In the Nation's Interest
Who is the “Middle Class?”
by JOE MINARIK January 20, 2016
The state of America’s middle class – together with its first cousin, the inequality issue – is all the rage of late. Everybody has an opinion. However, many of these many opinions conflict, often sharply. Given the importance of this issue to our policy choices as well as our national mood, it might be worth some careful thought.
Here is the fundamental point: “Middle class” is not a scientific concept. It is highly subjective. Different people will have significantly different conceptions of what “middle class” means, and as a result, they will reach all manner of different conclusions about how the “middle class” is doing. In fact, one or another alternative definition of the “middle class” might preclude asking a particular question about it, or might make another particular question trivial or tautological.
Beyond the definition of the “middle class,” the definition of income (the usual criterion for middle-class status) is also important. Do you judge a person’s situation on the basis of spending power? (If so, you care about disposable income, after taxes and public transfers.) Or do you believe that membership in the middle class depends upon how much a person can earn in the marketplace, whether from labor or from the return on accumulated wealth? (Then, instead, you want to know about market income.) And what is your preferred unit of analysis? Are you interested only in traditional families? Or do you want to include “households” (like single people, or unrelated twenty-somethings sharing living quarters)? And how do you want to compare the incomes of smaller with larger families or households? Are you troubled by comparing working-aged householders with others who are older and retired? These are more than details, and they help to determine the conclusions you reach.
Who is Middle Class? Answer One: Those In and Around the Middle
If your conception of the middle class is those Americans at and around the middle of the current income scale, whatever that may be relative to past years, then the consequent definition is fairly simple: Take the middle one third, or the middle 50 or 60 percent.
So suppose that you choose the middle one third as your definition, and someone asks one of the hot questions these days: How many Americans are “middle class?” Then your answer is – 33 percent. (I knew I could fool you with that one. But now, for the bonus round: Who is buried in Grant’s tomb?)
So you might say that such a definition of the “middle class” is trivial and meaningless. But it ain’t necessarily so. You are lining yourself up to ask a perfectly reasonable question: How is the typical American doing? Times may have been good, or times may have been bad, but we all want to know whether the folks at the heart of the U.S. population are faring well. To learn that, we could do much worse than to take fixed percentile rankings on the income scale, and to see how the incomes at those rankings have changed.
Let’s try that. For multiple reasons – including data availability, and reliability and consistency of the source – we will consider the Congressional Budget Office (CBO) data on the distribution of before-tax income between 1979 and 2011 (all the years for which the CBO has analyzed the numbers). And let’s call the middle class those with incomes between the 20th and the 80th percentiles (so that the bottom 20 percent is the “lower class,” and the top 20 percent is the “upper class”).
The chart and table below show the annual levels and overall rates of growth of incomes of the “middle class” – those at the 20th, 40th, 60th, and 80th percentiles of the income scale – and for reference, those at the 90th, 95th, and 99th percentiles, in the “upper class.” This picture answers our question, but with some nuance. The incomes of the fixed middle 60 percent of the population have increased – but very slowly. The end markers of the “middle class” by this definition grew by 1.0 percent per year (at the 20th percentile) and 1.2 percent (at the 80th), with the middle markers growing by 0.9 percent (the 40th percentile) and 1.0 percent (the 60th). Meanwhile, the income at the 90th percentile grew by 1.5 percent per year, at the 95th by 1.7 percent, and the 99th by 2.1 percent. (CBO does not provide finer-grained data, but other sources indicate that the growth rates within that top 1 percent were progressively faster at ever higher levels of income.)
The 2008 financial crisis clearly hit these numbers. Stopping the calculation in 2007 would increase the middle-class-level growth rates by a little (0.1 percent to 0.2 percent per year) but the upper-class growth rates by a lot (0.6 percent per year for the 99th percentile).
Conclusion? Well, what is your standard? Middle-class incomes did grow since 1979, but that growth rate was quite modest. (Using a not-precisely-equivalent data set – from the Census Bureau – that reaches back further in time, the median (50th percentile) income grew by 2.8 percent per year from 1947 to 1972.) And middle-class incomes grew less rapidly than incomes at higher levels over the same 1979-2011 time period, meaning that the upper class – especially those within the very highest small fraction of the highest 1 percent – left the middle class behind.
Who is Middle Class? Answer Two: Those Achieving a Middle-Class Standard of Living
There is another potential view of what the “middle class” is. This alternative view is equally valid, and also useful – but for answering different questions.
Let’s assume that we have a strong conception of what we believe a “middle-class standard of living” to be. We can then follow the U.S. population over time, and see from year to year how many people attain that standard of living. So depending on economic progress, more or fewer people might meet that standard. Some people might do worse and fall out of the bottom of the middle class so defined. But others might do better and climb out into the upper class. So unlike the prior definition which held the size of the middle class precisely fixed, this definition would allow the middle class to grow or shrink. The questions we ask will have to change a bit, but with careful interpretation we still will seek to find out how typical Americans are doing in this economy.
So let’s ask the question, “How many Americans attain a middle-class standard of living?” But what is a “middle-class standard of living?” Whereas the first definition of the middle class was relatively straightforward, this one is much more complex, and far more highly subjective, and might even change over time. Everyone has his own idea, and I cannot say that yours is wrong; I can only disagree. However, I can say that some ideas that have been put forward recently are at least to me not useful, and yield conclusions that seem misleading. Here are two examples.
One recent interpretation, in a brief blog by Mark J. Perry of the American Enterprise Institute, defined the middle class as those with inflation-adjusted (2014 dollars) incomes of between either $35,000 or $50,000 (the lower bound) and $100,000 (the upper bound), and then counted how many households fell within, below, or above that interval since 1967. (The data source was the U.S. Census Bureau; the income concept was not specified, but given the source was most likely close to that used by CBO, that is, market incomes plus cash government transfers.)
The conclusion of Perry’s exercise is that middle-income households have migrated up, to the upper class; and that lower-income households also have migrated up, to the middle class. For example, using the $50,000 boundary between the lower and middle classes (see the next chart), the lower class shrank from 58.2 percent to 46.8 percent of the population over 1967 to 2014; and the middle class shrank from 33.7 percent to 28.5 percent of the population. The upper class correspondingly grew, from 8.1 percent to 24.7 percent of the population. This is portrayed as victory.
You might wonder how the results of this exercise could appear so satisfying, given the lackluster income growth among the middle-income population in the exercise I just presented above (over a similar though not identical time span). Good question. The reason why I do not believe that this finding is useful is that it is preordained by the choice of a definition of the “middle class.” Any real income growth, however slow, causes people near the borderlines to advance from a lower income group into a higher one. The only way that this exercise could have failed to find progress would have been for there to have been total income stagnation – no real income growth at all – for almost half of a century. That, of course, would be inconceivable.
Another issue with this definition is that it arguably becomes increasingly less meaningful as time passes, if one believes that identity in the “middle class” (or the “upper class”) relates at all to a household’s standing relative to the rest of society. For example, a $100,000 (2014 constant dollars) income in 1964 put that household at the 92nd percentile of income – perhaps lofty enough to be considered “upper class.” That same $100,000 income (in the same constant dollars) in 2014 put that household at about the 75th percentile. Would a person with such an income in both years still feel “upper class” in 2014, having been passed and surpassed by so many of his or her peers? Given that we are talking about “class,” a highly subjective concept – rather than the somewhat more quantitatively deterministic fixed standard of living – it might seem reasonable that over some period of time, probably more than two years but surely less than 50, the class borderlines should be adjusted in some way for changes in people’s perceptions of relative status.
So to me, Perry’s question is not worth asking. It is a one-way bet. It is as though I had told you that we would wager on a coin toss, and I would pay you if the outcome were either heads or tails. Yes, we could conceivably get an answer of “edge” or “sidewalk grate,” but neither is very likely.
The meaningful test that we could make of Perry’s calculations is not whether the middle class as he defines it will migrate to the upper class – we know it will – but rather how many will do so, and how fast. And what is the standard by which we judge the result?
An indirect illustration of the complexity of such a judgment is close at hand. The Pew Research Center produced a far more elaborate statistical exercise, entitled The American Middle Class Is Losing Ground: No longer the majority and falling behind financially. The Pew researchers defined the middle class as those households with incomes between 67 percent and 200 percent of the actual median income in any given year, with the analysis extending from 1970 through 2014, and based on Census Bureau data on pre-tax income. Income growth had been slow over pretty much that entire period. (The slower growth of typical incomes from 1979 on identified by CBO and reported above actually began with the recessions and the oil crisis in the early 1970s.) Therefore, the Pew report gives a downbeat interpretation of the status of the “middle class,” which in the report follows closely the tone of the report’s title.
Not so fast, responds a blog by James Pethokoukis, also of AEI. He points out the literal finding of the Pew report, similar to the Perry AEI blog but not much discussed in Pew’s public rollout: that over the 1970-2014 period both the Pew-defined middle class and the Pew-defined lower class have shrunk, with almost all of that net shifting going into the upper class. (See the chart and table below.) How can Pew be so maudlin, Pethokoukis asks, when so many of the former middle class have become upper class instead?
The answer, in my opinion, lies again in the definition of the middle class. Pew’s approach is more complex than Perry’s, in that it allows the income class boundaries to rise as incomes in general grow. However, this conception still may not keep pace with people’s standards and expectations. To wit: Card-carrying members of the middle class in 1970 had clear prior experience of incomes growing at a pace well above what they enjoyed thereafter. So still, though to a lesser extent than in the Perry exercise, many middle-class households who were graduated by Pew to the upper class likely had little personal perception of having reached such exalted status.
To provide a sense of the magnitudes involved: In 1970, the median income (for a three-person household) used to calibrate Pew’s exercise was $47,538 (in 2014 dollars). (For purposes of this calculation, I ignore Pew’s adjustments for changes in family size over time.) The upper class therefore had a lower income boundary – that is, the boundary between Pew’s middle and upper classes – of twice that amount, or $95,076. (So note that these class boundaries are not dreadfully far from one of Perry’s interpretations.) Then in 2014, the median income is $53,657. That is an annual growth rate since 1970 of less than 0.3 percent per year. Because the growth of the median income was so slow, the borderline between Pew’s middle and upper classes in 2014 (growing at the identical slow rate as the median) is only $107,314 – again, defined as twice the 2014 median income.
I will venture an unverifiable (the most-convenient kind) personal opinion: A significantly higher percentage of 1970 households with $95,076 (2014 constant dollars) incomes would self-identify as “upper class” than would 2014 households with $107,314 incomes (in the same constant dollars, and even though the real incomes were higher). My personal explanation: People’s expectations and standards rose far more than did incomes in the intervening 34 years.
And to illustrate just a bit further: Imagine that from 1970 to 2014, the median income had grown by 1.9 percent per year, rather than the actual 0.3 percent. Even 1.9 percent, you will recall, is well below the 2.8 percent annual growth rate of the Census Bureau’s median income from 1947 through 1972, and was probably approximately the expectation of all of those middle-income households at the beginning of the data analysis period in the Pew research. Well, growing at that 1.9 percent rate, the median income would have doubled by 2014. Therefore, because Pew defines the “upper class” as those with incomes greater than twice the median, what Pew actually defines as the borderline of the upper class would in this happier counterfactual world have been only the median – the precise middle income. (The chart and table below make this thought exercise a little more tangible.) To reach the upper class with just this actually sub-par post-War-style growth, a household would have needed an income of $214,628 – double Pew’s boundary in its actual report. And again, this is not a fairy-tale caricature. All it would take to turn Pew’s upper-class boundary into the dead-center middle is an alternative rate of income growth about one-third less than what people had enjoyed on average over the preceding quarter century.
So do you buy either the Pew characterization of a shrinking, financially stretched middle-class, or Pethokoukis’s alternative vision of a robust middle climbing in large numbers into “upper-class” ranks? Well, personally, I would answer “only in part” to the former, and “no” to the latter.
Looking at Pew’s findings, note that their measured “middle class” is shrinking only because people are found to have graduated into the upper class; and the middle class is strapped because its incomes have not grown very much. Ironically, if you believe that many of those new graduates to the upper class are not really upper class at all, but rather still middle class, then Pew’s “middle class” should be both larger and more affluent – precisely counter to Pew’s headline result. But to be fair: That is really a measurement curiosity, not a dispositive scientific finding. The fact remains that income growth around the middle level has been painfully slow since 1970 – much slower than it had been since the end of World War II – and so however you demarcate the boundary between the middle and upper classes, people anywhere near that income region likely would agree with the downbeat tone of Pew’s report, if not its precise quantitative findings.
With respect to Pethokoukis, again, I would tonally (but not quantitatively) agree instead with Pew. His notion of a substantial growth in the “upper class” over a period when the incomes of all but the most affluent grew at an historically glacial pace is hard to accept. (But to be fair, he got those numbers directly from Pew’s research.)
So to review the bidding: We can define “middle class” in many different ways. We can look at some fixed quantile of those who are in the “income middle” at any given time – but if we do, we cannot meaningfully ask whether the middle class is growing or shrinking. Alternatively, we can define some middle-range standard of living, and then ask how many Americans attain it or exceed it – but we had better think carefully about how that standard of living is measured, lest we wind up with conclusions that are tautological, or that lose touch with people’s realistic standards and expectations.
In short, doing research like this is really difficult. It seems to entail two stages. In stage one, you choose a rule for defining the “middle class,” and make it hard-and-fast so that the research process is scientific and unbiased. In stage two, you look at the implausible results, and consider that you reached them because your definition of the “middle class” was too hard-and-fast. If we are talking about the squishy concept of “class,” and we want to remain in touch with the population’s feelings about it, then science will have to yield at least to a degree to a sense of what those feelings are.
So What Does All This Mean?
So venturing into personal opinion, I cannot see any really good news about the state of the “middle class” (by any reasonable definition) over the last 35 years. It boils down to slow rates of income growth generally. Yes, the upper tail of the income distribution – in particular, the top fraction of the top 1 percent – has done extremely well. But that affects very few people, and if we think of “the greatest good for the greatest number,” we need to be concerned about how widely that growth is shared.
Now, clearly, different people draw different conclusions about the state of the middle class, and my take has no more standing than anyone else’s. It is easy and tempting to use this issue as a platform for politicking. Stating that the middle class is doing just fine is one way to fend off recommendations for substantial changes in policy. Decrying the state of the middle class is a way to justify big changes. But those big changes could go in different directions. Some people in that camp would blame the recent economic management and to campaign for their replacement. Some want to justify supply-side tax cuts, or supply-side spending (infrastructure spending being a prime example) or large income transfers. (The common element of both supply-side spending and supply-side tax cuts is that both would be financed out of assumed future revenue increases resulting from assumed future increased economic growth – with the accent on the “assumed.”)
Personally, I fear that these comparatively easy answers will leave us worse off. Rather, several nearly intractable forces are at work, and counteracting them will be far from easy. They include:
The immediate post-World War II period, upon which our expectations have been based, was extraordinarily felicitous for economic growth and is unlikely to be repeated. The post-War boom powered by wartime forced saving and the return of the troops (among other economic forces) ended with the recessions and the oil crisis of the early 1970s – and in some respect with the entry of the baby boom into the labor force, which intensified competition for jobs and bid down wages.
The financial crisis of the last decade further worsened the economic environment. Competition over weaker consumer demand intensified on all fronts, driving down prices and wages. The financial system was knocked off track, interfering with the financing of new businesses. Middle-class home values fell, deterring spending and therefore job creation and wage increases.
Globalization has implicitly created foreign competitors to the U.S. workforce. Through either off-shoring or the creation of new foreign firms, there inevitably will be more competition for U.S.-based employers and workers.
Technology has replaced labor in many fields, with the likelihood of increasing sophistication of that technology in the coming years. That technology, whether it is employed by U.S.-based or new foreign firms, again inevitably will compete with U.S. workers.
All of these factors (and more) are roadblocks for middle-class income growth, even though technology and globalization likely will increase total output.
The only real solution is faster broadly based economic growth. And there is no magic bullet for growth (in my opinion). Neither supply-side tax cuts nor supply-side spending increases will yield significant additional growth (in my opinion). And for those who have any doubts about the efficacy of such policies, the prospect of their failure when we already have a dangerously swollen federal debt burden is chilling.
In my personal view, we need to do more of the blocking and tackling of fundamental economic policymaking. That includes education and workforce development, technological research and infrastructure investment (in a long-term well planned, rather than a “crash,” program), and tax reform – but all paid for, not enacted under the assumption that they somehow will pay for themselves.
The “middle class” is apparently becoming a political football. Of course, in recent years evenly divisible by four, the candidates and the body politic have seemed ready to kick just about anything.
Dealing with such fast-and-loose policy advocacy looks like a job for CED, and for business statesmanship. Thanks to all of our CED members for your commitment to sound, thoughtful policy choices. The nation needs you now more than ever.
*Credit to Dustin Hoffman (http://www.imdb.com/title/tt0067980/).