Sunday, October 12, 2014

Disinflation, Here We Come

Update (10/14)

It seemed worth bringing this question to the data. I built a simple three-variable vector autoregression of monthly PCE headline and core inflation and the monthly change in the Brent crude spot.

Assuming a price of $84/barrel -- yesterday's close -- I get that headline PCE inflation should bottom out at 0.9 percent in June 2015, incorporating in base effects. So the back-of-the-envelope calculations below check out nicely. We'll see if the model's forecasts hold up.

For future reference, the rule of thumb that comes out of the impulse response function is that for every 10-percentage-point increase in the spot price of Brent crude, expect 0.8 percentage points more in annualized headline PCE inflation for the next four months.

*     *     *

Spot prices for oil have dropped 20 percent in the last three months, from $110 to $90 a barrel. If they remain at these levels, inflation in the United States will slow quite a bit, and quickly at that. My estimate is that headline PCE inflation will fall to just under 1 percent within the next three months of data.

The estimate isn't that difficult to reach. Gas prices track closely with Brent crude, and some of the drop has already filtered into average prices at the pump, and some of that might already be in the August PCE data -- it's not clear exactly how much.

Suppose gasoline, as counted in PCE, end up 10 percent cheaper in September than in August. And energy goods and services is about 5 percent of personal consumption, of which 3 percentage points of that is directly gasoline. The other energy prices moves in tandem, and there are plenty of other prices in the economy that are sensitive to gasoline -- so that gives you about a half percentage-point drop in PCE inflation.

Recall that PCE inflation is 1.5 percent year-over-year. So, one surprise from energy markets, and we could be below one percent. At a time when we are supposed to be a couple months away from a rate hike, this could complicate the exit plan.

Saturday, September 27, 2014

What I Am Up To

Upperclassmen at Princeton must complete independent research, and Ilyana Kuziemko will be my adviser for junior year. I am still figuring out my research topic, but she focuses on inequality, criminal justice, public health insurance, and education -- so it is likely my topic will be in that general area. I usually do not share updates from my life at Princeton, but in case anyone has suggestions for things I should read or specific areas I might explore, I would very much appreciate it if you could reach out to me.

Update (10/29):

So I am doing research on that classic topic of labor economics: job training programs. It turns out that some questions about job training have been buried in the Basic CPS since 1994 to present and in the March Supplement from 2001 to 2009. I found them.

Using those data, I'll be examining the evolution of these programs since 1994, including the major 1998 reform, how the programs differ by type, and how the programs respond to business cycles.

Since so much work has been done estimating the treatment effects of these programs -- do they reduce unemployment? do they raise wages? -- I'm also going to do some work on another question using data from the American Time Use Survey: What's the elasticity of total search time with respect to participation? That is, do these programs substitute for other search activity, or does participation actually increase search time?

The goal here is to work towards answering the big question of how, not just if, job training programs work in the context of labor market search.

More on this soon.

Update (12/1):

It turns out there are insufficiently many observations of job training program participants in ATUS data (~300 pooled over 2003 to 2013). For what it's worth, which isn't much, my estimates of the elasticity of search time with respect to participation was about 2. That is, participation doubled search time -- depending on the kind of labor market search model in which this is contextualized, I imagine that is sufficient to explain all of the gains in labor market outcomes. Perhaps even more than sufficient, so that one would find that participation is effectively requiring the use of an inferior search technology. Interesting stuff, still, but one needs a bigger survey. If you find this question interesting and have the resources to make such a survey happen, please do reach out.

All this is to say I changed the topic of my research. Doing work now on the Supplemental Nutrition Assistance Program (SNAP, formerly the Food Stamp Program), specifically in the vein of Raj Chetty's 2008 paper on unemployment insurance, illiquidity, and moral hazard.

Monday, September 1, 2014

Tipped Workers and Hours

A couple summers ago, I worked as a busboy at a seafood restaurant on the Jersey Shore. The restaurant no longer exists. Hurricane Sandy wiped it off the map. But I was reminded of that summer on Labor Day.

Raising the federal minimum wage, now $7.25, has become a greater priority for public policy over the last couple years. My home state of New Jersey raised it by referendum to $8.25 for 2014 onwards, and we're hardly alone.

Here's an important reminder: More Americans (1.98 million) earn less than the minimum wage than earn the minimum wage (1.56 million), according to the Bureau of Labor Statistics. Tipped workers, like I was, earn a base pay of $2.13 an hour. They are expected to receive $5.12 an hour in tips to bring them up to $7.25.

Why I'm writing this post is to connect that importance of sub-minimum-wage workers to something else. Jodi Kantor had an excellent report last month on the role of erratic schedules for part-time workers who are also trying to care for a family. Kantor details the particularly painful experience of "clopening," that is, working until the close of a restaurant late into the night and then waking up early to reopen it the next morning, an experience I enjoyed a couple times.

Claudia Goldin's work has emphasized the role of hours -- in fact, she established a clear the flexibility of hours in an industry to the magnitude of the gender pay gap in that industry. It's likely that the tipped minimum wage is linked to these crazy hours. Let me explain why.

Let's imagine that I worked Saturday night, which is particularly busy on the Shore. Counting tips plus $2.13-an-hour base pay, I usually made well over the $7.25 minimum wage. But I'd often get assigned to work Monday mornings as well. Not many people like to eat seafood at 9 a.m. The consequence of this is that tips ran well below the $5.12-an-hour they'd need to be to keep me at a wage of $7.25 for every marginal hour. Instead, what was happening was that my tips from the Saturday night were being pulled forward in time, so that all I was really earning, on the margin, was the $2.13 an hour.

Some labor economists think that fraud and underpayment is one reason why the restaurant industry defends the tipped minimum wage so dearly. That's possible, but the story I just gave you provides the intuition for another reason why the tipped minimum gets defended: It gives the restaurant industry another pool of workers whose marginal wage is $2.13, because they've happened to pick up a shift that was more lucrative.

The key point is that this allows restaurants to stay open during hours when, at $7.25 an hour, it would not be otherwise profitable to be open. On the margin, those hours don't make sense -- they only make sense because the tipped minimum wage framework allows employers to average the cost of labor over a pay period. Blame the crazy hours, and perhaps the "clopener," on the tipped minimum wage.

Wednesday, August 20, 2014

Housing, the Weakest Link

Let's run down the macroeconomic checklist. Inflation? It seems to be heading slowly back to the Fed's target of two percent. Unemployment? We're two quarters away from full employment, at least as the Congressional Budget Office defines it. Labor markets more generally? Also tightening, now quickly. Output? Steady overall.

Housing? If there is anything that looks worse than before interest rates jumped in summer 2013 -- that is, something that would suggest a reassessment of where monetary policy is headed -- it's housing. Everything else looks soon to be ready for monetary policy's exit.

One-unit residential construction had begun crawling upwards from historic lows in 2011 and 2012. Then comes the fear of the taper and -- wham -- the rebound in construction just stalls out.

It's no less apparent if you include multifamily housing. Real residential construction spending is also ugly. So it's no surprise that residential fixed investment as a share of GDP has been flat for the last four quarter at its super-low level. New one-family home sales are stalled out, too.

Why should all that be a subject of concern?

Keep in mind that for much of that trough, exceptionally weak housing was one of the main factors the Fed cited in its monetary-policy statements as cause for keeping its policy rate at zero. And it was the essential reason for why the Fed got involved on the longer end of the yield curve. "The housing sector continues to be depressed," or a similar phrase, sat in the top paragraph summarizing economic conditions for years. The current phrase is: "The recovery in the housing market remains slow." That seems a considerable overstatement of where we are. The recovery in the housing market is not slow. It's nonexistent. It has been since July 2013.

That suggests the housing sector is at the moment substantially more dependent on relaxed monetary conditions than I, and quite a lot of other analysts, thought it was when mortgage rates jumped. For a while since then, I have been worried about the apparent disconnect between prices and construction, as resurgent home values should push a lot of buyers into the market for newly-built homes. That wasn't happening. Now, however, the contradiction may be resolved in an adverse way: Home prices have stopped rising for the last few months.

There may be more to the weakness in housing than just the jump in interest rates. The U.S. is still amid the deleveraging process for mortgage debt. And, as John Carney and Justin Lahart write, it's hard to see why young people would be taking on new mortgage debt, what with their student loans and their at-best modest income prospects for now. Jed Kolko of Trulia makes the point even more convincingly with MSA-level data, showing that job growth has surpassed rebound effects as the main determinant of home-price appreciation. I have no doubt that all of these bigger forces are at play -- but look at the graph again. Does the housing recovery not stall out right in summer 2013?

Housing seems to provide a more unambiguous case for a slow exit than labor markets at this point, even if that's what the Fed has chosen to talk about at Jackson Hole this week. This rate of construction is not at all what the U.S. should be seeing if monetary easing can rightly declare mission accomplished.

Monday, August 18, 2014

Vacation Is All I Ever Wanted

Here I am in Vox, investigating the decline of the American vacation:
Nine million Americans took a week off in July 1976, the peak month each year for summer travel. Yet in July 2014, just seven million did. Keeping in mind that 60 million more Americans have jobs today than in 1976, that adds up to a huge decline in the share of workers taking vacations.

Some rough calculations show, in fact, that about 80 percent of workers once took an annual weeklong vacation — and now, just 56 percent do.
While I've got the dataset on hand, I figure I might as well figure out when Americans take weeklong vacations, by month. So, here you go.

Part of the reason these calculations are "rough," as I wrote, is that we're operating from the assumption that the reference week (which includes the 12th of the month, per Current Population Survey rules). It is obviously wrong to use the week that includes December 12 to estimate absences in December, because, you know, Christmas. But this is may well be cancelled out for similar, opposite phenomena in other months.

I've made my data available here for easy replication. And, in case you're wondering why I know that song reference in the title -- the year it came out minus my age is a negative number -- you may find it amusing to know that this was one of the classics of my childhood.

Sunday, August 17, 2014

Secular Stagnation, Meet Data

There is a new e-book out on secular stagnation, with contributions from Larry Summers, Barry Eichengreen, Paul Krugman, Olivier Blanchard, and others. It offers a lot to discuss, but I'm struck by something mostly missing: implied forward real interest rates.

This seems like it should be important data for the hypothesis, but it's only discussed briefly in the chapter by Frank Smets and two other economists. The graph, fittingly enough, is on the very last page of content.

If real interest rates are going to be low for a long period of time, as secular stagnation implies, then the market should say that, and the way it would is through forward rates. So what do we see?

The first graph shows the instantaneous forward real interest rate -- basically, what the market thinks the cost of overnight funding should be in real terms at some future date -- as derived from TIPS yields. There are more precise ways to do this, but it's close enough for a blog post.

Markets anticipate that real interest rates will return slowly to one percent. For reference, prior to the recession, these rates averaged about 1.5 percent, at least on an ex-post basis.

Another way to look at this is to figure out the implied real forwards for longer-term bonds. This builds in market views on what the term premium will be in the future, of course. The market's message is generally the same: Real interest rates will be the same as, or modestly lower than, where they have been in the past. Most of the depressed position of interest rates comes from the short run.

It's early to come to any solid conclusions about what the secular-stagnation discussion has and hasn't accomplished. Yet the thought that lingers for me is: Why do we need "secular stagnation"?

For all the interesting discussion so far, much of it in this book, nothing has yet convinced me that secular stagnation is a necessary and unique component to my understanding -- my mental model -- of the post-recession global economy. What someone who is convinced needs to do is identify the key facts that cannot be explained by deleveraging and the huge shock to nominal income.

The closest anyone has come, in fact, was Summers when he discussed the mid-2000s as a period of mediocre growth despite pedal-to-the-metal monetary policy. Krugman's "observation #2," that there seems to be a long-term trend towards ever-lower real interest rates, is also getting there. The bar, generally speaking, for a new theory to get into a model is that the model's predictions must be otherwise inconsistent with reality.

There just doesn't seem to be such a need for an additional "secular stagnation" component, as I interpret the data above. There is nothing outrageously low about forward real interest rates. If secular stagnation is merely an organizing device, the problem is that it seems to want to say something about the future, whereas the base case of deleveraging and the nominal-income shock speak mainly about the past. That's why the issue must be cast in light of forward interest rates -- and why I remain sympathetic but unconvinced.