Wealth and Income

Recently released data from the 2019 Survey of Consumer Finances (SCF) show a highly unequal distribution of wealth in the US, both between and within racial groups. This is not new, but the numbers are always staggering. The typical white family has eight times the wealth of the typical black family. Within racial and ethnic groups, more than eighty percent of wealth is owned by the wealthiest twenty percent of families.

Having wealth is a huge advantage and not having wealth is a huge disadvantage, yet people argue, incorrectly in my view, that wealth statistics are not as important as they seem. I want to respond to some common arguments, which generally revolve around the connection between wealth and income.

Common argument #1: Wealth is not a good predictor of poverty

The basic argument here is that some people with low or negative net worth actually have very high income. Therefore, since poverty is defined based on income, people with low wealth are not necessarily poor. There’s nothing wrong with this argument; because wealth is measured as net worth, assets minus debts, these cases do exist. A recent medical school grad with lots of student debt and few assets will have negative wealth but a very high salary and is unlikely to be in poverty. But these are corner cases. All the argument is saying is that there exist some cases where the relationship between net worth and income falls apart.

Applying this argument to anything other than the specific corner cases is misleading. There is a strong correlation between having low wealth and having low income and nearly all poor people have low wealth. Likewise, few wealthy people are in poverty on an income basis.

More importantly, the argument does not explain differences between demographic groups. We know that the difference between black and white wealth is not in any way compensated for by black people having higher income. Overall, the story of the black-white wealth gap is an assets story, not a student debt story. Black families have substantially fewer assets, substantially less wealth, and substantially less income. We can see this by looking at net worth, financial assets, and pre-tax income:

Common argument #2: Wealth is not a permanent source of income

Again, this is an argument mostly based around corner cases. If someone has equity in their home, they may have some wealth but not necessarily have an income advantage over someone who rents. Stocks might temporarily skip paying dividends because of weak economic conditions. There are even cases where people lose money by having wealth comprised of assets that lose value. The horror.

But the bigger picture here is that assets generally do provide income or utility and having a positive net worth does mean more income when viewed from the aggregate data. I want to illustrate this in two ways. First, the pre-tax income of various groups by their wealth percentile shows a clear relationship between income and wealth. People with more wealth have higher income:

Additionally, if we look at aggregate personal income data for the US in 2019, the overall growth from 2018 is an impressive 6.8 percent after adjusting for inflation. The second largest source of that growth, explaining 1.5 percentage points, is larger payments to property owners in 2019 than in 2018:

I wouldn’t say that wealth is a permanent source of income but neither is labor. We can see this in the above chart from the change in the number of recipients of earnings (labor income). The chart shows that aggregate personal income went up 1.1 percentage points because more people were employed in 2019 than 2018. In other words, a (disproportionately black) portion of the population was still re-entering the labor force a decade after the great recession had ended. And as we’ve seen recently, people lose jobs all the time and sometimes these job losses last for many years, so what is a permanent source of income?

Perhaps a better question is whether wealth is consequential and reliable as a source of income. I’d argue that it is. Various forms of ownership income are equivalent to annualized income of more than $16,000 per person in August, despite the very severe ongoing recession.

Common argument #3: The concentration of wealth is a life cycle story

The following chart is useful for thinking about income and consumption over a person’s life:


The argument related to the life cycle story shown in this chart goes something like “wealth is concentrated because of age and reflects peoples’ retirement savings”. People argue that wealth reflects accrued earnings–it comes from the portion of the chart where labor income exceeds consumption. If wealth is simply accrued earnings used for consumption in retirement, then its okay that some (older) people have wealth and other (younger) people don’t (yet).

But the issue here is that large gaps still exist between families in the same age group and other large differences exist in ways that age does not explain. For example, some young people receive inheritance, in-vivo transfers, or other transfers like a car, help with college tuition, or help with a home down payment. Other young people simply do not receive these things. The gap by race is very wide when it comes to these kind of inter-generational transfers–they are much more common among white families than black families.

The accrued earnings story just doesn’t hold up to the data. From the Federal Reserve, “by some estimates bequests and transfers account for at least half of aggregate wealth (Gale and Scholz 1994), have recently averaged 3 percent of total household disposable personal income (Feiveson and Sabelhaus 2018), and account for more of the racial wealth gap than any other demographic or socioeconomic indicator (Hamilton and Darity 2010)”

Also, from 2016 SCF data, a large and growing share of people nearing retirement age (those 55-64) have no or negative wealth outside of home equity, and many also have no equity stake in their home. Social Security payments are the main source of income for a large portion of the retirement-age population and this should be expected to grow as defined benefit pensions plans become more and more rare. The end result of the current system of using private wealth to resolve income and consumption differences in the life cycle is massive gaps between people in access to goods and services at all points of the life cycle.

Should we focus on wealth?

First, if we think about wealth as assets rather than net worth (because the net worth differences between and within racial groups are largely the result of differences in assets), we can easily see the advantage of having money to cushion against an unexpected drop in income or unexpected expense. Black families are less likely to have emergency savings, and those with some savings have less than 1/5th the emergency savings of white families, on average. Black families are also substantially less likely to be able to borrow money from a friend or relative. For these reasons and many others, wealth is important, not just because it can be a source of income.

That said, income is still critically important. A policy focus on full employment, increased worker bargaining power, and compressing the wage distribution would make a huge difference in reducing poverty and reducing income inequality, and eventually, in reducing wealth inequality. But increasing the share of total income that goes to labor won’t solve one major source of inequality and poverty: the unequal distribution of workers in families. Some families contain one or more people who cannot or should not be working (children, students, elderly, persons with disabilities, unpaid caregivers, etc.), while other families do not. This situation is more common than people realize (around half of the country are non-workers during full employment) and is precisely where the welfare state shines. The US could use public ownership of wealth as a supplement to taxes in funding income payments to non-workers. As many countries already demonstrate, such payments can be designed to efficiently solve the inequality caused specifically by the uneven distribution of workers in families.

Importantly, whether the US manages to redirect a portion of its capital income to new welfare programs or not, the US government should make reparation payments to American descendants of slavery. People want to believe that wealth results from accrued earnings because it is a meritocratic story. The accrued earnings of millions of black people were stolen during slavery. Just because the victims died does not mean the wealth wasn’t stolen from their descendants. The theft of these earnings has contributed greatly to the massive racial wealth gap and this gap should be closed through both policies that close the racial income gap and through reparations and increased public ownership of assets.

Adding texture to the jobs report

The monthly jobs report shows a widely-cited spike in unemployment, starting in April 2020, as a result of the COVID-19 pandemic. It’s worth illustrating that employment effects go far beyond what is measured by unemployment, and additional analysis of household survey data give a more complete picture. The large number of people affected support a case for broad eligibility for government financial support.


The age 16 or older population can be divided into three categories: those who are employed in any amount, those who are not employed but are on temporary layoff or are looking for work (the unemployed), and those who are out of the labor force. The out of the labor force group includes students, caregivers, persons with disabilities, and the elderly, as well as those who have given up looking for work but would like a job.

Screenshot from 2020-08-16 20-44-39

Because the three categories in the chart above sum to 100% in any give month, a change in one category must be offset by a change in another category. For example, during a recession (shaded gray area) the employed share of the population falls as people become unemployed and leave the labor force.During the great recession in 2008-09 the job losses were protracted. From October 2006 to October 2007, the employed share of the age 16+ population fell by 0.6 percentage point. From October 2007 to October 2008, the employed share fell by 1.0 percentage point, and from October 2008 to October 2009 it fell by another 3.2 percentage points. In contrast, the 2020 recession started very abruptly. The March report covered the week before many places had shut down, but the April 2020 report showed a 9.3 percentage point drop in the employed share of the age 16+ population from April 2019. Each subsequent month, with the most recent available being July, showed a smaller one-year rate of change than the previous month.Screenshot from 2020-08-16 20-45-00To get a bigger sample size and an overview of the April 2020 onward data, the following sections combine the monthly survey data for April through July 2020 and compare it to the corresponding combined survey data covering April to July 2019. The average decrease in the employed share of those 16 or older was 7.2 percentage points, with three-quarters of that shift moving into unemployment (a 5.2 percentage point one-year change) and one-quarter leaving the labor force (2.0 percentage point one-year change).Screenshot from 2020-08-16 20-45-21

Overall Change by Age

Data show the effects in labor force status are disproportionate to young people. Among three-year age groups used for these calculations, the age 19-21 group was the hardest hit, with a decrease in the age group employment rate of 16.8 percentage points in April through July 2020 compared to the same four months the previous year. The age 19-21 group was also disproportionately likely to leave the labor force, with a 7.2 percentage point one-year change in the share of the population that is not in the labor force, compared to a 9.6 percentage point change in the share that is unemployed. The effect was similar but less pronounced for those ages 16-18, 22-24, and 25-27.Screenshot from 2020-08-16 20-45-34Next we look at changes within each of the three categories discussed above.

Changes within types of employment

Shifts within the employment category are important. For example, some full-time workers have their hours cut but are still employed, and some people lose a second job. Additionally, the Bureau of Labor Statistics reported that some people who should probably should have been classified as unemployed were instead described as “employed, but not at work” in the data, particularly in April 2020. In these three examples the person is classified as employed and wouldn’t show up in the previous chart.Therefore, the next chart divides those who are employed into four nonoverlapping categories: multiple jobholders (people who have more than one job), those fully-employed (defined as working 35 or more hours or voluntarily working fewer) and at work during the reference week, those working fewer than 35 hours but who want more hours (part-time for economic reasons), and those who are employed but not at work. The one-year change in each category shows that the employment effect is substantially larger when considering the deterioration within employment categories.Most age groups saw a decrease in the “fully-employed, at work” category that was larger than the overall change in employed defined in the previous sections. This is because of the reduction in second and third jobs, the increase in people not at work, and the shift of millions of people to part-time status, despite their preference for full-time work. The reduction in multiple jobholding rates was highest among those age 22-24 (1.9 percentage point decrease). The age 49-51 group had the biggest increase in the share of the age group that works part-time but wants to work full-time, with a 3.4 percentage point increase.Screenshot from 2020-08-16 21-45-08

Changes in Unemployment by Age

A bright spot in the four months of historically bad data is that most of the unemployed are on layoff. Unlike other recent recessions, the reduction in economic activity was a decision made for public health purposes. Arguably, the US did not do enough or move quickly enough in reducing employment and non-essential economic activity, causing the country to have many more cases and preventable deaths than western Europe.It is tough to know whether a job loss will actually be temporary, but the expectation during the early months of the shutdown was that people stay home and that policymakers provide enough income to keep most families financially stable without them going out and looking for work. The data show that people very disproportionately did report their unemployment as a layoff. That said, there was still an increase in the share of the age 16+ population that is actively looking for work. Those in their 20s are disproportionately likely to be looking for work rather than on temporary layoff, with the peak being for those 22-24, who have an increase of 1.8 percentage points in 2020 from the previous year rate of looking for work.Screenshot from 2020-08-16 20-46-05

Changes in Labor Force Participation by Age

Perhaps the most interesting chart, and the reason I initially looked into the topic, shows changes in categories of labor force non-participation (those out of the labor force). The source for this data asks people about their major activities and their reasons for not having a job and not looking for one. Those 16 or older can be out of the labor force because of school, a disability or illness, to care for home or others, because of retirement, or they can be discouraged because they do not believe work is available for them.The data show a mixture of effects, with the most prominent being an increase in the share who are discouraged. Some other effects are more age-specific. Young people are increasingly reporting school enrollment as their reason for non-participation, but also are disproportionately increasingly non-participants in the discouraged and other categories. There were also people age 22-45 who left caregiving roles, ostensibly to work or look for work. Additionally, the retired share of those 55 and older increased.Screenshot from 2020-08-16 20-46-23It’s worth noting here that age groups can partially offset each other in these data. For example, a portion of young people in college who would prefer to work are offset by people in their 20s and 30s who would go to school if they felt confident about leaving a job.Among those age 34 or older, there is actually a decrease in the share that report disability or illness as their reason for non-participation. It’s not clear what is happening here, though one possibility is that reductions in non-COVID-related medical procedures mean fewer people were sidelined temporarily by a surgery. Another possibility is that some portion of those who reported disability or illness found work or moved into other categories of non-participation or unemployment.

Summing up the categories

Far more people have labor market effects from the shutdown than those counted as unemployed. The unemployed average 19.5 million over the four month period and BLS reports 16.3 million unemployed in July. However, counting the people who are no longer fully employed plus those who are no longer multiple jobholders yields a total of 26.8 million people on average for the four month period.Screenshot from 2020-08-17 00-39-43The data also show that those age 16 to 24 are disproportionately affected. The last table includes only those in this combined age group. About one in seven in this group, about five million people, fell out of the fully employed category or the multiple jobholder category. About half of those are unemployed and the other half are underutilized or left the labor force.Screenshot from 2020-08-17 00-46-43

Teacher summer jobs

I’m curious about how jobs are changing for the roughly 5.5 million teachers in the US. Teaching doesn’t pay as well as jobs that require similar education, but teaching traditionally includes a summer break and retirement and healthcare benefits. This post looks at trends since 1997 in how teachers use their school summer break and shows that they are becoming more likely to work during the break.

Summer break

School summer breaks show up in data on the teaching share of total hours worked. In the school year, teaching comprises 3-3.5 percent of total work hours, compared to around one percent in the summer months.


Second jobs

Panel data show that the share of teachers working a different job in the summer is pretty stable, but that teachers are more likely to be at work as teachers over the summer months (June, July, and August) than they were in the past.


Earlier start to school year

The individual monthly data show an earlier start to the school year in many places (August rather than September) shortening the summer break for teachers. In August 2019, teaching comprised about 2.5 percent of total work hours, compared to 1.2 percent in August 1997.


Poverty and age

Mismatches between the timing in life of labor income and major expenses cause unnecessary poverty in the US. Labor income (earnings from working) is most common between the ages of 25 and 54 and tends to be highest in amount for people in their 40s and 50s. In contrast, education expenses accrue disproportionately to young adults, and healthcare expenses are higher for the elderly. As another example, new parents tend to have lower wages than parents of teenagers while childcare expenses are very high in the first few years of a child’s life. These mismatches between the timing of income and expenses create financial stress and poverty.

One way to see this is to look at poverty by age:

Screenshot from 2019-12-27 19-39-38

The blue area shows the dollar amount of poverty reduced by the combined welfare programs and tax credits. The green area shows the amount of poverty remaining after taxes, welfare, and medical, work, and childcare expenses. The large blue area on the right of each graph shows that market income leaves enormous poverty for the elderly, but that this poverty is reduced to below-average rates by programs such as social security, supplemental security, and medicare. These programs are absolutely critical to poverty reduction but still leave some elderly in poverty.

Two age groups that stand out as facing higher-than-average levels of poverty by disposable income are young adults (students and new parents disproportionately) and those before social security retirement/medicare eligibility age. These groups correspond to the bump in disposable income poverty (green area) around age 20 and again around age 60. These age groups are less likely to work relative to 25 to 54 year olds and therefore have less labor income. Unlike 25 to 54 year olds, those who are younger or older have expenses higher than their income, hence the poverty caused by the mismatch in the timing of income and expenses.

Technical note:

Total poverty is defined as the dollar amount of resources that people are below their Census-defined “poverty threshold”. In the case of market income it is the total amount of money that would need to be given to people for their before-tax labor and capital income to equal their poverty threshold, according to the Supplemental Poverty Measure (SPM). Total poverty by disposable income corresponds with the SPM-poverty-rate-based amount of poverty.

How far can full employment get us?

Full employment–a version of the US economy where anyone who wants a job has one–is an important complement to welfare for poverty reduction, but ultimately not a substitute for welfare.

Why is full employment important?

Full employment means more labor income for people. A single person working full-time and full-year will have income above the poverty threshold, even if earning the minimum wage.

Typically full employment is also associated with less discrimination in hiring and with workers getting a stable share of output growth. Achieving full employment also better positions the US to provide for people who cannot or should not be working.

Are we at full employment?

The US is likely close–within a few million jobs–to full employment. Many economists had expected runaway inflation as the US unemployment rate fell below 5.5% but inflation has been well-contained. Unemployment is now around 3.5% which is near to historical lows.

What can’t full employment do?

People who live in income-sharing arrangements (for example a family) where the number of non-earners (for example: children, elderly, work-limited disabled, caregivers, students, etc) exceed the number of earners are at a major disadvantage, even under full employment, relative to those who live with more earners than non-earners. That is, a single mother of three kids will be relatively better off with full employment than without full employment, but is likely to struggle financially even with full employment.

The previous example is fairly intuitive, but what people don’t realize is just how few people actually work full-time and full-year, even in a near-to-full-employment economy.

In 2018, no part of the US had even 42% of its total population working full-time and full-year:

Screenshot from 2019-12-18 13-22-23

The map shows commuter zones–groups of counties that form local labor markets. Even with the economy near full employment in 2018, no part of the country has a majority of its population fully-employed. That is, even when the US is maximizing its number of workers, there are still large gaps in labor income over time and within any given point in time. And reaching the people not reached by labor income in full employment is precisely what welfare is for.

Data reject the argument that welfare is not needed in full employment. Non-earners comprise a large share of every country’s demographic picture, even at full employment, and non-earners are unevenly distributed within families. This combination means poverty cannot be solved through labor income alone. Existing welfare programs in the US, like unemployment insurance, provide income for earners who lose their job; separately, there are welfare programs that provide income to non-earners regardless of the state of the economy, such as low-income, low-asset elderly reached through Supplemental Security Income. Other countries show that adopting additional programs, for example a child allowance, would effectively and immediately reduce poverty from its stubbornly high rate.

Who gets credit for the wage growth of low wage workers?

Last month, I pointed out 7% wage growth at the first decile (10th percentile). The CEA noted this same figure in a recent report, claiming credit for the uptick at the bottom of the wage distribution. My piece, in contrast, had assigned credit to the state, local, and company minimum wage increases. AEI has a piece that thanks both the long expansion and the minimum wage increases.

The AEI piece points to Atlanta Fed analysis showing that states without state minimum wage hikes have seen increases in the median wage of the first quartile (bottom 25%) of wage earners. This makes a strong case that state minimum wage increases only explain part of the story. But it’s worth testing the argument that if additional states had raised their minimum wage those states would have seen larger wage gains for low-wage workers. That is, by making the wrong decision the no- and low-minimum wage states fared worse, but despite making the wrong decision, they still fared reasonably well.

A test of my claim comes from comparing total hours worked in different wage groups and in two state groups: those that raised the state minimum wage from 2016 to 2019 and those that didn’t. To better capture total wage income, aggregate hours are used instead of number of workers. In the baseline data which cover 2016, slightly more than half (55.4%) of hours worked in the US were from people in the 27 states with minimum wage increases during 2016-2019. This is how the total hours worked are distributed by wage group and state group in 2016:


Notably, the share of hours that are paid the federal minimum wage or less is low; several times as many hours of work were paid between $7.26 and $12.00.

In the latest year of data, which ends November 2019, the story has changed:


The share of work hours paid less than $12 has fallen substantially since 2016. The effect is more pronounced in states that increased the minimum wage but also apparent in those that didn’t. As a result, states that didn’t raise the minimum wage now disproportionately have low wage work relative to the states that did.

It’s worth noting that total hours of work increased by five percent over the period. The concern that raising the minimum wage will cause job losses doesn’t show up in the data. The states that raised their minimum wage maintained a fairly constant share of total hours of work, claiming 55.2% of total hours over the past year.

Looking more closely at the data, the long economic expansion does deserve some credit for wage growth at the bottom of the distribution. Importantly, states that acted to raise minimum wages better shared the economic expansion with low wage workers.

Wage growth is in the pipeline

If you haven’t looked at the US wage distribution recently, you might be surprised to see nominal growth of 7.0 percent in 2019 Q3 and 6.7 percent in Q2 in first decile usual weekly earnings. My own calculations show an even stronger 8.7 percent year over year increase in October 2019. The increase in the first decile wage seems to be coming from state, local, and company increases to the minimum wage. It is also an indication that higher wages may soon be coming to the workers in the middle of the income distribution.

First, the data:

First decile wages and growth

The dark blue lines are the BLS quarterly data on first decile usual weekly earnings for full-time wage and salary earners. The light green lines (bd CPS) calculate the same series on a monthly basis.

The BLS and bd CPS series both show first decile wage growing at its strongest rate since the late 1990s. One conventional story here is that a tight labor market, as measured by a low unemployment rate, makes it harder for employers to find qualified replacements for employees who quit, which makes existing employees more likely to get raises. As a result, wage growth is usually strongest when the labor market is tight.

But there’s a catch. BLS uses the same sample to report nominal median wage growth of 3.6 percent in Q3 and 3.7 percent in Q2. BLS data on real average hourly earnings, which adjusts nominal earnings data for inflation, shows 1.2 percent real wage growth in October (nominal wage growth of 3.0 percent and inflation of 1.8 percent). This isn’t terrible, but it’s no seven percent. With the unemployment rate historically low, why isn’t median and average wage growth stronger?

Leaning on unemployment to fully explain wage growth is clearly not working. An alternative measure of whether a labor market is tight is the employment rate for people age 25-54. This measure is currently at it’s highest level since 2007 but is still the equivalent of more than a million jobs away from its late 1990s and early 2000s level. In other words, there’s evidence that the labor market is still over a million jobs away from full employment.

Maybe it’s better to ask instead why wages are growing so rapidly at the first decile and whether it means anything for wage growth more broadly. The data show that the number of full-time workers earning $400 per week or less has fallen to 7.5 million in the latest three months from 14.7 million in the same three months of 2014. The story here seems to be higher local and state minimum wages and higher company minimum wages at large employers. Several areas and employers moved the minimum wage to above $10 per hour ($400 for a 40-hour week) during the past five years.

Higher minimum wages could eventually translate into higher median wages. For example, the first decile and median wage tend to move together:

First decile and median wage growth

In the period of faster wage growth and tighter labor markets during the late 1990s, first decile wage growth accelerated and median wage growth accelerated shortly after. If employment trends continue, it seems reasonable to expect an increase in the median wage of five percent or more in 2020.

But the current period is not the late 1990s. Business fixed investment and labor productivity growth are both particularly weak now compared to then. Negative net capital investment and increasing payrolls suggest an aggregate situation akin to fewer tools per worker, which casts doubt on forecasts of sustained real wage growth.

Technical Note

Unlike the first two charts, I use some additional steps to smooth out the third chart comparing median and first decile wage growth. I calculate the wage in each month as the wage in the previous three months combined, so October 2019 is based on combined microdata from August, September, and October 2019. Additionally, I use the Census X13as program to seasonally adjust the results. The growth rate is based on this resultant seasonally-adjusted nominal wage.

In both sets of charts, I replicate the BLS process of taking a “binned median” to reduce how much the data reflect breaks around certain round numbers (like $400 per month).

Choropleth with matplotlib and basemap

When I was using excel to make all of my graphs, the choropleth map was out of reach and particularly alluring. Later, using Stata, I figured out how to make choropleth maps, but the results were never quite right. Now, much later, excel has tools to make these maps easily, and so do you, even if you don’t have a modern copy of excel, by using python!

What follows is an example that maps 2017 GDP growth by Norwegian county.

First, GDP growth data is collected from the statistics office website. Next, shapefiles that match the regions in the data (NO_Fylker_pol_latlng) are downloaded and, using an online tool, simplified to 10% (because Norway has an extremely complex coastline). The simplified shapefiles are saved in a folder called shapefiles.

Python imports:

# Import packages 
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.patches import Polygon
from matplotlib.colors import Normalize
from mpl_toolkits.basemap import Basemap as Basemap

Data from the Stats Office as dictionary d:

# data for choropleth
d = {'Østfold': 2.5,
 'Akershus': 2.5,
 'Oslo': 2.8,
 'Hedmark': 2.2,
 'Oppland': 2.3,
 'Buskerud': 2.0,
 'Vestfold': 2.1,
 'Telemark': 0.7,
 'Aust-Agder': 2.5,
 'Vest-Agder': 0.9,
 'Rogaland': -0.4,
 'Hordaland': 1.2,
 'Sogn og Fjordane': 1.6,
 'Møre og Romsdal': 0.7,
 'Sør-Trøndelag': 2.9,
 'Nord-Trøndelag': 2.9,
 'Nordland': 1.9,
 'Troms': 2.1,
 'Finnmark': 2.0}

Code creates the map:

# Create map with lcc projection and boundaries that tightly frame Norway
m = Basemap(llcrnrlon=5, llcrnrlat=57, urcrnrlon=33, urcrnrlat=71,
            projection='lcc', lat_1=57, lon_0=15)

fig = plt.figure(figsize=(8, 16))

m.drawmapboundary()   # Create space for drawing county shapes

# read shapefiles using latin-1 encoding and call shape data "no_co"
m.readshapefile('shapefiles/NO_Fylker_pol_latlng', 'no_co', 

ax = plt.gca()   # Call the current plot area "ax"
ax.axis('off')   # Turn off border on outer edge of map

# Map values between -2 and 4 to colors in the rainbow_r colormap
cm = plt.cm.rainbow_r
norm = Normalize(-2, 4)

# For each county, select the face color and add shape to the map
for info, shape in zip(m.no_co_info, m.no_co):
    fc = cm(norm(d[info['NAVN']])) 
    ax.add_patch(Polygon(shape, fc=fc, ec='white', lw=0.5))
# Add title and colorbar legend
plt.title(f'Norway Real GDP Growth by County, 2017', fontsize=16)
cb = fig.colorbar(ax.imshow([np.array([-2, 4])], cm), shrink=0.25, pad=-0.3)


An economic revival policy that fits rural areas well

Paul Krugman’s recent opinion piece in the New York Times argues that eastern Germany and southern Italy show that the US can’t help it’s rural citizens. I’m not sure I follow his argument in the piece, but he’s right that there are several reasons to believe that most large-scale economic policy proposals do not work well in rural areas. There is, however, one policy that would work particularly well.

Large-scale economic revival policies that have garnered a lot of political attention include: antitrust (essentially breaking up large corporations), raising the federal minimum wage aggressively, and instituting a jobs guarantee. I see each as working much better in cities and suburbs than in rural areas.

Expanding and enforcing antitrust law would push back against firm monopoly and monopsony power. Studies show that the consolidation of industries has been bad for workers. In the example of hospitals, when there is only one company running all of the hospitals in a city, it becomes harder for medical care workers to quit a bad job and move to a better one. As a result, hospitals can pay workers less, offer fewer benefits, and use more contractors. The issue here is that in a rural area there’s likely only one hospital anyway (or one factory or one chicken processing plant, etc). So it wouldn’t be efficient to split the rural hospital in two. There just aren’t enough patients to justify double the number of MRI machines, administrators, and accountants.

With the federal minimum wage unchanged for nearly a decade, and currently set at a poverty rate of $7.25 for an hour of work, raising the federal minimum wage seems like a no-brainer. The pushback to a higher federal minimum wage often comes from employers in rural areas. Their argument is that many jobs in rural areas would not make sense if the minimum wage was, say, $15.00 an hour, so the increase would disproportionately create job loses in rural areas. Thankfully, many states and cities are already moving towards a $15.00 minimum wage without federal intervention. But in many rural areas, the current optimal minimum wage is probably quite a bit lower than $15.00, perhaps $10.00 or $11.00 (granted federal wage increases would be gradual, but the point is that the optimal minimum wage is generally considered to be lower in rural areas, though there are exceptions). This is because the marginal productivity of a larger share of workers in rural areas is less than $15.00 per hour, compared to workers in cities and suburbs.

The jobs guarantee elicits comparisons to the Civilian Conservation Corps, which was very active in rural parts of the US. The idea of the jobs guarantee would be to turn the unemployment office into the employment office and to offer a $15.00/hour job to anyone who wants one. For rural areas the jobs guarantee wage faces the same problem as the minimum wage proposal discussed above, but additionally, there are highly-localized pockets of deep despair and poverty in rural areas. For example, my parents live in a town of about 5,000 people that has lost more than 2,000 jobs to trade since the 1990s. Perhaps I am not being creative enough, but I’m not sure there’s enough work (without adding equipment, tools, supervision, or expertise) to support even 400 new federal jobs in the area. Vegetable gardens, picking up trash, and childcare might be able to accommodate a few hundred people, but the problem is so large and so localized that I worry about what several thousand federal workers would actually do all day in the tiny town. In the case of my parents’ town, it’s probably better to just give people the money and not worry about what they do all day than to have them carry rocks back and forth between two piles.

There is one policy that fits rural areas better than it fits anywhere else: increasing who gets ownership income.

Ownership income is the main explanation for why the super-rich are so damn rich. Bill Gates, while known for Microsoft (which benefited massively from intellectual property protections), actually owes much of his wealth to diversified investments (getting paid for owning things other than Microsoft). Rich people own an enormous portion of the country’s stocks, bonds, and real estate, and these investments offer a return that equates to an increasingly large share of national income. With labor income, which has also seen the rich (CEOs, doctors, lawyers) pull away from everyone else, time is an equalizer. Everyone gets the same amount of time (24 hours per day). But when it comes to ownership income, whoever owns the most and the best gets the most money. The majority of people don’t own anything that offers a return (to keep with the analogy, they have 0 hours per day), while the wealthiest people own tens of billions of dollars of productive assets (they have millions of hours per day).

This is not a natural outcome but the direct result of US rules, which means it doesn’t have to be that way. A combination of two policy proposals would result in a very effective way to improve quality of life for people, that works particularly well in rural areas. First, Dean Baker proposed a scrip tax that would replace the US corporate income tax with the ownership of non-voting shares (say 25%) of each company. The idea here is that corporations do everything they can to avoid paying taxes, and usually are pretty good at it. But if the treasury owns a portion of the shares of those companies, then it eliminates the need to tax them. When the company does well and makes more money so does the treasury. The scrip tax could be used to create Matt Bruenig’s proposed social wealth fund (SWF). Essentially think of this as an index fund that is owned by every person in the US equally. Each person would receive dividend payments from this fund, just like rich people do now. For a family of four, this dividend would be several thousands of dollars a year that they would get the same way rich people get their money, just by existing in a system designed to benefit them. **In his report, Matt discusses several ways to pay for the SWF; I’m personally drawn to the funding idea from Dean, which is why I combine the two ideas in this blog post.**

Why would expanding ownership income work for rural areas? First, fewer people in rural areas have ownership income now, compared to cities and suburbs. We can see this in both the Survey of Consumer Finance and the Current Population Survey. Ownership income is rare in metro areas (cities and suburbs), but is even more rare in rural areas. That said, the scrip tax/SWF would not be a transfer from cities and suburbs to rural areas, but a transfer from the ultra rich to everyone else. There are very rich people everywhere, including in rural areas.

Second, rural areas face a problem with diversity of industry. On an aggregate level, rural areas have all sorts of industries, just as cities and suburbs do. But on a local level, there is often one dominant industry, an outcome that is much less common in cities and suburbs. For example, my parents town made nylon starting in the 1930s. This was the industry that employed pretty much everyone, directly or indirectly. The main street was “Nylon Boulevard,” and the town called itself “the nylon capital of the world.” When nylon production was moved to Mexico (and then subsequently Asia and Brazil), there was no other industry for the former nylon workers to move into. The people with the means to leave did, but many people were tied to their homes (now much more likely to have an underwater mortgage and much harder to sell) and families and were subjected to a downward spiral that continues today. Had the nylon factory been in a city, the workers could have stayed in their homes but more easily switched jobs, simply because there are so many more industries in a densely populated local area (there are counter-examples here, like Youngstown, OH, but on average this is the case).

With the scrip tax and social wealth fund, 25% of those nylon-producing assets would have been owned by the fund, while the nylon workers who own shares of the fund would own a tiny portion of every industry. Thus, the result is a diversification of assets. If one geographic area or one industry gets hit, everyone in the US loses a few pennies in the value of their share of the fund, but the local area’s total set of assets are at least slightly more diversified. Their shares of the fund are much more stable relative to their local economy. Certainly the job losses would still devastate the area, but at least one source of income would be largely unchanged (or even increase, as the outsourcing resulted in higher profits for the DuPont corporation).

Lastly, prices are 12 percent lower in rural areas compared to non-rural areas. This is one of the reasons the minimum wage proposals don’t work as well in rural areas, but it is also the reason that ownership income expansion works particularly well in rural areas. Since each person in the US would own the same portion of the fund, each person, no matter where they live, would get the same amount in their dividend payment. However, this dividend payment would buy more goods and services in rural areas than it would in cities and suburbs. The difference in prices is partially because land values differ so much between rural areas and cities/suburbs. An acre in southwest Delaware is never going to be worth as much as an acre in downtown Seattle.

While there are certainly details to be worked out for any large-scale economic policy, it is silly to suggest that the failure of Germany and Italy prove there’s no policy that would help US rural areas. Norway and Alaska offer a counter-example that would work well.

Six southern US metro areas: part 6 – unemployment

My sixth blog post on mid-sized cities near the southern section of the Appalachian mountains looks at unemployment (people who do not have a job but are actively trying to find one).

The six areas of interest are: the Chattanooga-Cleveland-Dalton, TN-GA combined statistical area, the Greenville-Anderson-Spartanburg, SC combined statistical area, the Asheville, NC center-based statistical area, the Johnson City-Kingsport-Bristol, TN-VA combined statistical area, the Huntsville, AL center-based statistical area, and the Knoxville, TN center-based statistical area. See the first post in the series for more background.

The source for these results is 36 months of aggregated Current Population Survey microdata, covering January 2016 to December 2018.

Unemployment rate

In a previous post, I calculated the unemployed share of the population. This post looks at the unemployment rate, which is the unemployed share of the labor force. Over the three-year period from 2016 to 2018, the US unemployment rate averaged 4.4%. The unemployment rate varied in the six areas of interest from 2.9% in Asheville to 5.1% in Huntsville. The unemployment rate in this three-year period averages 3.5% in Knoxville, 3.9% in the Greenville-Anderson-Spartanburg area, 4.0% in the Johnson City-Kingsport-Bristol area, and 4.6% in the Chattanooga-Cleveland-Dalton area.

BLS publishes high frequency estimates of unemployment in the Local Area Unemployment Statistics report. I’ve used the multi-year averages instead, to allow analysis of why people are unemployed and for how long they have been looking for a job.

Reason for unemployment

Unemployment can be grouped into four categories, based on what people were doing before they became unemployed: 1) people who quit a job and are looking for a new one (job leavers), 2) people who lost a job and are looking for a new one (job losers), 3) people who are looking for their first job (new entrants), and 4) people who were previously not in the labor force (for example: disabled or ill, taking care of family, retired, or had simply given up hope of finding work) and are now looking for work again.

From 2016-18, nearly half of US unemployment was because people had lost a job. This “job loser” category can be the most painful, as it is perhaps the least voluntary. The person had a job, and presumably wanted to keep it, but could not, and now they are trying to find a replacement job. In contrast, the re-entrant category, which makes up 1.3% of the US labor force, can be a positive indication. From 2001 to 2014, the US labor market was particularly poor for many people, and, as a result, many decided to stop looking for work. But over the past few years, the labor market has improved, and new jobs are pulling people off of the sidelines and encouraging them to look for work again. These people show up as “re-entrants”. Job leavers can also be an indication of a strong labor market. When people are confident that they can find a better job, they are more likely to leave their current job.

In four of the six areas of interest, the unemployment rate is below the US average for the “right” reasons. That is, people are less likely to be job losers or unemployed new entrants. In Knoxville and the Johnson City-Kingsport-Bristol area, more than half of the unemployed are job leavers and re-entrants. Re-entrants are also disproportionately common in the Greenville-Anderson-Spartanburg area.

The job loser share of the labor force is at or below the national average in all six areas, and particularly low in Knoxville and Asheville. The new entrant share of the labor force is above average in Huntsville and Chattanooga. Job leavers are more common in Chattanooga and Johnson City.


Duration of Unemployment

Another important determinant of whether unemployment is particularly painful is the duration of unemployment. If people are unemployed for a short time, they may be able to rely on unemployment benefits and, in some cases, their personal savings, to survive. However, a long period of unemployment can devastate savings, exceed the period where unemployment benefits are allowed, affect mental health, and even create a situation where people begin to lose their skills.

Long-term unemployed, measured as those whose unemployment has lasted 27 weeks or more, makes up one percent of the US labor force. The long-term unemployment rate is the same in Johnson City-Kingsport-Bristol and Knoxville. In contrast, it is more common in Huntsville (1.8%) and less common in the Greenville area (0.7%), Chattanooga (0.5%), and Asheville (0.3%).

Short-term unemployment (lasting a month or less) makes up 1.4% of the US labor force, 1.5% in the Greenville area, 1.1% in Knoxville, and 1.3% in Huntsville. The unemployed populations in both Huntsville and Knoxville are more likely to be long-term unemployed than the US average.


The next blog post in the series will look at which industries and occupations employ people in the six areas. The jupyter notebook used to create the analysis above is here.