A rating engine is the software that calculates an insurance premium by applying an insurer's rating plan (its base rates, rating factors, discounts, fees and minimum premiums) to the details of a risk, and then adding tax. It runs inside a policy administration system or as a separate service that other systems call through an API.
This entry explains what a rating engine calculates and in what order, with an illustrative worked example, where the rates come from, the rating factors used by line of business, the UK rules that restrict renewal pricing and the use of protected characteristics, and how US rate filing differs from the UK approach.
What a rating engine does
Actuarial texts describe the information needed to price a risk as four parts. The Casualty Actuarial Society's textbook Basic Ratemaking lists rules, rate pages (base rates, rating tables and fees), the rating algorithm and underwriting guidelines, and notes that the first three are generally found in an insurer's rating manual while underwriting guidelines are kept in a separate proprietary manual.1 A rating engine holds the rate pages and the algorithm as data and code, and applies the rules to classify each risk before it calculates.
| Part | What it contains | Examples |
|---|---|---|
| Rules | Qualitative information needed to apply the algorithm, including how to classify a risk | Definitions such as what counts as a primary residence, policy forms offered, minimum premium, down payments, refunds on cancellation |
| Rate pages | The numbers the calculation uses | Base rates, rating tables of relativities, fees |
| Rating algorithm | How the components combine into a premium | The order of rating variables, whether each is multiplicative or additive, maximum and minimum premiums or discounts, rounding |
| Underwriting guidelines | Company-specific criteria that act before or within the calculation | Accept, decline or refer decisions, placement in a company or rating tier within a group, schedule rating credits and debits |
The textbook explains that before computers were used for quoting, insurers typically calculated the rates for the most common combinations of rating variables to save agents and brokers calculation, and that as algorithms grew more complex and computers more common, rate pages came instead to hold the building blocks from which rates are calculated.1 It also distinguishes two ways of covering fixed costs: an explicit expense fee in the algorithm, or a provision within the base rate, in which case an insurer may set a minimum premium to make sure the premium covers its expenses.1
How a rating engine turns a risk into a quoted premium
A worked rating calculation
The table below rates two home insurance policies through the same simplified multiplicative algorithm. Every number in it is invented to show the mechanics, and the factors do not come from any insurer's rating plan.
| Step | Policy A factor | Policy A running total | Policy B factor | Policy B running total |
|---|---|---|---|---|
| Base rate for the cover | £300.00 | £300.00 | ||
| Territory relativity | × 1.15 | £345.00 | × 0.70 | £210.00 |
| Construction type relativity | × 0.95 | £327.75 | × 0.85 | £178.50 |
| Sum insured relativity | × 1.20 | £393.30 | × 0.60 | £107.10 |
| Claims-free discount of 10% | × 0.90 | £353.97 | × 0.90 | £96.39 |
| Multi-policy discount of 5% | × 0.95 | £336.27 | × 0.95 | £91.57 |
| Policy fee | + £25.00 | £361.27 | + £25.00 | £116.57 |
| Minimum premium of £150.00 | Not applied | £361.27 | Applied | £150.00 |
| Insurance premium tax at 12% | + £43.35 | £404.62 | + £18.00 | £168.00 |
Three features of the example are general. First, a multiplicative structure means each factor scales everything before it, so Policy B's low territory, construction and sum insured relativities compound to a premium below the minimum, and the minimum premium then sets the price. Second, the way discounts combine is a design choice written into the algorithm. Applied one after the other, as here, a 10% and a 5% discount give a combined reduction of 14.5%; added together before they are applied, they give 15%, which would have rated Policy A at £334.31 before the fee (£334.305 rounded half up) rather than £336.27. Third, rounding after each step can change the final penny, which is why an algorithm specifies where rounding takes place.
The 12% used in the example is the standard rate of insurance premium tax, and a higher rate of 20% applies to some insurance sold with goods and services such as motor cars and domestic appliances, and to travel insurance.2 The algorithm in the example follows the pattern of the illustrative homeowners algorithm in Basic Ratemaking, which multiplies a base rate by relativities for amount of insurance, territory, protection class and construction, underwriting tier and deductible, applies new home, claims-free and multi-policy discounts, then adds charges for optional extra cover and a policy fee.1 In that example the claims-free and new home discounts are added together inside one bracket, and the multi-policy discount is applied separately.1
Rating factors by line of business
The characteristics that move a premium are called rating variables, and their effects are held in rating tables as relativities, factors or multipliers when applied multiplicatively, or as addends when added.1 The number and nature of rating variables varies significantly by line of business and from insurer to insurer.1
| Type of insurance | Typical rating variables |
|---|---|
| Personal motor | Driver age and gender, model year, accident history (in the UK, sex can no longer be used in this way: see below) |
| Home | Amount of insurance, age of home, construction type |
| Workers compensation | Occupation class code |
| Commercial general liability | Classification, territory, limit of liability |
| Medical malpractice | Specialty, territory, limit of liability |
| Commercial motor | Driver class, territory, limit of liability |
Underwriting guidelines sit beside the rating variables. The CAS textbook describes guidelines that can decide whether a risk is accepted, declined or referred to a senior underwriter, place a risk in one of several companies within a group, assign it to a rating tier, or set schedule rating credits and debits.1 The textbook notes a trend, especially in personal lines, to designate new explanatory variables as underwriting criteria used to place risks in rating tiers or separate companies, so that the line between rating and underwriting characteristics has become blurred.1
Where the rates come from
A rating engine applies rates; it does not decide what they should be. Rates come from pricing models built by actuaries and analysts. The CAS Statement of Principles Regarding Property and Casualty Insurance Ratemaking, as quoted by the NAIC, sets out four principles, the first being that a rate is an estimate of the expected value of future costs, and the fourth that a rate is reasonable and not excessive, inadequate or unfairly discriminatory if it is an actuarially sound estimate of the expected value of all future costs associated with an individual risk transfer.3
A common modelling tool is the generalised linear model (GLM). The CAS monograph on GLMs for insurance rating explains that when a GLM uses a log link, it produces a multiplicative rating structure: each term becomes a multiplicative factor (the exponential of its fitted coefficient), and the factors multiply together, which is the same shape as the calculation in the worked example.4 The NAIC's white paper on predictive models states that before GLMs, rating plans were built using univariate methods, that many insurers now consider those too simplistic because they ignore dependencies between variables, and that the majority of predictive models used in US personal motor and home rating plans are GLMs.5
Pricing can go beyond the cost of the risk. The FCA's market study of home and motor insurance found that firms optimise their margins, defined as the amount of the price charged above or below the cost of underwriting the risk and serving the policy, and that insurers and intermediaries with delegated underwriting authority typically use different models for risk pricing and for margin optimisation.6
| Model | What it estimates, according to the FCA |
|---|---|
| Risk pricing | The cost of underwriting the risk and serving the policy, against which the margin is measured |
| Conversion model | Expected sales of the core policy and ancillary products compared with the number of new business quotes, and how price changes affect conversion |
| Retention model | Expected sales compared with the number of renewal quotes, and the expected effect of price changes on retention, a core input to the price firms wish to charge |
| Ancillary product model | Income from add-ons or premium finance sold alongside the core policy, which firms may incorporate into pricing decisions |
| Lifetime value model | The expected long-term profitability of a customer; most firms in the FCA's sample used lifetime value and propensity models |
Rate adequacy is also a prudential concern. In its 2026 priorities letter the PRA observed a continuing softening of the underwriting cycle in many lines of general insurance, which may place pressure on pricing, terms and conditions and reserving levels, and said it expects boards to ensure they are pricing and reserving adequately.7
Price walking and the UK renewal pricing rules
The FCA's market study found that most firms in its sample operated a price walking strategy: new customers received low prices, sometimes below cost, and firms recovered the discount by raising the margin, and so the price, at renewal.6 It found that 6 million policyholders paid high prices in 2018 and would have saved £1.2bn if they had paid the average for their risk, and that 10 million home and motor policies were held by people who had been with their provider for 5 years or more.6
| Product | New customers | Customers of more than 5 years |
|---|---|---|
| Motor | £285 | £370 |
| Buildings | £130 | £238 |
| Combined buildings and contents | £165 | £287 |
| Contents only | £56 | £138 |
The FCA's response, set out in policy statement PS21/5, was a pricing remedy requiring that a renewal price be no greater than the equivalent new business price (ENBP) the firm would offer a new customer.8 The rules on pricing, auto-renewal and reporting came into effect on 1 January 2022.8 The pricing rules are in chapter 6B of the Insurance Conduct of Business Sourcebook (ICOBS) and apply to home and motor insurance and related additional products sold to consumers.9
| Provision | Type | What it says |
|---|---|---|
| ICOBS 6B.1.59 | G | The rules are not intended to affect how risk is priced for home and motor insurance |
| ICOBS 6B.2.110 | R | A firm must not set a renewal price higher than the ENBP, applied at the point the renewal notice is prepared |
| ICOBS 6B.2.210 | R | For a combined home and motor package, the home element, the motor element and the bundled price must each be no higher than the ENBP |
| ICOBS 6B.2.510 | R | The ENBP assumes the customer approached the firm through the channel they used when they first bought the policy |
| ICOBS 6B.2.910 | R | The ENBP must include cash or cash-equivalent incentives given to new customers that the renewing customer would be eligible for |
| ICOBS 6B.2.3910 | R | A firm must not systematically discriminate against customers based on their tenure when determining the ENBP |
| ICOBS 6B.2.4010 | E | A pricing model used to determine the ENBP should not generate prices that are systematically higher the longer a customer's tenure |
To apply the rule, a firm needs two prices for each renewing customer: the renewal price, and the ENBP for the same risk, channel and incentives, which the renewal price may not exceed. Guidance at ICOBS 6B.2.3 states that the rule applies to net-rated business as well as gross-rated, so insurers or managing agents writing net-rated business arrive at an ENBP on a net-rated price basis.10 For non-investment insurance products generally, PROD 4.2.14AR requires a firm's product approval process to identify whether the product provides fair value to customers in the target market, including whether it will continue to do so for a reasonably foreseeable period following renewal, and PROD 4.2.14BR applies the same test to each component and to the package as a whole where the product is sold with additional products.11
What the FCA's evaluation found
The FCA evaluated the remedies in Evaluation Paper 25/2, using its insurance pricing data for 2019 to 2024.12
The FCA's evaluation of its general insurance pricing remedies
The FCA found the remedies statistically significantly associated with a decrease in motor prices, and attributed the width of its range to changes in economic conditions such as inflation during the period.12 It did not find a statistically significant relationship between the remedies and home prices, and said this should not be read as evidence that home customers were worse off, only that it could not establish a causal link.12 In motor, where renewing customers pose significantly lower risks on average, the FCA expected new customers to pay more than existing ones once price walking ended, and found the gap had widened from new customers paying £20.76 more to £109.19 more.12
| Measure | 2019 to 2021 | 2022 to the first quarter of 2024 |
|---|---|---|
| Motor, new business customers | £458.24 | £569.38 |
| Motor, existing policyholders | £437.48 | £460.19 |
| Home, new business customers | £176.64 | £222.36 |
| Home, existing policyholders | £272.02 | £271.53 |
| Home, expected cost of claims for the core policy | £92.26 | £137.51 |
The FCA noted that the 49% rise in the expected cost of claims for home policies may partly reflect insurers' expectations of higher claim frequency or severity, and is also likely to be influenced by inflation, particularly in the cost of repair materials.12
Protected characteristics as rating factors
UK equality law limits which characteristics a rating plan may use. Part 5 of Schedule 3 to the Equality Act 2010 sets out when doing something in connection with insurance and other financial services is not a contravention of section 29, the provision on services.13
| Characteristic | Position | Provision |
|---|---|---|
| Sex | The insurance exception in paragraph 22, which implemented the Gender Directive's permission for proportionate differences in premiums based on sex, was omitted from 21 December 2012. It continues to apply to contracts concluded before that date14 | Schedule 3, paragraph 22 (omitted), and SI 2012/2992 |
| Age | No contravention in connection with a financial service, but where a risk assessment considers age, it must use information relevant to the assessment from a source on which it is reasonable to rely13 | Schedule 3, paragraph 20A, inserted from 1 October 2012 |
| Disability | No contravention where the thing is done by reference to information relevant to the risk from a reasonably reliable source, and it is reasonable to do it13 | Schedule 3, paragraph 21 |
The change for sex followed the judgment of the Court of Justice of the European Union in Case C-236/09, Test-Achats, of 1 March 2011, which ruled that Article 5(2) of the Gender Directive, permitting proportionate differences in premiums where sex is a determining factor in the assessment of risk based on relevant and accurate actuarial and statistical data, was invalid with effect from 21 December 2012.14
Characteristics need not be rating factors to affect prices. In PS21/5 the FCA recorded respondents' concern that pricing models may use rating factors correlated with race or other protected characteristics, and stated that firms need to ensure the data they use in pricing does not discriminate against customers based on any protected characteristic under the Equality Act 2010, unless permitted by the Act.8 It said it was exploring work on whether there are correlations between profit margins and the racial composition of local geographic areas that could result from pricing algorithms, and that age is the main factor correlated with tenure.8
US rate filing and the UK approach
In the United States, rates are regulated state by state. The NAIC's predictive models white paper summarises that state rating laws are usually grouped into prior approval, file and use or use and file, no file, and flex rating, and that rate standards typically require that rates shall not be inadequate, excessive or unfairly discriminatory.5 The NAIC has three model laws on rate regulation, and two of them describe unfair discrimination as price differentials that fail to reflect equitably the differences in expected losses and expenses.3
US regulators have also addressed price optimisation, which the CAS defines as supplementing traditional actuarial loss cost models with quantitative customer demand models.3 In its 2015 white paper the NAIC's Casualty Actuarial and Statistical Task Force reported that numerous states had issued bulletins prohibiting the practice as they defined it, recommended that two customers with the same risk profile be charged the same premium for the same coverage, while acknowledging that some temporary deviations between new and renewal customers might exist because of capping or premium transition rules, and included a potential state bulletin listing price elasticity of demand, propensity to shop and individual retention adjustments among practices inconsistent with the requirement that rates not be unfairly discriminatory.3
| Question | United States | United Kingdom |
|---|---|---|
| Is the rating plan filed with a regulator? | Depends on the state: prior approval, file and use or use and file, no file, or flex rating5 | No prior filing or approval of rates under the provisions cited here. Guidance at ICOBS 6B.2.46 says that under Principle 11 firms should notify the FCA of any change in their pricing model where there is a material risk of harm for customers, and guidance at ICOBS 6B.1.5 states the rules are not intended to affect how risk is priced109 |
| What standard must rates meet? | Not inadequate, excessive or unfairly discriminatory5 | For home and motor, a renewal price no higher than the ENBP; for non-investment insurance generally, fair value under PROD 4.2.14AR1011 |
| How are pricing models scrutinised? | The NAIC's 2020 white paper sets out best practices and state guidance for regulators reviewing rate filings based on GLMs5 | Firms report home and motor pricing data to the FCA, and the FCA uses that reporting and attestations to assess compliance8 |
The NAIC's white paper lists information a regulator may request when reviewing a GLM-based filing, including a data dictionary for every variable in the model and a rate disruption analysis showing the distribution of percentage and dollar impacts on renewal business from rerating the current book.5
Show References Hide References
- Casualty Actuarial Society. Basic Ratemaking, Fifth Edition. G. Werner and C. Modlin, May 2016.
- HM Revenue and Customs. Notice IPT1: Insurance Premium Tax. Read 10 November 2026.
- National Association of Insurance Commissioners. Casualty Actuarial and Statistical (C) Task Force Price Optimization White Paper. 19 November 2015.
- Casualty Actuarial Society. CAS Monograph No. 5: Generalized Linear Models for Insurance Rating. Second edition, 2025 revision.
- National Association of Insurance Commissioners. Regulatory Review of Predictive Models White Paper. Casualty Actuarial and Statistical (C) Task Force, adopted 2020.
- Financial Conduct Authority. MS18/1.3: General insurance pricing practices, Final Report. September 2020, updated December 2020.
- Prudential Regulation Authority. Insurance Supervision: 2026 priorities. Letter, 15 January 2026.
- Financial Conduct Authority. PS21/5: General insurance pricing practices market study, Feedback to CP20/19 and final rules. May 2021.
- FCA Handbook. ICOBS 6B.1: Application and purpose. Read 10 November 2026.
- FCA Handbook. ICOBS 6B.2: Setting renewal prices. Read 10 November 2026.
- FCA Handbook. PROD 4.2: Manufacture of insurance products. Read 10 November 2026.
- Financial Conduct Authority. Evaluation Paper 25/2: An evaluation of our general insurance pricing practices remedies. July 2025.
- Equality Act 2010. Schedule 3: Services and public functions: exceptions. legislation.gov.uk, read 10 November 2026.
- The Equality Act 2010 (Amendment) Regulations 2012. SI 2012/2992, with Explanatory Note. legislation.gov.uk.
Frequently Asked Questions
How is insurance premium tax added to an insurance quote?
The rates of insurance premium tax (IPT) have changed over time: HMRC's Notice IPT1 records that the standard rate was 10% from 1 October 2016 to 31 May 2017 (lower before that) and the higher rate was 17.5% from 1 April 1997 to 3 January 2011. All travel insurance is subject to the higher rate, but for insurance sold in relation to motor vehicles or domestic appliances the higher rate applies only when the insurance is sold by or through a person who also supplies the goods. Some contracts are exempt, including reinsurance and long-term insurance such as life insurance, except medical insurance.
The tax is calculated on the chargeable amount, which according to the notice includes any commission paid to or retained by brokers and other intermediaries out of the premium. Where one premium covers taxable and exempt risks, only the part attributable to taxable risks is chargeable, and where it covers risks at both rates there are two separate chargeable amounts. To extract the tax from a tax-inclusive premium, HMRC gives a tax fraction for each rate: 3/28 at 12% and 1/6 at 20%. Where sections of one policy fall under different rates, the tax is calculated on each chargeable amount rather than on the total alone.
Can telematics or credit data change an insurance premium?
Yes, where the rating plan uses it and the law allows it. In the UK, the renewal rules expect data gathered during the policy to count. ICOBS 6B.2.19 requires a firm to include in a customer's equivalent new business price any risk information acquired during the current policy that raises or lowers that price, including information it would not normally hold for new customers, such as telematics data from usage-based insurance or fraud risk indicators. Guidance at ICOBS 6B.2.18 allows a firm to decide how to treat information it has for new customers but not for existing ones, such as behavioural factors like the time between quote and start date, provided it can still show that it does not discriminate by tenure and that it meets the fair value requirements in PROD 4.2.14AR and, where relevant, PROD 4.2.14BR.
Credit information is used in some US rating plans. In an illustrative homeowners example in the CAS Basic Ratemaking text, a policy is placed in an underwriting tier based on the insured's credit score, tenure with the company and prior loss history; the text explains that tiers are used in jurisdictions that permit different rates within a single company. The NAIC's predictive models white paper gives a credit-based insurance scoring model as an example of a composite score, and lists among the information a regulator may obtain the predictor variables an insurer considered but did not use, with the reasons for removing them.
How are commercial insurance premiums rated?
Much commercial rating starts from a manual rate, calculated from the rate tables and factors in the rating manual, and the CAS Basic Ratemaking text says commercial lines products often use schedule rating to vary the premium from it. In schedule rating, credits and debits are applied to the manual rate for the presence or absence of characteristics. Some schedules are specific and allow no judgement, while others include subjective factors that let the underwriter choose the credit or debit. Experience rating generally applies only to larger policies: the insurer compares the policy's prior loss experience with what would be expected for the same classes, and moves the manual premium up or down accordingly.
The exposure base differs by line. In the textbook's illustrative US workers compensation example, the premium for each employee class is a rate per $100 of payroll, the totals across classes form the manual premium, and the algorithm applies a schedule rating factor and credits, then adds an expense constant, subject to a minimum premium of $1,500. The textbook notes that the National Council on Compensation Insurance recognises over 400 classes. For large commercial risks, it describes methods that calculate a premium unique to the risk, including loss-rated composite risks, large deductible policies and retrospective rating.
Can an MGA or broker set the price in a rating engine?
Yes. Where an MGA or broker sets the price the customer pays, for example on net-rated business, where it sets the final retail premium, the FCA calls it a price-setting intermediary. In PS21/5 the FCA required both insurers and price-setting intermediaries to report home and motor pricing data, saying this would help it understand potential issues at different points in the distribution chain. It reduced the burden so that only the price-setting intermediary that sets the final price reports data for that business, and removed reporting by intermediaries on business where they do not set the price.
The renewal rules follow the price wherever it is set. ICOBS 6B.2.34 requires an intermediary involved in setting any portion of the renewal price to set its portion no higher than it would for a new customer, and guidance at ICOBS 6B.2.20 says that where an intermediary re-brokes a customer to another insurer on its panel, the customer is a renewal for the intermediary but new business for the insurer. On the prudential side, the PRA's 2026 priorities letter said delegated authority business looks set to grow across the London Market, raising additional risks of inadequate pricing and reserving if not properly managed or overseen, and asked firms to have strong oversight arrangements, including how they would exit unprofitable arrangements.
What did the FCA's pricing rules cost insurers to implement?
The FCA asked firms to report their compliance costs on a voluntary basis, and its Evaluation Paper 25/2 compared them with the original cost benefit analysis. Small firms reported higher average one-off costs than estimated (£5.0m against £2.2m), while large firms reported lower ones (£6.3m against £9.4m). Ongoing costs were higher than estimated for both: £0.8m against £0.3m for small firms and £1.2m against £0.7m for large firms. The FCA cautioned that because reporting was not mandatory, the results may be influenced by self-selection bias, since firms with higher costs may have been more inclined to respond.
On the benefit side, Evaluation Paper 25/2 records that the FCA had estimated that consumers in home and motor would save £4.2bn over ten years from the pricing remedy, of which £2.5bn related to motor. In PS21/5 it recorded that a report by one firm's consultants, submitted with that firm's consultation response, estimated the benefit at £1.4bn in its fully competitive scenario, largely because of different model assumptions, and said the remedy remained proportionate whichever figure was right. When the FCA later investigated individual firms that reported price discriminatory trends, the evaluation reports that the FCA generally concluded that breaches were triggered by technical pricing errors, or that firms could explain why average margin might increase with tenure without amounting to discrimination.
We'd rather show you than tell you.
We'll start by getting to know your business. Then we'll show you exactly how Genasys handles it.
Insurance People
Zero Fluff
Tailored to You