Un- vs. Under- (K. Bryant in the house!)

June 7th, 2019 at 12:13 pm

[My CBPP/Full-employment-project colleague Kathleen Bryant has a new paper coming out next week with Dartmouth econ professor Danny Blanchflower (et al) on measures of labor market slack and their correlation with wage growth. A key focus of the paper involves the underemployment rate, and since we learned today that u6 hit a cyclical low last month, I asked Kathleen to dash off a note to get folks ready for the paper’s release.]

By Kathleen Bryant:

Today’s jobs report showed that the U6 rate – a measure of underemployment including the unemployed, involuntary part-time workers, and those who aren’t currently employed or job-hunting but still interested in working – hit its lowest level since December 2000, at 7.1%. While the unemployment rate is the slack variable that receives the most attention from economists and the popular press alike, Professor David Blanchflower (of Dartmouth University) and I argue in a forthcoming paper (out next week!) that monetary policymakers should pay more attention to the underemployment rate as they assess whether the macroeconomy still has “room to run.”

We analyzed Labor Force Survey (LFS) data in the U.K. from 2002-2017, and we regressed wage growth on unemployment and underemployment rates to test whether one measure of slack exerts more influence on wage gains than the other. We quantify underemployment in the U.K.  using an index that starts with the unemployment rate and factors in the aggregate number of reductions and additions in work hours preferred by workers.

In the wage equations, we found that the coefficients on underemployment were large and consistently significant while the coefficients on unemployment were small and insignificant, suggesting that underemployment has a much stronger relationship with wage growth. Monetary policymakers that monitor the relationship between unemployment and wages (known as the “Wage Phillips Curve”) should therefore rewrite the Phillips Curve into the underemployment space.

Furthermore, we emphasize the importance of focusing on underemployment using a somewhat unorthodox data source in the monetary policy literature – survey data on mental health found in the U.K.’s LFS. We find that the underemployed consistently score lower on measures of happiness, satisfaction, and life worthwhileness and higher on an anxiety measure than full-time workers or those working part-time voluntarily. We also find that the incidence of depression has increased more rapidly since 2011 among under-employed workers than among all workers on average. These findings illustrate the consequences of a labor market with too much slack at a personal level, beyond the traditional macroeconomic consequences that are commonly studied.

If you’re interested in more discussion of Phillips Curves, underemployment, and survey data on psychological wellbeing, we hope you’ll read our paper – “A Case for Full Employment: Underemployment, The Falling NAIRU and the Costs of Excess Slack,” which will be published through CBPP’s Full Employment Project next week.

[JB here again. Here’s a figure related to this stuff. I ran a very simple model to predict wage growth–yr/yr % change regressed on slack variable and two lags of DV–and ran the model through Jan 2010, forecasting thereafter. The forecasts aren’t that different, but towards the end of the series, the unemployment rate over-predicts wage growth relative to the under-employment rate, which is what Blanchflower/Bryant would predict.]


Headwinds in the job market? Payroll gains slow and wages fail to accelerate

June 7th, 2019 at 9:34 am

Payrolls were up 75,000 last month, less than half of what was expected, though the unemployment rate held steady at its 50-year low of 3.6 percent. Despite the low jobless rate, wage growth failed to accelerate, and has been stuck around 3 percent for a few months now. Downward revision shaved 75,000 jobs off of payrolls over the previous two months and our smoother, below, shows a mild deceleration in the pace job gains.

All told, it’s a weaker jobs report than we’ve become used to seeing, but what is it telling us? Is it a warning signal of weakening demand or is it more that the job market is closing in on full capacity? This sounds like a technical distinction–are the constraints on the demand or supply side?–but it’s a critical one. If we’re hitting supply constraints then the threat to the economy is overheating. Conversely, if we’re seeing the first hints of weakening demand, then the threat is a forthcoming slowdown. Note that the implications to the central bank are the opposite in these two scenarios.

It is clearly too soon to tell, and the labor market, and thus the American consumer, remain in very good shape. There is no near-term recessionary scenario in these data. In fact, the under-employment rate–U6–hit a cyclical low last month of 7.1 percent.

But headwinds have accelerated and the possibility of weakening demand is real and, to my thinking, poses a greater risk than overheating (after all, the latter would have to show up in inflation data). Both the U.S. and European central banks are thinking harder about rate cuts to offset perceived weaknesses (U.S. futures markets believe there’s 70 percent chance of a rate cut this summer). Obviously, the trade war and its potential escalation–opening the Mexican front in the war–are in the mix, as is the related weakness of U.S. business investment numbers. Germany, Europe’s alleged powerhouse economy, just posted big drops in industrial production and exports. As of this morning, second quarter U.S. GDP is tracking well south of 2 percent.

Turning to the guts of today’s report, and given the low signal/noise ratio in these monthly data, our smoother averages job gains over 3, 6, and 12 month periods. It shows a mild deceleration to 151,000 over the past three months. While that’s significantly below the near 200,000 average over the past 12 months, it’s still a strong enough number to hold down the jobless rate. In fact, we expect this number to drop as we close in on full employment.

The wage story is especially germane right now, especially given my long-held view that it takes persistent, high-pressure labor markets like this one to provide workers with the bargaining clout they otherwise lack. This improvement should show up in real wage gains, and, as I’ll show, that has clearly been the case. But while we might expect nominal wage growth to accelerate with unemployment at a 50-year low, the two figures below–hourly wage growth for all private sector workers and for middle-wage workers–show, at the end of each series, what could turn out to be a flattening in the pace of wage growth. Note the solid line–a 6-month moving average–which is either flattening or falling slightly.

Even so, the next figure shows that the gap between nominal wage growth for middle-wage workers and inflation, the difference being real wage growth. For this group–about 80 percent of the workforce–real pay is up about 1.5 percent, a solid clip and compelling evidence of the wage-side benefits of high-pressure labor markets.

Though the flattening of wage gains is worth watching, this is a bit of a Goldilocks scenario–not too hot, not too cold–as wages for middle and low-wage workers (not shown, but I’ve documented this elsewhere) are rising at a cyclically high clip, yet inflation remains tame. In fact, this is a microcosm of an important macroeconomic development in recent years: a lot less path-through from wage gains to price pressures.

Finally, there is an important weakening trend in recent jobs reports that belongs high on the watch-list: the flatlining of manufacturing jobs. Last month, factory jobs were up just 3,000 and so far this year, they’re up just 30,000 compared to 110,000 over the comparable period last year. I strongly suspect this falloff is related to Trump’s trade war as as such, it is a good example of how his actions are hurting a key constituency of American workers.


Trump and the Mexican tariffs: How far is this administration willing to go to achieve their protectionist, anti-humanitarian goals? Maybe farther than we thought.

May 31st, 2019 at 9:54 am

As you know if you’ve looked at any morning paper, the Trump administration has proposed an escalating tariff on all imports from Mexico, starting at 5 percent on June 10th and rising by five percentage points each month until it reaches 25 percent. The tariffs are intended to force Mexico to take actions to reduce the flow of migrants into the U.S. Trump said the tariffs will remain in place until Mexico “substantially stops the illegal inflow of aliens coming through its territory.”

Here’s a Q&A on this proposed action. Initially, it may not look like a big deal for us (much more so for Mexico). But if it doesn’t fizzle quickly, and I don’t think it will, it could turn out to be important along various dimensions.

Q: Isn’t this is an unusual use of tariffs?

A: It is. The majority of tariff cases stem from countries arguing about trade, as is the case with China. Country A objects to country B “dumping” a specific export (“rubber tires, grade c”) at below cost in order to corner market share and Country A imposes a “countervailing duty” to level the playing field. Or, as with China, we object to their trade practices (though I’ve argued this attack is somewhat overblown).

Yes, tariffs have been used as a geopolitical tactic, to protect what Hamilton called “infant industries,” and to support the buildup of domestic industries to achieve import substitution (tariffs were also the main source of government revenue in early America). But I’m not aware of a case where tariffs have been used to block immigration.

Q: Ok, it’s an unusual idea. But is it a bad idea?

A: Yes, for two broad reasons. First, I have the same objection to this tariff as to any other sweeping tariff (versus the more targeted “dumping” example above): by disrupting broad trade flows and indiscriminately raising costs on swaths of industries and consumers, it is a blunt policy tool that may have been useful in Hamilton’s day but is no longer so. Trump envisions widespread import substitution, but his vision is atavistic. Trade flows and inter-country commerce are too far advanced to be wholly rewired. I don’t think the globalization omelet can be unscrambled but even if it could, the victory would be a Pyrrhic one on all sides of the borders.

We’re especially integrated with Mexico. The WSJ reports that “about two-thirds of U.S.-Mexico trade is between factories owned by the same company.” Those are largely auto manufacturers, as we import $93 billion in cars and parts from Mexico (as a share of our imports, that’s 5x our China share), computers, food, and hundreds more goods. According to Goldman Sachs researchers, 44 percent of our air conditioners and 35 percent of our TVs are imported from Mexico. After China, Mexico was our largest source of imports last year (we imported $350 billion from them last year, and exported $265 billion).

Second, it is a well-documented fact that unauthorized immigration from the Mexico has declined in recent years. What’s gone up is asylum seekers from Central American countries torn by violence and gangs. In this regard, the “crisis” at the border is of the Trump administration’s own making. Suppose this tariff got Mexico to do more to shut its southern border to asylum seekers. On legal, humanitarian grounds, that should be no one’s definition of success.

Q: What about the economic costs?

A: The direct costs start out too small to matter to our economy, but indirect costs could be steeper. Initially, $17 billion (5% * $350bn) is less than 0.1% of U.S. GDP. Tariffs work like a sales tax on U.S. consumers, but few would notice this initial installment. It is, however, worth pausing for a second to consider the weirdness of this aspect of the proposal: U.S. consumers are paying an anti-asylum-seeker tax. But as the Trump administration is aware, the U.S. is much more insulated from trade than those with whom we wage trade wars. We import 15 percent of GDP and export 12 percent. Those shares are much larger for our trading partners.

This could, however, be a bigger direct problem for Mexico, as their economy is already flat; Mexican GDP growth was about zero in the first quarter of this year, and their exports to the U.S., their largest trading partner, account for about 37 percent of their economy. It’s true that U.S. consumers pay the direct costs of tariffs, but it’s also true that exporters targeted by tariffs will also feel some pain, especially when we’re talking about such deep, large (from Mexico’s perspective) supply chains that cannot be handily redirected.

But there’s also a plausible scenario where this Mexico tariff seriously dings our economy, through at least two related channels: financial markets and investment. Equity and bond markets are initially reacting predictably negatively to the proposal, with auto shares taking a beating. I don’t worry about day-to-day market swings, but worse financial conditions, if they persist, clearly bleed through to growth, and thus to jobs, wages, and especially investment, where investors have increasingly been complaining about the “uncertainty engendered by the escalating trade war.

Relative to many economists, I’ve downplayed the “uncertainty” card; economies, like life itself, are always uncertain. But while I haven’t done the analysis (I will), I think there’s a signal building from trade uncertainty to the weak numbers we’ve been posting on business investment. This is especially the case given factors pushing the other way, such as low borrowing rates, strong consumer demand (albeit with recent hiccups), and high corporate profitability.

I’m not predicting recession, of course. Economists cannot reliably do so and the unemployment rate remains at a 50-year low. But the trade war was already a headwind and if this Mexican tariff goes through, that wind will gain velocity.

Q: Wait a minute. Trump may be crazy, but surely, he doesn’t want to undermine the economy, especially with a reelection campaign in the offing.

A: You’d think so—I certainly have—but this is yet another thing many of us have gotten wrong about him. The Trump recipe, according to observers including myself, has been: create chaos, capture the media, propose a nothing-burger solution, claim victory. And do all this before the sh__ hits the fan, i.e., before there’s real economic damage. And, in fact, there’s some evidence that Trump largely plays with house money, meaning he creates economic chaos when the economy and markets are strong enough to shake it off.

But lately, he and his team seem more committed to sticking with their interventions, even when there are clear costs, as in the market and investment data. True, labor markets, job growth, real wages—those most fundamental indicators—remain solid. If that were to change, perhaps we’d see the same outcome as when the air-traffic controllers said, “it’s over,” re the government shutdown.

But those of us whose theory of the case is that when it comes to damaging the U.S. economy, Trump will only go so far, may need to update our priors. If this Mexican tariff goes into place and then escalates, we may be looking at an administration that is willing to sustain a lot more damage to achieve their wrongheaded, protectionist, anti-humanitarian goals.

The Trump administration’s proposed “update” to the poverty threshold is a wolf in sheep’s clothing

May 22nd, 2019 at 1:37 pm

The Trump administration’s proposal to change the way the poverty line is annually adjusted for inflation is the policy equivalent of a wolf in sheep’s clothing. Sounds technical and weedy, but a new paper by CBPP’s Aviva Aron-Dine and Matt Broaddus shows just how damaging the change will be for the health coverage and benefits of millions of low- and moderate-income people. (An earlier CBPP report focused on other forms of assistance that would be less accessible under the change.)

Because the change will make the poverty line grow more slowly than it would otherwise, fewer people will be counted as poor and thus, fewer will benefit from government programs whose eligibility or benefits are keyed to the poverty line. Over time, millions will lose health coverage or face benefit reductions.

Importantly, note that none of these people will be better off due to the proposed change. Their actual (nominal) income will grow however it’s going to grow, regardless of the change in the price index. But because the new index will make the poverty threshold grow less quickly, some who would have been counted as poor or near-poor by the current system will no longer be so under the new system.

In other words, the Trump administration is proposing to lower the poverty rate without lifting a finger to help the poor.

In so doing, as Aviva and Matt point out, eligibility and benefits for many health programs would be cut, including:

–A Medicare subsidy that helps “low-income seniors and people with disability afford prescription drugs.”

–Medicaid and CHIP coverage, including maternal/child health and family planning.

–Premium tax credits that help people afford ACA marketplace coverage.

The table below shows how, after being in place for a decade (since its damage builds over time) at least hundreds of thousands of people would lose coverage or benefits from the change. Millions buying coverage in the exchanges would lose receive smaller credits against the cost of their premiums. Relative to today’s cost-sharing levels, “more than 50,000 people who would see their deductibles increase from about $850 to about $3,200.”

Source: Aron-Dine, Broaddus.

The switch to the slower-growing price index is being pitched as an innocent attempt to make the poverty line more accurate. But the phoniness of that argument is immediately apparent. The official poverty line is already too low and this change makes it even lower. The official threshold is based on an outdated methodology derived in the 1960s, and as such, it hasn’t evolved to incorporate the costs faced by low-income families today, like child care costs or the increase in housing relative to other costs. For these reasons, as Aviva and Matt report, “households just above the poverty line have high rates of material hardship: for example, high uninsured rates and difficulty affording health care, as well as high rates of food insecurity.”

Right now their proposal is in a 45-day comment period. We’re tracking this closely and will keep folks posted on its progress.

For now, just recognize that far from a benign, technical adjustment, this change threatens the economic and health security of millions of vulnerable people. I guess I should be inured to their regulatory (i.e., non-legislative) actions in this space (e.g., work requirements to hassle people off of SNAP or Medicaid rolls), but once again, the Trump administration is laser focused on solving the problem that America’s poor have too much and the rich have too little.

Pushing back gently but firmly on Michael Strain’s non-stagnation argument

May 16th, 2019 at 2:08 pm

A few folks have asked me about my friend Michael Strain’s recent Bloomberg piece where he argues against wage stagnation (it’s “more wrong than right”). It’s an old argument but one worth having, and Michael makes some important points and misses some big ones too (5, to be precise).

Larry Mishel and I counter a much shorter-term version of Michael’s case here but similar issues pertain. Certainly, the evidence he presents doesn’t change the basic wage story that I and many others carry around in our heads.

I think Michael’s most germane point is that nobody defines “stagnation.” If you think stagnation means real wages for low-wage workers have never gone up in the past four decades, you’re wrong. The figure below, from a recent piece I published (one I’ll get back to re a key point Michael misses), shows real wages for low and moderate wage workers stagnated through the 1970s, 80s, and 2000s.

But, in periods of very tight labor markets—the latter 1990s and now—they grew at a decent clip. This is key insight #1 about real wage growth for too many workers. It’s not that they’ve never grown. It’s that their growth periods in recent decades have been few and far between. And it’s largely dependent of achieving persistent full employment, a condition that’s also been too rare in recent years (see this exciting new paper on precisely this point!).

Key insight #2 is that, sure, switching to a slower-growing deflator leads to faster wage growth and there are good arguments for various choices (see Mishel/Bivens’ cautions re Michael’s choice of using the PCE for wages). But it doesn’t wipe out long periods of stagnation. Here’s the real 20th percentile wage (2018 $’s) using both the CPI-RS (used in the figure above) and the PCE. Just like the above figure: periods of growth, but longer periods of stagnation.

Key insight #3 is especially important and I’d urge fair-minded conservatives to think more about it. If you’re trying to understand why a lot of people have long been unhappy about their paychecks, you can’t just look at wage trends, you must look at their wage levels. That’s what I do here, and I argue that given what a lot of people are taking home in their paychecks, it’s awfully hard for them to make ends meet when paying for child care, health care, housing, and maybe even saving a little afterwards.

Insight #4 is that non-wage benefits don’t change the story, and probably make it less favorable for the “no stagnation” argument. We know, for example (because Larry Mishel always tells us), that the average benefit share of compensation has not accelerated over the stagnation periods shown above. Thus, non-wage comp cannot have offset slower real growth.

But it’s also likely the case—we don’t have long time series on this—that low- and moderate wage workers are no more likely, and I suspect are less likely, to have improved benefit packages over time. That is, if the average hasn’t accelerated, my bet is that the median and below have done worse.

Insight #5 provides what I suspect is another big reason that many workers feel left behind: the rise of wage inequality. As Larry shows here, from 1979-2017, real earnings of the top 1% grew 135% faster than those of the bottom 90%. And such disparities would remain no matter which deflator you use (because both low and high wages would be deflated by the same values).

Finally, I’d urge the no-stagnation crowd to consider why, as Michael notes, stagnation is frequently asserted as fact, by “[p]residential candidates,” “commentators and other opinion leaders….” Why does this resonate with audiences, especially in places feeling the pinch of globalization and deindustrialization?

It could be that they’re using the wrong deflator. But I’ll bet it cuts a lot deeper than that. I’ll bet it’s because they’re right.