All right.
So, this next panel is called Non-SSA Program Interactions.
And, as we all know, there's lots of programs out there that affect the likelihood that
people with disabilities will participate in social security disability insurance and
SSI, and we also know that there's lots of programs that affect the wellbeing of beneficiaries.
And this panel today will look at interactions between the Children's Health Insurance Program
and SSI, Medicare Part D and its effects on the wellbeing of DI beneficiaries -- and I'm
getting the middle one -- kids who use vocational rehabilitation services in their long-term
employment and social security disability program participation.
So, the first paper is going to look at the long-term effects of CHIP eligibility expansion
on SSI enrollment, and my Mathematica colleague, Mike Levere, will be presenting, and his discussant
will be Steve Hill, who I've known more than any -- longer than anyone else in this room,
almost 30 years.
We went to grad school together.
We started at, what, age ten.
So, take it away.
Thank you very much.
I'm going to speak today about the contemporaneous and long-term effects of CHIP eligibility
expansions on SSI enrollment.
This is joint work with Shawn Orzol and Lindsay Leininger at Mathematica, and Nancy Early
at SSA, and I just particularly wanted to thank Nancy for our instrumental work with
the putting together the data that we were able to use for this paper.
So, there's a current policy debate about the appropriate level of generosity for public
health insurance eligibility, where recent legislation considered by congress would really
substantially cut the level of Medicaid eligibility.
And there's been a lot of research in economics about how changes in Medicaid eligibility
affect health and how they affect labor market outcomes, given the kind of incentives, and
I think it's also really important to think about, well, what are the spillover effects
of Medicaid eligibility into other programs, and in particular, into disability insurance
programs.
That may have, potentially, very large fiscal consequences given that if you were to, say,
reduce Medicaid eligibility but at the same time that were to increase participation in
SSI or SSDI, that may have a significant effect on what those reductions in fiscal savings
are.
So, there is a resent literature on the relationship between health insurance eligibility and SSI
and SSDI application that is mostly focused on the adult program, and I'm going to focus
today on children's SSI.
And so, our main research question is how do changes in Medicaid eligibility affect
participation in children's SSI.
And we're going to focus on CHIP-area expansions in Medicaid that happened in the late 1990s
and early 2000s, and I'll tell you a bit more about that in one minute.
And we think that increases in Medicaid eligibility are likely to reduce SSI applications.
The reason for this is that, in most instances, in most states, when you qualify for SSI,
you also automatically receive Medicaid coverage as a result of that SSI award.
And so, if you have Medicaid coverage from somewhere else, that's going to reduce the
potential value of a new SSI award, where if you have a reduction in value, but there
are still some costs associated with submitting an application, if not monetary, then certainly
in terms of administrative burden, then a reduced benefit and a constant cost would
mean that that net cost benefit analysis could lead some people to reduce their applications
into SSI.
And to give you a very quick overview of our results, we're going to find that overall,
there's a very small relationship between Medicaid eligibility and SSI applications,
but in particular states where there is not an automatic link between SSI and Medicaid,
where, in states where there's an additional criteria to get Medicaid eligibility after
you get an SSI award, we're going to find that increases in Medicaid eligibility significantly
reduce SSI applications.
So, as a bit background on the CHIP-era expansions, in the late 1990s, the Children's Health Insurance
Program was established in order to expand public health insurance coverage to near poor
children, where in order to qualify for Medicaid you have to have an income below a certain
threshold of the federal poverty level, and so, for people who had incomes that were too
high to qualify for Medicaid, but that their parents' incomes with too low in order to
afford private health insurance coverage, the establishment of these CHIP programs helped
to cover that gap.
Like I said, this happened between 1997 and 2002, and the overall effects varied by a
child's age, where the reason for that is that before CHIP was established, the income
eligibility limits were higher for younger children so that more children were covered
who were younger age, and therefore, when the programs expanded, older children saw
relatively bigger increases in their Medicaid eligibility.
So, this is supposed to be a two-part graph, and it is not.
So, I would like you to focus, first, on the black line in the top left graph, just so
I can single out and try to explain what this picture is showing.
Essentially what we're trying to show is if you focus on that black line on the top left
graph, what we're estimating is the person of children at a given age in a given year
in a particular state that are eligible for Medicaid.
And so, for children who are ages one through five in Delaware in 1997, about 35% of them
were eligible for Medicaid.
And we estimate that using the 1996 CPS, and we used a fixed national cohort, where we
take everybody throughout the U.S., so we're not going to be subject to potential changes
in people choosing where they live affecting this estimate.
So, in 1997, for children ages one to five in Delaware, the income threshold was 133%
of the federal poverty level and 35% of kids were eligible.
That was the same in 1998.
And then in 1999, you see this increase to about 50% of kids be eligible when Delaware
established a CHIP program that increased the threshold to 200% of the federal poverty
level.
The rest of the figure shows that there's variation in terms of when Medicaid was expanded
and by how much, and to which children, given that there's this variation across time and
across states and across ages.
So that you can see that, for example in Vermont, the expansions in Medicaid happened in 1998.
In New Jersey it happened over a couple different years, and in North Dakota, there were much
smaller expansions at all.
And so, we're going to use that variation, and when Medicaid was expanded and by how
much as part of our estimating strategy.
So, our overall goal is to estimate the impact of increased Medicaid eligibility on children's
SSI applications, and we use that variation in state policy as a type of natural experiment
for expansions in Medicaid.
So, this is similar to -- this is a simulated eligibility measure that's been used by Currie
and Gruber, and a lot of other studies.
And the key thing is that variation in when and where these expansions take place is entirely
due to differences in the state-specific generosity, where, because we use that fixed national
cohort, it's not that people potentially move to a state where there is higher eligibility
for Medicaid and, therefore, that's going to affect this estimate.
There's not any changes due to the economy evolving over time, all of these differences
are due to the state-specific rules.
And so, essentially what that means is that where and when you are born is going to randomly
affect your Medicaid eligibility, and we're going to use that variation in Medicaid eligibility
to estimate the impact on SSI applications, where our main outcome measure is going to
be the number of children at a particular age in a given year in a given state that
applied for SSI, and we get that using the supplemental security record from the SSA
administrative data.
Our estimating strategy is going to control for any cross-state and cross-year variation
using state and year fixed effects, and we're also going to control for a general age trend
in SSI applications, because the older that you are up until age seven the more likely
you were to apply for SSI, whereas after age seven, the older that you are the less likely
you are to apply for SSI.
So, all of our findings are going to deviations from that general trend.
And so, these are our main impact estimates on SSI applications, where we find that a
10-percentage point increase in Medicaid eligibility is going to reduce SSI applications by one-one-thousandths
of a percentage point.
So that's very small in magnitude, and also relatively -- a 10-percentage point increase
is about a 20% increase in Medicaid eligibility given that the average was about 46%, and
that one-one-thousandth of a percentage point corresponds to a .02% reduction in SSI applications,
so the average SSI per capita was about half a percentage point.
And that's a very small effect, but it's also fairly precisely estimated, because we can
rule out with our confidence intervals any increasing or decreases of more than 3%.
And we also rather than use state and year fixed effects, we also use state by year fixed
effects, and that doesn't really affect our results, and I'll tell you a bit more about
some other robust checks in a minute.
So, when we started the project, we wanted to look into state heterogeneity, which I
alluded to earlier, because there is a different relationship between SSI and Medicaid in different
states.
So, in most states, which are called 1634 states, when you apply for SSI, you will also
-- you're also simultaneously applying for Medicaid.
And so, if you get an award of SSI, you will automatically receive Medicaid coverage.
There are some other states that have an additional criteria to get Medicaid, where in some of
those states you have to just file a separate application that will be accepted with certainty.
And in some of the states there's a stricter income criteria that you might not actually
get Medicaid, even if you get SSI.
And you can see a list of the states up here.
And really importantly, it's not these are like red states or blue states, it's kind
of a random grab bag of states.
And so, what we initially anticipated was that in the states where you automatically
get Medicaid after an SSI award is where we would see the biggest reduction.
And the reason that we thought this was that in those states there's a stronger link, so
when you get Medicaid eligibility from elsewhere, that has the biggest reduction in value of
a new SSI award.
In fact, we find the exact opposite.
And what you can see in this table is that in states where there is an additional criteria
to get Medicaid after you qualify for SSI is where there's the biggest reductions in
SSI application, so if you have that same 20% increase in Medicaid eligibility, that
reduces SSI applications by 12% in these states.
Now, these states are definitely different.
You can see in the second line here that, on average, many viewer people apply in these
additional criteria states.
Importantly, the acceptance rates are approximately the same across the two types of states, and
so what we think may be the case is that the type of person who is still going through
the process of applying in the states where there's an additional criteria is the type
of person who particularly values that Medicaid award, and so there's the biggest effect for
them, that there is a reduction in SSI applications after increases in Medicaid.
But if people want to talk about this later, I'd be very happy to discuss some other reasons
that you think this would be the case.
Our estimates are pretty robust.
We do a few different specifications checks, where rather than look at variation coming
from state and year, we estimate the variation in a couple of different ways, using either
age and state or age and year, and we find the same overall result, that there is no
overall effect of Medicaid eligibility on SSI applications, and that there is a big
reduction in states where there is this additional criteria to qualify for Medicaid.
We also vary the base year that we use to calculate the simulated eligibility.
I said we use the 1996 EPS.
We use different years.
It doesn't affect the estimate.
And really importantly, we also use old-age applications as a placebo test, where you
would think that for people who apply for SSI who are older than 65, their eligibility
is entirely determined by income rather than any disability income.
They also are likely to have Medicare coverage, and so you wouldn't think that changes in
children's Medicaid eligibility would affect those old-age applications, and they don't.
We find no overall effect and none of the same state heterogeneity.
So, to conclude, we find that overall, increases in Medicaid eligibility don't really affect
SSI applications.
But in particular states where SSI awards did not automatically lead to Medicaid coverage,
we find that there is a big reduction in SSI applications from increases in Medicaid eligibility,
suggesting that in some states there is some substitution between these programs.
Moving forward from here, we have some additional outcomes that we're going to look into for
this contemporaneous analysis, which is what I've been talking about, but, as you may have
seen in the title, we also have a long-term analysis, and that is still fairly preliminary.
Our results are somewhat mixed.
What we do is we look at the effects of the total years of Medicaid eligibility during
childhood and see how that affects long-term SSI applications.
That may have a particularly important effect, because people who receive SSI tend to do
so for a very long time, and if there are potential changes in SSI awards and applications
from Medicaid eligibility in childhood, those effects would be magnified in the long-term.
We might think that reductions in -- that increases in Medicaid could reduce applications
in the long term if, you know, increases in Medicaid improve health, but at the same time,
it might also increase SSI applications in the long term if you're more likely to see
a doctor and have a diagnosis of some type.
And so, our preliminary results kind of, depending on which specification looks positive or negative,
and so we're still looking into that, but we will keep you posted.
So, thank you very much.
I forgot to mention -- oops it's not on.
Steve's from Agency of Healthcare Quality and Research research, so he's sort of qualified
to talk about the paper.
About healthcare; right?
Yeah.
So, my comments have a disclaimer, and that is exactly as Gina said, they represent my
comments, not those of the Agency for Healthcare, Research, and Quality or the Department of
Health and Human Service.
And so, I really like the data and the methodology in this paper, and I think this is really
fun to read, and I just want to review the main findings of the paper.
So, the effects of Medicaid and CHIP, and public insurance in general, eligibility on
SSI applications for children, and awards for children, overall, they found a small
but not statistically significant effect, and there was -- when they broke it into two
groups of states, states that auto-enrolled kids into SSI and states that required a separate
Medicaid application, they found no effect in states that auto-enrolled SSI enrollees
in Medicaid, and then for the states that required a separate application, they found
a decrease in SSI applications and awards.
And one thing about those states that they point out in the paper that didn't get brought
up here is that many of the states that have separate applications also had stricter eligibility
rules than the SSI program.
And so, the results, again, are robust, several specification checks.
So, let's see.
So, I have three kinds of comments.
The first is, I think that it would be helpful to tie the paper more -- to explain better
how SSI income eligibility criteria relate to CHIP eligibility criteria.
Second, I'm going to talk about some of the mechanisms that could increase or decrease
SSI enrollment in response to changing in CHIP, and then I'll talk about some of the
factors that affect substitutability between CHIP and SSI and maybe explain it in a slightly
different way than Michael did.
So, SSI income eligibility determination is probably pretty familiar to this audience,
so I won't talk about it, but it's different from CHIP eligibility, in particular, that
there are some very generous disregards for earned income.
For CHIP, the typical state had a net income test, and the net income test had a small
disregard for earnings, but there was also a disregard for childcare expenses.
And I put together a simple spreadsheet so I could sort of sort this through a little
bit for myself, and what I found is there is one disabled child in the family, and a
family size, in total, is two to four adults, and their earnings are the only source of
income, then the child would be eligible if gross parental earnings were less than approximately
185 to 220% of poverty, depending on the family composition, and that's because of the general
disregard.
So, if you count for the CHIP childcare disregards, earning 220% of poverty is really about a
net income of 200% of poverty for CHIP, so the SSI threshold is within the range of CHIP,
and so that's actually helpful for thinking about this study, and thinking about substitutability
between the programs.
So, there are a couple of reasons why CHIP could have increased SSI awards, and the first
one is the welcome mat effect.
And when states reported that when they went out with their outreach efforts and get kids
enrolled in CHIP, they got a lot of kids enrolled in Medicaid.
There were all these kids who were eligible but not enrolled.
And you could have had a similar effect for SSI as well.
The government accountability office also put forward a hypothesis that when kids get
CHIP, they're going to see their providers, the providers are going to screen them, and
diagnose them, and then the diagnoses could be used for SSI applications, and that's kind
of a dynamic model.
Where is Dave now?
But, in fact, since there wasn't any evidence of an increase, I think we can move on from
there.
So, the paper really focuses on this substitution between the programs, and I'll, I think, explain
it a little differently, or how I see it a little differently.
And, so enrolling in CHIP is easier than enrolling in SSI, so families go for the less burdensome
approach, and CHIP applications were definitely easier.
SSI requires a disability determination.
It requires a complicated income and asset counting rule, and if you don't get there
through on the first try, there can be a lengthy fields process.
But the big thing missing from this story, I think, is SSI's cash benefit.
So, for whom is getting CHIP's insurance coverage equivalent to getting Medicaid cash plus benefits?
Well, it would have to be people who are eligible for only small cash benefits, I think.
So, these would be people with maybe incomes between 200 and 220% of poverty, so maybe
this isn't a big population.
Maybe if you think about people who have a lot of fluctuations in their income, SSI would
get really complicated, and so the families would be better off on CHIP, and I think it's
an interesting question how many families would be in that situation, but, anyway.
So, also the program substitution hypothesis, families with incomes above 220% of poverty,
gross income, they really wouldn't have capacity to substitute between programs because they
really aren't eligible.
So, a sensitivity test for the kind of results we've seen, is there an impact of eligibility
increases above that limit in CHIP and on applications and awards for SSI?
So, the paper focuses on one feature of that states -- that could affect the substitutability
between the programs.
Again, that's separate applications and that's correlated with states that have stricter
income eligibility rules.
But another factor is that some states, including some of the 209D states with stricter eligibility
rules, had pathways of eligibility for people with disabilities that had higher income thresholds,
so maybe they didn't need to go to CHIP.
These pathways do require the disability determination, so they do, you know, it's not a perfect substitution
with CHIP, but that also could be as important as having a separate Medicaid application.
States were giving Medicaid to people who received state finance supplemental disability
payments.
The 209B states were doing that.
During the time period of the study, a growing number of states were using poverty-related
eligibility pathways for people with disabilities.
Hawaii is a 209B state, and it also covered people with disabilities up to a hundred percent
of the federal poverty line.
Now, an important thing about this is, it doesn't sound so great, but, actually, that's
a net income threshold, and so all those SSI disregards apply.
So, a family -- again, same spreadsheet I was using before -- they could have gross
incomes of around 240 to 310% of poverty.
So, that really does put you -- makes CHIP a little less relevant for people in those
states.
So, the paper focuses on variation in CHIP income eligibility through their eligibility
rates, but there are other factors related to CHIP that could affect eligibility.
And there's a lot of research on premiums in CHIP and how that can be a barrier to take
up, and there might be other factors, like asset tests.
And I really like the method used in this paper.
Michael spent some time talking about the Currie Gruber approach, and I think that could
possibly work for thinking about income eligibility thresholds for the disability pathways.
If you just focus solely on the income rules and not thinking about the disability issue,
disability determination.
But I think it's going to be difficult to account for policy endogeneity if you're looking
for CHIP premiums and assets.
But I think it would still be worthwhile.
And the paper used two separate regressions when looking at types of states, and so that's
basically an interaction effect.
And I think it would be useful to look at just entering these policy variables directly
and, you know, do they have any effect on the substitution of application rates on their
own?
I think that would be useful for policymakers.
So, in summary, overall, the note, the small and statistically insignificant effect is
consistent with the competing mechanisms for increasing and decreasing applications, and
for the small population, it's likely to not to have many small cash benefits.
I think there are a lot of studies about geographic variation, and I think that the additional
features related to the substitutability between SSI and Medicaid and CHIP could be important
to take those into account rather than just looking at the one.
And I think this study is going to help policymakers understand the complex relationships between
health insurance and disability benefits, especially in a time period where CHIP is
up for reauthorization.
Congress needs to decide what to do.
Thank you.
All right, next up is Priyanka Anand from Mathematica, soon to be at North Mason University.
She's leaving up, but we'll still talk to her.
She is going to talk about long-term outcomes for transition-age youth with mental health
conditions who receive postsecondary education support.
That's a mouthful, but you'll understand what she's talking about soon.
And her discussant is Manasi Deshpande from the University of Chicago.
Well, thank you all for being here.
As Gina mentioned, today I'm going to be talking about long-term outcomes for transition-aged
youth with mental health conditions who receive postsecondary education support, and this
is joint from Todd Honeycutt from Mathematica.
So, state vocational rehabilitation agencies, or VR agencies, help people with disabilities
achieve their employment goals, and they do this by providing a wide variety of services.
These include paying for assistant technology, job placement services, and counseling services.
One service in particular that we are interested in this paper is providing support for post-secondary
education.
So, this would be providing tuition assistance, paying for books or other supplies to attend,
college or vocational training.
And we're interested in support for post-secondary education because it's thought that it may
improve employment outcomes.
It's also a goal of many people with disabilities who apply for VR support.
We're also interested in youth with mental health conditions for a couple of reasons.
The VR population, it's youth with mental health conditions that comprise a large portion
of the VR population, and, also, they toned to have lower employment outcomes.
And finally, past work has shown that youth with mental health conditions are less likely
to receive the VR services in general, and, in particular, they are less like to receive
college supports that youth with other type of disabilities.
So, for this reason, the objective of this paper is to examine the relationship between
receiving VR support for a postsecondary education and long-term outcomes for youth with mental
health conditions.
We go about this by answering two main research questions.
The first research question is, how do long-term employment and earnings outcomes vary by receipt
of postsecondary education support for transition-aged youth with mental health conditions?
In the paper, we also compare this to this question for youth development without mental
health conditions, but given time constraints, I'm just going to focus on youth with mental
health conditions here.
Our second research question is the same as the first, but instead of looking at employment
and earnings outcomes, we look at the receipt of federal disability benefits as an outcome.
This would be receipt of SSI or SSDI.
The past literature on this topic tends to focus on people with all types of disabilities
and not just focus on those with mental health conditions.
But even among this literature, the results are mixed.
Some of the papers find that receiving postsecondary education services has no effect on employment
upon VR exit.
Others find that there's actually a slight decrease in employment upon VR exit.
The paper that does look at people with mental health conditions was Dean and his co-authors,
and they found that there was a decrease in employment two years after receiving VR services,
but they only looked at one state, which was Virginia.
So, in our paper, we make a couple of contributions to this literature.
First, we do focus on youth with mental health conditions.
We also examine outcomes nine years after VR application, so most of the literature,
as I mentioned, looks at outcomes upon VR exit or two years afterwards, but we're able
to examine a much longer time period of nine years after VR application.
We also look at three types of employment outcome.
We look at employment, we look at earnings, and we look at receipt of federal disability
benefits, so SSI or SSDI.
And finally, we are able to control for individual characteristics and state characteristics
in our analysis.
Oh, sorry.
I
think I'm pushing too many buttons.
All right.
This is the slide.
So, we used three datasets in our analysis.
The first dataset that we used is the -911, and this is an administrative dataset that
provides us with information about the VR services that are received, and we use this
data from 2002 to 2013.
We then merge this data with the 2013 disability analysis file, the DAF, and the DAF provides
us with information on the federal disability benefits that are received by our population
for all years, from 2002 to 2013.
And then finally, we merged the RSA-911 and the DAF with the master earnings file, which
provides us with earnings information, and this is also administrative data.
Our final sample size is approximately 437,000 VR applicants.
Our sample consists of first-time VR applicants who submitted their application from 2002
to 2004.
And because we have data that goes through 2013, we're able to follow each of these three
cohorts for nine years each.
Our sample is limited to those who are ages 16 to 24 at application, so these are the
transition-age youth, and we just look at those that are eligible for VR services.
There are two main components of our empirical methods, so the first thing we do is we provide
descriptive statistics of our main outcomes by service receipt.
So, we look at the outcomes for those who received support for post-secondary education
and compare it for those who receive support for other services, but not postsecondary
education.
We then want to see if the relationships we observe in our descriptive statistics hold
after controlling for individual and state factors, and so we estimate a regression between
the main outcomes of interest on VR service receipt, while controlling for those other
characteristics.
I seem to be having more trouble with this than most.
Okay.
So, here are the employment rates for youth with mental health conditions who receive
services other than postsecondary education support.
And we show the employment rates for each of the nine years, one year before application,
and the nine years after VR application.
And you can see that in the ninth year after VR application, about 45% of youth with mental
health conditions who receive services other than postsecondary education support are employed.
And we now compare this to the employment rates of those who do receive postsecondary
education support.
So, the blue line on top, the dashed blue line, are those who receive college support,
and the green dotted lines are those who receive vocational training.
And you can see that those who receive postsecondary education support have higher employment rates
in every year than those who receive other types of support.
But this includes the year before application, which suggests that there is a strong selection
among those who receive postsecondary education services.
Nonetheless, we do see that the gap in the employment rates between those who receive
postsecondary education support and those who do not does get wider over time.
Next, we look at the average annual earnings of youths with mental health conditions, and
we start with those who are receiving services other than postsecondary education support
and compare it to those who do receive postsecondary education support.
So, again, it's even more clear here that the gap in earnings between those who do and
do not receive postsecondary education support is really growing over time, especially for
those who receive college support, which is the dash blue line on top.
As I mentioned, we want to make sure that these -- or we want to check to see if these
relationships still hold after we control for individual and state characteristics.
So here, we show the results from two sets of regressions.
The two bars on the left show the results when the outcome is whether the individual
was employed in the ninth year after VR application, and the two bars on the right side show when
the outcome is the log earnings in ninth year after VR application, and the blue line show
the coefficient on receiving college supports, so the estimate is the difference in the outcome
from receiving college support versus not receiving college support after controlling
for individual and state characteristics, and the green lines show the same for receiving
vocational training.
So, here, we can see that the regression estimates confirm what we saw in the descriptive analysis,
which is that youth with mental health conditions who are receiving postsecondary education
support, that they tend to have higher earnings, and they are more likely to be employed in
the ninth year after VR application than those who do not receive those supports.
Next -- I know.
Should I be pointing it at something?
Next, we look at the receipt of federal disability benefits.
For these analyses, we split our sample into two groups.
We look at those who were receiving benefits at the time of VR application and look to
see whether they had those benefits foregone in the subsequent nine years, and then we
look at those who were not receiving benefits at the time of application and look to see
if they started to receive them in the following nine years.
So, I'm going to start with the results for youth with mental health conditions who were
not receiving benefits at the time of VR application.
In the first row, you can see that those who received college support, a smaller percentage
of them received federal disability benefits in the subsequent nine years than those who
did not receive college support.
And the same is seen for vocational training.
So about 12 to 13% of those who receive postsecondary education support subsequently receive benefits,
compared to 19% of those who did not receive any type of postsecondary education services.
However, among those who did receive benefits in the following nine years, the number of
years they receive benefits for was fairly similar between those who did and did not
receive postsecondary education services.
Once again, we checked to see whether these relationships still hold after controlling
for individual and state characteristics.
On the left side, the outcome variable we're looking at is whether the individual received
SSA benefits any time during the nine years after VR application, and on the right-hand
side we look at the number of years of benefit received, conditional on receiving benefits.
And just like we saw in the descriptive analysis, we can see that those who receive postsecondary
education services are less likely to receive federal disability benefits in the nine years
than those who did not receive postsecondary education services.
We also see a slight decrease in the number of years of benefits for those who receive
postsecondary education services, but these numbers are quite small, and they are not
always statistically significant.
And our last group that we look at is youth with mental health conditions who were receiving
federal disability benefits at application, and we look to see whether those benefits
were suspended at any time, or over the next nine years.
And we can see that the percent of youth that had their benefits suspended were fairly similar
between those who did and did not receive postsecondary education services; however,
among those that did have their benefits foregone for work we see that those who received college
support or vocational training had a much larger amount of benefits foregone for work
than those who received other types of services.
Again, when we conduct our regression analysis we find very similar findings.
We find that those who receive postsecondary education support had a much larger amount
of benefits foregone for work than those who received other types of services.
So, to summarize, we found that for youth with mental health conditions receiving postsecondary
education support was associated with a higher likelihood of being employed and higher earnings
in the ninth year after VR application, and we found that for those who were not receiving
benefits, federal disability benefits at the time of application, that there was a lower
likelihood of receiving benefits in the subsequent nine years, and among those who were receiving
benefits at the time of application, we found that there was a larger amount of benefits
foregone for work for those who receive postsecondary education services.
So, I want to emphasize that the relationships we found here are not causal.
They're merely intended to set a baseline for the relationship between postsecondary
education support and outcomes, and we found that, based on our descriptive analysis and
our regression analysis, that there does appear to be a positive relationship between postsecondary
education and outcomes.
However, in order to determine whether this is a causal relationship, it's important to
conduct a rigorous evaluation, possibly through a demonstration with some form of randomization.
Another thing that will be important to determine, besides causality, is really thinking about
the cost and benefits of providing this type of support.
So past research that we've done has shown that providing support for postsecondary education
is actually quite expensive relative to other services.
The cost is estimate today be about $2,600 to $7,000 higher compared to other youth and
mental health conditions who receive other types of supports.
However, as we showed in this project, that the benefits foregone for work is along a
similar amount, about $2,100 to $5,000 higher for those who receive postsecondary education.
And this doesn't include all the other benefits from working; for example, the income that
they earned.
So, when think thinking about next steps in looking at VR support for postsecondary education,
it's really important to determine causality through rigorous evaluations, and also consider
these costs and benefits.
So, thank you very much.
All right.
Thanks very much.
Can you hear me?
I'm Manasi Deshpande from the University of Chicago and NBER.
Thanks very much for inviting me to discuss this paper.
So, first, I want to start just by providing some context for this paper and the broader
research and policy context, and then I'll talk specifically about the paper, and then
tie it back to some of these bigger questions.
So, you know, as Priyanka said, the VR program provides services to individuals with disabilities
to promote employment.
In 2016, the federal government spent $3.1 billion in federal grants to state agencies
to implement the VR program, and services include counseling, training, referrals, transportation,
and postsecondary education support.
Now, the 2014, the new VO requires VR agencies to spend at least 15% of that federal funding
on youth services, and they're probably people in the room who know more about this than
I do.
But at least in my work with agencies in the past few years, this appears to be a binding
constraint for most of these agencies.
So, most of these agencies are doing a lot of work to increase their preemployment transition
services for youth with disabilities to be able to meet this 15% requirement.
So, the sort of broader research and policy context here, I think, there are two major
questions with respect to VR and youth right now.
The first is how cost effective is VR in improving outcomes for youth with disabilities, and
the second is how do we connect more youth to VR services?
So, I feel obligated as a researcher to say that, ideally, we would know the answer to
the first question before working on the second question, but, you know, if we only funded
or did things for which we had evidence, there would probably be three government programs
in the United States.
So, with respect to the first question of how cost effective is VR, this is something
where we don't really have, to my knowledge, any causal experimental evidence of this question.
I think it's an incredibly important question, given how much money we spend on VR.
The PROMISE Program, the Promoting Readiness of Minors in SSI, will be evaluating a bundle
of services for SSI youth that includes both rehab, but also includes a lot of other services,
so I think that's kind of the sort of most promising evaluation right now to get a sense
of how effective VR is.
The second question of how to connect more youth to VR service, you know, so SSA doesn't
provide VR referrals to SSI youth, and there's a recent GAO report that found that schools
are supposed to be the primary connection between these youths and both rehab agencies,
but schools are often doing a poor job.
And in other work with my co-author Rebecca Dizon-Ross, we've conducted a lot of focus
groups on SSI youth and their families, and, specifically, what do they know about the
Voc Rehab Program, and we found it's somewhat heterogenous across states.
But in a lot of states, these families don't seem to be aware that VR services are even
available.
When you tell them what services are available through their VR, they express strong preferences
about what they think would be appropriate for their child and what services probably
wouldn't help their child.
And so, it does seem like there is a lot of capacity to provide more information about
voc rehab to these families.
So, with respect to the Priyanka and Todd's paper, just to provide an overview, the main
question is how do employment earnings and disability receipt vary by receipt of college
supports and by mental health conditions.
And they're using VR administrative data linked to SSA earnings and disability receipt data.
And I think, you know, the administrative data linkage here is a big contribution.
It allows them to look at long-term outcomes.
It allows them to look at employment and disability receipt nine years after the VR application.
So, their main findings are that youths who receive college supports are more likely to
be employed and less likely to receive disability benefits nine years later than youth who receive
other types of VR services, and that these differences are larger for youth with mental
health conditions than youth without mental health conditions, so Priyanka, you know,
emphasized in her presentation, mental health conditions, but in the paper, they also have
results for other youth.
So, what we learned from this paper, I think, so first, as I said, I think it's a very important
topic, and, you know, especially the linked administrative data lets us look at long-term
outcomes, and so their descriptive results allow us to compare long-term outcomes across
groups.
So, in particular, what they find is that employment rates nine years after VR applications,
employment rates are about 50% for those who apply for VR, among those with mental health
conditions.
They're much higher for youth who apply for VR and receive college supports, and then
for youth who receive other VR services, employment rates are similar to the rates for those who
just applied for VR.
And then for non-mental health conditions, you can see that overall employment rates
are higher and the patterns are similar.
So, you know, as the authors say, and I think it's important to reiterate that these results
are not causal, the differences in these groups in the graph that I just showed reflect a
lot of different things, so they reflect selection into college, they reflect the treatment effects
of college itself, they reflect selection into college supports, and then they reflect
the treatment effect of the college support.
The treatment effect of the college supports is, I think, really, what we want to know
in in paper, ideally.
But we're also going to get a lot of other factors that account for these differences,
and this the context in which selection is probably very important; right?
But youth who apply for and receive college supports probably did so because they were
planning to go to college; right?
They are selecting into college and then they're also, because they went to college, many of
them went to college, they will also have higher earnings and employment rates for that
reason.
And, you know, as Priyanka mentioned, you can sort of get at this by looking at the
difference in the year of VR application, or just before VR application, and you can
see that employment rates are higher for those who are already receiving -- applying for
college support.
Even that is likely to understate the amount of selection, right, because the youth who
apply for college supports might already be in school full time, and so they might not
be employed because they're in school.
And so, even that sort of methodology of using the preapplication difference in employment
rates is probably going to understate the amount of selection.
So, in terms of interpretation, two thoughts; one is, it would be helpful to know more about
what the comparison treatment is.
So, the authors are comparing college supports to other VR services, and the main reason
they give for this, which makes sense, is that they want to avoid picking up selection
into VR.
So, people who apply for VR and don't apply for VR are different in various ways, and
so they're just focusing on the group of people who apply for VR and, then, comparing within
that group, people who receive college supports versus people who receive other types of services.
But, probably, you know, ultimately what we want to know is what is the effect of college
supports versus no VR services.
And one way to get at that might be to do some kind of matching between the youth who
receive college supports and youth who don't apply for VR, and I realize that in administrative
data that's difficult because you just have such limited covariates.
And another question is what is the comparison group.
So, I'd like to know more, in particular, about this non-mental health condition group.
So, my sense is that it combines youth who have non-mental health disabilities for whom
we might expect similar outcome or worse outcomes than the youth with mental health conditions,
but also, youth with no disabilities or less severe disabilities for whom we might expect
better outcomes.
And if the non-mental health comparison is combining those two groups, it's a little
hard to know how do I compare the mental health group with the non-mental health group, so
some clarification on composition and characteristics would be helpful.
Okay.
And then just returning to those two questions from the first slide, how cost effective is
VR in improving outcomes for use with disabilities, you know, reiterate what Priyanka said, that
there is a need for causal evidence in this area, especially given how much we spend on
VR, and especially given these new [inaudible] requirements to dispend more on these services
for youth.
And I think that the new [inaudible] requirements provide an opportunity for federal and state
agencies to evaluate youth VR services.
So, as I said, a lot of these agencies are ramping up or investing in their preemployment
services, and they are really interested in what works and what doesn't work, and I think
there is an appetite for evaluating some of these services for youth to understand what
works.
And then, with respect to the second question of how to connect more youth to VR services,
you know, we know that it seems to be that families have pretty limited information about
the availability and quality of VR services.
The recent GAO report recommends that SSA options for increasing connections between
these SSI youth and VR agencies and services.
Thank you.
All right.
Last up on our panel is Seth Seabury from the University of Southern California, and
he's going to talk about the impact of Medicare Part D on Expenditures and utilization of
social security disability insurance beneficiaries.
And his discussant is Gal Wettstein from the Retirement Research Center at the Boston College.
And, you guys, red light means stop.
Green light means go.
Send an electric shock.
Well, thank you.
It's great to be here.
This is joint work with Amitabh Chandra from Harvard, and Ning Su, a post-doc at USC with
me, and this is very much research in progress, so I'm looking forward to getting your input
and sharing with you, and also getting your thoughts on other directions we can take it.
So, the goal of this research is to look at how the introduction of Medicare Part D affected
people on SSDI.
We're looking at under 65 Medicare beneficiaries specifically.
So, in 2006, as probably almost everyone here knows, Part D offered prescription drug coverage
through Medicare to beneficiaries for the first time, and this included people who were
eligible based on being over 65, as well as people eligible through their disability status
on SSDI.
And, so there's reason to believe that extending prescription drug coverage or improving prescription
drug coverage for people with chronic health conditions, disability, could have a big impact,
bigger than, say, even the general population.
Just by the nature of their disability, they often have greater health needs.
Prescription drugs can potentially play a big role in managing their health condition,
and also these are individuals or families with relatively low income for whom out-of-pocket
spending on drugs could prove a bigger financial burden.
So, we might expect whatever effects that we think Part D had on the Medicare population
more generally, we might expect effects to be even bigger on the under 65 disabled population.
However, you know, like any sort of worthwhile endeavor, it's always a little more complicated
than sort of a simple story; right?
And part of the problem here, that quickly became apparent to us as we started working
on this area, is that the nature of the individual's disability expanded their eligibility for
other types of programs, even prior to part D. So, in particular, this Medicaid dual eligibility
is what we're focusing on mostly here.
But this was something that became apparent to us, was a big deal that we needed to consider.
Medicaid typically offers prescription drug coverage, and they did so prior to Part D.
But that drug coverage varies a lot across states in terms of the generosity and what's
covered and how much they'll pay for, and so there's potential sort of confounding variation
that we're not going to observe very well amongst people with dual eligibility, and
there's also this issue that dual eligible beneficiaries were switched automatically
onto Part D so that it's not just that there's potential heterogeneity, there was a change
around 2006.
So, we want to be careful thinking about other program status when trying to estimate this.
Now, this Part D was obviously one of the biggest social insurance changes in many,
many years, especially prior to the ACA, and so it generated a lot of work amongst economists
and health policy scholars.
But most of that work has focused on aged beneficiaries, people over 65.
And I'll just give a very quick high-level summary of some of the existing findings related
to what we're doing here.
Part D was associated with what I would call a modest uptick in drug spending and utilization
for over-65 beneficiaries.
Part of the reason that it's called "modest" is that there appeared to be a lot of crowd
out of private drug insurance, but there was a large reduction in out-of-pocket spending
for the over-65 beneficiaries.
And the studies I've seen had relatively little to no impact on health outcomes or healthcare
utilization, so what we think is that there was some increase in spending, some increase
in utilization, not obvious impact on health.
But there really hasn't been a great deal of study of this in SSDI beneficiaries.
So, what we're doing is we're looking at the impact, so we're going to be looking at the
introduction of Part D in 2006, and how it affected prescription drug spending, the number
of prescription fills for the under-65 Medicare population.
We're also going to look at hospitalizations, a kind of key outcome, and also look at out-of-pocket
spending.
And some secondary outcomes we're going to look at include self-reported health, receipt
of samples, free samples, type of pharmacy used, and office-based provider visits.
Talk about, you know, how to interpret those in light of our primary outcomes in a bit.
And then, the second part of the study where we want to get to, I'm not going to talk about
it a lot today, because we still have much more to do on that.
But because we're basically going to be doing a difference-in-difference analysis, we're
certainly worried about potential changes in the beneficiary pool, and, in particular,
whether the availability of prescription drug coverage changed who is applying to SSDI.
And we've done a little bit of -- we're very early stages of work on that.
We're not finding obvious compositional changes, but there's much more to do.
So, I'm not going to talk a lot about that today.
Not because it's not important or we're not concerned, but it's just we're definitely
we're still working on that.
So, the data that we're using come from the Medical Expenditure Panel Survey.
We're using a set up very similar to the Engelhardt and Hoover paper on Part D, from 2011.
We're combining data from the full year consolidated and the prescribed medicine files.
We're focusing on 2001 to 2009.
So, we stopped at 2009 in our data, partly because of the ACA and all the -- or primarily
we stopped in 2009 because we wanted to stop pre-ACA and all the potential changes that
could have been introduced from that.
But, also, actually, in light of what Kathleen and Nicole said, it's actually helpful for
us to stop there, happy accident, I guess, because we weren't thinking about this, but
there could have been a lot of compositional changes from the great recession, and the
two-year waiting period for SSDI to get on Medicare helps us avoid those problems up
to 2009.
And, so, we're focusing on individuals 18 plus, with 12 months of Medicare coverage,
so we're looking at people covered in Medicare for the full year.
We're going to stratify based on over and under 65.
So, I want to be clear, that I'm calling them SSDI beneficiaries.
There are some under 65 Medicare who will not be on SSDI.
They could qualify through end-stage renal disease or ALS, but the majority will be on
SSDI.
And we're also taking kind of a loose definition of dual eligibility.
There are very different levels of Medicaid eligibility.
It's going to vary by state.
And a lot of ways we can't handle in the MEPS, so we're taking just kind of very expansive
definition and we're saying if you have at least one month of Medicaid coverage in the
sample then we're going to sort of call you different than the people who are 12 months
with no Medicaid.
So, that's how we're breaking dual eligibility.
And then we're also going to look at individuals 18 to 64 with 12 months of private insurance
coverage.
So, you can think of four distinct groups that we're interested in.
We have the under 65 Medicare only, under 65 with also Medicaid, over 65 Medicare only.
We're simplifying and excluding the over 65 duals, and under 65 privately insured, so,
four different groups of people based on their insurance coverage.
And the point is, we're taking them and we're using those different groups in two sets of
difference and difference regression based on comparing under-65 beneficiaries as the
treated group to controls, pre- and post-2006.
So, you can think of we're doing regressions, we're comparing under 65 Medicare to privately
insured and now we're thinking of it as a treated group compared to an untreated group.
Then we're also going to look at under 65 Medicare to over 65 Medicare, two treated
groups but in sort of more of a dose response type of model.
We're curious as to whether the impact of introducing Part D had a bigger effect on
the under 65 Medicare population compared to over 65.
And the other covariates include standard demographics; age, sex, race, education.
We're also looking at income and other chronic comorbidities that are reported in the MEPS,
so self-reports for ever being diagnosed with heart disease, stroke, emphysema, diabetes,
or asthma.
It's just we're trying to control for possible changes in the underlying health of beneficiaries.
And just we're also using MEP sample ways, and for any regressions we exclude 2006 as
a transitional year, and we're looking at 2001 to 2005 compared to 2007 through 2009.
A few figures here just to show.
They're basically the -- it's probably hard to distinguish some of these, but the top
two lines here represent prescription spending in the year for the under 65.
The red represented those with Medicaid.
The blue represents those without.
And you can see where the next down represents the over 65 Medicare, and the lowest spending,
as you would expert, are the under 65 private.
And most of the level differences here are due to differences in underlying health, driving
spending, and you can see there's really no change post-2006 in spending for the under
65 private.
Given the scale, it's a little bit hard to tell, but there is a bump in the over 65 Medicare,
and while there's definitely a pre-trend that we need to be concerned about, there seems
to be a trend break around 2006 for the under 65 Medicare.
Now, if we look, what's interesting is we see what we think is an increase in spending
for the under 65 Medicare, but we don't see a change in the number of prescriptions.
So, these are the average number of fills in the year for the same insurance groups,
and what you see basically is, again, private very flat.
There's a slight uptick in fills for the over 65 Medicare.
But for the under 65 Medicare, we basically see no evidence of any trends.
Now it's a little bit more variable, so it could be that there's an increase that we're
not able to capture.
But we're seeing an increase in spending but no evidence of increased utilization.
Now, finally, when I look at hospitalizations, there, actually, we see a pretty clear decline
for the under 65 Medicare, but not for either of the other two groups, so hospitalizations
for private are completely flat.
It's a little noisier for over 65 Medicare, but there's no obvious decline.
But for both under 65 Medicare only and under 65, what I'm calling dual, you see a drop
in the percent of people with a hospitalization in the year in post-2006.
So, I'll talk about what I think that means in just a second.
I'll show you the regression findings, what are the magnitudes we're talking about here.
If we look at spending versus privately insured, in the top panel, there are two rows in each
panel.
The top row is the difference.
The bottom row is the difference and difference.
You can see that Part D was associated with a $944 increase in prescription drug spending
for the under 65.
It's actually higher for the dual eligibles, although in percent terms it's pretty similar
between the Medicare only and the dual eligible.
So, we definitely see an uptick in prescription drug spending, as we saw in the figure.
And if we compare the under 65 to over 65 Medicare, you also see an increase, about
half as large.
So, you see definitely a larger increase amongst the under 65 compared to over 65 in terms
of the higher spending.
Now, also, though, as we saw in the figure, there's no evidence of any increase utilization,
and if anything compared to the over 65, you see a decline in the utilization.
These are large standard errors.
We can't rule out somewhat of an increase, but we're certainly not finding any evidence
of it.
So, on the surface, this is potentially problematic; right?
We're expanding drug coverage to a group because we'd like to see them get more access to medication.
And so, if you see higher spending, but no increase of higher utilization, then it's
hard to see whether you're actually getting a real increase in access.
So, unless you think they're getting better drugs, this would just seem like you're driving
up costs with not necessarily any clear benefit.
Now, we do see a drop in hospitalization, so you might wonder, are you seeing a health
benefit from that?
You know, as I'm used to work with claims data, we always use claims as a measure of
health, as a proxy for health.
But, in reality, hospitalizations reflect both health, as well as patient preferences,
as well as provider preferences.
So, one of the hypothesis, this is somewhat speculative, we've got some supporting evidence;
that we think people are maybe going to the hospital less to get medication, going to
the hospital, using the hospital pharmacy might have been a way that people with poor
drug coverage were actually getting access to their medication.
So, we think more than a clear improvement in health, it might be a behavioral change.
And this might still be good.
You're saving money by keeping people out of the hospital.
But I'm reluctant to claim it's a big improvement in health, particularly given that we're not
seeing an improvement in actual drug utilization.
One thing we are seeing, however, is a clear financial benefit to the SSDI, beneficiaries
in terms of lower out-of-pocket spending.
Now this conditional on having at least one prescription fill, but the annual out-of-pocket
spending for the Medicare-only group fell by a thousand dollars, so they're spending
a thousand dollars less on out-of-pocket drugs.
That's like giving them essentially 7% of their family income.
So, it's actually a pretty big financial for them from the introduction of drug coverage.
This is about to blow up on me, so I've definitely gone too long.
We find some evidence supporting the idea that we think that there's change in provider
behavior.
Sorry.
We also did see some improvement in self-reported health.
But whether that's actual physical health or just perceived wellbeing, I think, is an
open question.
So, probably raised more questions than answers with this work so far.
We think it's interesting findings, but there's definitely more than we need to do to understand.
Okay.
So, hello, and thank you to the organizers and the authors for giving me the opportunity
to give some comments on this work.
I won't spend too much time summarizing, because you just heard it.
But, basically the paper looks at the utilization of prescription drugs and other health services
by those on SSDI and how that changed with Part D.
They used two control groups.
One is the under 65 who are privately insured as a control group to estimate the total effect
of Part D, and the other is those who were over 65, and what I'm going to call traditional
Medicare to estimate the differential effect for those on SSDI, relative to the more elderly
population.
And their findings were, as we just heard, an increase in prescription drug expenditures
for those on SSD, both those who were newly insured, and those who were, as the authors
called them, dual eligible.
There was decline in hospitalization and a slight increase in self-reported health, but
no change in fill.
And so most of my comments are going to focus on some ideas on what might be going on.
Methodologically I don't really have any comments that the authors didn't already raise, except
that when it comes to the differences between the dual and newly eligible, I think it's
correct, as Seth said, that it's hard to differentiate completely between these two groups, and so
any such differences might be understated.
So, why could there be an increase in expenditures without more fills?
One possibility the authors raise in the paper is that it could result from increased brand
name drug use as opposed to generics.
This could be consistent with the fact that they see an increase in observed expenditure
for the dual eligible, because even though those duals already had, presumably, prescription
drug coverage through various state Medicaid programs, those often had restrictions on
the generic drugs that could be gotten through that insurance, so an increase could be due
to this.
Well, another possibility that I thought of, given the improvements in hospitalization
and self-reported health, although, as Seth said, there's some different interpretations
of those results as well, there could be a switch to using drugs that cure conditions
rather than just reading symptoms, and the example I had in mind is the new hepatitis
C drugs that sort of came on the market.
Clearly, the increased use of that drug is not due to Part D, it's because the drug was
developed.
But other such similar stories could have been going on, and that could lead to improved
health in cases where people are less price sensitive
to these more expensive drugs, like if they get more drug insurance.
And then similarly, but not exactly the same, because of the decline in fills, even though
it wasn't statistically significant, one thing that stood out to me is it could be people
are using more combination drugs when they get more drug insurance.
Often these combinations of drugs are marketed when the component drugs, patents, are about
to expire, and so they would remain generics for -- sorry -- they would remain brand name
even past the time the components drugs were already generically available.
And, so, if Part D increases these combinations drug usage, it could be a useful tool to study
how effective these are relative to their costs, which is a hot policy topic.
There's a decreased pill burden, perhaps better usage of prescribed drugs.
But on the other hand, the price is higher because there's less competition.
Let me wrap up with just a couple of further potential questions.
The study finds larger SS effects on utilization, and mostly on health, for the SSDI population
than previous literature has in the general population.
I think that raises questions, as Seth said.
Perhaps when we want to look at the impacts of Part Ds, it would be useful to look at
more heterogeneity in those impacts by health status and income, which are obviously correlated
with SSDI status.
And the similar effects for dual and newly eligible is interesting, and I think raises
interesting questions on not just whether people are insured but on the quality of that
insurance, what kind of products are covered and how generous the cost sharing is.
So those are my comments for now.
Thank you.
All right.
We've got about ten minutes for questions.
Anyone have any?
All right.
Microphone girls.
Hi.
Good afternoon.
Just a little bit of background, I'm here today because I have a Smartphone app that
changes the way companies source, hire, and train people with disabilities, and so my
question is for the second one, is we have found, because we have this big dataset, that
we don't know how to analyze, so if anybody wants to use our big dataset, that we don't
analyze very well, we found that it's actually more connected to somebody's network, so,
the idea that it's a lot easier to get a job if you know someone.
And so, when we, in our mediocre basic understanding of statistics, analyze people that come with
a college education that sign up for our platform, versus somebody that didn't attend college,
they actually don't -- that doesn't really change things.
More so the network does.
And so, are there ways for you to accommodate or considerate the network effect that happens
when people are going through, or is that something that's in anybody's research?
That would not be something we could do with this data that we have now.
But I think it's a great point that there are many benefits to receiving a postsecondary
education, and I think the network effects through going to college is one of them.
Yeah, the administrative data we have wouldn't tell us exactly, you know, who are the networks
that they're coming across while they are receiving their education, but I imagine that
there are other ways that -- not through our data, but there are other ways that we could
look into this really important question, because I think it's a great point; that part
of the advantage of getting this education is the network you come across, so I think
it's a great point.
My name is Sharon, for the first speaker, what is the proposed legislation for the CHIP
program.
I have a nephew who is on the CHIP program because he's a juvenile diabetic, so what
are the proposed legislations?
Would it be lessening, and if they had to lose that, would they be able to roll over
to the Medicaid program, or what could they do?
I might refer you to Steven, who might know more about that.
I know more of the disability side than the CHIP side.
I'm not sure what legislation is currently on the table.
But what is happening is that CHIP is only authorized for spending through the end of
the fiscal year, but states can continue to keep the money that they have and spend it.
And then I think, depending on the state, they'll run out of money.
So, it's a question of reauthorizing the same program, changing it.
The Medicaid Payment and Access Commission has been looking at some ideals of what could
be going, what would be solutions to coverage for kids, but, generally speaking, it's not
clear what's on the table.
Okay.
Manasi.
Thanks.
I had a question for Mike, actually.
Great paper.
I was wondering what you know about the effect of these Medicaid and CHIP expansions on just
any health insurance coverage for potential SSI kids?
I think in the general child population, we know that the increase in public insurance
coverage is much larger than the increase in having any health insurance, because a
lot of kids are going from private health insurance to these public programs when they're
expanding, and so a lot of that effect is just in income effect, of, you know, putting
the kid on public insurance instead of private insurance.
And, so, is that also true for the SSI child population and how should we think about the
income effect versus the gain in health insurance?
I'm not sure that I have a great answer.
We haven't looked that much into like how much -- kind of the first stage of how health
insurance changes as a result of the estimated changes in eligibility, and kind of speaking
to the income effect as well, it's hard.
Is that something you're doing in your work with survey data?
So, we can look at that in -- we have some control variables in our regression in the
CPS, where everything is, like, aggregated to the population level, so we could look
at how the changes in simulated eligibility affect the likelihood of having any health
insurance coverage to get at, is people who are newly qualifying for coverage versus people
who are, you know, just changing where that source of coverage comes from.
So that's something I think we can look into, and we will.
Thanks.
Did I see Gil's hand up there?
Go.
Hi.
So, Gilbert Gimm from George Mason University.
This is a question for Seth Seabury with the third paper.
So, have you considered looking at Medicare Part D's six protected classes of medications
to help explain your findings?
In essence, when Part D was authorized under the MMA, six medications that fall under,
HIV, mental illness, epilepsy, cancer, and I think organ transplantation, were prohibited
from being excluded from any formularies that private insurers developed through Part D
contracts.
And given that you appear to have an expenditure increase but no change in utilization, you
could argue that it's largely price that's driving your result.
If a state has Medicare prescription coverage that's pretty crappy for adults with disabilities,
moving to Medicare Part D would significantly increase the number of antidepressants and
anti-antipsychotic medications available, because they could not be ex excluded based
on price.
So, I guess the two-part question; one, have you considered that; and then, two, in the
MEPS data, do you have the ability to identify people with a mental illness, cancer, or some
of these categories that could help you to tease out the effect.
Yeah, that's a great point.
I think that the answer is, yes, one of the things that we want to take -- one of the
directions we want to take is to look at the actual types of medications that people are
spending and whether that changes pre and post.
I hadn't specifically thought of the protected class, but I think that's a great question.
We'll definitely want to look into that.
And in terms of mental illness, one thing that the MEPS introduced in 2004 was the K
Six screening tool, and so we actually looked at that to see whether there was any change
pre or post, and tried to see if there were any compositional changes.
But we could also try to see what the evidence, psychological distress, and see if there was
any kind of heterogenous effects in that population.
That would be one way to try to get at that issue.
I think that's a great suggestion.
All right.
We have time for one more.
I haven't done this side of the room.
Okay.
Hi.
My name is James Smith from BR Agency and this question is for the second presenter.
First, I just appreciate you looking at long-term outcome data for VR services.
That's really needed.
I also appreciate you pointing out that it's not necessarily causal relationship.
One thing I just ask you to consider is, one of the things we observed in Vermont is that
we definitely have young folks who come to us, where the VR intervention is critical
in terms of supporting going to college, and then there are definitely another group of
young people where it's a supplemental support, but, you know, whether VR existed or went
away tomorrow, they would go to college.
And so, I think teasing that out in your -- I mean, that may be possible to tease that out,
but that's a key thing.
Also, another sort of issue that does come up for us is, if state VR agency, for us in
Vermont, if we provide direct tuition assistance, the State Tuition Assistance Grant is immediately
reduced dollar for dollar, so there's no actual benefit to VR funding tuition for the student.
And so, you know, it's never as clean as it seems.
The last, just quick comment, was on preemployment transition services, I just wanted -- people
keep saying it's for youth.
It's not.
It's for students.
It's for high school students, and it specifically cannot be used to fund postsecondary education.
So that's 15% that's actually been taken out of the VR resources to actually fund postsecondary
education or any other vocational services.
It's specifically for youth, for students, so I guess that wasn't a question.
Sorry about that.
Those are really helpful observations and suggestions, so I appreciate very much.
Thank you.
All right.
Give a hand to the panel.
Come back in 15 minutes.
For more infomation >> 노현정 관상 눈길…"두툼한 콧방울과 귀, 전형적 부자 여인의 상" - Duration: 2:29. 
For more infomation >> Malgré son tatouage, Miss Provence reste qualifiée pour Miss France : les internautes - Duration: 3:22. 
For more infomation >> Shania Twain On Rediscovering Her Voice & If She'd Join The 'American Idol' Reboot - Duration: 7:53. 
For more infomation >> Episode 2 (cont'd) - Pizza Cheers! - Duration: 5:05. 
For more infomation >> Episode 3 (cont'd) - Newcomers - Duration: 6:29. 
For more infomation >> Episode 4 (cont'd) - Hanks And Neruda - Duration: 5:25. 
For more infomation >> Carla Bruni et sa fille Giulia complices lors d'une virée en bateau - Duration: 1:49. 
For more infomation >> Mitsubishi L200 L 200 2.5 DI-D Club Cab Invite 4x4 - Duration: 0:54. 
For more infomation >> Un exemple d'application de l'approche inclusive : Frédéric Bussières - Duration: 3:38. 
For more infomation >> Un exemple d'application de l'approche inclusive : Claude Désy - Duration: 3:21. 
For more infomation >> Doutzen Kroes pose sans maquillage et elle est sublime - Duration: 1:48. 
For more infomation >> The Fantastic Action of Hollow Knight - Duration: 5:21. 
For more infomation >> Qui est Natali Husic, la nouvelle chérie de Louis Sarkozy ? - Duration: 2:28. 
For more infomation >> GoogleMap Oyunu :D - Duration: 6:47. 
For more infomation >> ATTENTAT À BARCELONE : UNE FOURGONNETTE PERCUTE LA FOULE ET TUE AU MOINS DEUX PERSONNES - Duration: 3:43. 
No comments:
Post a Comment