Law of Large Numbers

The law of large numbers is a statistical principle stating that as the number of similar, independent exposure units increases, the average loss per exposure tends to more closely approximate the expected average loss, enabling insurers to predict losses with greater accuracy.

Get personalized explanations

🎬 Video Explanation

Checking video availability…
💡

Exam Tip

Larger pools of similar independent exposures improve AVERAGE-loss predictions. Individual losses remain variable. Correlated catastrophes and changing assumptions limit the pooling benefit.

What is the Law of Large Numbers?

The law of large numbers is a fundamental statistical principle that forms the mathematical foundation of insurance. It states that as the number of independent, similarly distributed observations grows, their average converges toward the expected value. In insurance, this means that larger pools of similar risks allow for more accurate loss predictions.

Core Concept

ElementDescription
PrincipleLarger samples yield more predictable results
ApplicationMore similar, sufficiently independent exposures improve average-loss predictability
ResultEnables actuarial pricing of premiums
FoundationSupports risk pooling; it does not eliminate uncertainty

How It Works in Insurance

Sample SizePrediction AccuracyExample
10 similar driversRealized accident rate can fluctuate widelyA few claims can strongly affect the rate
1,000 similar driversSmaller random fluctuation in the average, under the same assumptionsMore stable estimate of the underlying rate
100,000 similar driversGreater average-rate precision if risks remain sufficiently independentNo guarantee of an exact match in a given year

Mathematical Foundation

ConceptDescription
Expected ValueAverage outcome predicted statistically
VarianceMeasure of spread from expected value
Standard Error of the MeanFor independent losses with common finite variance, equals sigma / sqrt(n)
ConvergenceAverage loss approaches its expectation; individual losses remain variable

Insurance Application

StepApplication
1Collect historical loss data
2Calculate expected loss frequency and severity
3Insure large number of similar risks
4Average loss tends toward expected loss under the model assumptions
5Set premiums to cover expected losses + expenses + profit

Why Larger Pools Work Better

FactorSmall PoolLarge Pool
Random variation in average lossLargerSmaller, for comparable independent exposures
Predictability of average lossLess preciseMore precise under the same model
Total dollar exposureSmallerMay be larger; pool size does not guarantee solvency
Premium estimatesLess credibleMore credible if data and assumptions remain applicable

Actuarial Science Connection

ElementRole of Law of Large Numbers
Rate MakingAccurate premium calculations
ReservingPredicting future claim payments
UnderwritingGrouping similar risks
ReinsuranceSpreading catastrophic risk

Requirements for Law of Large Numbers

RequirementDescription
Homogeneous RisksSimilar exposure units
Independent RisksOne loss doesn't affect others
Random LossesNot predictable individually
Sufficient Sample SizeMore observations reduce sampling uncertainty; no universal cutoff

Limitations

LimitationDescription
Catastrophic EventsAffects many policies simultaneously
Non-independent RisksCorrelated losses (hurricanes, pandemics)
Unique RisksInsufficient similar exposures
Changing ConditionsHistorical data may not predict future

Examples of Law of Large Numbers

ScenarioIllustration
Life InsuranceMortality tables predict deaths accurately for large groups
Auto InsuranceAccident rates predictable across millions of drivers
Health InsuranceMedical costs more stable with larger insured populations
Property InsuranceFire losses follow patterns in large portfolios

Catastrophe Exception

IssueChallenge
HurricanesAffects thousands of policies at once
EarthquakesConcentrated geographic impact
PandemicsCorrelated health claims
SolutionReinsurance, catastrophe bonds, modeling

Exam Alert

Law of Large Numbers = a larger pool of similar, sufficiently independent exposures improves prediction of AVERAGE loss. It does not make individual losses predictable or guarantee an exact annual total. Common catastrophes and changing risk conditions limit the benefit. The standard error of an average shrinks with sample size; the standard deviation of individual losses does not.

Sources: NAIC actuarial principles, NAIC risk-pooling concepts, pp. 7-9 and NIST precision of a sample mean.

Study This Term In

Related Terms

Where shall we go next?

Your AI study companion · 10 free interactions a day

Stay Updated

Get free exam tips and study guides delivered to your inbox.

Free exam tips & study guides. Unsubscribe anytime.