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 23, 2017


More Ways to Watch Wages

The Atlanta Fed's Wage Growth Tracker slipped to 3.2 percent in January from 3.5 percent in December. The Wage Growth Tracker for women was 3.1 percent in January, down significantly from what we saw in late 2016, when gains topped 4 percent. For men, the January reading was 3.4 percent, very close to its average for the past 12 months. As I noted last month, I did not think the unusually high female wage growth was sustainable, and that proved to be the case. Since 2009, the Wage Growth Tracker for women has averaged about 0.3 percentage points below that for men—the same as the gap in the latest data.

Understanding why the Wage Growth Tracker slowed last month highlights the importance of being able to look beyond the top-line number. To provide Wage Growth Tracker users with more information, we have now added several additional cuts of the data to the Wage Growth Tracker web page. The amount of detail we can provide is limited by sample size considerations, and as a result, the additional data are reported as 12-month moving averages. The new data provide more detailed age, race, education, and geographic comparisons, as well as comparisons across broad categories of occupation, industry, and hours worked. As an example, here is a look at the (12-month average) median wage growth data for those who usually work full-time versus those who usually work part-time.

Have fun with these new tools, and we encourage you to comment and let us know what you think.

February 23, 2017 in Employment , Labor Markets , Unemployment , Wage Growth | Permalink | Comments ( 1)

February 21, 2017


Unemployment versus Underemployment: Assessing Labor Market Slack

The U-3 unemployment rate has returned to prerecession levels and is close to estimates of its longer-run sustainable level. Yet other indicators of slack, such as the U-6 statistic, which includes people working part-time but wanting to work full-time (often referred to as part-time for economic reasons, or PTER), has not declined as quickly or by as much as the U-3 unemployment rate.

If unemployment and PTER reflect the same business-cycle effects, then they should move pretty much in lockstep. But as the following chart shows, such uniformity hasn't generally been the case. In the most recent recovery, unemployment started declining in 2010, but PTER started to move substantially lower beginning only in 2013. The upshot is that for each unemployed worker, there are now many more involuntary part-time workers than in the past.

Regarding the above chart, I should note that I adjusted the pre-1994 data to be consistent with the 1994 redesign of the Current Population Survey from the U.S. Bureau of Labor Statistics (see, for example, research from Rob Valletta and Leila Bengali and Anne Polivka and Stephen Miller ). This adjustment amounts to reducing the pre-1994 number of PTER workers by about 20 percent.

The elevated level of PTER workers has been most pronounced for workers in low-skill occupations. As shown in the next chart, PTER workers in low-skill jobs now outnumber unemployed workers who left low-skill jobs. Prior to the most recent recession, low-skill unemployment was always higher than low-skill PTER.

The increase in PTER workers is also mostly in the retail trade industry, as well as the leisure and hospitality industry, where low-skill occupations are concentrated. The PTER-to-unemployment ratio for the goods-producing sector (manufacturing, construction, and mining) has remained essentially unchanged. In those industries, unemployment and PTER move together.

Some researchers, such as our colleagues at the San Francisco Fed Rob Valletta and Catherine van der List, have argued that the increase in the prevalence of involuntary part-time work relative to unemployment suggests the importance of factors other than overall demand for labor. Among these factors are shifting demographics (a greater number of older workers who are less willing to do part-time work) and industry mix (more employment in industries with higher concentrations of part-time jobs). Such factors are almost certainly playing a role.

Recent analysis by Jon Willis at the Kansas City Fed  suggests that the elevated levels of PTER in low-skill occupations may reflect that during the last recession, firms reduced the hours of workers in low-skill jobs more than they cut the number of low-skill jobs. In other words, firms still had some work that needed to get done, probably with peak demand at certain times of the day, and those tasks couldn't readily be outsourced or automated.

As the following chart from Willis's research shows, between 2007 and 2010, low-skill (non-PTER) employment actually increased slightly overall, but the mix of employment shifted dramatically toward part-time.

Since the recession, the pace of (non-PTER) low-skill job creation has been modest (about 20,000 jobs per month compared with 60,000 jobs per month in the years preceding the recession). Initially, this trend helped reduce low-skill unemployment more than the incidence of PTER—one reason why the ratio of PTER to unemployment continued to increase.

But the number of PTER workers in low-skill jobs has since been declining as more people have been able to find full-time jobs. At the current pace of job creation and (net) transition rates out of PTER, Willis estimates it would take until 2020 to return to prerecession levels of low-skill PTER. That seems a reasonable guess to me.

February 21, 2017 in Employment , Labor Markets , Unemployment , Wage Growth | Permalink | Comments ( 0)

February 13, 2017


Does a High-Pressure Labor Market Bring Long-Term Benefits?

Though it ticked up slightly in January , the U.S. unemployment rate is arguably at, or near, its long-run sustainable level. At least that is the apparent judgment of Federal Open Market Committee participants, the Congressional Budget Office (CBO), and others. Not surprisingly, this consensus is leading to some speculation that a combination of policy and the economy's natural momentum may result in unemployment rates moving well below sustainable levels—a circumstance some have referred to as a "high-pressure" economy.

Though lower-than-normal unemployment rates may have benefits, at least in the short-term, it is generally recognized that these circumstances also carry risks. Specifically, if the demand for resources (including labor) expands beyond the economy's capacity to supply them, the risk of undesirable inflation, financial imbalances, and other negative developments may grow—a point that Boston Fed President Eric Rosengren emphasized late last year. In recent history, high-pressure episodes have generally ended with the economy entering a recession; soft landings appear to be elusive.

That said, some have outlined potential labor market benefits to individual workers during high-pressure episodes—including higher labor force attachment, higher wages, and better job matches (see for example, here, here and here ). But could these types of labor market benefits persist and actually improve a worker's ability to also withstand an economic downturn?

To investigate this possibility, I ask the following question: Do high-pressure economies at the state level reduce the probability that a worker enters into unemployment during a subsequent downturn?

The details of my approach, using cross-sectional data from the monthly Current Population Survey, can be found in this appendix .

The following three charts illustrate the moderating impact a high-pressure economy can have on the probability of unemployment during a recession for various demographic groups. Chart 1 shows the impact on different age groups. The data tell us that the probability of unemployment for 18- to 34-year olds is 3.2 percentage points higher during recessions than during expansions, relative to how much higher the probability of unemployment is during recessions for 55- to 64-year olds (the excluded age group). This estimate is an average across all recessions between 1980 and 2015. Those who are 45- to 54-years old have only a modestly higher probability of unemployment (0.4 of a percentage point) during recessions than 55- to 64-year olds.

However, we also see from chart 1 that the effect of the recession on each age group is moderated by the state's high-pressure economy. Specifically, for each average percentage point by which the state's unemployment rate fell below the state's natural rate of unemployment prior to the recession, the probability of unemployment facing 18- to 34-year olds falls by 2.4 percentage points. Simply put, the hotter the state's prerecession economy, the lower the impact of the recession on workers' probability of unemployment.

We see the same impact across education groups in chart 2. Whereas those with some college face a probability of unemployment during a recession that is 0.7 percentage points higher than that of a college graduate, a prerecessionary high-pressure episode just 1 percentage point higher will wipe out the disadvantage that those with some college face during a recession relative to those with a college degree.

Chart 3 shows that black non-Hispanics experience even greater benefits from a high-pressure economy. A high-pressure period just 1 percentage point greater prior to a recession more than erases the average impact of the recession, relative to white non-Hispanics. (Note that these results are averaged across all recessions since 1980 and hence don't say anything about the labor market outcomes during any particular recession.)

The evidence I provide here suggests that a high-pressure economy may have some longer-term benefits in terms of improving labor market outcomes during economic downturns. If this is indeed the case, understanding how and why will be an important step in assessing the risk/reward calculus of high-pressure periods.

February 13, 2017 in Employment , Labor Markets , Unemployment , Wage Growth | Permalink | Comments ( 0)

February 07, 2017


Net Exports Continue to Bedevil GDPNow

Real gross domestic product (GDP) grew at an annualized rate of 1.9 percent in the fourth quarter, according to the advance estimate from the U.S. Bureau of Economic Analysis (BEA), 1.0 percentage point below the Atlanta Fed's final GDPNow model projection. This was a sizable miss relative to other forecasts. Both the consensus estimate from the January Wall Street Journal Economic Forecasting Survey and the January 20 staff nowcast from the New York Fed were expecting 2.1 percent growth last quarter.

The miss was also large relative to the historical accuracy of the GDPNow model. As the table below shows, almost all of GDPNow's error for fourth quarter growth was concentrated in real net exports. For the other broad subcomponents, GDPNow was more accurate than usual, as the last two columns of the table show. But net exports subtracted 1.70 percentage points from real GDP growth last quarter, whereas GDPNow forecasted they would only reduce growth by 0.64 percentage points. All but 0.02 percentage points of this error was in the "goods" category as opposed to services.

Three months ago, I wrote a macroblog post showing that nearly all of GDPNow's 0.8 percentage point error for third-quarter growth was concentrated in goods net exports. That analysis explained how GDPNow's goods net exports forecast is a weighted average of two forecasts. One of these forecasts is a "bean counting" model that uses monthly source data on nominal values and price deflators for goods imports and exports. The other is a quarterly econometric model that uses subcomponents of real GDP for prior quarters. In the GDPNow model, the "bean counting" model gets nearly 60 percent of the weight just before the advance GDP release.

To see how this approach matters for the GDP forecast, the following chart shows the "real-time" forecasts of the contribution of goods net exports to growth just before BEA's advance GDP estimate from the two models alongside the advance estimate of the contribution and the final GDPNow forecast.

We see that the "bean counting" forecast has been much more accurate than the quarterly econometric forecast, particularly for the last two quarters of 2016. Not surprisingly given its name, the "bean counting" model was able to largely capture the 0.75 percentage points that soybean exports contributed to third-quarter real GDP growth and the just over 0.5 percentage points they likely subtracted from fourth-quarter growth. The econometric model was not.

The final forecasts of goods net exports from the "bean counting" model have also been more accurate than GDPNow since forecasts were first posted online in mid-2014. Does this imply that an alternative "bean counting" version of GDPNow would be preferable? The answer is less obvious than you might think. Not putting any weight on the quarterly econometric model for any GDP subcomponents yields an average error for GDP growth (without regard to sign) of 0.635 percentage points, and the same statistic for GDPNow is 0.589 percentage points. This is despite the fact that the "bean counting" approach has been more accurate than GDPNow in its forecasts of net exports and about as accurate, on balance, for the other GDP subcomponents.

The final forecast of real GDP growth last quarter of this alternative "bean counting" model was 2.8 percent—only slightly more accurate than GDPNow. (For each GDP subcomponent, I include the "bean counting" and quarterly econometric model forecasts in this excel spreadsheet.)

However, if variants like the aforementioned "bean counting" approach continue to outperform the GDPNow model in one or more dimensions, we may consider regularly reporting their forecasts along with the GDPNow forecast.

February 7, 2017 in Forecasts , GDP | Permalink | Comments ( 0)

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)

Google Search



Recent Posts


March 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 29 30 31  

Archives


Categories


Powered by TypePad