Environment and Planning A 2000, volume 32, pages 1841 ^ 1856
Mismatch unemployment and local labour-market efficiency: the role of employer and vacancy characteristics John Adams, Malcolm Greig, Ronald W McQuaid
Department of Economics, Napier University, Edinburgh, EH10 5DT, Scotland; e-mail: [email protected]
; [email protected]
; [email protected]
Received 25 March 2000; in revised form 21 June 2000
Abstract. Much of the theoretical and policy debate on local labour markets has focused upon improving the educational, training, and other `supply-side' characteristics of job seekers. However, complementary employer `demand-side' factors, in particular the characteristics of employers, job openings, and recruitment practices, are also important in local labour markets but have been relatively neglected in the literature. The authors investigate such factors and argue that job vacancies in an environment of high unemployment are not only the result of traditional `structural' or `skill mismatch' between local labour markets, but may result from a mismatch between employer and jobseeker expectations combined with an asymmetry of information between these groups, resulting in `frictional mismatch' within local labour markets. Skill shortages can therefore be seen as a subset of wider recruitment difficulties. The authors survey a sample of employers to examine their characteristics, vacancies, and recruitment practices. The results provide evidence on the importance of employer and job factors on the flow of people into job vacancies. They also indicate that certain recruitment practices and inherent characteristics of vacancies and employers result in job offers that are either unattractive or inaccessible to the unemployed and hence increase the duration of job vacancies. It is argued that research into and design of local labour-market policies need to take more explicit account of the employer and job characteristics when discussing skill mismatch.
1 Introduction Conventional labour-market research and policies have highlighted deficiencies in the skills and other supply-side characteristics of job seekers as major causes of difficulties in filling job vacancies and of unemployment. However, the importance of employer and vacancy characteristics in prolonging vacancy duration in local labour markets has been given less consideration. These need to be more fully considered in order to provide additional understanding of their causes so as to increase the speed of flow of job seekers into new employment and reduce the costs to employers of unfilled vacancies. One set of standard explanations for persistent unemployment has particularly focused on the supply side (CEC, 1994; OECD, 1994; 1996). In the context of the United Kingdom the supply-side approach to improving international competitiveness is rooted in the argument that one of the fundamental economic constraints is an inflexible, lowskilled, low-mobility labour force, and that this represents a key supply-side constraint to economic growth which should be addressed through long-term investment in human capital (for example, DfEE, 1996a; 1996b). However, while supply-side issues are important, after nearly two decades of supply-side-orientated labour-market policy in the United Kingdom, unemployment in many local labour markets remains high and persistent, and competitive models have had only partial success in explaining involuntary, cyclical, or structural unemployment (Manning, 1995). Supply-side policies such as the New Deal in the United Kingdom have been criticised for failure to take account of local demand-side conditions (Campbell and Duffy, 1992; Fine, 1998; Peck, 1999; Turok and Webster, 1998). It has been argued that greater understanding of demand-side factors, such as the role of employer expectations and practices is necessary (Blanchflower and Freeman, 1994; Shackleton, 1995).
J Adams, M Greig, R W McQuaid
The supply-side emphasis (in both theory and policy) tacitly accepts a view of the demand for labour as largely given. For example, it is controlled by the profit-maximisation criterion and presents de facto the true number and true quality of workers which firms wish to employ. However, it is extremely rare in the mismatch or search literature that the declared ex ante needs of employers are critically assessed (Osberg, 1995) yet the ex post skill profile of many employers in the United Kingdom does not often conform to these ex ante requirements. That is not to suggest that employers are in some sense deliberately giving the wrong signals to the education and training sectors but rather that the true labour-skill requirements of a particular vacancy tend to be misspecified for a number of reasons. For example, many of the long-term unemployed may not be seriously considered by employers for jobs because it is presumed they lack certain characteristics which the employers consider important, such as work preparedness or experience. Different local labour markets are characterised by different flows of employed and unemployed people in and out of the labour market and jobs (or `labour-market flow regimes') which are partly a result of the underlying economic and industrial conditions and characteristics of the local economy (Martin and Sunley, 1999). These conditions and characteristics include both labour supply and local employer demand factors, as well as more general low aggregate demand, and are particularly important in local labour markets with persistently high levels of unemployment. (1) Hence skill shortages, and indeed many supply-side factors, can be considered as a subset of more general recruitment difficulties rather than the problem being one of skill shortages per se or purely of skill mismatch. So specific non-supply-side factors need to be examined in order fully to explain the extent of local labour-market inefficiency. In this paper we aim to build upon existing knowledge of the demand side of the labour market by offering a detailed analysis of the importance of employer demand factors through an examination of vacancy duration at a local level. The next section presents a theoretical analysis of unemployment and assesses the extent to which causal factors of unemployment have been examined in the economic literature. The third section details the specification of the model used to examine the influence of employer and job-vacancy factors upon vacancy duration and outlines the methodology used in the empirical research. Section 4 presents the results, both descriptive and analytical, from the survey of employers. In section 5 we present the conclusions and examine the policy implications. 2 A theoretical perspective on unemployment in local labour markets Unemployment can be caused by a shortfall in labour demand, for instance, as a result of insufficient aggregate demand in the economy, or a wage level above that required for market clearing (sometimes termed disequilibrium unemployment). Unemployment may also occur where supply of and demand for labour in the economy as a whole are equal, but vacancies and unemployment coexist because of a failure of supply and demand to clear either within or between specific submarkets (sometimes termed equilibrium unemployment). Inefficiencies or failure within a submarket can be termed `frictional' unemployment, while failure between submarkets can be referred to as `structural' or `mismatch' unemployment. In any local labour market there may be demand-deficient, frictional, and structural/mismatch unemployment. (1)
Low aggregate demand in a local economy may be caused by factors such as low national demand or an adverse industrial structure of the local economy. In this paper, however, we concentrate upon characteristics of labour demand by the existing local employers, which we term `employer demand factors'. These, of course, are influenced by national and sectoral factors as well as local labour-supply factors.
Mismatch unemployment and local labour-market efficiency
The term `mismatch' is traditionally used to describe a structural or skills mismatch of labour-force skills and employer requirements across or between labour markets, for example, a simultaneous surplus of job seekers with steelworking skills and excess demand for those with information technology skills. However, there may also be an asymmetry between the requirements of employers and job seekers within labour markets (frictional mismatch). For example, an employer recruiting in a market where there exists an adequate supply of labour may still experience recruitment difficulties because of its reputation amongst job seekers for low pay or poor working conditions, or because of the type of contract (as opposed to the type of job) offered. Another important form of frictional unemployment is spatial mismatch due to geographical immobility within local labour markets (Clark and Whiteman, 1983; Holzer, 1991; Ong and Blumenberg, 1998; Stewart et al, 1998; Van den Berg and Gorter, 1997), although there may also be geographical immobility (structural unemployment) between labour markets. Preston and McLafferty (1999) argue that spatial mismatch is a contingent phenomenon influenced by the attitudes and practices of employers, as well as the characteristics of the job seekers and the local economy, and a greater understanding of each of these is required. The causes of frictional and structural unemployment result from imperfections in both the supply (job-seeker) and the demand (employer) sides, and also from factors which may be regarded as exogenous to this employer ^ job seeker relationship. This is a dynamic process with the flows in and out of employment varying over time and across space (Clark et al, 1986; Martin, 1997). The speed at which submarkets clear is determined by the efficiency of the job-matching process, so the efficiency of the local labour market will determine the level of equilibrium unemployment. Conventional models of mismatch unemployment usually rest upon the assumption of what we have termed structural mismatch. They tend to concentrate on supply-side factors, and focus predominantly upon structural elements of mismatch, highlighting areas such as skill shortages, wage demands in specific markets, interregional spatial mismatch, and the search channels used by job seekers (Holzer, 1988; Layard et al, 1991). In many cases, underlying this is the assumption of the labour market as a commodity market with supply of and demand for labour and a point of equilibrium. In such models, an increase in skill levels of some job seekers will increase their employability without affecting the employability of any others. In other words, the supply of labour in the economy will increase and unemployment will fall. This can be seen as a consequence of the presence of unmet labour demand or essentially an extension of Say's law that supply will create its own demand, that is, that increasing the effective supply of skilled labour in the economy will generate sufficient economic growth (for example, through increased firm competitiveness) to create a market for this labour. However, this view has been challenged by the job competition theory (Thurow, 1975), which argues that a policy of increasing human capital levels for part of the workforce will not add to the overall employability of the workforce, but will merely displace those who have less human capital further down the `queue' for jobs. Hence unemployment will remain unchanged and supply-side policies in isolation will serve only to redistribute jobs. In essence (at the extreme) there is no structural mismatch as there are currently sufficient skilled job seekers available and training more job seekers will not increase the numbers of people in employment. So the job competition theory suggests that supply-side policies would be more effective if implemented alongside complementary macro and micro (employer) demand-side policies. However, supplyside policies that increase labour-market efficiency may still reduce frictional mismatch by increasing the speed at which people in Thurow's job queue are absorbed into employment.
J Adams, M Greig, R W McQuaid
We propose, therefore, that a complete explanation of local labour-market efficiency involves analysing both frictional and structural elements and that these elements require an in-depth examination from a demand-side perspective to complement research carried out on the supply side. Figure 1 illustrates the conceptual basis of the approach taken in this paper, although some factors may overlap both frictional and structural elements. This figure shows that there are several possible demand-side factors that may contribute to unemployment. We classify the predominantly frictional demand-side factors, represented by the upper left set, into four principal areas: vacancy characteristics, employer characteristics, the recruitment practices (search channels) used by employers, and discrimination by employers. In addition, the third factor (employer search channels) and the efficiency of matching technologies, such as job centres, will impact upon efficiency both within and between local labour markets (that is, will influence both frictional and structural unemployment). Industrial change or turbulence will contribute primarily to structural and skill mismatch. These are further described below. On the supply or job-seeker side, other parts of the model include personal transferable skills, job-seeker search channels, job skills, and geographical mobility. While we acknowledge the importance of these supply-side factors, it is the effect of frictional demand-side factors that are examined in this study. The four principal demand-side factors are now considered in turn. Characteristics of employers such as size, sector, and reputation may affect the attractiveness of vacancies to job seekers. Bosworth (1993) examined the effect of certain firm-specific variables on skill shortages and found that firms which are large, independently owned, and in the hotel and catering sector are associated with higher levels of skill Frictional
Vacancy characteristics Employer characteristics Employer (demand) side
Industrial structure and turbulence
Job seeker (supply) side
Personal transferable skills
Job-seeker search channels Geographical mobility
Figure 1. Determinants of equilibrium unemployment.
Occupational mobility (job skills)
Mismatch unemployment and local labour-market efficiency
shortages whereas retail, distribution, and business service firms experience fewer skill shortages. Additionally, studies in the USA have shown that employers with a poor reputation amongst workers tend to attract fewer applicants (Fombrun and Shanley, 1990; Gatewood et al, 1993). Russo et al (1996) analysed the effect of employers' characteristics and practices on the distance between employees' place of residence and work, arguing that harder-to-fill vacancies will exhibit a greater distance. They found that large firms are associated with greater recruitment distance, whereas firms in the manufacturing sector are associated with a shorter distance. (2) The relationship between firm size and recruitment distance may suggest that a large firm is likely to have less variance in its requirements than a group of smaller firms employing the same number of people. Second, the nature of vacancies includes factors such as job requirements set by employers based upon their needs (for example, education level, experience, flexibility, and personal skills). A misspecification of requirements can arise through information asymmetry between employers and job seekers (Polacheck and Yoon, 1987), leading to unemployment. Van Ours and Ridder (1993) found that vacancy duration is higher when high levels of education and experience are required, while Bosworth (1993) found that professional/management and part-time vacancies are both associated with high levels of skill shortages. These longer vacancy durations may reflect mismatch or the lower density of workers per job requiring higher educational or skill levels. It is important to note that, although in general longer vacancy duration may reflect labour-market inefficiencies, in some cases it may result in more suitable job seekers being employed or lead to fairer recruitment practices and so may be associated with greater labour-market efficiency. Another characteristic of vacancies requiring high levels of qualifications and experience is that they are associated with greater recruitment distance, whereas vacancies which require few or no qualifications and impose no age restriction are associated with a shorter travel distance (Russo et al, 1996). Discrimination by employers for reasons such as gender, race, or employment status is also important. This is often linked to the major problem in some of the United Kingdom's local labour markets that many of the long-term unemployed, and even those in low-paid employment, are effectively excluded from mainstream society (Green, 1997). First, discrimination on race has been widely studied, but it also affects groups such as married women who are regarded by employers as less likely to be committed to work than men or unmarried females (Collinson et al, 1990). Second, Griffith et al (1991) show that long-term unemployment sends negative signals to potential employers and this can create hysteresis effects with a high level of pathdependent long-term unemployment in a region (Jones and Manning, 1992). This is a dynamic process in which Van den Berg et al (1999) show that exit rates from unemployment to employment exhibit a negative duration dependence after the first quarter. Third, the type of job contract offered with a vacancy may also be influential in discouraging certain groups in the population. Campbell and Baldwin (1993) argue that skill shortages are only a subset of recruitment difficulties in general, and found that vacancies offering unsocial hours and low wages experienced the highest levels of recruitment difficulty. For example, they found that unsocial hours were a factor in 21% of hard-to-fill vacancies. This is a growing source of recruitment difficulty as firms adopt increasingly flexible work patterns (Hakim, 1993; Watson, 1994). (2) A longer distance implies a larger geographical catchment and potential labour force for such jobs and hence lower expected mismatch than in a smaller geographical area. Conversely potential workers living in areas of low demand or who have hard-to-place skills may be expected to look further for jobs.
J Adams, M Greig, R W McQuaid
The third principal category concerns the recruitment practices or search channels used by employers. These are an important determinant of conventional frictional unemployment in an economy and therefore cannot be ignored in our wider model of mismatch unemployment. Recruitment practices vary between firms and the use of methods such as employment agencies is increasing by both employers and job seekers (Watson, 1994). Roper (1988) examined the effects of employers' recruitment practices on vacancy duration, and found that firms which use more than one recruitment method, along with those which use informal recruitment methods and the government's Employment Service experience the shortest vacancy duration, whereas firms who advertise in national newspapers experience the longest duration.(3) Russo et al (1996) found that firms with efficient personnel departments and those which use informal interactive recruitment methods are associated with a smaller recruitment distance. Formal recruitment practices adopted by employers are also associated with a more accurate requirements specification, which would imply a shorter vacancy duration (Collinson, 1988). Finally, the nature of the external economic environment surrounding the firm may influence the level of mismatch unemployment and thus needs to be included. Labour supply and demand within the local market, measured directly through unemployment (U ) and vacancy (V ) rates, will influence unemployment and vacancy-duration levels. A relatively high U=V ratio implies more job seekers per job and therefore a shorter vacancy duration. In addition, high unemployment levels in the local labour market may reduce voluntary quits, although labour-market size and density are also likely to influence this (see below). The model described above contrasts with more limited models (H0 ) that concentrate on supply-side factors, leading to the following hypotheses regarding the causal factors of equilibrium unemployment, U. H0 : U f
s, w, c j , i, m , H1 : U f
s, w, c j , i, m, e, v, c e , d, l , where s is job-seeker skill levels, w is workforce flexibility (mobility), c j is search channels used by job seekers, i is industrial turbulence, m is efficiency of matching technology, e is characteristics of employers, v is vacancy characteristics, c e is search channels used by employers, d is discrimination by employers, l is local labour-market characteristics. The relative effects of both macroeconomic and efficiency factors upon the change in unemployment are often represented in the literature by a Cobb ^ Douglas matching function (Dur, 1999; Gorter et al, 1997; Lindeboom et al, 1994; McQuaid, 1986) of the form: F lJ a V b ,
where F is the flow of job matches (hires) in a given time period, l is a measure of (3)
National newspapers may be used because local media are expected to generate too limited a field of applicants who meet the minimum requirements. However, it is difficult to control for the specific job-skill requirements of jobs when comparing different search channels.
Mismatch unemployment and local labour-market efficiency
labour-market efficiency which will vary inversely with the level of mismatch, J is the number of job seekers (including both unemployed and employed job seekers), V is the number of vacancies available in the period, and a and b are geometric weights indicating the relative importance of J and V, respectively, where 0 < a, b < 1.(4) The number of successful matches, F, in a given time period, will be greater when the labour market is more efficient, and the greater the number of both vacancies and job seekers in the economy. In an area of relatively high unemployment the voluntary turnover of workers will be lower as fewer workers quit their jobs (Jones, 1992) and this will limit the movement or flows of both the employed and the unemployed. This interrelation among the different labour flows in the local labour market will be determined by the history and characteristics of the local economy but it will also help determine the `labour market regime' in the area (Martin and Sunley, 1999). Therefore, longer vacancy duration may result in greater unemployment (ceteris paribus) but may also be a consequence of higher unemployment. It is not possible to estimate F directly; however, data on vacancy duration are widely accessible. Vacancy duration is related to F, as the greater the vacancy duration, the slower will be the outflow, F, from the unemployment stock and those seeking to change jobs. Vacancy duration can be defined as V/F, the number of vacancies divided by the speed of outflow. Dividing both sides of equation (1) by V we obtain an expression for the matching rate per vacancy: y lJ a V bÿ1 ,
where y is the matching rate per vacancy, the inverse of which will give an expression for vacancy duration, T: T
1 ÿa 1ÿb J V . l
So the vacancy duration (the inverse of the matching rate per vacancy) is a function of the efficiency of the labour market and the stocks of job seekers and vacancies. As mentioned previously, the level of labour-market efficiency will depend partly on the level of frictional mismatch between job seekers and employers. A high level of frictional mismatch will reduce efficiency ceteris paribus. Efficiency can therefore be expressed as a function of the form: l CM ÿg ,
where M represents demand-side determinants of frictional labour-market mismatch (defined below in the empirical model), g is the geometric weight attached to this mismatch, 0 < g < 1, and C is a constant representing other determinants of efficiency, namely supply-side and structural mismatch factors [some other models also include employed job seekers under C rather than J (see Burgess, 1993)]. Equation (3) can now be rewritten as: T
1 J ÿa V 1ÿb , CM ÿg
which, for estimation purposes can be expressed in linear form: ln T ÿ ln C g ln M ÿ a ln J
1 ÿ b ln V . (4)
Weight a can be interpreted as the percentage change in the flow of job matches (hires) for a 1% change in J with V, l constant. Burda and Wypolosz (1994) estimated a matching function for four European countries, including the United Kingdom, and found an elasticity of matches (including exits from the labour force) of 0.5 ^ 0.7.
J Adams, M Greig, R W McQuaid
Equation (6) allows an estimation of the impact of demand-side frictional mismatch inefficiency upon vacancy duration while controlling for other inefficiency factors and job-seeker and vacancy levels within the local labour market. As no data are available on the numbers of employed job seekers in the local labour market under investigation, the unemployment level, U, is used as a proxy for J.(5) Equation (6) can then be approximated by: ln T ÿ ln C g ln M ÿ d ln U
1 ÿ b ln V .
The next section presents the detailed specification of the empirical model (7) and specific related hypotheses. 3 Empirical model 3.1 The survey and descriptive statistics
The areas covered by the survey were the former Edinburgh travel-to-work area (TTWA) and contiguous Bathgate TTWA, situated in east central Scotland. Edinburgh represents an urban labour market with a large financial and services employment base and Bathgate covered both rural and urban areas and had a relatively larger manufacturing base. The Lothian Business Index (LRDA, 1995) and the Lothian Region Development Authority data set provided an extensive database of employers covering both areas. The firms were in nonretailing industries and were stratified by SIC. A random sample of 20% in each industrial sector was selected from each area, and a total of 126 interviews were conducted. This sample consisted of 77 firms from the Edinburgh TTWA and 49 from the Bathgate TTWA, representing 17% of the database. Data were collected by means of a structured questionnaire administered via a series of face-to-face interviews. Employers were questioned on firm type and occupational structure, the nature of their most recently filled vacancy (for example, job type, contract, search channels used, qualifications required), and the duration of that vacancy. A brief overview of the descriptive statistics shows that 44% of vacancies were for unskilled and semiskilled manual occupations, 23% for skilled occupations, and 20% for management positions, highlighting a greater demand for unskilled workers. In terms of duration, 17% of vacancies were filled within a week. By four weeks 63% of vacancies were filled and only 8% were unfilled after twelve weeks. Employment contracts showed 82% of vacancies were for permanent full-time posts. The modal salary offered was between »10 000 and »14 999, although the distribution of salaries was uneven with nearly 40% of the vacancies offering wage levels which would be eligible for some form of state welfare support if the person taken on was a head of a household with no other income earners. In both areas, a large share (45%) of vacancies required no recognised academic or other qualifications which largely explains the wage structure in the vacancies available. However, nearly one fifth of the vacancies required degree-level candidates. Between these two extremes of the education/qualification spectrum very few of the vacancies required even the most basic formal qualifications in numeracy and communication ability. It is also notable from a detailed breakdown of the responses that a technical qualification (BTEC or HNC) was only required by 8% of the vacancies. In addition, vocational qualifications (SVQ or GNVQ) were stated to be minimum requirements for only 1% of the vacancies. Employers were asked to state the personal characteristics they were looking for in a job applicant. The majority of skills stated by employers were personal qualities (5)
If the proportion of unemployed within the total numbers of job seekers remains constant across regions and time (Gorter et al, 1997) then U should vary directly with J.
Mismatch unemployment and local labour-market efficiency
such as teamworking skills (cited by 95% of interviewees) and good communication skills and timekeeping (93% each). However, 27% of firms were resistant to employing someone who has been unemployed for six months or more. For hard-to-fill vacancies most employers cited more than one reason. The largest proportion of employers in the sample identified unsocial hours (39%), poor promotion prospects (34%), and `other' factors (38.3%) as the main barriers to recruitment in these vacancies. Skill shortages were cited by 18% of those experiencing a vacancy that was hard to fill. The model specifications and analysis are discussed in the next section. 3.2 Model specification and hypotheses
Vacancy duration provides a measure of intensity of a hard-to-fill vacancy and, unlike a dichotomous variable, allows an examination of the extent of recruitment difficulty. The model uses the log of the duration of employers' last vacancy as the dependent variable.(6) Previous studies have used similar measures of labour-market efficiency, for example, Roper (1988) in a study of the effects of employers' recruitment practices upon vacancy duration. Interestingly, Roper notes that if the duration of vacancies in the 5.5 million engagements in 1981 ^ 82 had been reduced by a day then unemployment would have fallen by around 20 000. Campbell and Baldwin (1993) examined the Table 1. Variables used in the model. Variable
Expected effect on vacancy duration
Local labour market lnUNEMRAT Unemployment rate in each area lnVACLEVEL Vacancy rate
Employer characteristics lnSIZE Number of employees Manufacturing industry lnMAN IND lnINDEP Independent firm Percentage of part-time workers lnPC PT Percentage of skilled workers lnPC SKIL lnUNION Unionised workforce
increase decrease decrease decrease increase decrease
Vacancy characteristics lnQUALEVEL Level of educational qualifications required lnPROFQUAL Professional or vocational qualifications required lnEITHER Neutral between male or female to fill vacancy Prefer employed or short-term unemployed applicants lnEMP STU lnPERM Permanent vacancy Vacancy involves shiftwork and/or unsocial hours lnSHIFT UN lnFTIME Full-time vacancy lnBENEFITS Employee benefits in addition to salary Manual vacancy lnMAN VAC Skilled vacancy lnSKIL VAC
increase increase decrease increase decrease increase decrease decrease decrease increase
Recruitment methods Recruit through government employment service lnJOB CENT lnAGENCIES Recruit through private agencies lnPRESS Recruit through newspapers lnUK Recruit throughout the United Kingdom lnINTERVS Number of interviews held Firm has personnel department lnPER DEPT
either increase either increase increase either
If a firm's vacancies are of various durations then there is a danger of some limited bias in estimating the average vacancy duration if only the last vacancy is taken.
J Adams, M Greig, R W McQuaid
probability that a firm will experience a vacancy which was hard to fill, but did not study the severity of this difficulty. Bosworth (1993) measured both the existence and the intensity of skill shortages, although on the current paper we go further by aiming to assess the impact of a range of possible determinants upon equilibrium unemployment and therefore require a direct and neutral measure of unemployment of which skill shortages are only a component. Duration is measured in weeks for an employer's most recently filled vacancy rather than for an ongoing vacancy and as such is not influenced by the timing of the interview relative to the date the vacancy was first advertised. This therefore represents the most current and accurate measure of vacancy duration. One vacancy per firm is measured (as opposed to multiple vacancies for fewer firms) partly to ensure that the data refer to the most recent case, as above, but also because this gives a more representative picture of the economy as a whole rather than the idiosyncrasies of individual firms. The independent variables chosen represent the employer, vacancy, and recruitment methods discussed in the previous section plus the local unemployment and vacancy levels. These variables are expected to affect the difficulty of filling, and hence the duration of, the employer's most recent vacancy. Table 1 defines these variables. The results from the model are now presented and discussed. 4 Model results The results generally support the descriptive data and add, we believe, considerable weight to our basic proposition that frictional demand-side factors are an important determinant of local labour-market efficiency. The logged value of vacancy duration is a continuous variable, therefore OLS multiple linear regression was used to estimate the model described in equation (7). The results are presented in table 2. The sample size falls to 102 when all regression variables are included. If the model is correctly specified then this should not be a source of bias in the parameters. First, examining the employer characteristics, vacancy duration increases significantly (at the 10% level) with a greater number of employees (lnSIZE) possibly because large firms may operate more formalised and bureaucratic recruitment practices which will extend the time required to fill vacancies (see discussion above which suggested possible opposing influences of firm size). As expected, firms with a greater percentage of skilled employees (lnPC SKIL) were also significantly (5% level) and positively associated with longer vacancy duration. Assuming such firms recruit a similar proportion of skilled workers, this may reflect a skill shortage amongst the more skilled professions or may be indicative of stricter recruitment practices at such levels or the longer notice that skilled employed job seekers must give to their current employers. It would appear that higher skilled vacancies require high levels of human capital and are generally positions of greater responsibility. The cost involved in an incorrect appointment for such a vacancy in terms of both salary paid and negative impact on firm performance would be far higher than for a lower level vacancy; hence firms will spend longer selecting an appropriate high-level candidate. Employers with a trade union presence in the workforce (lnUNION) experienced significantly (5% level) shorter vacancy duration times. This may simply be the result of trade unions pushing up wages within the firm relative to other (ununionised) firms and hence increasing the attractiveness of vacancies. Also it may be that a union presence increases the attraction of vacancies because of other factors such as better conditions of employment and/or that a union presence indicates an employer which enjoys a favourable reputation for employee welfare in general or that jobs are more clearly specified. The signs of the other employer characteristic variables were as expected, but were not significant.
Mismatch unemployment and local labour-market efficiency
Table 2. Analysis of determinants of vacancy duration. Variable Constant
lnSIZE lnMAN IND lnINDEP lnPC PT lnPC SKIL lnUNION
0.225 0.261 ÿ0.476 ÿ0.049 0.190 ÿ0.839
1.945* 0.655 ÿ1.363 ÿ0.832 2.154** ÿ2.169**
lnQUALEVEL lnPROFQUAL lnEITHER lnEMP STU lnPERM lnSHIFT UN lnFTIME lnBENEFITS lnMAN VAC lnSKIL VAC
0.218 ÿ0.450 0.732 ÿ0.301 0.725 0.295 ÿ1.310 ÿ0.136 ÿ1.090 0.592
0.697 ÿ1.040 1.980* ÿ0.807 1.606 0.869 ÿ2.331** ÿ0.380 ÿ2.191** 1.553
lnJOB CENT lnAGENCIES lnPRESS lnUK lnINTERVS lnPER DEPT
0.339 0.777 0.035 1.332 0.556 ÿ0.382
0.826 1.767* 0.094 2.424** 1.229 ÿ1.063
n 112 R 2 0:489 adjusted R 2 0:349 F 3:503 ** significant at 5% level; * significant at 10% level.
Second, examining the vacancy characteristics, a gender-blind attitude to recruitment (lnEITHER) had a significant (10% level) positive effect on vacancy duration. This may be because employers are using a formal, open recruitment process (based perhaps on their equal opportunities policies) in the expectation of getting better candidates (and greater long-term efficiency and fairness). Employers which are seeking a specific gender may have a more focused picture of their requirements of those they will consider, which could speed up the recruitment process, but may in some cases indicate potential discrimination (or lack of equal opportunities). Alternatively, the longer duration may reflect a higher level vacancy, for which the traditional gender roles associated with manual positions would be less important. Employer attitudes towards the short-term versus long-term unemployed had no significant effects, in contrast to expectations. The offer of a full-time contract (lnFTIME) significantly (5% level) reduces vacancy duration as expected, implying that such positions are easier to fill because of a greater willingness for job seekers to undertake full-time work. This may imply that a prevalence of part-time work represents a qualitative mismatch of requirements between employers and job seekers. Manual vacancies (lnMAN VAC) have a significant (5% level) negative association with vacancy duration, which may imply that there exists a greater supply of those able to carry out manual work compared with skilled work, and also
J Adams, M Greig, R W McQuaid
may reflect the less complex recruitment procedures involved in recruiting at such levels. The signs of the other vacancy variables were mostly as expected. Third, certain types of recruitment practices identified also significantly increase vacancy duration. The use of recruitment agencies (lnAGENCIES), and advertising the vacancy across the United Kingdom (lnUK), both significantly (at 10% and 5% levels, respectively) increase duration. Both of these recruitment practices could be seen as a reaction by the employer to a failure or expected failure to recruit by other initial methods. There may be several explanations for this. There may be an insufficient supply of labour locally to meet demand, the vacancy may be fundamentally unattractive to job seekers, or it may be that the initial channels are themselves ineffective. Alternatively, it is possible that the use of agencies and national advertising reflects a more thorough recruitment process for a higher level appointment, and the use of such methods is therefore only partly responsible for vacancy duration. The unemployment rate in the local area (lnUNEMRAT) had no significant association with duration. In a sample such as this, where few employers state a need for formal qualifications and so few report a shortage of skills as a problem, an abundance of unemployed labour combined with little reported evidence of skill shortages should result in vacancies being filled more quickly. Perhaps this is caused by non-supply-side factors influencing vacancy duration or by there being a lower worker quit rate in areas of high unemployment (as employers will generally prefer to hire those in employment and a lower quit rate will reduce the availability of this type of labour). When the insignificance of the vacancy level (lnVACLEVEL) is also taken into account, it could be argued that it is not the level of labour demand, but the qualitative nature of this demand which is partly responsible for frictional mismatch unemployment. The factors influencing vacancy duration appear, therefore, to have common themes: there is the established issue of possible skill shortages amongst job seekers; attractive contracts have the effect of decreasing vacancy duration as might be expected; there is the possible effect of the intensity and extent of recruitment policy; and some recruitment channels may be less efficient job-matching technologies than others or may be associated with recruiting different types of jobs. It is argued here that a shortage of skilled labour is not necessarily the principal cause of higher vacancy duration. There are three main reasons to support this. First, the descriptive statistics showed that less than one fifth of employers mentioned skill shortages as one of the difficulties they encountered in filling vacancies. Second, there was a large number of employers who expressed no requirement for formal qualifications when hiring, although they did express a desire for personal transferable skills. Third, and perhaps most importantly, the adoption of more thorough recruitment methods also increased vacancy duration and it is when attempting to fill higher skilled vacancies that these methods are most commonly utilised by employers. (7) Such longer vacancy duration need not necessarily mean an inefficient labour market as it may reflect fairer recruitment methods and so result in the employment of better candidates (resulting in greater efficiency for the firm, particularly when hiring expensive labour who are expected to add relatively high value-added to the firm).
(7) In order to investigate the existence of such an effect further, the sample was split between professional (managerial and other professional) and nonprofessional (manual and clerical) vacancies and two separate regression models were run in order to test for any differences in the causal factors of vacancy duration. However, no evidence of any differences was found. A possible explanation for this is that criterion used to define `professional' was the Standard Occupational Code which does not have any direct impact upon either salary or the method of recruitment.
Mismatch unemployment and local labour-market efficiency
5 Conclusions and policy implications This paper outlined a model of local labour-market efficiency based upon the distinction between a mismatch of job-seeker skills and employer requirements across or between labour markets (leading to structural or skills unemployment), and mismatch between the requirements of employers and jobseekers within labour markets (frictional unemployment). Even in equilibrium labour markets, information imperfections and asymmetries mean that the markets may not operate efficiently. If the labour markets are not in equilibrium, then filling vacancies more quickly may reduce overall unemployment. It was argued that the dominant supply-side policies had concentrated upon structural mismatch and largely ignored the role of local demand-side factors yet these may be significant. The empirical results provide evidence on the importance of employer and vacancy characteristics on the flow of people into jobs. The results show that a larger firm size, higher percentage of skilled workers, gender-blind attitude to recruitment, use of recruitment agencies, and UK-wide vacancy advertising were all associated with longer vacancy duration. The presence of trade unions, an offer of a full-time contract, and manual vacancies were associated with a shorter duration. Hence the hypothesis that employer demand-side factors are a significant contributor to vacancy duration and to labour-market efficiency and inefficiency is supported. Where longer vacancy duration results in more suitable job seekers being employed or in fairer recruitment practices, then labour-market efficiency may be enhanced. However, where this is not the case then longer duration leads to inefficiency. The results indicate that certain recruitment practices and inherent characteristics of vacancies and of employers result in job offers that are either more or less attractive or accessible to the unemployed and hence alter the duration of job vacancies. This suggests that factors influencing frictional unemployment are significant and theoretical and empirical models of mismatch should include both labour-supply and employer demand factors. The finding that the offer of a full-time contract has a negative coefficient indicates that vacancies for part-time posts take longer to fill. Part-time employment is attractive to employers as it allows them extra flexibility in expanding and decreasing their workforce to match their current demand. While the increase in part-time employment has benefited some workers for whom flexibility is paramount, the lower status and benefits compared to full-time employees makes part-time work unattractive for many, especially for those who are the only economically active household member. There has been a steady growth in the proportion of part-time employment as a percentage of total employment in the United Kingdom over recent years, and if this continues to raise the demand for part-time workers then it may reduce labour-market efficiency and increase frictional unemployment as it is not the type of work being sought by many job seekers. The job competition theory suggests that increasing supply-side skills will result in people moving up the `job queue' in the labour market, rather than decreasing unemployment levels. The lack of a significant relationship between the vacancy's duration and it being for a skilled employee (although the sign of the relationship is positive as expected) would suggest that overall there was little difficulty in filling most skilled posts, as also suggested by the descriptive statistics. This may provide some limited evidence that there is a degree of `job queuing'. However, contrary to this a greater percentage of skilled workers in a firm was positively associated with longer vacancy duration, so there may be difficulties in getting skilled workers to `fit into' such firms, or alternatively such firms may have more careful recruitment mechanisms, so the evidence is not clear.
J Adams, M Greig, R W McQuaid
There are a number of implications of giving a greater emphasis to non-supply-side factors. Some evidence exists of a screening out effect on the long-term unemployed so it would seem useful to relatively shift the emphasis away from training and towards job placement (although there will remain some people who will require training in basic and/or higher skills to enter the job market at an appropriate level). This has the advantage of reducing both the stigma associated with unemployment and the rate of depreciation of work-experience skills, which training programmes seem unable to do, and also of helping to overcome initial mismatch due to employer demand-side misspecification of job-seeker and vacancy characteristics. This is particularly important if employers themselves do not place significant value on the training programmes provided or on the vocational qualifications they generate (such as SVQ/GNVQs). The skills required by employers emphasise the importance of basic education rather than higher level or vocational skills. This is consistent with evidence in the USA which argues for a policy emphasis towards minimising unemployment spells through work, as opposed to training, and which suggests that this strongly maintains the reemployment wage rate (La Londe, 1995; Seninger, 1997). In terms of policy, a supply-side approach focusing primarily upon skills training would therefore be less likely to be fruitful than focusing upon basic education and transferable skills. One approach currently suggested is that which argues that unemployment in local communities can be addressed directly through the creation of intermediate labour markets (ILMs) (Emmerich, 1997; McGregor et al, 1997). These ILMs essentially represent a transient stopover between unemployment and reemployment in so-called `social economy organisations' providing standard wages in jobs which are of direct benefit to the local community in which the jobs are created. Such programmes can have the advantage in stressing work continuity and experience over training for low-skilled jobseekers, which may be more likely to be an attractive prospect for those employers who so far have not shown a great deal of interest in the output of past training schemes. Finally, it is argued that research into and design of local labour-market policies needs to take more explicit account of the employer and job characteristics when discussing skill mismatch. It is worth noting that a fundamental assumption with the UK approach appears to be that the demand for labour in the economy is buoyant enough to absorb the unemployed into jobs if only the rigidities on the supply side were removed. While supply-side policies such as skills training are important, these will be more effective if implemented in conjunction with complementary non-supply-side policies. These should not be limited to measures designed to increase aggregate demand at the macrolevel, but should include more specific policies to ensure a qualitative match of the requirements both of employers and of job seekers. Failure to implement such parallel policies may result in an underutilisation of labour and higher frictional and structural unemployment levels. Acknowledgements. The authors gratefully acknowledge the useful comments of two anonymous referees and the research assistance of Andrew Charlesworth-May and Beverley Christy. All errors remain those of the authors. References Blanchflower D, Freeman R, 1994, ``Did the Thatcher reforms change British labour market performance?'', in The UK Labour Market: Comparative Aspects and Institutional Developments Ed. R Barrell (Cambridge University Press, Cambridge) pp 51 ^ 92 Bosworth D, 1993, ``Skill shortages in Britain'' Scottish Journal of Political Economy 40 241 ^ 271 Burda M, Wypolosz C, 1994, ``Gross worker and job flow in Europe'' European Economic Review 38 1287 ^ 1315
Mismatch unemployment and local labour-market efficiency
Burgess S, 1993, ``A search model with job changing costs, `Eurosclerosis' and unemployment'' Oxford Economic Papers 44 75 ^ 88 Campbell M, Baldwin S, 1993, ``Recruitment difficulties and skill shortages: an analysis of labour market information in Yorkshire and Humberside'' Regional Studies 27 271 ^ 280 Campbell M, Duffy K, 1992 Local Labour Markets: Problems and Policies (Longman, Harlow, Essex) CEC, 1994 Growth, Competitiveness and Employment White Paper (Commission of the European Communities, Brussels) Clark G L, Whiteman J, 1983, ``Why poor people do not move: job search behavior and disequilibrium amongst local labour markets'' Environment and Planning A 15 85 ^ 104 Clark G L, Gertler M S, Whiteman J E M, 1986 Regional Dynamics: Studies in Adjustment Theory (Allen and Unwin, London) Collinson D L, 1988, ``Barriers to fair selection: a multi-sector study of recruitment practices'', Equal Opportunities Commission (HMSO, London) Collinson D L, Knights D, Collinson M, 1990 Managing to Discriminate (Routledge and Kegan Paul, London) DfEE, 1996a,``Labour market and skill trends, 1996 ^ 97'', SEN 266, Skills and Enterprise Network, Department for Education and Employment, Moorfoot, Sheffield DfEE, 1996b, ``Meeting the national skills challenge'', SEN 267, Skills and Enterprise Network, Department for Education and Employment, Moorfoot, Sheffield Dur R, 1999, ``Mismatch between unemployment and vacancies in the Dutch labour market'' Applied Economics 31 237 ^ 244 Emmerich M, 1997, ``Making a virtue of necessity? The role of intermediate labour markets'' Local Economy August, 98 ^ 103 Fine B, 1998 Labour Market Theory: A Constructive Reassessment (Routledge, London) Fombrun C, Shanley M, 1990, ``What's in a name? Reputation building and corporate strategy'' Academy of Management Journal 33 233 ^ 258 Gatewood R D, Gowan M A, Lautenshlager G J, 1993, ``Corporate image, recruitment image, and initial job choice decisions'' Academy of Management Journal 36 414 ^ 427 Gorter C, Nijkamp P, Pels E, 1997, ``Vacancy dynamics and labor market efficiency in the Dutch labor market'' Growth and Change 28 173 ^ 200 Green A E, 1997, ``Exclusion, unemployment and non-employment'' Regional Studies 35 505 ^ 520 Griffith G,Wood S, Knight J, 1991,``The Bristol labour market'', DP81, Department of Employment, Moorfoot, Sheffield Hakim C, 1993, ``Segregated and integrated occupationsöa new approach to analysing social change'' European Sociological Review 9 289 ^ 314 Holzer H J, 1988, ``Search method use by unemployed youth'' Journal of Labor Economics 6 1 ^ 20 Holzer H J, 1991, ``The spatial mismatch hypothesis: what has the evidence shown'' Urban Studies 28 105 ^ 122 Jones D R, 1992, ``Unemployment resistence and labour mobility in the UK: a spatial perspective'', in On the Mysteries of Unemployment Eds C H A Verhaar, L G Jansma (Kluwer Academic, Amsterdam) pp 166 ^ 178 Jones D R, Manning A, 1992, ``Long-term unemployment, hysteresis and the unemployment ^ vacancy relationship: a regional analysis'' Regional Studies 26 17 ^ 29 La Londe R J, 1995, ``The promise of public sponsored training programmes'' Journal of Economic Perspectives 9 149 ^ 168 Layard R, Nickell S, Jackman R, 1991 Unemployment, Macroeconomic Performance and the Labour Market (Oxford University Press, Oxford) Lindeboom M, Van Ours J, Renes G, 1994, ``Matching employers and workers: an empirical analysis on the effectiveness of search'' Economic Papers 46 45 ^ 67 LRDA, 1995 Lothian Business Index Lothian Region Development Authority, City Chambers, Edinburgh McGregor A, Ferguson Z, Fitzpatrick I, McConnachie M, Richmond K, 1997 Bridging the Jobs Gap: An Evaluation of the Wise Group and the Intermediate Labour Market (Joseph Rowntree Foundation, York) McQuaid R W, 1986,``Production functions and the disaggregation of labor inputs in manufacturing plants'' Journal of Regional Science 26 595 ^ 603 Manning A, 1995, ``Developments in labour market theory and their implications for macroeconomic policy'' Scottish Journal of Political Economy 42 250 ^ 266
J Adams, M Greig, R W McQuaid
Martin R, 1997, ``Regional unemployment disparities and their dynamics'' Regional Studies 31 237 ^ 252 Martin R, Sunley P, 1999, ``Unemployment flow regimes and regional unemployment disparities'' Environment and Planning A 31 523 ^ 550 OECD, 1994 The Jobs Study: Facts, Analysis, Strategies (OECD, Paris) OECD, 1996 The OECD Jobs Strategy: Enhancing the Effectiveness of Active Labour Market Policies (OECD, Paris) Ong P, Blumenberg E, 1998, ``Job access, commute and travel burden among welfare recipients'' Urban Studies 35 77 ^ 93 Osberg L, 1995, ``The missing linködata on the demand-side of labour markets'' WP 77, Department of Economics, Dalhousie University, New Brunswick, NJ Peck J, 1999, ``New labourers? Making a New Deal for the `workless class' '' Environment and Planning C: Government and Policy 17 345 ^ 372 Polacheck S, Yoon B, 1987, ``A two-tiered earnings frontier: estimation of employer and employee information in the labour market'' Review of Economics and Statistics 69 296 ^ 302 Preston V, McLafferty S, 1999, ``Spatial mismatch research in the 1990s: progress and potential'' Papers in Regional Science 78 387 ^ 402 Roper S, 1988, ``Recruitment methods and vacancy duration'' Scottish Journal of Political Economy 35 51 ^ 64 Russo G, Rietveld P, Nijkamp P, Gorter C, 1996, ``Spatial aspects of recruitment behaviour of firms: an empirical investigation'' Environment and Planning A 28 1077 ^ 1093 Seninger S, 1997, ``Jobless spells and re-employment wages'' Applied Economics 29 1169 ^ 1177 Shackleton L, 1995 The Skills Mirage (Employment Policy Institute, London) Stewart J M, McQuaid R W, Charlesworth A C, Adams J, 1998, ``A summary of the results of a survey of unemployed workers in the Auckland eastern suburbs TTWA'' New Zealand Labour Market Bulletin (Nos 1 and 2) 145 ^ 161 Thurow L C, 1975 Generating Inequality (Basic Books, New York) Turok I, Webster D, 1998, ``The New Deal: jeopardised by the geography of unemployment?'' Local Economy 12 309 ^ 329 Van den Berg G J, Gorter C, 1997, ``Job search and commuting time'' Journal of Business Economic Statistics 15 269 ^ 281 Van den Berg G J, Van Lowell A G C, Van Ours J C, 1999, ``Individual variation in exit rates from unemployment: a nonparametric multivariate analysis using aggregate data'', paper presented to European Association of Labour Economists 11th Annual Conference, Regensburg, Germany; copy obtainable from the authors Free University Amsterdam Van Ours J C, Ridder G, 1993, ``Vacancy durations: search or selection?'' Oxford Bulletin of Economics and Statistics 55 187 ^ 198 Watson G, 1994, ``The flexible workforce and patterns of working hours in the UK'' Employment Gazette 102 239 ^ 246
ß 2000 a Pion publication printed in Great Britain