• Job lock: Job mobility

    Links to all posts in the series to which this post belongs are in the introductory post

    My last several posts described research relating health insurance to labor market participation. That’s one vector for job lock—health insurance incentivizing entering or staying in the labor force. But there’s another, and more commonly studied, vector for job lock—staying with a particular firm for the coverage it offers, stifling job mobility (aka, affecting job choice or job turnover).

    Gruber and Madrian surveyed 18 papers in this area, finding a mixed literature. Six studies found statistically significant results consistent with job lock. Six returned only statistically insignificant results. Results in six other studies were mixed or could not be evaluated.

    A principal challenge in the study of the effect of health insurance on job mobility is that it’s difficult to disentangle whether someone has declined to switch jobs because of health insurance or, instead, because of other job-related factors. Confounding arises because some job-related factors (e.g., other benefits) are correlated with the availability of employer-sponsored insurance. If an employee stays at a firm that offers ESI, is that because he prefers the other benefits of working at that firm? Or because of the ESI? How could you tell?

    The identification strategy pursued in almost all of the other analyses of job turnover has been to compare the probability of job departure or turnover of otherwise observationally equivalent employees who differ only in the value that they are likely to place on a current employer’s health insurance policy. Various measures of the value of health insurance have been used. These include: [1] Health insurance coverage from a source other than one’s current employer, most often through a spouse or some sort of continuation coverage such as COBRA; [2] Family size; [3] Health conditions; [4] Health status. [References to papers employing each of these measures omitted.]

    These approaches have their strengths, but no study is ideal; the authors discuss various limitations of work in this area (omitted here for brevity). However, one 1994 paper by the two authors is singled out as particularly strong because it “uses a completely exogenous source of non-own-employment based health insurance.” The study exploited variations in state laws that mandated continued access to employer-provided health insurance for the non-employed (state laws akin to COBRA). They found that continuation coverage increased turnover by 10% and interpret it as a lower bound because the high cost of continuation coverage policies make it unlikely that the state laws fully alleviated job lock.

    Gruber and Madrian summarize the disparate findings in this area by using them to bracket the likely size of job lock. Their own work based on continuation coverage policies provides a lower bound, while work based on spousal insurance coverage provides an upper bound.

    Our view is that the approach of using alternative sources of insurance is more credible. Both approaches suffer from potential endogeneity problems, but the health/expected expenditures approach has a host of additional difficulties that do not arise with the alternative insurance approach. Moreover, within this alternative insurance approach the research by Gruber and Madrian (1994) provides an estimate which is likely free of endogeneity bias, by using variation in state and federal continuation of coverage mandates. So a conservative approach to reading this literature would be to take the results of Gruber and Madrian (1994) identified from continuation of coverage laws as providing as lower bound 10% estimate of the magnitude of job-lock, and the results from the spousal insurance approach as providing an upper bound estimate of 25-35% (Madrian, 1994b; Buchmueller and Valletta, 1996). [Links added.]

    GAO (2011) reviewed studies from 2001-2011 and found most consistent with job lock including Adams (2004) and Gilleskie et al. (2002). The former found that among 25-55 year old, married men, ESI reduces job mobility by 22.5% for those without alternative coverage. The latter, that among 24-35 year old, married males, ESI reduces job mobility by 10-15%.

    Despite their potential methodological weaknesses, many (though a minority) of studies of the effect of health insurance on job mobility did not find consistent, statistically significant results indicating job lock. If one was motivated to argue against job lock, this is where one should look, though it requires willfully ignoring the majority of studies that do find a statistically significant job lock effect. Of course, it’s important to keep in mind that the effect of health insurance on job mobility is only one kind of job lock. It tells you nothing about its effect on labor force participation, covered earlier in the series.


    Comments closed
  • Job lock: Labor force participation (prime-aged workers)

    Links to all posts in the series to which this post belongs are in the introductory post

    Older adults aren’t the only ones considering health insurance options when making labor force participation decisions. Younger adults do so as well. And, it should not be terribly surprising that spouses’ access to coverage can affect those decisions too. A wife or husband may be less likely to work or work less if her or his spouse has secured coverage for the family. And an unhealthy worker who has a greater relative need for coverage than a healthy worker may be more likely to work if her retention of health benefits depends on it.

    Gruber and Madrian found seven studies of the labor force participation of “prime-aged workers who are not single mothers.”* All seven reported statistically significant evidence consistent with the notion that employer-sponsored insurance (ESI) affects labor force participation decisions among married couples and results consistent with job lock for men.

    Four studies—Buchmueller and Valleta (1999)Olson (1998)Schone and Vistnes (1997), and Wellington and Cobb-Clark (2000)—examined the labor force participation of married women. As Gruber and Madrian explain, they all found “strong evidence that the employment and hours decisions of married women do in fact depend on whether or not health insurance is available through a spouse’s employment.”  GAO (2011) reviewed studies from 2001-2011 and found many consistent with job lock. Kapinos (2009) and Murasko (2008), for example, both found that married women worked less if they had coverage through their spouses. And Royalty and Abraham (2006) found that workers with spousal coverage were less likely to work full-time.

    One might be concerned, however, that a married man may be more likely to work and obtain employer-sponsored insurance (ESI) if his spouse has a distaste for market work. In other words, causality could run the other way. Gruber and Madrian read the evidence to suggest that this is unlikely.

    First, Buchmueller and Valletta (1999) find that the effect of husbands’ health insurance on wives’ labor supply is strongest in larger families, which is consistent with the notion that it is the value of health insurance that is driving the results and not simply tastes for market work. Second, Buchmueller and Valletta (1999) find that wives employed in jobs without health insurance work longer, rather than shorter hours, if their husbands have health insurance. In addition, Olson (1998) shows that conditional on working at least 40 hours per week, wives have a very similar distribution of hours regardless of whether or not their husbands have health insurance. Finally, both Buchmueller and Valletta (1999) and Olson (1998) find that husband’s health insurance reduces the probability of full-time employment for their wives quite substantially, but has only small effects on the probability of part-time employment. These findings taken together provide support for a causal explanation for the effect of husbands health insurance on the labor force participation of their wives, rather than a story based on unobserved correlations with tastes for market work.

    Of course, the conclusion that married women are less likely to work if their spouse has ESI coverage doesn’t say much about job lock. Such women are not in any sense “locked” into work. More broadly, however, the studies lend support for the intuition that the presence of health coverage affects the labor market.

    Those labor-market effects can manifest in job lock for prime-aged men, especially for those who have spouses or dependents who rely on that coverage  Two studies have examined the effect of health insurance on the labor force participation of prime-aged men—Wellington and Cobb-Clark (2000) (mentioned above) and Gruber and Madrian (1997). They include the following statistically significant results:

    • Among 25-54 year old men, continuation coverage (i.e., COBRA) increases the probability of exiting and the time out of the labor force by 15%.
    • Among working-age, married women, spousal health insurance reduces labor force participation by 6-12 percentage points, increases part time work by 1.6-3 percentage points, decreases full time work by 7-13.8 percentage points, and reduces hours worked per week by 15-36%.
    • Among married couples with both partners 24-62 years old, spousal health insurance reduces labor force participation by 23% for women and between 4-10% for men. It reduces annual hours worked between 8-17% for women and 4% for white men.

    More recent work by has focused on the effect of health shocks on employment for workers with and without ESI coverage Bradley et al. (2007) examined breast cancer-diagnosed, married women in Detroit. Those with ESI were ten percentage points more likely to remain with their jobs six months after diagnosis than those with coverage from another source; after 18 months, they were 17 percentage points more likely to stay in their jobs. Tunceli et al. (2009) examined cancer survivors 2-6 years after diagnosis, compared to a non-cancer sample. Those with ESI had a higher employment rate after diagnosis, compared to those with another source of coverage or no coverage. Bradley, Neumark, and Barkowski (2013) found evidence that women with own-job ESI reduce their labor supply by 8 to 11% less after a diagnosis of breast cancer compared to women less dependent on own-job ESI for coverage. All these results are consistent with job lock.

    * Single mothers are covered separately, usually in the context of “welfare lock.”


    Comments closed
  • Will this paragraph make you want to read about diagnosis-related groups?

    Imagine a government initiative that was supported by Republicans and Democrats alike, saved billions of dollars, improved health care, and was adopted around the world. It happened in 1983, and it continues today. 1 October 2013 marked 30 years since Medicare began paying hospitals by diagnosis-related group (DRG), arguably the most influential innovation in the history of health care financing.

    -Kevin Quinn, Annals of Internal Medicine

    Whether you agree with the assertions or not, that’s a very good opening paragraph, far better than just about anything you’ll read in an academic journal. (Stylistically, the “1 October 2013″ leading off a sentence isn’t so good. Editorial convention, I gather.)


    Comments closed
  • Job lock: Labor force participation (retirement decisions)

    Links to all posts in the series to which this post belongs are in the introductory post

    Theory tells us that job lock can affect labor force participation, but not its extent. For that we need evidence.

    It’s natural to hypothesize that job lock might be more pronounced for older workers who might otherwise consider retirement. Medical issues and costs increase with age, making employer-sponsored insurance (ESI) more valuable to, say, a 60 year old than, say, a 30 year old, on average. Poorer health at an older age makes working more difficult and retirement more attractive. However, before the age of Medicare eligibility, ESI generally provides the best route to affordable coverage.*

    Even after the age of Medicare eligibility, the existence of more generous ESI could influence participation in the workforce. Consequently, workers who might prefer retirement might continue to work exclusively for ESI benefits. That’s a form of job lock. A corollary is that offers of employer-sponsored retiree health insurance (RHI) might reduce job lock. Older workers not yet eligible for Medicare but who place high value on health insurance might be more likely to retire early if they can do so and still be covered by RHI.

    Gruber and Madrian found 16 papers on the effect of health insurance on retirement. The studies considered one or several of the situations described above. Based on them, Gruber and Madrian concluded,

    Despite using a variety of estimation techniques and several different types of datasets, almost every examination of the topic has found an economically and statistically significant impact of health insurance on retirement.

    Indeed, of the 16 papers surveyed, 12 found a statistically significant result consistent with job lock, one found a statistically significant result inconsistent with job lock, and the three others included results that were mixed or that could not be evaluated. In summary, findings consistent with job lock include:

    • RHI delays retirement until age of eligibility for it and accelerates retirement thereafter.
    • RHI reduces the age of retirement by 3.9-18 months, depending on data source.
    • RHI increases the probability of retirement before age 65 by 4.3-15 percentage points, depending on data source.
    • RHI decreases the probability of working past age 62 by 5.3%.
    • Medicare increases the probability of retirement by 280%.
    • Each year of continuation coverage (i.e., COBRA) increases retirement hazard by 30 percent, increases probability of self-reported retirement by 5.4%, and increases probability of not being in the labor force by 2.8%.
    • EHI increases the probability of working past age 65 by 5.3%
    • Poor health increases probability of retirement by 88% and decreases full time work by 5.1-6.3%, depending on age.
    • Private health insurance, RHI, and Medicaid decrease full-time work by 12-25%, depending on age.
    • A 10% decrease in the cost of health insurance in retirement increases retirement hazard by 1.1-1.4% for men and 1.4-1.9% for women.
    • The reduction in retirement health insurance cost associated with RHI increases retirement hazard by 25% for men and 28% for women.

    (The 2011 survey of the literature by GAO found six more studies all consistent with these results.)

    That’s a lot of results, and some cover a wide range of effect sizes. A formal meta-analysis would be handy, but none has been done and one is beyond the scope of this series. It’s worth mentioning, however, that all studies have limitations. Gruber and Madrian discuss some potential pitfalls. For example,

    Many companies have pension plans that encourage retirement at ages before individuals are eligible for Medicare. These companies, however, are also more likely to offer retiree health insurance benefits. Thus, the pension-related incentives for early retirement are correlated with the health insurance incentives for early retirement. [...] If pension-related early retirement incentives are positively correlated with retiree health insurance provision, it is likely that [] reduced form estimates of the effect of retiree health insurance on retirement are too large. Similarly, the selection of individuals with strong preferences for leisure into jobs that offer retiree health insurance will also lead to an upward bias in the reduced form estimates of the effect of retiree health insurance on retirement.

    This critique applies to studies by Madrian (1994)Karoly and Rogowski (1994), Headen, Clark and Ghent (1997) [unpublished], Hurd and McGarry (1996) [also unpublished], and  Rogowki and Karoly (2000). Some, though not all, of the mid-range and higher estimates listed above are from these studies.

    Before concluding, it’s worth mentioning that delaying retirement due to ESI is, from one point of view, not different from delaying it due to wages. That is, ESI, like wages, is a form of compensation. People like working for compensation, and would otherwise—wait for it—not work. The difference in the case of insurance is that, in principle, people could demand additional wages instead of ESI and then buy individual-market coverage. Were that practical, one could divorce labor market participation decisions from health insurance. In practice, that’s very hard for some older workers given the per-ACA dysfunctions of the individual-market. It’s on this very margin that “job lock” is meaningful with respect to labor force participation. That is, we want to know to what extent people continue working because that’s the only practical way to obtain coverage, not for the compensation effects of ESI. It’s not immediately clear to me to what extent this distinction is made in the literature.

    So far, I have considered the effect of health insurance on retirement decisions. Another strand of the literature addresses the effect of health insurance on the pre-retirement labor supply of married couples, which I consider in the next post.

    * Here and throughout, we’re considering a pre-ACA world unless otherwise specified. In a future post, we will consider how the ACA changes the landscape.


    Comments closed
  • Citation analysis, in one “sobering fact”

    It is a sobering fact that some 90% of papers that have been published in academic journals are never cited. Indeed, as many as 50% of papers are never read by anyone other than their authors, referees and journal editors. We know this thanks to citation analysis, a branch of information science in which researchers study the way articles in a scholarly field are accessed and referenced by others.

    Sadly and ironically, this sobering fact from Lokman Meho is not associated with a citation. Not that I necessarily doubt its veracity, but I would love to be made aware of the body of work that supports it. Anybody know? (Comments open for one week for leads only. Email/Twitter fine too. I have also emailed the author.)


    Comments closed
  • Pirate-themed CT scanner

    At a New York children’s hospital and via Powerful Pictures:

    pirate CT


    Comments closed
  • The latest prostate cancer screening literature review

    Julia Hayes and Michael Barry published a new prostate cancer screening evidence review paper at JAMA. This is one of the (admittedly few) areas of clinical research I follow fairly closely. So, I feel some obligation to post about the paper. On the other hand, I’ve got some job lock posts and other work to attend to. So, I’ll make this brief.

    The literature review updates prior ones found here and here (possibly gated). Though it is primarily a review of studies of prostate cancer screening, it also briefly covers some recent treatment, adverse effects, and statistical (simulation) modeling literature.

    Here’s the abstract by Hayes and Barry:

    Importance: Prostate cancer screening with the prostate-specific antigen test remains controversial.

    Objective: To review evidence from randomized trials and related modeling studies examining the effect of PSA screening vs no screening on prostate cancer–specific mortality and to suggest an approach balancing potential benefits and harms.

    Evidence Acquisition: MEDLINE, EMBASE, and the Cochrane Register of Controlled Trials were searched from January 1, 2010, to April 3, 2013, for PSA screening trials to update a previous systematic review. Another search was performed in EMBASE and MEDLINE to identify modeling studies extending the results of the 2 large randomized trials identified. The American Heart Association Evidence-Based Scoring System was used to rate level of evidence.

    Results: Two trials—the Prostate, Lung, Colorectal and Ovarian (PLCO) screening trial and the European Randomized Study of Screening for Prostate Cancer (ERSPC)—dominate the evidence regarding PSA screening. The former trial demonstrated an increase in cancer incidence in the screening group (relative risk [RR], 1.12; 95% CI, 1.07-1.17) but no cancer-specific mortality benefit to PSA screening after 13-year follow-up (RR, 1.09; 95% CI, 0.87-1.36). The ERSPC demonstrated an increase in cancer incidence with screening (RR, 1.63; 95% CI, 1.57-1.69) and an improvement in the risk of prostate cancer–specific death after 11 years (RR, 0.79; 95% CI, 0.68-0.91). The ERSPC documented that 37 additional men needed to receive a diagnosis through screening for every 1 fewer prostate cancer death after 11 years of follow-up among men aged 55 to 69 years (level B evidence for prostate cancer mortality reduction). Harms associated with screening include false-positive results and complications of biopsy and treatment. Modeling studies suggest that this high ratio of additional men receiving diagnoses to prostate cancer deaths prevented will decrease during a longer follow-up (level B evidence).

    Conclusions and Relevance: Available evidence favors clinician discussion of the pros and cons of PSA screening with average-risk men aged 55 to 69 years. Only men who express a definite preference for screening should have PSA testing. Other strategies to mitigate the potential harms of screening include considering biennial screening, a higher PSA threshold for biopsy, and conservative therapy for men receiving a new diagnosis of prostate cancer.

    I’ve already blogged about many of the studies and related papers referenced above and in the literature review, as well as some that are not. You’ll find it all under the prostate cancer tag.


    Comments closed
  • Job lock: Theory

    Links to all posts in the series to which this post belongs are in the introductory post

    On what basis does one have a reason to believe job lock exists? As Gruber and Madrian wrote,

    The very notion that health insurance is responsible for imperfections in the functioning of the U.S. labor market is somewhat curious. After all, health insurance is a voluntarily provided form of employee compensation. There is little discussion of the distortions to the labor market from cash wages. Why is health insurance different?

    As Gruber and Madrian show, in a perfectly competitive labor market in which employers can select which workers to whom they offer health insurance, employer-sponsored health insurance (ESI) does not impose any distortions on job mobility. Any firm wishing to hire a given worker either must offer that worker insurance or offer wages increased by precisely the cost of coverage for that specific worker. Since, in this simple model, employers can freely choose to whom they offer coverage, if an individual wishes to change jobs, she simply asks her new employer for health insurance—if she wants it—or the equivalent in wages of the cost of her coverage—if she prefers that. Since the cost is the same, it makes no difference to the employer.

    Perhaps you can already identify some of the ways in which the real world deviates from this simplified model. Gruber and Madrian run through them. First, for all practical purposes employers are constrained in their ability to offer health insurance to some workers and not others. For one thing, according to Gruber and Madrian, the IRS will not grant favorable tax treatment of ESI unless “most workers are offered an equivalent benefits package.” (More about this and other legal constraints in a future post in the series.) For another, it would be prohibitively costly for firms to assess the precise cost of coverage for each worker.

    Second, the labor market is not perfectly competitive. Firms face different prices for labor in part because they face different prices of coverage. In particular, large firms can obtain coverage more cheaply than small firms for several reasons (fixed administrative costs, lower risk due to the pooling of greater numbers of individuals). A consequence is that workers cannot obtain the same benefits across jobs, and they will attempt to match their preferences with employers that can accommodate them, a precondition for job lock.

    Gruber and Madrian consider a situation in which a worker holding a job at firm A, which offers health insurance that the worker finds valuable. Assume that the worker would be more productively employed at firm B. However, insurance costs at firm B are much higher than at A, perhaps because it is a smaller business. As a result, firm B doesn’t offer coverage. Even if it would like to hire the worker, firm B can’t just offer coverage to him alone, for reasons discussed above. Unless firm B can offer a wage such that the worker can purchase insurance on his own and still be as well off as the combination of wages and insurance from firm A, the worker will not switch jobs.

    Gruber and Madrian then raise an interesting possibility. Firm A, knowing that the worker values insurance and won’t switch to firm B on that basis, could reduce the worker’s wages. However, there has been no study to suggest that such worker-specific wage adjustment according to preference for insurance occurs. Moreover, it would probably be administratively difficult to administer such fine-tuned compensation packages. If we assume that firms rarely engaged in such fine-tuning of wages, then the forgoing provides a theoretical reason to expect job lock, at least with regard to job mobility.

    But what about labor force participation?

    Although we have framed the preceding discussion around the specific issue of job mobility, the same model developed above can be applied to labor supply decisions as well. In this case, the choice for the worker is not between one job and another, but between employment and nonemployment, where the health insurance availability across these two states may possibly differ. For example, consider an older worker thinking about whether to retire. Even if his value of leisure is greater than his marginal product of labor, a less healthy older worker may be unwilling to retire from a job that offers health insurance if health insurance when not employed is either not available or prohibitively expensive. This is a form of “lock.”

    Gruber and Madrian go on to discuss theoretical considerations pertaining to “welfare lock” (that potential workers may not enter the labor force in order to maintain means-tested welfare benefits) and job lock within the context of a secondary worker (e.g., with a spouse who may be the source of health insurance coverage). These are not that different from what has been discussed above. So, for brevity, I will omit them.

    So, the theory seems fairly clear that job lock can occur. The question is, is there evidence in support of that theory? That’s the subject of subsequent posts in this series, the next of which will be on Monday.


    Comments closed
  • Stand Up! – March 19, 2014

    Aaron is a frequent guest on Stand Up! with Pete Dominick, which airs on Sirius/XM radio, channel 104 from 6-9AM Eastern. It immediately replays on the channel, so those on the West Coast can listen at the same times. But today, I joined Pete instead.

    My talk with Pete included discussions about exchange enrollment and its stability, the future of health GOP plans for health reform, and we took some calls.

    You can play the audio right here, after the jump…


    Read the rest of this entry »

    Comments closed
  • Job lock: Introduction

    Ask an economist about employer-sponsored health insurance (ESI) and it won’t be long until s/he tells you it distorts the labor market. To most health economists, “job lock,” the idea that workers work more or face constraints in job mobility due to provision of work-related health insurance, is a real and important phenomenon. It’s one reason why many advocate limiting or ending the exclusion of ESI from taxation, among other reforms.

    But why, you might ask, do some so firmly believe in job lock? What’s the conceptual or theoretical explanation? Where’s the evidence that it exists and is substantial? If it exists, what laws and regulations help keep it in place? Finally, how does the Affordable Care Act (ACA) begin to address it, if at all?

    We’ve touched on some of these questions and their answers on this blog over the years, but never addressed them in full. That’s about to end. In a series of posts and with some help from Nick Bagley and Daniel Liebman, I will tackle these questions. Daniel’s annotated literature review lists some sources that will inform the series. I will not exhaustively cover the papers he described, but will attempt to provide a representative overview. For this purpose, I will rely most heavily on the literature reviews contained in Gruber and Madrian (2002)Fairlie, Kapur, and Gates (2011)Bradley, Neumark, and Barkowski (2013), and GAO (2011). Nick’s forthcoming post will cover the legal landscape. He has also provided helpful feedback on early drafts of my posts.

    By way of overview, the nature of ESI can affect the labor market in two ways. It can affect labor force participation (LFP), by creating disincentives for retirement—particularly before the age of Medicare eligibility—and affect the decisions of secondary workers (e.g., the spouse of a family’s primary wage earner). It can also affect job choice and create disincentives for job mobility, including entrepreneurship.

    Gruber and Madrian wrote a job lock literature review in 2002, covering the literature on the subject that had developed through that year. Their review, divides the literature into studies of job choice and analyses of LFP for three types of adults: older adults (near retirees), prime-aged men/married women, and single mothers. The last group is largely considered in the context of Medicaid, and the studies are more properly viewed as analyses of “welfare lock” (nonparticipation in the labor force so as to maintain eligibility for means-tested benefits). Gruber and Madrian concluded that “there is a fairly consistent case to be made that health insurance matters quite a bit for decisions such as whether to retire or to change jobs.” A systematic literature review published in 2011 by the GAO concurs with Gruber and Madrian. It found that of the 31 studies reviewed, “29 presented evidence consistent with job-lock.”

    This introductory post will serve as an index to the series’ posts. All expected posts are listed below, and links will be added as they posts go live.

    1. Job lock: Introduction [this post]
    2. Job lock: Theory
    3. Job lock: Relevant laws and regulations
    4. Job lock: Labor force participation (retirement decisions)
    5. Job lock: Labor force participation (married couples)
    6. Job lock: Job mobility
    7. Job lock: Entrepreneurship lock
    8. Job lock: What the ACA does
    9. Job lock: Conclusion


    Comments closed