude students who are continuing their education and /or have not completed a credential. Table 19: Job to Training Relatedness by General Program Area Is Job Rela ted to Tra in ing? (a) Genera l P r o g r a m A r e a Very related Somewhat related N o t very related N o t at a l l related T o t a l A c a d e m i c 34.9 22.6 10.8 31.7 100.0 (333) (216) (103) (303) (955) Career 56.4 25.2 6.8 11.7 100.0 (1657) C740) (200) (343) (2940) V o c a t i o n a l 59.7 19.1 5.2 15.9 100.0 (2907) (932) (254) (774) (4867) T o t a l 55.9 21.5 6.4 16.2 100.0 (4897) (1888) (557) (1420) (8762) (a) excluding those continuing their education and / or not completing a credential The second and third dimensions discriminating between the adequately employed and the underemployed are both related to the occupational labour market destinations of college leavers. The first occupational dimension is a public sector/private sector service dimension. Occupations in Art / Culture / Recreation and Sport and Social Sciences / Education / Government and Religion (as well as the occupational reference group, Health) reflect the public sector funded service pole of this dimension while occupations in Business / Finance and Administration and Sales and Services reflect the mid-point and private sector funded service pole. College leavers employed towards the public sector services pole are more likely to report adequate levels of employment while those employed in the private sector services are more likely to be underemployed. 69 The second occupational dimension may reflect the level of technology used on the job. On the high technology end of this dimension are the occupations in Natural and Applied Sciences with higher rates of adequate employment. At the other end of this dimension are the occupations unique to the processing and utilities industries as well as those unique to primary industries. Both occupational groups associated with higher levels of underemployment may make lower use of technology and reflect low-level manual labour. The remaining variables may be part of one of these three discriminating dimensions or they may reflect other dimensions that have not been adequately covered in this analysis. It is probable that the Consumer Service industries dummy variable is associated with the private sector service pole of public/private sector service dimension. The Sales and Services occupational dummy variable and the Consumer Services industries variable are highly correlated. Compared to coefficient for Sales and Service occupations (coeff. = 0.600), the weak standardized coefficient for the Consumer Services industry dummy (coeff. = 0.100) reflects the large proportion of discriminating power which must be shared by these two variables. Credential completion, as discussed previously, is significantly associated with General program area and thus the applied/non-applied program dimension. This association is due, in part, to the nature of the survey population and the educational structure of academic, university transfer programs in B.C. colleges (see earlier discussion under Program Completion). Previous post-secondary experience may be a function of the applied/non-applied program dimension or it may be part of an unmeasured dimension. There is a significant association between General program area and previous post-secondary experience. Students 70 from Vocational (44.5%) and Career programs (58.9%) are one and a half to two times as likely to have had previous post-secondary experience compared to students from Academic programs (28.2%). In the absence of an age variable or another proxy for general experience, previous post-secondary experience, which contributes to higher levels of adequate employment, may be part of a weakly measured, general experience dimension. Summary of Discriminant Analysis The multivariate analysis defined the key characteristics which discriminate between adequate employment and underemployment. General Program type proved to be the most significant discriminating characteristic with students from Academic programs likely to experience higher levels of underemployment. While occupation also proved important in distinguishing between the adequately employed and underemployed, industry did not. This may be due to the nature of the new National Occupational Classification system. When forming the occupational types, Statistics Canada took into account, where appropriate, the industry location of specific occupations. The incorporation of industry information within the occupational classification schema may have resulted in a large portion of discriminating information being shared by the occupation and industry dummy variables as indicated by the small industry coefficients. Finally, the personal characteristics, gender, visible minority and disability status, proved less important than structural characteristics when distinguishing between the adequately employed and the underemployed. 71 Chapter 5 Conclusions In this final chapter I examine the adequacy of human capital theory, labour market segmentation theory and credentialism as explanations of underemployment. I follow this examination with a summary of the methodological and substantive limitations of this study. Finally, I conclude the chapter with a brief look at some of the policy implications arising from this research. The labour market segmentation perspective appears to be more appropriate than either human capital theory or credentialism in accounting for the underemployment of B.C. college leavers. Human capital theory seems least able to account for the high and varied rates of underemployment while a loose interpretation of the results suggests a tendency towards credentialism in the labour market for B.C. college leavers. The labour market for B.C. college leavers is clearly segmented with regards to underemployment. The markedly uneven rates of underemployment experienced by college leavers in the core and peripheral sectors, given similar levels of post-secondary education, supports labour market segmentation theory. Although the B.C. economy does not have many of the large, monopolistic industries associated with the core sector, it does have extensive core, public sector, employment in government, education and health. The rate of underemployment for college leavers employed in the primary sector (i.e. health, applied sciences, education, government) is significantly lower than the rate of underemployment experienced by those in the peripheral sector (i.e. sales and service, and business, finance and administration). The large role played by the peripheral sector, service economy in BC is clearly evident in the significant number of college leavers employed service occupations and industries. 72 If the labour market segmentation perspective is correct, then, in addition to high levels of initial underemployment, those leavers employed in the peripheral sector are likely to experience difficulty moving into the more advantageous primary labour markets of the core sector. This is particularly true for those former students employed in sales and service occupations where the internal job markets and chances for advancement are negligible or non-existent. Not only are these individuals disadvantaged now, they will likely find it difficult to improve their market standing. It would be difficult to explain how the results of this study could support the central tenet of human capital theory. Overall, the finding that a third of college leavers are underemployed is sufficient in itself to demonstrate that an investment in higher education does not automatically guarantee employment at a level commensurate with that education. Human capital theory could account for this discrepancy by saying diat the supply/demand would return to balance over the long run (but see below). While human capital theory may be useful to inform a discussion of differential economic returns to education at highly aggregated levels (i.e. post-secondary education vs. high school education) it is insufficient to account for the wide variation in underemployment given similar levels of post-secondary education. The finding that structural market conditions and field of study are more important in determining underemployment than personal characteristics, like gender and visible minority status, does seem, at first glance, to support the human capital notion that markets for labour operate in a rational manner. According to this notion, the prominence of structural characteristics over personal characteristics in explaining underemployment would suggest that the labour market is responding in a rational manner to changes in the supply and demand for various types of skilled labour. Conversely, the weakness of gender and visible minority as 73 explanations of underemployment suggests that the labour market is not shaped by irrational, long-standing inequities based on personal characteristics. However, given the conflicting evidence on salaries and part-time employment, the notion of rational labour markets is also questionable. The existence of a marked salary gap between male and female college leavers, within and between market segments, belies such an interpretation and suggests that the underemployment construct may simply not be appropriate as a measure of inequity within the labour market (see Limitations for a further discussion). While the evidence for credentialism is not strong, I do not believe that these findings allow me to entirely reject a credentialist interpretation. At the very least, the results of this thesis suggest that employers are not unwilling to hire college leavers with more education than is required to perform the job at hand. Additionally, there is some evidence to suggest that B.C. employers may actually prefer over-qualified employees. A recent article in the Vancouver Sun on the increasing demand for university graduates noted that "employers are now hiring university grads for jobs that once would have been sneered at by those with degrees. Starbucks, for instance, likes to hire university graduates for jobs serving coffee" (Alden 1997:A9). Although this evidence is largely anecdotal, it would be a mistake to assume that employers would choose applicants with only the minimum essential qualifications over those more highly qualified. Limitations The principle methodological limitation of this study is common to most research on mismatched underemployment: although the concept of underemployment is relatively straight forward, it is very difficult to create a valid operational measure (see Chapters Two and Three). The measure of underemployment I used here relies, in turn, on the comparison of two 74 relatively weak measures of skill: skills acquired through formal education and skills required by an entry level job. The measure of skills acquired through education I used in this thesis provides only a small improvement over the simpler measure of years of education. The program type variable I used to operationalize acquired educational skills is peculiar to the BC post secondary system and could not be easily used outside the province. In addition, while program type reflects both the duration and the broad type of skills acquired (the latter by credential offered) it still does not define just what those skills might be. By default, this omission suggests that one vocational certificate or technical diploma is as good as another. This is clearly not the case. To the extent that college leavers are employed in jobs that are commensurate with the level of education they have acquired but not with the type of skills they learned, this measure underestimates underemployment. The measure of entry level job-skill requirements is equally problematic particularly since they are measured according to an educational metric. The skill level criteria of the new National Occupation Classification system (NOC) is a proxy of job skills measured by duration and type of formal education. While this treatment provides for a convenient comparison of acquired and required skills it also extends all the problems associated with measuring educational skills to the measure of entry-level job skill: namely, the false assumptions of job-skill homogeneity and the transferability of job-skills. The operationalization of underemployment used in this thesis is also limited by the exclusion of an individuals' total educational experience. As noted in Chapter Four, 41.7% of the employed college leavers in this study had previous post-secondary experience. While I 75 believe there are good reasons to exclude previous educational experience from the underemployment variable as it is operationalized here, they are methodological rather than substantive reasons (see Chapter Four). However, it is a weakness of the current operationalizations of underemployment that they do not adequately account for those students with a patched history of educational experience. If we believe that education does have a skill transmission function beyond the credentialist viewpoint then all educational experience is important if we are to understand the match between education and job skills. The operationalization of underemployment is not the only methodological limitation of this thesis. The operationalization of occupational type may also be problematic. The ten broad occupational or skill type categories of the NOC used in this thesis are not comparable to the earlier more traditional occupational groupings used in Canadian research on overeducation specifically or employment in general (Hunter 1986; Myles et. al, 1988; Krahn and Lowe 1990; Redpath 1990). The N O C skill type groupings collapse occupations across the type of work performed, educational background and industry. Most significantly, whereas professional and technical occupations are treated in the literature as distinct occupational groups (Myles et. al., 1988; Hunter 1986), the N O C includes them within each occupational skill type. The result is the combining of; (i.e.) mail and message distribution occupations with auditors and accountants; doctors and dentists with supporting occupations in health services; and marketing managers with cashiers. This treatment likely reflects Hunter's concern when he noted the failure of the aggregate level CCDO data to provide "...sociologically meaningful sets of categories" (Hunter 1986:52). Finally, I believe this study highlights a substantive limitation of the underemployment construct. Mismatch underemployment is highly insensitive to labour market inequities. I draw 76 this conclusion based on the finding that both gender and visible minority status are either weakly or insignificantly associated with underemployment. While female B.C. college leavers experience significantly and substantively higher levels of part-time employment and lower salaries they report only marginally higher levels of mismatch underemployment. These limitations, noted, it is important to stress that underemployment is a much discussed but little researched topic. Although there are limitations to my research, the evidence I introduce is significant in being some of the first research in B.C. on this important topic. Policy Implications The results of this study are pertinent to two broad policy issues affecting higher education: accountability and equity. The expansion of the Canadian post-secondary system that occurred in the sixties and seventies has given way to frozen funding and an intention by governments to re-engineer higher education to better suit the labour market. With a firm belief in the link between education and employment but suspicious of the value of the returns, governments have abandoned investment in favour of "testing and monitoring, efficiency and effectiveness, standards and targets" (Gaskell 1993:4). The Economic Council of Canada's final report summarizes the governments current position. A nation can achieve "more bang for the buck" from its education and training systems if there is coherence with the labour market and with economic performance... (ECC 1992:40). The measure of underemployment provided by the skill-level criteria of the National Occupational Classification schema is likely to become a key feature in governments' push to ensure the relevance of higher education to the labour market. While the results of this thesis 77 demonstrate that there is indeed underemployment among college graduates they also suggest that some of the primary determinants of underemployment are not under control of the educational system. Structural characteristics of the labour market proved to be the key feature in determining underemployment among B.C. college leavers. So long as the service sector continues to be a major source of new jobs in the B.C. economy, we can expect to see endemic underemployment, particularly among occupations in the sales and service sector. If underemployment is used as a measure to compare the labour market relevance of different programs it should only be done with extreme caution and not without longer-term research. The governments' desire for a demonstration of short-term labour market outcomes favours applied over academic programs. While students from academic programs show higher rates of underemployment in the short-term, there is evidence to suggest that the long-term employment returns to an academic education are better than those for applied programs. Policy decisions based only on short-term accountability measures are likely to seriously underestimate the longer-term relevance of academic programs. In as much as policy makers are concerned with issues of equity as well as accountability, there is a danger that a measure of underemployment will lead to false positive conclusions about labour market equity. While gender appears to have little effect on the extent of underemployment, an examination of salary data and part-time employment status demonstrates that a significant salary gap still exists between male and female college leavers. We must be careful in our desire to improve equity in education and the labour market that we do not simply provide ghettoized training for ghettoized jobs. 78 In closing, I believe that underemployment is a useful, though limited construct: while the concept appears deceptively simple, the practice of measurement is not. Such a measure should only be used in conjunction with other measures of employment outcomes like unemployment, salary and full/part-time employment status. In addition, the temporal relevance of education to work should be accounted for whenever possible. It is likely, however, that this measure will become more prominent given the more accessible measures of job-skill requirements in the NOC. Every effort should be made to improve the operationalization of underemployment and to clearly delineate its appropriate uses and limitations. 79 B I B L I O G R A P H Y Anisef, P., F. Ashbury and A. Turrittin 1992 Differential effects of university and community college education on occupational status attainment in Ontario. Canadian Journal of Sociology. 17(1):69-Anisef, P., and P. Axelrod 1993 Universities, graduates, and the marketplace: Canadian patterns and prospects. in P. Anisef and P. Axelrod, eds., Transitions: Schooling and employment in Canada, Toronto: TEP. Ashton, David 1988 Sources of variation in labour market segmentation: A comparison of youth labour markets in Canada and Britain. Work, Employment & Society. 2:1-24. 84. Berg, Ivar 1970 Education and Jobs: The Great Training Robbery. Boston: Beacon Press. Blaug, M . 1982 Where are we now in the economics of education} Directorate for Social Affairs, Manpower and Education, Study group on educational development. Paris. Boyd, Monica 1990 Sex differences in occupational skill: Canada, 1961-1986. Canadian Review of Sociology and Anthropology. 27(3):285-315. Burris, Beverly H . 1983 The human effects of underemployment. Social Problems 31(1):96-110. Burris, Val 1983 The social and political consequences of overeducation. American Sociological Review 48:454-67. Clairmont, D, R. Apostle and R. Kreckel 1983 The segmentation perspective as a middle-range conceptualization in sociology. Canadian Journal of Sociology 8:245-271. Clairmont, D, MacDonald, and F. Wein 1980 A segmentation approach to poverty and low wage work, in John Harp and John Hofley, eds. Structured Inequality in Canada. Englewood Cliffs, N.J.: Prentice-Hall. 80 Clark, W, and Z. Zsigmond 1981 Job Market Reality for Postsecondary graduates: employment outcomes by 1978, two years after graduation. Catalogue 81-572E, Statistics Canada. Ottawa: Supply and Services Canada. Clogg, C , and J. Shockey 1984 Mismatch between occupation and schooling: a prevalence measure, recent trends and demographic analysis. Demography 21(2);235-257. 1985 The effect of changing demographic composition on recent trends in underemployment. Demography 22(3);395-414. Clogg, C , and T. Sullivan 1983 Labor force composition and underemployment trends, 1969-1980. Social Indicators Research 12:117-152. Clogg, C , and T. Sullivan and J. Mutchler 1986 Measuring underemployment and inequality in the work force. Social Indicators Research 18:375-393. Boylan, R. 1992 Structure in external labor markets. Research in Social Stratification and Mobility. 11;97-125. British Columbia Ministry of Education, Skills and Training. 1996 1995 B.C. College and Institutes Student Outcomes: Survey Results by Program for Former College and Institute Students - Appendices. MESTL British Columbia Ministry of Skills, Training and Labour. 1993 1993 B.C. College and Institutes Student Outcomes Report. MSTL Economic Council of, Canada. 1990 Good Jobs, Bad Jobs: Employment in the Service Economy. Ottawa: Ministry of Supply and Services. 1992 A lot to learn: education and training in Canada. Ottawa: Ministry of Supply and Services. Freeman, Richard 1975 Overinvestment in College Training. Journal of Human Resources 10:287-311. 1986 Demand for Education, in O. Ashenfelter and R. Layard, eds. Handbook of Labour Economics, Volume One, BV:Elsvier Science Publishers. 1976 The decline in the economic rewards to college education. Review of Economics and Statistics 59:18-29. 81 Gaskell, Jane 1993 Introduction in P. Anisef and P. Axelrod, eds. Transitions: Schooling and employment in Canada, Toronto: TEP. Guppy, Neil 1992 Does school matter: An invited comment on Anisef, Ashbury, and Turrittin's "Differential effects of university and community college education on occupational status attainment in Ontario" Canadian Journal of Sociology. 17(l):85-99. Halaby, Charles 1994 Overeducation and Skill Mismatch. Sociology of Education 67:47-59. Hachen, D. 1988 Industrial labor markets and job mobility rates. Research in Social Stratification and Mobility 7:35-68. Hunter, Alfred 1988 Formal education and initial employment: unravelling the relationships between schooling and skills over time. American Sociological Review 53:753-65. Hunter, Alfred and Leiper J. McKenzie 1993 On formal education, skills and earnings: The role of educational certificates in earnings determination. Canadian Journal of Sociology 18(l):21-42. Hunter, Alfred and M . Manley 1986 On the task content of work. Canadian Review of Sociology and Anthropology. 23(1):47-71 Krahn, Harvey and Graham Lowe 1990 Young workers in the service economy. Ottawa: Economic Council of Canada, Working Paper No. 14. 1993 Work, Industry and Canadian Society. Scarborough: Nelson Canada. Livingstone, David W. 1987 Job skills and schooling. Canadian Journal of Education 12(l):l-30. Myles, J. 1988 The expanding middle: Some Canadian evidence on the deskilling debate. Canadian Reivew of Sociology and Anthropology. 1988;25:335-364 Myles, J., G . Pico and T. Wannell 1988 The changing wage distribution of jobs. The Labour Force. Statistics Canada, (October) :85-l 38. 82 Redpath, Lindsay 1991 The causes and consequences of education-job mismatch: a study of underemployment among Canadian university graduates, 1985-87 diss., Faculty of Arts, Department of Sociology, University of Alberta. Shultz, T. 1961 Investment in Human Capital. The American Economic Review 51(1):1-17. Shockey, J. 1989 Overeducation and Earnings: A structural approach to differential attainment in the U.S. labour force (1970-1982). American Sociological Review. 54(October):856-864. Smith, Herbert 1986 Overeducation and Underemployment: and Agnostic Review. Sociology of Education 59(April):85-89. Tipps, H. , and H . Gordon 1985 Inequality at work: race, sex and underemployment. Social Indicators Research. 16;35-49 Teichler, Ulrich 1990 Occupational Structures and Higher Education in Burton Clark ed. International Encyclopedia for Higher Education. 83 A P P E N D I X A 84 oo Correlation Matrix of Discriminant Variables 00 X tj' I a •a r •a o in "a . a -a 00 T3 a • 3 T3 -a p •a o o 0 5 T3 & t C O a -o a a & oo H 0 O 3 T3 00 C 3 ffl TS •3 T3 C O p -a 5 •3 •3 S T3 w I 5 .g in 3 m T3 I w o3 3 00 00 a. i •9 Complete HS 1.00 Taken p-sec .05 1.00 Complete Cred -.08 .18 1.00 Disability -.05 03 M 1.00 Visble Minority .00 04 M -.01 1.00 Employed PT ~M -.04 -.09 02 .06 1.00 Avg. Mnth Sal -.05 .18 19 M -.08 -.41 1.00 Gender U» 01 -m ,01 -.03 .15 -.25 1.00 Ind - Goods Prod Ind -0} 02 ,(>0 -04 -.16 .27 -.30 1.00 Ind - Distr Srv Ind w W $1 -01 M -.06 .05 -.13 -.15 1.00 Ind - Consmr Srv Ind .04 -.17 -23 M 03 .13 -.36 .02 -.33 -.21 1.00 Ind - Busn Srv Ind .04 .04 .04 -02 .03 -.06 .02 ,00 -.18 -.12 -.25 1.00 Ind - Pub Adm Srv Ind .02 03 -M .01 -.05 .09 .02 -.12 -.07 -.16 -.09 1.00 Ind - E d / H t h / W e l Ind -.06 14 20 01 -.03 .11 .05 .33 -.27 -.17 -.37 -.21 -.13 1.00 Occ - Bus/Fin/Adm .041 M -&1 -01 .03 -.02 -.06 .19 -.08 .08 -.17 .34 .11 -.12 1.00 Occ - Nat/Appl Sci .03 .09 .09 00 ,09 -.11 .16 -.19 .11 .03 -.18 .22 .08 -.14 -.14 1.00 Occ - Soc /Ed/Gov/Rel -.01 .09 .04 .02 -02 .02 .02 .14 -.13 -.09 -.16 -04 ,00 .38 -.13 -.09 1.00 Occ - Art/Cul/Rec .03 .03 -.03 M -.02 % .03 -.05 •01 -.05 ««t 06 .07 -.05 -.10 -.06 -.06 1.00 Occ - Sales/Srv .04 -.15 -.21 -,oi .03 .17 -.34 .10 -.27 -.07 .58 -.17 -.08 -.17 -.31 -.20 -.19 -.14 1.00 Occ - Trd/Trans/Equip -.08 -.03 .07 .02 -.04 -.13 .23 -.39 .35 .17 -.08 -.13 -.07 -.21 -.19 -.12 -.12 -.08 -.27 1.00 Occ - Primary Ind ,0& -.03 -.06 -.81 -.03 -.06 .09 -.11 .23 -.05 -.06 -.06 .02 -.08 -.08 -.05 -.05 -.03 -.11 -.07 1.00 Occ - Proc/Util 0*» 1H: -.02 00 .02 -.05 .06 -.12 .31 -.04 -.09 -.06 -.04 -.10 -.08 -.05 -.05 -.04 -.12 -.07 -.03 Occ - Health -.04 .11 .19 -01 -,01 .05 .10 .22 -.14 -.10 -.21 -.12 -.06 .56 -.16 -.10 -.10 -.07 -.23 -.14 -.06 Gen Prog - Acad .12 -.22 -.66 -.03 ,0© .13 -.23 .06 -.07 -.03 .23 -.05 01 -.14 *m -.10 .06 .04 .22 -.13 .05 Gen Prog - Career .06 .20 .24 -.02 .04 -.09 .16 -.03 -.03 fl& -.14 .15 .06 .03 .04 .21 -.03 .06 -.13 -.14 -.04 Gen Prog - Vocat -.17 .04 .45 .05 -.03 -.06 .08 -.03 .09 .03 -.11 -.08 -.07 .11 -.02 -.08 -.03 -.09 -.10 .26 -.02 Prg - Agri/Nat Res .05 .08 m -.01 -.05 .09 -.05 .11 -.03 -.05 .60 .05 -.05 -.05 .22 -.03 -.02 -.06 -.04 16 Prg - Cns/Mech/Mtl 08 .08 .22 02 -.01 -.12 .19 -.35 .24 .10 -.03 -.07 -.06 -.17 -.14 .03 -.09 -.05 -.16 .52 01 Prg - Tranport OS .06 .03 -.04 -.03 .13 -.12 .08 .12 -.05 -.04 -.02: -.06 -.04 m -.02 -.03 -.06 .12 10 Prg - Eng/Elec M .10 .14 .02 -.07 .12 -.18 .17 .04 -.11 .05 M -.11 -.07 .27 -.05 -.04 -.10 .10 -.02 u o u O 1.00 -.06 01 -.02 M -M 07 0£ .06 Gray shaded boxes are statistically insignificant (p>0.1). OQCrQOQQfQCTOCfQOQCrQCrQ » 2 e 3. O 3 n i» &. o a B o 5 e ^ 2 s r _ ^ e B " g R « S K * 2 £. * I' E » ? i < f § b o b 4 m o O $ m _• O Cnmnlere HS . . . . IjKi:. . t m O O O *(35w O O Ni :-:Gjt: Tob-pri r»-<:er •:•:+:•:•:• i K S S S K & S S S Complete Cred 8 S ^ S 2 g ^ 2 3 Disability g S S S ^ S I S V i s b l e Minority O fc* St. CN to to Ui :;:(M*:;:;:<3*:-:-:«+:-: L»J <_n o p p o p f a ' ^ H - o Employed PT g ^ I § 2 3 ^ 2 Avg. Mnth Sal to o i—• o o o •:•©*•: o Gender l / i o to ~ J l / i to SOte Ln O VJCUUCI w ° 3 S ° 2 ^ S 3 111(1 - G o o d s P r o d 1 1 1 ( 1 .' .' .' + .' >* i . « . i— O O vtSfi O SSB*: O SJS*S O T n r l . r)Utr Srv Tnd K S S S i t S ' ^ K ^ I n d - Consmr Srv Ind >- o o i - o o sfiSi o to TnH . Busn Srv Ind to Ln Ln to Ln to 1 1 1 1 1 u Li Ml oiv lliu o o m m m © © :33S O © T n t l . p , , * , Adm Srv Ind Ln (O to '"V-** to 4*- iT-Ulll ^1V mu S S ^ g g g J S S ; Ind-Ed/Hlth/Wellnd K S S S ' S S ^ i o Occ - Bus/Fin/Adm § S S £ £ £ & £ § Occ-Nat/ApplSci 2 5 2 2 2 2 g Occ - Soc/Ed/Gov/Rel ^ i H & S S S g O o c - A r t / C u l / R e c C 8 & g 5 & 2 B & Occ - Sales/Srv ^ ^ 2 ^ 2 g b s O c c - Trd/Trans/Equip .' .' .' 1 /•* / / 1 b b b © : :C* eg » o o Occ - Primarv Ind to to to ^ H u i V L L 1 l u l l* 1 1y m u o © o < & C * o O - < a > o Occ - Proc/IJtil c v t o t o » * | t * t o m * » * . i i"W urn Correlation Matrix of Discriminant Variables Complete HS Taken p-sec Complete Cred Disability Visble Minority Employed PT Avg. Mnth Sal Gender Ind - Goods Prod Ind Ind - Distr Srv Ind Ind - Consmr Srv Ind Ind - Busn Srv Ind Ind - Pub Adm Srv Ind Ind - Ed/Hlth/Wel Ind Occ - Bus/Fin/Adm Occ - Nat/Appl Sci Occ - Soc/Ed/Gov/Rel Occ - Art/Cul/Rec Occ - Sales/Srv Occ - Trd/Trans/Equip Occ - Primary Ind Occ - Proc/Util Occ - Health Gen Prog - Acad Gen Prog - Career Gen Prog - Vocat Prg - Agti/Nat Res Prg - Cns/Mech/Md Prg - Tranport Prg - Eng/Elec -a 1.00 -.17 1.00 .13 -.46 1.00 .05 -.60 -.43 1.00 -.04 -.10 .12 -to 1.00 -.12 -.29 -.19 .47 -.05 1.00 -.04 -.10 -.02 .12 -.02 -.05 -.08 -.18 .18 .03 -.03 -.09 Gray shaded boxes are statistically insignificant (p>0.1). Correlation Matrix of Discriminant u iviu S 3 t l o t >. Variables -a n a u 0 is/Mech/ lg/Elec a 5 X \ u •a a. a a o health < 00 U o° > 00 jri/N. is/Mech/ anpor lg/Elec a rts/Sc • Fine/Vi •c/Hc ft iucati S ft 2 ft 8 ft 7 w U P W PQ -5 u • Fine/Vi Pi U W u u I a V EP EP EP EP EP EP EP EP EP EP EP o o 0 0 ft ft ft ft ft ft ft ft ft ft ft Prg - Busn Mng -.14 -.22 .21 .03 -.06 -.16 -.06 -.11 1.00 Prg - Arts/Sci/Hum -.16 .85 -.37 -.53 -.09 -.26 -.09 -.17 -.32 1.00 Prg - Communicat -.03 -.06 .11 -.04 -,0i -.03 ~*l -.02 -.04 -.06 1.00 Prg - Fine/Vis/Perf -.05 -.11 .23 -.10 -.02 -.07 -.02 -.04 -.08 -.13 -.02 1.00 Prg - Rec/Hosp Srv -.06 -.15 -.05 .19 -.02 -.07 -.02 -.04 -.08 -.13 -.02 -.03 1.00 Prg - Comp Inf Sys -.04 -.09 .18 -.07 -.02 -.05 -.02 -.03 -.06 -.09 -m -.02 -.02 1.00 Prg - Education -.06 .05 -.08 .02 -.03 -.08 -.03 -.05 -.10 -.16 -.02 -.04 -.04 -.03 1.00 Prg - Leg/Soc -.04 -.09 .07 .02 -.02 -.07 -.02 -.04 -.08 -.13 -.02 -.03 -.03 -.02 -.04 Prg - Health .72 -.28 .13 .17 -.05 -.14 -.05 -.09 -.17 -.27 -.03 -.07 -.07 -.05 -.08 00 00 Gray shaded boxes are statistically insignificant (p>0.1). 89 Discriminant Analysis Classification Matrix Actual Group Membership Underemployed Adequately Employed Predicted Group Membership No. of Under - Adequately Cases employed Employed 5011 4295 716 85.7% 14.3% 9990 1644 8346 16.5% 83.5% Percent of "grouped" cases correctly classified: 84.27% 90 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items