Actuarial Outpost > SoA Task 1 EOM Exercise
 Register Blogs Wiki FAQ Calendar Search Today's Posts Mark Forums Read
 FlashChat Actuarial Discussion Preliminary Exams CAS/SOA Exams Cyberchat Around the World Suggestions

 Health Foundations Module Old Health Systems Overview Forum

#1
12-19-2018, 04:32 PM
 Sarah02421 Join Date: Aug 2008 Posts: 18

Task 1 – Health Foundations Module EOM Exercise

Hello. The google group that I started has grown so large that we’ve decided to take our conversation back to the Actuarial Outpost. Below, I have copied all of the posts for Task 1. To make the conversation easier to follow, I’ve put “New Post” before each post. The conversation began on November 5, 2018. I hope you find the conversation as helpful as I have. Thanks.

New Post
Hi. I'm going to start a new topic for each of the tasks. Hopefully that will keep confusion to a minimum as we talk.

New Post
“otherwise, about task 1, I am not sure what does the prior assumption mean. does that mean that I should assume the range of each element before check the data? what do you think”
I think that explaining prior assumptions means talking about what we expected to see for the data. I don’t think it means giving specific numbers. One example might be that I would expect the demographic index to increase as a person gets older.
“also task 1, whether the year is distinguished by policy year? or calendar year? in fact, I had assumed it is policy year, so I am confused about why the year 2's member months is greater than year 1's.”
I don’t think it matters if it is policy year or calendar year. Since employees come and go, I would expect that the member months would not match between years 1 and 2.

New Post
Hey, I'm just getting into the module task questions. As always, lots of confusion.
For this question, it seems like they are just asking:
1) whether we agree with the calculated YOY normalized PMPM trend of 3.3%, and
2) to state any assumptions we have made.
From what I can tell, the normalized PMPM & trend calculations both seem fine. For the second part of the question, I guess I am assuming that all of the data elements have been summarized correctly in the table (i.e. mem mths, net PMPM, demo index, and benefits index were all calculated correctly). The only other prior assumption I feel comfortable addressing is that the 3.3% trend is outside the 8%-10% range assumed in the PDF. Does this seem too simple?
Also, I agree with Sarah that the calendar/policy year distinction is not important to the question, although you could note in your assumptions whether you choose to recognize one or the other.

New Post
Hi. I've been thinking that assessing the quality of the data means checking if it is correct and reasonable.

- For member months (MMOS), it's hard to know if the they are correct since the data from the DW doesn't match the data from ABIS.
- The demographic index does not seem reasonable to me since the value drops at age 80. I would expect the index to increase as the members age since they are becoming less healthy.
- I also noticed that the input to the benefit index model does not match between the product area and the ABIS input data.

What do you think?

New Post
I agree with all three of your bullets, and you are probably right that we should address the quality of each data element. In that case, I think am going to explain what is wrong with each of the elements in Task 1, and then recommend changes & recalculate in Task 2. Both seem like they will have pretty straightforward and brief responses.

New Post
I agree with you, but I have different thought about the demo index. In my country, this result is appeared when the business group is small. there is many reasons to cause that, such as exposure, policyholder behavior, or grouping method used to analysis.

New Post
I've gone through most of the workbook for Task 1 and found several issues that I think need to be fixed (e.g. wrong benefit index factors, demo index 1.0 for ages > 79, benefit plan data output issues, etc.). However, my concern is with the provided Net PMPMs. It appears that the Net PMPMs provided are for SomeState and OtherState combined, but the Benefit Index factors are only for SomeState (i.e. there's no adjustment for plans G, H, and I from year 1 to year 2). Are you making any additional adjustments for OtherState? Or are you planning on leaving it and stating some assumption about it?

New Post
I am not making any adjustments for Otherstate plans, and assuming that since we are supposed to be focusing on Somestate, no members should be enrolled in any of the Plans G-I anyway. That raises the question of why total membership in the ABIS report doesn't match the membership data gathered by Calluna, so I am also assuming that although total membership does not tie to the ABIS report, the member mix by plan should not vary materially.
And for the Net PMPMs, I am taking from the PDF that Calluna has "competently gathered" the Net PMPM data (even though it doesn't appear that she competently gathered the member data from the same source). It is annoying that they say we can make this assumption in the memo, but then the first two tasks are basically asking us to pick apart her work.

New Post
I have another question. The instructions say that the demographic index is a new field to me "as decribed on Page 26 of Section 4 of the module."

Page 26 of Section 4 (1.4.26) is titled "Is the field populated accurately?" This page talks about 1 of the 5 steps to check when using a new field. The 5 checks are listed on page 1.4.23 which is titled "Things to check".

Page 1.4.39 which is titled "New field recommended steps activity" also lists 5 steps to follow when using a new field.

Is anyone including a reference to the demographic index as a new field? If so, are you citing either list of 5 and which one?

New Post
I did not reference it as a new field or cite the list, but I did end up working some of those points into my memo. It's don't think it's used to adjudicate claims, and it seems inconsistent / inaccurate. I dunno that there is much more to mention beyond that.
- show quoted text -

New Post
I have different understanding. with old version, page 26 of section 4 is titled " thing to check." so, I think the question ask us to think about those 5 step before using the demo index, and then I have replied the 5 step in the memo. maybe it make more sense to use the description with old version.

New Post
I did quote slide 26 and went through the 5 steps. I started my module a while ago so I'm not sure if my slide 26 is the same as you guys. I just got my DNMMR today. Not sure if this is the reason why.

New Post
Thanks for the responses on Page 26. I am going to reference the 5 "things to check".

xaznstylegrix - I'm sorry to hear about the DNMMR. I hope you don't mind my asking... are you re-doing the same EOM project or did you get a new one?

New Post
Unfortunately, it's a new one. I would like to join a google group for that if one exists or emerges.
Good luck everyone!

New Post
Hello all, just joined the group. Thank you for bringing me in, Sarah.

I am taking a similar approach to this note below. There are numerous issues with the data/process that have largely been summarized already throughout this thread. After those fixes I am getting a result in the 8-10% range mentioned in the intro of the EOM (closer to 10% than 8%) which I feel to be reasonable given the increase in Net PMPM claims in Y2, despite large #'s of enrollment pouring into plan E in Y2.

As far as the Somestate vs Otherstate conversation, I am comfortable noting that for purposes of this exercise that the relative # of individuals enrolled in plans in Otherstate is quite low compared to Somestate (and they also have lower demo factors). While we would much prefer to omit Otherstate enrollees in the Net PMPMs, I think it is not the end of the world if Otherstate experience is included for purposes of trend calculation.

New Post
Hello all, just joined the group. Thank you for bringing me in, Sarah.

I am taking a similar approach to this note below. There are numerous issues with the data/process that have largely been summarized already throughout this thread. After those fixes I am getting a result in the 8-10% range mentioned in the intro of the EOM (closer to 10% than 8%) which I feel to be reasonable given the increase in Net PMPM claims in Y2, despite large #'s of enrollment pouring into plan E in Y2.

As far as the Somestate vs Otherstate conversation, I am comfortable noting that for purposes of this exercise that the relative # of individuals enrolled in plans in Otherstate is quite low compared to Somestate (and they also have lower demo factors). While we would much prefer to omit Otherstate enrollees in the Net PMPMs, I think it is not the end of the world if Otherstate experience is included for purposes of trend calculation.

New Post
I noticed that ABIS membership is based on contract state whereas the data from the DW is pulled based on zip state. So the DW membership has some members from Otherstate based on contract state. I am stating that these populations need to be aligned but noting that I am limited in my ability to correct this because I don't have access to the source data. Also, the effect should not be material. I agree. Do you think this is adequate or is more thorough work needed to be done?
#2
01-04-2019, 03:42 PM
 Sir Issac Member SOA Join Date: Jun 2014 Posts: 406

"- I also noticed that the input to the benefit index model does not match between the product area and the ABIS input data."

Can someone comment on what those differences are exactly? I cannot spot any.
#3
01-08-2019, 04:17 PM
 Sir Issac Member SOA Join Date: Jun 2014 Posts: 406

nvm I found them. For anyone curious in the future, double check the destructible for plan E and coinsurance for additional services for plans E and F