Workers Comp Algorithms – Who Is Pushing the Model Buttons?

March 18, 2021 By JL Risk Management Consultants

Workers Comp Algorithms – How Far Have They Progressed?

The panel discussion yesterday with Frank Pennachio covered one hot topic – Workers Comp Algorithms.

I do agree that almost everything in existence today has an algorithm attached to it. All one has to do is pick up their smartphone. You are generating hundreds of algorithmic functions by using your phone.

Workers Comp algorithms are no different. Every time an injury is reported, a policy is placed, or even when a premium audit is performed an algorithm takes over the calculations. Let us step back to first define algorithms – you may be surprised with this one.

graphic of Euclidian Workers Comp algorithm

Licensed by Wikipedia Somepics

Algorithms Defined

A finite set of unambiguous instructions that, given some set of initial conditions, can be performed in a prescribed sequence to achieve a certain goal and that has a recognizable set of end conditions. (c) American Heritage Dictionary

The Wikipedia and Merriam Webster Dictionary (c) definition – In mathematics and computer science, an algorithm is a finite sequence of well-defined, computer-implementable instructions, typically to solve a class of problems or to perform a computation. Algorithms are always unambiguous and are used as specifications for performing calculations, data processing, automated reasoning, and other tasks.

The oldest algorithm is from 4,500 years ago by the Ancient Babylonians. The algorithm was called the Euclidian algorithm. I will not bore you with the details.

Check out the thumbnail of the Euclidian algorithm on the right side. If you click on it, it will resolve to a much larger graphic. The main things to remember are the words Entry and Input A, B.

Workers Comp Algorithms vs. Predictive Modeling

Are predictive modeling (analytics) and algorithms the same for Workers Compensation? Yes, they could be viewed as the same thing.

Frank brought up a great point that I think needs to be mentioned – there are 42 million claims that are digitized to some degree.

We could build workers comp predictive models with all that data. The largest single owner of workers comp data, from talking with data analysts, is Liberty Mutual. I have spoken with more than one of their analysts on the size of the bundled data sitting on servers somewhere – one word – humongous. By the way, their data analysts were impressive young men and women.

Simplest Algorithm Shows Major Modeling Unknown

Going back to the oldest algorithm quickly shows a major question mark. What data will be fed into any predictive model or algorithm. Remember, Input A, B from above? Who inputs those numbers? What do they decide the inputs should be for the world’s oldest model?

Workers Comp Algorithms = Same Model Problems

The inputs to the model would be almost the most crucial part. Check out the next section for the most crucial. The difficulty with workers’ compensation is that each state has a different set of rules. Would you need 50+ sets of algorithms/models with one for each state? (Probably so).

Wikimedia Commons – Jeff Keyzer

Take a look at the NCCI rating manuals, there are so many individual states that have modified each rule that the inputs to each model would have to almost be state-specific.

(Additional Rules: AZ, CO, CT, CT(A/R), DC, FL, GA, IL, KS, KY, LA, MT, NE, NV, NV(A/R), OR, OR(A/R), TN(A/R), TX, UT, VA, WV)

Now take the independent bureaus into account (15 of them) and you have a model that has many changes to the main Experience Mod calculation algorithm.

If one looks at the rating formula (an algorithm) in any of the rating bureaus, the number of state exceptions can be somewhat shocking.

Main Hidden Problem With Predictive Analytics and Modeling

The one area that I have not covered in any article is a very simple one that is passed over when discussing taking numbers, modeling them, then making predictions with them. Who decides what goes into the model and even more important – who writes the algorithms?

Until the day that software/computers can generate their own algorithms and models, bias (as Frank P. astutely pointed out) will always enter into the numbers.

I will not rehash the articles I have written previously on predictive analytics or workers comp algorithms. Feel free to follow the above links for the predictive modeling articles.

©J&L Risk Management Inc Copyright Notice

Workers Comp Predictive Analytics Changed Loss Run Reviews Forever

March 4, 2021 By JL Risk Management Consultants

Workers Comp Predictive Analytics – Do Not Ignore Old Open Claims

One of the big changes with loss run reviews started with the carriers and now the whole insurance industry not ignoring the old open claims. Workers Comp predictive analytics has changed how old open claims are evaluated in underwriting.

picture of workers comp claim analytics Aztec calendar

Public Use Wikimedia License – Keepscases

Years ago, I wrote numerous articles that pointed to only considering claims that were going to be fed into the Experience Mod system. I even said that predictive systems would not work in workers’ comp. That statement still stands. I will cover it in the next heading.

The scenario has changed so much in the last 10 years that I had to change my advice that I give to claims adjusters, employers, agents, and others.

Does this discount the Experience Mod system? No, it does not as the Mod is still the most common way to assess the risk of an employer.

Times are changing though, on how the Workers Comp insurance industry evaluates employers to underwrite them. One cannot ignore the Schedule Modification factors that carriers assign to employers.

Read up here on Schedule Mods as more carriers are using algorithms to calculate your risk factor in addition to looking at your Experience Mod. Schedule Mods can swing your premiums up to 50% in certain states.

Two Types of Workers Comp Predictive Analytics

The two areas of analytics as I view them are pre-claim analytics and post-claim analytics. Pre-claim analytics seem to work well. Post-claim analytics have not moved forward that far yet.

Pre-claim Analytics

Analytics on the front-end of the Workers Comp insurance process has become almost the norm in most underwriting departments. Carriers, Third Party Administrators, Captives, and other entities that provide coverage can now import the claims data, analyze it, and come up with a score that may vary somewhat from the Experience Mod system.

Many companies have sold and now sell analytics packages that have simplified the front-end rating processes on insureds.

Work flow of Workers Comp Predictive Analytics tools

Wikimedia Commons – Foxmetrics

Agencies are now more prone to have a set of algorithms that analyze current and potential clients for many types of loss ratios.

Inside of pre-claim analytics, the old claims count significantly more than with the Experience Mod rating systems used by the rating bureaus (NCCI, WCIRB, etc.).

Numerous old open claims on a loss run (more than five years old) may have little effect on your E-Mod/X-Mod. They will have a greater effect on pre-claim analytics. Closing out the old claims has become more important over time.

Check out this article on reopened claims causing problems if they are included in any analytic evaluations. Reopened claims sometimes have the same numbers reset on the claim, as when the claim was originally closed. These reopened claims can spike an E-Mod or any type of workers comp predictive analytics.

Post-Claim Analytics

Post-claim analytics, according to my opinion, attempt to predict claim values using various factors surrounding the claims. I still have not seen a package that can analyze the healing of the human body with a predicted return to work date or the amount of medical treatment required for a return to gainful employment.

Over my career, I have reviewed thousands of claims and loss runs. I have often seen this situation – a group of injured employees have the same:

Employer
Socioeconomic factors
Age, Sex, pre-injury health
Treating physician
Injury type
Other similar factors.

The healing and return to work period vary widely for a similar group as the human body does not heal the same. No algorithm that I have seen predicts the outcome of these claims.

I still want to be proven wrong. I have reviewed some of the systems that were good at pre-claim analytics including the level of safety provided to the employees as a predictive factor. I have not seen one to date taking post-claim data and using it for a model.

Old Claims Now More Critical – Workers Comp Predictive Analytics

Predictive Statistical Studies of Workers Comp Predictive Analytics Diagram

Wikimedia Commons – Norton, John Pease, 1877-

Workers Comp predictive analytics looks at more of a longer horizon than the usual. Even most actuarial analyses are giving more weight to older claims.

In 2010 – 2011 and before, I would usually ignore claims that were more than five years old. Why should I ask an employer or insurance agency to pay me to close files that made no difference to the Experience Mod or a self-insured Loss Development Factor?

Any claims that appear on loss runs, including old open claims, now require treatment as if they were open new claims. Workers Comp predictive analytics now give them much more weight when analyzing an insured.

©J&L Risk Management Inc Copyright Notice

Workers Comp Claim Analytics – Looking For Miracles Under Every Rock

February 22, 2019 By JL Risk Management Consultants

Workers Comp Claim Analytics – Hot Subject Leaves Me Cold

Most workers comp claim analytics posts I have written was in search of a good claim analytics package. Unfortunately, my search ended up like my search for workers comp apps – disappointed with still a hope for the future.

picture of man swimming workers comp claim analytics

Wikimedia- Oleg Dubyna

Company names have been left off my critiques. My basic point is for the claim analytics field for workers comp in total.

My review started with a claims analytics package in the 1990s. Yes, a few rehabilitation companies offered workers comp claims analytics software which was very limited in what it could produce for a claims or risk management department.

My review of these packages now involves consulting for a few carriers, TPA’s, and mainly self-insured/ large deductible employers. In the 1990’s I ran a very large governmental based TPA operation. The software used had a few modules that actually had a few good variables.

The variables had to be extrapolated a few times. However, the results were a very basic “yellow flag” on overcharges for certain pharmaceutical concerns. The PBM (Pharmaceutical Benefit Management) software covers that well. This, to me, was not real claim analytics. With an actuarial and statistical background, I hungered for more. I knew it was only a matter of time.

Now – 20 years later – the search for true workers comp claim analytics still leaves me wanting. The data is there, but one fault remains.

The psychological component and healing factors post-injury cannot be drawn on a graph or forecasted for any group of claims. Why? – No two people are alike with the butterfly effect driving the statistics.

I still today challenge any company that has claim analytics to please let me log in or send me a disc for offline software where I could at least have a few good red flags. If you see this article and have workers comp claim analytics software – please no beta versions. If your software creates results which are helpful to a claims or risk management department, I will laud your software.

I have sampled a ton of claims “processing” software with analytics as a backdrop at the NWCDC Conferences over the years. The claims processing end of most TPA/employer claims packages were beyond excellent.

The Best Analytics Right Now

The best analytics package right now sits at their desk adjusting claims for years. An experienced and talented claims adjuster usually can predict how a claim turns out long before claims development (5 – 7 years experience).

Why? – At five years – an adjuster has seen most claims situations. The adjuster has a claims load of familiar employers. At seven years in the claims business, most, not all, claims adjusters become “claims whisperers.” They have developed a gut instinct superior to any software of any type of claims reviewer.

One caveat – the adjuster must have performed a full (no loose ends) investigation into the claim within 48 hours after the injury. If this does not occur, the results will degrade from the start.

Secret/But Not So Secret Software

Hold on, one work comp claim analytics software does exist. I use it all the time. The software remains very reliable to date. The 1990’s claims processing package used this software successfully for data analytics. The answer is: (hold it, think about it)

Excel(R) or Open Office (R) <<Open Office is free for non-commercial operations. Yes, Excel (R) – thanks Bill Gates- remains one of the best workers comp claim analytics packages today. Why? – Because I can search and sort the data any way that I want to find those data outliers or trends.

I have used it as an Expert Witness on claims lawsuits. Make sure you have the statistics package downloaded into the software. I use Office 365 – so the statistics package is automatically included for my use. What statistics should you use? The most powerful one that I use the most is Regression for forecasts and Custom Sort. Custom Sort can be found in the main Excel ribbon.

Over the weekend, I will be sampling an online workers comp claims analytics package. Wish me luck!

©J&L Risk Management Inc Copyright Notice

Predictive Analytics Is Now Opioid Buzzword Replacement

May 18, 2018 By JL Risk Management Consultants

Predictive Analytics Replaces Opioid Buzzword

Predictive Analytics for workers compensation claims adjusting still has not proven itself as a cost saver. Please note that I am referring to workers compensation only.

picture wizard predictive analytics crystal ball

Public License – Wikimedia-The Wizard

Predictive analytics for other lines of casualty insurance such as auto and property damage has shown some promise over the years. However, workers compensation adjusting and reserve analytics have not reached the point as a viable option.

One of my longtime workers comp compatriots – Mark Walls of Safety National went as far as calling analytics a myth. That is very strong language, but I cannot find one analytics package, app, program, or cloud-based analytics that would allow me to challenge Mr. Walls’s assessment.

Mark and I agree down the line on many facets of workers compensation. With my actuarial background, I often give any type of statistics a wide berth to enhance the accuracy of workers compensation benefit management.

The bottom line is that I am still waiting for a great analytics package outside of an Excel(r) spreadsheet that would make me feel very comfortable with professing that “workers compensation claim and reserving analytics have arrived.” I still cannot even begin to utter that phrase. Why?

Two different random variables enter into the equation. Every injured employee has a different “healing window” of time to physically and mentally recover from an injury and return to the workplace. The other variable is the mental healing capacity of the injured worker. No two people look at a fractured arm in a cast with the same outlook.

So let us look at what I am talking about in a formula for the number geeks such as myself:

Known and predictable statistic (1) * known and predictable statistic (2) * known and predictable statistic (3) = accurate predictive analytic

but

Known and predictable statistic (1) * known and predictable statistic (2) * known and predictable statistic (3) * physical healing random variable * mental healing random variable = shot in the dark at best

Man With Predictive Analytics Neck Injured

Extended License Stockunlimited.com neck injury

As I have said, and Mark Walls says in his article – Show me the analytic. I will do an article at length on any software, etc. if I can be presented with one package that can predict claim outcomes.

That is why Reserving is an Art, not just a formulaic contrivance of a few numbers on a spreadsheet. As Mark mentioned in his article, good experienced adjusters still survive as the best analytic tool for foretelling how a claim will turn out. Hold on, though, please see the word good in the prior sentence.

Unfortunately, not all experienced adjusters are great at reserving. Then again, they can be the best claims handlers on earth using one technique. It is called The Golden Rule.

As with my search for a decent Workers Compensation app, I invite any person or company to see me a claims predictive analytics package that can predict workers compensation claims outcomes.

©J&L Risk Management Inc Copyright Notice

Insurance Trends vs. Paper Clips – Lets Get Ready To Analyze

December 21, 2016 By JL Risk Management Consultants

Insurance Trends vs. Paper Clips Analytics

The analysis of most insurance trends involves many innocuous spreadsheets that only gives the reader headaches. I was looking over an old article today concerning a NCPRIMA presentation from 2014.

StockUnlimited

BTW, we are upgrading every article on this website to make it Google searchable and rankable. That task has taken over 6 months as we have over 1,620 articles to upgrade. It is ongoing. Whew!

I digress – one of my old bosses always said do not analyze paper clips. You have people to do that for you. Analyze the big trends to see the direction your organization is heading (to and fro). Many of the groups to which he was presenting consisted of a large number of financial types (CFO, Financial Officer, etc.).

The financial types – I guess I am one – would squirm in their seats when he would say – Analyze trends not paper clips. Insurance trends are very similar.

I have pointed out many times in presentations and in the articles that Excel(R) is your best friend when analyzing trends. In fact Microsoft has provided Statistical Analysis packs for Excel. This is one of the best bargains in the analysis world.

Apache Open Office Insurance Trends Computer Screen

Wikipedia – Apache Foundation

If you are looking for a more economical (free) data analyses pack, you can try the OpenOffice by Apache installation that is a good analysis pack. The work you do must be personal and not business for the OpenOffice packs. You may want to throw a few dollars their way as the software is a treasure for non-business analysis.

Correction – According to Apache

“You may use it on home machines, in the office, with your small business, school, church, gardening club, etc.”

The ones I use are Regression, Stepwise Regression, Coefficient of Correlation, Z-test, T-test, F-test, Chi-Square, and Regression Graphs. I use regression often as I like to use more historical data than most insurance actuaries. Business cycles and insurance trends are 20 year – not 10 or less.

Google has long recognized long-tailed data for their keywords. Regression is a long-tailed analysis.

Analyzing insurance trends can be tedious at first. Once you set up a framework for analysis – depending on what you are analyzing – the task can become simple to a point.

©J&L Risk Management Inc Copyright Notice

Do Work Comp Claim Analytics Really Work As Promised

June 8, 2016 By JL Risk Management Consultants

Have Work Comp Claim Analytics Arrived?

A Work Comp Claim Analytics package is something that I have been in search of for many years. Can software actually replace the thinking process of an experienced adjuster?

Hand Illustrating Work Comp Claim Analytics Concept

StockUnlimited

Last year at the 2015 WCRI Conference I met a few analysts for one of the big insurance carriers. They had informed me the carrier was working on a database for “internal” work comp claim analytics. The package was not going to be sold to the public.

Last week, I noticed an article from a major third party administrator that is selling a claims analytics package. A few weeks ago I wrote about not being able to find any accurate analytics or predictive model packages.

Vector Graphic Of Work Comp Claim Analytics With Dollar Sign

StockUnlimited

Some of the comments on the article suggested that there are great analytics on the front end (policies, loss portfolio, underwriting) . That may be true but I have yet to see any worthy claim analytics package.

However, if a TPA or insurance carrier had enough data, one would have to think they would be able to at least see if an injured employee fit some type of profile such as drug dependent personality or if a person from a certain zip code would take longer to heal than the expected norms.

I have demoed so much software at the National Workers Comp Conferences without really finding an analytics program for claims. One reason is the limited amount of data that the vendors would have had – except for the large TPA’s and carriers.

I will eventually have to let go of the old school way of thinking that an experienced adjuster is the best way to analyze a workers comp file and its associated reserves. Once I see a package that works, I will keep quiet about work comp claim analytics. Until then, I will keep commenting and searching.

©J&L Risk Management Inc Copyright Notice

Work Comp Predictive Models vs Gut Feelings – The Winner Is(?)

May 2, 2016 By JL Risk Management Consultants

Work Comp Predictive Models

There are so many articles on Work Comp predictive models out in the blogosphere. Many taut their ability to see into the future. I am not a big fan of predictive models as I have yet to see and am still waiting on one that is consistent in nature.

Wikimedia – Khadra A

One of the larger carriers in the nation had a predictive algorithm for setting claim reserves. The reserves were off approximately 60% when I attempted a loss run review. The carrier has since quit using the algorithm.

Some types of insurance can be held up to a predictive model such as automobile or property loss. Workers comp predictive models attempt to predict how the human body heals as part of its forecasting.

The healing of the human body is such a random event. In that same light, I came across an article that seemed to say gut feelings is actually a learned predictive algorithm learned since birth. The article published in Inc magazine was an interesting take on how we all carry around a built in algorithm.

A great passage from the article – “the more we experience- the more accurate our gut instincts become.”

The great Cris Carter- (former Vikings receiver) said the players that practice more seem to have more luck.

Medical Algorithms Work Comp Predictive Models Chart

Wikimedia – National Heart

If you follow the link to the article, there is a great example of how Gary Player the golfer used gut instinct as part of his luck.

Workers comp adjusters develop this “gut instinct” after 5 – 7 years of experience on which claims to question; how much medical reserves to place on a file; which claims need another medical opinion; etc.

As I have often mentioned, if I happen to come across any Work Comp predictive models or algorithms that work, I will laud it heavily- but then again I am still waiting.

I will publish a similar article on how this all fits in with a business owner next time.

Predictive Model For Workers Comp Misunderstanding

November 12, 2013 By JL Risk Management Consultants

Prior WC Predictive Model Article Was Referring To Claim Reserves

Last Friday, I published a blog article on a predictive model in Workers Compensation. A few emails were received that indicated there were many great software products for predicting E-Mods and other variables in Workers Compensation.

Wikimedia Commons – Camelia.boban

In the original article, I was mainly referring to the fact I have yet to see any predictive model type program that could ever come close to forecasting the reserves for the life of a WC claim. We have reviewed a large number of WC files where a predictive model was used to set the reserves on a file.

The best analytic program model for worker’s compensation reserves is an experienced adjuster handling the same type of injury, employer, and state jurisdiction.

Those reserves had to always to overridden by the adjuster (usually with supervisory approval) at some later point in the file. The healing time of the human body is almost a random length between injured employees. Claims have too many almost random variables in them for claims analytics.

If Liberty Mutual ever shares their massive WC database, we may actually be able to have predictive modeling for claims.

My next article is on Statistics Madness.

Workers Compensation Predictive Modeling Will Never Work

November 8, 2013 By JL Risk Management Consultants

Workers Compensation Predictive Modeling

One of the more popular buzz phrases now is Workers Compensation predictive modeling. If one looks at any of the WC groups on LinkedIn, there are many discussions on this very matter.

Wikimedia – Jaberi-Douraki M

There are many predictive models on WC. Those models usually leave the observer with an empty feeling. Yes, you can forecast the Indemnity and Expenses almost to the penny. There are many modeling programs that will suffice in this case.

The inherent problem is valuing and forecasting medical benefits on any file. A health profile and background of the injured employee may shed some light on the subject. However, we are all very genetically unique individuals.

Woman Workers Compensation Predictive Modeling Carrying Files

StockUnlimited

A ruptured spinal disc can vary in treatment from two months to two years or longer.Predictive modeling requires some sort of stability in the variables. As any adjuster will tell you, it is hellish to try to predict medical treatment at the outset of the file. There are medical outcome determining factors that are not known for many months.Workers compensation claims adjusters walk a very fine line with medical forecasting. Sixty days into the file development, the adjuster is supposed to make a medical reserve determination for the life of the file. This is not possible. Reserving files is an art, not a science. We analyze many reserves on files on a daily basis. Reviewing reserves is a complete exercise of hindsight which is usually 20-20. Discussing what reserves were placed on a file one or two years ago is an easy task. The reverse of discussing what the reserves should be on a file in one or two years is difficult at best.Communicating with the claims adjusters about the status of an employee can reap large dividends, but as they happen, not three months later. Emails work best for claims departments.

Bottom line – a well-trained experienced adjuster is the best predictive modeler in Workers Compensation.

Possible 75% Savings on Workers Comp Premium From My Studies

September 20, 2012 By JL Risk Management Consultants

Possible 75% Savings – Study Results Below

It’s possible to experience a 75% Savings according to the results of my WC studies.

Yesterday, I blogged on predictive modeling in Workers Compensation. I had mentioned there were two studies that I performed in the recent past on a group of 7,000 and 5,000 claims respectively.

Picture of Jar Full Of Coins With Plant 1600% Savings of Workers Comp

The results of the studies were:

Predictive claims modeling relies on psychographics and the healing time of the human body. These two stats were so variable, no statistical conclusions could be drawn
Fraud in workers comp claims could not be predicted as claimant fraud was seen to be a random variable, there were no solid “red flags”
How the employer functioned post-injury did show some relationship to the cost of the claim. The correlation between the action of the employer and the cost of the claim was related

The keys to Workers Comp savings lies with the employer. I came up with the first three on the list over 20 years ago. They saved employers a large number of claim payments. The following is a list of the keys to Workers Comp savings proven by the two studies:

First Reports of Injury need to be filed ASAP. I used to understand delays due to paperwork flow, but that has virtually been eliminated as almost all insurance carriers and TPA’s have online filing of claims.
Treatment network needs to be in place. This does not mean having a PPO network from your bill review provider. Even if you are in a state that does not allow the employer to control the medical treatment, most employees with still go to your physician recommendation
(c) 123rf.com
Return to work program is very critical. The communication between your company and medical providers is very important. Medical providers are not just physicians. Physical therapists are also important providers. Providing a PT with your return to work program can save you a large amount of claims $.
Employee treatment has improved. When I first came up with this list over twenty years ago, many injured employees were treated as outsiders. This area has improved over the years as many employers have come to realize an injured employee is still an employee.
Top management consideration was the latest addition to the list. If an employer’s top management does not take their Workers Compensation seriously, the aforementioned four keys will never be taken that seriously. This is a 100% related statistic.

The first four keys will save an employer 400% each. That is a 75% total savings. They are for the most part all controllable by the employer. I have seen an improvement in those areas over the last decade, but there is still room for more attention to those areas.

Predictive Modeling For Workers Compensation – Data On Horizon?

September 19, 2012 By JL Risk Management Consultants

Workers Comp Predictive Modeling – Big Data Is The Key

The Workers Comp predictive modeling prognosticators say that predicting workers comp data is on the horizon. I was just reading over Mark Wall’s Work Comp Analysis Group in LinkedIn. If you are in Workers Comp and you do not follow the group, you are missing out on a number of great discussions.

Graphic of Silver Letter of Big Data on Top of Clouds Predictive Modeling Concept

123RF

There was a PHD’er that had a very-well written article on another Workers Comp blog concerning Workers Comp predictive models (PM). I chimed in and then decided to give Mark’s group some exposure – like it needs it with over 11,000 members.

I am for Workers Comp PM as much as anyone else in the Workers Comp business. However, as I mentioned in a posting to the group, property casualty claims (other than crop) have predictive models that are based on fixed variables. (property, auto, etc.)

Workers Compensation has to deal with an individual’s psychographic makeup along with the least predictive variable which is the healing of the human body. I have always noticed the variability of how people heal is nothing other than amazing. There are no two people alike in this respect.

Wikimedia Commons – Camelia.boban

Liberty Mutual is assembling the largest database in history. I hope they share some of their prognostications. Their research will likely be earth-shaking. I am not sure when their PM model will be finished to the point where they can publish any of the results. They may use it as a competitive advantage in the marketplace.

I have undertaken general research on a group of 7,000 and 5,000 files in the past. One group was of public employees while the other was random private companies. I will share what I found in my post tomorrow concerting PM. One hint of what I found was posted by me on Mark’s blog.