In the US, there are two parallel systems of data collection. The Bureau of Labor Statistics (BLS) collects information on the state of the labor market and, in doing so, includes only minimal measures of the health status of the working age population. Even so, the principal BLS survey, the Current Population Survey, collects much of the detailed information on the state of the labor market in separate supplements. This makes it difficult to put together a comprehensive picture of how individuals experience labor market phenomena. The National Center for Health Statistics, in contrast, collects information on the health status of the population and, in doing so, includes only minimal measures of employment.

To overcome the bifurcation in U.S. statistical agencies, two waves of the California Work and Health Survey (CWHS), in 1999-2000 and 2022-2023, were administered to provide a comprehensive data source on a wide range of labor market phenomena as well as their impact on the health status of the working age population in terms of symptoms, chronic conditions, and the individual’s assessment of overall health. However, CWHS-2000 had a smaller sample size, 2417 for the years 1999-2000, and the respondents were solicited from a random sample of landlines, whereas the respondents for the CWHS-2023, 4014 in all, were solicited from a random sample of cell phones.

The different sampling methods may have effects on estimates about employment. When, in the CWHS-2000, a landline number was reached, the person answering the phone may not ultimately have been the individual interviewed since when two or more met the study criterion of being 18-64, there was a protocol to choose the one interviewed. In contrast, when in the CWHS-2023 a respondent was reached, because cell phones are individual devices, respondents were universally answering for themselves. In addition, by the time the CWHS-2000 was administered, many individuals, particularly younger ones, had segued to the exclusive use of cell phones. CWHS-2000 may underrepresent younger individuals of working ages.

The results of both surveys were weighted to reflect the known characteristics of the California population, the CWHS-2000 using the State of California Department of Finance annual population estimates [15] and the CWHS-2023 using the results of the California portion of the US Bureau of the Census American Community Survey [16].

The CWHS integrates measures of employment from the Bureau of Labor Statistics’ Current Population Survey (including the CPS Supplements) [17, 18] and of health from the National Health Interview Survey [19] and Behavioral Risk Factor Surveillance System of the Centers for Disease Control and Prevention [20]. The employment measures capture status in the week prior to interview and, among the employed, the extent of work over the past year, whether the work was done as a hired employee, independent contractor, or through a temporary agency, and whether a job was contingent, that is, not expected to last beyond the current year. Respondents were also queried about working conditions on their jobs. The health measures collected information on overall health status, activity limitation, body mass index (BMI), and health behaviors, including smoking history and status, and alcohol consumption. Copies of the study instruments from the CWHS-2000 and 20223 and the source of each employment and health item may be found in several references [10, 13, 21].

In the results presented below, we compare labor market measures included in the surveys in exactly the same wording across the two timeframes of 1999-2000 and 2022-2023. This comparison can show broad trends, but due to the difference in the sampling methods, small differences should not be interpreted as reflecting secular changes in the labor market. In addition, prior to launching the CWHS-2023, measures were developed to capture the episodic nature of many contemporary employment phenomena by asking details of second jobs and expanding the scope of inquiry from the week prior to interview customary in labor market surveys to the month. Because these new measures were not available in the CWHS-2000, with one exception in which we show the further erosion of traditional terms of employment when incorporating the larger set of parameters measured in the CWHS-2023, the comparison is limited to measures included in the CWHS-2000.

Owing to the aging of the working age population and the increased participation of older groups in the labor force, we show changes in health measures for all working age Californians and for several age strata.

Because of the different sampling methodologies underlying the two surveys, rather than integrate the responses and test for differences across the surveys in labor market and health measures, for each survey we estimated the mean and 95% confidence interval for each measure. When the magnitude of change in a measure from the CWHS-2023 is such that the mean and 95% confidence interval for that measure lies outside the confidence bounds for the same measure in the CWHS-2000, we can say with greater certainty that the change is meaningful, although it may or may not be statistically significant [22].

The supposition that time-trends in the nature of employment represent something more than the growth of individual elements in the way individuals are hired, in the extent of their work, and in its quality rests in part on the assumption that such changes are the result of a conscientious policy to change the relationship between individual workers and the organizations for whom they work, and that the effects of the policy can be measured in a way that summarizes the change [8, 23-25]. Although there has certainly been progress in developing synthetic measures of change since the 1999-2000 CWHS was administered, particularly notions about the shifting of risk to workers and the organization of work to implement that shift [5, 6], we ae limited here to the use of measures established as of that survey. Two were incorporated into the 1999-2000 CWHS and repeated in the 2022-2023 CWHS: a measure called “traditional employment” that encapsulates both being formally hired by the organization for which work is done and working full-time, full-year on a regular shift [13] and the “employment continuum” [26] which merges information on economic rewards from work with that about its psychological characteristics, specifically the interplay of discretion and demands on the job.

Table 1 shows the demographic composition of the working age population and of the portion employed at the time of the CWHS-2000 and CWHS-2023. The female percentage of the employed population increased slightly, while the average age of the working age and employed population increased by two years each and the median age by a year. The aging of the working age and employed populations becomes even clearer by observing the increases in the oldest age category analyzed, persons 55-64. Overall, the percentage of the working age population 55-64 increased from 11.8% to 21.1%, or by about 80 percent in relative terms, while the percentage employed more than doubled, from 8.9% to 18.7%. The CWHS-2023 implemented suggested changes in the way information about racial and ethnic identity are collected, with large numbers in the survey indicating that they were from multi-racial backgrounds. Even so, the percentage of Non-Hispanic whites and African Americans in the working age and employed populations of California declined substantially while the percentage of Hispanics and Asian Americans increased. Finally, the fraction of California’s employed population with a high school degree or less is at least as large in 2022-2023 as 1999-2000, while the percentage with a BA or at least some graduate school has grown only slightly despite the demands from the tech sectors for workers with high levels of education, from 37.1% in the earlier dyad of years to 39.6% in the latter dyad.

The percentage of the working age population currently employed, employed at any point in the year prior to interview, or ever employed was remarkably similar between the two time periods, differing by less than 2.7 percentage points on any of these parameters (Table 2). Despite the considerable changes in the composition of the working age and employed population of the State, as described above, and changes in the industrial mix across California, employment rates have not eroded. In fact, the percentage employed in the year prior to interview and ever employed grew between 1999-2000 and 2022-2023.

Labor market analysts consider not just the percentage employed, but whether individuals work full or part-time and over the full or part-year and whether any part-time work is by the individual’s choice (Table 3). The fraction of the employed population of California working part-time increased by 4.0 percentage points, with the increase concentrated among those working part-time, part-year. The fraction of part-time workers indicating that their part-time schedules were not by choice grew to 6.8% of all employed in the State, although the confidence intervals surrounding the estimates of the two years overlap. The substantial decrease in full-time employment was largely fueled by a substantial decrease in those working full-time, full-year. There has also been a 3.4% decrease in the percentage working greater than full-time, defined as more than forty hours a week.

The net effect of the increase in part-time work and the decrease in full-time and greater than full-time work is that the mean usual hours of work has declined by 2.5 and the weeks worked in the past year by 2.6, resulting in a decline in the mean hours of work over the entire year across all Californian workers of 215 hours (the median reduction was 160 hours, the equivalent of about four weeks a year of full-time employment).

Between 1999-2000 and 2022-2023, the percentage in episodic employment (working fewer than 40 weeks in the year prior to interview) and in contingent employment (not expected to last beyond the current year) both increased, although the confidence intervals surrounding these estimates in the two periods overlap. The percentage with two or more jobs changed only slightly to compensate for the growth in part-time work and declining hours of employment.

Job churning, as reflected in the percentage with job tenure of a year or less, did increase over the two dyads of years, from 24.5% in 1999-2000 to 29.8% in 2022-2023. Unfortunately, the churning of jobs and the churning of positions within jobs within the same firm did not result in an increase in the percentage stating they had a better job than in the prior year; instead, the percentage with a better job decreased from 37.8% to 31.2% between administrations of the CWHS.

Decreasing hours of employment, decreasing mobility in terms of attaining better jobs, and increased episodic and contingent work has resulted in an increase in the percentage of respondents whose households had incomes at or below 125% of the Federal Poverty Line, from 12.9% in 1999-2000 to 15.0% in 2022-2023, although the confidence intervals for these two estimates overlap. In the CWHS, respondents report on several measures of financial strain. The one that was used in both the 1999-2000 and 2022-2023 surveys asked about difficulty living on household income. There was a 11.9 percentage point increase in those reporting such difficulty between the two administrations of the CWHS, to 30.0 percent in 2022-2023.

The CWHS-2000 was fielded at a time when analysts began to emphasize change in how people were hired to work, from being hired as formal employees to the various kinds of alternative employment spanning independent contracting, working for a temporary agency, being an on-call worker, or having one’s work subcontracted from one employer to another. CWHS-2000 included all but the last two categories of employment, permitting comparison of several forms of employment. Analysts were also noting the erosion of working conditions, in part attributable to the changes in the terms of employment.

Table 4 tabulates both the terms of employment and working conditions across the two timeframes. Self-employment and working as an independent contractor both grew slightly, by 1.1 percentage points for the former and 2.6 percentage points for the latter. The other form of alternative employment for which data were collected in both survey waves, temporary agency workers, actually decreased by 0.6 percentage points.

Another trend in employment that has garnered much attention is the effect of irregular work schedules. Working a regular day shift is considerably less common among CWHS-2023 respondents, declining by 9.1 percentage points from the CWHS-2000 levels. In contrast, and almost tautologically related, rotating or variable shifts have increased in frequency (slight differences in the wording of the items may account for some of this growth).

Perhaps the most profound effect of the pandemic was to increase the percentage of Californians working from home at least some of the time, from just under a third of all workers in 1999-2000 to just under a half in 2022-2023. Concurrently, the percentage indicating that they supervised the work of others declined by 10.1 percentage points, probably reflecting the growth of self-employment, independent contracting, and other forms of alternative employment. Between the two surveys, there was almost no change in the percentage of the employed population indicating that they were covered by a union contract on the job. An increasing fraction of California’s employed population work in firms of a thousand workers or more, while a decreasing fraction work in firms with fewer than a hundred workers.

The CWHS surveys asked a series of questions about the nature of working conditions. There was a very small increase in the percentage reporting flexible working hours, the large growth in at-home work notwithstanding. There was also very little change in the mean scores on items tapping job demands and job control. However, the percentage reporting a high level of physical demands in at least one domain of work activity declined substantially over time.

Labor market analysts have attempted to integrate data across three axes to synthesize the changes in employment, including employment rates, quantity of employment among those who are working, and quality of employment in terms of the form of hiring and working conditions once hired. The 1999-2000 CWHS included a set of measures that combine work quantity with terms of employment [13]. The Traditional Employment measure incorporates working full-time for the full-year on a regular day shift, hired on a permanent basis and not working as a consultant or for a temporary employment agency (Table 5). Definition One, below, incorporates all of those aspects of employment. The percentage meeting all criteria declined from 35.5% in the first dyad of years to 31.7% in the second, or by 3.8 percentage points. Definition Two expands the criteria from the first definition to incorporate the proviso that the work not be done from home, as noted the signal change in employment between 1999-2000 and 2022-2023. When that proviso is included in the definition, the proportion meeting all criteria declined by almost half, from 25.7% in the first dyad of years to 13.6% in 2022-2023. The CWHS-2023 collected information on a wider array of alternative employment arrangements than the CWHS-2000. When all forms of alternative employment are added to the criteria for alternative employment, the percentage meeting the revised Definition Three is only 29.7% of all workers.

The traditional employment criteria are objective and as such do not provide an evaluative element to the assessment of work. Grzywacz and Dooley [26] developed their Employment Continuum which integrates economic sufficiency with psychological sufficiency to characterize jobs; optimal jobs provide good wages and good working conditions; inadequate jobs do not meet the minimal standards of economic sufficiency. Inadequate work grew slightly, by 3.1 percentage points to 24.1% of workers between the two dyads of years, while work which is barely adequate expanded by 6.4 percentage points to 21.5% (the confidence intervals for the estimates of inadequate work do not overlap). Concurrently, work which is economically good but not necessarily psychologically sufficient declined, slightly, by 2.6 percentage points and work which is optimal declined substantially from 15.7 to 8.7% of all workers (the confidence intervals for the latter measure do not overlap). Combining the objective measures in the traditional employment definitions with the evaluative criteria of the Employment Continuum, the combination of traditional and optimal work declined by 4.6 percentage points between the two dyads of years, but the decrease in the percentage whose work was not traditional and optimal was even larger, 8.0 percentage points, to only 6.1%. The percentage with traditional and inadequate employment declined slightly, by 0.4 percentage points; those in non-traditional work experienced a 4.1 percentage point increase in this kind of work. Workers with terms of employment meeting criteria for traditional jobs were more than twice as likely to report optimal working conditions (13.9 vs. 6.1% in 2022-2023); they were less than a sixth as likely to experience inadequate work in that dyad of years, 5.3% vs. 33.6%.

With the exception of the percentage of the working age population consuming some alcohol in the month prior to interview, all of the measures of health status changed to the extent that confidence intervals for estimates for each measure in the two timeframes do not overlap (Table 6). Among all working age Californians, the percentage reporting fair or poor health increased by 8.9 percentage points, to just under a quarter; among the employed, the percentage doubled. Supplementary Table 1 (http://links.lww.com/JOM/B936), which displays the health measures stratified by age, shows that the increases happened across the age spectrum. Similarly, the proportion of the working age population reporting limitation in activities, the principal disability measure in the CWHS, also increased, from 15.7 to 22.7 percent. Body mass index increased by two points between the two administrations of the CWHS; the increase in BMI occurred in every age stratum. The CWHS data indicate the extent to which the prevalence and quantity of smoking has decreased: current smoking was about half as common in 2022-2023 as in 1999-2000, while the percentage who have never smoked increased by 18.1 percentage points to just under three-quarters. Packyears, a measure of lifetime exposure to the hazards of smoking, decreased from 14.3 to 9.8. Although there was little change in the percentage of the working age population who had at least some alcohol in the month prior to interview, the frequency of binge drinking (having five or more drinks on an occasion) increased by 3.8 percentage points (confidence intervals for estimates of binge drinking did not overlap).

Many jurisdictions have made attempts to regulate the working conditions associated with non-traditional employment by such strategies as tightening criteria for what constitutes independent contracting, the most common form of alternative employment, bringing greater transparency or even mandating minimal standards for on-call work and shifts, and to bring such workers under the umbrella of such 20^th Century protections as minimum wages, access to workers’ compensation, and rights to organize jobs when the relationship between the firm and the worker is neither a traditional “hire” or, if one, impermanent [27]. California has been in the forefront of such efforts with the passage of AB5 which defined the criteria for being an independent contractor [28] and Proposition 22 which limited the reach of the legislation for ride sharing and delivery work [29]. Cities within the State, along with several outside of California are expanding the reach of minimum wage laws to incorporate such features of non-traditional employment as waiting time and mileage [30]. The observation from this study that not only is non-traditional employment expanding in California, but the fraction of such employment meeting criteria for optimal is declining suggests that the ferment in the legislative and regulatory realms is well-placed even though there has been a falloff, albeit smaller in relative terms, in optimal employment among those in traditional work, too. Thus, as a society we need to be vigilant in ensuring that the 20^th Century agenda of protecting the health and well-being of those hired in traditional ways continues to be actively pursued, even as we expand efforts to deal with the non-traditional forms of employment in which those protections are lacking, or at best minimal.

At a time of low unemployment and with the recent decline in immigration, the fact that several key measures of health status worsened may mean that the demand for workers may be difficult to meet in the years to come. This is especially true given that the increase in the proportion with only fair or poor health or reporting activity limitation was not limited to those in the traditional pre-retirement years, but extended throughout the age spectrum. Given that persons with disabilities have only about two-thirds the employment rate of those without, the increase in the activity limitation rate of 7 percentage points may itself account for a decrease of the population available for employment of several percentage points. Improving the accessibility of rehabilitation services and providing Americans with Disabilities Act-mandated accommodations may be necessary reduce the impact of worsening health on employment.

The research community continually and rightfully appraises the measures to be used to gauge changes in employment and health [11]. Similarly, biometricians have used the principles of Rasch analysis to improve the efficiency and validity of health status measures [31]. Few would argue that improved measures of race/ethnicity should not be chosen. However, doing so means that the declining fraction of the working age and employed populations who identify as non-Hispanic whites and Hispanics, respectively, is underestimated since some of those who might have classified themselves in these categories when forced to do so in 1999-2000, chose other classifications in 2022-2023.

Another issue concerns the extent to which results for California may be applicable to the rest of the nation. California has always been in the forefront of using alternative ways of hiring, from the use of a seasonal workers in its large agricultural sector to using project-based employment in the entertainment industry and independent contracting in its tech sectors. However, the strategy of shifting risks from firms to workers [5] and to use the fissuring of labor as an important method to do so [6] would seem to extend far beyond the State’s borders. Furthermore, as a major center of technology companies, the policies used by these firms perforce extend beyond California to incorporate all of these firms’ employees. Indeed, often California firms outsource certain functions to less expensive regions of the country and outside the country while retaining certain functions within the State [32].

With the measures available in both 1999-2000 and 2022-2023, we observed a substantial decline in full-time/full-year employment, at a population level the equivalent of an average of four weeks a year when translated into hours, a substantial increase in employment of short tenures, work that is less likely to represent an improvement within a job or a new better job and more likely to be subject to irregular shifts. All of these changes result in greater frequency of economic strain in the household, despite employment. Both the employed and not employed working age populations in California are more likely to report fair or poor health and activity limitation. The importance of serial measurement of both employment and health measures is that, to determine the extent to which poor working conditions result in poor health or whether persons with health issues are less likely to gain or maintain employment, it is necessary to array health and employment over time.