JP2002373259A - Net premium calculation method in property insurance or the like using individual risk model and system therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a rational net premium calculation method corresponding to the individual risk characteristics of objects to be insured. SOLUTION: The net premium calculation method is provided with a step 1 of receiving one or more objects to be insured for each contractor and preserving them in a storage means by using a computer, a step 2 of calculating a generation probability for each risk factor on the basis of the result of analyzing a risk profile indicating the characteristics of the risk factor beforehand by a statistic approach or a structure model approach by using an individual risk mode, a step 3 of calculating a loss ratio for each risk factor on the basis of the result of the previous analysis, a step 4 of receiving insurance merchandise design data for each contractor, a step 5 of calculating risk distribution on the basis of the generation probability, the loss ratio and the insurance merchandise design data, and a step 6 of calculating a net premium on the basis of the risk distribution.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for calculating net premiums for non-life insurance and the like using an individual risk model. More particularly, the present invention relates to a method for calculating net premiums for individual or integrated insurance based on an individual risk model for the insured.

[0002]

2. Description of the Related Art A net premium is a portion of premium income of an insurer that is allocated to insurance payment to a policyholder (insured). Insurance premiums (also called operating premiums)
It consists of this net premium and an additional premium that will be allocated to expenses necessary for operating the insurance business. Traditionally, for most non-life insurance and some life insurance, such as group life insurance, this net premium has been calculated based on a collective risk model.

In the method of calculating net premiums based on this collective risk model, first, a risk profile corresponding to a risk factor does not separately sort homogeneous policyholders (insured persons), but sorts the same policyholders as a whole. Are set as random variables, etc., as a component of a risk distribution. For example, when considering the events of insurance accidents, we focus on how many insurance events occur in a homogeneous group of policyholders, rather than on a model of what frequency of accidents occur to which policyholders. Set the random variables. Conventionally, the majority of non-life insurance and group life insurance have used a model in which the product of the number of accidents and the average amount of loss per case is set as a random variable expressing the risk. Non-life insurance companies especially use this model for FD (Frequency Dama)
facility method).

[0004] Here, the risk factor is a primary factor that causes an insurance accident. For example, this risk factor may be death of a person, fire in a house, occurrence of an earthquake, arrival of a typhoon, etc. The risk profile is a characteristic of a coverage object that describes a probability of an insurance accident occurring for each risk factor and a loss ratio due to the occurrence of an insurance accident.

[0005] For example, when the risk factor is fire, the risk profile includes a place, a structure of a building, and the like. The accident occurrence probability and damage ratio change depending on the location and the structure of the building. Conventionally, even in a collective risk model, a part of the calculation of a probability distribution using a structural approach has been performed for the average amount of incurred damage per case based on an individual risk profile. In addition, the reason why the insurance population is modeled as a unit in the collective risk model is that a large number of rules (if the sample size in the insurance population is large enough, If the average is close to the original average for the event to occur, the error from the individual risk model will be small if the population is large enough, and it will be smaller than the individual risk model. This is because processing such as statistical processing can be simplified.

Next, the risk amount of the insurance population is measured for the entire population. Insurance claims paid are estimated according to the measured risk amount. Therefore, net premiums are calculated using a premium calculation principle based on the principle of the income and expenditure that the total premium income in the insurance population is equal to the total paid insurance claims. For example, in Japan, in accordance with the principle of premium calculation based on the collective risk model, for non-life insurance, the net premium rate is calculated according to the calculation method of the rate calculation society, and for life insurance, the Life Insurance Association Based on the calculated mortality rate based on the standard mortality rate, the company calculated uniform net premium rates without assuming a competitive market.

[0007] In recent years, insurances in which the insurance population is more subdivided than conventional insurance called risk subdivision insurance have been sold. An example of this risk segmentation insurance is risk segmentation type automobile insurance. Conventional car insurance, whose premium was determined by "vehicle type, driver age, accident history", is equipped with gender, residential area, purpose of use, ABS and airbags, and even passenger seats in this risk-divided car insurance. It is intended to further divide the conditions of the policyholder (insured person) such as the presence or absence of the safety equipment such as the dual airbag and the like so as to reduce the burden of insurance premiums on low-risk users.

However, this risk segmentation insurance is also calculated using the premium calculation principle based on the collective risk model described above. Therefore, the policyholders (insured persons) are classified according to the conditions of the policyholders (insured persons) such as gender, area of residence, purpose of use, and the presence or absence of safety equipment such as ABS and airbags and dual airbags also installed in the passenger seat. Insured persons) and set up a collective risk model that is subdivided in each case. In other words, the risk segmentation insurance can be applied only to insurance such as car insurance having a large sample number so that the collective risk model is effective even if the insurance population is segmented.

[0009] However, individual and corporate policyholders (insured persons) want to design individualized insurance products that meet their individual risk characteristics at a reasonable price.
However, as described above, conventional insurance including risk segmentation insurance is calculated based on a collective risk model. For this reason, non-standardized insurance products cannot use a collective risk model to establish a random variable with a reliable and stable probability distribution, and have a reasonable net premium insurance policy that matches the individual risk characteristics of individuals and corporations. There is a problem that insurance companies can not provide.

[0010] Conventionally, individual and corporate policyholders (insured persons) have to separately purchase standardized insurance such as fire insurance, car insurance, and life insurance. This is because the insurance administration regulates products that can be sold by business type, and statistical processing based on a large insurance population stabilizes the net premium rate. There are many policy reasons, such as the need to use standardized products for use.
However, if several insurances can be integrated, the risk of overlapping parts of standardized security can be eliminated, and the overall effect of portfolio effects can be reduced, resulting in lower premiums. Should be possible. However, even if individuals and corporations want such more freely designed insurance products, the methods based on collective risk models to date have sufficiently reflected individual risk profiles for individually designed products. It is difficult to calculate premiums.

[0011] Further, there are the following problems as other problems related to insurance product design. The coverage period of insurance contracts is on a yearly basis, especially for property and casualty insurance, and the coverage period is usually one year. When a collective risk model is used, a random variable that expresses risk in a term structure is built for insurance design in units of coverage other than year in life insurance and for multi-year contracts in non-life insurance. Because it is difficult to calculate, for example, the calculation of a reasonable insurance premium that reflects the term structure of the interest rate (yield curve), and in the case of building fire insurance, the change in the probability of fire occurrence due to aging of the building There is a problem that it is difficult to take into account. In addition, insurance such as insurance for fires caused by earthquakes
There is also a difficult problem with the insurance premium calculation technology for insurance that guarantees secondary damage. Only with the method based on the collective risk model, in the quantification of risks such as fires due to earthquakes, calculation of insurance premiums for products that secure secondary damage (fire),
It is not always sufficient as a rational method that can consistently explain the relationship with the calculation of insurance premiums for products that secure only normal primary damage (earthquake).

[0012] Conventionally, the coverage of an insurance product is an area of a loss amount equal to or greater than a deductible amount and within an insurance amount or a payment limit. Therefore, in the case of segmentation in the area of the damage amount equal to or greater than the deductible amount and within the insurance amount or the payment limit, it has been conventionally performed by performing the statistical processing by going back to the damage amount statistics, and the statistical processing is performed. There is no established method for calculating reasonable premiums for individual insurance products for which there is not enough data to do so. For example, even if a policyholder wants an insurance contract that guarantees a loss of 10% to 50% of the insurance amount in individual property fire insurance, it is not possible to perform statistical processing by going back to the loss statistics and tailor-made insurance No reasonable premium calculation method has been established for products, and the process by which policyholders and insurers reach agreement on premiums is diversified, often requiring more labor on both sides than selling general goods. Is required.

[0013]

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and provides a method for calculating a reasonable net premium in non-life insurance or the like according to individual risk characteristics of an insured. The purpose is to do. Specifically, the present invention measures the amount of risk according to the individual risk characteristics of the insured, such as non-life insurance, using an individual risk model, and provides the insurance company or the policyholder (insured) with the insurance product desired. The purpose is to provide a reasonable calculation method of net premiums for design contents.

[0014]

To achieve the above object, the present invention provides at least an input means, an arithmetic means,
Accepting one or more insured objects by an input means for each contractor by using a computer including a storage means, and individually identifying a risk profile representing a characteristic of a risk factor which is a cause of the loss of the insured object. Calculating the probability of occurrence for each of the risk factors based on the results preliminarily analyzed by a statistical approach or a structural model approach using a statistical risk model and stored in storage means; and The calculating means calculates the loss ratio for each of the risk factors based on the result stored in the storing means in advance, the step of receiving insurance product design data for each contractor by the input means, the occurrence probability, Perform risk distribution based on the loss ratio and the insurance product design data. Provided calculating means, the net premium calculation method comprising the steps of pure premium calculation means calculates based on the risk profile.

[0015] In order to achieve the above-mentioned problem, the present invention further comprises a step of receiving one or more insured objects for each contractor by the input means, using a computer including an input means, an arithmetic means, and a storage means. And a result obtained by preliminarily analyzing a risk profile representing a characteristic of a risk factor that is a cause of the loss of the insured object by a statistical approach or a structural model approach using an individual risk model and storing the risk profile in a storage means. Calculating the risk rating based on the risk rating based on the result or the risk rating analyzed in advance and stored in the storage unit based on the risk rating. Based on the result or the risk rating analyzed in advance and stored in storage means, Calculating the loss rate for each risk factor by the calculation means, receiving insurance product design data for each policyholder by the input means, the occurrence probability stored in the storage means, the loss rate, There is provided a net premium calculation method including a step of calculating a risk distribution based on insurance product design data and a step of calculating a net premium based on the risk distribution.

In addition to the steps of the above net premium calculation method,
Accepting the risk tolerance corresponding to the risk-adjusted rate of return of the insurer by the input means, and recalculating the calculated net premium based on the risk tolerance input by the input means by the computing means. It is preferable to further include Further, in the step of calculating the occurrence probability by the calculating means, the calculating means based on a risk hierarchy set in advance and stored in the storing means, representing a causal relationship of the occurrence of the loss between the risk factors stored in the storing means. It is preferable that the probability of occurrence and the damage ratio are calculated.

Further, the insurance product design data includes one or more items selected from a group including a coverage range, a coverage period, a coverage reason structure, and a premium payment method. I do.

The risk profile is analyzed in advance by a statistical approach or a structural model approach using an individual risk model, and the result stored in the storage means is a probability set in the individual risk model stored in the storage means. Based on past loss data for each of the risk factors corresponding to the variables, or natural science or engineering data of the insured object corresponding to the risk factors, all items that are candidates for a risk profile for each of the risk factors and , And a result obtained by calculating the degree of correlation with the risk factor by the calculation means. This engineering data may include social engineering data. Natural scientific or engineering data can also include a wide range of expert knowledge qualitatively and quantitatively estimated from these past data.

A risk profile representing a characteristic of a risk factor, which is a cause of the loss of the insured object, is analyzed in advance by a statistical approach or a structural model approach using an individual risk model and stored in a storage means. Based on the result, the calculating means calculates the probability of occurrence for each of the risk factors, the calculating means digitizes statistical data representing the risk profile, and a pre-analysis and setting stored in the storage means. Accepting a selection of a first statistical model for obtaining an occurrence probability by an input unit, and using the numerically statistic data stored in a storage unit to perform a first statistical model
Calculating means for calculating the parameters of the statistical model, and using the parameters calculated by the calculating means,
And calculating the probability of occurrence for each risk factor by the selected first statistical model.

A risk profile representing a characteristic of a risk factor, which is a cause of the loss of the insured object, is analyzed in advance by a statistical approach or a structural model approach using an individual risk model, and the result is stored in a storage means. Calculating the risk rating by the calculating means, receiving, by the input means, a selection of a preset risk rating function stored in the storage means for each of the risk factors, and a risk selected in the risk profile. Applying a rating function and calculating the risk rating by the calculation means.

A risk profile representing a characteristic of a risk factor, which is a cause of the loss of the insured object, is analyzed in advance by a statistical approach or a structural model approach using an individual risk model and stored in a storage means. A step of calculating the risk rating based on the result by the calculating means, a step of receiving a selection of a third statistical model set in advance by the input means corresponding to the type of the risk factor, and a step of receiving the selected third statistical model Calculating means for pre-processing statistical data of the risk profile according to the statistical model; performing statistical analysis processing on the pre-processed statistical data by the calculating means; The calculating means calculates a risk rating function that converts A step that applies the rating function to the risk profile, it is preferable to include a step of calculating the risk rating calculation means.

The causal relationship of the occurrence of the loss between the risk factors is calculated by the calculating means on the basis of the part where the occurrence of the loss between the risk factors is logically configurable and a chain of conditional probabilities of the occurrence of the loss. Preferably.

The calculating means calculates the loss rate for each of the risk factors based on the result analyzed in advance and stored in the storage means. The step of calculating the loss ratio corresponds to the risk factor and stored in the storage means. A step of accepting selection of the set first loss rate estimation statistical model by the input means; a step of performing preprocessing of the risk profile data by the arithmetic means; and a step of performing the risk profile estimation by the first loss rate estimation statistical model. Calculating the function of converting the calculated risk profile into a loss rate, and calculating the loss rate based on the risk profile by the calculating means. It is suitable.

The calculating means calculates the loss rate for each risk factor based on the risk rating, and inputs a selection of a second loss rate estimation statistical model corresponding to the risk factor stored in the storage means. Means for receiving the risk rating, means for pre-processing the risk rating, means for calculating the function for converting the risk rating into a loss rate by the second loss rate estimation statistical model, and The calculating means preferably calculates a loss rate based on the risk rating using a function that converts the risk rating into a loss rate.

The calculating means calculates a risk distribution based on the probability of occurrence, the loss ratio, and the insurance product design data stored in the storage means, based on the calculated probability of occurrence. Calculating the probability of occurrence corresponding to the insurance reason structure and the insurance period;
When calculating the loss ratio of the secondary loss for each of the risk factors, the loss ratio of the secondary loss for each of the risk factors is adjusted in consideration of the insured range, Combining the probability of occurrence corresponding to the coverage period, the modified loss ratio in consideration of the coverage range for the loss ratio for each of the risk factors, and the probability of occurrence corresponding to the coverage event structure and the coverage period. Combining the loss ratio, the modified loss ratio, and the probability of occurrence corresponding to the warranty event structure and the insurance period, based on the coverage range and the insurance event structure. Calculating the risk distribution using

The present invention further provides a program for causing a computer to execute the above-described steps. The present invention also provides a computer-readable recording medium recording the program.

Further, the present invention provides an input means for receiving one or more insured objects for each contractor, a storage means for receiving and storing insurance product design data for each contractor by the input means, A storage means for preliminarily analyzing a risk profile representing a characteristic of a risk factor, which is a factor causing a loss of an object, by a statistical approach or a structural model approach using an individual risk model and storing a result stored in a storage means. And calculating means for calculating the probability of occurrence and its loss rate for each of the risk factors based on the result stored in the storage means and pre-analyzed and stored in the storage means, and storing the calculated loss probability in the storage means, Based on the stored probability of occurrence and the loss ratio and the insurance product design data stored in the storage means, a risk Providing calculation means for calculating a cloth, a premium computing system including a calculating means for calculating a pure premiums based on the risk profile.

Further, the present invention provides an input unit for receiving one or more insured objects for each contractor, a storage unit for receiving and storing insurance product design data for each contractor,
A storage means for storing a risk profile representing a characteristic of a risk factor which is a cause of the loss of the insured object, a result of analyzing in advance by a statistical approach or a structural model approach using an individual risk model, A computing means for calculating a risk rating based on the result analyzed in advance and stored in the storage means, and a calculation based on the result or the risk rating analyzed in advance and stored in the storage means, for each of the risk factors. The calculation means calculates the occurrence probability and the damage rate, and stores the calculation result in the storage means; the occurrence probability and the damage rate stored in the storage means; and the insurance product design data stored in the storage means. Computing means for calculating a risk distribution based on the risk distribution, and computing means for calculating a net premium based on the risk distribution. To provide a net insurance premium calculation system.

In the above net premium calculation system, the storage means for receiving and storing the risk tolerance corresponding to the risk-adjusted rate of return of the insurer by the input means, and the net premium calculated based on the risk tolerance. It is preferable to further include an operation means for re-calculating.

When the calculating means calculates the probability of occurrence and the loss ratio, the calculating means may calculate based on a risk hierarchy which is set in advance and stored in the storage means and which indicates a causal relationship of occurrence of the loss between the risk factors. preferable. Preferably, the insurance product design data includes one or more items selected from a group including a coverage range, a coverage period, a coverage reason structure, and a premium payment method.

The terms mentioned above are defined below. The input means is an input means operated by an operator such as a keyboard or a mouse, an input by a recording medium such as a floppy (registered trademark) disk, or a communication line or an L.
Means for receiving an input by an electric signal such as an AN (local area network) is included.

The storage means is means for storing data such as characters and numbers, programs, and the like. For example, a random access memory or a hard disk in a computer. The arithmetic means is means for executing a program. For example, it is generally called a central processing unit, that is, a CPU. The computer-readable recording medium is a medium that records programs and data and is readable by a computer. For example, a floppy disk,
CD-ROM, MO, ROM chip, etc.

The policyholder is a natural person or a legal person or a group of those who have the right to enter into an insurance contract with an insurance company. An insured object is an object whose insured profit is guaranteed by an insurance contract. For example, real estate, such as buildings, or economic value that would be lost due to personal property damage, such as cars, ships, and airplanes,
It is the economic value lost in the event of a corporate accident.

An individual risk model is a model for calculating a risk distribution by setting and analyzing a random variable for each probabilistic event to be insured in setting a model for estimating a risk distribution. It is. For example, in the case of fire insurance, in the individual risk model, even if there are a plurality of fires, the probability variable is set individually for each individual fire probability event as in Equation 1. For this reason, the amount of damage (cash flow movement) for each individual fire probability event can be considered, and the design of various products can be elaborated. For example, if an insurance product is designed using an individual risk model, the coverage (partial area of the coverage) can be set arbitrarily according to the needs of the user, or the risk of a fire occurring after an earthquake Insurance of hierarchy combination can be designed reasonably. In other words, by adopting an individual risk model, calculating the probability of occurrence and the loss ratio for each risk factor separately, incorporating them into the content of the desired insurance product, and calculating the premium Will be able to At this time, even the correlation between the probability of occurrence and the loss ratio between the individual risk factors can be taken into account.

(Equation 1) The calculation of premiums by the individual risk model has not been applied to actual non-life insurance design because the calculation is generally very complicated and financial engineering has not been developed as it is now. In actual insurance products, a collective risk model with few parameters and described later has been used.

A collective risk model is a method of estimating a risk distribution, in which a stochastic event to be insured is not distinguished from one having the same property,
A model that calculates the risk distribution by analyzing global properties such as the sum of those random variables. For example, in the case of fire insurance, in the collective risk model, random variables are set for a plurality of fire-related probability events collectively as in Expression 2. In the past, the calculation of premiums for property and casualty insurance based on the collective risk model, when starting sales as a new product, takes into account various past statistics that are considered to be closely related, while fully considering safety. After calculating the insurance premium and after the sales performance of the product has been obtained, recalculate the premium as appropriate based on the payment performance of the sold product and adjust the excess and deficiency between the insurance premium and the payment performance Is often done. For this reason, basic statistical data such as the frequency of insurance accidents and the amount of damage are collected and managed for each type of product sold by many insurance companies. At present, it is not possible to adjust the relationship between the above data and the data provided by earthquake insurance. In other words, it has been operated in a form that cannot separate insurance design and risk factors. Therefore, when trying to design a highly individualized insurance, it is difficult to set a reasonable random variable that can reflect the risk profile, and it is extremely difficult to design a free product as enabled by the present invention. With. Nor can the risk hierarchy be considered in a collective risk model. In addition, the so-called risk segmentation insurance is to simply design the insurance by segmenting the population into several categories. It is a simple derivation of the collective risk model in that the stochastic event makes a homogeneous assumption and is essentially different from the individual risk model described above.

(Equation 2)

The risk factor is a primary factor that causes an insurance accident. For example, deaths of people, fires in houses, earthquakes, typhoons, liability,
For example, the occurrence of a loan default. The risk factors in the present invention include death of a person, personal injury, retirement of a company, eruption, business interruption, bad weather, insolvency of business partners,
Fluctuations in lending rates, strategic investment failures, bill defaults,
Suspension of money exchange, government control, insolvency of financial products, occupational accidents, loss of credibility, liability for employers, poisoning, political transactions, separation of key managers, confidentiality of products, kidnap /
Ransom demand, trade secret disclosure, employee misconduct, inconsistency in inventory, theft, changes in business practices, intellectual property infringement, product liability, breach of contract, error or negligence, illegal employment practices, Includes all factors that could cause loss to natural persons and corporations, such as liability for tripartites, liability for motor vehicle accidents, bankruptcy of inbound and outbound contractors and contractors, liability for broadcasting, and liability for officers.

The risk profile refers to the characteristics (and a set of characteristics) of the coverage objects that explain the probability of an insurance accident occurring for each risk factor and the loss ratio due to the insurance accident occurrence. For example, when the risk factor is fire, the location of the building, the structure of the building, the purpose of use of the building, the type of business of the individual or corporation using the building, the presence or absence of fire extinguishing equipment, and the like are items that constitute the risk profile. For each of these items, for example, the presence or absence of fire extinguishing equipment, the risk of fire is reduced if there is a fire extinguishing equipment, the risk profile is determined quantitatively or qualitatively such that the risk is high if not. be able to.

The results of analyzing the risk profile in advance by a statistical approach using an individual risk model or by a structural model approach are as follows. With respect to risk factors, only items of a risk profile having strong explanatory power are selected, and a set of calculated numerical values or indices indicating the degree of correlation between each item and the risk factor is meant. For example, about fire
If the structure of the building and the purpose of use account for 80% of the probability of fire occurrence, select these two risk profiles, Determining a set of coefficients for converting a given point into a probability corresponds to this.

More specifically, the statistical approach is:
This is a method of estimating the probability of occurrence and loss ratio by processing the risk profile using a statistical method. For example, if the loss due to fire is insured, the records of past fires (past loss data) are examined and statistically processed to estimate the probability of occurrence corresponding to the elements of the risk profile. This is to estimate the probability of occurrence and the like corresponding to the risk profile of the individual case as a combination of elements.

The structural model approach is a method of constructing a model for associating a physical structure and a logical structure inherent in an insured object with a risk profile, and estimating an occurrence probability and a loss ratio by simulation or the like. . In constructing this model, natural science or engineering data is used. This engineering data includes social engineering data. Natural scientific or engineering data also includes expert knowledge obtained from these data in the past. This is because qualitative or quantitative estimations and judgments can be made from expert knowledge even for risk factors for which there is no experience and no loss data exists in the past.

For example, this structural model approach is:
When the loss due to fire is insured, the probability of occurrence and the loss ratio are estimated by modeling the building reflecting the risk profile or modeling the logical structure of the building fire occurrence. It can also be said. Here, modeling a building as a space means that when a fire shelter is installed between the first room and the second room, a fire in one room is unlikely to cause a fire in another room. In other words, a model is created based on natural science or engineering data, such as being less likely to be completely burned in the event of a fire and having a high probability of being partially burned. Also, modeling the logical structure of a building fire occurrence means that if the building is built with reinforced concrete and the type of business used is a chemical manufacturer, the cause of the fire is likely to be a fire from a chemical reactor. , Made of reinforced concrete,
Modeling is based on past natural science or engineering data, such as being less likely to burn down and having a higher probability of being physically burnt down.

The engineering data includes social engineering data. This is because, for example, if the risk factor is fire, social engineering data such as the number of fire stations in the area where the insured building exists, the number of fire brigade, population density, etc. Is greatly affected. Therefore, such social engineering data is also included in the engineering data. The social engineering data shall also include general market data such as the amount of compensation for liability insurance such as product liability, and the amount of damages in case of death in a car accident. These amounts may change due to rulings and social circumstances. Therefore, these are also included in the social engineering data. The social engineering data shall include not only social engineering statistical data but also qualitative social engineering considerations based on changing social situations.

The probability of occurrence is the probability of occurrence of a loss due to the risk factor set by the insured. In the present invention, the occurrence probability can be treated as different for each insured object, but it is also possible to set the same for a plurality of insured objects. Also,
At each point in time, it is also possible to construct the occurrence probability in a form having a period structure for the insured period as a value obtained by multiplying the instantaneous occurrence probability by the non-occurrence probability up to each point.

The loss ratio is a ratio of a loss amount to an insurance amount or a payment limit. Here, the insurance amount is the amount of insurance against insurance accidents contracted by the insurance company and the insurance subscriber. In addition, the payment limit is the maximum guarantee amount for an insurance accident. Since the magnitude of the damage amount is stochastically distributed, the loss ratio also has a probability distribution. The probability distribution of loss ratios differs for each insured.

The insurance product design data includes the coverage area, coverage period, coverage reason,
This is data that sets specifications as financial products by combining or structuring the contents of insurance premium payment methods.

Here, the coverage refers to a partial area of the coverage of an insurance product having a single or a plurality of risk factors. If the exemption amount is set for a single risk factor, the insured range is equal to or greater than the exempted amount, and the range of the damage amount up to the insurance amount or the payment limit is the insured range. For example, if a special insurance contract that guarantees only damage of 10% to 50% of the insurance amount out of the insurance amount is made at the request of the policyholder, the area of the loss amount of 10% to 50% of the insurance amount is obtained. Is the coverage. (Especially, such a subdivided coverage area is called a layer (security category) in terms of reinsurance, etc.)

The coverage period is a period during which an insurance accident is guaranteed by insurance. Also called insurance period. Period is years,
The unit may be any of six months, quarters, months, days, hours and minutes.
In the case of non-life insurance, insurance contracts that cover multiple years on a yearly basis are called multi-years. Insured events are defined as insurance contracts.
It is the nature of an insurance accident, which is defined as the event under which the obligation to pay insurance claims occurs. In integrated insurance for multiple risk factors, such as fire and damages, the coverage is based on these natural risk factors and can be determined artificially. Such integrated insurance for multiple risk factors is particularly called multi-line.

In addition, it is also possible to design an insurance in which, when the primary damage is an earthquake, an insurance obligation is required for a secondary damage fire caused by the earthquake. In this case, the insured event is a risk factor having a risk hierarchy. Such a combination of insurance events such as multi-line and secondary damage is called an insurance event structure. The insurance premium payment method is a method in which a contractor pays an insurance premium to an insurer. Annual payment, monthly payment, semi-annual payment, etc. can be considered.

The risk distribution is calculated based on the probability of occurrence and the loss ratio, and is the probability distribution of all insured payouts of the insurer which occur stochastically for each policyholder (insured). In the risk hierarchy, when a risk factor is composed of a plurality of factors and the factors have a hierarchical structure composed of a chain of causal relationships, this hierarchical structure is called a risk hierarchy. For example, in the case of an insurance accident "fire caused by the occurrence of an earthquake", the risk hierarchy has a two-layer structure due to a chain of causal relationships between the earthquake and the fire. By taking into account such a risk hierarchy, for example, the relationship between the net premium for products that cover secondary damage (fire) and the net premium for products that cover only normal primary damage (earthquake) can be determined. Because it can be explained consistently, the net premium for insurance products against earthquakes and fires can be more reasonably calculated. At that time,
This has the effect that the coverage of primary damages and secondary damages can be set freely according to customer needs.

The risk rating is an index integrating these risk profiles when the risk profiles related to the risk factors are different from each other or when the risk profiles are large. Normal,
It is used in the form of a symbol (A, B, C, etc.) having an order of about 10 or an integer. Like a normal risk profile, a risk rating is a characteristic that describes the probability of occurrence and the loss ratio.For example, the risk rating has a relative correlation such that the lower the rating by number, the lower the probability of occurrence and the loss ratio. Use the one that is
In other words, the risk rating is an index indicating the likelihood of loss of the insured. Based on the risk profile, it is calculated as a parameter representing the characteristics of the probability event for each policyholder's insured object. Adopting such a risk rating concept simplifies the relationship between multiple risk profiles that represent the characteristics of the coverage and the risk distribution, and allows for intuition to be associated, so expert knowledge Is easier to reflect. In addition, there is an effect that the risk distribution is not affected by a slight change of the risk profile which has essentially no influence, and the robustness is obtained.

The pre-processing means that the data is arranged by replacing qualitative data with numerical values, supplementing missing data, or standardizing the data, and then performing statistical processing such as principal component analysis and calculating a statistical model. It is to obtain desired data and functions. Statistical models are models that are created or logically created based on past data. For example, a linear regression model, a logit-probit model, and the like.

The risk-adjusted rate of return is calculated as the value obtained by subtracting the expected loss and expenses from the profits and dividing it by the required capital. It is an indicator that is used mainly for appropriate capital allocation. The risk tolerance refers to the tolerance of the insurer (insurer) for underwriting risk of insurance contracts. If an insurance contract is underwritten with a net premium calculated based on the individual risk of the insured, the composition of the underwriting portfolio may cause the insurer to deviate from the net premium calculated based on the principle of income and expenditure. You will take risks. Such tolerance is a control variable due to the bias of the underwriting portfolio distribution.

The risk tolerance corresponding to the risk-adjusted rate of return is the risk tolerance set to obtain the expected maximum loss required when calculating the required capital amount of the risk-adjusted rate of return. That is. When calculating net premiums based on this risk tolerance, the calculation is made using the value of return on equity, dividend rate or hurdle rate in capital allocation, assuming a competitive market. Here, calculating on the assumption of a competitive market means that the insurance market is a competitive market, that is, assuming that there are always competitors, and calculating the lowest net premium limit that the insurer can underwrite. It is to be. In this way, taking into account risk tolerance, insurers calculate objective and standard net premiums that take into account management and competition risks when designing any complex insurance product. There is an effect that can be done.

According to the method for calculating a net premium according to the present invention described above, a reasonable net premium can be calculated according to the individual risk characteristics of the insured. According to the present invention,
Measure the amount of risk according to the individual risk characteristics of the insured, and request the insurance company or policyholder (insured)
Calculate reasonable net premiums for insurance product design. Further, according to the present invention, several insured objects and risk factors can be collectively insured. Therefore, it is possible to set a low net premium as a total.

[0055]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of a net premium calculation method according to the present invention will be described below with reference to the drawings. FIG. 1 is a flowchart outlining the steps of the net premium calculation method according to the present embodiment. As shown in FIG. 1, the net premium calculation method according to the present embodiment is roughly divided into six steps 1 to 6 in which a net premium is calculated using a computer. First, the setting of the insured object is received by the input means (step 1). Then, the calculation means calculates the occurrence probability (step 2). Next, the calculation means calculates the loss rate (step 3). The setting of the insurance product design data is received by the input means (step 4). The calculation means calculates the risk distribution (step 5). The calculation means calculates the net premium (step 6).
These are the above six steps. These steps 1-6
Also, the net premium calculation method according to the present invention, which can be understood in the steps described below, can be implemented by a system using a computer including at least an input unit, a calculation unit, and a storage unit. Although not described in detail, a program for implementing the net premium calculation method according to the present invention and a storage medium storing the program are also one of the embodiments of the present invention.

Next, an embodiment of the net premium calculation method according to the present invention will be described in detail. 2 and 3 show an embodiment of the method for calculating a net premium according to the present invention. Step 1 in FIG. 1 of receiving the setting of the insured object by the input means corresponds to step 10 in FIG. Step 2 in FIG. 1 where the calculation means calculates the occurrence probability corresponds to steps 11 to 14 in FIG. Next, step 3 in FIG. 1 where the calculating means calculates the loss rate corresponds to steps 15 to 19 in FIG. Step 4 of receiving the setting of the insurance product design data of FIG. 1 by the input means
Corresponds to step 20 in FIG. Step 5 in which the calculation means calculates the risk distribution in FIG. 1 corresponds to step 21 in FIG. Step 6 in which the calculating means calculates the net premium in FIG. 1 corresponds to steps 22 to 25 in FIG.

[Setting of Insured Target] First, the insurance premium calculator inputs an insured target for each policyholder (insured) using input means such as a keyboard electrically connected to a computer. At this time, one or more insured objects are input to the computer. The computer accepts one or more settings of the insured object set for each contractor, and stores them in the storage means (step 10). The insured object is an object whose insured profit is guaranteed by an insurance contract.
For example, economic value that is lost due to damage to real estate such as a building, or property such as a car, a ship, or an airplane, and economic value that is lost when an accident occurs in an individual or a corporation.

[Setting of Risk Factor and Preliminary Analysis of Risk Profile] Next, the setting of a risk factor, which is a cause of the loss of the insured object, is received by the input means. For this risk factor, a risk profile representing a risk characteristic is analyzed in advance by a statistical approach or a structural model approach using an individual risk model.

The risk factor is a primary factor that causes an insurance accident. For example, deaths, personal injuries, fires in houses, earthquakes, typhoons, floods, liability obligations, loan defaults, etc. Further, when the insured object is a loss due to a building disaster, a fire, an earthquake, a wind disaster, a flood disaster, a lightning strike, a hail disaster, a snow disaster, and the like are appropriate as risk factors. The relationship between the insured object and the risk factor can be stored in advance in a storage unit in the computer as a table showing the relationship.

The risk profile indicates the probability of occurrence of an insurance accident for each risk factor, and the characteristics of the coverage object that describes the loss ratio due to the occurrence of the insurance accident. For example, when the risk factor is fire, the risk profile may include items such as the location, structure, purpose of use, and presence / absence of fire extinguishing facilities that affect the loss of the insured building. Also, for example, when the risk factor is death of a person, the risk profile includes, in addition to age and gender, physical medical information and family composition that are individual attributes of the individual, which have been used for conventional life insurance. ,
It also includes the place of residence, housing environment, occupation, qualifications, income, educational background, history of rewards and punishments, and psychological and behavioral characteristics of life risks and health.

The individual risk model is a model for calculating a risk distribution by setting and analyzing a random variable individually for each probable event to be insurance in setting a model for estimating a risk distribution. It is. This is one of the core ideas of the present invention. For example, in the case of fire insurance, in the individual risk model, even if there are a plurality of fires, a random variable is individually set for each individual fire probability event. Furthermore, one can consider the correlation between these individual fires, or the correlation between the occurrence of other events and the occurrence of a fire (in a very simple example, a fire is more likely to occur if an earthquake occurs). . On the other hand, in the collective risk model, a plurality of fires are collectively considered to calculate an occurrence probability or a damage rate of the entire fires.

The above-described risk profile is analyzed in advance by a statistical approach using an individual risk model or a structural model approach using a structural model of occurrence events. This is because if the factor is fire, it is necessary to confirm that the risk profile is appropriate for the risk factor of fire.

Further, preferably, in a statistical approach using an individual risk model, a relationship between a fire and an item of a risk profile is indicated by using loss data due to a past fire, and a degree of correlation is shown for each risk profile. The calculation means calculates in the form of an index or a numerical value.

In the structural model approach using an individual risk model, a model for associating the physical structure and the logical structure inherent in the insured object itself with the risk profile is constructed, and the probability of occurrence and the loss ratio are calculated using a computer or the like. Estimate by simulation using. In constructing this structural model, natural science data or engineering data is used. This engineering data includes social engineering data.

The natural science or engineering data also includes expert knowledge obtained from these past data. For example, in the structural model approach, using natural science or engineering data on buildings, the calculation means calculates the relationship between fire and risk profile items in the form of an index or numerical value indicating the degree of correlation for each risk profile. I do. Specifically, a risk profile item having a correlation with a risk factor equal to or higher than a predetermined numerical value or level is defined as a fire risk profile. The natural science or engineering data on the building in the structural model approach using the individual risk model is the material of the building (steel, concrete, wooden)
In the past, based on past engineering data (including expert knowledge), whether fire-prone materials are used, how much they spread, or the degree of concentration of artificial fire and the number of fire stations from a social engineering perspective Also includes social engineering data.

As described above, engineering data widely includes social engineering data. For example, if the risk factor is fire, the number of fire stations in the area where the collateral is located,
Social engineering data, such as the number of fire brigade and population density, can also have a significant effect on the amount of loss after a fire. Therefore, such social engineering data is also included in the engineering data. The social engineering data shall also include general market data such as the amount of compensation for liability insurance such as compensation for product liability, and the amount of compensation for damages in the event of death in a car accident. Since these amounts may vary depending on the judgment or social situation, it is necessary to take into account the risk rating, the loss ratio using the risk profile, and the like when calculating by the calculation means. Therefore, these are also included in the social engineering data.

Incidentally, the natural science or engineering data widely includes not only pure data but also expert knowledge obtained from these data. For example, even if you use a risk profile that has no experience or no corresponding data in the past, based on past natural science or engineering data, experts can use this knowledge to You can judge whether there is. Thus, natural science or engineering data also includes knowledge widely.

FIG. 4 is a flowchart showing a method of analyzing a risk profile. According to this flow,
The analysis is performed by a computing means on the computer. First,
The risk factor is set in the storage means of the computer from the input means (step 30). Next, all items that can be candidates for the risk profile for each risk factor are stored in storage means, past loss data for each risk factor, or the insured natural science or engineering corresponding to the risk factor. It is set from the input means based on the target data (step 31).

Then, a selection of whether to use the past loss data or the natural science or engineering data of the insured object corresponding to the risk factor for each item is received by the input means (step 32). Here, both the loss data and the natural science and engineering data may be used.
Further, for each item, a selection of whether to obtain the degree of correlation with the risk factor as a numerical value or as an index is received by the input means (step 33).

Then, the degree of correlation for each item is determined using past loss data for each risk factor or insured natural or scientific data corresponding to the risk factor stored in the storage means. Is calculated by the calculating means (step 34). The calculating means determines whether the finally calculated numerical value or index is equal to or higher than a predetermined numerical value or level (step 35). If the numerical value or index is equal to or higher than the predetermined numerical value or level, this item is adopted as a risk profile.
If the value is equal to or less than a predetermined value or level, the calculation means analyzes that this item is rejected as a risk profile.

[Calculation of Occurrence Probability] Steps 11 to 11 in FIG.
As shown in FIG. 14, the calculation by the calculation means of the occurrence probability is
The calculating means may calculate directly based on the risk profile, or the calculating means may calculate the risk rating once based on the risk profile and then calculate the probability of occurrence. The insurance premium calculator sets in the storage means of the computer from the input means (step 11).

In the case where the calculation means directly calculates the occurrence probability from the risk profile in step 11, the risk profile is based on the result of analyzing the risk profile in advance by a statistical approach or a structural model approach using an individual risk model. Then, the occurrence probability is calculated by the calculation means for each risk factor (step 12). When the calculation means does not directly calculate the probability of occurrence from the risk profile, the calculation means calculates the risk rating based on the result of analyzing the risk profile in advance using a statistical approach or a structural model approach using an individual risk model. Calculation is performed (step 13). Then, the calculation means calculates the occurrence probability for each risk factor based on the risk rating (step 14).

[Calculation of Occurrence Probability Reflecting Risk Profile] In step 11, when the input means accepts a selection to calculate the occurrence probability directly from the risk profile, the risk profile is used by an individual risk model. The calculation means calculates the occurrence probability based on the result of the analysis performed in advance by the statistical approach or the structural model approach (step 12).
In other words, based on the results analyzed in advance by the statistical approach or the structural model approach using the individual risk model, the statistical data corresponding to the risk profile of the insured is stored in the storage means. , Calculating means calculates the occurrence probability. For example, in the case of a statistical approach using an individual risk model, a logit / probit model is applied to statistical data representing a risk profile stored in a storage unit.

In the logit / probit model, linear regression analysis is performed on the statistical data of a set of scores indicating the risk profile after converting the values using an inverse function of a logistic distribution or an inverse function of a standard normal distribution. Then, when the value obtained by the regression analysis is converted by the inverse function of the logistic distribution, the value is substituted into the logistic distribution function, and when the value is converted by the inverse function of the standard normal distribution, the value is substituted into the standard normal distribution function. The calculating means calculates.

The calculation of the probability of occurrence based on the risk profile is performed, for example, on a computer as shown in the flowchart of FIG. First, the statistical data representing the risk profile stored in the storage means is digitized (step 40). Next, selection of a statistical model for obtaining an occurrence probability is received by the input means (step 41). Then, using the quantified statistical data, the estimated values of the parameters of the selected statistical model are calculated by the calculating means (step 42). Finally, the calculating means calculates the probability of occurrence using a statistical model using the parameters calculated by the calculating means (step 43).

[Calculation of Occurrence Probability by Risk Rating] In step 11, when the input means accepts a selection that the calculation means does not calculate the occurrence probability directly based on the risk profile, the risk profile is calculated using an individual risk model. The calculation means calculates the risk rating based on the result analyzed in advance by the statistical approach or the structural model approach (Step 1).
3). Next, the calculation means calculates the occurrence probability based on the risk rating (step 14). In other words, the calculating means calculates the occurrence probability using the data corresponding to the risk rating of the insured object, which is stored in the storage means.

Here, the risk rating is an index integrating these risk profiles when the risk profiles related to the risk factors are different from each other or when the risk profiles are large, and the risk It is an index that indicates ease. For example, in the case of a statistical approach using an individual risk model, principal component analysis is performed after standardizing statistical data of a set of scores indicating a risk profile stored in a storage unit.

Next, a coefficient corresponding to each item of the risk profile thus obtained is obtained by linear regression analysis, and a weighted average is used using the coefficient. The calculating means calculates the risk rating of the insured object by associating the range obtained by dividing the probability distribution of the obtained loss amount with the probability of the damage amount calculated by the same calculating means from the risk profile of the insured object. It is done in the manner of doing. For example, the calculation of the risk rating is performed on the computer by calculation and dispersion according to the flowchart of FIG. The calculation of the risk rating is roughly divided into two stages and is calculated by the calculation means. One is a pre-processing step of determining a function for assigning a risk rating from a risk profile (steps 50 to 53), and then assigning a risk rating to a given specific risk profile (step 54). To 55).

First, selection of a statistical model corresponding to the type of risk factor is received by the input means (step 50). The pre-processing of the risk profile data according to the statistical model is performed by the calculating means (step 5).
1). Here, pre-processing is to quantify the risk profile, analyze whether it is independent of other risk profiles, and perform principal component analysis on whether or not this risk profile is the main component for loss It is.
Next, statistical analysis is performed on the preprocessed risk profile (step 52). This statistical analysis performs a linear regression analysis or the like on the risk profile. From the statistical analysis result, a risk rating function for converting the risk profile into a rating is determined (step 53).

According to the type of the risk factor, selection of a risk rating function from the risk rating functions calculated in step 53 is received by the input means (step 5).
4). Apply the risk rating function to the risk profile,
The calculation means calculates the rating (step 55).

The calculation by the calculating means of the probability of occurrence based on the risk rating is performed, for example, by comparing statistical data or natural science or engineering data corresponding to the risk rating stored in the storage means for each risk rating. The frequency of occurrence (intensity) is calculated by the calculating means, and is associated with the probability of occurrence. For example, on a computer, it is calculated by the arithmetic means according to the flowchart of FIG. In the calculation by the calculation means of the probability of occurrence based on the risk rating, the calculation by the calculation means of the probability of occurrence by the statistical approach using the individual risk model, the calculation by the calculation means of the probability of occurrence by the structural model approach using the individual risk model There is.

In the calculation by the statistical approach using the individual risk model of the probability of occurrence based on the risk rating,
First, statistical data relating to the occurrence of a loss of a risk factor is input (step 60). Next, the pre-processing for each risk rating of the statistical model is performed by the calculating means (step 61). Here, the preprocessing is to recalculate the statistical model for each parameter called risk rating. From the analysis result, a function corresponding to the occurrence probability is determined from the risk rating (step 62). With the function corresponding to the occurrence probability from the risk rating, the calculation means can calculate the occurrence probability from the risk rating.

In the calculation of the occurrence probability based on the risk rating by the calculating means by the structural model approach using the individual risk model, first, the selection of the structural model corresponding to the risk factor is received by the input means (step 63). Here, the structural model refers to natural science or engineering data,
Alternatively, it is a model for each risk factor that is determined qualitatively or quantitatively using knowledge based on these. For example, if the risk factor is fire, if a certain event occurs based on natural science or engineering data (including expert knowledge) of the building materials and the structure of the rooms, the fire may occur in a specified room. This is a model that simulates whether an accidental fire will occur.

Preprocessing is performed by the arithmetic means for each risk rating (step 64). Here, the preprocessing is to delete statistically undesirable data when deriving a function corresponding to the occurrence probability from the structural model. Next,
The function corresponding to the occurrence probability is determined from the risk rating by the structural model (step 65). With the function determined in step 65, the occurrence probability is calculated from the risk rating by the calculation means.

Step 1 where the calculating means calculates the occurrence probability
In steps 2 and 14, the calculation means may calculate the occurrence probability in consideration of the risk hierarchy in addition to the risk profile and the risk rating. The risk hierarchy is a hierarchical structure in which a risk factor is composed of a plurality of factors and the factors have a hierarchical structure composed of a chain of causal relationships. There is generally a causal relationship between the risk factors and the probability of occurrence. For example, an earthquake easily causes a fire. In order to reflect this causal relationship in a risk distribution to be described later, a hierarchy of risk factors is a risk hierarchy.

By considering the probability of occurrence of fire (conditional probability) due to an earthquake based on the setting of the risk hierarchy, these causal relationships can be reflected in the model of the probability of occurrence. The conditional probability is calculated by calculating the probability of occurrence of another risk under the occurrence of a certain risk based on statistical data between risk factors. For example, the calculation is performed by calculating the simultaneous distribution by the calculating means and dividing the simultaneous distribution by the probability distribution in which the condition itself occurs.

As the risk hierarchy, a quantitative numerical value such as a correlation coefficient (for example, rank correlation) may be used, or a relationship indicating a qualitative correlation may be set artificially. The risk hierarchy is, for example, as shown in FIG.
On a computer, it is calculated by arithmetic means. First, a setting of data expressing a mutual relation with a risk factor defining a risk hierarchy is received by an input unit (step 70). The part that can be logically constructed from the element statistics of each risk factor, such as product events such as fire (secondary event) that is likely to occur due to an earthquake (primary event), and the part where the chain of conditional probabilities and others are estimated as one The setting of the data to be classified is received by the input means (step 71).

Here, the logically configurable portion means that a primary event (primary loss) and a secondary event (primary loss) that cause a fire due to an earthquake but are not logically considered to occur due to the fire. Secondary loss) and the damage ratio and probability of occurrence. The conditional probability indicates a relationship with a probability of occurrence of a secondary event due to a primary event contracted by a contractor's request. Then, a function that gives an occurrence probability and a loss rate corresponding to the risk hierarchy is determined (step 72).

[Calculation of Loss Ratio] Steps 15 to 1 in FIG.
As shown in FIG. 9, the loss ratio is calculated by a computer based on the risk profile or the risk rating. The premium calculator accepts the setting from the input means in the storage means of the computer as to whether to calculate the loss ratio directly from the risk profile when the calculation means calculates (step 15). If the loss ratio is to be calculated based on a result previously analyzed by a statistical approach using a specific risk model of the risk profile or a structural model approach, the calculation means proceeds to step 16. Alternatively, if the calculating means does not directly calculate the loss ratio based on the risk profile, that is, if the calculating means calculates the loss ratio based on the risk rating, the process proceeds to step S17.

If the risk rating is calculated when the calculating means calculates the probability of occurrence in step 13, the calculating means proceeds to step 19, and the calculating means calculates the loss ratio for each risk factor based on the risk rating. Calculation is performed (step 19). If the calculation means does not calculate the risk rating, the calculation means calculates the risk rating based on the result of analyzing the risk profile in advance using a statistical approach using an individual risk model or a structural model approach. After that, the calculation means calculates the loss ratio for each risk factor based on the risk rating (steps 18 to 19).

[Calculation of Loss Ratio Based on Risk Profile] FIG. 9 is an image diagram of the loss ratio when the loss ratio is specified or patterned as a probability distribution depending on the risk profile. The horizontal axis represents the loss ratio, and the vertical axis represents the probability density of the loss ratio. Curve L in FIG. 9 indicates a fire profile in the case where the risk profile is as follows (location: Tokyo, building structure: reinforced concrete structure, building usage purpose: sales, fire protection equipment: fire protection compartment: fire extinguishing equipment: sprinkler). The probability distribution of the loss ratio is shown. This curve L indicates that a large loss occurs with a small loss ratio (the loss amount is relatively small with respect to the insurance amount), but a large loss ratio occurs (the loss amount is relatively large with respect to the insurance amount). It shows that it is difficult.

The curve N in FIG. 9 has a risk profile such as (location: Tokyo, building structure: wooden, building use purpose: manufacturing, fire protection equipment: no fire protection compartment, fire suppression equipment: no sprinkler). The probability distribution of the fire damage rate in the case is shown. Contrary to the curve L, this curve N is unlikely to cause a loss with a small loss ratio (the loss amount is relatively small with respect to the insurance amount), but has a large loss ratio (the loss amount is relative to the insurance amount). Loss) is likely to occur. In addition, a curve M in FIG.
(Location: Tokyo, Building structure: Concrete block structure, Building use purpose: Manufacturing, Fire protection equipment: Fire protection compartment available, Fire extinguisher equipment: No sprinkler) Shows the probability distribution of fire damage rate. The curve M has a smaller loss ratio (the loss amount is relatively smaller than the insurance amount) and a larger loss ratio (the loss amount is relatively larger than the insurance amount) as compared with the curves L and N. ) Loss is unlikely to occur, but the loss rate is likely to be around 0.5.

The relationship between the risk profile and the loss rate can be obtained quantitatively or qualitatively from the loss rate data corresponding to the risk profile. For example, the calculation of the loss ratio based on the risk profile by calculation means
The calculation is performed by the calculation means in such a manner that the statistical data representing the risk profile stored in the storage means is accumulated and a distribution is created. Alternatively, using natural science or engineering data corresponding to the risk profile, it is calculated by a calculating means using a tree diagram reflecting all possible events. For example, the loss rate based on the risk profile is calculated by the calculation means of the computer according to the flowchart of FIG. The calculation of the loss ratio based on the risk profile can be divided into a calculation using a statistical approach using an individual risk model and a calculation using a structural model approach using an individual risk model.

In the case of calculating a loss ratio by a statistical approach using an individual risk model based on a risk profile, first, a selection of a preset loss ratio estimation statistical model corresponding to a risk factor is received by an input means ( Step 80). Next, the pre-processing of the risk profile data is performed by the calculating means (step 81). A loss rate function for converting a risk profile into a loss rate by a statistical model is represented by a function type (for example, a function Z = ax + by + c) that best represents the tendency of data using risk profile data, and parameters and constants of this function. Is calculated and estimated by the calculating means (step 82). With the above, the pre-processing is completed, and the selection of the loss rate function according to the type of the risk factor is accepted by the input means as each processing (step 83). Further, the calculation means calculates the loss ratio relating to the risk profile (step 84).

In the case of calculating a loss ratio by a structural model approach using an individual risk model based on a risk profile, selection of a structural model corresponding to a risk factor is received by an input means (step 85).
Next, the risk profile is pre-processed by the calculating means (step 86). The calculation means simulates the damage ratio distribution using the structural model (step 87).

[Calculation of loss ratio based on risk rating]
FIG. 11 is an image diagram of the loss rate when the loss rate is specified or patterned as a probability distribution depending on the risk rating. 11A to 11E show loss ratios when the risk ratings are A to E. The horizontal axes of FIGS. 11A to 11E show the loss ratio, and the vertical axes show the probability density. For example, FIG. 11A means that a loss with a low loss ratio (the ratio of the loss amount to the insurance amount is low) is likely to occur, and a loss with a high loss ratio (the ratio of the loss amount to the insurance amount is large) is unlikely. Is shown. FIG. 11E shows that, conversely, a loss with a low loss ratio (the ratio of the loss amount to the insurance amount is low) is unlikely to occur, and a loss with a high loss ratio (the ratio of the loss amount to the insurance amount is large) is likely to occur. I have. FIGS. 11B, C, and D have a tendency intermediate between FIGS. 11A and 11E, and from FIG. 11B closer to FIG. 11A, the closer to FIG.

The loss ratio in FIG. 11 is calculated based on the risk profile and calculated based on the risk rating. The relationship between the risk rating and the loss ratio is obtained quantitatively or qualitatively from the loss ratio data corresponding to the risk rating. For example, the input means accepts the setting of the probability density function of the beta distribution as shown in Expression 3 for the distribution of the loss ratio.

(Equation 3) The relationship between the risk rating and the loss ratio is obtained by determining the parameters α and β of the beta distribution in Expression 3 by a statistical approach or a structural model approach using an individual risk model according to the risk rating.

Also, for example, when the statistical data obtained by the statistical approach is abundant, or when the natural science or engineering structure of the risk is relatively simple in the structural model approach, the loss rate distribution of the risk rating is relatively simple. f (X) may be received by the input means. Equation 4
The relationship between the risk rating and the loss ratio is obtained by determining the parameters of the Escher transformation as shown in (1) using a statistical approach or a structural model approach using an individual risk model. A rating higher than a general rating (a loss with a high loss ratio is relatively unlikely to occur) is obtained by converting the loss ratio distribution f (x) with a negative parameter α (f
(X) × φ (x)) is a loss rate distribution, and those with low ratings (high losses are relatively likely to occur)
Parameter α is a value obtained by converting f (x) with a positive value (f
(X) × φ (x)) is the loss rate distribution.

(Equation 4)

The calculation of the loss ratio based on the risk rating is performed by the calculation means according to, for example, the flowchart of FIG. The calculation of the loss ratio based on the risk rating is based on the statistical approach using the individual risk model based on the risk rating and the calculation of the loss ratio using the structural model approach using the individual risk model based on the risk rating. There is.

In the calculation of the loss ratio by a statistical approach using an individual risk model based on the risk rating, first, as an advance process, selection of a loss ratio estimation statistical model corresponding to a risk factor is received by an input means (step 90). ). Next, the risk rating data is pre-processed by the calculating means (step 91). A loss rate function for converting a risk rating into a loss rate using a loss rate estimation statistical model is estimated by calculating a function type that best represents the tendency of risk rating data, and parameters and constants of this function by using arithmetic means (step 92). Next, as each process, the selection of the loss rate function by the type of the risk factor is received by the input means (step 93).
The calculation means calculates the loss ratio relating to the risk rating (step 94).

The calculation of the loss ratio by the structural model approach using the individual risk model based on the risk rating is as follows:
First, selection of a structural model corresponding to a risk factor is received by the input means (step 95). Here, the structural model of the damage ratio means that, for example, if the risk factor is fire, the building materials and the structure of the rooms are calculated from natural science or engineering data (including expert knowledge) from a certain room. This is a model that simulates how much it spreads when a fire starts. Next, the risk rating data is pre-processed by the calculating means (step 96). Using the structural model, the damage ratio distribution is simulated by the calculating means (step 9).
7).

[Insurance Product Design Data] Next, the insurer stores the insurance product design data for designing the insurance product in the storage means by the input means (Step 2 in FIG. 3).
0). Insurance product design data mainly includes coverage, coverage period, coverage reason structure, insurance premium payment method, and the like. Although not mentioned here, the insurance product design data may include data relating to other insurance product designs.

[Setting of Coverage Range] The coverage range refers to the coverage set for the deductible amount. It may be set for each insured object, each risk factor, and each coverage period. The exemption may be set by the amount of the exemption,
The exemption amount may be set by an exemption rate which is a value obtained by dividing the exemption amount by the insurance amount or the payment limit. This exemption rate is 10% ~ 2
Layers (security divisions) for every 10%, such as 0%, 20% to 30%, etc., may be set. The exemption rate may be set to a layer having a wide range such as 10% to 50%.

[Setting of Coverage Period] The coverage period is a period in which the insured object is secured. The period may be in year, month, half year, quarter, day, hour and minute units. For example, it may be multiple years.

[Setting of Covered Reasoning Structure] The term of the covered reasoning structure refers to a combination of the covered reasons. Here, the insured event refers to the content of an insurance accident that is defined as an event in which an obligation to pay an insurance claim occurs in an insurance contract.

An example of the insurance reason structure is a contract causal table as shown in FIG. In the contract causal table of FIG. 13, for example, when the primary damage is an earthquake, a contract is to be made as to how much of the fire, which is one of the secondary damages due to the earthquake, is secured. For example, when the primary damage is four types of fire, earthquake, wind disaster and flood disaster and the secondary damage is four types of fire, earthquake, wind disaster and flood disaster, the combination of primary damage and secondary damage is primary and secondary damage. There are 12 cases except for 4 cases where the damage is the same.
Therefore, 12 contracts are made on the contract causal table. For example, a value of 0.5 is described in the column when the primary damage in FIG. 13 is a wind disaster and the secondary damage is a flood disaster.
This 0.5 is the ratio of collateral for flood damage that occurs indirectly as secondary damage after a wind damage of primary damage occurs.
(50%).

[Setting of Insurance Payment Method] The insurance payment method is a method in which a contractor (insured person) pays to an insurance company as an insurer. For example, it is determined whether to pay in a lump sum or in installments.

FIG. 14 is a diagram showing an image of a security cover based on insurance product design data according to the present embodiment. The calculation means creates a table as shown in this image and stores it in the storage means. The column a in the table of FIG. 14 indicates the risk factor, the columns b to f in the table of FIG. 14 indicate the coverage period, and the column g of the table in FIG. The layer (%) shown in column g of the table in FIG. 14 indicates the percentage of the exemption amount with respect to the insurance amount or the payment limit. Note that in FIG. 14, the columns that are blacked out indicate the secured layers (guaranteed divisions).

For example, the portion of the table shown in FIG. 14 where the column b and the row h overlap each other indicates the coverage period of the first year, and covers 25% to 100% of the insurance amount against the risk factor fire. Means that. At this time, the insurance amount is 1
When the amount is 100 million yen, the insurer is in the exemption from having to pay the insurance money to the policyholder for the loss of 1 billion yen.

Further, the portion of the table shown in FIG. 14 that overlaps column d and row k indicates the third-year coverage period, and covers 50% to 75% of the insurance amount against the risk factor of flood disaster. Means that. At this time, the insurance amount is 100
When the payment is 8 billion yen, the insurer will pay the policyholder 2.5 billion yen of 5 billion yen to 7.5 billion yen in the collateral range.

[Calculation of Risk Distribution] Steps 12 and 14
The calculation means calculates the risk distribution based on the probability of occurrence calculated by the calculation means in step, the loss ratio calculated by the calculation means in steps 16 and 19, and the insurance product design data stored in the storage means in step 20. (Step 21).

The risk distribution is calculated based on the probability of occurrence and the loss ratio, and is the probability distribution of all insured payouts of the insurer which occur stochastically for each policyholder (insured). The risk distribution is calculated on a computer, for example, as shown in FIG.
According to the flowchart shown in FIG. First, the occurrence probability calculated in step 43 is based on a predetermined insurance reason structure and an insurance period. Therefore, the occurrence probability corresponding to the insurance event structure and the insurance period desired by the insurer or the contractor is recalculated by the calculating means based on the occurrence probability obtained in step 43 (step 100). Next, a setting as to whether or not the calculating means calculates the secondary damage (secondary loss, secondary event) set in the insurance reason structure is received by the input means (step 101). When the secondary damage is calculated by the calculation means, the corrected damage ratio and the occurrence probability in consideration of the coverage range are combined with the damage ratio of the secondary damage for each risk factor (step 102).

When the secondary damage is not calculated by the calculating means, the corrected damage rate and the probability of occurrence are combined by the calculating means in consideration of the damage ratio for each risk factor in consideration of the coverage (step 103). Whether the risk distribution can be analytically calculated by the calculation means is set by the input means (step 104). If the risk distribution can be calculated analytically by the calculation means, the risk distribution is calculated analytically by the calculation means using the combined data and the data of the coverage reason structure and the coverage (step 105). If the risk distribution cannot be calculated analytically, the combined data and the
The risk distribution is approximately calculated by the calculation means using the data of the coverage range (step 106). In addition, the calculated risk distribution is compared with the results from other numerical analysis methods, the results of similar products, and the like to confirm the approximation accuracy (step 107).

The results from other numerical analysis methods and the results of similar products are compared, and if the approximation accuracy is not the desired accuracy, the results are combined again using the numerical analysis method in step 106. The risk distribution is approximately calculated by the calculation means using the data and the data of the coverage reason structure and the coverage area.

[Calculation of Net Premium] A typical calculation of net premium (steps 21 to 25) will be described below. The net premium is calculated by the calculating means based on the risk distribution calculated by the calculating means in step 21 (step 22). The average amount of damage (pure risk amount) μ is given by Equation 5
As shown by, the expected value E [X] of the random variable X according to the risk distribution L is defined as the average amount of damage.

(Equation 5)

The average amount of this loss amount (pure risk amount) μ
Is the net premium that does not take into account the risk tolerance corresponding to the risk-adjusted rate of return of the insurer.

[Recalculation of net premiums according to risk tolerance]
It is set in the storage means by the input means whether the insurer wants to calculate by the arithmetic means in consideration of the risk tolerance corresponding to the risk-adjusted rate of return of the insurer (step 2).
3). Here, the risk tolerance is the tolerance of the underwriter for the underwriting risk, and in the present embodiment, is a control variable related to the bias of the portfolio distribution of the underwriting risk. When the risk tolerance is not considered, the net premium is the net premium calculated by the calculating means in step 22. If the setting is made in consideration of the risk tolerance, the calculation means proceeds to step 24. First, accept the risk tolerance corresponding to the insurer's risk-adjusted rate of return,
It is stored in the storage means (step 24). Based on the risk tolerance, the net premium calculated by the calculating means is recalculated by the calculating means (step 25).

The recalculation of the net premium according to the risk tolerance corresponding to the risk-adjusted rate of return is specifically performed based on the calculation formula (6).

(Equation 6) Here, the risk amount indicates an expected maximum loss amount, and for example, a value-at-risk (VaR) or an expected shortfall is used. Value-at-risk is an indicator that measures the maximum loss within an event excluding outside the confidence interval. Expected shortfall is a measure of the average loss of events where the loss exceeds the value at risk. Here, the risk tolerance corresponding to the risk-adjusted rate of return of the insurer represents the percentile of the confidence interval, and generally uses 95% or 99%.

Here, the confidence interval means a range having a certain probability and being on the left side of the probability percentage point in a certain probability distribution. Here, the percentage point is a value at which the upper probability (lower probability, both-sided probability) is 100 × α% in a certain probability distribution. α is a number from 0 to 1. Usually, 100 × α% often has a lower probability, and is expressed as “90 percent point” or “99 percent point”. The lower probability is called a reliability (or a confidence level), and a range of possible values corresponding to the reliability is called a confidence interval.

Equation 6 is derived as follows. First, the risk-adjusted profit from the insurer such as an insurance company taking on the risk X is calculated from the net premium p (X) by the net risk amount E
It is a value obtained by subtracting [X].

(Equation 7)

At this time, the insurer's capital requirement for X is calculated by calculating the increase in the cost-adjusted profit from the increase in the expected maximum loss (R (Y + X) -R (Y)) for the entire portfolio of underwriting risk, that is, This is a value obtained by subtracting the net premium p (X).

(Equation 8)

Here, the risk-adjusted rate of return of the insurer is defined as in Expression 9.

(Equation 9)

The marginal value of the risk-adjusted rate of return is the return on shareholders 'equity, the dividend on shareholders' equity, or the hurdle rate in capital allocation.

(Equation 10)

By transforming equation (10), the equation for calculating the net premium in equation (6) is obtained. When transferring a part of the layer by reinsurance or the like, it is necessary to take into account the risk transfer cost. For example, a net insurance premium in the case of transferring a risk with a loss amount of H or more can be expressed by Expression 11.

[Equation 11]

Equation 11 is based on the assumption of a competitive market.
This is because there is no arbitrage opportunity even through the reinsurance market, and the net premium p (X) has the additivity as shown in Expression 12 with respect to the layer.

(Equation 12)

[Eliminating Effect of Duplicate Event and Portfolio Effect] By using the net premium calculation method of the present invention to secure not all individual risk factors but all of the risk factors in an integrated manner, Due to the following two effects, the net premium can often be reduced rationally.

The first effect is an effect by eliminating the duplication of stochastic events to be covered by insurance. For example, consider insurance that integrates life insurance (death insurance) and car insurance. The set of events of the target insurance accident is Ω. In A, a set of events covered by life insurance is A, and a set of events covered by car insurance is B. Then, a set of events of death due to a car accident is A∩B. For life insurance, X is a random variable that represents the amount paid from the insurance where each event occurred.
Assuming that Y is for automobile insurance and X = Y = Z for A∩B, the expected values (pure risk amounts) E (X) and E (Y) used for calculating the net premium for each insurance are as shown in Expression 13. expressed.

(Equation 13)

In the case of using different insurances,
Is duplicated in the calculation of net premiums. If collateral can be secured in an integrated manner, net premiums in the overlap will be reduced. To enable such integration,
It is necessary to divide and hierarchize the risk factors of car accidents and construct an individual risk model like the net premium calculation method of the present invention. The second effect is due to the correlation of stochastic events. The set of events of insurance accidents is Ω
Let X and Y be new random variables that represent two insurance payments with no event overlap defined by. A net premium formula that reflects the risk tolerance corresponding to the insurer's risk-adjusted rate of return, for example, can be used to calculate the amount of risk
If the risk is expressed using the standard deviation, the variance V (X + Y) of X + Y becomes

[Equation 14] Becomes Cov (X, Y) is the covariance of the random variables X, Y. The relationship of Expression 15 also holds for the standard deviation σ (= √V). Such a property is called a portfolio effect.

(Equation 15)

Generally, the portfolio effect means that the total risk amount is smaller than the sum of the individual risk amounts. Let R and G be random variables representing individual loss amounts, and R
If (G) and R (H) are the risk amounts of G and H, respectively, the total risk amount R (G + H) is the individual risk amount R
(G) The fact that the inequality of Expression 16 that is smaller than the sum of R (H) is satisfied is referred to as “satisfying the subadditivity”.
The inequality in equation (16) formulates the portfolio effect.

(Equation 16)

Therefore, in many cases, the net premiums are reduced by integrating the net premiums in consideration of the risk tolerance corresponding to the risk-adjusted rate of return of the insurer.

[0131]

[Example 1] Earthquake-secured fire insurance for a cosmetics company factory in Shizuoka Prefecture As an example, the method for calculating the net premiums according to the present invention is based on an earthquake-secured fire insurance for a cosmetics company factory in Shizuoka Prefecture. Calculated net insurance premiums. The insurance amount is 1 billion yen, the coverage period is 5 years, the deductible amount is 50 million yen, and the earthquake collateral ratio is 5 in the contract causal table.
It was 0%. The risk profile of the insured factory of the above cosmetics company, which is the contractor (insured), is listed in Shizuoka Prefecture, the structure of the building is steel, the type of industry is chemical, and the purpose of the building is chemicals. Manufacturing and fire extinguishing equipment had private fire extinguishing equipment.

As a result of analyzing this risk profile based on this risk profile by a statistical approach using an individual risk model or by a structural model approach, it was found that all risk profiles could be applied to a fire as a risk factor. First,
The risk distribution, which is the probability distribution of payments, based on the probability of fire up to five years including the conditional probability of fire in the event of an earthquake, the loss rate and the above product structure was calculated by the calculation means. The net premium was calculated by the calculation means, taking into account the risk tolerance corresponding to the risk-adjusted rate of return of the insurer. As a result, insurance services that cannot be provided by conventional insurance could be applied. For example, since the coverage period is set to five years (multi-year), insurance for each year is reduced to 5 years.
Net insurance premiums could be reduced compared to joining the year.
Also, net insurance premiums could be reduced compared to separately providing fire insurance and earthquake insurance. Heretofore, it has not been possible to evaluate such risks in an integrated manner and calculate the risk distribution using arithmetic means.

[Example 2: Integrated insurance for all facilities of a construction company in Kanagawa Prefecture] In order to show that free design of insurance is possible using the net premium calculation method of the present invention, the Shows integrated insurance for all construction company facilities. The insurance amount is 500 million yen for fire and 4 for earthquake.
100 million yen for flood disaster and 100 million yen for flood disaster. In the contract causal table, the security ratio for a fire caused by an earthquake is 50%, and the security ratio for a flood caused by an earthquake is 10%. For this insurance contract, the risk profile of the construction company that is the contractor (insured) is listed, the location is Kanagawa prefecture, the structure of the building is steel or reinforced concrete or wooden,
The type of business is construction, the purpose of the building is business and manufacturing, and the fire extinguishing equipment is without private fire extinguishing equipment. Based on this risk profile, analysis of this risk profile using a statistical approach or a structural model approach using an individual risk model revealed that all risk profiles could be applied to the risk factor fire.

A fire corresponding to this risk profile,
Based on the term structure of the damage ratio and the probability of occurrence (including conditional probability) for each risk factor of earthquake and flood disaster, the calculation means calculates the finally integrated risk distribution based on the product structure shown in FIG. The net premium was calculated by the calculation means taking into account the risk tolerance corresponding to the risk-adjusted rate of return of the insurer.

The blacked columns in FIG. 16 are the layers that have been assigned. For example, in the first year, the exemption rate indicating the exemption rate from the fire insurance amount is 25% or less, indicating exemption. As a result, insurance services that cannot be provided by conventional insurance could be applied. For example, since the insurance period is set to be multiple years (multi-year) of 5 years, the net insurance premium can be reduced as compared with the case where the insurance for each single year is subscribed for 5 years. In addition, net insurance premiums could be reduced compared to separately providing fire insurance, earthquake insurance, and flood insurance. Conventionally, the risk distribution of such risks could not be integratedly calculated by the calculation means. As shown in the present embodiment, in the net premium calculation method of the present invention, a so-called portfolio effect can be generated by integrally securing all risk factors instead of individual risk factors, and the net premium can be reduced. Reasonably lowered.

[Example 3: Death insurance on the way to work for a company worker] Not only the age and gender of the worker, but also the location of the house, commuting route, attendance time, average return time, work location, affiliation, Attributes such as the average weight of portable items, medical history, friendship, alcohol consumption, and drinking frequency at home are used as risk profiles.

Using the net premium calculation method according to the present invention, a model attendance pattern and a model return home pattern include an accident rate of a vehicle or the like occurring on a commuting route, a morbidity rate of a vehicle or the like (typhoon / earthquake), and walking by medical history. Calculate the mortality rate for each individual by applying an estimate of the sudden death rate at the time. In this way, different premiums can be calculated for the same age, gender, and insurance design.

[0138] Such insurance cannot be performed by conventional statistical methods, that is, analysis of data collected by the life insurance association and insurance company. The first is that data with such information has not been aggregated in the past, and the second is that the number of samples is extremely small even if the relevant case is found in some way, so that it can be applied to insurance. This is because the accuracy of the mortality cannot be ensured.

[Example 4: Specific illness insurance based on medical close-up examinations] Investigation of lifestyle as a result of medical examinations (including DNA analysis) of blood and the like using the net premium calculation method according to the present invention. Possibility of improving lifestyle by psychoanalysis,
Analyze family medical history and other data as a risk profile, and configure insurance to cover the economic losses associated with the occurrence of specific diseases in the future.

Conventionally, the probability of occurrence of such a specific illness is extremely small, and the more a specific illness becomes, the more it cannot be analyzed by policyholder statistics possessed by an insurance company. In addition, depending on medical knowledge, it is considered that there are many cases where an insurance company rejects a contract due to risk selection, and in many cases, effective insurance design cannot be performed because statistics do not exist.

Example 5: Insurance Covering Loss of Company Project Due to Retirement of Employees for Voluntary Retirement Using the net premium calculation method according to the present invention, all employees of a business company are Age, years of service, position, position,
Analyze risk profiles such as salary treatment, past medical history, number of family members, presence or absence of family members who need nursing care, presence or absence of family-owned companies, psychological analysis of psychological resistance to voluntary retirement, etc. Estimate the probability of retirement at your own convenience. Based on this probability, we calculate the expected value of economic loss when employees engaged in projects in the company retire due to voluntary reasons and calculate insurance premiums.

In the conventional method of calculating the voluntary retirement rate as described above, the retirement rate is calculated based on the number of people by the age of three to five years in the past or by the age of service. However, the personal circumstances of past and future retirees cannot be the same, and it is not possible to estimate the likelihood of retirement for each individual involved in the project.

[Example 6: Reinsurance of the longevity risk of a pensioner of a corporate pension] Using the net premium calculation method according to the present invention, the type of business of the company in which the corporate pension is established and the Estimate the mortality rate of the beneficiary group of the company based on assets, medical history, and housing conditions, and provide reinsurance to cover the additional costs of pension benefits accrued to the company pension when longevity can be expected from the average. Constitute.

Conventionally, the mortality rate of the pensioners of the tax-qualified pension and welfare pension fund is almost always uniform according to regulations. As a countermeasure, only ex post pension finance adjustment can be made.

[0145]

As is apparent from the above description, according to the present invention, it is possible to calculate a reasonable net premium according to the individual risk characteristics of the insured. According to the present invention, the risk amount according to the individual risk characteristics of the insured object is measured, and the insurance company or the policyholder (insured person) desires a reasonable net premium based on the design conditions of the insurance product. Can calculate.

That is, by adopting the calculation method or system of the net premium calculation using the individual risk model according to the present invention, the insurance design and the risk distribution calculation can be separated, and the risk hierarchy is also calculated in the calculation of the risk amount. Can be used, and a more rational and consistent method can be used, so that premium calculation is possible even in highly individualized insurance product design, which has been difficult to calculate premiums up to now. As a result, it is possible to develop a highly flexible insurance product that meets the needs of the consumer.

[Brief description of the drawings]

FIG. 1 is a flowchart showing one embodiment of a net premium calculation method according to the present invention.

FIG. 2 is a flowchart showing one embodiment of a net premium calculation method according to the present invention.

FIG. 3 is a flowchart showing an embodiment of a net premium calculation method according to the present invention.

FIG. 4 is a flowchart showing a risk profile setting method of the net premium calculation method according to the present invention.

FIG. 5 is a flowchart showing calculation of an occurrence probability based on a risk profile in the net premium calculation method according to the present invention.

FIG. 6 is a flowchart showing calculation of a risk rating in the net premium calculation method according to the present invention.

FIG. 7 is a flowchart showing calculation of an occurrence probability based on a risk rating in the net premium calculation method according to the present invention.

FIG. 8 is a flowchart showing the setting of a risk hierarchy in the net premium calculation method according to the present invention.

FIG. 9 is a graph showing the probability distribution of the loss ratio based on the risk profile of the net premium calculation method according to the present invention.

FIG. 10 is a flowchart showing calculation of a loss ratio based on a risk profile of the net premium calculation method according to the present invention.

FIG. 11 is a graph showing a probability distribution of a loss ratio based on a risk rating in the net premium calculation method according to the present invention.

FIG. 12 is a flowchart showing calculation of a loss ratio based on a risk rating in the net premium calculation method according to the present invention.

FIG. 13 is a schematic diagram showing a contract causal relation table of the net premium calculation method according to the present invention.

FIG. 14 is a schematic diagram showing an image of a product structure of the net premium calculation method according to the present invention.

FIG. 15 is a flowchart showing risk distribution calculation of the net premium calculation method according to the present invention.

FIG. 16 is a schematic diagram showing an insurance product structure of an embodiment of a net premium calculation method according to the present invention.

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme court ゛ (Reference) G06F 17/18 G06F 17/18 Z (72) Inventor Shosuke Nakamura 1-5, Otemachi, Chiyoda-ku, Tokyo 1-5 1 F-term in IBJ Daiichi Financial Technology Co., Ltd. (reference) 5B056 BB00 BB21 BB64 HH00

Claims (20)

[Claims] 1. A step of receiving one or more insured objects for each contractor, and statistically using an individual risk model to represent a risk profile representing a characteristic of a risk factor that causes a loss of the insured object. Calculating the probability of occurrence for each risk factor based on the results analyzed in advance by an approach or a structural model approach, and calculating the loss rate for each risk factor based on the results analyzed in advance. Calculating, accepting insurance product design data for each contractor, calculating a risk distribution based on the probability of occurrence, the loss ratio, and the insurance product design data, based on the risk distribution Calculating a net premium. 2. A step of receiving one or more insured objects for each contractor; and statistically analyzing a risk profile representing a characteristic of a risk factor which is a cause of the loss of the insured object by using an individual risk model. Calculating a risk rating based on a result analyzed in advance by an approach or a structural model approach; and calculating a probability of occurrence for each of the risk factors based on the result analyzed in advance or the risk rating. Calculating a loss ratio for each of the risk factors based on the result of the pre-analysis or the risk rating; accepting insurance product design data for each contractor; Based on the loss ratio and the insurance product design data, the risk distribution Net premium calculation method comprising the steps of calculation, and calculating the net premiums based on the risk profile. 3. The method of claim 1, further comprising: receiving a risk tolerance corresponding to the risk-adjusted rate of return of the insurer; and recalculating the calculated net premium based on the risk tolerance. 2. Net premium calculation method described in 2. 4. The method according to claim 1, wherein in the calculating the occurrence probability, the occurrence probability and the loss ratio are calculated based on a predetermined risk hierarchy, which represents a causal relationship of the occurrence of the loss between the risk factors. 1-3
Net premium calculation method described in any of the above. 5. The insurance product design data includes one or more items selected from a group including a coverage range, a coverage period, a coverage reason structure, and a premium payment method. The net premium calculation method according to claim 1. 6. The risk profile in which a risk profile has been analyzed in advance by a statistical approach or a structural model approach using an individual risk model, and the result is calculated for each risk factor corresponding to a random variable set in the individual risk model. The degree of correlation between all items that are candidates for a risk profile for each of the risk factors and the risk factor based on past loss data of the risk factor or natural science or engineering data of the insured subject corresponding to the risk factor. 6. The net premium calculation method according to claim 1, wherein the result is obtained by calculating the following formula. 7. A risk profile representing a characteristic of a risk factor, which is a factor that causes the loss of the insured object, based on a result of previously analyzing by a statistical approach or a structural model approach using an individual risk model. Calculating the probability of occurrence for each of the risk factors; digitizing statistical data representing a risk profile; and accepting selection of a first statistical model for analyzing and setting a predetermined probability of occurrence. Calculating the parameters of the first statistical model using the quantified statistical data; and using the calculated parameters, the probability of occurrence for each risk factor using the selected first statistical model. Calculating.
7. The method for calculating a net premium as described in any one of items 6 to 6. 8. A risk profile representing a characteristic of a risk factor, which is a factor that causes the loss of the insured object, based on a result of previously analyzing by a statistical approach or a structural model approach using an individual risk model. Calculating the risk rating, for each of the risk factors, receiving a selection of a preset risk rating function, and applying the selected risk rating function to the risk profile, and calculating a risk rating. The net premium calculation method according to any one of claims 2 to 7, comprising: 9. A risk profile representing a characteristic of a risk factor which is a cause of the loss of the insured object, based on a result of previously analyzing by a statistical approach or a structural model approach using an individual risk model. Calculating a risk rating, receiving a selection of a preset third statistical model corresponding to the type of the risk factor, and statistics of a risk profile according to the selected third statistical model. Preprocessing data; performing statistical analysis on preprocessed statistical data; calculating a risk rating function that converts the risk profile to the risk rating based on a result of the statistical analysis processing; Apply the rating function to the profile to calculate the risk rating Net premium calculation method according to any one of claims 2-8, characterized in that it comprises a step. 10. The causal relationship of the occurrence of a loss between the risk factors is calculated based on a part where the occurrence of a loss between the risk factors is logically configurable and a chain of conditional probabilities of the occurrence of the loss. Claims 4 to
9. The insurance premium calculation method according to any of 9. 11. A step of calculating a loss ratio for each of the risk factors based on the result of the analysis in advance, wherein the loss ratio of the first loss ratio estimation statistical model corresponding to the risk factor is set in advance. Accepting the selection, pre-processing the data of the risk profile, calculating a function for converting the risk profile into a loss ratio by the first loss ratio estimation statistical model, and calculating the calculated risk profile Calculating a loss ratio based on the risk profile using a function that converts the data into a loss ratio. 12. A step of calculating a loss ratio for each of the risk factors based on the risk rating, receiving a selection of a second loss ratio estimation statistical model corresponding to the risk factor, Pre-processing; calculating a function for converting the risk rating into a loss ratio according to the second loss ratio estimation statistical model; and converting the calculated risk rating into a loss ratio by the function. Calculating the loss ratio based on the rating. 13. A step of calculating a risk distribution based on the probability of occurrence, the loss ratio, and the insurance product design data, comprising: Calculating the probability of occurrence corresponding to the warranty period; and calculating the loss ratio of secondary damage for each of the risk factors, taking the coverage into consideration in the loss ratio of secondary damage for each of the risk factors. Combining the adjusted loss ratio with the probability of occurrence corresponding to the insured reason structure and the insured period; and for each of the risk factors, the insured loss ratio, the adjusted loss ratio, the insured reason structure, and Combining the probability of occurrence corresponding to the insured period, the combined loss ratio, the modified loss ratio, the insured event structure and the Based on the occurrence probability corresponding to the coercive period, using the insured event structure as the insured scope, claim 5 including the step of calculating the risk distribution
12. The method for calculating a net premium as described in any one of 12 above. 14. A program for causing a computer to execute the steps according to claim 1. Description: 15. A computer-readable recording medium on which the program according to claim 14 is recorded. 16. An input means for receiving one or more insured objects for each policyholder, a first storage means for receiving and storing insurance product design data for each policyholder, and a loss of the insured object occurs. A second storage means for storing a result of analyzing a risk profile representing a characteristic of a risk factor, which is a factor to be performed, by a statistical approach or a structural model approach using an individual risk model; A first calculating means for calculating an occurrence probability and a loss rate for each of the risk factors based on the stored pre-analyzed results and storing the calculated probability in a third storage means; A second calculating a risk distribution based on the probability of occurrence and the loss ratio and the insurance product design data stored in the first storage means;
A net premium calculation system that includes a calculation unit and a third calculation unit that calculates a net premium based on the risk distribution. 17. An input unit for receiving one or more insured objects for each contractor, a first storage unit for receiving and storing insurance product design data for each contractor, and a loss of the insured object occurs. A second storage means for storing a result of analyzing a risk profile representing a characteristic of a risk factor, which is a factor to be performed, by a statistical approach or a structural model approach using an individual risk model; A risk rating is calculated based on the stored pre-analyzed result, and based on the pre-analyzed result or the risk rating, an occurrence probability and a loss ratio are calculated for each of the risk factors. A first calculation unit stored in a storage unit, the occurrence probability and the damage ratio stored in a third storage unit, A net premium calculation system comprising: a second calculation unit that calculates a risk distribution based on insurance product design data; and a third calculation unit that calculates a net premium based on the risk distribution. 18. A fourth storage means for receiving and storing a risk tolerance corresponding to the risk-adjusted rate of return of the insurer;
18. The net premium calculation system according to claim 16, further comprising: fourth calculating means for recalculating the calculated net premium based on the risk tolerance. 19. The method according to claim 19, wherein the first calculating means calculates the occurrence probability,
19. The loss ratio is calculated based on a risk hierarchy that is set in advance and stored in a fifth storage unit and that indicates a causal relationship of occurrence of a loss between the risk factors. The net premium calculation system described in Crab. 20. The insurance product design data includes one or more items selected from a group including a coverage range, a coverage period, a coverage reason structure, and a premium payment method. 20. The net premium calculation system according to any one of claims 16 to 19.