Sorry, no resources...

While reading "The Savings and Loan Crisis and Its Relationship to Banking" (about the S&L debacle incurred in US in 1980s), a paragraph really caught my attention:

"...Accordingly, when much of the S&L industry faced insolvency in the early 1980s, the FHLBB’s examination force was understaffed, poorly trained for the new environment, and limited in its responsibilities and resources.7 Qualified examiners had been hard to hire and hard to retain (a governmentwide hiring freeze in 1980–81 had compounded these problems)..." (Page 170)

Does this sounds familiar to you? If you never heard of anything like "limited resources" from your management or your boss, you must be happily working in your current company and working on long OT is once in a blue moon.

Many bosses think that their staffs are "supermen" - they introduce new workloads to the existing staffs, decline to provide sufficient resources and still expect for marvelous results. Is it possible? Well, I think most of the time this overoptimism will end up with disappointment - either the work is not done well or more problems are created. Such disappointment can be avoided if they face the music at the first place, namely the staffs are just ordinary humans and NOT "supermen".

I think sometimes many bosses have abused the usage of "challenge". For example, when a first line manager asks for additional headcount, the boss will tell him: "Currently there is no budget for extra headcounts. By the way, you should take this as a 'challenge'." How should the first line manager response? Can he say "NO, I can't take the challenge"? An assignment will be perceived as a "challenge" if it has certain level of difficulties and the staff will have satisfaction upon accomplishment. If a there is only 4 persons to do 8 persons' work, how can the team do good work? Even though they can complete the work, it is luxurious to expect for good work quality - the work is completed merely for the sake of finishing the work. Since the team is unable to produce good work, how can they get any satisfaction? A challenge which never bring satisfaction is not a challenge.

Sufficient resources should be viewed from 2 perspectives - the "quantity" as well as "quality". Yes, a  fresh graduate is "cheaper" if we measure his remuneration in term of monetary term. However, we often overlook (or ignore) that this fresh graduate requires "indirect investments" for the lack of experience and knowledge. The first line manager or senior executives need to allocate time and resources to train new staffs. As a result, the remaining time and resources available for other work are lesser. The productivity of 1 senior + 7 fresh graduates team is definitely lower than 4 senior + 4 fresh graduates team. Don't say "I already give him extra headcounts" if you don't give him the types of headcounts he needs.

We have to accept the fact that we are ordinary humans. If we do not have resources, don't start a new project.

Nuclear Power Station Accident, Tsunami... No longer Improbable

When I went through a few chapters of "Catastrophe, Injury, Insurance: The Impact of Catastrophes on Workers Compensation, Life, and Health Insurance" published by Risk Management Solutions (RMS) in 2006, two paragraph about nuclear power stations and tsunami caught my attention - the first paragraph commented the chances of a major radioactivity release from the nuclear power station:

Scenarios that fall into these plausible mass casualty events but are improbable include a large-scale accidental release of radioactivity from a nuclear power station. Estimates suggest that such an event could affect several million people and make many hundreds of thousands sick. Operating safety standards for nuclear power stations require that the chances of a major radioactivity release are maintained at below 1 in 10,000.

The second paragraph commented large-scale tidal waves:

A tsunami or coastal tidal wave affecting a major stretch of coastline on either the Atlantic seaboard or the Pacific coast could be destructive and deadly for populations living along the coast. Large-scale tidal waves could potentially be caused by an ocean floor earthquake, a submarine landslide, or a meteorite in mid-ocean, but estimates of the return periods of these types of events are measured in many thousands of years.

Sometimes we take it easy for the events with low possibilities, by presuming like it will not happen - low possibilities do not mean that the event will not happen (the tail effect)... Japan just suffered massive earthquake and tsunami at the beginning of 2011.

I really hope the Lynas' Advanced Materials Plant project in Kuantan, Malaysia  will be called off - no matter how well the operating safaty standards are followed, there is still non-zero possibility for accident of radioactivity release - it will be already TOO late when the accident happens, and the price to be paid will be extremely large...

One Asia Rare Earth tragedy in Bukit Merah already taught Malaysia enough lesson.

Government Servants' Retirement Age Extended to 60

The Prime Minister’s Department presented the following government bills for first reading at Dewan Rakyat of Parliament Malaysia on 8 November 2011, which related to the retirement benefits (including pension) for the government servants:

• Pensions (Amendment) Bill 2011
• Statutory and Local Authorities Pensions (Amendment) Bill 2011
• Pensions Adjustment (Amendment) Bill 2011
• Judges’ Remuneration (Amendment) (No.2) Bill 2011

As revealed in the recent Budget 2012, the retirement age of the government servants will be extended to 60 years old starting next year (2012). Currently, the retirement age is at 58 years old - although they can opt to retire earlier at age 55 or 56. Furthermore, in the event of a retiree passes away during 20 years (previously 12.5 years) after his retirement, his family members will continue to receive the full amount of his pension and other relevant retirement benefits.

According to Budget 2012, the government expenditure on pensions and gratuities comprise 5.2% of the total development expenditure (of RM230,833 mil). I am afraid that this figure will keep increasing in the future years, given that it is already a well-known fact that Malaysia has a lot of government servants (recently there were some political disputes on reducing government servants). If the the retirement benefits are using pay-as-you-go as funding method, these expenditures may appear as a treat to the country financial stability in the future - to make matters worse, the pension amount will need to adjust from time to time to reflect the increasing cost of living (the annual pension increases 2% as per Budget 2012).

Although extending the retirement age may delay the commencement of retirement benefits, but this effect will be partly offset by the larger pension amount - which is directly correlated to the final salary (final salary at age 60 is definitely larger than the salary at age 58). I would think that it is high time for the government to evaluate the management of government servants' retirement benefits and find another better approach to fund and pay the benefits (such as defined contribution plan... Well, I know it is a bit difficult...) - I don't want to see in the future that a large portion of the income tax I pay is used to pay retirement benefits. I couldn't imagine what the liabilities will be if the government asks the actuaries to do valuation on the retirement benefits.

Don't use luxurious retirement benefits for political reason to get votes, we will suffer in the future. It is easy to offer better retirement benefits, and it will be extremely difficult to cut the benefits - no one will like to receive lesser retirement benefits, although he may know that the country is facing financial problems. Just take a quick look on what has happened in Greece - people who used to have comfortable living will definitely object for anything that will make their living less comfortable (even though just "slightly" uncomfortable).

Don't wait until it is too late to correct - and the cure will be very painful.

Related news:

• Family members will receive full amount of retirement benefits if a retired government servant passes away during 20 years after his retirement (9 November 2011; Chinese Language; Sin Chew Jit Poh)
• 5%-6% expansion expected in 2012 (8 October; English Language; The Star)

GST in Malaysia... Around the Corner?

The Malaysian Deputy Finance Minister Dato' Donald Lim Siang Chai has recently told the press that the government is expected to gradually implement Goods and Services Tax (GST) by end of 2012, and followed by the full implementation in 2013.

Hopefully Dato' Donald's comment is the real indication of the implementation date of GST in Malaysia. The government has been discussing GST for quite some years - the Royal Malaysian Customs even set up a GST dedicated website to educate public on GST. Again, the 2012 Budget did not mention on the GST implementation - perhaps this may not be a good time for the Barisan Nasional government as the general election is around the corner.

Different from the current single stage sales tax and service tax, GST is a multi-staged consumption tax that covers most supplies and goods. Since the GST implementation is once delayed, those corporates which are not ready yet should take this opportunity to revamp their systems (including their accounting systems); otherwise, the last minute job is always painful nightmare. Hopefully, by implementing GST, the personal tax and corporate tax is expected to reduce significantly, in order to increase Malaysia business competitiveness in this region - and increase the tax revenue! (the country's budget has been running in deficit for so many years...). Of course, I believe some living essential goods (such as rice, sugar, oil etc.) should be exempted from GST, otherwise will increase rakyat's burden (especially the lower-income families).

Similarly, the takaful operators in Malaysia should also enhance their policy admin systems to cater for withholding tax implementation, which affects the surplus sharing / mudharabah payments and profit declaration for Participant Account (PA) / traditional Participant Investment Fund (PIF).

Related news:

• GST may be implemented by end of next year (29 October 2011; Chinese Language; Sin Chew Jit Poh)
• When should Malaysia implement GST? (15 October 2009; English; Business Times)
• Embracing GST to Boost Revenue (19 september 2011; English; The Edge)

Everyone Assumes Health Plans are the Bad Guy

When I read "Everyone wants to assume that the health plans are the bad guy", I couldn't help keep nodding my head.

This interesting statement was written by David Axene (President & Consulting Actuary, Axene Health Partners) in his article "Is California leading states in health care reform?", posted in SoA Blog. Many people have an perception that insurance companies are the greedy devils which always look for opportunities to get excessive profits by increasing the health insurance's premium rates - but do they really complain on the increasing medical costs charged by the medical providers (like private hospitals)? They do not understand the premium increase is mainly attributed to the increase in medical costs, or they refuse to understand?

"Public" here refers to those public who have bad perception on insurance companies
selling health insurance

Perhaps this disappointing public behaviour can be explained by the above chart - public needs to pay premiums to insurance companies, but they receive treatments from the hospitals. We can easily understand with common sense that it is human nature to consider the party which they need to pay money to as "bad guy", especially when there is premium increase. Even though the insurance companies explain that the increase in medical cost is the main culprit, they fail to (or refuse to) comprehend the key problem as the medical bills are not paid out from their pocket directly. To make the matter worse, sometimes the government also think the insurance companies are the "bad guy" and impose a lot of restrictions and controls - while they do not take sufficient measures to control the root cause.

On the other, the medical providers are the one help the public by providing medical treatments and they do not pay or only pay small amount of deductibles / co-insurance. Since the medical bills are settled by the insurance companies and not from their own pocket, they don't concern what the amounts are. Since the Malaysian Government is currently working on healthcare reform, they must invite expertises from insurance industry to join the working committee - the public behaviour will change dramatically according to how the healthcare policy is established.

It is human's nature for not appreciating insurance companies - My friend, Jane, has a relative who just received treatments in a hospital, was complaining that her insurance company did not fully reimburse the pre-hospitalization medical bills (i.e. the prescription drugs costed a few hundred ringgit). Even though the insurance company has paid a few thousand hospital bills. Of course, no gratitude to the insurance company.

Spreadsheet Tips (1): More Flexible MOP Template

In case you are looking for some Excel / Spreadsheet tips on how to make your templates more efficient, perhaps you can get some ideas from the tips I share below. Afraid of bombastic command or VB coding? Don't worry, the tips I share below are very simple and you don't need to be an expert to use the tips.

The following tips use examples from the template I used to compile movement of policy (MOP) statistics (i.e. new business, in force business & termination/alteration) . However, you may apply the same ideas for other types of templates as well.

Tips (1): Produce results by plan code level, not product group

Normally, the MOP statistics are reported in specific product groups, such as "Whole Life", "Endowment", "Temporary" etc. in BNM Form L6-L8 (Conventional Insurance) / FT5-FT7 (Takaful). It is common for us to group our data by the required product groups using data processing applications (e.g. FoxPro, Data Conversion System (DCS)) before we transfer the results to the spreadsheet template.

Well, try to think a few conditions you may come across - Your superior finds out there is a mistake in classifying few product codes into incorrect product codes, and he needs you to update the last year production and current year production based on the correct product group. OMG, you need to update the FoxPro / DCS tables and rerun the programs to product revised new business summaries - to make matter worse, the last year new business data may be already gone due to hard disk crash (because you forgot to backup the data)...

Figure 1: Data Sheet

You will not face such problem if you produce the new business results by product code / plan code. Instead of doing grouping at FoxPro / DCS level,  you only do the product grouping at spreadsheet template level. As shown in Figure 1, I use column G (labeled as "Index") to do product grouping. For example, for "Ind_Trad_EndEdu_Reg" :

• Refers to "Individual Products + Traditional Business + Endowment + For Education + Regular Premium".
    
• I use SUMIF to sum up the no. of policies sum assured and annualized premiums for the product group, i.e. "=SUMIF('data_NewBiz (Ind)'!$G:$G,"Ind_Trad_EndEdu_Reg",'data_NewBiz (Ind)'!$C:$C)".

Figure 2: Reference Table

It is not difficult to set up the product group "code". In Figure 2, you can easily find out that the code is basically just the combination of several fields, and teh underscore ("_") is used separate the fields (you can omit "_" if you want your "code" to be shorter. You can use "&" to combine the values from different fields, instead of CONCATENATE, such as "=IF(ISBLANK(B6),"",B6&"_"&C6&"_"&D6&"_"&E6)" used in my template.

Now you can produce the revised report easily - change the product group in the current year and last year working file, and your revised report is done!

Tips (2): Too many worksheets in your spreadsheets? How to display the selected group of worksheets?

It is common for actuarial people to have spreadsheet template with a lot of worksheets. For example, the MOP template for Figure 1 & 2 contains 25 worksheets. Sometimes it may be a bit troublesome to go to the specific worksheets I would like to check.

Figure 3: Buttons used to hide unwanted worksheets

To overcome this problem, I use VBA to hide the unwanted worksheets. As shown in Figure 3, I used 4 buttons to hide the worksheets that I do not want to see. As the 9 data sheets are only referred when needed, I create an additional control to display/hide the data sheets. When I click "Display Monthly Sheets" button (without data sheets), only 6 worksheets are displayed and it makes my checking easier! 

Below is an example of the VBA coding I used in my template:

Sub DisplayMthlySheets()

    Application.ScreenUpdating = False

    

    For i = 1 To Sheets.Count

        If Sheets(i).Name = "input" Or _

           Sheets(i).Name = "FT5_mth" Or _

           Sheets(i).Name = "FT6_mth (ProdClass)" Or _

           Sheets(i).Name = "FT6_mth (Cause)" Or _

           Sheets(i).Name = "FT7_Close" Or _

           Sheets(i).Name = "Recon (Mth)" Then

            Sheets(i).Visible = True

        Else

            Sheets(i).Visible = False

        End If

   Next i

    

    Call DisplayDataSheets

    

    Application.ScreenUpdating = True    

End Sub

FoxPro for Actuarial (2)

I started to learn FoxPro when I started my first actuarial career after graduating from the university. My mentor in the actuarial department, Madam L, was also my FoxPro shifu (i.e. master / teacher / guru). Madam L was the reinsurance administrator in the department - by using FoxPro, she managed to do necessary reinsurance administration work for the entire individual life insurance business (more than 1 million policies) by her own, alone! By going through her programs and guidances, I picked up necessary skills to use FoxPro and create FoxPro programs for my reporting and experience studies tasks (my manager resigned three month after I joined the department - all his tasks became my tasks!) - until now I am still grateful to the guidance she provided to me previously.

When I was promoted to a team leader with junior staffs, similarly I needed to teach them the relevant FoxPro knowledge so that they had necessary skills to do the assigned tasks. At that time, there was no proper training set up to teach the juniors FoxPro skills - the juniors have to go through the existing programs and learn the FoxPro skills by their own, by going through a lot of try-and-error. Consequently, some FoxPro programs were not modified properly and eventually were discarded as they were already too messy; on the other hand, some FoxPro programs written by my "genius" colleagues cannot be understood by most people in the department. Hence, I decided to set up a proper "syllabus" for FoxPro to train my juniors.

I split my FoxPro training into 5 modules, and each module has its own training materials: (with quiz & exams!)

• Module 1: Introduction & Query
• Module 2: Query (2) - create & alter tables, multiple-table matching
• Module 3: Program (1)
• Module 4: Program (2) - looping, arrays, text merge
• Module 5: Program (3) - procedure & functions, advanced topics

I frequently reminded my staffs that "actuarial are NOT IT programmer" - we only need to learn the relevant skills that are required to carry out our actuarial tasks. In other words, we do not need to learn the entire FoxPro skill sets like IT programmers - such as the visual features of Visual FoxPro (as this is not really necessary to make the process efficient). Furthermore, the standard "syllabus" also helped me to ensure at least the same level of skill sets for the entire team - the FoxPro programs and coding would be set up in a similar manner, so that other users and future developers could understand the programs quickly and modifications could be done easily.

Does it mean that we cannot use commands not stated in the "syllabus"? Not really. For those commands which were considered "nice to have" and did not really improve the efficiency of the process, I normally advised them not to use it (I said: "...because you will not the person who uses the programs forever"). However, if any team member found out good commands that can help to improve the efficiency, I would amend the "syllabus" accordingly so that everyone could (well, actually "should"...) learn the new commands.

In fact, apart from FoxPro, my staffs and I also set up standard "syllabus" for Prophet training as well. Setting up such proper training modules (especially for the junior staffs) are crucial for the department's knowledge management. If the "knowledge" is managed well in the department, the department will be able to develop healthily - there is continuity in the skills developed and lessons learned in the past. Otherwise, the past mistakes will recur in the future and our successors will only learn what we have learned previously, instead of making progress on what we have developed and learned previously.

(In case you are looking for FoxPro training program for actuarial people, you may contact me at limcheebeng@hotmail.com to discuss further.)

FoxPro for Actuarial (1)

Recently when I helped my client to set up process to compile movement of certificate statistics (MOP) (i.e. new business, in force and termination/alteration), I advised them that Excel is not an ideal tool to do such actuarial exercise which requires a lot of data manipulation. "Then what is the better tool?" They asked me. Without any hesitation, I answered: "FoxPro".

Why do I recommend FoxPro as a good tool to be used for actuarial exercises that require a lot of data manipulation & matching (such as MOP reporting, experience studies and reinsurance administration)? What are Excel, MS Access or Data Conversion System (DCS) (a supplementary application comes together with Prophet) not considered good tools for actuarial exercise? Among the reasons are:

• Ability to do data matching / joining - as FoxPro uses SQL (Structured Query Language) as the primary programming language, we can combine the selected fields from 2 different tables/data easily based on multiple matching criteria, by using "SELECT" and "JOIN" command (the ones I used the most are "LEFT OUTER JOIN" and "FULL JOIN").
  
  Doing such data matching / joining in Excel or DCS? You will definitely think that I must be kidding - it is pretty tedious to use "LOOKUP" function to do matching, especially if we need to refer to multiple fields / part of a field / combination of fields as matching criteria. Furthermore, our Excel / DCS will extremely "tired" to look up values if we have a huge set of data...
    
• Highly automated process - We can set up highly automated process using coding the required steps in a "program" - from converting the raw data to FoxPro tables, perform necessary calculations and data manipulations to produce summaries for the intended results (such as in text file or Excel formats).
  
  Although Excel is able to summarize data like FoxPro by using Pivot Table function, it will be a bit complicated to set up coding for automated Pivot Table using VB in Excel (I just helped my client to set up one - I admit the relevant VB coding makes me a bit headache). FoxPro just needs a few lines of commands to do the required summaries.
    
• Ability to handle huge volume of data - I still remember that when I used to use FoxPro to do matching for 2 tables with multiple millions records each. Sometimes I did felt sorry to my computer for "torture" it with sure heavy tasks... :p
    
• Easy to debug - Although MS Access uses SQL as well, you may find out it may not be easy to debug. I still remember I used to help my staff to debug the MS Access template - which he totally had no idea where went wrong. The error message didn't give adequate information!
    
• Inexpensive - Comparing to other actuarial applications (such as Prophet), FoxPro is considered cheap (just a few thousands riggit) - especially for a newly set up company with limited budget. By having FoxPro, you can even set up models to calculate reserves.

I will discuss my experience in learning FoxPro and the FoxPro training I provided to my staffs (with proper syllabus and training materials. Got exams too!)

Medical Insurance for All Malaysian?

During a seminar on health care reform held recently in Kuala Lumpur, Malaysian Health Minister Liow Tiong Lai mentioned that the government would refer to the healthcare model in Taiwan (one type of social insurance, i.e. National Health Insurance (NHI)) as one of the bases to reform Malaysian health care system. The government will develop the National Health Financing Authority (NHFA) to manage the fund under the National 1Care Health Scheme (1Care) - which is expected to be funded by government, private sector, employers and the people.

Theoretically, it is a good idea to have a structured healthcare system that can provide healthcare services to all citizens in the country. In my view, expensive medical cost is one of the key financial risks that everyone should really concern on - almost everyday we can read news about unfortunate folks with serious illnesses begging the public to fund for the expensive medical costs. According to Liow, starting from 2004, the medical expenditure for private hospitals exceeds government hospital. Frankly, the main factor attributes to this trend is the government hospitals are unable to deliver the healthcare services promptly and efficiently - I believe some of you may experience the frustration in government hospital, such as spending hours of waiting or getting doctor appointments scheduled after many months. However, we have to admit that government hospitals still have the most complete range of equipments.

However, practically, it is extremely difficult to implement social healthcare insurance model as adopted in Taiwan. There are too many factors that directly affect the implementation of social healthcare insurance - you can find out many weakness in Taiwanese model from Wikipedia article I refer below (sorry I only managed to find out the Chinese Language version). Let me quickly quote a few examples:

• Difficult to increase insurance premium, despite the high medical inflation (have political consideration)
• High volume of abuse by "insured" (due to public's attitude)
• Overstatement of medical costs by hospitals

The responsible government agency needs to have a very stringent controls on all aspects of the healthcare, if they want to make this healthcare reform successful. It will be really a massive work to be done, furthermore on regular basis. Many expertises are needed to close monitor the "claim experience", from incidence rate, average claim paid, average length of stay up to loss ratio - many insurance companies have done a lot of controls on their medical claims but unfortunately quite a number of them the medical insurance is not a really profit making business. The agency should be granted sufficient authority to take actions on the hospitals if necessary (without political influence). Well, you may want to ask the the problem: "Where can we get the relevant expertises?"

We still have long way to go in healthcare reform. By the way, just wondering if the healthcare reform will also cover the delivery efficiency as well?


Reference:

• "Liow: Public Healthcare to Refer Taiwanese Model; Poor for Free, Rich Pays More" (Sin Chew Jit Poh, 21 August 2011, Chinese Language)
• "Mechanism to be Devised under NHFA to Manage 1Care Fund" (Bernama, 15 March 2011, English Language)
• "Nation Health Insurance" (Wikipedia, Chinese Language)

Is "Claim Contingency Reserve" Necessary for Takaful Business?

When I worked with one of my previous client, I came across "Claim Contingency Reserve" (CCR) in their operation model. According to "Takaful and Retakaful - Principles and Practices" written by Tobias Frenz, CCR (also known as Special Reserves or Equalisation Reserves in some markets) acts as a buffer against an adverse or volatile claims experience.

My client and I have a few discussions on CCR, especially on whether it is necessary to establish CCR in the participant risk funds. Normally, CCR is funded by surplus arising during the year and setup as a % of tabarru' or technical provision (This reminded me of the old days when I calculated solvency margin using 4% of reserves, ...). It was a a challenge to determine an appropriate % to be used and the approach to releasing CCR over time. Well, there is no prescribed % or approach available. 15% or 10%, which is better? Should we release over 3 years or 5 years? Should we release linearly or staggered (like commission payment pattern)? I'm afraid no one can really give you a good answer...

In my view, CCR has the same purpose as the solvency margin or capital (such as Risk-based Capital (RBC) Framework practised by the conventional insurers in Malaysia). Previously, there was no regulatory solvency framework (I was quite surprise when I found this during the employment with my first takaful employer), which required the takaful operators to setup solvency margin to protect the participants against adverse deviation. Hence, it might make sense to establish buffer and call it as CCR. (Personally I don't prefer to call this buffer as "reserve" - for me, this is not a form of "reserve" like the technical provisions we setup for the takaful liabilities).

Recently, Bank Negara Malaysia (BNM) has issued concept paper for RBC Framework for Takaful Operators - I was quite excited as I have been expecting this for some years! Once takaful operators adopt RBC Framework, I would think it is no longer necessary to have CCR in their operation model (otherwise you will have double provision).

Time to say goodbye to CCR! (Remember to revisit your surplus management policy, OK?)

Movement of Policy (5): Termination / Alteration

Note: Please note that the following discussion is only for movement of policy (MOP) of individual life/family takaful products only.

Among the three statistical reporting forms (L6, L7 & L8 for life insurers; FT5, FT6, FT7 for takaful operators), the termination / alteration statistics are considered as the most complicated. As all three forms need to be reconciled (please refer to the above diagram), many people have no choice to put the differences that cannot be reconciled as "Others" (another popular word to be used is "balancing item").

In fact, it is not complicated if you follow the correct approach to do these termination / alteration statistics. If you re-arrange the formulas in the above diagram, you can easily see that actually these statistics intend to explain the changes/movements occur previous month in force policies and new business incepted in current month, by comparing to the current month in force policies.

So, what is the best way to do termination / alteration statistics? I guess you already get the correct answer by now, i.e. comparing current month in force data (DATA 1) to previous month in force data plus current month new business data (DATA 2 + DATA 3) and the differences between DATA 1 and DAT 2 & DATA 3 are the statistics you need:

• Termination: The base plan / riders available in DATA 2 & DATA 3, but NOT available in DATA 1.
• Reinstatement / Addition of new riders: The base plan / riders NOT available in DATA 2 & DATA 3, but available in DATA 1.
• Alteration: The base plan / riders available in both DATA 1 and DATA 2 & DATA 3, but the sum assured, premiums or product code are different in DATA 1 and DATA 2 & DATA 3.

If we would like to split reinstatement and addition of new riders, we may need to have an additional step to check whether the attaching base plan are available in DATA 2 - if yes, we can consider those riders are newly added riders to an existing base plan.

In order to do the above-mentioned comparison, we need to first understand the base plan / rider's unique identification used in our policy admin system - i.e. unique field or combination of fields that differentiate one base plan / rider from others:

• Base Plan: Normally, policy number is the main identification (like "primary key" in MS Access table) that we can use to differentiate one base plan from others. If we have yearly renewable products, we may need to find out whether the system will use the existing policy number upon renewal (Hopefully your system will use the existing policy number, otherwise you will run out of policy number very soon...).
   
• Rider: We may need to use combination of multiple fields to form the unique identification. In one of the system I use previously, each rider has a unique "sequence code" - all riders under attached to a base plan have different sequence code. I used combination of policy number and rider code (e.g. "1234567A" (policy number) + "01" (rider code) for the 1st rider under base plan with policy number 1234567A)

How frequent we should prepare the statistics?
Similar to the in force statistics, termination / alteration statistics are rarely used for production performance or decision making. Hence, if our process is properly automated and it only take little resources & time to prepare the statistics, it will be nice to have the termination / alteration statistics on monthly basis. You may not need to submit the report to the Management if they don't really want to look at the statistics (they already have plenty of reports to go through).

So what's the point we spend so much efforts preparing the termination / alteration statistics? Well, let's think how we can make use of the statistics (the reports will remain as papers if we do not use them). Actually we can use these statistics as "early warning" system in order to trace any unusual trends promptly - we can use these statistics to do some quick calculation to get rough estimates on surrender / lapse rates or claim rates.

It will only be useful to the Management if we derive meaningful conclusions from the statistics.

How should the termination / alteration statistics to be reported?
Internally, I would recommend to list out all types of termination / alteration - please do NOT have a category called "Others" (Only for external reporting like L7 or FT6). Different from the new business statistics and in force statistics, we may not need to report the statistics by product (Otherwise our tables will be very big and difficult to study).

However, it will be useful if we can breakdown by base plan / riders and major product class (as investment-linked products may have different experience from the traditional products). The statistics may not be meaningful if we combine all products together (which we normally call "rojak").

What are the source data?
Apart from the new business data and in force data mentioned above, we need another set of data from our Data Warehouse / MIS for the policy statuses of the policies terminated during current month - so that we are able to identify why a policy is terminated (such as death / TPD / Critical illness, surrender, lapse, maturity / expiry, … - please refer to "Back to Basic - Movement of Policy (MOP)" for detailed discussion).

What is the tool?
Similarly, the first choice I would recommend is FoxPro. So far I have no idea how to use Data Conversion System (DCS) to easily compare 2 different sets of data (i.e. "LEFT OUTER JOIN" in FoxPro).

Movement of Policy (4): In Force

Note: Please note that the following discussion is only for movement of policy (MOP) of individual life/family takaful products only.

Compared to new business statistics, in force statistics are much simpler - basically, what we need to report are the policies as at end of particular period (month-end / quarter-end / year-end) which are considered as "in force" (i.e. the coverage is still valid).

Depending on the policy admin system we use, we can identify the in force policies by referring to their policy statuses. In one of the system I used previously, the "in force" business are those policies having the following policy statuses:

• In Force: Normal premium paying policies.
• Paid up: Applicable to the limited payment products only; policies which all required premiums have been paid.
• Reduced Paid Up (RPU): The policyholders decide not to continue paying the remaining premiums (or by other scenarios defined in the contracts) and covert the policies to "paid up" policies based on the available cash values - the sum assumed is reduced but the policy term remains the same.  If a RPU policy is previously a participating (Par) policy, it will cease from participating (i.e. change to non-participating (Non Par) policies)
• Extended Term Insurance (ETI): The difference between RPU and ETI is ETI remains the original sum assured, but the policy term is shorten.

Some systems may use different policy statuses for policies which the future premiums have been waived. However, in my view, instead of using policy status, it will be better to use "premium payment method" as indication - if you study the waiver of premiums carefully, you will find out the waiver policies are actually still "premium paying policies", just the premiums are paid from the provision account. Hence, it should be considered as changing "payment method".

Similarly, if our system uses different policy status for policies in grace period, we should also consider those policies as "in force". However, in my view, having different policy status for policies in grace period (or like "pre-lapse" status you heard previously) will unnecessary increase system workload - the system needs to change the policy status once the grace period starts and revert to the original status once the required payment is received (Please also remember that system will need to record the changes in the transaction history for audit trail).

How frequent we should prepare the statistics?
In force statistics are not as "popular" as new business statistics, as they are rarely used for production performance or decision making (the Management is more interested to see the Embedded Value (EV) rather than merely statistics) - furthermore, for an established company, the composition of in force business in the book do not vary significantly from one month to another.

If our process is properly automated and it only take little resources & time to prepare the statistics, it will be nice to have the in force statistics on monthly basis - especially if our valuation colleagues need to do valuation monthly (to provide additional comfort on the data quality, i.e. we can use the in force statistics to detect the usual changes in the composition of in force business or no. of policies / sum assured (total & average) / premiums(total & average)).

Of course, the regulatory reporting requirements will influence how frequent we need to prepare the statistics.

How should the in force statistics to be reported?
I would recommend to use the same formats as new business statistics, for easy comparison. Please refer to the discussions in "Movement of Certificate (3): New Business" - there are similar considerations for new business statistics and in force statistics.

What are the source data?
Our Data Warehouse / MIS should provide us data as at month-end, or account closing date (if the account closing date < month-end). The SAME data should be used for both reporting & valuation purposes - otherwise we will be wasting our time in doing reconciliation.

We should have in force data on monthly basis. The in force data are the most widely used data in Actuarial Department - apart from actuarial reporting & valuation, other functions

What is the tool?
Similarly, the first choice I would recommend is FoxPro. In my one of my previous assignment, my client has discarded Data Conversion System (DCS) and use FoxPro to prepare the model point files for Prophet runs. Frankly, it will be more efficient to use only one tool to carry task from manipulating the source data to producing model point files.

Please refer to the discussions in "Movement of Certificate (3): New Business" for details.

Movement of Policy (3): New Business

Note: Please note that the following discussion is only for movement of policy (MOP) of individual life/family takaful products only.

When we create a new entry in a policy admin system, normally we will create a "proposal" - so that the underwriters can perform necessary assessments and decide the proposed life can be accepted or request for additional information (such as medical checkup). After all the underwriting requirements have been fulfilled (and of course we have received "adequate" premiums), the proposal will be converted to an "in force" policy - and our agents will be happily informed their clients: "Congratulation! You are now covered by Company ABC!"

In simple words, our "new business" statistics are used to report the proposals that have been converted to in force policies in a particular period (month/quarter/year).

How frequent we should prepare the statistics?
I would recommend to prepare new business statistics on monthly basis. The statistics should consist of 2 sections, i.e. "Monthly" and "Year-to-date (YTD)".  For better comparison, we need to ensure that both sections should have exactly the same format (I really cannot think of any good reason why we should have different formats).

Normally, the Management & Sales Department are more concern on YTD statistics.

How should the new business statistics to be reported?
Normally, the new business statistics required by internal (e.g. Management & Sales Department) and external (e.g. regulator) are different:

• Internal: Require the new business statistics to be reported by product / product groups.
• External: Require the new business statistics to be reported by product class (such as "Endowment", "Term", …). We need to take note of additional reporting requirements (e.g. in L6 / FT5, we do not need to report the sum covered for personal accident riders).

The "Internal" report should serve as the "base" report which contain the detailed new business statistics; the "external" statistics should be summarized from the "internal" report and adjusted based on the reporting requirements. The "external" report may need to combine several "internal" reports (if you have separate reports for different lines of business).

"I spend so much time preparing the statistics, but they don't even read it!"
Frustrated of you readers who don't read your reports? Sometimes it's our fault for the target readers not reading the reports we produce.

Normally, the non-actuarial readers will face difficulties to digest a report containing multiple large tables with many rows & columns - our report are "useless" if they cannot be read & digested by the target readers, especially those who are involved in making decisions. Hence, we should prepare a 1-page summary (with some graphs for better clarification if possible) explaining how the company performs during this quarter/financial year - and our favorite statistics in large tables should serve as appendices for those who are looking for details.

The compilation process can be complicated, but the results should be addressed in a user friendly way.

What are the source data?

Ideally, we should request a separate set of monthly new business data from our MIS / Data Warehouse, instead of using the in force policy data. The new business data should extract all policies set in force during the reporting month, regardless of their month-end statuses.

If our policy admin system are setup properly, the new business can be easily identified by referring to the "issue date" instead of "effective date" / "risk commencement date" (RCD):

• Issue date: Date of a proposal is converted into an in force policy.
• RCD: Date of the coverage starts. For backdated policies, the "RCD < Issue Date". However, if you find out "RCD > Issue Date", it is either the data are incorrect or the logic in our system is not defined properly.

The advantage of using a separate set of new business data is we are able to capture the new business set in force and terminated in the same month (i.e. not available in our month-end in force policy data).

If MIS / Data Warehouse is unable to produce new business data, we have no choice but to compare the current month and previous month month-end in force policy data:

• Identify new records: Identify the policies that appear in current month data and NOT previous month data.
• Check issue date: The new records with issue date in current month are considered as "new business". The remaining records are considered as "reinstatements".
• This method is valid if the new business terminated in the same month is minimal or insignificant. You may want to do a study (say once a year) to validate your assumptions.

I would recommend you to still include the new business withdrawn during free-look period in your report - as excluding them will make the process remarkably complex. Alternatively, in order to provide better picture to the readers, you may want to include some info related to free-look cancellation in the summary of new business report.

What is the tool?
I would recommend you to use FoxPro (instead of Data Conversion System (DCS)) - especially its ability to do data matching.

We need to always remember that Actuarial executives are NOT IT programmers - they do not need to learn the programming language in much depth. The program written by an actuarial executive should be systematic and easy to understand (I will be very upset if I read a program that is without comment...)

For my previous employment, I have set up a set of training manuals used to train the juniors how to use FoxPro and do programming using SQL. I reminded my juniors that we should do the programming according to the approaches recommended in the manuals - so that the future successors will be able to pick up the program easily. If they find out any new command/approach which are useful in making the compilation process more efficient, they should revise the training manuals accordingly.

Additional notes...
We need to always note that among all actuarial reports, the new business statistics have the most readers - the Management & Sales Department use these statistics to measure their production performance. For life insurance industry in Malaysia, the life insurance companies submit the statistics to LIAM (Life Insurance Association Malaysia), in order to produce a consolidated report for industry production comparisons.

Movement of Policy (2): Actuarial? Data Warehouse?

If someone asks me: "Who should be the most appropriate person in charge in doing movement of policy (MOP) reporting - Actuarial or Data Warehouse (or IT Department)?" After some hesitation, I shrug my shoulders: "Frankly, I'm sorry that I don't know..."

Ideally, MOP reporting is basically summarizing & categorizing the data available in the policy admin system. If the system data are "clean" (what a luxurious wish), we can just apply some rules to segregate the available policies into "new business", "termination/alteration" and "in force" - just like what we do using Excel, FoxPro or DCS (Data Conversion System). Furthermore, the rules are actually easily understood by even non-actuarial colleagues. Now, it sounds like it is a bit too expensive to as Actuarial Department to do MOP reporting.

However, in reality, the data in the policy admin system in many companies are not clean - especially those with group insurance/takaful business. There are so many types of erratic data until it requires Actuarial to decide whether a policy is considered "valid" in MOP's perspective. Furthermore, some companies may have more than one policy admin system and MOP reporting requires combination of the policy data from multiple systems.

Sounds a bit tricky now? Not yet until you hear this: to make the matter worse, the policy data for some group insurance/takaful may not exist in any of the policy admin system! Some products may be manually managed by the Group Insurance/Takaful Department or Employee Benefit Department. Sometimes you may be informed by the sales personnels: "I'm sorry that the data you require are available at the third party's server and we haven't received those data yet..." (You may feel like screaming at him: "What? I thought the policy has been issued 6 months ago???")

We have to accept that the real life is imperfect. "By right, it should be simple, correct?" You're correct, "by right" it is. Sometimes the way we manage a supposed-to-be simple thing makes it becomes complicated.

Back to Basic - Movement of Policy (MOP)

If you ask me what a fresh graduate should learn first when he joins your actuarial department, I will definitely answer he needs to learn movement of policy (MOP).

In my view, MOP is the basic of all actuarial studies - it is about how a policy " behaves" during the entire life cycle, i.e. from when the policy is set in force until the policy is terminated (including maturity / expiry). By having a good understanding in MOP, it will be easier for us to visualize the relevant cash flows occur during policy term - especially when we setup an actuarial model to project cash flow. When I setup the Takaful Library for my client, the "base" variables I setup first are MOP related variables.

From what I learn from friends in the industry, many companies are struggling to get the "correct" MOP statistics - which are known as L6, L7 & L8 for conventional insurers and FT5, FT6 & FT7 for takaful operators (namely "New Business", "Termination/Alteration" & "In Force Business" respectively). These statistics are the most widely used statistics (especially in measuring industry performance & competitor analysis), unfortunately they are normally done by junior staffs without adequate guidance from their seniors (perhaps used to be the seniors' nightmare previously) to improve/correct the MOP statistics.

Well, in case you are looking for some info on how to preparing these statistics, I hope the summarized write-up below can provide you some insights. For the purpose of reporting, normally we will compile statistics on no. of policies, sum assured and annualized premium.

1. New Business (L6/FT5) - all new policies issued during the current reporting month. You should refer to the issue date of the policy instead of effective date / risk commencement date. By right, issue date should be the date when a proposal is converted to in force status.
    
2. Termination / alteration (L7/FT6) - all policies terminated & reinstated during the current reporting month, or the changes (i.e. alterations) done on the in force policies.
   
   Types of terminations are (1) maturity/expiry, (2) lapse (due to non-payment of premium / exhaustion of funds), (3) surrender (voluntarily withdrawal), (4) free look cancellation, (5) cancellation due to other reasons such as non-disclosure or administrative errors, (6) claim (that terminates policy, such as death, accelerated total & permanent disability (TPD) and accelerated critical illness (CI)).
   
   Alterations refers to (1) reinstatement, (2) the changes done on sum assured amount and premium amount to the in force certificates, (3) addition/termination of riders.
    
3.  In Force Business (L8/FT7) - the in force policies which are considered as in force as at end of the month. You may find out the "in force" policies may have various policy statuses, such as "In Force (normal premium paying)", "In Force (premium waived)" and "Paid-up", "Reduced Paid-up (RPU)" and "Extended Term Insurance (ETI)".

Hence, I think now you can easily understand the important MOP equation above - let's take no. of policies (NOP) as the example:

• NOP at beginning of current month (also NOP at end of previous month)
• Add: New business acquired during current month
• Minus: "Net" Terminations occur during current month (Termination NOP - Reinstatement NOP)
• Equal to NOP at end of current month

The same concepts apply to sum assured and premiums. Of course, the actual actuarial reporting work require more considerations - which I will share in my subsequent articles. You are most welcomed to your opinions/ideas on MOP here as well.

It depends...

If you ask someone: "Do you think you can complete all required tasks by this evening?" and the answer you get is "Well, it depends...", do you feel irritated? I would say most of us will answer "Yes", as it is not much different compared to not getting any answer at all!

This was one of the most valuable lesson I learned from Mark Saunders, the speaker of Presentation 4: "Actuaries and communication skills -- an oxymoron? Some thoughts and tips", during Jointly Regional Seminar 2011 recently held in Kuala Lumpur, Malaysia (19-20 July 2011). I totally agree with Mark that "it depends" is a really bad answer that does not give any meaningful / value-added feedback to the person who is looking for answer - especially if we would like to make the CEO upset.

I posted Mark's "it depends..." in my Facebook and my friends shared with me other irritating ways of answering a questions, such as:

• "I will get back to you ..." which most of the time no feedback after the meeting ends.
• "See how things go..." which also have the same meaning as "I don't feel like giving you an answer now..."

OK... I admit I used to answer in these ways before and I already promise to myself I will avoid these words in the future. Oops... Is this how you upset your bosses as well?

Talking about Claim

I have a great discussion with the HOD of life claim department this afternoon, and I have learned plenty of things. I always like to talk to the HOD of claim department (either in the previous company or current company I work), as he/she shares a lot of real life examples which make make me know better how insurance works in reality, and the problems and challenges he/she faces in managing insurance/takaful claims.

Actually, the main purpose I had the discussion with the HOD was to understand how the claim process is carried, so that later the system development team starts working on the claim module enhancement, the problems the claim assessors facing now (that related to system) can be resolved together. Well, I am glad that he shared some real life cases with me during the discussion, helping me to understand better how he & his staffs assessed claims.

I always agree that claim department is an important “gatekeeper” of insurance/takaful companies, apart from the underwriting department. Theoretically, claim assessor should only pay claims if the disabilities/events meet the definition in the contract and he should reject the claim if otherwise – unfortunately, this is too ideal in practical. Although some claims do not meet the definition stated in the contract, the customer may complain to some body/regulator or take legal action towards insurance/takaful companies, which eventually the insurance/takaful companies have to pay the claims which are initially rejected.

The reason we have contracts for insurance/takaful contracts because we want to properly draw up the exposure under a certain products, so that we can set the premium rates accordingly – if an insurance/takaful is very “lenient” in paying the claims (i.e. still pay claims even though the disability/event doesn’t really meet the definition stated in the contract), eventually the victims are the policyholders – they will have to pay a high premiums to cater for high claims. Insurance/takaful companies are not welfare bodies and they definitely will not charge low premiums for high claims.

If you find out an insurance/takaful company accepts risks easily and pay claims generously, and you are actually a healthy life, I would urge you to think twice if you intend to purchase a policy from that company… Remember you are the one who pay claims.

Chat with an Actuarial Junior Staff

I have a chat with my ex-staff, J, last Friday in UAT Room. I was assisting the Business Analyst (BA) in supervising the new product User Acceptance Test (UAT), and Jeff was one of the user from Actuarial Department.

Since J joined the company, he has been working in the system implementation team – well, in short, most of his work is to carry out UAT for new products and migrated products. I asked him: “How long you have been here already?” He answered 2 years, and he hasn’t had any chance to really involve in other actuarial works like pricing and valuation. I was a bit worried with his limited experience – I told him it was high time for he to start to learn other types of jobs, as he doesn’t even have any experience in preparing the statistical reporting (like new business report & termination report) or experience studies (like mortality study and persistency study). The situation will be more difficult when he becomes a senior executive but doesn’t know how to really do a junior executive’s work.

J seemed to be not so agree with my opinions… He told me some more senior actuarial people told him that it is OK to work in other areas for a few years and switch to Actuarial Department later, just like Ms. S who has been working as an underwriter before she joined actuarial department. I respected their opinions, but I told him I have another view: when Ms. S graduated from the university, the supply of actuarial talent was very low, hence she did not face any obstacle in switching to actuarial field; however, the current environment (in Malaysia’s actuarial industry) has changed, the supply seems to exceed the demand, especially for fresh graduates. If J continues to do system work for another year without learning any pricing/valuation work, I’m doubt if any other employer will pay a senior executive salary when he gets a pricing/valuation job in other companies – especially if he has passed many papers.

We didn’t really have any concrete conclusion for our conversation… Well, there is no right or wrong answer. But I do really hope he will progress well in his career.

Prophet Modeling (4): How did I setup my Takaful Library?

Hooray, I finally completed the Takaful Library! Working together with Ms. H, we finally setup the Takaful Library from scratch and created 32 “Products” (and another 19 “Same As” products). I am glad that the Valuation Team (headed by my ex-boss Mr. R) can now using a consolidated workspace to do the actuarial studies, especially asset share study and reserve & liabilities capital charges for new valuation basis and Risk-based Capital (RBC) submission.

After around 4 months of hard work, finally I can really sit down and “document” this exercise. I will share the experiences I have in this article and several subsequent articles later – hopefully they can help to provide some ideas to you…

The above flow chart summarizes the approach I used to complete the Takaful Library project for my Client:

Approach Papers
I started the project by preparing documentation (Excuse me, I think I heard someone just shouted “Oh no!!”…), which I generally called them as “approach papers” for convenience. There are few documents to be prepared a this stage:

1. Product documentation
   To document the product features and product specific rates that are required in the model.  I only documented those required in the product setup – I didn’t spend time to document items such as exclusions of a products.
    
2. Model approach
   To document how the model should be setup, such as movement of policy, cash flow, asset share, profit/surplus declaration and others. You also need to document the simplifications/assumptions used to setup the model – for example, I did not create a separate movement of policy for riders, i.e. assuming the policyholder will not terminate a rider while the base plan remains in force. Such assumptions need to be documented so that users and future Prophet Managers are aware of these assumptions (instead of wasting time to study the model all over again).
    
3. Reserve approach
   To document how the reserves (e.g. statutory reserves) & solvency are calculated. I did not combine the “reserve approach” and “model approach” in a single document because “reserve approach” is product specific and it requires more frequent updates.

These documents are similar to the BR (Business Requirements) of IT projects. However, I didn’t request my Client to sign-off the documents as we definitely need to further enhance the documents along developing the models (instead of wasting time on too many formalities).

Modeling
Although my focus was to develop the Takaful Library in Prophet, I still setup the similar model in spreadsheets – spreadsheet modeling is the basic form of actuarial modeling that every actuarial people should know.

The spreadsheet models are mainly used to check the results produced by Prophet. They helped me a lot in detecting errors which were difficult to identify by reading the variable definitions.

Of course, spreadsheet models were easier to be reviewed by my Client.

Review & Validation
As mentioned above, I used spreadsheet models to validate my Prophet results. In fact, I also reviewed the variable definitions one-by-one – I printed out the definitions on recycled papers (more than 200 pages!!) and went through every single definition. I admit that this was a very painful exercise.

The review & validation exercises were split into 2 stages: (1) single model point; (2) portfolio. The single model point run used spreadsheet models to validate the Prophet results (the differences due to different ways of rounding are extremely small); while portfolio run was done by my Client using the in force data for annual valuation exercise, mainly for reasonableness checking.

Apart from using a single deterministic scenario, ideally we should do sensitivity testing for single model point run & portfolio run as well – however, due to time & resources constraints, I only managed to do very limited sensitivity testing, mainly on movement of policy.

Sign-off
Of course, sign-off was the final stage of this project – my Client & I needed to sign-off not only on the Prophet Library & workspace, we also needed to sign-off documents. I have also prepared a “project closure document” to summarize what we have achieved in this project – this project was considered after we provided sign-off for this document.

I will share my experience learned from this project in more detailed in my coming articles. Feel free to provide your comments & suggestion.

More Heart Attack during World Cup Tournament?

Are you a hardcore football fans? Are you extremely stressed when you are watching the match which the team you support is about to lose? If your answer is "Yes", you better take care your heart well during the World Cup Tournament...

I learned an interesting fact during a seminar yesterday, i.e. a study reported that during World Cup Tournament time, hospital admission from people having heart attack increases tremendously due to emotional stress - the example given was when Argentina beat England in a penalty shoot-out during the 1998 World Cup, there was a 25% increase in hospital admissions for heart attacks in England (You may refer to "Can Watching Soccer Give You a Heart Attack?" for details).

Oops, look like football can appear as a hazard which affect the losses arising for health / critical illness insurance...? (I never understand how the emotional stress is because I have no interest in football...)

Prophet Modeling (3): Something About Documentation

I was a bit surprised when Ms. Z called me. She asked me for a favor if I could provide a brief explanation on how actuarial modeling is done, especially related to Prophet – because she needed to review of the actuarial modeling processes.

“OK, I can only share with you whatever I did in the past. But you need to check with Actuarial Department for the latest practice they have.” I am no longer with Actuarial Department, hence I couldn’t say on behalf Actuarial Department – furthermore, there is a new leader for the Modeling Team. What I could do was only sharing my past experience when I did my reserve workspace and simplified asset share workspace.

I met Ms. Z the next day, together with her colleague (Oops, I didn’t ask what her colleague’s name is…). I explained how Prophet works in a very brief way, so that they could have some ideas how Prophet helps actuarial people doing the job. “Then what were the documentations you had when you setup the actuarial model in Prophet?”

Yes, it was really a good question. I was trained by my previous Appointed Actuary to have a proper documentation in most of the works I have done – my AA told me: “You will not be forever here and other colleagues who take over your work in the future will not know what you think now”. So what are the documentations I had when I developed those workspaces?

1. Approach papers – this document states the methodology I would like to adopt when I setup my Prophet model. In this document, I listed the assumptions (not the projection assumptions we use to project cash flows) and formulas used to calculate a certain value. For example, I prepared a document explaining how the model calculated the statutory reserves for a mortgage plan.
   
   This approach paper could be “product specific paper” (like reserve paper) or “general paper” (for the calculations which the formula didn’t vary by product, such as asset share calculation).
    
2. Spreadsheet – Based on the methodology I stated in the approach paper, I also prepared a spreadsheet model so that I could use to counter check with the results generated by Prophet – for a single model point record. I found out this was very useful in helping me to detect the errors/mistakes which I didn’t realize when I checked the coding (I have to admit that checking spreadsheet was easier than checking coding). Of course, this spreadsheet was setup using recursive method like Prophet.
    
3. Reconciliation – Before I setup the Prophet models, the valuation work was done by an actuarial consultant. When I did my reserve workspace, I managed to reconcile the results generated from Prophet to the model the consultant used, and the major differences were explainable (e.g. the consultant used annual projections and the interpolations; Prophet used monthly projections). The comparisons were documented properly – so that we understood the impacts of changing model.

This reconciliation is particularly important if we discover any errors in an actuarial model, so that we can understand the impacts of correcting the errors, the implications to the financial statements (if any) and also learned a lesson (and not to repeat the same error in the future)!

Of course, there were some other documents I have prepared for the users and the developers, such as training manuals and user manuals. I agree that doing documentation is a painful experience but you will find out that the documents will help you a lot in the future – especially dealing with the auditors. It’s very true, trust me.

Prophet Modeling (2): How to Organize Variables?

During my discussion with Mr. L & Ms. H on Prophet modeling, I briefly touched about how to organize variables more efficiently into various Variable Groups (from A to Z).

In the existing reserving workspace and simplified asset share workspace (deterministic models) which set up by Ms. H & I (also Ms. B) previously, we also split the Variable Groups into “core variable groups” and “input variable groups”, which is similar to the standard Conventional Library. The major different between our approach and the standard Conventional Library is we have the same type of classifications for both core variables & input variables – i.e. A-M for core variables, and N-Z for input variables.

For example, we have created Variable Groups for decrements like mortality rates – say we used Variable Group “D” (the 4th alphabet starting from A) for core variables (D – Decrements (C)), and similarly we use Variable Group “Q” (the 4th alphabet starting from N) for input variables (Q – Decrements (I)). This is because for the same types of variables, they can appear as either core variables or input variables:

1. The variable used to define the valuation mortality table needs to be an input variable because different valuation mortality tables may be used for different products.
   
2. The variables used to calculate monthly mortality rate (i.e. 1-(1-qx)^(1/12)) needs to be a core variable because the same formula is used regardless of the valuation mortality table used (i.e. generic for all products).

Of course, we also use different way classifying the variables – which we think to be more helpful in term of managing the variables systematically. For example, we have a variable for policy profile which most of the variables are read from model points – such as entry age, gender and sum assumed. “Systematic” is very important especially once we have more and more variables in our library.

Prophet Modeling (1): Some Thoughts on Setting Up Takaful Library

I had an hour chat with Ms. H & Mr. L on Prophet Modeling yesterday. Ms. H is my ex-staff when I was the leader for the Modeling Team, she is quite brilliant in helping me setting up the Prophet models (simplified) the Valuation Team currently uses; while Mr. L is my replacement. The objective of the short chat was to exchange some thoughts on how to proceed the work to model a comprehensive cash flow projection model for the Takaful business.

Well, although I have partly did the projection model before I resigned from Actuarial Department, I suggested to Mr. L that he should start setting up the model from scratch – instead of continuing from my partially complete model or doing amendments from the Conventional Library. Well, doesn’t this consume more time? This sounds correct if we only have a quick thought, but if we really make a thorough consideration, actually setting up the model from scratch is easier and more efficient.

Let’s talk about doing amendment from Conventional Library first. When I was with my previous employer, the Prophet model I used in my valuation work was modified from the standard Conventional Library. I told Mr. L (who also previously worked in a conventional insurer) that it was alright for the models we used previously to use the standard Conventional Library as a base, because how the conventional insurance business works is close to what the standard Conventional Library has modeled. Mmm… The takaful business doesn’t work the same way as the conventional insurance business (not only different naming for the terms to be used) – modifying the standard library may take more time and may even tends to have more errors – furthermore, there are variables in the standard library that we may not have any idea what they are. Hence, it is better to create the Takaful model from scratch.

How about modifying from my partially complete model? In my view, Mr. L may have his own modeling philosophy and it may not be right to force him to follow my modeling philosophy – unless I have completed the model. Furthermore, he will have better control on the Prophet model he is going to setup – most importantly, he will have better understanding on what he is doing. I remember that one of the appointed actuaries I have worked with commented that there was a tendency that actuarial models worked like a “black box”, which the actuarial people who used the model didn’t understand how the model really worked. Consequently, many mistakes happened here and there in the actuarial analysis – and the errors were EXPENSIVE.

Mr. L also told me that he has some other tasks on his plate, apart from setting up the Takaful model. “Mmm, this is no good.” I commented honestly. Setting up model requires full focus, especially if he has a limited time to complete the model. I even told him that he should even lock himself up in a separate room which doesn’t have interruption.

“Otherwise, you will be in the same boat…”

Already Graduated for 10 years…

“We have graduated for almost 10 years…”

During the “pre Chinese New Year” dinner I have with a few Actuarial Science (normally we call “QA”) course-mates yesterday (20.01.2010) at Purple Cane Restaurant (Chinese Assembly Hall, Kuala Lumpur), I made the above remark. Yes, time flies, we have graduated from Universiti Kebangsaan Malaysia (UKM) for almost 10 years already.

It was really a great time to catch up with my old friends who I have known for more than 10 years. We also talked about other course-mates who were not there – I did hope that I was able to invite more (especially who work in Kuala Lumpur / Selangor), but unfortunately I do not have all contacts. Some of them work in oversea, such as Bee Chuan, Fong Siong (I just knew that he got married already!), Bee Chee and many more; some of them are we have totally lost contact (not even via Facebook), such as Eng Keong. I really appreciate the time I studied in UKM with these course-mates and it was a great thing that we still keep in touch and have such gathering after so many years.

Gim Chong (and wife), Kok Keong, Kian Sing, Chuen Chee (and wife & son), Ann, Ee Hoon and Kau (and wife), thank you for joining this “reunion” dinner.

“How about next activity?”

“Well… Maybe we can have a visit to UKM campus and meet our lecturers…”