Jobs Endangered by a $15 Minimum Wage in Seattle

Both candidates in the Seattle mayoral race support an effort to raise Seattle’s minimum wage to $15 per hour.  Mayor Mike McGinn says he would even support an effort to set the minimum wage even higher.  Mayor McGinn and challenger Ed Murray are foolish if they believe that the Seattle mayor or City Council can ignore the laws of supply and demand.  A mandate that workers in lower paying occupations receive higher wages will lead to substantial job losses in these occupations.  The labor force in the city of Seattle is about one-third of the King County labor force and one fifth of the labor force in the Seattle metro area. This means that when Seattle laws make it more expensive to operate a restaurant, coffee shop, retail outlet or other business inside city limits, businesses will relocate to the suburbs and shoppers and customers will follow.

The economics of a minimum wage for a city is quite simple.  Employers in Seattle are price takers in the market for unskilled and less skilled labor.  It doesn’t matter how inelastic the demand for less skilled labor is in the aggregate, all that matters is the elasticity of demand for workers within city limits.  A large increase in the cost of hiring dishwashers or cashiers within Seattle merely shifts demand for these services to businesses outside city limits where the minimum wage is $9.15 and costs are lower.  The $15 minimum wage will destroy jobs in Seattle but will increase employment in some businesses in the suburbs.  The best substitute for shopping or dining in the city is shopping or dining in the suburbs.  Customers will be inconvenienced, unskilled workers in the city will be harmed and have to commute further to work.  However, business owners and unskilled workers in suburban areas could benefit from a $15 minimum wage in Seattle.

Mayors and mayoral candidates who support large increases in the minimum wage should also be required to specify which jobs in their cities would be endangered by their policies.  Following the International Union for Conservation of Nature which designates species as endangered, vulnerable or near threatened, I believe that politicians should acknowledge when their policies threaten the viability of certain jobs.  Politicians should also be required to use the same sort of designation to indicate the severity of the threat posed by their actions.  Politicians can make jobs extinct by raising the minimum wage so much that workers are priced out of the market for their services.

I propose that in Seattle:

  • A job is endangered if 90% of current workers earn less than the proposed $15 minimum wage.
  • A job is vulnerable if 75% of current workers earn less than the proposed $15 minimum wage.
  • A job is near threatened if 50% of current workers earn less than the proposed $15 minimum wage.

The Bureau of Labor Statistics (in its OES data) lists 637 detailed occupations in the Seattle metro area.   In 120 of those occupations, employing 27.7% (over 390,000 workers) of the workforce, the median wage less than $15 per hour.  The following tables provides examples of occupations that are most at risk due to a $15 minimum wage.

There are 16 endangered jobs in Seattle.  These jobs are endangered because at least 90% of workers earn less than $15 per hour.  The following table lists some of the most common endangered jobs.  For example, there are 25,930 food preparation and servers in the Seattle metro area and 90% earn $14.07 or less.  A $15 minimum wage will likely cause restaurants in Seattle to lose business to suburban competitors.  Other jobs on this list are endangered by information technology.  For example, as the cost of hiring hotel and motel clerks increases, more businesses will use kiosks and encourage customers to check-in online.

Endangered Jobs in Seattle

At Least 90% of Employees Earn Less   Than $15.00 per Hour

Occupation Title

Number of Workers

90th Percentile Wage

Food Preparation and Servers, Including Fast Food



Personal Care Aides






Dining Room Attendants and Bartender Helpers



Home Health Aides



Hosts and Hostesses, Restaurants and Lounges



Hotel and Motel Desk Clerks



Baggage Porters and Bellhops



There are 33 vulnerable jobs in Seattle.  These jobs are vulnerable because at least 75% of workers earn less than $15 per hour.  The following table lists some of the most common vulnerable jobs.  For example, there are 12,590 cooks in the Seattle metro area and 75% earn $14.59 or less.  A $15 minimum wage will likely cause restaurants in Seattle to lose business to suburban competitors.  Other jobs on this list are vulnerable to technological change.  For example, as the cost of parking lot attendants and ticket takers increases, more businesses will use kiosks and other devices to substitute capital for labor.

Vulnerable Jobs in Seattle

At Least 75% of Employees Earn Less   Than $15.00 per Hour

Occupation Title

Number of Workers

75th Percentile Wage

Restaurant Cooks



Food Preparation Workers



Maids and Housekeeping Cleaners



Counter Attendants, Cafeterias and Coffee Shops



Packers and Packagers



Childcare Workers



Amusement and Recreation Attendants



Cleaners of Vehicles and Equipment



Parking Lot Attendants



Taxi Drivers and Chauffeurs



Ushers and Ticket Takers



Manicurists and Pedicurists






Laundry and Dry-Cleaning Workers



There are 71 near threatened jobs in Seattle.  These jobs are threatened because at least half of workers earn less than $15 per hour.  The following table lists some of the most common threatened jobs.  For example, there are 47,390 retail sales workers in the Seattle metro area and half of them earn $12.13 or less.  A $15 minimum wage will likely cause shops and stores in Seattle that employ these workers to lose business to suburban competitors.  Other jobs on this list are threatened by technological change.  As the cost of stock clerks and order fillers increases, more businesses will use computer and information technology to substitute capital for labor.

Near Threatened Jobs in Seattle

At Least 50% of Employees Earn Less   Than $15.00 per Hour

Occupation Title

Number of Workers

50th Percentile Wage

Retail Salespersons






Stock and Material Movers



Stock Clerks and Order Fillers






Nursing Assistants






Security Guards



Landscaping Workers






Counter and Rental Clerks



Hair Stylists



Bank Tellers



Preschool Teachers



Cafeteria Cooks



Meat, Poultry and Fish Cutters






Sewing Machine Operators



File Clerks



The mayoral candidates in Seattle may think they help workers in their city who are struggling in today’s economy by advocating a $15 minimum wage.  In fact, the mayoral candidates’ policies will harm the workers they would like to help.  These candidates tell Seattle residents that if they can’t find an employer willing and able to pay at least $15 per hour for their services, they will be prohibited from working inside city limits.  A $15 minimum wage in the city will cause Seattle residents to commute to the suburbs to work in stores, shops and restaurants. The only voters and businesses that should support this silly policy are those located outside Seattle city limits.

How Long Until The Next Debt Ceiling Debate?

Everyone understands that the federal government debt ceiling will eventually be increased, but much of the policy discussion in the past week has focused on the harm from a default that could be triggered by a delayed increase in the ceiling.  Few journalists are asking by how much the debt ceiling will be increased and when Congress is likely to confront this problem again.  The debt ceiling was increased 50 times between June 1965 and 2002 and ten times in the past decade.  The following chart plots the number of years between each increase in the debt limit.


Debt ceiling increases have become larger in magnitude and less frequent.  From 1965 to 2002 the average debt ceiling increase was effective for 277 days, on average.  Over the past decade the average debt ceiling increase was effective for 380 days, an increase of 37% over the prior four decades.  Overall, 78% of increases in the debt ceiling were effective for one year or less, 50% required a subsequent increase in less than 8 months, and 25% required another increase within four months.  Only four increases in the debt ceiling were effective for at least two years.  If this Congress were to raise the debt limit comparably to the median previous Congress, it will need to address the question of another debt increase on about June 10 of next year.

Debt ceiling increases in the past decade have been more than twice as large, relative to GDP, as they were between 1965 and 2003.  The following chart plots the magnitude of each increase in the debt ceiling from 1965 to the present.

Debt2From 1965 to 2002 the average debt ceiling increase was equivalent to about 3.46% of GDP.  The past ten increases raised the debt ceiling by an average of 7.09% of GDP.  The distribution of previous debt increases is skewed; about half raised the debt limit by less than 2% of GDP and one-quarter by less than 1.2% of GDP.  If this Congress were to increase the debt limit comparably to the median previous Congress, in terms of the share of GDP, expect a $325 billion increase or enough to cover the deficit for about six months.

Many taxpayers and voters would like to avoid another government shutdown knowing that a shutdown means that monuments will be barricaded, citizens will be inconvenienced, and furloughed federal workers will receive back pay for days that they didn’t work.  While default on federal debt obligations is not an option, it may make sense for conservatives who prefer smaller government to link a larger than typical increase in the debt ceiling to tax and entitlement reform.  Half of debt ceiling increases since 1965 were effective for less than 8 months and many only covered the federal deficit for about four months.  A larger than typical increase in the ceiling would mean that another government shutdown wouldn’t occur for at least a few years.  Taxpayers should support such a large increase in the debt ceiling if it is accompanied by entitlement reform and tax reform that promotes economic growth.  If future tax and spending reform is not linked to the increase in the debt limit, expect another debate over debt, spending and taxes in 4 to 8 months when the Congress will need to raise the debt limit once again.

Note: The data on federal debt limit increases I used was found here and here.

The Decline in Construction Employment, Infrastructure Investment and the Davis Bacon Act

Reihan Salam has a smart piece in National Review Online and provides some good insights about how the construction sector is changing and why construction employment has not rebounded as it has in prior recoveries.  He notes that modular construction and technological change is likely to change the labor intensity of construction projects.  So even when construction activity rebounds, construction employment may never regain the share of total employment it reached during the housing boom of a decade ago.  Salam is correct; trends in construction employment may begin to look a bit more like manufacturing.  In U.S. manufacturing it has been quite common to see output increase despite stagnant or even declining employment because of technological change.

Salam writes that construction projects in the U.S are often inefficient and I agree, especially when it comes to infrastructure investments.  The federal regulations make public sector projects far too expensive to taxpayers even at a time when a record downturn in construction employment should mean much lower costs.  The federal government has not taken enough advantage of the considerable slack in construction employment to build and repair infrastructure.

By law, federal government projects must pay the prevailing wages of construction workers, and these wages are often union scale.  This regulation, known as the Davis Bacon Act, has artificially inflated the price of construction labor on public sector projects.  Even in states and counties  where construction employment has been depressed for the past five years, government contractors are sometimes required to pay wages in construction trades that exceed the average in the area by at least $10 per hour.

Davis Bacon wages do not rely on carefully designed samples of workers, such as the Bureau of Labor Statistics (BLS) Occupation Employment Statistics (OES) Survey to determine the wage distribution in construction trades in a local area.  Instead, Davis Bacon wages are determined in the Labor Department’s Wage and Hour Division which over-samples unions and obtains much higher construction wage estimates.  Only 6.6% of private sector workers are union members so the special treatment of unions in Wage and Hour Division surveys leads to unrepresentative prevailing wage estimates.

As an example consider Riverside County, California where the unemployment rate in July was 11.1%.  The most recent OES survey reports that the average wage for a carpenter is $27.25 per hour and 75% of carpenters earn $36.39 per hour or less in Riverside.  Yet the Davis Bacon Act mandates that federal contractors pay at least $48.43 per hour to carpenters in Riverside, in wages plus fringe benefits, on government construction projects.  The Davis Bacon prevailing wage for carpenters is $37.35 per hour and prevailing fringe benefits are $11.08 per hour.  (The BLS National Compensation Survey reports that the average cost of fringe benefits is $10.52 per hour nationwide.)  Similarly inflated compensation is required for brick masons, electricians, plumbers and equipment operators.

Although reasonable economists can disagree about the level of public spending on infrastructure, ideally we should make more public investments in infrastructure during a downturn when opportunity costs are lower.  The Davis Bacon Act interferes with such a common sense policy.  Conservatives should have proposed a repeal of Davis Bacon in the waning months of the Bush Administration or early in the Obama Administration as a way to more efficiently utilize slack resources during the recession.

Requiring taxpayers to pay inflated prices to construction labor makes as much sense as paying inflated interest rates to government bondholders even though market interest rates have declined.  The federal government currently pays lower interest rates on government debt because it pays market rates on new debt issues.  Fortunately there is no equivalent to the Davis Bacon Act requiring that the federal government pay inflated non-market interest rates to protect retirees and pension funds that hold government bonds.  It is time to change the law so that taxpayers can also pay market wages on construction projects.

Do Stand Your Ground Laws Deter Violent Crime?

Florida’s “stand your ground” law is coming under attack after Saturday’s acquittal of George Zimmerman for murder and manslaughter charges in the death of Trayvon Martin.  Stand your ground laws expand the legal justification for the use of lethal force in self-defense.  Since Florida passed its law in 2005, 20 other states passed similar laws between 2006 and 2009.  A recent paper by Cheng Cheng and Mark Hoekstra of Texas A&M University presents empirical evidence that these laws “increase homicides by a statistically significant 8 percent, which translates into an additional 600 homicides per year” in the states (including Florida) that passed “stand your ground” laws from 2005 to 2009.  Moreover, they find that these laws do not appear to deter burglaries, robberies or aggravated assaults.

The Cheng and Hoekstra study compares crime rates before and after laws were passed in “treatment” states to crime trends in the “control states” where there was no change in the law.  This is a standard practice in social science policy evaluation studies.  Cheng and Hoekstra control for differences in policing and law enforcement, changes in economic conditions, and changes in the demographic composition of states and provide several different empirical specifications of their model.  Their careful study suggests that their key results are robust to alternative specifications of the empirical model.

I remain skeptical of the Cheng and Hoekstra results for three reasons:

  • The raw decline in homicides per 100,000 residents between 2000 and 2011 is very similar for states that did and did not enact “stand your ground laws” from 2005 to 2009.  The Cheng and Hoekstra result relies on percentage changes in homicide rates.  The percentage decline was smaller in “stand your ground” states because they had higher homicide rates in 2000.
  • Cheng and Hoekstra did not present results for rapes.   The raw decline in the number of rapes per 100,000 residents between 2000 and 2011 is larger for states that enacted “stand your ground laws.”  In other words, a simple comparison of crime rates suggest that “stand your ground” laws prevented more than 1500 rapes per year in the 21 states that passed these laws.
  • Homicide rates and trends differ substantially across states which makes this kind of policy evaluation quite difficult.  For example, I compare homicide trends in Nevada, ground zero for the real estate bust, and North and South Dakota where the economy has been booming because of natural gas and oil exploration.  A consistent estimate of the impact of “stand your ground” must control for cross-state differences in homicide levels and trends.

Homicides Declined by Similar Amounts in States That Did and Didn’t Pass “Stand Your Ground” Laws

The following chart compares annual population-weighted average homicide rates across two groups of states, those that passed “stand your ground laws” between 2005 and 2009 and those with no change in the law since 2000.


 In states that would eventually pass “stand your ground” laws the average homicide rate was 27% higher in 2000 than it was in states that would not change their law.  The reasons for these cross-state differences in homicide rates are likely to persist regardless of a change in the law.

Homicide rates fell in both groups of states, but the timing of the decline was somewhat different.  Homicide rates started declining about two years earlier in states with no change in the law.  A simple comparison of three-year average homicide rates from 2000-02 to 2009-11 indicates:

  • The average homicide rate in states that passed a “stand your ground law” declined by 0.79 per 100,000 residents between 2000-02 and 2009-11.
  • The average homicide rate in states with no change in the law declined by 0.89 per 100,000 residents between 2000-02 and 2009-11.

Homicide rates fell less in states that passed a “stand your ground law.”  The magnitude of the difference is about 1 homicide per year per 1,000,000 residents.  There were 7,272 homicides in 2011 in the 21 states that passed “stand your ground.”  The simple calculations above suggest that there might have been 135 fewer homicides in these 21 states had the law not been passed.  This difference is neither practically or statistically significant.

Cheng and Hoekstra estimate a much larger effects of the law, about 4.5 times as large as the raw difference in homicide rates presented above.  They contend that “stand your ground” was responsible for an additional 600 homicides in 2011.  Their effect is larger in part because the percentage decline in homicide rates was larger in states that did not pass the law (because these states already had a lower homicide rate).  In addition their methodology attempts to control for other factors that could explain homicide trends.  It is not clear which of these factors caused the regression-adjusted results to be larger than the simple comparisons of mean homicide rates.  They also attempt to exploit differences in the timing of changes in the law to measure it’s effects.  Unfortunately 13 of the 21 states that passed “stand your ground” did so in 2006 and 4 more in 2007, so it is difficult to ascertain whether the impact of the law occurs with a lag.

Rapes Declined More in States That Passed “Stand Your Ground” Laws

Cheng and Hoekstra attempted to estimate the deterrent effect of “stand your ground” on burglary, robbery and aggravated assault crime rates.  They found no effect.  They did not, however, attempt to measure the impact of these laws on rapes.  Had they done so they would have seen the following pattern:


The rate at which rapes occurred was 28% higher in 2000 in states that would eventually pass “stand your ground” relative to the states that would not change their law.  Rapes declined in both groups of  states between 2000 and 2011.  A simple comparison of three-year average rates from 2000-02 to 2009-11 indicates:

  • In states that passed a “stand your ground law” rapes declined by 5.9 per 100,000 residents between 2000-02 and 2009-11.
  • In states with no change in the law rapes declined by 4.7 per 100,000 residents between 2000-02 and 2009-11.

The decline in rapes was more substantial in states that passed a “stand your ground law.”  The magnitude of the difference is about 1.2 rapes per year per 100,000 residents or more than 10 times the impact of the law on homicides as described earlier.  Taken at face value this implies a reduction of 1,579 rapes due to the “stand your ground” law in 2011.  Despite the larger decline in rapes in “stand your ground” states, the percentage decline is slightly larger (16% compared to 15.6%) in states with no change in the law because these states started with a lower crime rate in 2000.

State Difference in Crime Trends

The purpose of the policy evaluation study is to impute state-specific crime rates but for the passage of the “stand your ground” law.  This is an extremely difficult task and requires an understanding of why crime trends differ from one state to another.  As the previous charts have shown murders and rapes were generally declining in states regardless of whether a “stand your ground” law was passed.  It is clear that “stand your ground” was more likely to be enacted in higher crime rate states.   A similar correlation between state-specific crime trends and the passage of “stand your ground” would invalidate the empirical approach used by Cheng and Hoekstra to evaluate this law.

As an example of the difficulties encountered when comparing cross-state differences in crime trends consider the case of Nevada and the Dakotas.  The following chart shows that the homicide rate in Nevada fell by 43% between 2006 and 2011 after increasing from 2000 to 2006.  In contrast the homicide rate in North and South Dakota (combined) increased by 265% between 2000 and 2011.


South Dakota and North Dakota passed stand your ground laws in 2006 and 2007 while Nevada did not change its law.  Clearly there are many other important differences between these states and the economic conditions in these states have changed substantially over the past decade.  Nevada suffered more than any state from the real estate collapse and recession.  Between 2006 and 2011 its unemployment rate jumped from 4.2% to 13.2%.  Between 2006 and 2011 North and South Dakota benefitted economically from an energy boom with some of the lowest unemployment rates in the U.S.  The boom also brought big population growth and an increase in crime rates.

The comparison of Nevada and the Dakotas illustrates why one should be skeptical of studies that compare percentage changes in homicide rates across states.  In the Dakotas in 2000 there were only 0.8 homicides per 100,000 residents per year.  Thus even a massive percentage increase in the homicide rate in the Dakotas led to an increase of about 2.2 homicides per 100,000 residents per year by 2011.  In contrast the 43% decline in the homicide rate in Nevada between 2006 and 2011 led to a decrease of about 3.9 homicides per 100,000 residents per year.  It makes more sense for a policy evaluation study to compare changes in homicide rates per 100,000 residents than percentage changes in state-specific rates, because even modest percentage declines in high crime rates states will save more lives than very large percentage changes in lower crime rate states.  Finally, it is worth noting that even after the large percentage increases in homicides in the Dakotas and the more modest percentage declines in homicides in Nevada, the homicide rate in Nevada remains 74% higher than in the Dakotas in 2011. 


Although it is tempting to evaluate the impact of “stand your ground” laws by comparing the trend in homicides in states that passed laws since 2005 to trends in states with no change in the law, one should do so with great caution.  First, homicide rates differ widely among states and there are equally important underlying differences in crime trends among states.  “Stand your ground” laws were passed in states with higher violent crime rates, on average.  This means that roughly equal declines in homicides per 100,000 residents in states with and without changes in the law translated into smaller percentage reductions in homicide rates in “stand your ground” states.  I consider this to be weak evidence that “stand your ground” laws increase homicides because it presumes that in the absence of the law homicides per 100,000 residents would have declined proportionately more in the states with the highest homicide rates.

This post also presented new evidence that rapes per 100,000 residents declined more substantially in states that passed “stand your ground” laws.  While I am skeptical of cross-state comparisons in crime trends, if one believes the evidence that “stand your ground” increased homicides one should also believe that rapes were reduced by more than 10 times as much as the increase in homicides after the passage of “stand your ground.”

Washington DC’s $12.50 Living Wage Will Harm Working Class Residents of the District

Yesterday the Washington D.C. City Council approved a $12.50 living wage that would apply to retail establishments operating in spaces with at least 75,000 square feet if they are owned by companies with at least $1 billion in annual sales.  If Mayor Gray signs the bill the cost of operating big box retail establishments in the District will increase substantially.  The living wage ordinance is bad news for workers and consumers in Washington, D.C.  There is no doubt that the ordinance will reduce job and shopping opportunities for people who live and work in the District.  The biggest losers will be working class District residents who would have worked and shopped at the retail establishments that would have been built but for the ordinance.

The Washington Post’s Wonkblog erroneously claims that: (1) “the literature” suggests that raising minimum wages in cities has no negative effect on employment and (2) large corporations “are better equipped to absorb higher labor costs.”  Businesses are not sponges that exist to absorb costs – they operate to generate income for their owners, investors and shareholders.  If Washington D.C. adopts laws that raise costs for retailers, there is no shortage of alternative locations elsewhere in the U.S., Mexico, or China to locate retail outlets.

Wonkblog also believes that big box retailers can afford a nationwide $12 minimum wage because some Berkeley professors said so.  Apparently a $12 wage would lead to a “pretty negligible” increase in costs equivalent to 1.1% of revenue.   It is clear that Wonkblog does not understand how razor-thin profit margins are for big box retailers as they compete with each other and with online retailers.  Combined profits at Wal-Mart, Target, Home Depot, Lowes and Best Buy last year were 3.6% of their combined annual sales.  An increase in costs equivalent to 1.1% of revenue would put a massive dent in the profitability of brick and mortar retailers when one considers that online retailers would be largely immune to the higher costs caused by a $12 minimum wage.

The D.C. living wage will kill jobs in the District because retailers in D.C. compete with suburban retail outlets.  According to the Department of Labor the most common jobs in the retail sector are “cashiers” and “retail salespersons.”  The D.C. ordinance would require big box retailers in the District to pay wages (even to beginning entry-level employees) that are much higher than wages paid by other retailers in the metro area.  The most recent data from the Bureau of Labor Statistics Occupational Employment Survey indicates that 75% of cashiers in the Washington D.C. metro area earn less than $11.65 per hour and approximately 64% of retail salespersons in the Washington D.C. metro area earn less than $12.50 per hour.  In other words, the ordinance will cause big box retailers to continue to locate stores in suburban Virginia and Maryland and avoid the District.  Wealthier District residents, who rely less on public transportation, are inconvenienced less by commuting to the suburbs to shop.  Working class District residents will find it much less convenient and much more expensive to commute to the suburbs to work as a cashier or retail salesperson or shop at a discount store.

The ordinance will keep big box retailers from locating in the District and cause the few big box retailers already in D.C. to leave for the suburbs.  Washington, D.C. has 230 convenience stores and 173 liquor stores but only one big box discount department store.  The number and type of retail establishments in the District with at least 100 employees are listed below (These data are from the Census Bureau’s County Business Patterns and Zip Code Business patterns):

Retail Establishments With At Least 100 Employees in Washington, D.C.
Supermarkets 21
Non-discount Department Stores 2
Baked Goods 2
Electronics/Television 2
Discount Department Stores 1
Home Center 1
Computer 1

In contrast, Davidson County Tennessee (Nashville), with approximately the same population as Washington, D.C. has 22 supermarkets, 6 discount department stores and 5 non-discount department stores with at least 100 employees.    Nashville residents have many more options when it comes to shopping for discounts than D.C. residents.  The living wage ordinance will mean that District residents will continue to be deprived of discount shopping opportunities in their own neighborhoods.

Let’s hope that Mayor Gray has the courage to veto the living wage ordinance.  It is bad public policy that harms the working class residents of the District.

Public Debt, Growth and Exaggeration

A Working Paper released earlier this week by Herndon, Ash and Pollin (HAP) of the University of Massachusetts identified some data processing errors in the widely cited 2010 paper “Growth in a Time of Debt”, by Reinhart and Rogoff (RR).  These errors cast doubt on some of RR’s conclusions.  Unfortunately, in the Working Paper and an April 17th opinion piece in the Financial Times by Robert Pollin and Michael Ash, claims about the empirical relationship between public debt and growth are exaggerated.  HAP’s conclusion that countries with public debt ratios in excess of 90% grow no more slowly than countries with public debt ratios of 30% to 90% is incorrect and based on an inappropriate statistical test.

RR’s thesis is that there is a threshold for the debt-to-GDP ratio beyond which a country’s growth rate drops markedly. Although their previous work suggested that debt thresholds are “importantly country-specific”, their 2010 paper suggested that for many countries the debt threshold was about 90% of GDP.  A conclusion from the simple empirics in their 2010 paper is that a country’s growth rate is likely to drop once its public debt exceeded 90% of its GDP, but debt ratios below 90% had no significant impact on growth.

The following Table is based on the corrected RR data used by HAP in their Table 4.  It shows an inverse relationship between public debt to GDP ratios and GDP growth.  Growth rates are significantly slower for countries with debt ratios between 30% and 90% relative to countries with debt ratios less than 30%.  There is no significant difference, however, in the growth rates of countries with debt ratios in the 30% to 60% and 60% to 90% ranges. 

Debt /GDP

Avg. Growth

Avg. Growth Relative to <30% Debt/GDP



< 30%






-1.05%   (5.74)




-1.08%   (5.43)




-0.99%   (3.93)




-2.01%   (6.41)




-1.77%   (4.93)




-2.61%   (4.80)


In the Financial Times Pollin and Ash state: “Using the Reinhart/Rogoff data, we found that the average GDP growth rate for countries carrying public debt levels greater than 90 percent of GDP was either comparable to or higher than those for countries whose debt ratios ranged between 30 percent and 90 percent.”

This assertion is based, in part, on the statistical test described in the notes to their Table 4.  HAP tested the joint hypothesis that the 3.19% growth rate in countries with a debt ratio of 60% to 90% and the 2.41% growth rate in countries with debt ratios of 90% to 120% are both equal to the 3.09% growth rate for countries with debt ratios of 30% to 60%.  HAP fail to reject this joint hypothesis, with a p-value of .11. 

HAP conducted a statistical test that is inappropriate for evaluating the RR claim of a debt ratio threshold at 90%.  RR do not claim that debt ratios of 60% to 90% are associated with lower growth rates than debt ratios of 30% to 60%.  It is unclear why HAP conducted a joint test and compared growth rates among countries with moderate debt.  HAP failed to find significant differences in growth rates among countries with public debt ratios from 30% to 120% because they conducted a joint statistical test that included a hypothesis unrelated to the 90% threshold.

Using the same data, I conducted alternative statistical tests of the Reinhart-Rogoff claim that debt ratios in excess of  90% reduce growth compared to debt ratios in the 30% to 90% range:

  • The 2.41% growth rate in countries with debt ratios of 90% to 120% is significantly lower (p-value=.041) than the 3.12% growth rate in countries with debt ratios of 30% to 90%.
  • The 2.17% growth rate in countries with debt ratios of 90% or more is significantly lower (p-value=.002) than the 3.12% growth rate in countries with debt ratios of 30% to 90%.

In the countries and time periods used in the HAP study, average growth rates were significantly lower when debt ratios were above 90%.   This empirical finding merely reflects an association between growth rates and debt/GDP.  The threshold effect for debt ratios that RR describe is weakened by the adjustments and corrections noted by HAP.  However, it is inaccurate to say that average growth rates in countries with debt ratios in excess of 90% are comparable or higher than growth rates in countries with 30% to 90% debt ratios.

The data used by RR, as corrected by HAP, show that growth is significantly slower in countries with debt ratios in excess of 90% compared to countries with debt ratios of 30% to 90%.  Growth rate differences around a 90% debt threshold are less dramatic after making the corrections suggested by HAP.  Although simple comparisons of average debt ratios and growth rates across countries and over time are illustrative, at best, the HAP critique has not changed the evidence; high public debt is associated with slower growth.

The New 39.6% Tax Bracket Will Directly Impact 9.5% of Aggregate Personal Income

I estimate that the new top marginal income tax rate of 39.6% on annual income in excess of $400,000 per year will have a direct impact on about 7/10 of one percent of taxpayers.  However, the income earned by these taxpayers in excess of the $400,000 threshold accounts for about 9.5% of aggregate personal income.  Had President Obama prevailed in the fiscal cliff negotiations, and the marginal income tax rate increased for incomes in excess of $200,000, higher marginal rates would have directly affected about 2.6% of taxpayers.  The income earned by taxpayers in excess of the $200,000 threshold accounts for about 14.7% of aggregate personal income.

Changes in marginal tax rates have incentive effects as well as indirect effects on workers and consumers not subject to the higher marginal rates.  The economic debate over raising marginal tax rates on high income earners centers on the magnitude of the disincentives for work, saving, investment, and job creation caused by higher marginal tax rates.  As high income earners change their behavior in response to the reduced incentives to earn income, all consumers and workers will be impacted by the new top marginal rate.

Unfortunately there have been few empirical studies that have focused on income earners directly impacted by the 39.6% marginal rate.  This is because less than one in one hundred earners generate enough income to pay the new top marginal rate.  Second, little information about these earners is known because most publicly available databases protect the confidentiality of survey respondents by “top coding” personal income.  When the income variable in a database is “top coded” it is impossible to tell whether an individual’s annual income is $500,000 or $500 million.  Finally, while the income generating activities of middle-income earners is typically summarized by annual hours of work, the productive activities of high income earners are less likely to be captured by surveys that merely report hours and weeks worked per year.  High income earners engage in risk-taking and entrepreneurial activities that may be more or less responsive than hours worked to changes in marginal tax rates.  In other words, although there are many economics studies measuring how the hours worked by low and middle-income earners respond to changes in marginal tax rates, these studies tell us little about how income earners will change their investment and risk-taking behavior now that the top income tax rate is 39.6% not 35%.  Whatever the incentive effects of this new tax rate will be, they will be felt by all Americans not just taxpayers earning in excess of $400,000 per year.

Note: My calculations are based on the summary statistics presented by economists Emmanuel Saez and Thomas Piketty, using the 2010 Statistics of Income data from the IRS.  These data are not top coded and provide the best information about high income earners.  Unfortunately tax returns data have no information about how many individuals earned income in the taxpaying household, how many hours were worked, or how much risk-taking and entrepreneurial activities were taken by the household in the tax year.  Saez and Piketty reported that in 2010, excluding capital gains from income, the top:

  • ½ of one percent of tax returns reported annual income of $488,000 or more and accounted for 13.4% of all personal income.
  • 1% of tax returns reported annual income of $336,000 or more and accounted for 17.4% of all personal income.
  • 5% of tax returns reported annual income of $148,000 or more and accounted for 33.7% of all personal income.

Using these summary data, and the conditional means of income reported by Saez and Piketty, I fit a simple model for the top percentiles of the empirical distribution of personal income in order to estimate the fraction of aggregate personal income above the $200,000 and $400,000 thresholds reported above.

The 6.5% Solution? Why the Unemployment Rate is an Unreliable Target for the Fed

The Federal Reserve Board is making a mistake by using the unemployment rate as a target for monetary policy.  I write this as a labor economist not an expert in monetary policy.  My criticism is not about whether the Federal Reserve should have a dual mandate.  Instead, I question whether the official unemployment rate is the best measure of labor market activity that Ben Bernanke and the Fed could use as their target.

Last week the Federal Reserve Board said that it would continue quantitative easing and keep interest rates low by purchasing $85 billion in Treasury securities per month at least until the official unemployment rate falls to 6.5%.  The Department of Labor reports six different measures of labor underutilization denoted U-1 through U-6.  The official unemployment rate, U-3, only include jobless workers who are willing and able to work and actively searched for a job in the past four weeks.  As Gary Becker noted on his blog, the unemployment rate is a flawed indicator of the state of the labor market because jobless workers may have become discouraged and stopped looking for work

Unemployment measures are sensitive to movements of jobless workers from the official category of “unemployed” to jobless and non-employed because they have, at least temporarily, stopped looking for work.  This means that the unemployment rate can fall for the wrong reasons; i.e. because unemployed workers gave up their job search and not because more people found jobs.  Consequently most labor economists use some variation of the employment to population ratio as a more robust measure of labor market activity.  At Welch Consulting we have constructed such an employment index that corrects for changes in the age distribution of the population and the difference between part-time and full-time employment.  The Fed should consider using a similar employment index as their labor market policy target.

For example, it appears that more than half of the decline in the unemployment rate in the past two and a half years, from 9.6% in May 2010 to 7.7% today, has been for the wrong reasons.  Since May 2010 the labor force participation rate has fallen from 64.9% to 63.6%; the adult population increased by 6.68 million and the labor force grew by only 1.08 million.  In other words there are 5.6 million more adults classified as “Not in the Labor Force” compared to May 2010.  Some of these adults are discouraged workers (and therefore counted in U-5 and U-6).  Others gave up looking for work more than twelve months ago and are not counted in any official Labor Department measure of underemployment.  Moreover, the employment to population ratio is 58.7% today, the same as it was in May 2010.  This means that in the past two and a half years employment has grown just in proportion to the adult population.

However, the official employment to population ratio reported by the BLS understates the gains in the labor market in the past 30 months for two reasons.  First, a higher percentage of jobs today are full-time than in May 2010.  Second, the aging of the workforce means that a larger fraction of adults are leaving the labor force for retirement.  The Welch Consulting Employment Index indicates that aggregate employment has grown about 1% faster than population after adjusting for the aging workforce and gains in full-time employment.  In contrast a decline in the unemployment rate from 9.6% to 7.7% over the past 30 months would have resulted in a 2.1% increase in employment had the labor force participation rate remained at its May 2010 level.  Consequently more than half of the apparent gain in the unemployment rate over the past 30 months is due to people leaving the labor force as they gave up their job search.

The last time the U.S. unemployment rate was 6.5% was October 2008.  Ben Bernanke has indicated that monetary policy could change once the unemployment rate drops to 6.5%.  Unfortunately, the difference between the health of the labor market today and in October 2008 is grossly understated by the 1.2% higher official unemployment rate today.  For example, the Welch Consulting Employment Index has declined by 4.25% since October 2008.  This means that in order to reach the same labor market activity as in the fall of 2008 full-time equivalent employment would have to increase by 4.25% (holding constant the adult population) and more than that as the population grows. Put simply the economy is short 6.1 million full-time jobs relative to October 2008.

The Fed does not expect the unemployment rate to decline to 6.5% until the end of 2015.  But changes in fiscal policy could cause the unemployment rate to drop rather suddenly, not because people find jobs, but because the long-term unemployed may stop looking for work.  What could cause this change?  Congress and the President could curtail the EUC2008 extended unemployment insurance program which allows unemployed workers to continue collecting unemployment insurance benefits for up to 99 weeks (in some states and some cases).  Typical benefits expire after 26 weeks, but the EUC2008 program has extended benefits for many jobless workers for at least 78 weeks over the past four years.  Currently about 2.2 million long-term unemployed workers collect EUC2008 benefits.  If these benefits are curtailed and 2 million of these jobless workers stop looking for work, the unemployment rate would immediately fall to 6.5%.

It is unlikely that all recipients of extended benefits will stop looking for work immediately after their EUC2008 benefits expire.  But once active job search is no longer a prerequisite for collecting benefits many unemployed workers will curtail their search, leave the labor force and lower the unemployment rate.  Once the EUC2008 program ends, it is quite conceivable that the unemployment rate will decline to 6.5% within a year despite a very weak labor market for many workers.

I leave it to macroeconomists and monetary economists to debate whether more quantitative easing can help stimulate real output and employment growth.  However, the Fed is wrong to link monetary policy to an unemployment rate that can increase when the labor market improves (and people resume their job search) and decrease when the labor market weakens (and people give up their job search).  Looking ahead it is more important how the economy reaches a 6.5% unemployment rate rather than the rate target itself.  The Fed’s policy guidance would be easier to comprehend if a better target was used for labor market activity.  A properly constructed employment index, that accounts for the changing demographics of the workforce, provides a much more accurate measure of the labor market’s strength.

Private Sector Hiring is Slowing Down

According to the Bureau of Labor Statistics JOLTS data (Job Opening and Labor Turnover Survey), private sector hiring, while still ahead of 2011, has slowed.  The JOLTS data record information on the number of persons hired by establishments and not merely total employment.  Thus the JOLTS data can provide information on whether companies are hiring new workers and replacing workers who have left their jobs or whether they are holding back on hiring.

According to the JOLTS data there were 12.73 million workers hired by private sector employers from August through October in 2012.  This represents only a 1.34% increase from the same three months in 2011.  The increases for 2009-2010 and 2010-2011 were 5.03% and 6.92% respectively, for the same three-month period.  Moreover the number of workers hired remains 16% lower than it was in 2007, before the recession.  Hiring in goods-producing industries (mining, construction and manufacturing) actually fell by 4.3% in the past year to the lowest level since 2009.

The slowdown in hiring may be due to uncertainty about the economy, including the budget negotiations in Washington.  Another sign that economic uncertainty may be impacting the labor market is that the pace at which workers are quitting their jobs has slowed down as well.

The number of workers quitting their jobs is an important indicator of how sure workers are that they can find a new higher-paying job.  When the labor market is booming more workers are willing to quit their jobs for better opportunities.  From 2009-2010 and 2010-2011 the number of private sector workers who quit their jobs increased by 12.2% and 9.7% respectively (based on data from August through October of each year).  In the past year the number of workers quitting their jobs increased by only 1.5% and remains about 26% below pre-recession levels.

The non-farm total payroll employment data from the BLS indicate that the economy has added about 136,000 private sector jobs per month over the past six months – or just enough to keep pace with population growth.  The JOLTS data indicate that the pace at which companies are hiring workers and the pace at which workers are quitting their jobs has slowed, after two years of solid increases in both hires and quits.

The labor market recovery is still fragile.  Both employers and workers may be holding back on decisions awaiting the outcome of Federal budget negotiations and the resolution of uncertainty about tax rates and government spending.  If the policy compromises by Congress and the President raise the cost of doing business, including the hiring and retention of workers, the gains in employment we have seen over the past two years could be reversed in 2013.

1.7 Trillion Weeks of Unemployment Benefits

In the past four years Federal and state unemployment insurance programs paid about 1.7 trillion weeks  (32.7 million years) of unemployment insurance benefits to jobless workers as they continued their job search.  32.7 million years is a remarkably long time period that is usually reserved for events measured on the geologic time scale (South America fully detached from Antarctica about 32.7 million years ago during the Oligocene Epoch).  Unemployment benefits were paid to an average of nearly 8.2 million workers per week, every week, for the past four years.  The unemployment insurance rolls have been quite high for an unusually long time because of the depth of the recession, the weakness of the recovery and because Congress and the President extended unemployment benefits so that job losers could collect benefits for up to 99 weeks. 

More generous benefits undoubtedly provided greater financial support for job losers and their families, but also encouraged jobless workers to be more selective in their job search and remain unemployed longer.  Many Democrats and Keynesian economists view the unemployment insurance program, food stamps and other social safety net programs as economic stimulus.  On the other hand conservatives, such as Casey Mulligan of the University of Chicago, argue that the work disincentives of the unemployment insurance program and other safety net and entitlement programs increased the depth and length of the recession.

One way to quantify the opportunity cost of providing unemployment insurance benefits to approximately 8.2 million jobless workers per week is to consider how many employees could have been hired using those resources.  Although unemployment insurance benefits vary by state, the typical weekly benefit is about one half of a worker’s previous weekly wage.  This means that the cost of insuring 8.2 million jobless workers per week is about the same as the wage and salary costs of employing 4.1 million workers per week. 

Many economists have complained that the government stimulus didn’t include enough investment in infrastructure or purchases of goods and services.  Our representatives in the Federal government chose to pay people to search for work rather than employ them directly for public works projects.  But how many roads, bridges, schools and other valuable public sector investments could have been completed instead of paying for 1.7 trillion weeks of job search?  Instead of paying half of the typical weekly salaries of 8.2 million people looking for work each week, we could have instead:

  • Paid the salaries of every worker employed in the construction and repair of streets, highways and bridges for the next century
  • Paid the salaries of every elementary and secondary school teacher in the U.S. for four years.
  • Paid the salaries of all workers in the motor vehicle (and parts) industry for two decades.

Democrats and Keynesian economists lament that state and local government employment has fallen 1.8% over the past five years instead of the 3.9% growth from 2002 to 2007.  The relatively small decrease in state and local government payrolls pales in comparison to the cost of jobless benefits over the past four years.  The money paid to unemployed workers per year over the past four years is equal to about 1/5 of the annual salaries of all state and local government employees combined.

Hoover Dam, The Grand Coulee Dam, LaGuardia Airport, The Lincoln Tunnel and many other public works projects were built during the Great Depression when many of the workers on these projects had few other job options.  The economic approach to dealing with the 2008-2009 recession has been quite different.  In 2013 we will reach 2 trillion weeks of unemployment benefits paid since the recession began.  When the history of this recession and recovery is written it will be clear that we did not use this time of excess capacity and idle workers to re-build and re-tool our infrastructure.  We will not be able to point to the dams, bridges, highways, schools and hospitals that were built during the recovery even though about two million construction workers lost their jobs after the residential real estate market collapsed and many of them are still out of work.  Instead the approach of this Administration and this Congress has been to pay people who lost their jobs to look for work, even though many of the jobs that were lost in the recession are no longer there.

%d bloggers like this: