What makes a good job? Job quality and job satisfaction

Employee, employer, and government interest in job quality

Individual well-being depends on a number of life circumstances. One of the most important for most people is work. While unemployment sharply reduces well-being, having a job entails much more than simply not being unemployed. Jobs vary across multiple dimensions, and the associated well-being from working (referred to here as job quality) varies just as widely from job to job. Job quality is a key component of overall individual well-being and an integral part of the broader question of how well our societies are doing.

As well as being a central part of worker well-being, job quality is likely to be of interest to firms as well. First, job quality can increase worker productivity, as shown in both survey and experimental studies. Meta-analysis has revealed that job satisfaction and employee engagement are related to firm performance (which includes profitability, productivity, turnover, and absenteeism) [1]. In the efficiency wage theory of wage-setting, worker effort explicitly depends on the firm’s investment in job quality (which this theory equates with the wage level). As such, higher wages will produce higher firm profits, for at least some level of wages. It is easy to imagine a generalization of this set-up in which workers respond to higher job quality supplied by the firm with greater effort or less turnover, both of which are in the firm’s interest. Along these lines, research has shown that more satisfied workers are less likely to quit [2]. That workers have a certain amount of leeway in their behavior produces a potential congruence of interest between workers and the firm.

But workers’ and firms’ interests in job quality are not entirely aligned. While some people may believe that the best things in life are free, job quality comes at a price to firms: most of the job aspects that workers value are costly for firms to provide. There is a return to this investment, through higher worker productivity, lower turnover, and other factors. But even so, there is a point at which firms’ returns to greater job quality will be outweighed by the costs. Firms that maximize only worker job quality will go out of business.

Because cost restricts the aspects of job quality that firms can offer, workers have to choose between jobs with different aspects, giving rise to that common economics phenomenon, trade-offs. A worker who wants more of one dimension of job quality will likely have to make do with less of another. Under the compensating differentials argument, firms compensate workers for tougher work conditions through higher wages. But this general principle can be extended to other work aspects that matter for workers in addition to wages, such as hours, security, danger, and intrinsic interest. Workers’ willingness to trade off one job aspect for another depends on the relative value workers assign to the different aspects.

Job quality is also of interest to policymakers. It provides a barometer of how well people are doing in the workplace, which is a key element of quality of life. A side product of this measurement is the ability to identify workers in low-quality jobs. Labor market interventions to increase well-being often target workers who are doing less well, either for the cost–benefit reason that it is easier to increase their well-being or because we care about the distribution of well-being, not just the level.

Some ways of increasing job quality are costly to the government; for example, cutting taxes on low-income workers and subsidizing public transport or childcare. Others need not cost much at all, as in some cases of mismatch between the kind of jobs that firms offer and those that workers want. Such mismatches might come about because firms do not know exactly what their workers value or because distortions introduced by the tax code make some dimensions of job quality “too expensive” for firms to supply relative to others (compared to the value that workers put on them). In the case of a mismatch, policymakers could try to improve job quality through taxes or subsidies that change the relative prices of job quality domains, or through support to firms and workers on exchanging information about job domains (through works councils, for example).

It could also be the case that the negotiations between the firm and the workers that determine job quality are one-sided. Firms may have so much power that they can reduce job quality below the level that maximizes social welfare (which includes both worker and firm well-being). An analogous argument can be made if workers are too strong. Here public policy might intervene to restore the balance of power.

But before anything can be done to improve job quality, information is needed on what makes a good job.

Measuring job quality

It used to be fairly simple to answer the question, what makes a good job? Good jobs were those that were well-paid, with perhaps some attention to the length of the work week. There is a correlation between income and job satisfaction, but that correlation is far from perfect (see the Illustration). More recent work has emphasized that wages and hours are not sufficient statistics for assessing job quality. Many other aspects of work matter too. This increasing realism in the analysis of job quality has come at the price of greater complication of the analysis: while wages and hours of work are at least potentially observable, many of the newer job quality aspects (job insecurity and relations with co-workers, to take just two examples) are much less so.

The challenge is to find information on a potentially large number of dimensions of work and add them together to yield a composite figure for job quality. This composite figure is both a barometer of worker well-being and an indicator for firms and the government on where to improve job quality.

There are parallels between measuring job quality and measuring individual and societal well-being more broadly. In measuring societal well-being, the multidimensional well-being approach considers several life dimensions in which individuals may be doing well or less well. Each relevant life dimension must be separately measured. These well-being scores can then be weighted and summed into a single number (as in the UN’s Human Development Index) or left as a battery of individual “traffic light” indicators (as in the OECD’s Better Life Index).

An alternative is to subcontract the task of measuring well-being to the individuals themselves, as it were, by asking them to evaluate how well they are faring overall using a predetermined evaluation scale. One widely used question in this vein is on overall life satisfaction. For example, the British Household Panel Survey asks respondents to measure their overall life satisfaction on a scale of one (“not satisfied at all”) to seven (“completely satisfied”). Other response scales are also possible. This approach provides a single number reflecting respondents’ overall evaluation of their own life (reflecting their own implicit aggregation over all of the dimensions of their life that they find relevant).

An analogous process applies to job quality. We can measure various aspects of an individual’s job (wages, hours of work, insecurity, autonomy, relations with others, and so on) and add them up into a “good job index,” or we can simply ask individuals how satisfied they are with their job overall, producing a single job satisfaction score that parallels the life satisfaction score mentioned above. As has been noted, “the chief attraction of this subjective approach is that it avoids having to lay down in advance, or from above, what really matters in a job” ([3], p. 10).

If we have such information on job satisfaction, are we then home free? Not necessarily. There is a suspicion that subjective measures like job satisfaction, despite their increasing use, are less trustworthy than a measure of wages or hours worked. Equally, by nature, such measures are on a bounded scale. In the British Household Panel Survey example above, workers can never score job satisfaction higher than seven, no matter how good their job is.

There is, however, a strand of the literature that validates job satisfaction as a reliable measure of workers’ real sentiments about their job. As noted above, job satisfaction is correlated with firm performance. Another strand of work relies on panel data to confirm that the job satisfaction reported by individuals in a given year is a good predictor of their future decisions of whether to stay or leave that job and even whether to leave the labor market altogether (retirement). The boundedness problem may have been overstated too. In most surveys, individual responses are not concentrated in the top well-being category: in fact, both the top and bottom life satisfaction and job satisfaction categories are becomingly increasingly depopulated [4].

To evaluate the different aspects of a good job and, as argued later, to assess policies in this area, we need to know what it is that individuals value in their work.

What constitutes a good job?

A good job, like a good life, consists of multiple dimensions that an individual values. The relative importance of these dimensions is not necessarily the same for everyone or even for the same person at different times.

A fair amount of information on job dimensions is available from several general surveys, as well as bespoke ones such as the three Work Orientations waves of the International Social Survey Programme (ISSP) in 1989, 1997, and 2005 (a fourth wave is in the field in 2015). The ISSP Work Orientations modules contain a considerable amount of cross-country information on both job values (what workers think is important) and job outcomes (what they actually get) [5]. The job values and outcomes can be diced up into six broad dimensions [6]:

• pay;

• hours of work (including any mismatch between actual and desired hours);

• future prospects (self-reported promotion opportunities and job security);

• hard work (self-reported exhaustion, hard physical work, stress, and working in dangerous conditions);

• job content (self-reported interesting job, job helps other people, job is useful to society, and autonomy); and

• interpersonal relationships (with management and co-workers).

While these six dimensions are wide-ranging—and go far beyond many other taxonomies of good jobs—they may still not cover everything that matters to workers. One obvious omission is how well a worker’s skills match the job. Empirically, this is often picked up by measures of over- or under-education or over- or under-skilling. Research has emphasized the effect of such mismatches on wages, so the pay domain may partly reflect the skill match. A second omission is the quality of the person–organization fit in dimensions such as risk-aversion and job security [7] or sense of identification with the employer’s mission [8]. For example, intrinsically-motivated workers are more likely to work in the non-profit sector, and extroverted workers are more likely to choose jobs with opportunities for social interaction (the “personality fit” of the worker to the job). Well-being at work may also be affected by what others in the same firm earn. Knowing that other workers are paid more can reduce a worker’s well-being through an envy effect [9], but it can also boost job satisfaction if the higher pay is viewed as a signal of what the worker may aspire to earn. Other potential missing dimensions are trust in management and skill development.

Several international organizations have developed job-quality indices that cover most of the dimensions mentioned above (see Table 5.1 in [10] for a summary). Some add unemployment insurance and pension coverage, as measures of labor market income risk. Others suggest incorporating aggregate measures of labor-market health, such as job creation, inflation, and income inequality. Only individual-level job-quality measures are discussed here, however.

While we cannot be certain whether such taxonomies have captured all the relevant aspects of job quality, this may not even be the most important problem. Aggregating all the job dimensions in order to make an overall statement about job quality requires information on which job dimensions matter more than others.

Job values: Discovering which job aspects matter most

Not all aspects of a job are equally important. Some people are especially risk-averse, others have a higher marginal utility of income, and still others do not mind working long hours. Knowing which aspects of the job are more important is key for policymaking. If we want to make jobs better, what aspects should we concentrate on? Knowing individuals’ job values (the weights, or what matters most) enables the calculation of overall job quality as a weighted sum of the different job outcomes. As a by-product, information on individuals’ job values can also shed light on which workers end up with which jobs in the labor market (showing, for example, that the relatively risk-averse shun self-employment).

Information on job values can be gleaned in various ways. The simplest way, as mentioned, is to ask workers directly. A revealed-preference approach looks at which kinds of jobs workers leave or go to, although job attraction requires information on expected job outcomes, whereas job quits are a function of observed job outcomes in the current job. Other approaches examine the relationship between overall job satisfaction and the dimensions of job quality, such as income, hours, and job security, or between overall job satisfaction and the dimensions of job satisfaction (satisfaction with hours, satisfaction with job security, and so on).

Several surveys take the direct approach. The ISSP Work Orientations questionnaires ask workers how important eight different job aspects are to them, on a scale of one (“not at all important”) to five (“very important”): high income, flexible work hours, good opportunities for advancement, job security, interesting work, autonomy, helping others, and useful to society. Respondents’ rankings of job values do not change much over time and are broadly similar for men and women. Job security and having an interesting job receive by far the highest ranking (50–60% of respondents rank these as “very important”; see [5]). The third most important job aspect is autonomy, at 30%. The other aspects (including high income) languish at around 20%. Over time, the shares of respondents identifying high income, promotion opportunities, helping others, and useful to society as important have risen slightly, but these movements have typically been only a few percentage points over the three survey waves.

Job-values information is also available from waves 1 (1991), 9 (1999), and 14 (2004) of the British Household Panel Survey. For people who are currently working, the surveys ask: “Here are aspects of a job that people say are important. I’d like you to look at this card and say which is the most important to you about a job.” The job aspects covered are promotion prospects, total pay, good relations with the manager, job security, using initiative, nature of the work, hours worked, and something else. Unlike the question in the ISSP Work Orientations survey, this is a question about relative not absolute values: respondents choose the first most important job aspect out of the eight choices on the card.

In wave 14 of the British Household Panel Survey, for men, the top-ranked job aspects are job security, pay, and the nature of the work, with each selected as the most important by about 25% of respondents [11]. For women, the nature of the work is cited as the most important by a third of respondents, with job security and pay attracting 17% each. The only other aspect selected as most important by a substantial share of respondents is use of initiative, at about 10% for both sexes. These rankings have been reasonably stable over the three waves of the survey in which the question has appeared. There is evidence, however, that pay has been rising in perceived importance and that job security (especially for men) has been declining over the three waves. Since unemployment fell across the three survey years, one possible explanation is that the value accorded to pay is pro-cyclical, while that accorded to job security is countercyclical.

The second approach, revealed preference, does not seem to have been systematically investigated. It is commonplace to say that pay and hours help drive labor-market transitions, but their role is only rarely compared with the other dimensions of job quality listed above. The satisfaction scores in the different job dimensions in the British Household Panel Survey (satisfaction with hours, satisfaction with pay, and so on) can be compared for their predictive power regarding future quits, with the satisfaction dimension that best predicts quits argued to be the most important [2]. By this measure, job security is the most important, followed in a block by pay, use of initiative, the nature of the work, and hours of work; relations with others and promotion opportunities bring up the rear. This ranking differs somewhat by sex and age.

The third approach is to see how overall job satisfaction correlates to the various job outcomes in the ISSP Work Orientations questionnaires [12]. The largest correlations were found for relations at work and job content, followed by promotion opportunities, income, and job security. Hard work had the smallest estimated correlation. This ranking was similar for men and women.

Last, overall job satisfaction can be correlated to the various job satisfaction dimensions that appear in the British Household Panel Survey [13]. The nature of the work and hours of work have the largest correlations, and job security, promotion, and pay have the smallest (but still significant) correlations.

While these last two approaches are entirely valid, they pose a question for the measure of job quality: if the weights in job quality are derived from the correlations with overall job satisfaction, then why not just use the job satisfaction score itself as a direct measure of job quality?

That the job-value results are not all consistent with each other perhaps reflects the different data sets, type of question asked (whether the aspect is important or whether it is the most important), and the different methods used to infer workers’ values. Nonetheless, job security and what workers actually do during their workday (in terms of the interest and content of the job) often appear among the top-ranked aspects of work.

With this information on job outcomes and job values, we can calculate job quality in the economy. But we should almost certainly not do so by using the figures on average values and average job outcomes. Job values and outcomes both differ across individuals. In the ISSP Work Orientations data, individuals who say that promotion is important are more likely to have high promotion opportunities. The same holds for a number of other job outcomes [12]. The correct approach is to calculate job quality for each individual (from their values and outcomes) and then sum the individual job quality results to calculate the figure for the firm or the economy.

Cross-section and time-series evidence on job quality

The response to the question “who has the good jobs?” actually turns out to be similar for job quality and job satisfaction in ISSP Work Orientations data. Job quality can be calculated as a simple sum of dichotomous versions of the different job outcomes, with “1” representing a good outcome. Job satisfaction can be measured by the single job satisfaction question.

By these measures, women, older workers, and workers with higher incomes have good quality/satisfying jobs; union members have worse jobs according to both overall measures [12]. In addition, countries with higher estimated job quality have workers who report higher job satisfaction.

Time trends in all the components of job quality mentioned above are also illuminating. Analysis of ISSP Work Orientations data between 1989 and 2005 shows that both subjective income (“income is high”) and job security fell over this period, while promotion opportunities and reported hard work rose [5].

The different job-quality dimensions did not move in the same direction. As might be expected, any “traffic light” approach will show improvement in some dimensions but deterioration in others, so no overall statement is possible without information on how to weight the different job aspects. Some data sets (the ISSP, for one) contain information in this respect, but arguably not enough. We need to know how much workers value each job aspect, including hard work, job content, and relations with co-workers. But even if that information were available, some aspects that matter might have been overlooked. This is a general problem in the multidimensional measurement of job quality.

The measures of job outcomes in the British Household Panel Survey refer mainly to workers’ levels of satisfaction with different dimensions of work. Individuals report overall job satisfaction but also satisfaction with four job dimensions: total pay, job security, the work itself, and hours worked. All job satisfaction questions are answered on a scale of one to seven. From the early 1990s to the late 2000s, satisfaction with pay and job security broadly improved in the UK, while satisfaction with the work itself fell [11]. There is no particular time trend in satisfaction with hours of work. Thus, the same issue arises as with job outcomes. When reports of satisfaction with individual job dimensions move in different directions, how should they be summed up into an overall statement?

In the UK, overall job satisfaction has broadly trended downward, with some evidence of a plateau in the later years of the data. In terms of the job satisfaction dimensions then, either the decline in satisfaction with the work itself outweighed the rise in satisfaction with pay and job security, or there was another unmeasured important job dimension that deteriorated over this period. The phenomenon of falling overall job satisfaction over time is not particular to the UK. Similar movements can be seen in a number of other OECD countries [4].