Using Actuarial Information To Improve Claims Outcomes 

Introduction 

Actuaries require accurate and timely information for them to calculate the reserve loss development pattern as well as the ultimate exposure for losses.  

Even though management may not always be happy with the information, knowing the correct ultimate exposure is always ultimately good for the insurance company or the employer.   

Due to company silos, process or ignorance, most claims operations do not regularly engage with the actuary process.   

Quality and focused claims handling can directly and positively impact the projected actuarial outcomes.  Directed information from the actuaries can help focus the claims process to maximize the outcomes from their efforts.  Many claims’ handlers do not understand the actuarial process and as a result, they do not appreciate the impact which the claims administration and reserving process has on the financials of the insurance company or of self-insured employers.  Nor do most claims operations regularly receive targeted actuarial reports from the actuaries.  A lack of information keeps them from focusing on the claims which will have the most positive impact on the actuarial outcomes.    

Additionally, many actuaries do not receive timely and accurate claims trending information which would improve the accuracy of their work. This is also due to the fact that the claims operations do not always appreciate the impact their claims process have on the ultimate outcomes.   Without appropriate insight, actuaries make assumptions which may not always result in the most accurate calculations.    

This paper: 

1.   Provides insight into the actuarial process for claims handlers. 

2.  Educates claims handlers on the impact that actuarial results have on the employers’ financials.   

3.  Encourages appropriate communication between the claims operations and the actuary to foster accurate actuarial reporting and to avoid costly actuarial mistakes. 

Actuarial science is the discipline that applies mathematical and statistical methods to assess risk in insurance, finance and other industries and professions.  

Actuaries are professionals who are qualified in this field through intense education and experience and certification. 

The actuarial profession was originally developed for insurance companies to help determine the pricing “cost of goods sold” of the insurance product as well as their total exposure of their liabilities (losses).  Insurance commissioners require insurance companies to maintain an appropriate ratio of “surplus” (unrestricted assets) to their projected losses in order to ensure that the insurance companies remain solvent for their investors as well as for the insured customers.  In the last several decades, this science has also been used to help determine the need for collateral by insurance companies who sell large deductible programs as well as State governments who have collateral requirements for self-insured employers. 

Actuaries do this by developing “loss triangles” out of reserves and claims payments. From these triangles, they determine the best “loss development” factor to use for their employer. 

I have attached several appendices explaining how they develop a loss triangle and how they use the triangle to manufacture a loss development factor to determine the ultimate projected losses for an employer.  I highly recommend that if you have never worked with or seen a “loss triangle” or are not clear on how a reserve loss development factor is calculated, that you read these appendices. 

Improved communication between the claims handler and the actuary can provide important information back to the claims handlers on where to focus the claims effort or what information may be helpful for the actuary to do a better job. For instance when there is an increased focus on settlements, the actuary may only see the increased in paids and not recognize the reduced long term fiscal impact by closing the files early unless the change in process is communicated.  Additionally the claims handler may not know it is a specific accident year which is driving much of the negative results.  A re-focus on a specific accident year may make a big positive difference for the employer. 

Collateral requirements 

Self-insured companies are required to post their anticipated losses as part of their listed liabilities on their annual financial reports.  These requirements come under the Sarbanes Oxley financial reporting requirements promulgated by the Federal Government.  

Most states require self-insured companies to post enough collateral to cover the potential ultimate losses should they become insolvent or become unwilling to pay the workers compensation benefits to the legitimately injured workers.  Bankruptcy does not always indicate that a company is unwilling or unable to meet their obligations to pay workers compensation benefits.  For instance, during the last economic depression several airline companies were in Chapter 7 but continued to pay benefits and maintained their self-insurance certificates in most jurisdictions. 

If a company is using a reinsurance company to insure aggregate or large losses, or if there is a large deductible insurance policy, most insurance companies require the self-insureds to post collateral for their potential exposure.  This exposure is usually calculated using actuaries. 

Some insurance carriers use the outstanding collateral they are holding as a method to help retain the insured on risk. They do this by retaining some of the collected collateral to cover potential IBNR (Incurred But Not Reported) claims, (even if the all of the accident years claims are closed).  

Usually, collateral is determined by a combination of the Ultimate Loss figure and the financial rating of the employer.  

Collateral can be in the form of cash, letters of credit, surety bonds, Credit Default Swaps or other convertible assets. 

Although several different methodologies are available to determine the appropriate Ultimate Loss, the most common are via “development methods”, whereby Paid Losses, Incurred Losses, or Claim Reserves are developed to their ultimate value based on development factors.  The actuaries will then make a decision on using one or a combination of the three data sources.  

There have been a few employers who reported their exposure based only on the outstanding reserves of the cases.  Historically the states have independently applied a factor to these reserves. 

A few years ago there was a large retail employer who calculated and posted on their annual financial report the undeveloped outstanding reserves.  When they changed TPA’s, the files were all re-reserved based on a different reserving philosophy.  The result was an increase of $26M in outstanding liabilities.  The increase in exposure resulted in four cents a share hit to the company’s bottom line in that quarter. The use of an actuary would have significantly mitigated the surprise.  

In most jurisdictions, the practice of only reporting and posting based on outstanding reserves has been supplanted by requirements for an actuarial analysis. 

Public Agencies and Collateral Requirements 

Most public agencies in the United States are not required to post collateral for their outstanding reserves.  This is because technically they cannot become bankrupt and even if they do become “insolvent”, they will not stop paying the benefits to their injured workers. They usually project their costs as part of an annual budget process. Therefore many do not use actuaries or bother to project their ultimate exposures.   

Reserving practices have a significant impact on the financial health of insurance companies, as well as on employers who retain some workers’ compensation risk.  Within some states the oversight of the reserves and reserving process may differ significantly between the insured claims and the self-insured claims.  Usually the State Department of Insurance oversees all insurance company activities including reserving practices.  While many states have a dedicated self-insurance department which regulates the reserving and collateral requirements for self-insured companies.    

States have specific reserving and claims handling requirements but there are many variables, when applied, which can impact the projected ultimate claims costs.  This can be problematic for self-insured companies who have exposure across the United States.  For instance, some States allow claims to be administratively closed (when future medical care has been awarded) after two years of inactivity.  Other factors impacting reserves can be which “life expectancy table” to be used to calculate Life pensions, or future medical care. Some states do not recognize medically reduced life expectancy reports from doctors or allow this information to be factored in the reserves.  

Most jurisdictions do not formally recognize that there is significant ongoing medical inflation in the reserving process.  They only require the medical reserves for calculating the “future medical” cases to be calculated using  the recent years average annual expenses times the injured workers’ life expectancy rather than also include an annual inflation multiplier.  Many actuaries include a medical inflation rate in their calculations of the calculation of the ultimate loss.  

Changes in exposure which may impact actuarial results 

Claims handlers have a direct impact on most (not all) of the factors which impact the actuarial projections of the claims.  

One of the most important factors for accurate actuarial results in consistency in the reserving practices.  This is why the timing of the reserve changes and accuracy of the reserves is considered one of the fundamental jobs of the claims handlers.  

It is important for examiners to recognize when there are potential changes in exposure or claims handling practices and communicate these changes with the employer and the actuary to help avoid surprises.  

The first and most important action is to make sure that the data provided to the actuary is consistent (in definition) and timely. 

If there is a change of claims administrators, understanding the differences in claims handling practices and reserving philosophy and communicating any differences in practice or definition to the actuary will help avoid surprises from the report.  

It is also important to make sure that the data is always collected at the same time.  A few days in certain times of the year can make a significant difference in outcomes. 

Factors which can change the actuarially predicted outcomes for the employers (development factor): 

• When there is a change of claims system (particularly true when changing Claims Administrators) (this is because different systems may have different definitions for opened or closed claim) (some claims operations do not reopen claims to pay minor medical bills after the claim is closed and some do  re-open all claims to pay any bills). 
• When there is a change of definitions within the claims systems (sometimes if there are changes in the claims systems these changes are not conveyed to the actuary with an adequate explanation of what was changed and why it was changed). 

• The definition of a closed claim can be problematic.  Some organizations require a re-opening of all claims to pay old bills. Some allow up to $2500 to be paid on a closed claim without re-opening.  
• When there is a change of claims handlers. (either change of TPA or within the TPA). 
• When there is a change of settlement philosophy (Stipulations vs C&R). 
• When there is a change in the laws rules or regulations (changing medical cost containment programs from Loss to ULAE). 
• When there is a change in the fundamentals of the system (the opioid epidemic). 

• When there is a change in the exposure at the employer (new hires, layoffs, implementation of a new safety program, 
• When someone “messes” with the claims reserves (CFO reducing reserves on the top 10 cases in each office by 20%) 
• Present value and discount rate. (Some companies state their reserves after using a present value discount rate some don’t) 
• When the actuarial society puts out a notice (advent of Medicare Set Aside). 
• Fraud gambits. 

• Changes in IBNR claims, and changes in claims closure patterns 
• Calendar year vs accident year (IBNR again) 

Conclusion 

Prior to the production of the annual actuarial report, there should be at least one meeting between the claims management, the actuary and the employer to discuss the overall claims handling process, if there have been any changes in potential exposure, and what the claims handlers are doing to mitigate any identified exposures.  

Appendix I (Paid loss triangles) 

A loss triangle is the primary method in which actuaries organize claim data that will be used in an actuarial analysis. The reason it is called a loss triangle is that a typical submission of claim data from a client company shows numeric values forming a triangle when viewed. 

The easiest loss triangle to explain is the accident year paid loss triangle. This loss triangle totals paid loss data in each of two categories. The first is accident year. Simply put, all paid losses from all claims occurring in the accident year are totaled. 

Now keep in mind, these losses are cumulative, not incremental. This means loss payments are included regardless of when the payment was made. The only thing that determines the accident year of a claim payment is the year of the accident. 

For example, James Slipsalot fell down on the job in 2004 and became permanently injured. All loss payments for Mr. Slipsalot’s claim are included in accident year 2004 paid losses. If you are instructed to calculate accident year 2004 losses, you must add all of the payments to that worker that have ever been made, not just the payments made in 2004, or the payments made in the current year. You must include the payments to Mr. Slipsalot even if the claim has been closed for several years. 

The second category that claim data is organized by is called “age”. Age, in this context, is the number of years from the beginning of the accident year to the moment of the last transaction in your database. So, let’s say $75,000 was paid for Mr. Slipsalot’s claim in 2004(mostly medical); then in each subsequent  year an additional $30,000 (mostly indemnity) has been paid; he is relatively young, so you expect that these $30,000 claim payments will continue for the next 30 years. For this claim the information can be presented as follows: 

James Slipsalot’s Claim Data
Accident Year	Claim Age	Valuation  Year	Paid       Losses
2004	1	2004	75,000
2004	2	2005	105,000
2004	3	2006	135,000
2004	4	2007	165,000
2004	5	2008	195,000
2004	6	2009	225,000
2004	7	2010	255,000
2004	8	2011	285,000
2004	9	2012	315,000
2004	10	2013	345,000
2004	11	2014	375,000

In the first row of the table above we can see that when the 2004 claim was 1 year old $75,000 had been paid for this claim. The second row shows that when the claim was two years old $105,000 had been paid (the original $75K plus the $30K in the 2^nd year). The paid loss in each row after that is $30K higher reflecting the $30K that gets paid in each subsequent year. 

On this table, you’ll notice that there is another field called “Valuation Year”. This field does not contain any new information. It just reflects the year in which the claim is valued (i.e. the latest year in which data is included for the analysis). Valuation year is equal to accident year plus age minus 1. The reason for the “minus one” is because the age of an accident year starts at the beginning of the accident year (i.e. 1/1/2004), and the valuation year is always done at the end of a year (i.e. valuation year 2004 contains data through 12/31/2004). As a result when the valuation year and accident year are the same, the age is actually one year. 

Finally, as a minor point, please remember that all claims with accidents in the same year have the same age regardless of when in the year the accident took place. Accidents occurring in January and December, 2004 both have an age of 1 on December 31, 2004. 

Now that you have all of the terminology down, this is the resulting paid loss triangle: 

	Age
AccidentYear	1	2	3	4	5	6	7
2008	403,082	688,542	808,102	917,258	1,069,538	1,124,462	1,167,216
2009	229,968	367,469	401,836	427,278	428,350	428,350
2010	175,320	303,919	348,486	356,487	393,186
2011	226,710	326,284	364,840	367,698
2012	275,881	405,696	445,860
2013	137,245	238,268
2014	40,913

This loss triangle was created using data available through December 31, 2014. 

The loss triangle above shows losses from accidents between 2008 and 2014 valued at each year between 2008 and 2014. So now you may wonder, why does the presentation of claim data always end up looking like a triangle? Well, the parts of the triangle with no numbers represent valuation years that are in the future. These values are unknown, and therefore blank. 

As an example, $403,082 had been paid for accidents occurring in 2008 through age 1 (valuation year 2008, valuation date 12/31/2008). We know that by looking at the first row and column of the table above. Then, the total amount of payment on these claims grew, albeit more slowly, with each passing year. By the time the 2008 claims were aged 7, a total of $1,167,216 had been paid for these claims. We have 7 independent annual observations for accident year 2008 and those observations are in the first row of the triangle. 

Accident year 2014 is a completely different example. $40,913 had been paid for accidents occurring in 2014 through age 1 (valuation date 12/31/2014). We have no idea what accident year 2014 losses will be at age 2 through 7 because those ages are in the future. Therefore, for the 2014 accident year, Age 2 and all later ages are blank. 

Accident year 2013 has information for ages 1 and 2, but nothing after that, accident year 2012 has ages 1, 2, and 3. You get the point. Each older year has one more element of data than the year before resulting in a triangle shape whenever historic information is presented by accident year and age. 

Appendix II (Loss Development Factors) 

Loss development factors are typically organized in loss triangle format. 

Loss development factors are significantly relied upon in many actuarial loss reserving methodologies. They can be calculated entirely from loss triangle data– no additional data needs to be provided to the actuary to calculate loss development factors. A loss development factor is the loss value in a loss triangle divided by the value immediately before it in the loss triangle. 

For example in the loss triangle example, the first loss development factor for accident year 2008 is equal to $688,542 divided by $403,082 (the value right before it). That loss development factor (LDF for short) is 1.708. It is called the accident year 2008 age 1 to 2 loss development factor. Generally, the first thing that an actuary will do with a loss triangle is create a loss development triangle. 

Here is the one created from the loss triangle example. 

Accident Year		1-2	2-3	3-4	4-5	5-6	6-7
	2008	1.708	1.174	1.135	1.166	1.051	1.038
	2009	1.598	1.094	1.063	1.003	1.000
	2010	1.734	1.147	1.023	1.103
	2011	1.439	1.118	1.008
	2012	1.471	1.099
	2013	1.736

You probably notice a few interesting things with this triangle. First off, it is smaller than the loss triangle. There is one fewer row and one fewer column. This is because the LDF is calculated from two numbers; so there is no loss development factor available for 2014 where there is only one loss observation. Also, there are only 6 loss development factors that can be calculated for the 2008 year. 

The next point of interest to most people is that all of the factors are over 1. This isn’t a surprise as paid losses will generally be higher in each passing year as more and more losses get paid. For accident years that are very old and have no claim activity, the loss development factors will drop to 1.000 meaning that losses are unchanged between successive valuations. 

If you take the first number in a row of a loss triangle and multiply it by all the values in that row in the loss development triangle, you will get the last number in the loss triangle. 

For example, for accident year 2008:  $403,082 x 1.708 x 1.174 x 1.135 x 1.166 x 1.051 x 1.038 =  $1,176,216. 

The loss development method, assumes that future loss development factors for a particular age will be the same as those in the past. In other words, the loss development factors for a particular column in a loss development triangle will be the same. 

Appendix II (Ultimate Loss Development Methodology) 

Loss Development Methodology is the final step in the most commonly used actuarial method to estimate loss reserves. 

The final step of the loss development methodology is constructing a table that looks something like this: 

	1	2	3	4	5
Accident Year	Paid Losses	Age in Years	Cumulative Loss Development Factor	Estimated Ultimate Losses (1) x (3)	Estimated Ultimate Outstanding Losses (4) – (1)
2008	1,167,216	7	1.160	1,353,917	186,701
2009	428,350	6	1.183	506,804	78,454
2010	393,186	5	1.242	488,460	95,273
2011	367,698	4	1.329	488,770	121,073
2012	445,860	3	1.502	669,716	223,856
2013	238,268	2	1.787	425,897	187,629
2014	40,913	1	3.217	131,635	90,722
Total	3,081,491			4,065,200	983,708

This part is definitely not rocket science. It is just putting calculations from all the other steps together. Let’s go through it. Column 1 is the latest observation of paid losses. These values are the rightmost value for each row on the loss triangle. They form a diagonal line on the loss triangle and represent the data for the 2014 valuation year. 

Column 2 is just the current age of the accident year. This report is using data through December 31, 2014, so the age of 2014 is 1 year and each year prior is a year older. 

Column 3 is the cumulative loss development factor based on our selected LDFs in the table here. 

Column 4 is just simple math. It’s column 1 times column 3 and represents an estimate of ultimate losses (all the losses that will ever be paid) for that particular accident year. 

Column 5 is also simple math. Ultimate losses minus paid losses is outstanding losses. This is an estimate of reserves for the accident year. Based on this methodology, the actuary estimates that ultimately $4,065,200 will be paid for all accidents occurring between 2008 and 2014. Since $3,081,491 has already been paid for these accidents, $983,708 must be reserved for future claim payments. 

The loss reserve approach is also commonly applied to incurred losses (cumulative paid losses plus outstanding loss reserves). It is the same exact methodology except the loss development factors are smaller (because incurred losses are larger). There is still an “ultimate development factor” and a projection of ultimate incurred losses. It is important to be aware that ultimate incurred losses will always equal ultimate paid losses (because ultimately reserves = 0). As a result the paid loss development methodology and the incurred loss development methodology are different techniques used to project the same number.  The actuary may rely on the paid or incurred loss development methodology more depending on whether he believes recent paid or incurred patterns are more predictive of the future. 

The final step for the actuary in a loss reserving study is to review the results for this method and other actuarial loss reserving methods and select a “best estimate” of ultimate losses and reserves. This “best estimate” will underlie the actuary’s loss reserve selection and actuarial opinion on loss reserves. 

 An example of an actuarial method: 

Bornhuetter-Ferguson actuarial method. This method is calculated as follows: 

1.  For each of the previous 10 years, the actuarially determined ultimate losses are divided by the payroll for each of these years. This result is inflated by 1% each year to account for payroll and workers compensation benefits inflation. For example, the first year has 10 years of inflation; the next year has 9, etc. This calculation yields the pure premium factor. 

2.  The actuary then calculates the following using the pure premium factors for each individual year: 

a.  Average of all years 

b.  Dollar weighted average of all years 

c.   Dollar weighted average of most recent 3 years 

d.  Dollar weighted average of the most recent 5 years 

e.  Time weighted average of all years 

3.  The actuary exercises discretion and selects a pure premium factor. For both years, the actuary selected the dollar weighted average for the most recent 3 years. 

4.  The selected pure premium factor is multiplied by hundred dollars of payroll. This result is the expected ultimate losses. 

5.  The expected % of losses unreported is determined from the incurred loss triangles for the years 2001 to present. The expected ultimate losses from step 4 is multiplied by the expected % of losses unreported. 

6.  The calculation from step 5 is added to the actual incurred losses reported to provide the expected ultimate losses.