Ban the Box: Recent critics of the policy are not nearly as convincing as they think they are

August 22nd, 2016 at 2:18 pm

As any student of policy analysis knows, when we analyze the impact of a real policy on real people, we must look for “unintended consequences,” a kind of backfiring where for some unforeseen reason, the policy hurts those it’s trying to help.

This came up recently in a couple of research papers evaluating the impact of “ban the box,” or BTB, an initiative intended to meet an extremely venerable goal: to help those with criminal records make their way back into the workforce. BTB does so by moving background checks from the beginning of the application process to the end, often after a conditional offer of employment has been extended.

The problem, according to some critics of the policy, is that while BTB might help those with criminal records get their feet in the door, employers without criminal record information will engage in “statistical discrimination.” That is, they’ll discriminate against applicants they believe most likely to have criminal records: young black and Latino men. Based on this dynamic, one author of a critical analysis of BTB concluded that the policy “does more harm than good.”

Not so, and here’s where policy analysts need to be a lot more careful. Are they identifying a problem with the policy or with employer behavior more generally? Especially if it is the latter, are there better ways to address that behavior than policy repeal? Do the policy’s beneficiaries outweigh anyone hurt by unintended consequences?

That’s why a new policy brief from Maurice Emsellem and Beth Avery (E&A) of the National Employment Law Project (disclosure: I chair NELP’s board) is so important. Media reports haven’t done enough to contextualize those studies’ findings and have been too quick to dismiss a policy that, while only a limited part of what’s needed to help disadvantaged individuals get a fair shake in the job market, generally appears to be having its intended effects.

As E&A note, evidence from cities across the United States – from Durham, North Carolina to Washington, DC to Atlanta, Georgia – indicates that ban the box is increasing employment among people with criminal records. The critical studies don’t challenge that claim directly, but, as noted above, argue that employers respond to ban-the-box policies by discriminating broadly against all young black and/or Latino men. But while employer discrimination surely exists, the studies fail to make the case that this is due to BTB.

One study, by Amanda Agan and Sonja Starr, relies on fictitious job applications for young men ages 21-22 submitted to New Jersey and New York employers both “shortly before and after the New York City and New Jersey private sector ban-the-box laws took effect in 2015.” The study found that the gap in callbacks between white applicants and black applicants (whom were given distinctively racial-sounding names) increased following the introduction of ban the box.

But does this prove unintended consequences? For firms that used and then banned the box, black callback rates were about 11 percent before and after the policy change, while white callback rates went up, from around 11 percent to 15 percent. It appears that these results were driven by increased callbacks for applicants with criminal records of both races, as black applicants with records had only an 8.4 percent callback rate and white applicants with records had only an 8.8 percent callback rate prior to the change.

In other words, there were increased callbacks for applicants of both races with records (though more so for whites): a good example of an intended consequence. The NYT also got this wrong, asserting that the paper found that “employers became much less likely to call back any apparently black applicant.”

The other study, from Jennifer Doleac and Benjamin Hanson, compared employment across the United States between metropolitan areas that did and didn’t enact ban the box laws. While the authors found reduced employment for black men without college degrees between the ages of 25 and 34, they also found increased employment for other black workers.

There are also some potentially serious modelling problems with this study. First, the authors consider every worker in an entire metropolitan statistical area (MSA) to be covered by a ban-the-box law “if any jurisdiction in their MSA has a BTB policy.” A BTB law passed in East Palo Alto, CA in January of 2005, for example – a city with around 30,000 people – looks to me like it would have caused the entire San Francisco-Oakland-Hayward, CA MSA, which has a population of closer to 4.7 million, to be considered covered, despite no other jurisdiction within that MSA adopting ban the box until San Francisco did so in October of that year.

What’s more, public-sector-only BTB laws count, and they’re all that’s in place in nearly 80 percent of months the authors examine in MSAs covered in this way (including in the aforementioned California MSA until July of 2013). It’s not clear why we’d expect public-sector BTB laws to affect the hiring behavior of private employers, or the behavior of both public and private employers in neighboring cities and counties that don’t have BTB laws. The authors also neglect to control for local unemployment, which I’d guess matters in a model that predicts employment outcomes.

My conclusion from the studies and the NELP report is that it’s not clear that employer responses to ban-the-box laws are the problem. What is clear, from these and many other studies, is that there are serious discrepancies in job opportunities for people of different races. Those discrepancies have nothing to do with BTB and must be addressed. And if the critics are right—if some employers are actually responding to ban-the-box laws by ramping up statistical discrimination—then they’re doing so in violation of federal civil rights law.

E&A conclude correctly that “a comprehensive policy response is necessary to fundamentally increase job opportunities for people with records and reduce race discrimination in hiring.” That policy response must include ban the box, enforcement of federal civil rights law, policing reforms, prison reforms, education about implicit bias, and a suite of other policies. It takes a village of such policies to achieve economic justice.

What if the Fed is just really good with anchors?

August 22nd, 2016 at 11:30 am

I’ve got a piece over at WaPo that OTE’ers might enjoy on the Fed’s 5, 3, 2 problem. As in their unemployment and interest rate targets (~5 and 3 percent) are too high and their inflation target (2 percent) is too low.

Let’s talk about this last bit—the inflation target—a bit more, though this conversation also applies to the other stuff in the piece, as you’ll see.

First, and this isn’t the main point of this post, but a bit of venting. Actually, never mind—I dealt with this through a cathartic tweet (the 2% target is an average, not a ceiling! Can I please get some symmetry!).


The actual point of this post is just to reflect a bit more on the phlat Phillips Curve (PC), as shown in this recent analysis by Fed economist Michael Kiley (whose work on all this is thoughtful and compelling). One of Kiley’s figures, below, shows the extent to which the PC has flattened in recent years.

Source: Michael Kiley

Source: Michael Kiley

The question is “why so phlat?” and one answer that I don’t get into in my WaPo piece is that the Fed has gotten really good at convincing everyone that damn it, inflation is going to stay low and stable and that’s all there is to it. In Fed-speak, that’s saying “inflationary expectations are well-anchored” around their target of 2%.

Kiley and others provide some evidence to that effect, but what’s interesting to me is how this explains important findings like these which show the collapse of traditional statistical measures that used to explain the variance in inflation using measures of economic slack.

A friend provides a useful analogy: Trying to estimate the PC these days is a little like testing the impact of outside temperature changes on an inside room that’s climate controlled. You’re not going to pick up a lot of variance because the climate is effectively controlled by the thermostat. If you, say, regress the inside room temperature on the outside temp, your coefficient will wiggle around zero, because the thermostat is doing its job.

In other words, the PC is flat because the Fed is effectively controlling inflation.

This seems convincing (if it sounds really obvious, I assure you, as an old person, that wasn’t always the case) but one would like to disprove other explanations, especially since the extent of the anchoring would have be really strong to explain how little inflation has responded to output gaps either when they were really very large or when they were closing pretty quickly. Kiley takes you through numerous other suggested explanations, including basic rigidities in prices and wages.

But I’ve always wondered if there’s a globalization piece to all this. Surely increased global supply chains put downward pressure on prices. Also, inequality, low worker bargaining clout, and the decrease in collective bargaining have long diminished the link between productivity and real wages…and perhaps prices as well.

Last point: as I stress in the WaPo piece, the inflation target is too low—at 2%, it invokes possible zero-lower-bound problems the next time we hit a downturn, and especially with a…um…difficult Congress (meaning adequate countercyclical fiscal policy may well not be forthcoming), that’s a really serious problem.

If they can anchor so effectively at 2%, why not 4%?

It’s all connected…4 referrals from today’s papers.

August 19th, 2016 at 8:49 am

Three articles, one blog post.

First, Dean Baker points to this great Bloomberg article by former Fed regional bank pres Narayana Kocherlakota (NK) on how, since black unemployment typical runs 2x the overall rate, Fed policy is especially consequential for them (and other minorities). It’s really just the old adage that when the economy sniffles, less advantaged workers catch pneumonia, and–a key theme of my own work–less advantaged workers are disproportionately helped when the economy is strong. Full employment is especially important for blacks, as I’ll show again in a moment.

The piece points out that these relationships came up in the minutes from the recent Fed meeting, something that hasn’t happened much at all in the past. Some, and not a little, credit for that goes to the activist group Fed Up, which continues to have real impact on these critical debates.

One could certainly ask “what took them so long?” As NK’s figure shows, this 2x relationship persists through the long history of this data series. But it’s still progress.

Second, I wanted to link to a piece I recently did that ties a lot of NK’s insights together, adding the unemployment/wage dimension. I took that ~2x relation and mapped it onto a wage/unemployment elasticity for low-wage workers in the spirit of Val Wilson’s work.

Source: my analysis

Source: my analysis


Until recently, growth in this expansion has been a spectator sport for many disadvantaged workers. One way to help them is for the Fed to accommodate very low unemployment. That could conceivably trigger inflationary concerns, but based on how weak that correlation is these days, lower unemployment seems an extremely favorable trade-off to the low-wage workers who would benefit disproportionately in terms of faster wage growth.

But I also don’t want to forget the larger picture:

Of course, life is more complicated than these relatively simple connections imply. It will take a lot more than just the Fed holding off on interest-rate increases to generate true racial economic justice. Getting there will also require, as a basic starting point, both criminal justice reform and direct job creation in neighborhoods that have historically been left behind (even when the rest of the country is at full employment).

Third, speaking of that larger picture, here’s a little piece in today’s WaPo (print edition) that thinks about the structural forces driving inequality–taxes and transfers can help and are helping. But we need to do much more and think much bigger to deal with the power imbalances behind these inequities.

Finally, one reason productivity is so low is that investment is so low. And one reason investment is so low is that public companies would rather do stuff that boosts their near-term stock price–share buybacks and dividend payouts–than their long-term productivity. It’s a big, serious, long-term problem that relates to the structural shifts alluded to in my WaPo piece, not to mention the institutional forces that NK and I document.

In other words, Buddha was right: it’s all connected…


The dollar’s makin’ me holler, and other tales of the macro muddle

August 17th, 2016 at 3:40 pm

Old gold bugs know the old plaint, “gold, gold, you’re makin’ me old.” Well, today’s version is “dollar, dollar, you’re makin’ me holler!”

I’ll be brief—I’ve got something longer on this coming out soon—but I’ve been struck by both the recent jigs and jags in the dollar and other currencies. A central reason for those movements is that it’s awfully hard to figure out what the Fed is up to, as I’ll recount in a moment.

First, currency movements have been following some unusual patterns – for example, rising after central bank rate cuts (Japan, Australia; typically, we expect currency values to fall after rate cuts) and jumping around here in the US with more volatility than usual, highly sensitive to winks and nods from our Fed about their next rate move.

The WSJ this AM featured a story about the falloff in the dollar year-to-date, which opened thusly:

Federal Reserve officials are trying to signal that another rate increase is likely while at the same time questioning whether the economy can expand fast enough to justify lifting them much beyond that.

It is a confusing combination that is sapping the Fed’s influence over markets.

Then, a few minutes ago, the Journal tells me that the dollar’s rallying on the suspicion that when the minutes from the Fed’s July meeting come out, they’ll be leaning into a rate hike later this year. So the Fed is like, “we’re gonna raise rates,” but the extent to which market investors believe them are changing on a daily basis.

Actually, an hourly basis. Check out the summersault in the dollar index upon the release of the Fed’s minutes. Basically, it goes from “they’re gonna raise!” to “no they’re not!” to “maybe…who knows?!” all in the course of a few New York minutes.

Source: WSJ, my animation

Source: WSJ, my animation

In fact, it’s all a muddle. Read this interview with Fed governor John Williams. The thrust of his argument is that interest rates need to go up as the Fed’s been “adding enormous policy accommodation over the past several years” and, even while they’ve long been missing their inflation target on the downside, there’s a risk of getting “significantly behind the curve.” At one point he makes a distinction between “letting up on the gas” and “tapping the brakes,” one that left me and I suspect others going “wait…wuh?”

But Williams’ own work, and his isn’t the only that finds this, suggests that the Fed’s so called neutral long-term rate—the rate consistent with full employment and stable prices—is zero right now, meaning they haven’t been enormously accommodative. In that same interview, he seems to be reaching to square these contradictions, by suggesting that the Fed’s current model—targeting 2% inflation, a Fed funds rate of ~3%, and an unemployment rate of ~5%–is not reliable and that they should maybe move to a different targeting regime, like price-level or nominal GDP targeting. Both of those would lead the Fed to take rate hikes way off the table right now.

The point isn’t to pick on Williams or anyone else who’s clearly trying to figure out what’s going on, or more precisely, what’s changed in the basic macro-economic relationships such that prior guideposts are no longer reliable. The problems arise because of a level of confidence that these prominent figures believe they must exude. Part of their model is providing guidance and “we’re not sure what’s going on” is considered to be…um…a bit weak in that regard.

But the cognitive dissonance–asserting X despite the fact that my model and the evidence suggest Y–is leading to an incoherence that’s predictably generating volatility.

I’m less worried about investors than about workers. And for all the muddle, the one thing that seems clear is that the risks to the economy and particularly the labor market—which is generating solid job growth and even some wage gains (for which we should all give Chair Yellen and the Fed serious credit)—remain “asymmetric:” there’s a greater risk of needlessly slowing non-inflationary growth than there is of inflation accelerating. The notion of being “behind the curve” in that regard seems indefensible.

Like I said, more to come on this. As Williams’ comments re alternative targeting suggests, this is an important time to be thinking about pretty different models of how the world works.

When fiscal policy lags, the one-two fiscal/monetary punch doesn’t land

August 15th, 2016 at 2:12 pm

Two recent pieces looked at where we are in this recovery and why, seven years into an expansion, we’re still not at full employment. One is by my main man Josh Bivens from the Economic Policy Institute and the other is by David Mericle and Avisha Thakkar from Goldman Sachs (GS) research (sorry; no link to that one).

The figure below–not from either of these pieces–motivates the discussion. It shows three lines of real GDP: an estimate of potential GDP pre-great-recession (2007), the most recent estimate of the same, and actual real GDP (potential GDP is CBO’s estimate of what GDP would be with fully utilized resources; it’s the value of the economy when it’s firing on all cylinders). I–somewhat artfully, you’ll surely admit–call this figure: The runner who can’t cross a goal line that’s moving towards her.

Sources: BEA, CBO

Sources: BEA, CBO

Why is that?

The GS folks take a unique and informative approach: they compare a spate of economic indicators in the US to the “Big Five” advanced economy financial crises (taken from the work of Carmen Reinhart and Kenneth Rogoff) as well as to the 2008 European crisis economies. Basically, GS asks: how has the US recovery gone compared to prior recoveries elsewhere from financial-crisis-induced recessions? (The “Big Five” include Spain in 1978, Norway in 1987, Finland in 1990, Sweden in 1990, and Japan in 1992.)

Josh provides a different but also useful comparison: prior US downturns.

The GS team concludes that the labor market in particular has outperformed their historical comparison groups, as shown through the unemployment rate comparison below. True, our unemployment rate is biased down due to the weak performance of labor force participation and still-elevated underemployment, but as I’ve extensively documented, the US job market has been tightening up for awhile, driven by solid employment growth, now averaging around 200,000/month. See Bivens’ Figure A on this point.

Source: GS Research

Source: GS Research

Still, as the first figure shows, the output gap is yet to close, even as potential GDP’s been downgraded. That reflects both slower labor force growth (some of which is demographics but some of which is weak labor demand) and our most serious outstanding economic problem, very slow productivity growth.

What’s most interesting to me about both the Bivens and GS studies is in regard to policy responses. Initially, US policy makers hit back hard with both monetary and fiscal policy. I’ve argued that the combination is important and complementary: monetary policy lowers the cost of borrowing but if households are both deleveraging and, based on the loss of housing wealth, suffering from lower net worth (i.e., a negative wealth effect), they’re less likely to take advantage of those lower rates. That’s when you need fiscal policy to put more money in people’s pockets such that they’ll tap the monetary stimulus. Ergo, the one/two punch.

Both studies show the following: the one/two punch of monetary/fiscal policy weakened when fiscal support for the recovery faded. The GS figure is striking (sticking with my artful theme, let’s call it Paul Krugman’s nightmare). Fiscal support started strong both here and in Europe, as did (see second figure) monetary policy (the negative numbers reflect the Fed’s lowering and holding down the Fed funds rate). But while the monetary authorities kept their stimulus going, austere fiscal policy kicked in with a vengeance.

Source: GS

Source: GS

Source: GS

Source: GS

Bivens austerity figure is also illuminating, and includes all three levels of government spending (fed, state, local).

Source: Bivens

Source: Bivens

Over to Josh:

[The above figure] shows the growth in per capita spending by federal, state, and local governments following the troughs of the four recessions. Astoundingly, per capita government spending in the first quarter of 2016—27 quarters into the recovery—was nearly 3.5 percent lower than it was at the trough of the Great Recession. By contrast, 27 quarters into the early 1990s recovery, per capita government spending was 3 percent higher than at the trough, 23 quarters following the early 2000s recession (a shorter recovery) it was 10 percent higher, and 27 quarters into the early 1980s recovery it was 17 percent higher.

His judgment is harsh and unequivocal…practically Old Testament:

If government spending following the Great Recession’s end tracked spending that followed the recession of the early 1980s for the first 27 quarters, governments in 2015 would have been spending an additional trillion dollars in that year alone, translating into several years of full employment even had the Federal Reserve raised interest rates. In short, the failure to respond to the Great Recession the way we responded to the other postwar recession of similar magnitude entirely explains why the U.S. economy is not fully recovered seven years after the Great Recession ended.

I think that’s probably right and I also think it helps to at least partially explain what may be the most negative development in any of the above pictures. I’m gonna let you guess what that is…I’m waiting…hint: it’s in the first figure.

It’s the sharp decline, from the 2007 to the 2016 vintage, in CBO’s estimate of potential real GDP. Hard to believe that downshift is a demographics story, because CBO estimators knew the population’s demographics back in 2007, when they had potential growing a lot more quickly. In fact, CBO assigns most of the reduced potential to “reassessed trends:” a more pessimistic outlook re hours worked and productivity than what they believed before the crisis. By taking half the punch out of the one/two fiscal/monetary punch, dysfunctional and mindlessly austere policy makers have contributed to the loss of hundreds of billions in value-added.

Still, sticking with the first figure, soon the aging runner will likely finally cross the goal line, and we’re doing better here than most other advanced economies. We’re nudging towards full employment and I’m even finally seeing a little pressure on wages. But it would be foolish to ignore the mistakes we’ve made and what they’re actively costing us in lost output, jobs, and living standards.