Wednesday, April 18, 2018

Research on how diabetes impacts retirement age


How Does Diabetes Influence the Impact of Aging on the Probability of Employment?



Abstract
This paper examines how diabetes and complications from diabetes affect the impact of age on the probability a person nearing retirement age remains employed. The results presented here indicate that diabetics, especially those with complications tend to leave the workforce prior to 62 and becoming eligible for Social Security benefits. Diabetes and complications from diabetes also reduces the ability of people to remain in the workforce to increase their Social Security benefit. Increases in the eligibility age for receiving Social Security benefits would impose substantial hardships on diabetics. Programs that reduce the number of people with diabetes and eliminate diabetic-related complications could expand the workforce and stimulate economic growth. These benefits should be counted when considering the cost of programs to reduce diabetes.
Keywords: Diabetes, Health, Social Security
JEL Classification: I14, J26, H51, H55


Tuesday, April 10, 2018

The Economic Divide Among Democrats

The Economic Divide Among Democrats

In 2016, there were huge difference between the economic position of Clinton and Sanders.  However, most of the discussion and the political debate in 2016 was on non-economic issues.  The amount of air time on issues like gun control, racial justice, police shootings, fair treatment of women, and Trump misogyny, far outpaced the amount of air time on the ACA, Medicare, the future of Social Security, the environment, and student debt.

There are huge economic divides inside the Democratic party, which were fully displayed but not resolved by the 2016 nomination contest.  The positions of Senator Sanders were always clear.   He wanted complete government funded universal health insurance, free college at public universities, and trade policies similar to the ones Trump has now implemented.

Secretary Clinton’s proposals were more complex.  She, in my view correctly, viewed Senator Sanders proposals as too expensive.  She supported modifications to the ACA rather than single payer, moved towards a less expensive version of Sander’s college debt plan and altered her views on the Trans Pacific Partnership.

There were problems with the policies that Clinton espoused.   She underestimated the angst in the country about changes in health insurance, many of which were not caused by the passage of ACA.  One of her proposals to subsidize loans to early employees and founders of startup firms was very hard to defend.   She appeared to waffle on TPP and other trade issues, an approach that hurt her in Michigan, Pennsylvania and Wisconsin

Clinton won the nomination but her victory was not based on her economic policies.   She won because she like her husband before her had a lot of support from black voters, she painted Sanders as soft on guns and she had a lot of support inside the Democratic Party due to her years of service.

The economic divides inside the Democratic Party illustrated by the 2016 contest are being ignored today.  Turn on MSNBC today and you will get a discussion of Russia, Trump corruption and Stormy Daniels.  Most voters are rightly appalled by Trump’s behavior, which is arguably at a different corruption level than any scandal in previous history.   Trump, unlike people at the center of previous scandals, seems to be asserting that he is above the law.

However, scandals often do not lead to political debacles.   The Republicans survived Teapot Dome and the Democrats survived Whitewater and Lewinsky.

The Democrats can’t count on anger against Trump leading them to victory. Harding would not have won reelection in 1924.   I suspect that Trump, like Harding, will not seek reelection.

Racial justice is important.   Police shootings of unarmed young black men are disturbing.   Neither issue is decisive with white swing voters in Wisconsin, Pennsylvania, or Michigan.  When people tangentially, associated with BLM disrespect police or break laws while protesting some whites vote Republican.

Unjust deaths will always dominate the news cycle.   However, for many voters, economic issues broadly defined almost always trump issues like gender inequality, racial justice and corruption.


Four economic issues – health care, college costs, trade, trade, and Social Security – could determine the outcome in 2018 and in 2020.   Memos on how the Democrats should approach these issues will be created soon.

Wednesday, April 4, 2018

Impact of disease on employment for people nearing retirement age

Impact of disease on employment for people nearing retirement age

Question:  How might one model the impact of disease on the employment status for people nearing retirement age?  How does disease impact the ability of a person to stay in the workforce prior to age 62 and after age 62?

Short Answer:   Disease causes people to leave the workforce prior to age 62, before becoming eligible for initial Social Security benefits.  Disease also results in people leaving the workforce prior to age 66 and becoming eligible for the full retirement benefit.

Data:   These questions were examined with data from the 2015 Medical Expenditures Panel Survey.  

The study covers 3314 people between the ages of 58 and 59.

The dependent variable employed was based on EMPST53 which is set to 1 if the person had connection with a job during the round options 1, 2, or 3.

The database has questions on whether respondent had diabetes, complications from diabetes, stroke, arthritis, asthma, coronary heart disease emphysema, or cancer.

The data base also has information on sex, education level, and age of each respondent.

Methodology:  A three-step procedure was used to estimate the impact of disease on the relationship between age and employment status.

Step One:  Estimate the logit regression model of employment status on disease variables and other explanatory variables.    The odds ratio from this logit regression model are presented below.

First Stage Logit Model
Variable
Odds Ratio
p-vlaue
age5859
1.61
0.001
age6061
1.23
0.141
age6364
0.82
0.161
age6566
0.43
0.000
male
1.43
0.000
ba_deg
1.70
0.001
no_college
0.92
0.469
diabetes
0.68
0.000
complications
0.35
0.000
stroke
0.38
0.000
arthritis
0.63
0.000
asthma
0.85
0.217
coronary
0.64
0.002
emphysema
0.40
0.000
cancer
0.63
0.000
_cons
1.75
0.001



N=3314


Log likelihood = -2037.3463





Step Two:   Use the estimated odd ratios from the disease variables and the observed disease status of respondents to obtain a disease vector variable. 

 Odds ratios that are lower than 1 indicate that the disease will cause people to leave the workforce.    Note that all odd ratio estimates for the disease variables are less than 1.0.  (Only Asthma was insignificant.) 

 The tendency for disease to force people out of the workforce is the weighted average of one minus the odds ratios for the disease variables with the disease status observations for all respondents.


STATA code for disease index here.

Generate

Disease

=

0.3236833*diabetes+0.6452645*complications+0.618275*stroke+0.3654761*arthritis+0.152522*asthma+0.3575031*coronary+0.6038477*emphysema+0.3699676*cancer

Detailed Summary of Disease Variable Here:


The Disease Statistic
Percentiles
0%
0
1%
0
5%
0
10%
0
25%
0
50%
0.37
75%
0.69
90%
1.06
95%
1.34
99%
2.09
100%
3.43
Mean
0.43
STD
0.48
Skew
1.50


Step Three:   Estimate a second logistics model for the employment variable where the model includes explanatory variables sex, education, the age of the respondent and the interaction of the age and the disease index.   The coefficients of the logit regression model are presented below. 



The Second Logit Model
Variable
Coef.
P>z
male
0.35
0.00
ba_deg
0.53
0.00
no_college
-0.09
0.43
age5859
1.16
0.00
age6061
0.77
0.00
age62
0.41
0.03
age6364
0.41
0.01
age6566
-0.22
0.16
disease_age5859
-1.56
0.00
disease_age6061
-1.28
0.00
disease_age62
-0.92
0.00
disease_age6364
-1.38
0.00
disease_age6566
-1.42
0.00

The logit coefficients reveal decreases in employment as people age and larger decreases in employment for people with disease at each age group.

Note the omitted education group is some college.  A BA degree has a larger impact on employment relative to the base group than does no college.

I can use these logit coefficients to get predicted employment rates at each age group for a specific individual at different levels of the disease index.  

A separate post detailing these calculations were made will be created.   Link will be placed here. Go here for a discussion of the method applied to a similar question.




These calculations are presented below.


Age and Employment Probability
Age
Employment Probability (Disease Index is 0)
Employment Probability Diseases Index is 75th Percentile value of 0.69)
58-59
0.885
0.595
60-61
0.839
0.573
62
0.785
0.579
63-64
0.785
0.456
65-66
0.659
0.300


Observations from estimated employment probabilities. 

Many people a high disease index leave the workforce prior to age 62, the age where people become eligible for a Social Security retirement benefit. 

The estimate probability of being employed for people at the 75th disease percentile is lower at 62 than 60-61.  

The drop off in employment after age 62 is substantially steeper at the disease 75th percentile level than at the disease 0 percentile level.

Concluding Remark:   A member of the Senate Finance committee once joked that he was fearful that scientists would create a pill that got rid of all disease because such a pill would bankrupt Social Security.   The results presented here suggest that the eradication of diseases or improvement of health would increase employment.   The increase in employment would increase economic growth.  Part of the costs of new health care advances should be offset by this additional economic growth.