About


The Atlanta Fed's macroblog provides commentary and analysis on economic topics including monetary policy, macroeconomic developments, inflation, labor economics, and financial issues.

Authors for macroblog are Dave Altig, John Robertson, and other Atlanta Fed economists and researchers.


February 06, 2017


Examining Changes in Labor Force Participation

The Labor Department announced on Friday that January's unemployment rate was 4.8 percent, only 10 basis points below the level in January 2016. You can be forgiven if looking at a graph of the unemployment rate since 2007 makes you think of a roller coaster, because it showed a very steep climb, followed by a swift decline. From a distance, it may seem like the car's descent stopped about a year ago and has merely been bumping around a bit as it approaches the elevation of the platform.

But the unemployment rate alone does not fully account for improvement in the labor market. During the past three years, the labor force participation (LFP) rate has become a particularly important metric to look at. The overall share of the population that is working or actively seeking work has been essentially flat during this period, which is striking because there is a powerful demographic trend—an aging population—that is pulling it down with tremendous force.

Many factors are behind LFP's relative flatness, some of which undoubtedly relate to the labor market's strength. The opportunities available in the labor market affect an individual's decision to enter or leave the labor force. For example, it can affect when a person chooses to retire, enroll in college, apply for disability insurance, or stay home to care for family instead of looking for employment.

On a quarterly basis we update our web page with analysis of how these reasons for not being in the labor market have changed during the past year, and we also look at the extent to which these changes affect the overall LFP rate. Between the fourth quarter of 2015 and the same period in 2016, the LFP rate rose 0.14 percentage points (not seasonally adjusted). The chart below breaks out this increase and shows how much the various reasons for nonparticipation account for the increase (holding the age composition of the population fixed) versus the downward pressure exerted by an aging population.

Let's briefly look at the relative contributions to the change in labor force participation in more detail:

Aging of the population: During the last year, the aging population was the only significant factor continuing to depress the LFP rate. In line with this factor's contribution from previous years, it accounted for about 0.15 percentage points of the decline in the LFP rate.

Retirement: Retirement rates ticked down over the year, resuming a trend that had stalled in the past few years. Later retirement was the largest influence on LFP in the past year and completely offset the effect of aging population, boosting the rate by 0.15 points.

Shadow labor force: The share of the population not technically counted as "unemployed" because they are not actively searching but say they want a job fell slightly over the past year. This decline boosted the LFP rate by 0.04 percentage points. (A decline in this category is usually associated with a strengthening labor market.)

Health problems: The share of the population who said they are too chronically ill or disabled to work declined for the second year in a row, reversing the trend of the prior eight years. This decline put upward pressure on LFP (0.04 percentage points) and could partly be a reflection of a stronger job market with more opportunities for those with disabilities (see this report  from the U.S. Bureau of Labor Statistics for more information).

Rising education: The share of the population not in the labor market because they are in school increased slightly, lowering the LFP rate by 0.03 percentage points. School enrollments rates rose for decades and accelerated during the last recession. The small contribution of schooling to the change in the LFP rate during the past year likely brings it closer to alignment with the long-term trend.

Family responsibilities: The share of the population not participating in the labor force because of family responsibilities declined during the last year, boosting the LFP rate by 0.13 percentage points.

An interactive chart on our website allows users to choose their own time period for comparison for all those 16 years old and above, those 25–54 years old, as well as for men and women separately. You can see how various factors have contributed to that roller coaster effect—strap yourself in!

February 6, 2017 in Employment , Labor Markets | Permalink | Comments ( 1)

January 23, 2017


Wage Growth Tracker: Every Which Way (and Up)

As measured by the Atlanta Fed's Wage Growth Tracker, the typical wage increase of a U.S. worker averaged 3.5 percent in 2016. This is up from 3.1 percent in 2015 and almost twice the low of 1.8 percent recorded in 2010. As noted in previous macroblog posts, the Wage Growth Tracker correlates tightly to the unemployment rate. As median wage growth has risen, the unemployment rate declined from an average of 9.6 percent in 2010, to 5.3 percent in 2015, and to 4.8 percent in 2016.

What does this correlation suggest about the Wage Growth Tracker in 2017? Let's start with a forecast of unemployment. Based on the latest Summary of Economic Projections, the central view of Federal Open Market Committee participants is that the unemployment rate will end this year at around 4.5 percent, about 30 basis points below the median participant's estimate of the unemployment rate that is sustainable over the longer run.

With a modest further decline in the unemployment rate, other things equal, we might then also expect to see a modest uptick in the Wage Growth Tracker in 2017. But I think the emphasis here should be on the word modest. Speaking for myself, sustained Wage Growth Tracker readings much above 4 percent in 2017 would begin to worry me, especially without a compensating pickup in the growth of labor productivity, which has been stuck in the 0 to 1 percent range in recent years. Significantly higher wage growth—reflecting a tightening labor market more than larger gains in worker productivity—could make the inflation outlook a bit less sanguine than we currently think. (This macroblog post discussed the connection among productivity growth, wage growth, and inflation.)

Thus far, many firms appear to have been able to keep their labor costs relatively low by replacing or expanding staff with lower-paid workers. (Our colleagues at the San Francisco Fed have written about how changes in the composition of workers can mute changes in total labor costs.) However, it's not clear how long that approach can be sustained. Indeed, it's noteworthy that average wage costs appear to have accelerated recently. For instance, U.S. Bureau of Labor Statistics data  indicate that average hourly earnings in the private sector increased over the year by 2.9 percent in December—the fastest pace since 2009.

We haven't been hearing reports from firms where the typical worker's wage increase in 2017 is expected to be above 4 percent. However, we did get readings for the Wage Growth Tracker pretty close to 4 percent in October and November of last year. As the following chart shows, a sharp increase in women's median wage growth (hitting 4.3 percent in October 2016) drove the overall increase. In contrast, the median wage increase for men was 3.5 percent.

The jump in the relative wage growth of women came as a bit of a surprise. Female wage growth had been generally running below that of men since 2010, and analysis by my colleague Ellie Terry showed that gender-specific factors that are unlikely to change very rapidly explain a fair amount of that lag. Therefore, we suspected that the divergence in wage growth might not be sustainable—a suspicion that proved to be true. Median wage growth for women slowed to 3.5 percent in December, the same growth rate men saw.

Readers who can't get enough Wage Growth Tracker data will be delighted to note that in 2017 we plan on making further enhancements to the tool. These enhancements will include finer cuts by age, education, industry, and hours worked, as well as new cuts by occupation, race, and location. You can stay informed on all Wage Growth Tracker updates by subscribing to our RSS feed  or email updates .

January 23, 2017 in Employment , Labor Markets , Wage Growth | Permalink | Comments ( 0)

January 03, 2017


Following the Overseas Money

​​

Though the holiday season has come to a close, the forthcoming policy season may bring with it serious debate about a holiday of a different sort: a tax "holiday" that would allow corporations to repatriate accumulated profits currently held overseas.

As with many of the policy proposals that the new Congress and administration will consider, our primary interest here at the Atlanta Fed is to assess how the policy, if enacted, will likely affect our own economic forecasts and the environment in which future monetary policy will be made.

The best starting point is usually to just determine the facts as we know them. In this case, the question is what we know about the nature of the foreign earnings of U.S. corporations.

U.S. corporations' undistributed foreign earnings have been accumulating rapidly for more than a decade, as companies have expanded their foreign operations. The income earned by U.S. domestic corporations' foreign subsidiaries is generally not subject to U.S. tax until the income is distributed to the parent corporation in the United States. According to a November 25, 2016, Wall Street Journal article, over the past decade total undistributed foreign earnings of U.S. companies have risen from about $500 billion to more than $2.5 trillion, a sum equal to nearly 14 percent of U.S. gross domestic product.

Though it is not uncommon to refer to these sums as "a pile of cash," this sort of terminology is perhaps a bit misleading. For one, some of that "pile of cash" is not cash at all. According to a report from the Joint Committee on Taxation , "the undistributed earnings may include more than just cash holdings as corporations may have reinvested their earnings in their business operations, such as by building or improving a factory, by purchasing equipment, or by making expenditures on research and experimentation."

More important, the portion of foreign earnings that hasn't been invested in business operations is not necessarily "trapped" or "stashed" overseas. In fact, much of it is in the United States, already working (albeit while untaxed) for the U.S. economy.

U.S. companies do not routinely disclose what their foreign subsidiaries do with undistributed earnings. To better understand the situation, in 2011 the Senate Permanent Subcommittee on Investigations conducted a survey of 27 large U.S. multinationals. Survey results showed that those companies' foreign subsidiaries held nearly half of their earnings in U.S. dollars, including U.S. bank deposits and Treasury and corporate securities (see the table).

A couple of years later, a June 13, 2013, Wall Street Journal report also found that Google, EMC, and Microsoft kept more than three-quarters of their foreign subsidiaries' cash in U.S. dollars or dollar-denominated securities.

So it turns out, then, that a large fraction of undistributed foreign profits is held at U.S. banks or invested in U.S. securities. Even dollar deposits held by U.S. companies in tax havens such as Ireland, the Cayman Islands, and Singapore ultimately live here in the United States because foreign banks typically hold their dollar deposits in so-called correspondent banks in the United States.

In fact, U.S. dollar balances always stay in the United States, even if they are controlled from outside the country. Those dollars in turn are available to be lent out to U.S. businesses. And when U.S. companies' foreign subsidiaries invest their cash holdings in U.S. Treasury bonds, they are in effect lending to the U.S. government.

Foreign subsidiaries of U.S. companies choose to invest their profits in dollar-denominated assets for much the same reasons that make the U.S. dollar an international reserve currency:

  • the dollar maintains its value in terms of goods and services (the dollar is a global unit of account);
  • U.S. financial markets are deep and liquid, providing ample investment choices; and
  • U.S. government obligations are considered virtually risk-free, making them a safe haven during times of global stress and risk aversion.

Companies also have operational reasons for keeping surplus cash in U.S. dollars. Most of the international trade invoicing is done in dollars, so U.S. companies' foreign subsidiaries hold dollars to pay suppliers and deal with customers. Also, nonfinancial companies prefer to avoid foreign exchange risk and volatility. Finally, holding most of the funds, which are not invested in foreign operations, in dollars mitigates potential accounting losses, since U.S. companies are required to report in dollars on their consolidated financial statements.

None of this is to say that a tax holiday for U.S. corporations on undistributed foreign profits is a good or bad policy choice. But even without passing judgment, it may fall to macroeconomic forecasters to estimate the policy impact on business investment, job growth, and the like. Understanding the facts underlying the targeted funds is a reasonable starting point for answering the harder questions that may come.

January 3, 2017 in Fiscal Policy | Permalink | Comments ( 4)

December 16, 2016


The Impact of Extraordinary Policy on Interest and Foreign Exchange Rates

Central banks in the developed countries have adopted a variety of extraordinary measures since the financial crisis, including large-scale asset purchases and very low (and in some cases negative) policy rates in an effort to boost economic activity. The Atlanta Fed recently hosted a workshop titled "The Impact of Extraordinary Monetary Policy on the Financial Sector," which discussed these measures. This macroblog post discusses the highlights of three papers related to the impact of such policy on interest rates and foreign exchange rates. A companion Notes from the Vault reviews papers that examined how those policies may have affected financial institutions, including their lending.

Prior to the crisis, central banks targeted short-term interest rates as a way of influencing the rest of the yield curve, which in turn affected aggregate demand. However, as short-term rates approached zero, central banks' ability to further cut their target rate diminished. As a substitute, the central banks of many developed countries (including the Federal Reserve, the European Central Bank, and the Bank of Japan) began to undertake large-scale purchases of bonds in an attempt to influence longer-term rates.

Central bank asset purchases appear to have had some beneficial effect, but exactly how these purchases influenced rates has remained an open question. One of the leading hypotheses is that the purchases did not have any direct effect, but rather served as a signal that the central bank was committed to maintaining very low short-term rates for an extended period. A second hypothesis is that central bank purchases of longer-dated obligations resulted in long-term investors bidding up the price of remaining longer-maturity government and private debt.

The second hypothesis was tested in a paper  by Federal Reserve Board economists Jeffrey Huther, Jane Ihrig, Elizabeth Klee, Alexander Boote and Richard Sambasivam. Their starting point was the view that a "neutral" policy would have the Fed's System Open Market Account (SOMA) closely match the distribution of the stock of outstanding Treasury securities. In their statistical tests, they find support for the hypothesis that deviations from this neutrality should influence market rates. In particular, they find that the term premium in longer-term rates declines significantly as the duration of the SOMA portfolio grows relative to that of the stock of outstanding Treasury debt.

The central banks' large-scale asset purchases not only took longer-dated assets out of the economy, but they also forced banks to increase their holdings of reserves. Large central banks now pay interest on reserves (or in some cases charge interest on reserve holdings) at an overnight rate that the central bank can change at any time. As a result, these purchases can significantly reduce the average duration (or maturity) of a bank's portfolio below what the banks found optimal given the term structure that existed prior to the purchases. Jens H. E. Christensen from the Federal Reserve Bank of San Francisco and Signe Krogstrup from the International Monetary Fund have a paper  in which they hypothesize that banks respond to this shortening of duration by bidding up the price of longer-dated securities (thereby reducing their yield) to restore optimality.

The difficulty with testing Christensen and Krogstrup's hypothesis is that in most cases central banks were expanding bank reserves by buying longer-dated securities, thus making it difficult to disentangle their respective effects. However, in 2011 the Swiss National Bank undertook a series of three policy moves designed to produce a large, rapid increase in bank reserves. Importantly, these moves were an attempt to counter perceived overvaluation of the Swiss franc and did not involve the purchase of longer-dated bonds. In a follow-up empirical paper , Christensen and Krogstrup exploit this unique policy setting to test whether Swiss bond rates declined in response to the increase in reserves. They find that the third and largest of these increases in reserves was associated with a statistically and economically significant fall in term premia, implying that the increase did lower longer-term rates.

Although developed countries' monetary policy has focused on their domestic economies, these policies can have significant spillovers into emerging countries. Large changes in the rates of return available in developed countries can lead investors to shift funds into and out of emerging countries, causing potentially undesirable large swings in the foreign exchange rate of these emerging countries. Developing countries' central banks may try to counteract these swings via intervention in the foreign exchange market, but the effectiveness of sterilized intervention is the subject of some debate. (Sterilized intervention occurs when the central bank buys or sells foreign currency, but then takes offsetting measures to prevent these from changing bank reserves.)

Once again, determining whether exchange rates are influenced and, if so, by what mechanism can be econometrically difficult. Marcos Chamon from the International Monetary Fund, Márcio Garcia from PUC-Rio, and Laura Souza from Itaú Unibanco examine the efforts of the Brazilian Central Bank to stabilize the Brazilian real in the aftermath of the so-called "taper tantrum." The taper tantrum is the name given to the sharp jump in U.S. bond yields and the foreign exchange rate value of the U.S. dollar after the May 23, 2013, statement by Board Chair Ben Bernanke that the Federal Reserve would slow (or taper) the rate at which it was purchasing Treasury bonds (see a brief essay by Christopher J. Neely). Chamon, Garcia, and Souza's paper  takes advantage of the fact that Brazil preannounced its intervention policy, which allows them to separate the impact of the announcement to intervene from the intervention itself. They find that the Brazilian Central Bank's intervention was effective in strengthening the value of the real relative to a basket of comparable currencies.

All three of the studies faced the difficult challenge in linking specific central bank actions to policy outcomes, and each tackled the challenge in innovative ways. The evidence provided by the studies suggests that central banks can use extraordinary policies to influence interest and foreign exchange rates.

December 16, 2016 in Exchange Rates and the Dollar , Interest Rates , Monetary Policy | Permalink | Comments ( 2)

December 05, 2016


Using Judgment in Forecasting: Does It Matter?

​​

Many professional forecasters use statistical models when making their near-term projections for real gross domestic product (GDP) growth. A 2013 special survey on the forecasting methods of the Survey of Professional Forecasters  found that 18 out of 21 respondents featured a statistical model prominently in their current-quarter economic projections. Nevertheless, there is fairly compelling evidence that many professional forecasters incorporate judgment in their forecasts of the first estimate of real GDP growth for a quarter—even when much of the source data used to construct the GDP estimate are available.

In the October 2016 Wall Street Journal Economic Forecasting Survey (WSJ), the most common panelist projection for annualized third-quarter real GDP growth was 2.5 percent, and the second most common one was 3.0 percent. The first digit after the decimal point, or tenths digit, of these two numbers are "5" and "0." Of the 58 individual forecasts of third-quarter growth in the survey, 21 had a tenths digit of "0" or "5," a total that is almost twice as large as we would expect if all tenths digits were equally likely to be submitted.

This pattern isn't unique to the most recent quarter's GDP forecast. The following chart shows the historical frequency of the tenths digit in past WSJ surveys for first estimates of real GDP growth over the period from the first quarter of 2003 to the third quarter of 2016, made about three weeks before the release.

Almost 40 percent of these 2,390 forecasts have a tenths digit of "0" or "5." In contrast, the historical distribution of published first estimates of real GDP growth from the fourth quarter of 1991 to the third quarter of 2016 and real gross national product (the most common measure of U.S. production in an earlier era) growth from the third quarter of 1965 to the third quarter of 1991 has a tenths digit of either "0" or "5" only 18 percent of the time. The historical Atlanta Fed's GDPNow forecasts have a "0" or a "5" tenths digit only 15 percent of the time.

More formally, one easily can reject the hypothesis at the 1 percent significance level that the tenths digit of the WSJ panelist forecasts are either uniformly distributed or follow the Benford distribution for tenths digits after rounding to the nearest tenth (see this paper by economists Stefan Gunnel and Karl-Heinz Todter, who found similar relative frequencies of "0s" and "5s" in professional forecasts of German GDP growth and consumer price index inflation).

If we assume that near-term GDP growth forecasts with a tenths digit of "0" or "5" typically involve more judgment than forecasts with another tenths digit, a natural question is whether these more judgmental forecasts are less accurate than others. Of the 2,390 WSJ growth forecasts mentioned above, the ones with a tenths digit of "0" or "5" (after rounding to the nearest tenth) had an average error of 0.786 percentage points without regard to sign, and the others had an average error of 0.743 percentage points. These accuracy metrics are not statistically different at even the 10 percent significance level. Moreover, because of the panel nature of WSJ forecasts, we can measure how often a forecaster has a tenths digit of "0" or "5" (after rounding). Of the 44 panelists who submitted at least 30 three-week-ahead GDP forecasts during the period of the first quarter of 2003 through the third quarter of 2016, the correlation of the panelists "0" or "5" tenth digit frequency and their average error without regard to sign is only 0.13 and not significantly different from 0.

Although at least some professional forecasters appear to make judgmental adjustments to their near-term GDP projections, the evidence presented here does not suggest it comes, on average, at the cost of accuracy.

December 5, 2016 in Forecasts , GDP | Permalink | Comments ( 1)

November 28, 2016


Does Lower Pay Mean Smaller Raises?

​​

I've been asked a few questions about the relative wage growth of low-wage versus high-wage individuals that are measured by the Atlanta Fed's Wage Growth Tracker. Do individuals who were relatively lower (or higher) paid also tend to experience lower (or higher) wage growth? If they do, then wage inequality would increase pretty rapidly as low-wage earners get left further and further behind.

The short answer is no. As chart 1 shows, median wage growth is highest for the workers whose pay was relatively low (in the bottom 25 percent of the wage distribution), and lowest for those who were the highest-paid (in the top 25 percent of the wage distribution). Median wage growth is reasonably similar for those whose pay was in the middle 50 percent of the wage distribution.

To understand what's going on, let's look at the construction of a Wage Growth Tracker sample. In simple terms, a person's wage is observed in one month, and then again 12 months later. But relatively low-wage workers are less likely to remain employed (and hence more likely not to have a wage when observed a second time) than other workers. Almost half of workers who are not employed 12 months later come from the lowest 25 percent of the wage distribution. For workers in a relatively low-wage job, a greater share who might otherwise have experienced a declining wage left their employment, resulting in a larger share of wage increases among those who remained employed.

In contrast, relatively high wage earners in the Wage Growth Tracker sample have a remarkably low median wage growth—zero in recent years. They also have a much greater chance of experiencing a wage decline than other workers (see chart 2).

However, getting a complete picture for high-wage individuals in the Current Population Survey is limited by the fact that observations are top-coded (or censored to preserve identifiable individuals' anonymity). For example, weekly earnings higher than $2,885 are currently simply recorded as $2,885. If a person in this circumstance gets a wage increase, it will still be reported as just $2,885, which would make it seem as if wages didn't increase, even if they did.

Top-coding itself has only a relatively small effect on the median wage growth for the whole sample because top-coded earnings aren't that common. But they are a reasonably large share of the upper part of the wage distribution, which makes the median wage growth pretty unrepresentative for people who were relatively high wage earners. In principle, one could try to surmount this problem by estimating the earnings for top-coded workers, but my experience has been that doing so is likely to add more noise than insight.

What about examining a worker's current wage instead of their prior wage? Is the median wage growth also higher for workers who are currently in the lowest part of the wage distribution? No. In fact, they are more likely than others not to have received a pay raise or even to have had the rate of pay reduced. Conversely, someone who is currently in the upper part of the wage distribution is more likely to have received a larger pay raise than other workers. Some workers move up the wage distribution—but not all.

The bottom line is that the point of reference matters a lot when looking at the tails of the wage distribution, and top-coding limits the ability to learn much about the wage growth of high wage earners. But for the middle part of the wage distribution, it doesn't matter so much. The median wage growth of the overall sample is pretty representative of the typical wage growth experience of workers in the heart the wage distribution.

November 28, 2016 in Employment , Labor Markets , Wage Growth | Permalink | Comments ( 0)

November 22, 2016


Outside Looking In: Why Has Labor Force Participation Increased?

​​

The labor force participation rate (LFPR) is an estimate of the share of the population actively engaged in the labor market. The LFPR has increased about 30 basis points over the past year (from the third quarter of 2015 to the third quarter of 2016)—a modest reversal in the precipitous decline in the LFPR that began in 2008. What accounts for this stabilization and—given the demographic and cyclical forces in play—how much longer can it last?

The following is perspective through the lens of the reasons people give for not participating in the labor force. Perhaps the component most responsive to changes in labor market conditions is what I will refer to as the "shadow labor force," which is made up of people who are not in the official labor force and are not actively seeking employment, but who say they want a job. (This group includes people discouraged over job prospects.) During tough times, the share of the population in the shadows rises, and during good times it falls. In the third quarter of 2016, about 2.3 percent of the population fell into this category—down from a high of 2.8 percent but still a bit above prerecession levels (see the chart).

But focusing solely on the decline in the shadow labor force to explain the recent reversal in the LFPR would be a mistake. In fact, high unemployment in the aftermath of the Great Recession was accompanied not only by a rise in the share of the shadow labor force, but also by an increase in the share of the population who said they didn't currently want a job—because of either a health issue or engagement in some other activity. Although some of this likely reflects trends already at work before the recession, some of it was also probably a cyclical response to weak job opportunities.

The chart below shows how these various factors cumulatively contribute to the decline in the LFPR between the third quarter of 2007 and the third quarter of 2016. It shows that, in addition to a larger share in the shadow labor force, the reasons for the decline between 2007 and 2016 also stemmed from a greater age-adjusted share who were too sick or disabled to work (purple) or in school instead of working (light blue). Interestingly, the share out of the labor force but wanting a job (dark blue) actually exerted the smallest downward force on LFPR of all of these three reasons. The green section represents the impact of the baby boomers: an increasing share of the population of retirement age. Partly offsetting this shift in the age distribution was a decrease in the propensity of these workers to actually retire (orange).

The next chart shows that almost all the nonparticipation factors that had put downward pressure on the LFPR since 2007 have reversed course and contributed positively to an increase in the LFPR during the past year. In particular, there was a decline in the share of the population who cited nonparticipation because of poor health or enrollment in school or were otherwise wanting but not looking for work. This decline in the schooling and illness nonparticipation rates is particularly noteworthy because it stands in contrast to the increasing trends that were in place prior to the recession (to read more, visit our LFP Dynamics page).

The only significant factor continuing to depress the LFPR during the past year is the impact of an increasing share of the population in age groups with relatively low labor force attachment. This factor brings me to the second question I posed earlier in this post: What will this picture look like going forward? Unfortunately, I think the answer is that it's very hard to say.

Other things equal, it seems reasonable to think that the nonparticipation rates attributable to age-adjusted schooling and poor health will eventually revert to the upward trends occurring before the recession, a reversion that will push down the LFPR. Probably the biggest wild card for the future is what will happen with decisions concerning retirement (and hence older individuals' LFPR). The trend toward retiring later in life has risen and fallen a couple of times during the past two decades. The positive role that later retirement has played in mitigating the overall decline in the LFPR in recent years—coupled with the steadily increasing share of the population approaching traditional retirement age—suggests that deferred retirement will be an especially important factor to keep an eye on. This point is nicely illustrated in this piece Adobe PDF file format by our colleagues at the Kansas City Fed, who look at the role of later retirement in reducing the rate of outflow of people from the labor force.

Note: Data on the reasons for not participating are available in the data download section of the Labor Force Dynamics page, which we will update every quarter (data for the third quarter are coming soon).

November 22, 2016 | Permalink | Comments ( 0)

November 15, 2016


Wages Climb Higher, Faster

​​

The Atlanta Fed's Wage Growth Tracker is a three-month average of median growth in the hourly earnings of a sample of wage and salary workers taken from the Current Population Survey. Last month in a macroblog post, I noted that the Wage Growth Tracker reading for September, at 3.6 percent, was close to where it had been hovering since April. However, I also noted that the non-averaged median wage growth for September was at a cyclical high of 4.2 percent, and so it would be interesting to see what the October data revealed. Well, the October data are in, and they do confirm a sizeable uptick in wage growth over the last couple of months. The median wage growth for October was 4.0 percent, which brings the Wage Growth Tracker up to 3.9 percent—a percentage point higher than a year ago, and now the highest level since November 2008.

In addition, nominal wage rigidity, as measured by the fraction of workers reporting no change in their hourly rate of pay from 12 months earlier, declined to 13 percent—the lowest since April 2008.

The rise depicted by the Wage Growth Tracker is consistent with the recent trend in average hourly earnings from the payroll survey (up 2.8 percent from a year earlier in October—the fastest pace since June 2009). This increase is occurring even though the unemployment rate has changed little in recent months and is only 10 basis points lower than a year ago. Perhaps employers are finally catching up to the realities of a low unemployment rate. Larger wage gains may also be behind why we are seeing fewer workers leave the labor force. Labor force participation is some 30 or so basis points higher than it was a year ago, and this is primarily because the flow out of the labor force has slowed.

Note: The Wage Growth Tracker website now contains data for the smoothed and unsmoothed series going back to 1983. Previously, the historical data started in 1997. You will notice gaps in the time series in 1995–96 and 1985–86 because the Census Bureau masked the identifiers used to match individual earnings during those periods.

November 15, 2016 in Employment , Labor Markets , Wage Growth | Permalink | Comments ( 0)

November 14, 2016


Is There a Gender Wage Growth Gap?

​​

The existence of the "gender wage gap" is well documented. Although the gap in the average level of pay between men and women has narrowed over time, studies conducted in the past few years find that women still tend to make about 20 percent less than men. Researchers estimate that between one half and three quarters of the gap can be accounted for by observable differences between men and women in the workforce such as labor market experience, educational attainment, as well as job characteristics (see here , here, and here). This estimation leaves one quarter to one half of the gap that is the result of other factors. While some pin the remainder on discrimination or unfair hiring practices, others suggest the remaining gap may reflect subtle differences in work preferences, such as women choosing jobs with family-oriented benefit packages or flexible work arrangements.

A related question is whether there are differences between the average wage growth of men and women. Since 2010 the Atlanta Fed's Wage Growth Tracker has revealed a disparity between the pay raises of continuously employed men and women, as depicted in the following chart.

Between 1997 and 2010, wage growth of men and women was about equal. Since 2010 however, a gap has emerged. On average, men have been experiencing about 0.35 percentage points higher median wage growth than women. Can differences in characteristics such as experience and job choice explain this gap?

To answer this question, I aggregated individuals into groups based on their potential labor market experience (0–5 years, 5–9 years, 10–24 years, and 25–48 years) education (degree or no degree) family type (married, whether your spouse works, and whether you have kids); industry (goods versus services) occupation (low, middle, or high skill); sector (public versus private); and if the person switched jobs recently. I then computed the median wage growth for each unique group in each year. Using a statistical technique called a Oaxaca Decomposition, I separated out the difference between men and women's wage growth that can be pinned on differences in the way men and women are distributed among these groups (the "endowment" effect).

The following chart shows median wage growth after removing this endowment effect.

After removing the difference in wage growth that is the result of differences in gender-specific characteristics, wage growth of men and women is much more similar. In particular, these differences appear to almost entirely account for the gap that had emerged after 2009. What explains the gap in wage levels between men and women is still an open question, but this analysis suggests that much of the difference in wage growth through the years has to do with family/job choices and other individual characteristics.

November 14, 2016 in Employment , Labor Markets , Wage Growth | Permalink | Comments ( 1)

November 07, 2016


The Price Isn't Right: On GDPNow's Third Quarter Miss

The U.S. Bureau of Economic Analysis's (BEA) first estimate of third quarter annualized real gross domestic product (GDP) growth released on October 28 was 2.9 percent. A number of nowcasts were quite close to this number, including the median forecast of 3.0 percent from the CNBC Rapid Update surveyOff-site link of roughly 10 economists. The Atlanta Fed's GDPNow model forecast of 2.1 percent? Not so close.

What accounted for GDPNow's miss? The table below shows the GDPNow forecasts and BEA estimates of the percentage point contributions to third quarter growth of six subcomponents that together make up real GDP. The largest forecast error, both in absolute terms and relative to the historical accuracy of the projections, was for the contribution of real net exports to growth. The published contribution of 0.83 percentage points was much higher than the model's estimate of 0.07 percentage points.

Why did GDPNow miss so badly on net exports? For both goods and services real net exports, the GDPNow forecast is a weighted average of two forecasts. The "bean counting" forecast uses the monthly source data on the nominal values and price deflators of exports and imports. The econometric model forecast uses published values of 13 subcomponents of real GDP for the last five quarters to predict real net exports for both goods and services. The statistically determined weights on the bean counting forecast increase as we get closer to the first GDP release and accumulate more monthly source data. (More details are provided here.)

For real net exports of services, 89 percent of the weight was given to the bean counting forecast. This weighting worked out well last quarter as the forecasts of the contribution of real services net exports to third quarter growth from both the bean counting and combined models were within 0.01 percentage points of the published value of −0.14 percentage points. But for real net exports of goods, the bean counting forecast received only 59 percent of the weight in the final GDPNow forecast. It projected that real net exports of goods would add 0.76 percentage points to growth—reasonably close to the BEA estimate. In contrast, the econometric model projected a subtraction of 0.58 percentage points from growth.

Since GDPNow had monthly price and nominal spending data through September on goods imports and exports, why didn't it place more weight on the source data? One of the important reasons is that it's difficult to match the quarterly inflation rate of the BEA's import price deflator for goods. The BEA constructs its price deflator  with detailed price indices from the Bureau of Labor Statistics (BLS) producer price index and import/export price index programs as well as a few other sources. GDPNow uses the BLS's import price data at higher levels of aggregation than the BEA uses and also differs in the manner that it handles seasonality. The chart below plots the difference between the BEA's quarterly goods import price inflation rate and the GDPNow proxy. These inflation measures have differed by 5 percentage points or more on a number of occasions. Since goods imports are 12 percent of GDP, a miss of this magnitude on the price deflator would lead to a miss on the real net exports of goods contribution to growth of 0.5 percentage points or more, even if the other ingredients in the calculation were all correct.

Are there any lessons here for improving GDPNow? Ideally, GDPNow would be able to closely map the monthly source data to real goods net exports so that most of the weight would go to the bean counting forecast once all of the data are in—much as it does with nonresidential structures and residential investment. The BEA's estimates of real petroleum imports are based on similar data in the monthly international trade data publication. Because petroleum imports account for so much of the volatility of inflation for goods imports, it may be better to use the monthly real petroleum imports data directly and only worry about replicating the price index for nonpetroleum goods.

That said, a previous macroblog post illustrated that the method GDPNow currently uses has a reasonable forecasting track record for net exports when compared with several consensus estimates from professional forecasters. Net exports may remain difficult to nowcast even with refinements to GDPNow's methodology.

GDPNow has established a commendable track record. But sometimes when it misses the mark, an analysis of the error can provide insight into how GDPNow works and the limitations of the model.

November 7, 2016 in Forecasts , GDP | Permalink | Comments ( 1)

Google Search



Recent Posts


February 2017


Sun Mon Tue Wed Thu Fri Sat
      1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28        

Archives


Categories


Powered by TypePad