If they are going to use technology to underwrite, they need to get it right. I had a property risk at least 95 miles from Galveston Bay that was declined because their program said it was 15 miles from the bay. I tried to tell the underwriter that if you went 15 miles from the bay, you still wouldn’t be anywhere near Harris County. He said they have to go by the program. Apparently, their program can’t read a simple map! And human common sense isn’t allowed any longer!
Very good point. I’m old enough to remember the old Sanborn map books. Insurance companies actually had a pictorial exhibit of their underwriting exposures. At a glance, you knew the construction, age, occupancy etc of all surrounding properties. And, how many other policies the company had issued in the area.
Also, in the old days, you had agents with real underwriting / binding authority, special agents and state agents that had real decision making authority. And, it took years to earn the title of underwriter.
Accountants? That implies you pre-dated the rise of actuaries into senior management. The current environment is a natural progression of analytics from viewing Sanborn maps to IT registers of exposures fed into IBM computers via paper punch cards, and displayed as columnar data gathered and compiled by Accountants and Early Generation Actuaries. Then, more sophisticated analyses were enabled by Gates’ and Jobs’ personal computers. The migration of IT support to Asian countries were a minor advance, mainly due to cost savings.
Finally, starting around 2010-ish, we see the most sophisticated ‘underwriting’ (watch for my new term for this professional endeavor in a later posting today) by the 2nd (3rd?) generation of Actuaries. These new ‘underwriters’ use advanced stat models to generate models of claims, risk target marketing by agents of the most profitable exposures, and for property catastrophe models that do a much better analysis than visually viewing a Sanborn map. We have arrived at a point in time where computers can replace underwriters, and some agents of homogeneous exposure lines (e.g. personal auto), as I have projected several times in the past.
The predictive modelers at State Farm managed to underwrite their way to a $7 billion loss on Personal Auto alone in 2017. Great job guys. Led to closing several offices and reducing employee count by 4,200.
And they still made about $400 million, after making between $3.5-6.5 billion the previous year. You don’t know the industry, you are a dinosaur.
March 26, 2018 at 5:15 pm
Agent says:
Like or Dislike:
3
4
UW, I know the industry far better than you do. Predictive Modeling has so many holes, it is a joke. Underwriters (not you) used to use much easier and better techniques to underwrite risk. This new underwriting is not good and does not bode well for the future. Why do you still have a job? The robots will be doing it soon.
March 27, 2018 at 7:53 am
Ron says:
Like or Dislike:
3
2
Agent,
Ever notice how I never makes comments on how to run an agency? You should stop commenting on underwriting and/or pricing of insurance for the same reason, lack of knowledge and experience.
Predictive Modeling was not designed to make sure companies are profitable every year. They are designed to reduce the volatility of the underwriting cycle. This means less work for you moving business from one carrier to another during changes in the cycle.
March 27, 2018 at 8:17 am
StorMonica DanieLewinskis says:
Like or Dislike:
1
4
@Ron; you fail to see the FULL potential of predictive modeling per GLMs, non-GLMs, and Machine Learning. Read something other than IJ and its politically slanted articles and learn something new about the devolution of underwriters into glorified marketing agents. They are gradually being replaced by Actuaries and CPAs atop insurance organizations and within underwriting departments. Many of their analytic duties are being usurped by Actuaries, CPAs, and Data Scientists.
March 28, 2018 at 10:47 am
UW says:
Like or Dislike:
2
2
Like I said Ron, he doesn’t know the industry. At best he knows selling policies in a small town. He’s shown over and over he does not understand on even a superficial level, risk, probability, or anything that makes up actual insurance and risk management. It’s a disgrace the site is basically built around people like him.
March 28, 2018 at 2:42 pm
StorMonica DanieLewinskis says:
Like or Dislike:
3
1
@UW; you are providing proof that the article has merit.
March 29, 2018 at 9:06 am
UW says:
Like or Dislike:
1
2
Durrrrrr, idiot.
March 29, 2018 at 5:39 pm
DNCs Coll(F)usion GPShip Strzok an IceberGowdy says:
Like or Dislike:
0
1
@Agent; I think SF Agents/ UWers varied from what the Predictive Analytics indicated what they should do. I can’t understand how their large datasets and the advanced PA techniques could fail so badly unless the UWers ignored the indications to use their subjective judgement.
UWers listen to, but occasionally ignore, what actuaries tell them, so why should anyone expect them to follow GLMs’ rates by class/ territory/demo group when they come from Actuaries, CPAs, and/ or Data Scientists?
March 26, 2018 at 11:35 am
SWFL Agent says:
Like or Dislike:
2
2
You could also argue that these agents with “real underwriting/binding authority” helped pave the way for the demise of companies like USF&G, Aetna, and others. “In the old days”, how many PL underwriters buckled under the pressure to make an exception when an agent called and whined about how good or big an account is or why the company shouldn’t non-renew the PL because of the commercial account.
I have found that good information with Loss Runs, pictures, good description of the account, pictures went a long way to get an account written if it was in the appetite guide of the carrier on a good Commercial account.
Pictures are pretty good but pre-inspections are better. Preferably, by the carrier’s own loss control team. No better way to vet a prospect.
March 29, 2018 at 5:41 pm
DNCs Coll(F)usion GPShip Strzok an IceberGowdy says:
Like or Dislike:
1
1
Often better; review their financial reports.
April 2, 2018 at 12:45 pm
Agent says:
Like or Dislike:
1
0
Companies that have failed have no one to blame but their management or lack thereof. Underwriters like UW will be a thing of the past soon. Poor choices, lack of understanding of the industry will do the trick every single time.
Planet, I do my own pre-inspections on Commercial Accounts. I have not been wrong about an account yet. I did have an account that was pre-inspected for WC and they recommended that new stripes be painted on the loading dock. How good was that to determine whether it should be written or not?
Agent, agents think they do pre-inspections but aren’t even close to being as qualified as a true loss control expert. Can you run a mock OSHA inspection? Can you provide a true ergonomic assessment? Are you a certified crop adviser? Are you CETP certified?
MARCH 26, 2018 AT 5:15 PM
Agent says:
LIKE OR DISLIKE:
1
1
Why do you still have a job? The robots will be doing it soon.
Yeah, it’s pretty laughable to think Agent can do inspections or that most insurers would put much weight on an agent’s inspection. He doesn’t understand the industry in any way, shape, or form. Unreal.
April 2, 2018 at 11:41 am
Agent says:
Like or Dislike:
2
0
Experience in writing commercial is what I rely on. Analyzing Loss Runs, Audits have proven to be very reliable, particularly when the account had not been given Loss Control advice in prior years. Companies are notorious for not doing a good job on Loss Control. They often farm it out to Independent people who really don’t know what they are doing.
March 29, 2018 at 5:44 pm
DNCs Coll(F)usion GPShip Strzok an IceberGowdy says:
Like or Dislike:
1
1
Who but OSHA does OSHA inspections? Why do a mock OSHA inspection when OSHA is available? Cost is a reason. What size accounts are you considering in your list for Agent to do? Renewals are a different situation, also.
DNC’s, I have often made submissions with excellent information and had underwriters ask me questions already answered on the applications and information. They are very stupid to keep asking the same questions over and over.
April 3, 2018 at 12:41 pm
??? says:
Like or Dislike:
0
0
Tons of companies have loss prevention experts come out and provide mock OSHA inspections! The company will find what you’re doing wrong and assist you in fixing it, instead of penalizing your for it.
Do me a favor, call all your insureds and tell them you have a great loss prevention measure for them. then call OSHA and request they come to their site for an inspection.
See how that is received.
April 3, 2018 at 1:48 pm
Captain Planet says:
Like or Dislike:
0
0
We get asked all the time to provide mock OSHA inspections. It’s a great way for them to be prepared for the real deal and avoid fines. Not to mention, enhance their safety efforts along the way. Account size can vary. I’ve seen it performed on $40,000 in premium all the way to $4M in premium. We are performing these on our existing book so our renewal insureds. We don’t look at our insureds as 12 month relationships, we look at them as a minimum of 3 years and hopefully longer than that. I will say, our acquisition costs are high on the new business side, too. Though, we’d much rather have our own risk control folks giving us an unbiased opinion than take an agent’s word for it. Trust but verify, right?
April 3, 2018 at 12:38 pm
??? says:
Like or Dislike:
0
0
You do your own pre-inspections? LMAO!
Yea Me2! “Hey, UW this is a great account, they pay a ton in premium”. “Oh yea, how are their loss controls?”. “Perfect! I looked them over myself, not quote this and give me exactly the price I need. Don’t bother inspecting, it’s just how i want it.”
Yup – I remember it! The tag line then was something like: “Underwriting is dead – all that is left is classification and pricing”. Read this in the 80’s. Data modeling and predictive analytics are a lot less fun than the old way we used to select and price risk – but I hate to say it – the data stuff usually works better (especially in standard Life and Auto lines). All part of the natural progression of things as data and analytics improve – as does AI.
There are a few points I would raise to this blog.
– The application of psychotropic behavior in risk has not been missed by the industry but is under study. The technology that enables this study is available but relatively nascent.
– Regulation
– Applying behavioral profiles to individuals sounds like a basket of legal challenges unless it is done with care and fairness.
– Insurance segmentation has a second edge; too much and you price yourself out of the target segment. (Developing other services for sale to mitigate risk is probably the right compliment.
This is a “Moneyball” concept applied to insurance. For years, choices were made in underwriting based on qualities everybody was taught. The predictive modeling just like baseball analytics, allows the user to reconsider old assumptions. With insurance, the constraints will be regulatory (Prop 103 in California) and social ( is it OK to apply certain characteristics to underwrite and rate a risk?). The other challenge will be to get enough data to properly break down certain risks.
“…without randomness; Insurance is not insurance.” ?? Say what? Insurance transfers pure risk. I think most chief underwriting officers, and maybe agents collecting contingencies based on loss ratios would beg to differ. This industry is still about covering claims that a carrier is betting NOT to happen. Further, simply because the independent agency model has increased competition from direct writers, does not mean that technology governs the industry. Who is taking service calls? Who is taking claim calls? Do computers adjust wind claims, or recommend adequate BI/EE coverage for a machine shop with fluctuating sales?
I don’t think so. Disagree with the premise of this article.
machines taking over the world and no more business decisions to be made by humans…good luck working with machines…..glad i won’t have to worry about that crap
Who remembers when the “a” rates were “a” rated? That’s when you needed 10-15 years of solid underwriting experience to land a new job as an underwriter. Now if it’s not in the “approved” manual, they tell you “We’ll take a pass.”
Remembering, correct. The appetite guide rules and if a risk does not fill the boxes just right, they say we will take a pass. The ability of an underwriter to make a judgment of accepting risk is gone. Most good underwriters have been run off or retired and replaced by modelers who have no idea.
Everything you write sounds like a broker that isn’t valued or respected by the underwriters they work with. At my company and previous companies I’ve been at, if 1 box isn’t checked, and it’s from basically anybody but a broker we don’t want to work with anymore or don’t care if they leave, we would ask for clarification or for another document that can give us a better idea of the risk.
8 hate to break it to you, but you might be viewed as a pain in the ass to the underwriters, and unless it’s perfect they might just prefer not working with you. “Not in our appetite” is easier than, “stop wasting our time, I can’t deal with your crap.”
Seems the author and the editor are both math illiterate. In the 3rd paragraph it is said that 15% is 300% greater than 5%. Wrong. 10% is 100% greater than 5% and 15% is 200% greater than 5%. A common but silly error in reporting.
Not here, but before the editors have appeared to be “math illiterate,” aka innumerate.
They said that an average 25 yr is has a 5% chance of dying by age 60. A 15% chance would be a 200% increase from 5%, and 300% higher.
Increase:(15-5)/5 = 2.0
2*100=200% increase
Higher:15/5=3.0
3.0*100=300%
More importantly as it relates to risk, there would be no reason to compare a second person’s probability of an event happening as in increase from a first, separate person’s, probability of an event happening. If it was 1 person and they discovered a new condition then it would make sense to increase their personal probability and view it as a percentage increase.
For the past 20 years I’ve watched some of the world’s top P&C carriers make a complete mess of things using data analytics.
Data models have a place but they cannot successfully replace (can only enhance) underwriting experience, instinct, and customer relationships.
Underwriting is first and foremost about building strong relationships with customers – agents and insured’s – aimed at effectively managing business risk and providing peace of mind for business leaders.
Underwriting is therefore first and foremost a human endeavor.
Going forward, companies that will be most successful at underwriting will be those that understand the human component while using data analytics to support underwriting quality.
If they are going to use technology to underwrite, they need to get it right. I had a property risk at least 95 miles from Galveston Bay that was declined because their program said it was 15 miles from the bay. I tried to tell the underwriter that if you went 15 miles from the bay, you still wouldn’t be anywhere near Harris County. He said they have to go by the program. Apparently, their program can’t read a simple map! And human common sense isn’t allowed any longer!
Very good point. I’m old enough to remember the old Sanborn map books. Insurance companies actually had a pictorial exhibit of their underwriting exposures. At a glance, you knew the construction, age, occupancy etc of all surrounding properties. And, how many other policies the company had issued in the area.
Also, in the old days, you had agents with real underwriting / binding authority, special agents and state agents that had real decision making authority. And, it took years to earn the title of underwriter.
But then the accountants took over the companies.
Accountants? That implies you pre-dated the rise of actuaries into senior management. The current environment is a natural progression of analytics from viewing Sanborn maps to IT registers of exposures fed into IBM computers via paper punch cards, and displayed as columnar data gathered and compiled by Accountants and Early Generation Actuaries. Then, more sophisticated analyses were enabled by Gates’ and Jobs’ personal computers. The migration of IT support to Asian countries were a minor advance, mainly due to cost savings.
Finally, starting around 2010-ish, we see the most sophisticated ‘underwriting’ (watch for my new term for this professional endeavor in a later posting today) by the 2nd (3rd?) generation of Actuaries. These new ‘underwriters’ use advanced stat models to generate models of claims, risk target marketing by agents of the most profitable exposures, and for property catastrophe models that do a much better analysis than visually viewing a Sanborn map. We have arrived at a point in time where computers can replace underwriters, and some agents of homogeneous exposure lines (e.g. personal auto), as I have projected several times in the past.
The predictive modelers at State Farm managed to underwrite their way to a $7 billion loss on Personal Auto alone in 2017. Great job guys. Led to closing several offices and reducing employee count by 4,200.
And they still made about $400 million, after making between $3.5-6.5 billion the previous year. You don’t know the industry, you are a dinosaur.
UW, I know the industry far better than you do. Predictive Modeling has so many holes, it is a joke. Underwriters (not you) used to use much easier and better techniques to underwrite risk. This new underwriting is not good and does not bode well for the future. Why do you still have a job? The robots will be doing it soon.
Agent,
Ever notice how I never makes comments on how to run an agency? You should stop commenting on underwriting and/or pricing of insurance for the same reason, lack of knowledge and experience.
Predictive Modeling was not designed to make sure companies are profitable every year. They are designed to reduce the volatility of the underwriting cycle. This means less work for you moving business from one carrier to another during changes in the cycle.
@Ron; you fail to see the FULL potential of predictive modeling per GLMs, non-GLMs, and Machine Learning. Read something other than IJ and its politically slanted articles and learn something new about the devolution of underwriters into glorified marketing agents. They are gradually being replaced by Actuaries and CPAs atop insurance organizations and within underwriting departments. Many of their analytic duties are being usurped by Actuaries, CPAs, and Data Scientists.
Like I said Ron, he doesn’t know the industry. At best he knows selling policies in a small town. He’s shown over and over he does not understand on even a superficial level, risk, probability, or anything that makes up actual insurance and risk management. It’s a disgrace the site is basically built around people like him.
@UW; you are providing proof that the article has merit.
Durrrrrr, idiot.
@Agent; I think SF Agents/ UWers varied from what the Predictive Analytics indicated what they should do. I can’t understand how their large datasets and the advanced PA techniques could fail so badly unless the UWers ignored the indications to use their subjective judgement.
UWers listen to, but occasionally ignore, what actuaries tell them, so why should anyone expect them to follow GLMs’ rates by class/ territory/demo group when they come from Actuaries, CPAs, and/ or Data Scientists?
You could also argue that these agents with “real underwriting/binding authority” helped pave the way for the demise of companies like USF&G, Aetna, and others. “In the old days”, how many PL underwriters buckled under the pressure to make an exception when an agent called and whined about how good or big an account is or why the company shouldn’t non-renew the PL because of the commercial account.
I have found that good information with Loss Runs, pictures, good description of the account, pictures went a long way to get an account written if it was in the appetite guide of the carrier on a good Commercial account.
Pictures are pretty good but pre-inspections are better. Preferably, by the carrier’s own loss control team. No better way to vet a prospect.
Often better; review their financial reports.
Companies that have failed have no one to blame but their management or lack thereof. Underwriters like UW will be a thing of the past soon. Poor choices, lack of understanding of the industry will do the trick every single time.
More than likely the program is saying they are 15 miles from tidal water, not necessarily the bay itself.
Planet, I do my own pre-inspections on Commercial Accounts. I have not been wrong about an account yet. I did have an account that was pre-inspected for WC and they recommended that new stripes be painted on the loading dock. How good was that to determine whether it should be written or not?
Agent, agents think they do pre-inspections but aren’t even close to being as qualified as a true loss control expert. Can you run a mock OSHA inspection? Can you provide a true ergonomic assessment? Are you a certified crop adviser? Are you CETP certified?
MARCH 26, 2018 AT 5:15 PM
Agent says:
LIKE OR DISLIKE:
1
1
Why do you still have a job? The robots will be doing it soon.
Yeah, it’s pretty laughable to think Agent can do inspections or that most insurers would put much weight on an agent’s inspection. He doesn’t understand the industry in any way, shape, or form. Unreal.
Experience in writing commercial is what I rely on. Analyzing Loss Runs, Audits have proven to be very reliable, particularly when the account had not been given Loss Control advice in prior years. Companies are notorious for not doing a good job on Loss Control. They often farm it out to Independent people who really don’t know what they are doing.
Who but OSHA does OSHA inspections? Why do a mock OSHA inspection when OSHA is available? Cost is a reason. What size accounts are you considering in your list for Agent to do? Renewals are a different situation, also.
DNC’s, I have often made submissions with excellent information and had underwriters ask me questions already answered on the applications and information. They are very stupid to keep asking the same questions over and over.
Tons of companies have loss prevention experts come out and provide mock OSHA inspections! The company will find what you’re doing wrong and assist you in fixing it, instead of penalizing your for it.
Do me a favor, call all your insureds and tell them you have a great loss prevention measure for them. then call OSHA and request they come to their site for an inspection.
See how that is received.
We get asked all the time to provide mock OSHA inspections. It’s a great way for them to be prepared for the real deal and avoid fines. Not to mention, enhance their safety efforts along the way. Account size can vary. I’ve seen it performed on $40,000 in premium all the way to $4M in premium. We are performing these on our existing book so our renewal insureds. We don’t look at our insureds as 12 month relationships, we look at them as a minimum of 3 years and hopefully longer than that. I will say, our acquisition costs are high on the new business side, too. Though, we’d much rather have our own risk control folks giving us an unbiased opinion than take an agent’s word for it. Trust but verify, right?
You do your own pre-inspections? LMAO!
Yea Me2! “Hey, UW this is a great account, they pay a ton in premium”. “Oh yea, how are their loss controls?”. “Perfect! I looked them over myself, not quote this and give me exactly the price I need. Don’t bother inspecting, it’s just how i want it.”
Hmmmmm. Seems I saw this article 28 years ago:
http://www.richardwdorman.com/pdf/underwriting-is-dead.pdf
I’m sure this article will be published again in another 75 years when the computers are writing the article for us. Some things never change!
Yup – I remember it! The tag line then was something like: “Underwriting is dead – all that is left is classification and pricing”. Read this in the 80’s. Data modeling and predictive analytics are a lot less fun than the old way we used to select and price risk – but I hate to say it – the data stuff usually works better (especially in standard Life and Auto lines). All part of the natural progression of things as data and analytics improve – as does AI.
332 year old industry – Lloyd’s started in 1686.
There are a few points I would raise to this blog.
– The application of psychotropic behavior in risk has not been missed by the industry but is under study. The technology that enables this study is available but relatively nascent.
– Regulation
– Applying behavioral profiles to individuals sounds like a basket of legal challenges unless it is done with care and fairness.
– Insurance segmentation has a second edge; too much and you price yourself out of the target segment. (Developing other services for sale to mitigate risk is probably the right compliment.
This is a “Moneyball” concept applied to insurance. For years, choices were made in underwriting based on qualities everybody was taught. The predictive modeling just like baseball analytics, allows the user to reconsider old assumptions. With insurance, the constraints will be regulatory (Prop 103 in California) and social ( is it OK to apply certain characteristics to underwrite and rate a risk?). The other challenge will be to get enough data to properly break down certain risks.
“…without randomness; Insurance is not insurance.” ?? Say what? Insurance transfers pure risk. I think most chief underwriting officers, and maybe agents collecting contingencies based on loss ratios would beg to differ. This industry is still about covering claims that a carrier is betting NOT to happen. Further, simply because the independent agency model has increased competition from direct writers, does not mean that technology governs the industry. Who is taking service calls? Who is taking claim calls? Do computers adjust wind claims, or recommend adequate BI/EE coverage for a machine shop with fluctuating sales?
I don’t think so. Disagree with the premise of this article.
machines taking over the world and no more business decisions to be made by humans…good luck working with machines…..glad i won’t have to worry about that crap
Who remembers when the “a” rates were “a” rated? That’s when you needed 10-15 years of solid underwriting experience to land a new job as an underwriter. Now if it’s not in the “approved” manual, they tell you “We’ll take a pass.”
Remembering, correct. The appetite guide rules and if a risk does not fill the boxes just right, they say we will take a pass. The ability of an underwriter to make a judgment of accepting risk is gone. Most good underwriters have been run off or retired and replaced by modelers who have no idea.
Everything you write sounds like a broker that isn’t valued or respected by the underwriters they work with. At my company and previous companies I’ve been at, if 1 box isn’t checked, and it’s from basically anybody but a broker we don’t want to work with anymore or don’t care if they leave, we would ask for clarification or for another document that can give us a better idea of the risk.
8 hate to break it to you, but you might be viewed as a pain in the ass to the underwriters, and unless it’s perfect they might just prefer not working with you. “Not in our appetite” is easier than, “stop wasting our time, I can’t deal with your crap.”
Thank you for the insightful post, UW.
Seems the author and the editor are both math illiterate. In the 3rd paragraph it is said that 15% is 300% greater than 5%. Wrong. 10% is 100% greater than 5% and 15% is 200% greater than 5%. A common but silly error in reporting.
Not here, but before the editors have appeared to be “math illiterate,” aka innumerate.
They said that an average 25 yr is has a 5% chance of dying by age 60. A 15% chance would be a 200% increase from 5%, and 300% higher.
Increase:(15-5)/5 = 2.0
2*100=200% increase
Higher:15/5=3.0
3.0*100=300%
More importantly as it relates to risk, there would be no reason to compare a second person’s probability of an event happening as in increase from a first, separate person’s, probability of an event happening. If it was 1 person and they discovered a new condition then it would make sense to increase their personal probability and view it as a percentage increase.
For the past 20 years I’ve watched some of the world’s top P&C carriers make a complete mess of things using data analytics.
Data models have a place but they cannot successfully replace (can only enhance) underwriting experience, instinct, and customer relationships.
Underwriting is first and foremost about building strong relationships with customers – agents and insured’s – aimed at effectively managing business risk and providing peace of mind for business leaders.
Underwriting is therefore first and foremost a human endeavor.
Going forward, companies that will be most successful at underwriting will be those that understand the human component while using data analytics to support underwriting quality.