The Atlanta Fed's macroblog provides commentary on economic topics including monetary policy, macroeconomic developments, financial issues and Southeast regional trends.
- BLS Handbook of Methods
- Bureau of Economic Analysis
- Bureau of Labor Statistics
- Congressional Budget Office
- Economic Data - FRED® II, St. Louis Fed
- Office of Management and Budget
- Statistics: Releases and Historical Data, Board of Governors
- U.S. Census Bureau Economic Programs
- White House Economic Statistics Briefing Room
January 15, 2015
Contrasting the Financing Needs of Different Types of Firms: Evidence From a New Small Business Survey
The National Federation of Independent Business's (NFIB) small business optimism index surpassed 100 in December, a sign that small business' outlook on the economy has now reached "normal" long-run average levels. But that doesn't mean that everything is moonlight and roses for small firms. One question from the NFIB's survey (one that is not used in its overall optimism index) concerns a firm's ability to obtain credit. The survey asks, "During the last three months, was your firm able to satisfy its borrowing needs?" The chart below shows the net percent (those responding "yes" minus those saying "no") of firms reporting improving credit access.
The chart suggests that credit access has improved significantly since the end of the recession but that conditions still appear to be tougher than typical. Given the importance of small firms to employment growth, we at the Atlanta Fed have been particularly interested in monitoring financing conditions for small businesses. For this reason, we've conducted a regular survey of small businesses in the Southeast since 2010. In the fall of 2014, we joined forces with the New York, Philadelphia, and Cleveland Feds to expand and refine the small business data collection effort. The results of that survey are now available on our website and include downloadable data tabulations by different types of firms. Specifically, data are available by criteria including states, industries, firm size (in terms of revenue), and firm development stage.
Our previous small business surveys have focused on the experiences of young firms, so I found the new survey's tabulation by firm development stage of particular interest. For example, here's a summary of the experience of startups' ability to access financing markets versus that of mature firms.
First, what constitutes a startup? For comparison purposes, we draw the line (somewhat arbitrarily) at less than five years old. For mature firms, they not only have to be at least five years old, but they also must have at least 10 employees and hold some debt. When I picture a startup, I imagine a new restaurant owner purchasing tables and chairs, or a tech company manufacturing a prototype to market to potential investors. These types of firms are unproven and risky and tend to need relatively small amounts of money. Which begs the question: where are they going to get funds they need to grow? Before answering that question, let's examine the recent business performance of startups in the survey. About half of startups operated at a loss during the previous 12 months, but only about 20 percent had shrinking revenues. Most were either increasing the size of their workforce or had the same number of employees as a year ago. The top challenge reported by these young businesses was nearly tied between "difficulty attracting customers" (reported by 27 percent of firms) and "lack of credit availability" (reported by 26 percent of firms).
So how do those behind startups fund their businesses? In 2013, nearly half relied primarily on personal savings, whereas about 18 percent primarily used retained business earnings. Without a solid revenue history to prove their creditworthiness, financing was understandably difficult to come by. Only about 38 percent of startups received at least some financing, compared with 93 percent of mature firms. Many startups assumed it would be a fruitless endeavor—about one-fifth of them assumed they would be turned down, the cost would be too high, or the search would be too time consuming. The number of people who sought financing was about equal to those who were discouraged, and most were seeking less than $250,000.
Where did they apply? Their search was much broader than used by their counterparts at mature firms. Although both types of firms sought mostly loans and lines of credit, applications for products backed by the Small Business Administration, credit cards, and equity investments were notably higher for younger firms compared to mature firms. When it came to loans and lines of credit, there were large differences not only in what types of insitutions they submitted applications to, but also where they were most successful. Startups were mostly likely to apply at large regional and large national banks, but their approval rates were higher with smaller banks and online lenders (see the table).
The differences between young firms and mature ones is only one way to look at the data. The full report details variations by firm size, industry, and state. For more on general business and finance conditions of small firms, visit the small business trends dashboard.
January 09, 2015
Gauging Inflation Expectations with Surveys, Part 3: Do Firms Know What They Don’t Know?
In the previous two macroblog posts, we introduced you to the inflation expectations of firms and argued that the question you ask matters a lot. In this week's final post, we examine another important dimension of our data: inflation uncertainty, a topic of some deliberation at the last Federal Open Market Committee meeting (according to the recently released minutes).
Survey data typically measure only the inflation expectation of a respondent, not the certainty surrounding that prediction. As a result, survey-based measures often use the disagreement among respondents as a proxy for uncertainty, but as Rob Rich, Joe Tracy, and Matt Ploenzke at the New York Fed caution in this recent blog post, you probably shouldn't do this.
Because we derive business inflation expectations from the probabilities that each firm assigns to various unit cost outcomes, we can measure the inflation uncertainty of a respondent directly. And that allows us to investigate whether uncertainty plays a role in the accuracy of firm inflation predictions. We wanted to know: Do firms know what they don't know?The following table, adapted from our recent working paper, reports the accuracy of a business inflation forecast relative to the firm's inflation uncertainty at the time the forecast was made. We first compare the prediction accuracy of firms who have a larger-than-average degree of prediction uncertainty against those with less-than-average uncertainty. We also compare the most uncertain firms with the least uncertain firms.
On average, firms provide relatively accurate, unbiased assessments of their future unit cost changes. But the results also clearly support the conclusion that more uncertain respondents tend to be significantly less accurate inflation forecasters.Maybe this result doesn't strike you as mind-blowing. Wouldn't you expect firms with the greatest inflation uncertainty to make the least accurate inflation predictions? We would, too. But isn't it refreshing to know that business decision-makers know when they are making decisions under uncertainty? And we also think that monitoring how certain respondents are about their inflation expectation, in addition to whether the average expectation for the group has changed, should prove useful when evaluating how well inflation expectations are anchored. If you think so too, you can monitor both on our website's Inflation Project page.
January 07, 2015
Gauging Inflation Expectations with Surveys, Part 2: The Question You Ask Matters—A Lot
In our previous macroblog post, we discussed the inflation expectations of firms and observed that—while on average these expectations look similar to that of professional forecasters—they reveal considerably more variation of opinion. Further, the inflation expectations of firms look very different from what we see in the household survey of inflation expectations.
The usual focal point when trying to explain measurement differences among surveys of inflation expectations is the respondent, or who is taking the survey. In the previous macroblog post, we noted that some researchers have indicated that not all households are equally informed about inflation trends and that their expectations are somehow biased by this ignorance. For example, Christopher Carroll over at Johns Hopkins suggests that households update their inflation expectations through the news, and some may only infrequently read the press. Another example comes from a group of researchers at the New York Fed and Carnegie Mellon They've suggested that less financially literate households tend to persistently have the highest inflation expectations.
But what these and related research assume is that whom you ask the question of is of primary significance. Could it be that it's the question being asked that accounts for such disagreement among the surveys?
We know, for example, that professional forecasters are asked to predict a particular inflation statistic, while households are simply asked about the behavior of "prices in general" and prices "on the average." To an economist, these amount to pretty much the same thing. But are they the same thing in the minds of non-economists?You may be surprised, but the answer is no (as a recent Atlanta Fed working paper discussed). When we asked our panel of firms to predict by how much "prices will change overall in the economy"—essentially the same question the University of Michigan asks households—business leaders make the same prediction we see in the survey of households: Their predictions seem high relative to the trend in the inflation data, and the range of opinion among businesses on where prices "overall in the economy" are headed is really, really wide (see the table).
But what if we ask businesses to predict a particular inflation statistic, as the Philly Fed asks professional forecasters to do? We did that, too. And you know what? Not only did a majority of our panelists (about two-thirds) say they were "familiar" with the inflation statistic, but their predictions looked remarkably similar to that of professional forecasters (see the table).
So when we ask firms to answer the same question asked of professional forecasters, we got back something that was very comparable to responses given by professional forecasters. But when you ask firms the same question typically asked of households, we got back responses that looked very much like what households report.
Moreover, we dug through the office file cabinets, remembering a related table adapted from a joint project between the Cleveland Fed and the Ohio State University that was highlighted in a 2001 Cleveland Fed Economic Commentary. In August 2001, a group of Ohio households were asked to provide their perception of how much the Consumer Price Index (CPI) had increased over the last 12 months, and we compared it with how much they thought "prices" had risen over the past 12 months.The households reported that the CPI had risen 3 percent—nearly identical to what the CPI actually rose over the period (2.7 percent). However, in responding to the vaguely worded notion of "prices," the average response was nearly 7 percent (see the table). So again, it seems that the loosely defined concept of "prices" is eliciting a response that looks nothing like what economists would call inflation.
So it turns out that the question you ask matters—a lot—more so, evidently, than to whom you ask the question. What's the right question to ask? We think it's the question most relevant to the decisions facing the person you are asking. In the case of firms (and others, we suspect), what's most relevant are the costs they think they are likely to face in the coming year. What is unlikely to be top-of-mind for business decision makers is the future behavior of an official inflation statistic or their thoughts on some ambiguous concept of general prices.
In the next macroblog post, we'll dig even deeper into the data.
January 05, 2015
Gauging Inflation Expectations with Surveys, Part 1: The Perspective of Firms
Central bankers measure inflation expectations in more than a few ways, which is another way of saying no measure of inflation expectations is entirely persuasive.
Survey data on inflation expectations are especially hard to interpret. Surveys of professional economists, such as the Federal Reserve Bank of Philadelphia's Survey of Professional Forecasters, reveal inflation expectations that, over time, track fairly close to the trend in the officially reported inflation data. But the inflation predictions by professional forecasters are extraordinarily similar and call into question whether they represent the broader population.
The inflation surveys of households, however, reveal a remarkably wide range of opinion on future inflation compared to those of professional forecasters. Really, really wide. For example, in any particular month, 13 percent of the University of Michigan's survey of households predicts year-ahead inflation to be more than 10 percent, an annual inflation rate not seen since October 1981. Even in the aggregate, the inflation predictions of households persistently track much higher than the officially reported inflation data (see the chart). These and other curious patterns in the household survey data call into question whether these data really represent the inflation predictions on which households act.
Even if you're unfamiliar with the literature on this subject, the above observations may not strike you as particularly hard to believe. Economists are, presumably, expert on inflation, while households experience inflation from their own unique—some would suggest even uninformed—perspectives.
We have yet another survey of inflation expectations, one from the perspective of businesses leaders. We think this may be an especially useful perspective on future inflation since business leaders, after all, are the price setters. Our survey has been in the field for a little more than three years now—just long enough, we think, to step back and take stock of what business inflation expectations look like, especially in comparison to the other survey data.
Our initial impressions are reported in a recent Atlanta Fed working paper, and the next few macroblog posts will share some of our favorite observations from this research.
We have been asking firms to assign probabilities to possible changes in their unit costs over the year ahead. From these probabilities, we compute how much firms think their costs are going to change in the coming year and how certain they are of that change (see the table). What we find is that the inflation expectations of firms, on average, look something like the inflation predictions of professional forecasters, but not so much like the predictions of households.
But we also find that there is a significant range of opinion among firms, more so than the range of opinions that forecasting professionals express. Some of the variation among firms appears to be related to their particular industries and are broadly correlated with the uneven cost pressures shown in similar industrial breakdowns of the Producer Price Index from the U.S. Bureau of Labor Statistics (see the table).
So what we have now are three surveys of inflation expectations, each yielding very different inflation predictions. What accounts for the variation we see across the surveys? Our survey allows us to experiment a bit, which was one of the motivations for conducting it. We didn't just want to measure the inflation expectations of firms; we wanted to learn about those expectations. In the next few macroblog posts, we'll tell you a few of the things we've learned. And we think some of our initial findings will surprise you.
December 23, 2014
Chances of Finding Full-Time Employment Have Improved
Today's sharp upward revision to the third-quarter GDP reading reinforces the view that the underlying strength of the U.S. economy has been sufficient to support more rapid improvement in the labor market. Last week we noted the solid and broad-based recent improvement in the involuntary part-time work (part-time for economic reasons or PTER) situation over the last year, noting significant declines in the stock of PTER workers across industrial sectors and occupational categories.
In this post we look at labor market improvement over the last year in terms of worker flows. Because the Current Population Survey is set up as a rotating panel, many of the people in the survey in any given month were in the survey a year earlier as well. This allows us to ask the question: if you were an unemployed prime-age individual (25–54 years old) or working PTER one year ago, what are you doing today? Have your chances of becoming employed full-time improved? Chart 1 shows the distribution of labor market outcomes of prime-age workers who were PTER one year earlier. Chart 2 shows the distribution of outcomes for those who were unemployed one year earlier. The data are 12-month moving averages to smooth out seasonal variation.
For both PTER workers and the unemployed, the chances of becoming employed full-time are up from a year earlier (and the chances of being unemployed are down). In November 2013 there was about a 45 percent chance of someone who was PTER a year earlier having a full-time job. In November 2014 that had improved to about a 48 percent chance. This full-time employment flow rate is still much lower than the prerecession average of around 55 percent, and the improvement appears to have stalled a bit in recent months, but it is a notable improvement from a year earlier nonetheless. For PTER workers, the picture along other dimensions is more mixed. The chances of becoming unemployed appear to have returned to around prerecession levels, which is good, but the likelihood of remaining PTER is still quite elevated.
For the unemployed, there has been an even more marked improvement in the full-time finding rate over the last year. In November 2013 there was around a 32 percent chance of someone who was unemployed a year earlier having a full-time job. In November 2014 the chances improved to close to 36 percent. Moreover, the improvement in the rate of finding full-time work is responsible for the similar-sized decline in the chances of remaining unemployed. The only negative here is that the likelihood of an unemployed worker becoming PTER, while low, remains elevated compared with before the recession.
All in all, we think these developments are encouraging and add to the view that the pace of labor market improvement has picked up over the last year.
December 19, 2014
Exploring the Increasingly Widespread Decline in Involuntary Part-Time Work
We at the Atlanta Fed have been arguing for some time that the unusually large number and share of workers employed part-time but wanting full-time work (counted in the Current Population Survey as part-time for economic reasons, or PTER) partly reflects slack in the labor market that is not reflected in the official unemployment statistics. We are in good company. Chair Yellen reiterated this view in her prepared remarks during Wednesday’s Federal Open Market Committee press conference. The good news is that the stock of PTER workers has declined by around 900,000 during the last year compared with a decline of fewer than 200,000 in 2013. Moreover, the CPS data suggest the decline is primarily because these workers have either found full-time work or are no longer wanting full-time work (that is, are working part-time for noneconomic reasons), and not because they have become unemployed or have joined the ranks of the discouraged outside of the formal labor market. Even better news is that the recent decline has been very broad based (see the charts).
Up until about a year ago, the overall decline in the number of PTER workers was driven primarily by those in middle-skill occupations in goods-producing industries and, to a lesser extent, in services-producing industries. But during 2014, the decline is also evident in services-producing industries among PTER workers in both low- and high-skill occupations—two categories that had not seen any material decline in their PTER ranks since the end of the recession. (A previous macroblog post discussed the various occupational skill categories.) There is still a ways to go, but these developments are very encouraging.
December 04, 2014
The Long and Short of Falling Energy Prices
Earlier this week, The Wall Street Journal asked the $1.36 trillion question: Lower Gas Prices: How Big A Boost for the Economy?
We will take that as a stand-in for the more general question of how much the U.S. economy stands to gain from a drop in energy prices more generally. (The "$1.36 trillion" refers to an estimate of energy spending by the U.S. population in 2012.)
It's nice to be contemplating a question that amounts to pondering just how good a good situation can get. But, as the Journal blog item suggests, the rising profile of the United States as an energy producer is making the answer to this question more complicated than usual.
The data shown in chart 1 got our attention:
As a fraction of total investment on nonresidential structures, spending on mining exploration, shafts, and wells has been running near its 50-year high over the course of the current recovery. As a fraction of total business investment in equipment and structures, the current contribution of the mining and oil sector is higher than any time since the early 1980s (and generally much higher than most periods during the last half century).
In a recent paper, economists Soren Andersen, Ryan Kellogg, and Stephen Salant explain why this matters:
We show that crude oil production from existing wells in Texas does not respond to current or expected future oil prices... In contrast, the drilling of new wells exhibits a strong price response...
In short, the investment piece really matters.
We've done our own statistical investigations, asking the following question: What is the estimated impact of energy price shocks in the second half of this year on investment, consumer spending, and gross domestic product (GDP)?
If you are interested, you can find the details of the statistical model here. But here is the bottom line: the estimated impact of energy price shocks is a very sizeable decline in investment in the mining and oil subsector relative to baseline and, more importantly, an extended period of flat to slightly negative growth in overall investment relative to baseline (see chart 2).
In our simulations, the "baseline" is the scenario without the ex-post energy price shocks occurring in the third and fourth quarters of 2014, while the "alternative" scenario incorporates the (estimated) actual energy price shocks that have occurred in the second half of this year. These shocks lead to a cumulative 8 percent drop in consumer energy prices and a 6 percent drop in producer energy prices by the fourth quarter of this year relative to baseline. By the fourth quarter of 2017, 2 percentage points of these respective energy price declines are reversed. In chart 2 above, each colored line represents the percentage point difference between the "alternative" scenario and the "baseline" scenario.
As for consumption and GDP? Like overall investment, there is a short-run drag before the longer-term boom, as chart 3 shows:
So is the recent decline in energy prices good news for the U.S. economy? Right now our answer is yes, probably—but we may have to be patient.
Note: We have updated this post since it was originally released, clarifying a sentence in the paragraph above chart 2 and providing the data for the charts. The original sentence stated: But here is the bottom line: the estimated impact of energy price shocks is a very sizeable decline in investment in the mining and oil subsector and, more importantly, an extended period of flat to slightly negative growth in overall investment (see chart 2).
November 24, 2014
And the Winner Is...Full-Time Jobs!
Each month, the U.S. Census Bureau for the U.S. Bureau of Labor Statistics (BLS) surveys about 60,000 households and asks people 15 years and older whether they are employed and, if so, if they are working full-time or part-time. The BLS defines full-time employment as working at least 35 hours per week. This survey, referred to as both the Current Population Survey and the Household Survey, is what produces the monthly unemployment rate, labor force participation rate, and other statistics related to activities and characteristics of the U.S. population.
For many months after the official end of the Great Recession in June 2009, the Household Survey produced less-than-happy news about the labor market. The unemployment rate didn't start to decline until October 2009, and nonfarm payroll job growth didn't emerge confidently from negative territory until October 2010. Now that the unemployment rate has fallen to 5.8 percent—much faster than most would have expected even a year ago—the attention has turned to the quality, rather than quantity, of jobs. This scrutiny is driven by a stubbornly high rate of people employed part-time "for economic reasons" (PTER). These are folks who are working part-time but would like a full-time job. Several of my colleagues here at the Atlanta Fed have looked at this phenomenon from many angles (here, here, here, here, and here).
The elevated share of PTER has left some to conclude that, yes, the economy is creating a significant number of jobs (an average of more than 228,000 nonfarm payroll jobs each month in 2014), but these are low-quality, part-time jobs. Several headlines have popped up over the past year or so claiming that "...most new jobs have been part-time since Obamacare became law," "Most 2013 job growth is in part-time work," "75 Percent Of Jobs Created This Year  Were Part-Time," "Part-time jobs account for 97% of 2013 job growth," and as recently as July of this year, "...Jobs Report Is Great for Part-time Workers, Not So Much for Full-Time."
However, a more careful look at the postrecession data illustrates that since October 2010, with the exception of four months (November 2010 and May–July 2011), the growth in the number of people employed full-time has dominated growth in the number of people employed part-time. Of the additional 8.2 million people employed since October 2010, 7.8 million (95 percent) are employed full-time (see the charts).
The pair of charts illustrates the contribution of the growth in part-time and full-time jobs to the year-over-year change in total employment between January 2000 and October 2014. By zooming in, we can see the same thing from October 2010 (when payroll job growth entered consistently positive territory) to October 2014. Job growth from one month to the next, even using seasonally adjusted data, is very volatile.
To get a better idea of the underlying stable trends in the data, it is useful to compare outcomes in the same month from one year to the next, which is the comparison that the charts make. The black line depicts the change in the number of people employed each month compared to the number employed in the same month the previous year. The green bars show the change in the number of full-time employed, and the purple bars show the change in the number of part-time employed.
During the Great Recession (until about October 2010), the growth in part-time employment clearly exceeded growth in full-time employment, which was deep in negative territory. The current high level of PTER employment is likely to reflect this extended period of time in which growth in part-time employment exceeded that of full-time employment. But in every month since August 2011, the increase in the number of full-time employed from the year before has far exceeded the increase in the number of part-time employed. This phenomenon includes all of the months of 2013, in spite of what some of the headlines above would have you believe.
So, in the post-Great Recession era, the growth in full-employment is, without a doubt, way out ahead.
Author's note: The data used in this post, which are the same data used to generate the headlines linked above, reflect either full-time or part-time employment (total hours of work at least or less than 35 per week, respectively). They do not necessarily reflect employment in a single job.
November 20, 2014
For Middle-Skill Occupations, Where Have All the Workers Gone?
Considerable discussion in recent years has concerned the “hollowing out of the middle class.” Part of that story revolves around the loss of the types of jobs that traditionally have been the core of the U.S. economy: so-called middle-skill jobs.
These jobs, based on the methodology of David Autor, consist of office and administrative occupations; sales jobs; operators, fabricators, and laborers; and production, craft, and repair personnel (many of whom work in the manufacturing industry). In this post, we don't examine why the decline in middle-skill jobs has occurred, just how those workers have weathered the most recent recession. But our Atlanta Fed colleague Federico Mandelman offers an explanation of why this has occurred.
So how have workers in middle-skill occupations fared during the last recession and recovery? Let's examine a few facts from the Current Population Survey from the U.S. Bureau of Labor Statistics.
Only employment in middle-skill occupations remains below prerecession levels
Chart 1 shows employment levels by skill category (using 12-month moving averages to smooth out the seasonal variation). From the end of 2007 to the end of 2009, the overall number of people working declined by more than 8 million. Middle-skill jobs were hit the hardest, declining about 10 percent from 2007 to 2009. As of September 2014, the level was still about 9 percent below the 2007 level. In contrast, employment in low-skill occupations is 7 percent above prerecession levels, and employment in high-skill occupations is about 8 percent higher than before the recession.
For full-time workers (working at least 35 hours a week at all jobs) the decline in middle-skill occupations is even more dramatic. From 2007 to 2009, the number of full-time workers whose main job was a middle-skill occupation fell more than 15 percent from 2007 to 2009 and is still about 11 percent below the level at the end of 2007.
Those in middle-skilled occupations were most likely to become unemployed
In the 2001 recession, the chances of being unemployed after one year were similar for those working full-time in middle- and low-skill occupations. During the most recent recession, the likelihood of becoming unemployed rose sharply for everyone, but much more sharply for those working in middle-skill occupations. At the recession's trough, almost 6 percent of people who were employed in middle-skill occupations one year earlier were unemployed, compared with about 3 percent of workers in high-skill occupations and 3.5 percent of workers in lower-skill occupations (see chart 2).
Underemployment has improved only slowly at all skill levels
The share of people who are working part-time involuntarily about doubled for workers in low-, middle-, and high-skill occupations. For middle-skill occupations, the share rose from around 1.7 percent to 4.3 percent and is currently around 2.4 percent. For low-skill occupations, involuntary part-time employment increased from 2.4 percent to 5 percent and was still 3.8 percent as of September 2014. And for those in high-skill occupations, the chances of becoming involuntarily part-time rose from 0.8 percent to 1.8 percent and are now back to about 1 percent (see chart 3).
Ready for some good news?
Those who held middle-skill jobs are more likely to obtain high-skill jobs than before the recession
Currently, of those in middle-skill occupations who remain in a full-time job, about 83 percent are still working in a middle-skill job one year later (see chart 4). What types of jobs are the other 17 percent getting? Mostly high-skill jobs; and that transition rate has been rising. The percent going from a middle-skill job to a high-skill job is close to 13 percent: up about 1 percent relative to before the recession. The percent transitioning into low-skill positions is lower: about 3.4 percent, up about 0.3 percentage point compared to before the recession. This transition to a high-skill occupation tends to translate to an average wage increase of about 27 percent (compared to those who stayed in middle-skill jobs). In contrast, those who transition into lower-skill occupations earned an average of around 24 percent less.
In summary, the number of middle-skill jobs declined substantially during the last recession, and that decline has been persistent—especially for full-time workers. Many of the workers leaving full-time, middle-skill jobs became unemployed, and some of that decline is the result of an increase in part-time employment. But others gained full-time work in other types of occupations. In particular, they are more likely than in the past to transition to higher-skill occupations. Further, the transition rate to high-skill occupations has gradually risen and doesn't appear directly tied to the last recession.
Authors' note: The middle-skill category of jobs consists of office and administrative occupations; sales; operators, fabricators, and laborers; and production, craft, and repair personnel. The other two broad categories of occupations are labeled high-skill and low-skill. High-skill occupations consist of managers, technicians, and professionals. Low-skill occupations are defined as those involving food preparation, building and grounds cleaning, personal care and personal services, and protective services.
November 13, 2014
A Closer Look at Employment and Social Insurance
The Atlanta Fed's Center for Human Capital Studies hosted its annual employment conference on October 2–3, 2014, organized once again by Richard Rogerson of Princeton University, Robert Shimer of the University of Chicago, and the Atlanta Fed's Melinda Pitts. This macroblog post summarizes some of the discussions.
Social insurance programs in the United States and other developed countries represent a large and growing share of expenditures relative to gross domestic product (GDP). Assessing the costs and benefits of the diverse programs that make up the U.S. social insurance system is a key input into the design and implementation of effective programs. This conference featured seven papers that dealt with various aspects of this assessment. Although each program is designed to address specific issues and hence needs to be studied in the context of those issues, many of the same basic economic questions arise in each context. For example, what is the rationale for social insurance programs? Do they address inefficiencies, or are they mainly designed to redistribute from one group to another? Who benefits from specific programs? How do programs designed to achieve specific objectives distort economic outcomes? These are the questions that featured prominently in the conference.
A classic question in economics concerns the extent to which markets cannot achieve efficient outcomes without government intervention. It is well known that the so-called "invisible hand" can achieve efficient outcomes in a wide range of standard settings, but do these results extend to situations in which information asymmetries exist? In 1976, Michael Rothschild and Joseph Stiglitz's article "Equilibrium in Competitive Insurance Markets" suggested that in the presence of certain kinds of private information, insurance markets could not achieve efficient allocations. In fact, they argued that competitive equilibrium might not even exist in these settings. In "Adverse Selection Is Not a Justification for Social Insurance," Ed Prescott challenges this result and shows that competitive equilibrium exists and achieves efficient allocations in settings that include information problems such as Rothschild and Stiglitz's adverse selection problem. Key to this result is the presence of mutual insurance companies, and how this presence influences the contracts offered by insurance companies in equilibrium. In the Rothschild and Stiglitz environment, insurance companies were effectively agents with deep pockets that were outside the model.
Providing insurance to individuals in situations in which they face bad outcomes may distort individual behavior and lead to negative outcomes that outweigh the benefits of the insurance. This basic issue was addressed by three of the papers at the conference in three separate contexts. Jason Abaluck, Jonathan Gruber, and Ashley Swanson examined how prescription drug coverage through Medicare influences prescription drug usage; Hamish Low and Luigi Pistaferri studied the disability insurance (DI) system; and Bradley Heim, Ithai Lurie, and Kosali Simon examined whether the extension of health benefits to young adults as mandated by the Affordable Care Act (ACA) influenced the behavior of young adults.
In "Prescription Drug Use Under Medicare Part D: A Linear Model of Non-linear Budget Sets," Jason Abaluck, Jonathan Gruber, and Ashley Swanson study how prescription drug use responds to price changes associated with social insurance through Medicare. At the conference, Gruber discussed one key objective of their analysis: uncovering the elasticity of prescription drug use with respect to price. A large elasticity implies that providing insurance in the form of lower prices will distort behavior and lead to much higher drug use, and some recent papers have argued that this elasticity may be quite large. Their basic strategy is to study how changes in the details of Medicare coverage over time influenced individual choices. A novel feature of the estimation strategy is to take advantage of the fact that the marginal price people face depends on their overall annual expenditure on prescriptions, so that individuals can be sorted into groups based on histories of usage, interacted with changes in the details of coverage. A first key finding of this paper is that the elasticity is relatively small. A second key set of findings concerns the extent to which individual choices (in terms of plan selection and yearly expenditure conditional on plan choice) reflect departures from rationality, such as myopia or salience. The paper finds an important role for both of these effects.
Disability insurance (DI) represents a clear and classic example of the tension between insurance provision and insurance. While one would like to provide insurance to individuals who are unable to work, it can be difficult to assess the true ability of an individual to work, thereby creating the opportunity for people who are not disabled to also collect. Luigi Pistaferri addressed this issue in the paper he coauthored with Hamish Low, "Disability Insurance and the Dynamics of the Incentive-Insurance Tradeoff." This paper builds and estimates a structural model that incorporates labor supply, health shocks, earnings shocks, and the key details of the DI application process. The authors conduct various counterfactuals and assess the tension between insurance and incentives in the context of the U.S. DI program. Several results emerge. First, making the review process less strict would enhance welfare despite worsening incentives for people to misreport their health status. This is because the current system denies too many truly disabled individuals from collecting. But decreasing generosity would also increase overall welfare by decreasing the incentives for false collection.
One of the first measures of the Affordable Care Act (ACA) to be enacted was the provision that allowed dependent individuals to remain covered by their parents' healthcare plans until the age of 26. The paper by Bradley Heim, Ithai Lurie, and Kosali Simon, "The Impact of the Affordable Care Act Young Adult Mandate: Evidence from Tax Data," aims to assess the extent to which this provision has affected outcomes for young adults in terms of employment, wages, schooling, and marriage. As Simon described it at the conference, the novel aspect of this analysis is that it tracks outcomes using administrative IRS data, which affords a large sample size. The main empirical strategy is to compare the change in outcomes from before and after the provision was enacted for individuals below the age threshold with the change in outcomes for individuals just above the age threshold. The paper also reports estimates based on triple differencing that uses information on parental health insurance status. The main message from the analysis is that one cannot find robust, statistically significant effects of this ACA provision on outcomes for young individuals. One important qualification is that despite the large sample size, standard errors are still quite large, so that the analysis cannot rule out the possibility of economically significant effects.
Naoki Aizawa and Hanming Fang also considered the effects of the ACA in their paper "Equilibrium Labor Market Search and Health Insurance Reform." However, in contrast to the above papers that focus on how a particular program feature might influence individual choices, this paper focuses on how the creation of health insurance exchanges and the individual insurance mandate would affect the overall equilibrium in the labor market, taking into account the firms' decisions on whether to offer insurance and the wages that they offer to workers. In his presentation, Fang discussed building a structural equilibrium model of the labor market and estimating it using a variety of data sets. The authors find that the ACA will reduce the uninsured rate from about 20 percent to about 7 percent. But interestingly, the paper finds that the uninsured rate would drop even further if the employer mandate were dropped from the ACA. General equilibrium responses are key to understanding this result, illustrating the importance of studying these effects.
One of the rapidly growing social insurance programs is Medicaid. Mariacristina De Nardi, Eric French, and John Bailey Jones assess the benefits of this program in their paper "Medicaid Insurance in Old Age." As French described at the conference, this paper uses a structural approach to assess the extent to which households with different income and health status benefit from Medicaid. The analysis focuses on individuals from age 70 and forward using data from the Health and Retirement Study, emphasizing the risks that individuals face as a result of health shocks. Medicaid offers partial insurance against these shocks, particularly the large expenditures associated with nursing home care, and the paper assesses the value of this insurance for individuals in different positions in the wealth distribution at age 70. The paper has two main findings. First, the insurance value of Medicaid is substantial, and decreasing the size of the program would entail large welfare costs in excess of one dollar for every dollar of reduced spending. Second, expanding the size of the program would offer significant insurance value only to wealthy households. The authors conclude that in terms of managing the risks of the elderly, the current scope of Medicaid seems appropriate.
As the above discussion emphasizes, a critical input into the design and assessment of social insurance programs are data that allow us to reliably document the outcomes and groups that the insurance program wishes to help, as well as measure the efficacy of existing programs in achieving desirable outcomes. In the paper "Welfare Programs and Survey Misreporting: Implications for Income, Poverty and Disconnectedness," Bruce Meyer and Nikolas Mittag documented the serious shortcomings of several standard publicly available data sets when it comes to measuring the resources available to the poorer segments of the population. Meyer presented the paper at the conference, and it uses administrative data from New York State that allow them to link income and transfer data, both cash and in-kind, and compare the measures obtained using these administrative data with the measures obtained using data from the Current Population Survey (CPS), which is a standard source for publicly available data on the income distribution. The results are striking. Relative to analysis based on data from the CPS, analysis using administrative data shows better outcomes in terms of inequality and disconnectedness and yield larger effects from existing programs in terms of their ability to affect these outcomes.
Full papers or presentations for most of these papers are available on the Atlanta Fed's website.
By Melinda Pitts, director of the Atlanta Fed's Center for Human Capital Studies, Richard Rogerson of Princeton University, and Robert Shimer of the University of Chicago
- Sales Flexing Muscle at More Firms
- All Eyes on the Consumer
- Signs of Strengthening Wage Growth?
- What the Weather Wrought
- Déjà Vu All Over Again
- Is Measurement Error a Likely Explanation for the Lack of Productivity Growth in 2014?
- What Seems to Be Holding Back Labor Productivity Growth, and Why It Matters
- Signs of Improvement in Prime-Age Labor Force Participation
- Could Reduced Drilling Also Reduce GDP Growth?
- Are Shifts in Industry Composition Holding Back Wage Growth?
- May 2015
- April 2015
- March 2015
- February 2015
- January 2015
- December 2014
- November 2014
- October 2014
- September 2014
- August 2014
- Business Cycles
- Business Inflation Expectations
- Capital and Investment
- Capital Markets
- Data Releases
- Economic conditions
- Economic Growth and Development
- Exchange Rates and the Dollar
- Fed Funds Futures
- Federal Debt and Deficits
- Federal Reserve and Monetary Policy
- Financial System
- Fiscal Policy
- Health Care
- Inflation Expectations
- Interest Rates
- Labor Markets
- Latin America/South America
- Monetary Policy
- Money Markets
- Real Estate
- Saving, Capital, and Investment
- Small Business
- Social Security
- This, That, and the Other
- Trade Deficit