My comments on CBPP’s Census Data press call today

September 13th, 2016 at 2:27 pm

Someone asked me to post them, so here they are:

First, just so this doesn’t get lost in the mix, let me say why I think these data are so important. After all, given that they’re for last year, they don’t exactly move markets. What’s the big deal?

The big deal is that these Census data give us the most granular look to date at the impact of the job market and public policy on the living standards of low- and middle-income families. You can get an update on the stock market every minute. We learn about the extent of child poverty once a year.

So, what does this granular look show us? Bob took you through some of the results we view as important, so let me add some brief observations on what’s behind these historically notable improvements.

Of course, there’s the tightening job market. In 2015, employers added almost 3 million jobs, on net. The unemployment rate fell almost a point overall, and almost two points for African-Americans. The real hourly pay of mid-wage workers went up 2% in 2015 compared to less than half that in 2014.

You know the old adage: when the economy sniffles, the least advantaged catch pneumonia. Well, that works in reverse, too. The benefits of closing in on full employment disproportionately flow to the least advantaged.

So we find in today’s data that while households throughout the pay scale saw real gains, the largest gains accrued to those households at the bottom of the income scale. I’m sure you’ve heard the topline income number: real median HH income up 5.2 percent, the largest one-year gain on record in this series, which starts in the mid-1960s, and the first real gain since 2007.

But real HH income went up 8 percent at the 10th percentile and 6 percent at the 20th percentile. Poverty rates for whites fell about one percentage point; the rate for blacks and Hispanics fell twice that much (though from much higher levels).

Meanwhile, income gains at the 90th and 95th percentiles were between 2 and 4 percent, below those of lower income households. This too is a familiar pattern of the benefits of full employment. It’s not that tight labor markets don’t help the wealthy. It’s that they tend to do well in good times and bad, while less well-off households depend on tight job markets to give them the bargaining clout they otherwise lack.

So a big part of the story today is that strong labor markets make a big difference in helping to connect low- and middle-income working families to the broader economy. That also points the way forward. I’m glad to see a great year in these data, but middle- and low-income HH’s need a lot more than one good year.

Another factor behind today’s very positive results is public policy. This is most evident of course with the health coverage results, as the fingerprints of health care reform are all over the sharp improvement in coverage we see both in today’s Census Bureau data as well as in other data sources.

In other words, in 2015, after six years of an economic expansion, both the economy and public policy were finally pulling for the middle class and the poor. Given the powerful forces of inequality pushing the other way, the results show the extent to which this one-two punch–full employment and progressive policies–can lift the living standards of working families.

While Bob and my review of the Census data has been largely positive I want to offer two caveats. First, one good year does not reverse decades of stagnation. This fact is particularly visible in the real median earnings trend of ft/fy men. In 1979, these guys earned about $52,000 per year in today’s dollars. In 2015, they earned about $51,000 (which includes their 1.5% real gain last year). The trend for women is much more favorable over time, which implies some partial closure of the gender pay gap, but their median pay has been largely flat since 2000 (which includes their solid 2.7% in 2015). So let’s be careful not to overlook the deep, long-term challenges still facing many middle-income workers.

Finally, inflation was very low last year, about 0.1%. Such uniquely low price growth helped to boost real incomes and lower poverty rates last year. This year, inflation has climbed a bit and is running slightly north of 1%. To be clear, based on continued improvement in the labor market, as well as the continuation of important, progressive policies, including the Affordable Care Act and minimum wage increases, I expect the positive trends you see today to be ongoing as we speak. But based on higher inflation, next year’s report may not be quite as stellar as this year’s version.*

*[I did not say this on the call, but there’s also likely to be a strong “regression to the mean” effect: outlier results are often followed by ones that are somewhat attenuated.]


Print Friendly, PDF & Email

7 comments in reply to "My comments on CBPP’s Census Data press call today"

  1. Smith says:

    Men’s income stagnant or dropping, while women’s income slowly works it’s way towards parity are related, I think. My theory is that downward nominal wage rigidity masks the effect of a women’s pay gap on men. When women are paid less, men may not get actual pay cuts. But they may not get raises either. The effect is less pronounced in higher pay scales, but college educated wages are stagnant since 2000. Wages actually fell for those with less formal education, partly due to glut of the college educated taking jobs not requiring degrees.
    Look for this anywhere, you won’t find it. The argument that men should want women to be paid the same never seems to notice the common sense notion that men won’t get raises when 1/2 the workforce are making less than their current wage.

  2. Smith says:

    When is a rise in income not a rise in income? When it’s a drop in unemployment.
    This, according to an article in today’s New York Times (front page, top right of hard copy)
    “Even in regions that are prospering, many workers have seen little wage growth in recent years. The rise in median income was driven mostly by increased employment rather than wage gains.”

    Previously I was guessing maybe the reported 5% plus increase was came from actual increases in middle wage earners outstripping other segments of income distribution. If so, then a small shift could move the median to explain the fact average wages went up less than 3%, and higher wage segments lagged. However, if the story above is to be believed, then not even that is so, because the reported rise relied on increased employment. It’s a welcome development, but it’s not at all what the headline of a 5% increase would have one assume. People might ask who are these people getting 5% raises? Now we know they don’t exist. That’s statistics. That’s why statistics get a bad rep. Not their fault. Blame the editors.

    What that says is anyone reading about a 5% pay increase and wondering how that can be when their own experience says the economy isn’t doing ok, is correct in disbelieving and dismissing the story.

    An accurate headline would be “Drop in Unemployment Moves Median Wage Metric” Of course, that really wouldn’t be news. Who cares about the side effect of increased employment causing the normally bellwether metric to jump? The real story then is actually that it didn’t mean wages rose at all.

    Again, we’re back to last year’s 2.5% to 3% real wage earnings, which occurred mostly due to falling energy (led by oil) causing .1% inflation. The boss (or HR) didn’t call in his workers and say the 2% fictitious raise you get every year that matches inflation is withdrawn because fracking caused a one time collapse on the spot market.

    Another effect I hadn’t considered is how the fracking boom boosted income in energy regions. I don’t know what percentage of national wage growth comes from this effect.

    Another consideration when attending to reality is a consideration of people’s experience who stay in their jobs, what’s their wage growth like?

    And to top it all off, the historic view is rather pathetic.

    People want this to change, and unfortunately the center left candidate isn’t selling change.

  3. Smith says:

    This means that more unemployed workers, more jobs, were added at levels above the median. How many?
    “Even in regions that are prospering, many workers have seen little wage growth in recent years. The rise in median income was driven mostly by increased employment rather than wage gains.”

    Assuming the 5%, half of that was due to no inflation, and then using the table found here, it could be no more than an extra .3% people who got jobs over the median. For example, if 2003 previously unemployed or new workers were hired, 1000 people got jobs paying less than median, 1003 got more, that and normal 2.5% raise which wasn’t negated by inflation gives you 5% hike in real median. The mix of whether the jobs come from unemployed vs. new workers can lower that difference too. That’s because the unemployed are nearly always making less than the median. If they are hired below the median, they don’t move the median, but hired above counts twice, you subtract from the lower half and add to the upper half.

  4. Smith says:

    How many? I asked without answering above. The answer is about 1 million (but importantly a net figure, above – below). But in 2015 the total number of jobs added was 2.45. So how could you get 1 million net increase of jobs over the median of about $50,000. Net means 2.5 million more jobs that paid above the median than the amount of jobs paid below the median the previous year. It’s easy, but tricky, meaning no complicated math, but the calculations are not totally obvious either. First, a bit how the 1 million figure is derived. The wiki income distribution table (url given below) shows nearest the median, every $5,000 dollar increase is about 4 million plus people. and $5,000 is 10% of $50,000. We need to derive the reported 5% increase, but only account for 2.5% or 1/4 of that number of people since near 0 inflation provided for 2.5% boost in salaries in 2015, hence 2.5% more people plus 2.5% raise = 5% bump in median. That assumes everyone got a normal 2 to 2.5% raise which unfortunately normally keeps wages stagnant. Unemployment fell in 2015 from 5.6 to 5% or 900,000 while 2.5 million jobs added (the other 1.6 million going to growing population of 1% per year as per 150 million workforce). 40% of new jobs previously added post recession were higher paying (see link below), so let’s assume higher means above the medium, and bump the .4 to .5 as 2015 reportedly added a better mix. .5 * 900,000 = 450,000 which does not a million make. But the 450,000 previously unemployed moved from bottom half to top half so the net is double (bottom – 450) – (top + 450) = 900,000 because median measures middle point of counts. 900,000/4,000,000 is x/10%, solve for x and you get 2.25% + 2.5% raise = 4.75% Where is the remaining .25% increase. A better mix of higher paying jobs in the 1.6 million new workers which is 1% of workforce can get you there, say that same .4 to .5 bump is 10% increase, so now 10% * 1% adds another .1, and well, we’re in the ballpark.

    I’d be interested to know if this is not the case and median income does not include income of unemployed. Note that unemployed who get jobs in the bottom half of median boost wages but do not change median one iota. If the unemployed were not counted, they would lower the median when hired.

    Here is where rough ratio of high paid to low paid jobs added is derived, fifth paragraph begins “But job losses and gains have been skewed.”

    Here are the distribution figures.

  5. Smith says:

    “Net means 2.5 million more jobs that paid above the median ”
    should read
    “Net means 1 million more jobs that paid above the median”

  6. Smith says:

    uh, oh, the above is incorrect, you do not double, because unemployed added to top half move the median proportionately, whereas additions to workforce in the top half move the median by half their proportionate share.
    This leaves 1.25% unexplained in the above analysis. Maybe a better mix than 50 50 high pay to low pay? Or greater increase than 2.5% for near median wages?

  7. Smith says:

    It’s household income, that’s the key. It unravels the mystery of how you get to 5% (I think). Again, no one got a 5% raise, and less unemployment of individuals who enter the top half explains 1/4 of the 5%. 1/2 is explained by normal 2% raise, but with no inflation. The other 1/4 is explained by the fact we are looking at household income. That means additional wage earners previously unemployed push a household from bottom half into top half and proportionately increase median. Many moving parts to the economy. I apologize for so many posts, a bit longish, but I think this is really important to understand.
    Staring at the charts on the link below helped and reading over and over the figures until it dawned on me what was going on was useful.
    All three charts are great, including the last, and you can cursor over lines to get figures.
    Cumulative percent change in average annual household income, by income group, 2007–2015

    Also worth noting the real wage increase in women’s pay was 2.7 vs. 1.5 for men, but women receive 80% of men’s wages. For $50,000 median wage it cost an employer $750 for the 1.5% and $1,050 for 2.7%. If those same increases held every year, it would take 20 years more for women to catch up to men.
    From that perspective wage increases are always overestimated due to the pay gap.

    Do I think Yellen knows this stuff about where the 5% increase comes from? That it reflects no wage increase at all, but rather continued elevated and partly hidden unemployment, and one time zero inflation bonus, and still less than pre recession level, let alone return to trend? Who knows. They don’t call it a lost decade for nothing.