Tag: unemployment

Hot September Employment

Posted on by peter.c.rupert

By Paul gomme and Peter Rupert

The BLS announced that payroll employment increased 254,000 in September (plus 72,000 in upward revisions over the previous two months), solidly beating the “forecasts” that hovered around 150,000. Before the report many had talked about the slowing of the labor market, such as this from CNBC:

September’s jobs picture is expected to look a lot like August’s — a gradual slowdown in hiring from earlier this year, a modest increase in wages and a labor market that is looking a lot like many policymakers had hoped it would.

Well, looks like policy makers didn’t get what they hoped for! In fact it looks more like a gradual increase in hiring over the past four months. The private sector led the charge, increasing 223,000, the second highest reading since May of 2023.

Private sector service jobs increased 202,000 with health services and social assistance rising 71,700 and leisure and hospitality jumping up 78,000, the highest since January, 2023. Declines were seen in manufacturing of both durable goods, down 3,000 and non-durable goods down 4,000.

Average hours of work fell from 34.3 to 34.2, so that total hours of work fell 1.5%. Average hourly earnings climbed to $35.36 from $35.23. The growth in hourly earnings continues to outpace CPI inflation, meaning real wages are rising.

The household survey shows an employment increase of 430,000 and the number of unemployed persons fell 281,000. The labor force increased 150,000. The unemployment rate declined from 4.22% to 4.05%. Curiously, in its press release, the Bureau of Labor Statistics said that the unemployment rate was little changed between August and September.

Earlier this week, the Job Openings and Labor Turnover Survey (JOLTS) was released and showed little change in job openings, hires and separations. There are still more job openings than the number of unemployed persons.

Overall, the labor market continues to defy the press who seem to be constantly trying to show the economy is softening. No signs here. What will this do to the outlook for the Fed’s next steps? The labor market is also at odds with the Fed, at least in terms of their last statement,

Recent indicators suggest that economic activity has continued to expand at a solid pace. Job gains have slowed, and the unemployment rate has moved up but remains low. Inflation has made further progress toward the Committee’s 2 percent objective but remains somewhat elevated.

Looking at the very first graph on this post, one can see that job gains have been increasing over the past several months and the unemployment rate actually fell this month. With the strong GDP numbers and this strong labor market outcome it may shift the thinking that inflation pressures could/might/will be increasing. Was the recent 50bp cut too much too soon?

August Employment Report

Posted on by Paul Gomme

On September 6, the Bureau of Labor Statistics released its employment report for August. According to the Establishment Survey, the US economy added 142 thousand jobs in August. In addition, the previous two months were revised down, -61 in June and -25 in July. The Household Survey put the gain at 168 thousand jobs.

From the Establishment Survey, the leisure and hospitality sector added 46 thousand jobs; construction 34 thousand jobs; and health care 30 thousand jobs. Sectors losing jobs were led by manufacturing (24 thousand) and retail trade (11 thousand).

Average weekly hours rose from 34.2 to 34.3 so that total hours of work increased by 4.7%. Average hourly earnings increased 0.4%, from $35.07 to $35.21.

The unemployment rate, derived from the Household Survey, fell marginally, from 4.25% to 4.22%. (The headline numbers are touting a decline from 4.3% to 4.2%.) However, a broader measure of the unemployment rate which includes marginally attached individuals rose between July and August, from 7.8% to 7.9%.

Policy Implications

Is strength or weakness in the eye of the beholder? Or rather, in the data of the beholder?

Given the signaling from members of the FOMC, it’s pretty clear that the committee will lower the Federal funds rate at its next meeting later in September. Speculation is growing that the committee may deliver a 50 basis point reduction rather that `just’ a 25 point reduction. The case for a more aggressive loosening of monetary policy lies on the real side of the economy. Where does this weak real side show up? Maybe in the employment numbers reported above. Yet, the same report shows a slight decline in the unemployment rate and a 48 thousand decline in the number unemployed. The labor force increased by 120 thousand. As mentioned above, wage growth was fairly strong and total hours increased substantially. Second quarter real GDP growth was revised up from 2.8% to 3.0%. Considering that US population growth is running around 0.5% per year, real per capita output growth for the second quarter is around 2.5% which is well above its long run average. Meanwhile, while PCE inflation has fallen, it is still a bit above the FOMC’s 2% target.

July Employment Report

Posted on by peter.c.rupert

By Paul Gomme and Peter Rupert

The BLS announced that employment rose by a very subdued 114,000 according to the establishment survey, 97,000 of which came from the private sector. The bulk of the increase came from Health Care and Social Assistance sector, increasing 64,000. Moreover, there were 29,000 fewer employees over the past two months than reported earlier, as May was revised down 2,000 and June revised down 27,000.

Average hours of work fell from 34.3 to 34.2. Given the weak increase in private sector employment and the decline in average hours meant that total hours of work fell about 2.6%.

To be sure, the July employment report is disappointing. The figure below plots the change in nonfarm payroll employment since 1947. To this figure, we’ve added a red dot when the change in employment was at most 114,000 (as in the most recent jobs report), and the economy was in an expansion. (We’ve excluded a few months around the start of the pandemic because these employment changes were so extreme.) The bottom line is that the US has often had weak job reports in the midst of expansions. The point being that looking only at one monthly report may be very misleading. Indeed the Fed has mentioned this in other contexts:

The Committee does not expect it will be appropriate to reduce the target range until it has gained greater confidence that inflation is moving sustainably toward 2 percent.

July 31 FOMC Statement

The unemployment rate, based on the household survey, rose to 4.25% from 4.05% in the previous month. Digging deeper into the unemployment data reveals that much of the increase in the unemployment rate in 2024 has been due to a combination of workers losing jobs, and workers reentering the labor market, the labor force grew by 420,000. Keep in mind that reentry may be an artifact of the rules for counting an individual as unemployed which includes a notion of active job search. According to the jobs report, in July there were 4.6 million individuals who were not in the labor force but who want a job, an increase of 346,000. When these marginally attached individuals are included in the ranks of the unemployed, the unemployment rate is nearly 8%, not the official 4.3%.

Once again there are many in the media showing the possibility that the economy is heading toward a recession. Note: definitionally, it has to be true that if we are currently not in a recession we are heading toward one since recessions do frequently occur (but have become much less frequent, see the recessions graph below. At any one time there are usually both positive and negative signs. For example, how much weight should one put on a one month decline? Or a one day decline in the stock market? Jeremy Piger uses data to infer the probability the economy is in a recession, as of July 26, the probability that we were in a recession in June was 0.26%. That said, there will be inflation reports before the next FOMC meeting, but at this point it looks more likely there will be a rate cut in September, barring any large increases in inflation.

Q2 GDP and June PCE

Posted on by peter.c.rupert

By Paul Gomme and Peter Rupert

On July 25 the BEA announced that the advance estimate of real GDP increased 2.8% in Q2 on an annualized basis. The gains were fairly widespread, except for residential and non-residential structures, that fell 1.4% and 3.3%, respectively. Personal consumption expenditures, PCE, increased 2.3% and was the largest contributor to overall growth, at 1.57 percentage points.

PCE price index

On July 26, the BEA announced that the personal consumption expenditures, PCE, price index increased 0.95% on an annualized basis. Our preferred trend measure came in at 1.9%.

The Fed’s preferred measure, the PCE ex food and energy came in at 2.2% on an annualized basis while our trend measure came in at 2.6% and has continued to fall for the last 5 months.

As we have mentioned many times before, we believe our trend measure better captures the path of inflation and, more importantly, implications for Fed policy. For example, the annualized monthly change was higher in June, 2.2%, than in May, 1.5%. It seems pretty obvious that the Fed will not change its current stance on policy given this one month blip.

Policy Discussion

No doubt, there will be a lot of chatter about whether the FOMC should lower the Fed funds rate at its July 30-31 meeting, or wait until September. Or something else. To wade through all this, it helps to have a framework to organize thoughts about the incoming data. Arguably, the so-called Taylor rule has the broadest acceptance in the economics-policy profession. Briefly, the Taylor rule says that the Federal funds rate should be set as: (a) the natural real interest rate plus (b) the target inflation rate (2%) with (c) an upward adjustment when actual inflation exceeds target and (d) a downward adjustment reflecting slack on the real side of the economy. Typically, this slack is measured by either the output gap, or the unemployment rate gap. It’s easiest to understand why one of these gaps is included in the Taylor rule by thinking about what happens when there’s a negative gap. In the case of the unemployment rate, the idea is the demand for labor is high. Consequently, either firms will have to offer higher wages, or workers have more bargaining power and can command higher wages. Either way, these nominal wage channels put upward pressure on prices through some sort of “cost push” channel. This could be as simple as firms pricing using a constant markup over their (marginal) costs. In the case of a negative output gap, the story is that demand is outstripping supply, and firms find it easier to raise their prices. Returning to the Taylor rule, the idea is that these gaps reflect future inflationary pressures, and that the FOMC should respond now to head off future inflation.

There’s a lot of wiggle room in the Taylor rule. First, one needs an estimate of the natural real interest rate. The Atlanta Fed’s Taylor rule calculator provides eight (8!) choices for the real interest rate, currently ranging from 0.7% to 2.5%. That said, a given measure of the real interest rate does not change much over time.

Second, how to think about the deviation of inflation from target? As mentioned above, we like our trend measure which has moved down in 2024. The Taylor rule would, then, prescribe a lower Fed funds rate. But that prescription depends on the FOMC having religiously followed the Taylor rule over the past few years — which it almost certainly hasn’t. Nonetheless, some commentators may suggest that it’s time to start lowing the Fed funds rate since PCE inflation has come down in 2024.

Third, how to measure real-side slack? The output gap is given by potential output less actual output. The problem here is the nebulous concept of “potential” output. The fact that FRED has a potential GDP series is of no comfort. (“Fake data!”) To be absolutely clear, the output gap is a made-up number. Similarly, the unemployment rate gap is the difference between the “natural unemployment rate” and the actual unemployment rate. Some may substitute NAIRU (the non-accelerating inflation rate of unemployment) for natural unemployment rate, but it’s the same basic idea. As with the output gap, there’s the problem of measuring the natural unemployment rate. (“More fake data!”) Between the output gap and the unemployment rate gap, the Atlanta Fed provides 18 (yes, 18!) measures of real-side slack. What’s does the Taylor rule say should be happening with the Fed funds rate based on recent real-side data? Currently, estimates of the output gap are positive: there’s slack in the economy which tends to push down the Taylor rule’s prescription for the Fed funds rate. The strong growth in the second quarter is likely to cut the size of this gap (unless of course, potential output is revised!) which calls for a higher Fed funds rate. On the other hand, the unemployment rate has increased, and so the unemployment rate gap has increased which, through the Taylor rule, would call for a lower Fed funds rate.

TLDR: Inflation is coming down; the Taylor rule dictates a lower Fed funds rate. The output gap has narrowed; raise the Fed funds rate. The unemployment gap increased; lower the Fed funds rate.

So, here is the rub, even with the most widely used model at hand, it offers little guidance as to what to do next. Indeed, there is way too much wiggle room to come to a coherent and consistent policy recommendation.

May Employment Report

Posted on by Paul Gomme

By Paul Gomme and Peter Rupert

The BLS reported 272 thousand new jobs according to its Establishment Survey, easily beating economists’ expectations of 190 thousand reported by Bloomberg. The job gains for May exceed the average for the previous 12 months, 232 thousand. Job gains for March were revised down 5 thousand while those for April were revised down 10 thousand.

In contrast, the Household Survey indicates that the economy lost 408 thousand jobs in May. Indeed, 5 out of the last 8 months have seen the two surveys going in opposite directions.

The BLS also reported that the unemployment rate rose slightly, from 3.86% to 3.96% in May.

Meanwhile, the labor force participation rate dropped slightly, from 62.7% to 62.5%.

Average hours of work remained at 34.3 and private employment rose 229 thousand, leading to a 2.1% increase in total hours of work. Average hourly earnings climbed $0.14 and continue to lie above year over year inflation, thereby increasing real wages.

While the headline employment numbers come from the establishment survey, the labor force participation rate is calculated from a household survey, and is calculated by the number of people unemployed plus the number of people employed relative to the age 16 and over non-institutional population. Since the labor force participation rate fell, it follows that the sum of employed and unemployed people must have fallen. We also know that the number of people unemployed rose from 6.5 million to 6.65 million. Consequently, for the sum of employed and unemployed people to fall, it must be that the number of people employed fell. And that’s exactly what the Household Survey tells us. But not the Establishment Survey.

Overall, the labor market shows continued strength and will make the Fed’s decision a little more difficult. Inflation is still above the 2.0% target and with the strong economy there seems little reason to lower the rate at the next meeting. However, as we discussed above, reading the labor market is not as easy as first appears.

March Employment Report

Posted on by peter.c.rupert

By Paul Gomme and Peter Rupert

The BLS announced that payroll employment increased 303,000 in March, another solid reading that will likely change the Fed’s stance concerning the timing of cuts in the Fed Funds rate. The private sector added 232,000.

The construction sector jumped up 39,000, the largest increase since May of 2022.

Average weekly hours of work rose from 34.3 to 34.4 leading to a 5.7% (annualized) increase in total hours of work.

The household survey also showed considerable strength with employment increasing 498,000. The labor force increase 469,000 leading to an increase in the labor force participation rate to 62.7 (was 62.5). The number of unemployed persons fell 29,000 and the unemployment rate fell from 3.86% to 3.83%.

On April 2 the BLS JOLTS data showed that job openings changed little in February, at 8.8 million and the rate of job openings remained at 5.3% for the third straight month.

Economic Week in Review

Posted on by peter.c.rupert

By Paul Gomme and Peter Rupert

It was a pretty busy week for incoming data. Bottom line: The economy continues to reveal strong economic growth and maybe we have not “landed” at all, we are still flying.

Employment Situation

The BLS establishment survey showed that employment rose 187,000 in August. Although the gain was higher than in the previous two months, the June employment numbers were revised down by 80,000 and July down 30,000…employment was 110,000 less than previously reported entering August. The gain was less than the 271,000 average gain over the previous 12 months. Private employment gains led the charge at 179,000 with the service sector adding 143,000. The biggest gain came in health care, up 97,000.

Average hours of work increased from 34.3 to 34.4. That, combined with the 179,000 increase in private employment led to a large increase in total hours of work.

The household survey showed an increase in the unemployment rate from 3.50% to 3.79%. The number of people unemployed did rise by 514,000, however, there was also a 736,000 increase in the labor force. The labor force participation rate has been steadily increasing but is still below the pre-pandemic level. Basically, once the effects of the pandemic have receded there has not been much of a change in the reasons for showing up as unemployed.

Since the unemployment rate is the ratio of the number of people unemployed divided by the labor force (the number of people employed and unemployed), the unemployment rate can increase either because the numerator (the number unemployed) increased, or because the denominator (the labor force) decreased. Which one accounts for the 0.29 percentage point increase in the unemployment rate in August? According to the Household Survey, the number of unemployed rose by 514 thousand in August while employment rose 222 thousand. In other words, the labor force increased by 736 thousand. For August, the increase in the unemployment rate came about due to an increase in the number of individuals unemployed.

The figure below digs deeper into the labor market flows. The arrows reflect flows of people between employment (E), unemployment (U), and not-in-the-labor force (N). The change in employment is obtained by adding the numbers with arrows pointing into E, and subtracting the flows associated with arrows out of E: a net increase in employment of 146 thousand. This number is different from the 222 thousand obtained directly from the employment data from the Household Survey. The reason being that there are some additional inflows and outflows found in this table that have to do with adjustments to population, teenagers turning 16, etc. We can similarly compute the change in the number of unemployed by looking at the flows in and out of U: an increase of 512 thousand (rather closer to the actual change of 514 thousand obtained from the number unemployed with the adjustments). Relative to the flows in and out of unemployment, 512 thousand is not a huge number: the total flows (regardless of sign) sum to 6,180 thousand, and so 512 thousand is 8.3% of the total flows.

The charts below show that the number unemployed rose by 99 thousand due to an increased number of people out of the labor force moving into unemployment, by 75 thousand owing to a decrease in those transiting between unemployment and out of the labor force, by 175 thousand by virtue of more employed people becoming unemployed, and by 281 thousand as a result of fewer unemployed shifting into employment.

Job Openings and Labor Turnover Survey

The JOLTS data came out 8/29 and showed the number of job openings declined to 8.8 million in July from 9.2 million and 11.4 million in July of 2022. Having said that, the difference between the number of people unemployed and the available jobs are still much higher than any time pre-pandemic. The quit rate has come down somewhat, but, like the openings rate, still higher than its pre-pandemic level. One interpretation of the quit rate is that quitting and moving to better jobs helps one climb the job and income ladder. Said differently, quit rates fall during recessions as there are fewer opportunities to move. Layoffs remain very subdued as well. The rate of job hiring as fallen considerably over the past year or so and now back to the average rate since 2014 (excluding the pandemic).

Output, Income and Consumption

The second estimate for Q2 real GDP was released by the BEA on August 30, and showed a downward revision from 2.4% to 2.1%. Consumption was revised up from 1.6% to 1.7%.

While the downward revision in Q2 real GDP was not small, the monthly consumption data released on 8/30 by the BEA showed a large increase in consumption for July. Consumption expenditures increased 0.8% in current dollars (9.9% annualized) and 0.6% in chained 2012 dollars, the largest increase since January. The personal savings rate as a fraction of disposable income declined by nearly a full percentage point, from 4.3% to 3.5%

Takeaways

The data describe a growing economy with little, if any, signs of braking. Looking at the headline numbers and article titles, such as this in the 9/1 WSJ: “Job Gains Eased in Summer Months; Unemployment Increased in August,” might suggest a faltering labor market. However, a deeper dive into the underlying data suggests no such thing.

May Employment Report

Posted on by peter.c.rupert

By Paul Gomme and Peter Rupert

The BLS announced that payroll employment climbed 339,000 in May. The private sector added 283,000. The bulk of the gain was in the service sector, adding 257,000. The goods sector increased 26,000 but almost all of it, 25,000, in construction.

Although employment climbed 339,000, average weekly hours fell from 34.4 to 34.3 (a decline of 0.1%), throwing a little cold water on the overall report. As shown in the chart below, average weekly hours have trended down since late 2020. Hourly pay, however, climbed 0.3%, from $34.33 to $34.44, and the year over year increase was 4.30%; unfortunately, that growth is still a bit below the year-over-year CPI inflation rate for April, 4.96% (the May CPI has not yet been released).

The household data showed almost exactly the opposite from the establishment survey with an employment decline of 310,000. The employment to population ratio fell slightly from 60.4 to 60.3, and still falls below the pre-pandemic level of 61.1.

The unemployment rate is based on data from the household survey. As mentioned above, employment fell by 310,000 according to the household survey. Combined with a 440,000 increase in the number of unemployed persons, the unemployment rate rose from 3.39% to 3.65%. The changes in the number of employed and unemployed left the labor force participation rate unchanged at 62.6%.

To be “officially” classified as unemployed, an individual must have “actively” looked for a job and available to begin employment. This definition excludes those who are deemed “marginally attached” to the labor force who would like a job, but have not taken sufficiently active measures to find one. The chart below plots the headline unemployment rate along with a measure that includes marginally attached individuals as well as those employed part time for economic reasons (the “U-6” definition). Whereas the headline unemployment rate is 3.7%, the broader measure of unemployment stands at 6.7%. The gap between the two, 3.0%, is about as low as it has been since 1994 when the U-6 measure of the unemployment rate starts.

Until the early 1980s, female unemployment tended to exceed that of men; since then, the pattern has reversed. Since 2021, the male unemployment rate has exceeded that of women by 0.17 percentage points.

Note: We use the terminology of the BLS so as not to add any confusion, in particular, Sex and Race. Also, the BLS uses the terminology Race and Hispanic or Latino Ethnicity.

Historically, Black and African American unemployment rates have exceeded that of other racial groups. Since 1973 (when the data becomes available), the Black and African American unemployment rate has averaged 6.1 percentage points higher than that of whites. Over time, this gap has narrowed; since 2021, it has averaged 3.1 percentage points. Hispanic and Latino unemployment rates lie between that of Blacks/African Americans and whites. (Note that we have seasonally adjusted the Hispanic and Latino unemployment rates using Python’s ARIMA X11 package with default settings; officially seasonally adjusted series are not available.) Over the available data, the Hispanic/Latino unemployment rate exceeded that of whites by 3.15 percentage points; since 2021, by 1.5 points. The Asian unemployment rate is only available since 2003. On average, their unemployment rate is 0.55 percentage points lower than that of whites; since 2021, the gap is only 0.09 points.

We can further look at unemployment rates by sex and race, albeit for those 20 years of age and older. As mentioned earlier, since the early 1980s, for the population as a whole the male unemployment rate has exceeded that of women. While the same is generally true for Blacks/African Americans and whites, the average unemployment rate for Hispanic/Latino women is 0.95 percentage points higher than that of Hispanic/Latino men. Data for Asians is not broken down by sex.

People enter unemployment through various channels. The largest component is for people who lose their job, that represents about half of all of the unemployed. The next largest category is from those who reenter the labor force after a spell of being absent; these are labeled reentrants. Then there are those that voluntarily leave their jobs and those who are just entering the labor force.

Looking across those unemployed, average weeks of unemployment has been trending up somewhat over time. Between 1950 to 1980 average weeks of unemployment hovered between 10 and 15 weeks. Indeed, average weeks never hit 20 weeks until after 1980. Since that time average weeks have hit 20 or more numerous times and today stands at just over 21. The Great Recession and the pandemic had massive effects on weeks of unemployment.

Overall, the labor market continues its strong performance. While the unemployment rate increased it still remains as low as the economy has seen for decades.

February Employment Report

Posted on by Paul Gomme

We are not seeing a lot in this employment report to warrant any large changes in beliefs about the current stance of the labor market. The headline number in the Establishment Survey showed a gain of 311,000, solid, but not remarkable. There were gains in: (1) leisure and hospitality (105,000), (2) retail trade (50,000), (3) government (46,000, of which 37,000 is local government) and (4) health care (44,000). It also mentions losses in: (1) information (25,000), and (2) transportation and warehousing (22,000). It also notes that the information sector has shed 54,000 jobs since November of 2022. There were 25,000 jobs lost in the information sector in February, 9,000 of which were in the Motion Picture and Sound Recording Industry, and 3,000 were in Telecommunications.

Throwing a little cold water on the report were sizable downward revisions to the Establishment Survey employment gains for December (down 21,000 to 239,000) and January (down 13,000 to 504,000). The figure below gives a historical perspective on the first (initial) release, the second (revised) release, and the third (final) release.

In addition to the downward revisions, average weekly hours (private non-farm payroll) are down 0.1 hours, to 34.5. However, average weekly hours are still “high” by (recent) historical standards. Having said that, the combination of the rise in employment with the decline in average hours led to an overall decline in total hours of work by about 4 million, or 1%.

Average hourly earnings were up $0.08 and up about 4.6% year over year. However, CPI inflation continues to erode real average earnings.

The Household Survey showed that the number of people unemployed (up 242,000) and the unemployment rate (from 3.43% to 3.57%) rose. The labor force expanded by 419,000, moving the labor force participation rate to 62.5%. All groups except for teens have been trending up slightly since the end of the pandemic.

Overall, the labor market remains one of the strongest in recent memory. Not only is employment growing at a solid pace, the number of job vacancies has stayed substantially higher than the number of people unemployed and is unprecedented since these data began at the end of 2000. The relationship between vacancies and unemployment, known as the Beveridge Curve, has had a strange journey due to the pandemic. During the Great Recession the unemployment rate increased substantially as vacancies continued to decline. The “counterclockwise” pattern, standard in the search and matching framework, saw a reduction in the unemployment rate as vacancies began to increase. The unprecedented closures to business at the onset of the pandemic drove the unemployment rate to 15% even though vacancies did not decline substantially. Today, vacancies are near an all time high and unemployment near an all time low.

December Labor Report

Posted on by peter.c.rupert

By Paul Gomme and Peter Rupert

The BLS announced December employment on January 6 with another solid month that continues to show one of the strongest labor markets in recent history. Payroll employment increased 223,000 overall and 220,000 in the private sector.

The service sector provided the largest increase, 180,000. Temporary help services fell 35,000 after falling 30,300 and 22,300 in the previous two months. On the goods side, construction and durable goods employment came in strong up 28,000 and 24,000, respectively; although nondurable goods employment fell 16,000.

Average hours of work fell to 34.3 from 34.4. The increase in employment and decrease in average hours worked have opposing effects on total hours worked. In the event, the decline in average hours worked dominated as total hours of work dropped for the second straight month.

The household survey showed an increase in employment of 717,000, once again showing a marked difference from the establishment survey.

The labor force increased 439,000 leading to an increase in the participation rate to 62.3. Moreover, the number of unemployed persons fell by 278,000. These changes led to a fall in the unemployment rate to 3.5%.

Overall, the labor market continues to show considerable strength. Consequently, any discussion of a recession needs to focus on other parts of the economy.