Tuesday, October 14, 2014

Alison Grime’s right to privacy?

Alison Grimes continues to refuse to state whether she voted for Barack Obama.  She claims a right to privacy because the Constitution guarantees a secret ballot.

My comments:

Comment One:  Alison Grimes is running for office as a Democrat.   Typically leaders when they are running for office let people know where they stand.   She has no legal obligation to announce her vote.  Candidates typically waive this privacy right.

Comment Two:  The right to a secret ballot protects people from persecution.  The Senate candidate from Kentucky does not need to be protected from prosecution.  

Comment Three:  President Obama is being vilified in a personal way that differs from the way politicians are typically treated.   Alison Grime’s refusal to state whether she voted for President Obama plays into the Republican narrative that President Obama is different – perhaps even evil.

Comment Four:  I have differences with President Obama but let’s face it he is not really as liberal as depicted.   Alison Grimes whiffed on a lob.   Here is a response:

Did you vote for President Obama?

Yes but this does not mean I support everything he has done.

I can’t imagine what America would be like if John McCain had won the 2008 election and the economy continued on the course that existed then.

Unlike my opponent I want to modify and not repeal the ACA.   Repeal of ACA would lead to many (give number) Kentuckians losing their health care.

Yes I voted for President Obama.  If in the future a Republican gets elected I will work very closely with the new President to help Kentucky and the nation.    My opponent chose to put his partisan agenda ahead of the state and national interest.   The results have been tragic. 

Comment Five: Her inability to state that she voted for a two-term Democratic President is demoralizing.    She is going to deflate her own turn out, especially among blacks.

Comment Six:  Alison Grimes, like Hillary, believes that one can run for office and not tell the people where she stands.     Where does Hillary stand on Keystone, on Social Security COLAs, on student loan debt proposals on charter schools, ……..  The only way we will find out is if strong candidates  -- A-list candidates – challenge her.

Comment Seven:  Alison Grimes despite her relative lack of experience was not challenged for the Senate nomination in Kentucky.  She was hand picked by the Clintons and is a child of privilege.   This is what happens when nepotism guides the candidate selection process and you don’t test your candidates.

Concluding thoughts:  Most Democrats don’t agree with President Obama on everything.  This is a main difference between Republicans and Democrats.   President Obama is a good man who took on a tough job and should not be vilified.   Alison Grime’s inability to state she voted for the President plays into a Republican narrative that the President is evil.  It is demoralizing.  

Paint Kentucky red. 

Monday, October 13, 2014

Thoughts on the Economic Costs of Pandemics

So it is clear, at least to the Republicans, that President Obama’s approach to Ebola has been lackadaisical.   The truth is that the government generally favors a go-slow approach to disaster preparedness especially when the response could disrupt the economy in some way. 

Most economists believe that policy actions should be based on a cost benefit analysis.   Several years ago the Congressional Budget Office estimated the economic costs of a pandemic.  

It is not clear to me that cost benefit analysis should guide our response to potential disasters, especially when worst-cases costs are much larger than likely costs.

The CBO found that the likely cost of pandemics were small.   The CBO report did not contain policy recommendations but this view continues to guide government policy.

Below I discuss the CBO view of pandemics.

A Discussion of the Report:

The CBO provides an assessment of the likely impact of an avian flu outbreak on the U.S. economy.  The CBO forecasts the likely impact of two pandemic scenarios on GDP:

(1)  A severe epidemic similar in size and scope of the 1918 pandemic
(2)  A milder scenario that resembled pandemics in 1957 and 1968.

The CBO found that the severe pandemic could produce a short run impact on the worldwide economy similar in depth and duration to that of an average postwar recession in the United States.  The CBO also found that a milder pandemic would have a much smaller impact, “which might be indistinguishable in the macroeconomic data.”

The CBO projects that in a severe pandemic 30% of workers in each sector (except for the farm sector) would get ill and that 2.5% of these workers would die.  The death rate is assumed to be lower in the farm sector because of the lower social interaction in this sector.  This disease rate and death rate is consistent with the death of more than 2 million people in the United States alone.  The CBO assumes that in this scenario, survivors would miss three weeks of work either because they were sick or because they needed to care for someone who was sick.

The CBO projections for the mild scenario assume that 25% of workers (except in the farm sector) got sick and that 0.1% of these workers died.  This death rate is consistent with the death of 100,000 people.  The CBO assumes that in this scenario survivors missed 4 days of work again either because they were sick or taking care of others who were sick.

The projections of changes in GDP from the pandemic involve a supply and a demand effect.  The supply effect is simply a function of the lost employment for the year.  Workers who die are assumed to lose the entire year. Workers who survive lose 3 weeks for the severe pandemic and 4 days for the mild pandemic. 

The demand effect is based on an examination of GDP by industry.  CBO assumed different declines in demand for different industries based on judgments about the degree of social interaction in each industry.  The report acknowledges that this process is subjective.

“Given that there is little historical evidence available to form these estimates, they are admittedly extremely rough.”   (Page 42 of the CBO report.)

Table A2 lists the assumed demand effects by industry.  The demand impacts for the severe scenario are:

  • 80% decline for arts and recreation, accommodations and food service.
  • 67% decline for warehousing.
  • 10% decline for most other private industries.
  • 5% decline in government services.
  • 0% change in information, finance, professional services, and education.
  • 15% increase in the utilization of health care.

The CBO subtracted the supply-side impact of the pandemic from the demand-side impact for each industry in order to avoid double counting.

A Discussion of Issues Raised by
 The CBO Report

Comment One:  The CBO report does not allow for any involuntary unemployment.  The entire supply effect is the result of voluntary employment stemming from missed work attributable either to personal or family illness.  However, some industries, (hotels, amusement parks, casinos, movie theaters, and life theater) are assumed to experience an 80 percent drop in demand.  Substantial layoffs would occur in these industries.  Involuntary unemployment will be larger than direct absenteeism due to health effects.

Comment Two:  The CBO report does not consider the impact of large regional impacts in particular regions, which would further exacerbate the declines in output.  The regional impacts would be especially large in cities and states that specialize in tourism and recreation.  In the midst of an epidemic that was killing 2 million people few people would travel to Las Vegas and visit a crowded casino and few families would travel to Orlando for an amusement park.  Epidemic rates would be especially high in crowded cities.  Involuntary employment could be substantial in the cities most affected by a pandemic.  The health situation and economic impacts of a pandemic might, like the impacts of Katrina, be highly concentrated in a few states and cities. 

Comment Three:  The report does not allow for the pandemic to cause decreased real estate prices and defaults on mortgages.  The severe pandemic scenario would adversely impact housing markets through several channels. First, the pandemic increases the number of empty houses through the 2 million flu-related deaths.  Second, increased unemployment from absenteeism or involuntary unemployment will increase mortgage defaults, especially among homes with little equity.  Third, decreased wages will lead to reduced demand for homes.  Fourth, mortgage default would increase because of household solvency problems attributable to increased health problems.  House values and other forms of wealth have a direct impact on consumption, the key component of aggregate demand.  The negative impact of pandemics on housing would further reduce consumption, the primary component of aggregate demand because consumption is a function of wealth.

Comment Four:  The report does not consider the possibility that a pandemic will decrease stock prices or cause corporations to default on their bonds.  A decrease in demand of 60% to 80% in some industries would lead to massive corporate defaults.  This increase in corporate defaults would increase the cost of capital and decrease investment.   An increase in corporate defaults and decreased stock prices reduce wealth and decrease consumption.  These corporate defaults could impact the solvency of life insurance companies, which could in turn impact mortgage defaults.   The CBO does not consider any of these potential impacts in their assessment of the potential impacts of a pandemic.

Comment Five:  The CBO analysis assumes that an increase in the demand for health care will partially offset decreased demand in other sectors.  However, the CBO report on page 28 to 32 points out that there is little or no excess capacity in the health care system and that a pandemic would reduce capacity because many health care providers would become sick.  The CBO discussion on strategies for providing additional health care includes increased training of home health care providers and an increased role for the military.  Neither of these channels results in increased GDP.  Home health care services are a non-market based transaction that is not included in GDP.  The military, as part of government spending, has already been included as part of GDP.  The inclusion of increased demand for health care results in the CBO overestimating GDP in a recession.

Comment Six:  The CBO assumption of a 0% change in demand for information, finance, professional services and education is unrealistic.  Even if demand for media remains unchanged this sector is impacted by decreased advertising revenue.  Second, large layoffs by industry will decrease demand for health insurance policies 401(k) enrollment and group life insurance sales.  Individual life insurance sales could rise because of the pandemic.  Third, the demand for education will fall because schools and universities are places where the virus can easily spread.

Comment Seven The calculation on the supply effect assumes that the pandemic increases real wages.  However, the impact of a pandemic on real wages is theoretically weak.  Why would workers be able to claim wage increases over inflation during a disaster when corporate profits were down?

Comment Eight:  The report says “the analysis ignores the possibility that productivity among workers who remain on the job would be likely to rise.”  (Page 42.) In fact, firm productivity could fall in a pandemic even if the actual productivity of survivors rose.  The replacement of key workers will lead to productivity decreases because of the resources needed to train the replacement.   Also, productivity could decrease because firms will spend resources to insure that infections are not transmitted on the job.

Comment Nine:  The decrease in GDP is only one of the economic costs associated with a pandemic.  It would be useful to calculate the impact of the pandemic on wealth, corporate and personal bankruptcy, and most importantly the value of the lost life.

Comment Ten:  The cover letter attached to this report refers to impacts on the worldwide economy.  The analysis inside the report pertains to the U.S. economy.  Presumably worldwide impacts could be much larger if avian flu is more severe in Asia and the Middle East. 

Comment Eleven:  The CBO assumes a lower death rate in the farm sector because of the lower social interaction in this sector.  However, the farm sector is where the birds are and cases to date have involved bird to human transmission.

Concluding Thoughts:  Many people believe the current Administration has under-reacted to the threat of Ebola.  Basically, economists who have studied this issue have concluded that the likely impact of Pandemics on GDP is manageable (around the size of a typical recession).   The approach taken by the current Administration is consistent with these cost estimates and the way past Administrations have reacted to the threat of pandemics like Avian flu. 

Monday, October 6, 2014

Landline Use Continues Its Decline -- Implications for 2014?

Ownership and use of landlines and cell phones – evidence from the NHIS

Background:   Pollsters have been a bit off in recent elections partially because their samples drawn from lists of people who own or use landline phones are less representative of the general population due to the growth of cell phones.

Question:  What are the most recent figures on the number of voting-age people without landline phones?   Have pollsters caught up?

My guess on the second question is NO!   More on this will follow.

The second question was partially answered by researchers at of all places the CDC.

Wireless Substitution:  Early Release of Estimates From the National Health Interview Survey, July- December 213.

 Recent data on landline and cell phone ownership:

Key data in the CDC report on cell phone use. 

Landline and Cell Phone Ownership by Houesholds
July-December 2009
July-December 2013
Landline With Wireless
Landline Without Wireless
Unknown Status
Wireless Only

The report also contains two tables on demographic characteristics for wireless-only households, a table on health prevalence statistics (landline, wireless-only and phoneless) and information on wireless-mostly households. 


Less than 50% of the households in the population now have a landline phone. 

 8.6 percent of people with landlines did not also have a wireless phone. 

 41 percent of the population is wireless only.

The proportion of the population that is phoneless went from 2.0 percentage points in 2009 to 2.5 percentage points in 2013. 

The obvious:  The trends documented in the CDC study reveal clear problems associated with the use of samples that are predominately based on landline samples to get reliable election results. 

Most households no longer have a landline.
The overwhelming majority of people with landlines also have a cell phone.
The use of cell phones has caused many people to limit their use of landlines and allows people to avoid unwelcome calls from pollsters.

The comments below suggest the obvious observations understate the impact of this problem for pollsters.

Specific Comments:

Comment One: Households can include multiple adults.   The NHIS also presents information by adults only.   The adult-only numbers may be more relevant for analysis of voting issues.   The NHIS interviewers may only have talked to one family per household.   Hence, it is really hard to get reliable information about preferences of people who live in homes with more than one family.  

Comment Two: The third table in the CDC paper about the demographic makeup of people with wireless only phones would be more useful if it also had information on landline only and land-line wireless segments.   Also, I need the share of people in each group in the general population.   I am limited by my current lack of easy access to the micro data.

Comment Three:  The percent of people with no phones 2.0 percent in 2009 and 2.5 percent in 2013 seems small.   The change may or may not be significant.   However, this issue could be a big deal is some areas like urban areas of Ohio, where some pollsters were startled by 2012 outcomes.

Comment Four:  Some wireless users have bare-bone plans with limited hours.   Other people may only use their wireless devise through Wi Fi.   If so, even a survey that includes cell users could misrepresent the general population depending upon how the sample is constructed and on the type of cell phone users included or excluded. 

Comment Five: The issue of bias due to growth of the number of wireless users is especially large in states where there are a large number of new arrivals.   People tend to keep their landlines until they move and decide it is not needed anymore.  A large number of new entrants may be a reason why the polls are a bit off in Colorado. 

Comment Six:  Table five reports that the percent of households, which are mostly wireless, went from 17.4 percent to 18.3 percent over the 2010 to 2013 time frame.  Is this change significant?   It seems a bit small.

 Regardless, the impact of an increase in wireless-mostly usage should be assessed in comparison to the landline/wireless population not the entire population

Finally, the mostly-wireless concept may not be the best measure of the impact of wireless use on pollster non-response rates or poll bias.

Concluding Remarks:  Cell phones have impacted the ability of pollsters to predict election outcomes.  See my post reviewing the PEW study.

Impact of cell phones on polls:

Many pollsters were shocked in 2012.  The reason I suspect is that they did not keep up with the growth of cell phones.  The growth of cell phones and the declining use of landlines is a moving target.  I suspect the problem has not been fixed and we will see a number of surprises come November.