Data issue: Module 4 Exercise

[DAG FIN]

Anyone else notice that the face amounts for the sales data are listed in terms of 50K, 100K, 250K (band 1,2,3 resp), but the lapse data are not in multiples of these? For example, band 1 = 50-99K face values. Year 1 shows 20K policies sold, for face amount totaling 1,000,000K. 50K being the lower bound implies all policies were sold at exactly 50K. However, the face value of lapses is 47,480K - NOT a multiple of 50K. Similar analysis applies to all bands. Are we to assume that the sales data is estimated or that the lapse data may be incorrect? I will probably just waive my hands and point it out as a possible data issue and move on. Just wanted to see if anyone out there more familiar with life had any guidance that could help explain this.

[glassjaws]

Is there a surrender charge?

[Hachover]

Originally Posted by glassjaws

Is there a surrender charge?

Let's see. It's a term product, so there are no cash values. A lapse means they don't pay premiums. Even if there were a surrender charge, I doubt anyone would come forward to pay it.

[Hachover]

Originally Posted by DAG FIN

Anyone else notice that the face amounts for the sales data are listed in terms of 50K, 100K, 250K (band 1,2,3 resp), but the lapse data are not in multiples of these? For example, band 1 = 50-99K face values. Year 1 shows 20K policies sold, for face amount totaling 1,000,000K. 50K being the lower bound implies all policies were sold at exactly 50K. However, the face value of lapses is 47,480K - NOT a multiple of 50K. Similar analysis applies to all bands. Are we to assume that the sales data is estimated or that the lapse data may be incorrect? I will probably just waive my hands and point it out as a possible data issue and move on. Just wanted to see if anyone out there more familiar with life had any guidance that could help explain this.

I'm not sure what you mean (because it's been a long time) but deaths would bring down the average. Deaths are not a surrender.

Anyway, you should move on. The exercise is pretty straightforward.

[DAG FIN]

Not sure what you mean here about deaths bringing down the average. The issue is that all policies are issued in face amounts of 50K, but lapses are not stated in amounts of 50K. I guess I'm wondering how can a policy have a partial lapse? (Same is true with the deaths data). In any case, I'm not too concerned with it, but it is a potential issue with the data. A major point in the readings is that good analysis with bad data is useless. Given that, I thought it was worth pointing out to anyone else working on this exercise.

[weinerda]

Hmm... What about lapse off anniversary? This would reduce the exposure by a partial face amount, no?

[Dahlia]

Originally Posted by weinerda

Hmm... What about lapse off anniversary? This would reduce the exposure by a partial face amount, no?

The assignment doesn't give policy issue dates and it states that all the lapses are at year end. Anwyay I just though this was a "special" mod4-exercise-term-product which has odd lapse amounts. I don't think it's one of the problems that the exercise wants us to focus on fixing, but I could be wrong...

[DAG FIN]

Originally Posted by weinerda

Hmm... What about lapse off anniversary? This would reduce the exposure by a partial face amount, no?

I thought about off-cycle terms, but I don't think this should impact anything. Exposure should be an off/on - ie the 50K is in the exposure base until the moment it terms (or 10 years pass). Once it terms, regardless of when, the exposure base should be reduced by 50K. I would assume that to measure the annual lapse rate, you'd want to consider the entire amount lapsing, regardless of when. If we were measuring against average exposure in the year, than I think you're right. Then lapses should also be weighted to reflect timing. However, since we're measuring against initial sales volume, I don't think any weighting is necessary.

[DAG FIN]

Originally Posted by Dahlia

The assignment doesn't give policy issue dates and it states that all the lapses are at year end. Anwyay I just though this was a "special" mod4-exercise-term-product which has odd lapse amounts. I don't think it's one of the problems that the exercise wants us to focus on fixing, but I could be wrong...

It seems the general consensus is that it's not a big deal. I just found it odd that on the module with so much emphasis on checking input data, there appears to be a problem with the input data. If it's not a focal point, maybe it should be?? They basically hold your hand throughout the rest of the exercise, so I assumed they're trying to lull us into a false sense of security! NEVER take candy from anyone that writes exams for the SOA!

[glassjaws]

Ha. I should think more before I speak. My bad.