Intro. [Recording date: December 8, 2016.]
Russ Roberts: Before introducing today's guest, I want to encourage listeners to go to EconTalk.org. You'll find a link there to vote for your favorite episodes of last year, of 2016. And give us some additional feedback about your listening habits. Really appreciate that.
Russ Roberts: Now on to today's guest, author and economist Mark Warshawsky.... Our topic for today is a recent working paper he's written, "Earnings Inequality: The Implications of the Rapidly Rising Cost of Employer-Provided Health Insurance." Mark, welcome to EconTalk.
Mark Warshawsky: Thank you. I really appreciate the opportunity to chat about his.
Russ Roberts: Well, it's incredibly important. And--this is an area of general interest to me, and in particular the focus on inequality, but also the standard of living. Which, whenever we try to measure compensation, income, these kind of issues that you raised in your paper come up. And I thought you had some insights and some facts that were rather startling and illuminating. So I'm very excited to talk about it. I want to start with some basic economic logic. What's the difference between, say, compensation and take-home pay for a worker?
Mark Warshawsky: Well, there are many, many differences between the two. For most workers, they get a whole package of both pay and benefits when they work for an employer. They may get a pension or they may get a contribution to a 401k plan. And what's particularly relevant here is they very frequently will get health insurance whose costs will be shared between the employer and the employee. But it's very important to remember that for most workers, most of the cost is paid for by the employer. And as we'll discuss, health insurance is very expensive. So this is a major cost for an employer and a major part of the compensation package. There are [?] other benefits that an employer will give to workers: certainly, vacation time and other types of leave, perhaps other types of benefits. And then, of course, added to the pay that a worker actually takes home. So, there are a lot of components to compensation, and part of the issue that is included here in this research and also in the politics of this matter is: take-home pay is pretty noticeable. And it's pretty easy to measure. It's in official records and tax records and earnings and so on. But the other parts of compensation, the benefits that workers get, are harder to measure and aren't as [?] measured, and so they are sometimes not noticed as much in terms of their cost, it's not noticed as much to the worker and it's not noticed as much to the researchers who look at these matters. But I can assure the employer knows very much what all this costs. And it's very much part of the employer's consideration in terms of how to design the compensation package and what they can afford to pay given the profits in their business--what they can afford to pay in terms of, you know, [?] earnings. The employer really pays attention to these costs very carefully.
Russ Roberts: So, there's a gap to start with between what the employer pays and what the employee receives--physically, literally there's a gap. Meaning, the employer has to pay for it; the employee takes home the cash part of it; and it's possible that the benefits that the employer has to pay for are worth the same amount to the employee, so the compensation that the employee receives is equal to the compensation that the employer pays. But it's also possible that the value of the benefits provided by the employer, which the employer has to pay for, may not be valued for all employees equally; and by some, not at all. So, if the employer provides, say, daycare on site for employees and I don't have any children, then that's a form of compensation--that's a cost to the employer--that I get no benefit from. Whereas, somebody who has multiple children might find that to be a wonderful value. So, first, talk about that and how the labor market works in dealing with these issues about compensation versus income for employees.
Mark Warshawsky: Yeah. So, there's a very complex mix of considerations to an employer in designing a benefits package. Sometimes it has to be done in a one-size-fits-all manner, and if they feel that some employees would appreciate, let's say as you indicate, you know, childcare, obviously it costs the employer quite a bit. But they give it to everybody; and they don't necessarily make an adjustment on a worker-by-worker basis to the rest of their pay as to whether they are using the child care or not. So, sometimes these things are done in true[?] approximations and averages. But there is another consideration--and this is very important to our topic--is that there also are a lot of government rules that pertain to these benefits. And in particular there's something called non-discrimination requirements in the tax code that particularly apply to retirement[?] benefits and to health benefits. And they state--they are very complex and so I'm not going to go through all the nooks and crannies of them. But in very broad terms they say that, regardless of whether the worker is high paid or low paid, everyone gets the same benefits. In retirement benefits this usually is a percent of pay; but for health benefits it means the same benefits, in dollar terms, so that the lower-paid worker gets the same health insurance benefits in terms of the cost to the employer as the high-paid worker. And that is a government requirement. Now, there are many other government requirements coming from different parts of federal and state law, and they have a big influence on how these packages are designed. You know, we could discuss whether they are good policy or bad policy, but at least for our purposes we have to recognize that they exist and that they have an important influence on how benefits are provided.
Russ Roberts: So, let's talk about the role of taxes. Most--almost all, maybe; you tell me--most benefits are not taxable in the same way that income is taxable. So, you could think about, if you gave me, let's say that you as the employer cover $5000 worth of health care insurance for me, one alternative would be you could raise my salary by $5000 and let me pick my own health care plans, not approved by the company, etc. And that's not as attractive to me in many cases because if you give me $5000 in income, I have to pay taxes on it. If you provide $5000 of health care coverage, or all kinds of other things that might be provided on the job, I tend generally not to pay taxes on that. Which seems like a good thing. But that has problems. So, talk about that just as a general aspect of benefits.
Mark Warshawsky: Yeah. You're 100% correct. Most, not necessarily all, but most benefits provided by employers are not taxed for the employee as income. The motivation for that--and this goes back, way, way back to the beginning of income tax--the motivation was that there [?] was a type of incentive or a type of encouragement to the employer to, and to the employee to demand these type of benefits. I think the notion is because it was thought that to have retirement benefits and to have health benefits is a good thing, and that employers and workers should be encouraged to have it. And so therefore they are not taxed. Again, in retirement benefits, the rules are very complex. It's not a totally open tax treatment, so there are limits. Some people may recall that, for example in their contributions to the 401k account, there are dollar limits that apply to restrict the amount of money that can be contributed tax-free to 401k accounts. Again, the important point here with our discussion of health insurance: there are no such limits for health insurance. So, even very, very generous plans--no deductibles, no copays, no employee contribution for the insurance premium. You know, and there are such plans out there. Totally tax free. So, that obviously is a contributing factor to, you know, having these plans and sometimes I think many people argue for them to just grow seemingly without limit.
Russ Roberts: So the last point, at least I think is the last point before we turn to health care in particular, that I think is so important and so unintuitive for most people is the idea that mandating benefits for workers is not a free lunch. I think a lot of people--certainly on Twitter, and pundits in the op-ed pages of our newspapers and publications, have a tendency to say, 'Well, if the government forces employers to provide Benefit X, then that's good for workers because that makes sure they have that benefit.' And what people neglect is that, that almost inevitably--I suppose we could cook up some scenarios where this is not true, but it's not easy to do--that almost inevitably means lower take-home pay: lower salaries in terms of cash for that worker. And it's imaginable, and often the case in my belief that workers would prefer the cash to the benefit. It's true that the benefit is tax free. But sometimes the benefit isn't a value. So, one example I always think of--and this is subtler, relative to the standard benefit discussion we've been having--one example would be safety regulation. So, we all--everybody likes the idea of a safe workplace. The question is: If you mandate certain expenditures by employers to make the workplace safe, that is a--may be--a way to help workers that is not as productive as paying them more money and letting them take their own steps to stay safe on the job. That is, it may be cheaper for workers to reduce the risk of injury than for employers. That's not always the case, of course. And employers have a natural incentive to make their workplace safe because they want to attract workers to a safe workplace. It's easier than to attract them to an unsafe one. But if you wanted a "unsafe one"--and it's not black and white, it's not 0-1, it's actually continuous--if you wanted a relatively less safe workplace, you might be able to attract workers there with higher pay who would then take their own steps to make sure that they are not hurt on the job. And that, that might be better for the workers--especially who are poor, who have low skills. And so when we mandate benefits for low-skilled workers in the name of helping them, we've taken away some freedom from them and some wellbeing, because they cannot use their money as they might choose. You can get into a debate about whether they can choose wisely. And I, of course, as listeners know, tend to be sympathetic to the idea that they can choose more wisely than the employer or the government. But that's the debate. The important thing is to recognize, I think, the unavoidable truth: That benefits aren't free. They come out of the salaries of workers. And that that is a cost that has to be taken into account.
Mark Warshawsky: Yeah. I mean, as you indicate, it really is Economics 101. There may be a very few economic circumstances where the employer, they are mandated to provide benefits can pass that cost along, you know, if they are a regulate monopoly or, you know, sell their products in, you know, in monopoly situations and they don't have much competition. But let's face it: When you are doing business in a global economy, there aren't many of those type of organizations around. So most employers, the vast majority of employers, you know, pay their workers; and it's pay--it's compensation. It's not take-home pay. They look at the cost of the employee as the cost of pay, as benefits as well as the mandated benefits that they have to provide. And if there's more mandated benefits, and--you know, and the going rate for workers is, you know, x, and you know, the profits that they make and you know, what economists call the marginal product, is you know, is y, and, you know, it's part of if x goes up because of mandated benefits, well that means that the take-home pay is going to have to go down. And that's--you can almost call it a law of economics. In other words, it's something that you just, you can't get around. You know, again, except in very special circumstances.
Russ Roberts: And it's not--I want to make the point--it's not just mandated benefits. It could be that--there are many, many benefits on the job that we don't think about. We tend, when we say 'on the job benefits' or compensation, we tend to think of these dramatic, large categories like health care, vacations, a company car, a cell phone, dental care, health care, retirement, etc. But of course employers provide all kinds of things on the job that are not obvious and we don't think about, like a warm place--there's heat. There's light. Sometimes there's a uniform. Sometimes there's training and explanations of things; and there's camaraderie and corporate culture. And all these things are part of the job. And if culture--or more importantly, competition--encourages employers to provide something for employees--and that thing gets more expensive, even if it's not mandated but it's just part of the competitive environment, employers are going to respond to that. And how we then observe--since we don't easily observe compensation but rather easily observe take-home pay--we may underestimate the wellbeing of workers or their relative situation to, say, 10, 20 years ago, simply because we are not really measuring the right thing. And although we're going to focus on inequality, I also want to think about how these kind of issues affect our perception of changes in the standard of living over time: not just the inequality issue. It is, of course, both, how they interact. You want to say anything about that before we go on?
Mark Warshawsky: No; I'm very sympathetic to what you are saying. There certainly have been some attempts over the years to try and measure these things, but they are hard. And that's why, you know, getting a little bit into my contribution here, when you get some unique data sources, that in fact does measure the cost of some of these benefits, you know, that gives us a great insight that we didn't have before.
Russ Roberts: So, let's start with just the logic of your argument, not the particular measurement--the ways you are going to measure it. Your claim, which is quite clever, and appears to be a matter mainly of just logic rather than bias or ideology--your claim is that rising health care costs affect our measurement of inequality if we only look at earnings rather than compensation. So, it's a little bit complicated, but it's just pretty much just basic algebra and logic. We're not going to do the algebra on the air, but we'll do the intuition and the logic. What's the argument you are making?
Mark Warshawsky: Yeah. It is, I think, just basically arithmetic. But, you know, maybe it's sort of--put this in context a little bit even before I go into the logic: The important thing to keep in mind is how much--and again, I don't think people realize how much these costs have gone up. So, here I'm going to cite a statistic from the Kaiser Foundation, which collects information on costs of health insurance. Just from 1999 to 2014--so, 15 years--the cost to an employer, and this is on the average, of health insurance, so this is what the employer is paying, not what the employee is paying--what the employer is paying for health insurance tripled. That's just in 15 years. And so--incomes didn't triple. The economy didn't triple. But the cost of health insurance tripled in just 15 years. And so when you have that sort of, I would say, I don't know what you call it, dislocation or disequilibrium or just sort of a massive change of an important part of the pay package--of course, it's going to have impacts on things which are more easily measured. Which are earnings. So, anyways, getting to the logical argument here: And we'll simplify it a little bit; it's of course more complicated when you look at the data and when you consider a lot of institutional and market considerations. But let's think of it in pretty simple terms. The cost of insurance, the health insurance, to an employer is the same whether the worker is high paid or low paid. And that's true, as I indicated, by law, because of nondiscrimination requirements. But it's also simply because that's simply the way, you know, pay packages way--that sort of, I'll even call it custom, or just for peace within the labor force: that everyone gets the same health insurance benefit regardless of whether they are high paid or low paid. Now, obviously there are high paid and low paid workers: people with more education or more value to the employer are going to be paid more as opposed to people who are, you know, either younger or don't have as much experience or simply don't provide as much value and profit to the employer are obviously going to be paid less. So, you know, that's just natural. And what everyone would expect. The employer looks at the whole cost. He doesn't look just at the take-home pay. He looks at: What does the insurance cost me and what does the take-home pay cost me? And they combine that. And then they say, 'Is the total package going to make sense for this worker and for that worker?' Now, when, you know, when health insurance is going up so much over a short period of time, that's going to change relationships that heretofore were occurring in terms of take-home pay. So that, let's say that at the initial point of our analysis, let's say, health insurance cost $4000 per worker. But for the worker who was earning $40,000, that represented 10% of pay. But for the worker who was earning $100,000, it was only 4% of pay. So, then let's say that cost doubles. So, it's $8000. So, for the worker earning $40,000, now it's going to be a lot higher percentage than for the worker that's earning $100,000. And even if everyone gets a compensation pay increase because, let's say, the company is doing better or the economy is doing better or time productivity increases, and so on, the employer still has to pay the money for the health insurance. And that represents more of a percent of compensation for the low-paid worker than for the high-paid worker. And so, the employer will increase take-home pay for everybody; but they are going to have to increase it less for the low-paid worker than for the high-paid worker. Because the cost of having the worker has gone up, you know, because of the cost of health insurance; and that represents more of the total compensation for the low-paid worker than for the high-paid worker. And you are going to see, you know, at the end of the day, that the take-home pay for the high-paid workers will have increased more than for the low-paid workers.
Russ Roberts: So, I want to take a step back and take a slightly different way of thinking about the numbers here and see if I'm doing it right.
Mark Warshawsky: Okay.
Russ Roberts: Part of the challenge of all this is that there are three million things happening at the same time: The economy is growing; there's increases and decreases in the demand and profitability of a product of a particular company; there's increased scarcity of certain types of labor. So to make it simple, let's say all that's flat and static. That's the kind of thing that drives non-economists crazy; but it helps you see what's actually going on. That's the reason we do this, as a mental exercise. So, let's say nothing has changed in the economy except health care is getting more expensive. And let's say that health care insurance doubles in the example you gave. So, I'm the low-paid worker: I was making $40,000 in take-home pay and I had this nice compensation for my health care insurance which was worth $4000. And you're the high-paid worker: You're making $100,000 in take-home pay, but your compensation is actually $104,000. So, the ratio of our compensation is actually 104/44. But we don't see that. We just see the 100/40.
Mark Warshawsky: Right.
Russ Roberts: Now let's say health insurance doubles: It goes to $8000 in the example you used. And I'm not any more productive, and neither are you. What's going to happen in the marketplace, approximately? And there's some wiggle room for this, but let's just take the extreme case. My take-home pay is going to go down. It's going to go down to $36,000 from $40,000, because my total compensation is going to stay at $44,000. I'm no more valuable to my employer than I was before, so their first impulse is going to--not their impulse--competition among employers is going to tend to drive the salary of people like me down to $36,000; and yours is going to go down to $96,000. And the ratio of 96/36 is not the same ratio as 100/40. And as a result--did I do that right?
Mark Warshawsky: You did it exactly right. And I'll just conclude your thought: Therefore, it will look like the lower-paid work is falling behind the higher-paid worker, proportionally. But that's not actually the case, because we know total compensation for both workers has not changed at all.
Russ Roberts: Right. So, my--the absolute gap between us is the same, but the ratio is what we tend to look at in these stories has changed. And now, let's move forward: let's have health care ratchet up steadily--right? Health care gets increasingly expensive. And if that's happening, the gap between you and me is just going to continually get bigger, if nothing else changes. Now, of course, something else is going to change: salaries are growing, in general, in an economy that's growing. But it's going to look like they are not growing very quickly because an increasingly large share of it is going toward health care. So there's two things going on that are deceptive when we only look at income rather than full compensation. The first is, it will look like--on average--forget the poor and the rich, forget the inequality issue. First thing that you're going to notice is that the growth in income is going to be relatively low compared to the growth in compensation. The second thing is that the inequality at any point in time is going to be growing because compensation--excuse me--benefits as a proportion of total compensation is increasing more rapidly than take-home pay is going to be. And that's always going to be true when, say, it's doubling or tripling, health care insurance costs, because productivity doesn't go up that fast. The other point to make is that I might--and this is just a reality; sometimes there's so much emotion around these issues I think it's really hard to think about them clearly--but the reality is, is that in that not-so-attractive scenario where my take-home pay falls by an additional $4000 to $36,000, I don't necessarily get $8000 worth of benefit just because health insurance is more costly. I do if I need it or really want it. If I really feel compelled to have health insurance, the fact that it's $8000 now and paid by the employer, that's worth it for me. It's true my take-home pay is down to $36,000 from $40,000, but I value it. And the more complicated case will be, in terms of measuring standard of living and inequality, is when I don't value it: if it's gone up in ways that are not productive for me but through either mandates or competition I'm forced to pay that for it. I mean, I think, if nothing else, listeners should realize that I'm paying for those increases in health care insurance, not the employer. Just because it's covered by the employer masks what's really going on underneath.
Mark Warshawsky: Yeah. That, leads into a related topic which I think is very relevant for politics and also the policy and the economics of the issue, is health care costs are exploding. They are rising very rapidly. And the question comes both at the individual level and at the company level and at the level of government and society, as to whether those benefits are worth it. In other words, whether they are on the margin providing us with a lot of increased health and increased longevity and so on. There's a strong debate about this, and there's some reason to think that they are not providing that marginal benefit, again, particularly when you consider the very rapid increase in cost. So, I think that becomes relevant when we think about this at a higher level. But you are certainly right: even at the individual level, people have different preferences for these type of benefits, compared to young, compared to old--all sorts of individual situations. And just general attitudes toward risk and attitudes towards even getting health care--some people are more interventionist in their situation and others don't necessarily go to the doctor at every sniffle. So, even at the individual level, not even considering broad societal considerations, people have different preferences for these things; and one size fits all is probably not the best solution.
Russ Roberts: So, just to clarify the two issues I think are easily confused--that I probably jumbled up in the last comment I made and while saying that we shouldn't confuse them. Let me make it clear. Philosophically, people may differ, and do differ, on whether government should mandate certain kinds of benefits. They can--I happen to be a person who is very opposed to that; but I understand the argument. It's not open and shut. You can be a fine human being and think that government can mandate health care, that government should make health care insurance deductible, etc., etc. For all kinds of reasons--political economy, philosophical, ideological reasons. At the same time, there's a separate issue that we're talking about here, which is: Once you do that, the measurement of wellbeing of workers and the inequality across workers is going to be much more difficult to observe because we don't regularly generate government reports on compensation in dollar terms. We instead generate reports on worker pay in dollar terms. We are missing a large part of compensation. And that affects our understanding of inequality. And it affects our understanding of wellbeing over time--changes in the standard of living, progress for low- and middle-income and high-income workers. And that has nothing to do with whether you think it's a good idea to have mandated health care or government benefits of this kind or another or the nondiscriminatory kinds of regulations that we face. We're simply talking here about the numerical part. And to me, the only thing controversial about that part of this is, "Who pays for it?" That is, the story I just told where salaries fell from $40,000 to $36,000--it is possible, under different models of the labor market, that the employer might have to eat some of those costs. They won't all be imposed on the employee. So, when we said earlier, when you said earlier, that between 1999 and 2014, health care costs to employers tripled, my first thought is: 'Well, that means that employees had to pay triple, because it comes out of their compensation.' Now, you can debate that. That is debatable. You can argue that it doesn't all come out of compensation. And economists--this has to do, we're not going to get into it, with the elasticity of supply and demand for labor of certain types. But that there is an effect on employees when costs go up to employers--you can't argue with that. You can argue about the size of it. You can argue how important it is. You can argue about whether employers have to bear some of those costs: they don't all fall on the employees; that markets aren't fully competitive, etc., etc.
Mark Warshawsky: At one time when these measurement systems and tax systems were designed, back in the 1940s and 1950s, you know, when these benefits were not very costly--health insurance was cheap. So maybe it didn't matter as much, because it was almost a rounding error. Or it wasn't that significant in terms of whether--perhaps we didn't tax it or didn't measure it in our income measurement systems. It didn't matter as much. But when you have these costs increasing as rapidly as they have, those systems of measurement become increasingly outmoded as ways of looking at ways of people's wellbeing and just in terms of what they are earning. What they are really earning. So, you know, that is also a reality that, when these costs have increased so much we just have fallen behind in our ability to both understand and measure these things.
Russ Roberts: Okay. So now let's come to the magnitudes. So, as you say, in the old days, a small proportion. So, this distinction between compensation and income was relatively unimportant. And then the other question is: Well, how important is it? And I think, when I argue with people about the importance of focusing on compensation rather than just income, one of the things you hear is, 'Well, okay, it's true, but low-income workers, middle-income workers, they don't really get many benefits. Most of the benefits are accruing to--it used to be the case that everybody would, say, get a pension. Now that's mostly gone. And only the higher-end workers are getting benefits.' So, let's, first just talk about the rough, crude magnitudes for who are--the magnitude of benefits for workers across different income levels in the economy today. What do we know about that?
Mark Warshawsky: Yes, and I'm just going to--yeah. I'll cite some statistics from my paper. So, I'm just taking two years: 1992 and 2010. So, looking at the lower-paid workers, those who were in the 30th percentile--so these are people basically earning, they are full-time workers but they are basically earning minimum wage--their health insurance as a share of compensation--and this is data from the Bureau of Labor Statistics (BLA), so this is based on a government survey; it's a very well-established survey, very carefully-designed--and it's a survey of employers. So, it doesn't require the [?] of workers--
Russ Roberts: [?] Yeah.
Mark Warshawsky: as to what they think they are getting or maybe they are getting. It's directly to the employer that knows what these things cost. So, anyway, so in 1992, the, at the low-income worker, their health insurance as a share of compensation was about 7-and-a-half percent of compensation. For the very highest-paid worker, in the 99th percentile--we are talking about the 1% that people talk about--their health insurance as a share of compensation was only 4-and-a-half percent. So, because, again, it's sort of, to a first degree approximation, it costs the employer the same thing to provide health insurance, whether it's a very high-paid worker or a low-paid worker. In reality there are distinctions. And there are differences by geography, by type of employer, and so on. But again, these are actually statistics from the survey of employers. Low-paid workers had a higher share of health insurance paid by their employer as a part of compensation compared to the high-paid worker.
Russ Roberts: And repeat the ratio again? This is 1992, it was 7-point something--to 4-point something.
Mark Warshawsky: Yeah. So, basically 7.5% to 4.5%.
Russ Roberts: Okay. Fast forward.
Mark Warshawsky: So health insurance benefits were more costly as a share of compensation, total compensation, for the low-paid worker than for the higher-paid worker. And these are, you know, incontrovertible statistics, again, based on the Survey. So it's not a matter of ideology or some claim that well-paid workers don't get health insurance. They do--
Russ Roberts: Or your attempt to try to measure what the health care is worth to a low-paid worker or a high-paid worker. These are just what the government gathered from employers that they paid out of pocket.
Mark Warshawsky: Yeah. What it cost them to have those workers working for them. Exactly.
Russ Roberts: It's just important to add--we've said many times that it's hard to measure these benefits. That's only because the way the government tends to report them. Right? So, we have surveys of workers where we ask people what we earn. We have surveys of employers; we ask them what they pay. The things that typically get reported and talked about in the media are income. These compensation numbers you are quoting--they are black and white. They are checks written by employers there--it's not like, 'I wonder what it is, how important it is, I'm going to guess.' These are costs to employers that are measured in this survey. Right?
Mark Warshawsky: Yeah. And it is indeed based on a survey of employers. Most of the time that people talk about pay, they talk about their income. And a lot of times that comes from income tax returns. And a lot of scholars that look at income inequality basically focus on income tax data. But again, that's take-home pay. It does not include these non-taxed sources of compensation. They are essentially not on the tax return.
Russ Roberts: Okay. Carry on.
Mark Warshawsky: So, yeah.
Russ Roberts: So, 1992, 7.5, to 4-something.
Mark Warshawsky: So we can--
Russ Roberts: Go forward to 2000.
Mark Warshawsky: Sorry?
Russ Roberts: You were saying. I interrupted you. In 1992, the ratio of the 30th percentile worker to the 1st percentile worker was 7.5-to-4.something. Go forward--
Mark Warshawsky: 4.5. Exactly. So, now we move to 2010. Pay has gone up but benefits have, costs of health insurance have gone up even more. So, in 2010, looking at the low-paid worker, their cost of health insurance paid by employer as a share of compensation for the low-paid worker has increased from 7.5% to 11.6%. So, that's a--what--3 percentage point increase. 3.1 percentage point increase.
Russ Roberts: 4.1.
Mark Warshawsky: Uh. 4.1. Yeah.
Russ Roberts: It's over 50%.
Mark Warshawsky: 4.1. Thank you for that correction. So, 4.1 percentage point increase in the share of compensation that is required to pay for the employer to pay for health insurance for the low-paid worker. From the very highest paid worker, the 99th percentile, that share of compensation that goes to pay for health insurance did increase but it only increased to 5.7%. So, that was an increase from 4.5 to 5.7--I think I'll do the math right now: 1.2 percentage points. So, the percentage-point increase as a share of compensation was almost 4 times the increase for low-paid workers as for high-paid workers. And so, and for low paid workers, we said 4.1% [percentage points--Econlib Ed.]; for high-paid workers it was 1.2% [percentage points--Econlib Ed.]. And those are the facts--and I don't think people can argue either with the logic or with the data in terms of that the consequences of rapid increases in health insurance are felt more by the low-paid workers than by the high-paid workers.
Russ Roberts: Well, they could argue--they could say it's not a good study; they could argue it wasn't representative; they could argue it left out this or that. But what I find important about what you've done is that at least you have some measurement of this. Let people argue about it if they want. Here's a paragraph in the paper I'm going to read that I found so striking--a much shorter time period but it really makes clear how these effects are important. I'm going to read it selectively, so that, again, we're bombarding listeners with a lot of numbers and I hope you guys out there are doing okay; but it's so important. We're talking about 1999 to 2006. This is a 7-year period. And you are quoting, again, data from the Bureau of Labor Statistics (BLS). So, the employer cost of health care for a low-wage worker went from 6.2% to 12.2% of compensation over that time period. That's just a huge increase. While total compensation for the group rose by 41% over that 7-year period--this is not corrected for inflation; that's not the point--41%; wages grew by only 28%. So if you looked at wages you'd say, 'Well, over 7 years, up 28%. It's pretty good. It's not great but it's okay.' And you'd have to correct for inflation to see how much better off they actually were. But their compensation went up 41%. That's a huge difference, 41 versus 28, trying to assess their wellbeing over that period. To go up to the high end, for a high-end worker, the top 1% over this period--smaller window than we just talked about--health care costs grew from 4% to 4.3%--a tiny increase in percentage terms. The more important point is: Earnings grew 35%; compensation grew 36%, for high end workers, because it's such a small portion--there's not that big a difference between compensation and earnings. However, if you just looked at earnings, the low-wage workers, their earnings only went up 28% compared to 35% for the high-end workers. So you'd say, 'Well, inequality is growing.' But if you looked at compensation, compensation went up 41% for low-wage workers and only 36% for high-end. So, actually, the true inequality is getting smaller. And that's really the point of your paper. A lot of numbers there; but again, the basic intuition is pretty clear, that if we looked at compensation rather than earnings we'd see a different picture. We are actually seeing the lowest wage workers in the data, the 30th percentile full-time workers, actually doing better: their standard of living is growing faster than people at the high end. But if you only look at earnings, you are misled. What about over the longer time period that you look at, 1992-2010?
Mark Warshawsky: Well, when you look over a longer period of time, of course there are many other factors that are coming into play. Although the advantage of looking at 1992-2010 is, they are similar points in the business cycles. So, they are both at the, just immediately following the low point in the economic cycle. But still, there is a lot--while over a longer period of time, the results aren't quite as striking, they are still there. So that about half the increase in inequality is explained by the increase in health care costs. But it's not a 1-for-1. So, to be more specific, looking at this 1992-2010 period, in the 30th percentile, earnings growth was 60%; but if you look at compensation which includes the growth of health care costs, a very rapid growth of health care costs, the growth is 70%. Whereas if you look at the 99th percentile, earnings growth was 78% and compensation growth for the 99th percentile was 82%. So you do have both earnings growth and compensation growth higher for the upper earners than for the lower earners. But--and this is the point of the analysis and the main finding here--is that compensation growth grows more for the lower-paid workers than looking at earnings growth alone. And it's relatively more for the lower-paid workers than for the higher-paid workers. So, we still have the result that a big part of the increase--not all--in earnings inequality is explained for by the increase in the, the rapid increase in the growth of health care. And I want to even move a little bit further in terms of thinking about it, in terms of what role the government or government policy would have. And that is: If you just think that there's something nefarious going on in terms of, I don't know, power relationships or just plain unfairness in the labor market between high-paid and low-paid workers, well, that, you know obviously indicates one set of policy prescriptions--increasing taxes on the high paid and so on. But if a large part of the explanation is the increase in the cost of health care, well, if that's the cause, then you presumably should address the cause and not the symptoms. And that clearly leads to an entirely different set of policy prescriptions and thoughts, in terms of how to deal with this issue.
Russ Roberts: And you are suggesting that, in fact, compensation inequality may be falling, while earnings inequality is rising. Is that a fair summary?
Mark Warshawsky: Yeah. Either it is falling or it has not increased, or it has not increased certainly as much as what just the earnings and income data show. It depends on the data source and the time periods. But the broader point, which is consistent across different time periods and different ways of looking at the data is that the increase in inequality is highly exaggerated when we look at just earnings or income as opposed to total compensation.
Russ Roberts: So, I want to raise one more issue related to this issue of who is going to pay for this and how it gets paid for. If my--going back to the example you used at the very beginning of the conversation--if I'm making $40,000 in salary and $4000 is in health care insurance that my employer is paying for me, and it doubles. Or let's make it really dramatic: it quadruples or quintuples and it's now $20,000 instead of $4000. So, it goes up to $60,000 is my compensation. If I just got a free ride in health care, my employer is not going to be able to afford that. So my employer is going to be tempted to take that whole $20,000 out of my income. And so, I've got $44,000 in compensation, and my employer says, 'Well, I'm sorry but my health care costs are really expensive. You're no more productive. I'm going to give you a salary of, now, $24,000 instead of $40,000.' I am going to respond by saying, 'I quit. I need to find a job that doesn't cover health care insurance. I'm going to go work for a firm that is homogeneous--only has low-wage workers.' So it isn't a discriminatory problem; and none of us are going to get health care. Or, if my employer already has low-wage workers mostly, they might just decide: 'Folks, you are not going to get health care any more. It's just too expensive. I can't afford it.' And they might even conceivably have to give me a raise in my income. So, react to that and how that might contaminate some of your results. Because it's kind of amazing that workers in the 30th percentile are still--11%, over 11% of their compensation is health care? You'd think a lot of them would want to work somewhere else, maybe, and have more income.
Mark Warshawsky: Yeah. It is an interesting and almost surprising result. But it is in the data. It's actually almost 12% of compensation as for the low-paid workers is the cost of health care. I would say there are a few things regarding [?]. One is, health care is--because it's so expensive, it is a risk to workers; and they do want insurance. They may not want to spend as much as it costs. But it's still nonetheless something that is of value to them, even if only as covering a risk, like they have car insurance or homeowner's insurance. So, there's number one. Number two is that it is still likely to be cheaper to be covered by the employer than it is to go out in the marketplace and buy insurance on your own, health insurance on your own, which could be expensive. You have to pass underwriting. Absent talking about the ACA (Affordable Care Act), which is a whole other subject. But here we're talking about historical data through 2014. So, the ACA was not a factor. So it still may be that it's better to be insured by your employer than to try to insure on your own. And it's also the case that you probably don't have an alternative in terms of Medicaid. You might try to fund your health care costs on your own, but you, you know, and maybe you can try to get charity care from the hospital. But that's not--all these are sort of second bests and not the most desirable outcomes. And so even though it's really a lot of money--it didn't happen overnight, also, we have to remember. This is a phenomenon that is long-standing and precisely because it didn't happen--one day it was $4000, the next day it was $20,000--where I think you would get your very strong reaction. But this has happened over time. So therefore it's more muted. The other consideration is, it is a tax advantage to get it through your employer as opposed to paying for it on your own. And that is a big advantage. And finally, I would say that it has happened, somewhat, you know, around the edges, and as you sort of note, particularly low-wage firms, some employers have indeed dropped health care coverage. It hasn't been dramatic, but there has been some of that as well. So, all of that taken in total I think explains the data. I mean, the bottom line as we still see it in the data that a very rapid increase in focus in percentage terms on the lower-paid for the cost of health care.
Russ Roberts: Now, one of the problems with the paper or the claim--I don't know how to say it--is that most of the people who have tried, maybe all of the people who have tried except for you, to measure these kind of effects have to bend themselves into pretzels to impute some kind of guess about health care costs to a certain class of workers or some measure of compensation. This data set that we're talking about that you've drawn these conclusions from--again, the conclusions, there's a theoretical set of claims here that I think is almost impossible to argue with; then there's an empirical question: How important is it? Your measure of importance comes from a data set that only you have, it appears. So, talk about where those data come from and why do you have access to it and other folks don't? It is the Bureau of Labor Statistics--that's good; it's not your uncle. But why aren't other people using these data to answer these questions?
Mark Warshawsky: Well, it [?] vindicated--the logic is so obvious that, obviously I'm not the only one to have thought of this. So, for example, the Congressional Budget Office (CBO) has looked at this question; some scholars at the Brookings Institution have looked at this question. And they come up with a similar conclusion but not as empirically strong as what I have. And the reason why, I believe, that they don't come with as strong, and careful, result is that they are using data sources which are, I guess you could characterize them as indirect. They have to do some imputations; they have to do some smoothing; they have to get data from different data sources which are not always consistent, consistently measured, sometimes they are measured with error because they rely on the recall of workers or households. So, those data sources are inferior to what I have, which is the same data source that the Bureau of Labor Statistics uses to report on very widely quoted statistics on what's going on with wages and what's going on with costs of hiring workers. These are very widely used and very widely respected statistics. I did have to ask for this data, in terms of, to get the detailed data on the numbers by compensation level. That is not something that the Bureau heretofore has reported. And I can't honestly say that I [?] others haven't asked for it. It wasn't a slam dunk, in terms of me getting it. It required a little bit of effort and a little bit of persistence on my part to convince them that this is a worthy study. But ultimately they did provide me with these statistics. And I have to be very grateful to them for doing so. But I can't fully explain why others haven't tried to get these statistics, which I think are, as I indicated, are very high quality and, you know, the same statistics are used for widely, widely recognized and widely cited. The unique thing here was to get some of the detail on these, on earnings and compensation by earnings level. That's what's unique here and what has not been heretofore published.
Russ Roberts: And can other scholars get access to those data now?
Mark Warshawsky: I assume so, if they ask.
Russ Roberts: Can you make those public?
Mark Warshawsky: Yeah--it wasn't sort of mine to give. So I'm a little reluctant to offer them on behalf of the Bureau of Labor Statistics, statistics that they--it required a negotiation to get. So, but, I think the question could be posted to the Bureau of Labor Statistics as to sharing it with others.
Russ Roberts: But the point is, is that: The BLS puts out a ton of data. I assume a lot of government agencies. And they put that out in various ways. So, just to take an example: When Piketty and Saez look at inequality over time using IRS (Internal Revenue Service) data, or they've done a similar kind of negotiation, I think--they get access to data from the IRS--they make public their data that they've created from the IRS. They don't make public the IRS data. They, in the course of--the very generously post all kinds of spreadsheets that an economist or interested person can play with themselves and check. And many people have. And that's great. But of course there's a lot of stuff going on under the surface. There's a lot of imputed and implicit assumptions that are made to get the certain numbers in those spreadsheets. And that part is much more opaque. What I'm wondering is, is that when you present the numbers in your paper, in what sense are those just a straightforward arithmetic, you know, ratio of things that the BLS [?] versus crunching and assumptions behind them that you had to do to get to the conclusions that you drew?
Mark Warshawsky: There's very little crunching and no imputations and no smoothing and no inferences. It is--the spreadsheets were provided to me entirely based on the raw data. So, there's very little of that. You know, I think the issue--certainly with IRS statistics, and I think also with the Bureau of Labor Statistics as well, is, these are surveys or actual data that come from actual companies and actual individuals; and we have to be mindful of their confidentiality. So, that's why there needs to be a negotiation. When you get to more refined levels of this data to do this analysis, you do have to be more concerned about the ability to identify, 'Oh! IBM (International Business Machines, Corp.) has these data.' You know, that's a commercial secret to IBM. And they share this information with the BLS on the promise that their data will not be shared, their specific data. And similarly with tax data. Obviously we have to be concerned about confidentiality. So, it's a balance, in terms of, you know, sharing all the data as opposed to being concerned about confidentiality. I think in my regard, the data, there was no--very little crunching. Almost no crunching. It's basically the raw data.
Russ Roberts: It's unfortunate, though, that we have to take your word for it, for now. I mean: I know you; you are a nice guy. You could have made a mistake, though, right? Even the nicest person makes spreadsheet errors. It would be useful if the BLS would make this more widely available so people could confirm your numbers and their reliability.
Mark Warshawsky: I think that's a balance which, you know, is a broader issue; and an argument certainly could be made for the BLS to do so. I think also, there's also a practical issue in terms of the BLS provides a lot of data. They have limited staff. And so I think there's that element as well. But I think mainly it's a question of, when you cut the data more finely--as you need to, in this exercise, to look at levels of compensation--you have a little bit more risk of revealing the confidential information. But that's a judgment and a balance.
Russ Roberts: Having said that, I just want to conclude, and then I'll give you the last word: It's always going to be a question of magnitudes. There's always a question of how important things are that you want to notice or observe. And inevitably there's imperfection in trying to measure those magnitudes. I spend a lot of time in this program talking about complex, sophisticated econometric analysis. That's not what this is. These are just some facts. Even facts can be tricky. You chose the 30th percentile on the grounds that that was basically the full-time worker cut-off. As we said before, it could be the survey is not that representative, for reasons that we don't understand. It could be it leaves out certain categories of workers that are important. There's always issues that are, even interpreting just basic "facts." But I think the fundamental thing that I learned from your paper and that I'd like listeners to consider is this idea that compensation and earnings are not the same; that our inevitable focus on earnings doesn't tell the whole story when we look at inequality.
Mark Warhsawsky: Well, I appreciate the opportunity to talk about this. And let me just conclude with a little bit, moving a little bit into the policy realm. As we talked about inequality, and also as we talk about the repeal and replacement of the Affordable Care Act, I think the contribution that I'm making is that the two are tied. That, particularly as health care has grown as a share of income, just as a share of our economy--now it's almost $1 out of every $5 dollars in our economy. When we talk about inequality, when we talk about the role of health insurance and the cost of health insurance, who should pay for it, you know, how those programs should be set up: The two are so intertwined and so importantly connected to one another that we really have to think about them in the same breath, almost. We talk about them in the same breath. And as we go forward in design or replacement for the Affordable Care Act, I think we really need to address the root cause of the problem--how the income inequality, which is the rising cost of health care cost--and I think we need to deal with that more directly than we have been. And we've been very indirect and very tentative about it. And I think we need to be more aggressive and more direct in addressing that problem.