Posted by: Josh Lehner | February 4, 2016

Tobacco Sales, 2015

Cigarettes sold in Oregon have been on a long-run decline since the early 1980s, if not longer, as the smoking rate and overall consumption and usage have plunged. Oregon’s trends have matched or exceeded the national ones over this time period. However, packs sold in 2015 were 2.7 percent higher than in 2014, or nearly 360,000 more.

Our office has a continuation of the long-run consumption decline built into our forecast. Our main variable is the number of packs sold per adult. As such, the increase in sales has resulted in more tax revenue than expected. So far in the first six months of the 2015-17 biennium, actual cigarette tax revenue has exceeded forecast by nearly $8 million. That’s a forecast error of over 7 percent. Unfortunately that’s my forecast error and I can’t blame someone else…

So it’s important to try and figure what is going on here. Overall, the tax differential between Oregon and Washington has had a large impact on the number of packs sold in each state. When Washington raises their tax and Oregon does not, Washington sales fall considerably and Oregon’s stabilize or increase. The opposite is true as well. However, you have to go back to the early 1990s to find a time when Oregon cigarette sales increased and the tax environment was stable, like it is today. Every other instance in the past 25 years of increasing Oregon sales was directly after Washington increased their tax per pack.

CigORWA1215

It can be hard to know what exactly is going on but we do know two things. Based on a recent conversation among our counterparts across the country, this increase is sales is a nationwide trend. The vast majority of states said their sales were above forecast, with many seeing small increases in sales like here in Oregon. The other states saw stabilizing sales or smaller declines then expected. So the driver of these trends is national in scope, and is not a local phenomenon. This could be the increase in disposable income due to lower gas prices finally showing up in the sales data. It could also be related to the overall improvement in the economy, with more plentiful economic opportunities, rising wages and relatively strong consumer sentiment.

It could also be related to consumer behavior and changing tastes or preferences, which is item number two. A recent Wall Street Journal article looked at e-cigarette sales and usage. It pointed out research indicating consumers did not like e-cigs as much, sales have slowed, manufacturers and retailers have a backlog of inventory, and increased scrutiny from states have all impacted the industry. Along these lines, it is very plausible that some smokers switched to e-cigs in recent years, but did not like them, and have switched back to cigarettes, thus the uptick in sales following previous declines.

The second chart focuses on more recent trends. The uptick in cigarette sales is more apparent here than in the long-run graph above. The moist snuff and cigar data comes from quarterly tax filings with the Department of Revenue. DOR Research has been compiling this information for us in recent years, which is quite helpful to see the various trends in the Other Tobacco Products (OTP) category.

Tob15q4

The big driver of sales and revenue in OTP is moist snuff, which continues to increase. Cigar sales can be noisy, but fairly steady over time, like loose tobacco sales as well. Overall OTP revenues so far in 2015-17 have actually come in a bit below forecast (-$1.2 million). Our outlook for OTP tax revenue is for annual increases of about 3 percent, which is a mix of moderate growth in moist snuff and flat trends in all other OTP products. Among our counterparts, there seems to be a consensus for moderate growth in moist snuff consumption (2-5% annually, depending upon who you talk to) until proven otherwise.

Posted by: Josh Lehner | February 2, 2016

Young College Graduate Migration by State

Last week our office issued a new research report on migration into Oregon among young college graduates.  These trends matters for a regional economy for at least two key reasons.

First, most migrants are in their root-setting years (25-34 years old). From the report:

One’s mid-20s through their mid-30s represents a time when most people settle down, begin their careers in earnest, get married, buy a house and have kids. This age group is also vital for longer run economic growth. Once a regional economy is able to attract such workers, they rarely leave, as migration rates decline considerably as an individual ages into mid-life. As such, a place like Oregon is able to grow its working age population through migration and raise the productive capacity of the regional economy.

Second, employment prospects are better for higher degrees of educational attainment. In the report, we divided the population into two nearly equally sized groups based on their major. One group consisted of STEM+ degrees, referring to science, technology, engineering, math plus business or health degrees. The second group was all other degrees – communication, education, liberal arts and the like. From the report:

College graduates have better labor market outcomes than those with less formal schooling. At each step along the educational attainment spectrum, an individual has a higher rate of labor force participation, a higher rate of being employed (and lower rate of unemployment) and higher average wages. STEM+ degree holders even more so than their college graduate counterparts in other fields.

The map below shows 2014 STEM+ migration rates across all 50 states and the District of Columbia. This map shows where young college graduates with scientific, technical or medical degrees are moving to, relative to the overall young working age population. The groupings are nearly perfect quartiles, but based on how the data lines up. The top group consists of 11 states plus D.C., the second group 15 states, the third group 11 states, while 13 states saw net out-migration among young college graduates holding such degrees.

STEMmap14

The scatter plot below shows migration rates across all states for 2014 but keeping the distinction between STEM+ degrees and all other fields of study. The equal value line is show in blue, indicating the migration rate among each degree grouping is the same. A state like Oregon sees nearly identical migration rates for all degree types, albeit skewing slightly more toward the scientific, technical and medical fields. However, there are a handful of state — California, Connecticut, Hawaii, Texas and Virginia — that see strong STEM+ gains but considerably slower gains in all other degrees, or even small losses. (Previously our office did a similar analysis for the largest MSAs.)

STEMscatter14

Finally, another reason these trends matter for a regional economy is the type of jobs being created and filled. As the report highlights, “STEM+ graduates have the highest rate of being employed in degree jobs. This category includes all individual occupations that require at least a Bachelor’s degree or more for an entry-level hire, according to the U.S. Bureau of Labor Statistics.” The Federal Reserve Bank of New York calls these college jobs and just launched a neat interactive feature for U.S. level data. While the first graph below is Oregon-specific, this overall pattern holds nationally and here in the Pacific Northwest as shown in the second graph.

STEMdegreejobPNWDegreeJob

This breakdown into the type of job young college graduates hold is a key reason that average wages are higher for STEM+ degree holders. On the New York Fed site, they do highlight some historical trends in the type of jobs held by young graduates which is quite interesting, along with a detailed table of unemployment, underemployment and average wages by college major. Our office’s work focusing on just the Pacific Northwest does line up with their national results.

Posted by: Josh Lehner | January 28, 2016

REPORT: STEM+ Trends in Oregon

Migration is fundamental to Oregon’s economy and is woven into the fabric of our state. Our office’s new report focuses on net migration among young college graduates and the types of degrees they hold.

.

Download the files here:

Report: Oregon STEM+ Report

Slides: Oregon STEM+ Slides

 

Posted by: Josh Lehner | January 25, 2016

Industrial Production and Recessions

Right now leading indicators are generally split between the manufacturing side which are flashing recession warnings and the non-manufacturing side which are mostly still in expansion territory. Much of the bad news comes from the global and Chinese economic slowdown, strong USD, and pullback in investment due to low oil prices. All of these are interrelated and speak to the manufacturing weakness.

One of the best indicators along these lines is industrial production. We use it in our Oregon Index of Leading Indicators. Right now industrial production is down nearly 2 percent over the past year. Every single time industrial production has been this negative in the past 40 years, the U.S. economy has already been in recession. Obviously this is a concern, to say the least.

However, given that we know much of the decline is due to the oil and gas industry, a key question is how broad of a slowdown are we really seeing or is it just a sector specific story weighing on the top line? The graph below takes a diffusion style look at the 20 subsectors within industrial production (3 digit NAICS level). Right now 7 subsectors are contracting year-over-year. While this is considerably higher than the 1 subsector contracting to begin 2015, it remains below what is typical in past recessions when 16 subsectors registered declines, on average (see table).

IndProRecGraph

The table shows the number of contracting sectors, out of 20 total, for each month when total industrial production first falls nearly 2 percent over the previous year.

IndProRecTableThe declining subsectors are apparel, paper, primary and fabricated metals, machinery, oil and gas, and other mining. The first two sectors no longer comprise much of U.S. manufacturing. Metals and machinery declines are much more worrisome. In our latest Governor’s Council of Economic Advisors meeting on Friday, it was mentioned that the metal weakness was coming in large part from pipes, which are a direct input into oil and gas exploration of course.

Taken together, all of this does paint a picture so far that is closer to an industry specific shock. The pain is not as widespread within industrial production as in past business cycles, or at least not yet. Of course this does not mean the economy is free and clear, rather, this one good indicator may be giving a false warning thus far. See Josh Zumbrun’s latest in the Wall Street Journal for more along these lines.

For the record, I do not believe we are in recession today as most indicators are not saying so, particularly given the labor market strength. As Tim Duy points out, probably the two best leading indicators are the yield curve and initial claims — both of which remain in expansion territory. That being said, our office is worried about manufacturing here in Oregon. Our outlook for manufacturing in the state is very subdued given the strength of the dollar and global weakness.

Posted by: Josh Lehner | January 22, 2016

Housing Bubble, Ownership and School Boundaries?

This is part 2 of 2 this week on housing, see here for part 1. Given that housing and affordability are at the forefront of many people’s minds these days, I have been doing quite a bit of data work to help myself understand what exactly is going on. While the data and focus here is on Portland, the results are generally transferable to other parts of the state, particularly other metropolitan areas.

As you may have expected, the housing bubble and its aftermath is still having big ramifications in the housing market today. First the bubble increased the number of sales during the mid-2000s given the lower underwriting standards but also the psychology of the market during the bubble era. Second, for the majority of households that did not get foreclosed on, they were effectively trapped in their homes due to the significant loss of value. Only now are we starting to see nearly all homes above water.

This is readily evident in the graph below showing how the current ownership pattern in Multnomah County compares to some work from the National Association of Home Builders from prior to the bubble. Their work was based on national data, but acts as a reasonable benchmark of how different today’s pattern actually is. Essentially, there are a whole lot more people still living in the same home for about 10 years then you would expect in a normally functioning market. Historically these likely would be part of the move-up buyer population which is missing today.

MultVintage15

While this second graph is a bit confusing — going from right to left to match the first graph — it compares current ownership to a couple years ago. Overall the pattern is very similar, just lagged 2 years, but the good news is the peaks and valleys are better today. As the market continues to return to normal, expectations are for this overall pattern of ownership to smooth out.

MultVintage1513

One final thought. I find it hard to imagine that these patterns are not impacting the ongoing Portland Public School boundary review and potential realignment. While it is natural for population patterns to shift over time, the housing bubble has resulted in a broken market, making it harder for families to move or relocate in the past decade. So we have more young families than expected in some neighborhoods and less in others (more empty nest households looking for those move-up buyers?). Expectations are for these issues to correct as the economy continues to improve and the housing market makes progress toward normal. Another possibility, as we are seeing it happen too, is the increase in remodeling. As some households were unable to move, they made due with what they had and improved it to meet their needs. To the extent that remodeling solves their housing needs, expectations are more there to be less moving.

Posted by: Josh Lehner | January 20, 2016

Housing Trends and Shifts

This is part 1 of 2 this week on housing. Given that housing and affordability are at the forefront of many people’s minds these days, I have been doing quite a bit of data work to help myself understand what exactly is going on. Affordability is a statewide issue, but our best and most timely data generally comes from the Portland MSA. While the data and focus here is on Portland, the results are generally transferable to other parts of the state, particularly other metropolitan areas.

First, what type of housing do people actually live in? Previously we took a quick look at the housing stock itself, broken down into more categories. However, here I wanted to know how housing changes by age. It’s probably what you expected with younger households renting apartments, with a slow build of single family ownership through one’s peak working and earning years. Overall 61% of Portland housing stock is detached single family homes (51% are owner occupied, 10% are renter occupied).

PDX14SF

What’s really important to understand in the market today and trying to make reasonable forecasts is how these trends have changed over time and where they’re going in the future. The graphs below compare 2007 to 2014 along the owner-renter and detached single family-multifamily spectrum.

 

PDXRenterSF0714

You can clearly see the shifts that have happened, with the largest changes seen among the younger ages. This is the population that felt the brunt of the foreclosure crisis and the largest homeownership rate decline since the bubble. Yet there are changes at nearly all ages as well

Overall there have clearly been massive shifts in the housing market in the past 15 years. First into ownership during the bubble and now back toward renting and multifamily. It is likely that much of these changes have already occurred today, as shown above. Of course they can continue a bit further in the near term. However over a medium or longer look, given demographics and finances, it is likely that both ownership and detached single family will make a comeback from where they are today. See our previous work on the impact of demographics for more. Our office does not expect the pendulum to swing all the way back to where it was during the housing bubble, but certainly for it to swing back part of the way.

One big reason are those demographics. Even if national projections show that all new households will be among the oldest Americans, that’s not true in a popular metropolitan area like Portland. If you simply take household formation rates by age from 2014 ACS data and apply those to our office’s county population forecast, you get this outlook.

HHProj1525

Yes, the older ages will see the most growth as the Boomers age, but there are sizable gains among the younger ages as well due to migration and the Millennials. This is also likely more of a lower bound as it takes 2014’s household formation rates and applies those moving forward. Expectations are formation rates will improve along with the economy. You can roll your own depending upon these assumptions and while the specifics will differ, the general pattern will remain the same.

Posted by: Josh Lehner | January 14, 2016

U.S. EPOP by Single Age

A lot focus in recent years has been on the labor force participation rate — the share of adults either with a job or looking for one — and with good reason. It provides a good estimate of how many individuals are engaged with the labor market and potentially looking at the productive capacity of the economy. With that being said, aging and shifting demographics have a massive impact on the LFPR. As such, it is also important to look the share of individuals that actually have a job. At the end of the day what matters, and what feels good for people is money in their pocket or bank account.

The first graph shows the share of Americans that have a job at each individual age for 2015*. Employment rates are highest, and pretty stable from one’s mid-20s through their mid-50s. This is the reason 25-54 years old is considered the prime working age population. From there, employment and labor force participation rates decline considerably as one ages. As the graph notes, the typical 70 year old is more than two-thirds less likely to be employed than the typical 50 year old.

EPOP15

The shape of the graph above is typical no matter what year you look at. What matters are specific differences between years in terms of their economic impacts. The second graph shows such differences for 2015 relative to the past two business cycle peaks in 2000 and 2007.

EPOP000715

Many economists use the prime working age EPOP as their gauge today given demographic trends, and with good reason. Right now, that measure shows the labor market recovery, in terms of employment rates, is about half-way back to pre-Great Recession rates. What surprised me however was how uniform these changes were for folks in their 20s, 30s and 40s — note the flat line above in this age range. I would have thought there would be more variation among this group. In fact, the Oregon data shows 25-34 year olds all the way back but that may just be sample noise for a small state.

What is really striking in the graph are the employment rates for older workers. Here, there is a larger share of Americans with a job today than in recent history. Some of this is due to the changing nature of jobs and better health. Increased office employment allows individuals to work later in life as the physical hardships are much less than manufacturing, construction and the like. However, certainly some of this increased employment among older Americans is financial. Many more are unable to retire due to the lack of financial security, thus they are working later in life.

Shifting dmeographics are having a big impact on the economy and labor market. Growth rates are expected to be lower moving forward due to them. As Boomers retire, Millennials are replacing them in the labor market, however for each retirement you need to hire more than one person to see positive growth rates. The latest BLS projections indicate 9.8 million more jobs in 2024 than in 2014. However BLS shows 46.5 million job openings during this time, primarily due to this labor market and generational churn. In talks and presentations, I tend to tell people that even with lower growth rates we will have plenty of jobs in the near future, as many of the jobs will be this generational churn below the surface.

* The 2015 data is technically the 12 month average from Dec ’14 through Nov ’15 due to data availability in DataFerrett

Posted by: Josh Lehner | January 11, 2016

Who Moves to Oregon? (Graph of the Week)

This edition of the Graph of the Week continues our focus on migration into Oregon. Migration is very important to the state and one of the two main reasons Oregon outperforms the typical state during expansions. This also sets the stage for our office’s upcoming research report on young, college educated migrants, to be released in a couple weeks.

In recent years, migration indicators like driver licenses surrendered, school enrollment, and moving van lines data, all pointed toward acceleration in population growth in the state. As the actual data has been released (Portland State and then Census/ACS) it confirms these trends. Net migration into Oregon in is now effectively as large as during the housing boom years, at around 40,000 net new migrants per year. But you already knew that if you follow our office’s work, or PSU’s or various media outlets across the state that have been trumpeting the latest van lines data.

What you might not know is who exactly is moving to Oregon. As you may have suspected, its pretty much everyone. Oregon sees a net influx of residents across all age groups. However, a disproportionate number are in their root-setting years; that important decade from mid-20s to mid-30s when most people settle down, begin their careers in earnest, get married, buy a house and have kids.

WhoMovesOR14

These young, working age individuals and households are vital for longer run economic growth. The reason is once a regional economy has such individuals, they rarely leave for another area as migration rates fall considerably from one’s 20s to their 30s and so forth. This means a place like Oregon now has a larger working age population, which can raise the productive capacity and growth rates of the local economy. (Of course, as discussed in our Rural Oregon report, the type of migrants do differ between rural and urban areas.)

Addendum 1:

The two big reasons people move are for jobs and housing. However our office has done some additional work that looks at other characteristics of migrants. Among the factors that lead to higher migration rates are: being single, being young, not being employed, earning lower amounts of income, and having higher educational attainment. Now, that may seem a little bit odd – that collection of characteristics. And it may give some credence to Portlandia’s stereotype, but it really does make sense. Migration is for the young. And migration rates are higher among those with more schooling. So 20- and 30-somethings are more likely to single, more likely to earn less money (they are still early in their careers) and more likely to be unemployed or not in the labor force (one’s peak working years are their late 30s through early 50s).

Addendum 2:

Kanhaiya takes these migration rates by age and by sex, along with some other economic factors — particularly relative unemployment rates, housing prices and the like — and runs them through his population models to create our office’s population and demographic forecasts. As the population’s natural increase (the number of births minus the number of deaths) is slowing considerably, and even turning negative in some locations, the state’s population is being driven more and more by migration.

Addendum 3:

Besides just the overall number of migrants, it is important to look at the relative incomes as well. Particularly for an office that, you know, forecasts state tax revenues. Below is a map Kanhaiya creates based on IRS migration and tax filer data. Oregon loses migrants and income to Washington each and every year. However the inflows from California are always larger and more than compensate for the losses. See slide 12 from Mark’s Destination Oregon presentation at the Oregon Economic Forum for more history on the California and Washington flows. Beyond the northern flow of migrants on the West Coast, Oregon does see net gains from the rest of the country and its regions. Depending upon the year, there are some net outflows to specific states but not a general pattern of losses over time.

IRS13Map

Given that young migrants account for the bulk of the total gains into Oregon, our office has a forthcoming report on the young, college educated migrants. Stay tuned for more in a couple of weeks.

Posted by: Josh Lehner | January 6, 2016

Harney County, A Brief Historical Perspective

Given the ongoing occupation of the Malheur National Wildlife Refuge, and recent articles in the New York Times and the Wall Street Journal about the broader topic of federal lands, our office just wanted to provide a bit of historical economic data on Harney County. Furthermore, Representative Cliff Bentz sits on the House Revenue Committee and regularly engages us in discussing trends in Eastern Oregon, by both informing us of his district and by asking some tough questions.

While big geographically, Harney County is sparsely populated. Employment has been essentially flat for the past forty years, fluctuating around 2,500 jobs. This trend differs considerably from the statewide economy and even the nation. Relative to the late 1970s — just before the state went into the severe early 80s recession and timber industry restructuring — the number of jobs today in Harney County is 10 percent below back then. Clearly, that is a really long time with essentially no growth.

HarneyQCEW7615

While much of the recent focus is on farming and grazing, one big issue facing Oregon, particularly the southern and eastern parts of the state, is the decline of the timber industry.  It started with the early 80s recession and accelerated following the environmental restrictions put in place a decade later. (See here for more on the industry’s history.)

WoodProducts0915

In a state that has lost nearly two-thirds of its direct wood products employment since the late 1970s, Harney County’s losses have actually been significantly larger. From 1978 to 2014, Harney County lost 99 percent of its wood product jobs, with just 6 reported logging jobs remaining today. That takes the industry’s share from more than 3 out of every 10 jobs back in the late 1970s to zero percent today.

HarneyWoodProducts

Now, many argue that the harvest levels and number of jobs in the industry back in the 1970s were not sustainable, due to over logging and the like. Furthermore, even before discussing the environmental impacts or the restrictions, the automation and mechanization in the logging process and in the mills, by itself, would and has resulted in a large employment decline. Yet, most of our good local data begins in the 1970s and it does show the full extent of the decline.

Lastly, the defining feature of the Timber Belt is the influx of migrants to a region that has suffered an economic hit as severe as seen in places like the Rust Belt, the Corn Belt or among the old textile towns across the South. Harney County specifically has seen population losses since its peak in 1979 or 1980, making it an outlier in Oregon. However its two neighboring counties on either side have seen better numbers, with employment and population gains in both Lake and Malheur counties. Overall, the region has certainly increased, even if Harney has seen the worst trends among the group.

TimberBelt12

The overall stagnation of rural economies is not unique to Oregon, and the decline of manufacturing and farm jobs is a nationwide trend. For more on Rural Oregon, see our office’s recent report. Also see our regional economic overview from a year ago and the Employment Department’s regional page.

Posted by: Josh Lehner | December 29, 2015

2016 Outlook: Labor Force Participation and Middle-Wage Jobs

Among the myriad economic trends to watch in 2016, two in particular stand out to me: labor force participation and middle-wage jobs. The reason is both have turned the corner in recent years — even if the conventional wisdom about them has not — and they’re indicative of the growing breadth or coverage of the expansion, something that was clearly missing in the early years of recovery. The good news is the new year should bring continued improvement for both measures of the economy.

First, labor force participation, which is important but can be misleading. It’s important because more workers, or more potential workers can raise the overall productive capacity of the economy, generating stronger growth. Conversely, fewer workers can shrink the capacity of the economy. However the massive demographic shifts in the U.S. and in Oregon are driving much of the underlying participation trends, which you cannot do anything about in the short-term. The LFPR is essentially at its lowest point in recorded history with data going back to the 1970s. However, given the demographics, the LFPR should be at its lowest point, or close to it. What matters from an economic perspective is what our office has been calling the participation gap — the difference between the demographically-adjusted LFPR and the actual LFPR.

LFPR1115

The household survey has been noisy in 2015, to put it nicely. However, given that job gains continue to outpace population growth, and is expected to continue to do so for at least 2016, our office expects some continued improvement in LFPR, including revisions to the 2015 data. We are not looking for a robust rebound, but about a 1 percentage point increase over the next year – part of the labor force response. As the demographically-adjusted LFPR is falling by 0.4 percentage points per year, this combination will cut Oregon’s participation gap nearly in half by the time 2017 comes around. The gap will still exist — in fact our baseline forecast does not have it fully closing as some of the Great Recession damage is likely permanent — but comparable in size to the gap back in 2007. (See here for more on Oregon’s LFPR.)

The second important trend to watch in 2016 is the growth in middle-wage jobs. Broadly speaking this includes occupations that pay roughly $30-50,000 per year — based on our office’s job polarization research (HERE & HERE, e.g.) — and represents the middle part of the income distribution, and employ roughly 60 percent of Americans and Oregonians. This group entails most blue collar occupations like truck drivers, plumbers, construction and production workers in addition to white collar ones like office support staff, sales teams, teachers and community service workers.

Middle-wage jobs accounted for more than 8 out of every 10 lost jobs during the recession. While both high- and low-wage jobs are currently at historical highs in terms of employment, middle-wage jobs are not. In fact they have recovered just about half of their recessionary losses. However growth has returned and appears to be accelerating a bit. We’ll have to wait a few more months for the official 2015 data, but rough estimates based on mapping industry data back to occupations shows sizable gains in Oregon this past year.

MWJobsOR15

Even as it appears 2015 was actually a good year for middle-wage jobs, our office does not expect them to fully regain their recessionary losses until 2017. Much of the growth in recent years has been more on the white collar side, as the blue collar occupations experienced larger losses and fewer gains. However, construction is now on the upswing, along with the housing market.

In our original report, our office discussed the drivers of middle-wage jobs. Some are driven largely by population growth. As migration flows have returned to Oregon and household formation increases, so too does the demand for housing, home repairs, health care, entertainment, teachers, and community services. The uptick in population growth bodes well for these middle-wage jobs in the near term.

The second group of middle-wage jobs can broadly be considered as business support occupations. From the report:

Administrative Support, Sales and Transportation all act as suppliers of labor and services to other businesses or employees. With increases in business operations, including headquarters, the demand for such occupations will increase even if technological advancements continue to eliminate a portion of these jobs. This provides an opportunity for continued investment into activities that foster both an entrepreneurial business climate and also recruitment and retention efforts of existing firms. The loss of significant headquarter operations in Oregon over recent decades has decreased the demand for some of these business support firms and workers.

All told, the outlook for the number of middle-wage jobs in Oregon is relatively bright today. However, over the longer term the relative share of jobs is expected to continue to decline as high- and low-wage jobs see stronger growth.

Older Posts »

Categories

Follow

Get every new post delivered to your Inbox.

Join 488 other followers