• Returning to Brooks’ comments on employee dumping

    On Friday I quoted David Brooks and promised to return to the issue he raised. Here’s what he wrote:

    The Congressional Budget Office projects that 19 million people will move to the exchanges at a cost of $450 billion between 2014 and 2019. But according to the economists Douglas Holtz-Eakin and James C. Capretta, costs could soar to $1.4 trillion if those who would be better off in the exchanges actually moved to them.

    I’ve since read a number of related documents, including those that likely informed Brooks’ column. The one that provides the most detail is “Labor Markets and Health Care Reform: New Results” by Holtz-Eakin and Cameron Smith. In it they illustrate that the number of exchange-subsidy eligible workers for whom employers would spend less by dropping coverage and paying a compensating wage (plus the employer penalty) is three times the number of individuals the CBO calculated would receive exchange subsidies: 57 million instead of 19 million. This explains Brooks’ numbers: 3 x $450 billion is about $1.4 trillion (with rounding).

    This assumes all employers who would theoretically benefit from dropping coverage to lower-wage workers would do so. Certainly some employers will do just this. I concede that. But to expect all to do so must be an overestimate.

    Anyway, if one assumes all employers would drop coverage if it were beneficial to do so, Holtz-Eakin and Smith write that, “the gross price tag would be roughly $1.4 trillion.” The key word here is gross. Gross! Brooks didn’t use this qualifier, but it is important. Why not net? What’s missing to arrive at the final net cost? One thing I believe is not included are the income and payroll taxes levied on the higher wages those additional exchange-enrolled people will receive. Holtz-Eakin and Smith show how much additional pay would be required to make employees whole, writing,

    [T]he evidence suggests that if one portion of that package [worker compensation] is reduced or eliminated – health insurance – another aspect – wages – will ultimately be increased as a competitive necessity to retain and attract valuable labor. Thus, the key question is whether the employer can keep the employee “happy” – appropriately compensated and insured – and save money.

    I don’t know how much in income taxes would be collected, but it is something Hotlz-Eakin and Smith could compute. It would somewhat offset the $1.4 trillion gross price. (The CBO likely calculated the offset, so one could just triple it. I haven’t yet found the CBO’s figure on this. Have you?)

    Another question is whether all employers who theoretically might benefit from “dumping” will actually drop offers of coverage firm-wide. Holtz-Eakin and Smith don’t assume they all will cut coverage in this way (firm-wide). That’s because if coverage is not offered to lower-wage workers (for whom dumping is economically beneficial to employers), it can’t be offered to higher wage workers either. But there’s another proposed route: “[T]here may be incentives for firms to ‘out-source’ their low-wage workers to specialist firms (that do not offer coverage) and contract for their skills.”

    Of course, this is something firms can (and do) do today. Moreover, it’s not costless to outsource. There are transaction costs that would partially offset the benefit. Thus, it is not likely that all firms for which outsourcing would appear beneficial (and which don’t already out-source) would do so. Finally, if this were something all firms that could theoretically benefit would do, we should see a considerable crowd-out effect in Massachusetts. So far, there is no evidence of it.

    I’m not saying that the ACA won’t contribute to erosion of employer-sponsored coverage. It will. But it isn’t likely to be as an extreme reaction as some predict, nor will it cost as much as Brooks thinks. The $1.4 trillion price tag he suggested ignores some offsetting factors. It’s just not plausible.

    • The biggest flaw in their analysis is the assumptions made in the computation of how many people there are to be dumped by employers:

      “In round numbers, at present there are 123 million Americans under 250 percent of the FPL. Roughly 60 percent of Americans work (the employment-population ratio is 58.8 percent) and about 60 percent of those receive employer sponsored insurance. This suggests that there are about 43 million workers for whom it makes sense to drop insurance if the health plan costs the employer $11,941.”

      The costs they calculate start in 2014, so using current numbers is a little odd. There will be more Americans in 2014, which would drive that base upwards. On the other hand, if the economy recovers, there will be fewer under that income level. Which effect will dominate, I don’t know.

      On the other hand, it’s almost certain that the employment ratio of households below 250% of FPL is below that of the general population. If they were working, they’d be less likely to have such a low income. What that ratio is I couldn’t find numbers for, so I’m willing to accept the 58.8% for now.

      The biggest hole is in the employer-sponsored coverage rate. Lower-earnings households are far less likely than the total population to get health insurance through their employer. Using the 2009 ACS data ), the percentage of 18+ individuals living under 300% of the FPL who get insurance from an employer is about 36%. There’s no breakdown for 200-250% of FPL, so I’ll grant them the higher number, while also noting that it’d be lower if you only included the employees who had coverage, and not their spouses (or children up to 25).

      Applying the employment ratio they give and the employer-provided insurance rate I calculated to the 123 million, we get a little over 26 million. Multiplying by the mean 10-year cost per subsidee calculated by the CBO makes $620 billion.

      Footnote 3 on page 4 there says that “the gross cost would be partially offset by the reciept of the $2,000 penalties.” $2000 times 26 million recipients is over 52 billion dollars in offset. So the net added cost, under extremely generous assumptions and employers dumping everyone that they profitably can, is $567 billion.

      Oh, and the extra pay that would be negotiated would push some people into lower subsidy brackets, reducing the costs further.

    • Wouldnt the Massachusetts experience be relevant?


    • @steve

      I caution against using Massachusetts statistics as any indicator vis a vis likely outcomes with the PPACA for a variety of reasons (always highly insured, concentration of teaching hospitals, economic changes since initiation of Romneycare, etc. etc) but if they are an indicator you probably won’t like their direction:

      According to “Health Insurance Coverage in Massachusetts: Results from the 2008-2010 Massachusetts Health Insurance Surveys,” put out by the state in December 2010, the number of people insured by employers dropped 1.2% between 2008 and 2010 and the report said “Employer-sponsored insurance (ESI) remains, by far, the most common type of coverage among Massachusetts residents, covering about two-thirds of all residents in each year. However, between 2009 and 2010, the share of residents with ESI coverage fell while the shares with Medicare and public or other coverage increased.”

      More useful in my opinion because it deals with absolute numbers, according to the August 2010 Key Indicators quarterly report from the Massachusetts Department of Healthcare Finance and Policy, the number of people insured in “private groups” (which I think is a good if not exact strawman for employer sponsored insurance) dropped by almost 100,000 during the two years in question while the number of individuals purchasing insurance directly only increased about 50,000.

      So the numbers go in the wrong direction (although I would guess those trend lines had more to do with the economy than Romneycare one way or the other).