Getting from Binomial to Poisson

In many binomial problems, the number of Bernoulli trials n is large, relatively speaking, and the probability of success p is small such that n p is of moderate magnitude. For example, consider problems that deal with rare events where the probability of occurrence is small (as a concrete example, counting the number of people with July 1 as birthday out of a random sample of 1000 people). It is often convenient to approximate such binomial problems using the Poisson distribution. The justification for using the Poisson approximation is that the Poisson distribution is a limiting case of the binomial distribution. Now that cheap computing power is widely available, it is quite easy to use computer or other computing devices to obtain exact binomial probabiities for experiments up to 1000 trials or more. Though the Poisson approximation may no longer be necessary for such problems, knowing how to get from binomial to Poisson is important for understanding the Poisson distribution itself.

Consider a counting process that describes the occurrences of a certain type of events of interest in a unit time interval subject to three simplifying assumptions (discussed below). We are interested in counting the number of occurrences of the event of interest in a unit time interval. As a concrete example, consider the number of cars arriving at an observation point in a certain highway in a period of time, say one hour. We wish to model the probability distribution of how many cars that will arrive at the observation point in this particular highway in one hour. Let X be the random variable described by this probability distribution. We wish to konw the probability that there are k cars arriving in one hour. We start with using a binomial distribution as an approximation to the probability P(X=k). We will see that upon letting n \rightarrow \infty, the P(X=k) is a Poisson probability.

Suppose that we know E(X)=\alpha, perhaps an average obtained after observing cars at the observation points for many hours. The simplifying assumptions alluded to earlier are the following:

  1. The numbers of cars arriving in nonoverlapping time intervals are independent.
  2. The probability of one car arriving in a very short time interval of length h is \alpha h.
  3. The probability of having more than one car arriving in a very short time interval is esstentially zero.

Assumption 1 means that a large number of cars arriving in one period does not imply fewer cars will arrival in the next period and vice versa. In other words, the number of cars that arrive in any one given moment does affect the number of cars that will arrive subsequently. Knowing how many cars arriving in one minute will not help predict the number of cars arriving at the next minute. Assumption 2 means that the rate of cars arriving is dependent only on the length of the time interval and not on when the time interval occurs (e.g. not on whether it is at the beginning of the hour or toward the end of the hour). The assumptions 2 and 3 allow us to think of a very short period of time as a Bernoulli trial. Thinking of the arrival of a car as a success, each short time interval will result in only one success or one failure.

To start, we can break up the hour into 60 minutes (into 60 Bernoulli trials). We then consider the binomial distribution with n=60 and p=\frac{\alpha}{60}. So the following is an approximation to our desired probability distribution.

\displaystyle (1) \ \ \ \ \ P(X=k) \approx \binom{60}{k} \biggl(\frac{\alpha}{60}\biggr)^k \biggr(1-\frac{\alpha}{60}\biggr)^{60-k} \ \ \ \ \ k=0,1,2,\cdots, 60

Conceivably, there can be more than 1 car arriving in a minute and observing cars in a one-minute interval may not be a Bernoulli trial. For a one-minute interval to qualify as a Bernoulli trial, there is either no car arriving or 1 car arriving in that one minute. So we can break up an hour into 3,600 seconds (into 3,600 Bernoulli trials). We now consider the binomial distribution with n=3600 and p=\frac{\alpha}{3600}.

\displaystyle (2) \ \ \ \ \ P(X=k) \approx \binom{3600}{k} \biggl(\frac{\alpha}{3600}\biggr)^k \biggr(1-\frac{\alpha}{3600}\biggr)^{3600-k} \ \ \ \ \ k=0,1,2,\cdots, 3600

It is also conceivable that more than 1 car can arrive in one second and observing cars in one-second interval may still not qualify as a Bernoulli trial. So we need to get more granular. We can divide up the hour into n equal subintervals, each of length \frac{1}{n}. The assumptions 2 and 3 ensure that each subinterval is a Bernoulli trial (either it is a success or a failure; one car arriving or no car arriving). Assumption 1 tells us that all the n subintervals are independent. So breaking up the hour into n moments and counting the number of moments that are successes will result in a binomial distribution with parameters n and p=\frac{\alpha}{n}. So we are ready to proceed with the following approximation to our probability distribution P(X=k).

\displaystyle (3) \ \ \ \ \ P(X=k) \approx \binom{n}{k} \biggl(\frac{\alpha}{n}\biggr)^k \biggr(1-\frac{\alpha}{n}\biggr)^{n-k} \ \ \ \ \ k=0,1,2,\cdots, n

As we get more granular, n \rightarrow \infty. We show that the limit of the binomial probability in (3) is the Poisson distribution with parameter \alpha. We show the following.

\displaystyle (4) \ \ \ \ \ P(X=k) = \lim \limits_{n \rightarrow \infty} \binom{n}{k} \biggl(\frac{\alpha}{n}\biggr)^k \biggr(1-\frac{\alpha}{n}\biggr)^{n-k}=\frac{e^{-\alpha} \alpha^k}{k!} \ \ \ \ \ \ k=0,1,2,\cdots

In the derivation of (4), we need the following two mathematical tools. The statement (5) is one of the definitions of the mathematical constant e. In the statement (6), the integer n in the numerator is greater than the integer k in the denominator. It says that whenever we work with such a ratio of two factorials, the result is the product of n with the smaller integers down to (n-(k-1)). There are exactly k terms in the product.

\displaystyle (5) \ \ \ \ \ \lim \limits_{n \rightarrow \infty} \biggl(1+\frac{x}{n}\biggr)^n=e^x

\displaystyle (6) \ \ \ \ \ \frac{n!}{(n-k)!}=n(n-1)(n-2) \cdots (n-k+1) \ \ \ \ \ \ \ \  k<n

The following is the derivation of (4).

\displaystyle \begin{aligned}(7) \ \ \ \ \  P(X=k)&=\lim \limits_{n \rightarrow \infty} \binom{n}{k} \biggl(\frac{\alpha}{n}\biggr)^k \biggr(1-\frac{\alpha}{n}\biggr)^{n-k} \\&=\lim \limits_{n \rightarrow \infty} \ \frac{n!}{k! (n-k)!} \biggl(\frac{\alpha}{n}\biggr)^k \biggr(1-\frac{\alpha}{n}\biggr)^{n-k} \\&=\lim \limits_{n \rightarrow \infty} \ \frac{n(n-1)(n-2) \cdots (n-k+1)}{n^k} \biggl(\frac{\alpha^k}{k!}\biggr) \biggr(1-\frac{\alpha}{n}\biggr)^{n} \biggr(1-\frac{\alpha}{n}\biggr)^{-k} \\&=\biggl(\frac{\alpha^k}{k!}\biggr) \lim \limits_{n \rightarrow \infty} \ \frac{n(n-1)(n-2) \cdots (n-k+1)}{n^k} \\&\times \ \ \ \lim \limits_{n \rightarrow \infty} \biggr(1-\frac{\alpha}{n}\biggr)^{n} \ \lim \limits_{n \rightarrow \infty} \biggr(1-\frac{\alpha}{n}\biggr)^{-k} \\&=\frac{e^{-\alpha} \alpha^k}{k!} \end{aligned}

In (7), we have \displaystyle \lim \limits_{n \rightarrow \infty} \ \frac{n(n-1)(n-2) \cdots (n-k+1)}{n^k}=1. The reason being that the numerator is a polynomial where the leading term is n^k. Upon dividing by n^k and taking the limit, we get 1. Based on (5), we have \displaystyle \lim \limits_{n \rightarrow \infty} \biggr(1-\frac{\alpha}{n}\biggr)^{n}=e^{-\alpha}. For the last limit in the derivation we have \displaystyle \lim \limits_{n \rightarrow \infty} \biggr(1-\frac{\alpha}{n}\biggr)^{-k}=1.

We conclude with some comments. As the above derivation shows, the Poisson distribution is at heart a binomial distribution. When we divide the unit time interval into more and more subintervals (as the subintervals get more and more granular), the resulting binomial distribution behaves more and more like the Poisson distribution.

The three assumtions used in the derivation are called the Poisson postulates, which are the underlying assumptions that govern a Poisson process. Such a random process describes the occurrences of some type of events that are of interest (e.g. the arrivals of cars in our example) in a fixed period of time. The positive constant \alpha indicated in Assumption 2 is the parameter of the Poisson process, which can be interpreted as the rate of occurrences of the event of interest (or rate of changes, or rate of arrivals) in a unit time interval, meaning that the positive constant \alpha is the mean number of occurrences in the unit time interval. The derivation in (7) shows that whenever a certain type of events occurs according to a Poisson process with parameter \alpha, the counting variable of the number of occurrences in the unit time interval is distributed according to the Poisson distribution as indicated in (4).

If we observe the occurrences of events over intervals of length other than unit length, say, in an interval of length t, the counting process is governed by the same three postulates, with the modification to Assumption 2 that the rate of changes of the process is now \alpha t. The mean number of occurrences in the time interval of length t is now \alpha t. The Assumption 2 now states that for any very short time interval of length h (and that is also a subinterval of the interval of length t under observation), the probability of having one occurrence of event in this short interval is (\alpha t)h. Applyng the same derivation, it can be shown that the number of occurrences (X_t) in a time interval of length t has the Poisson distribution with the following probability mass function.

\displaystyle (8) \ \ \ \ \ P(X_t=k)=\frac{e^{-\alpha t} \ (\alpha t)^k}{k!} \ \ \ \ \ \ \ \  k=0,1,2,\cdots

Advertisements

Relating Binomial and Negative Binomial

The negative binomial distribution has a natural intepretation as a waiting time until the arrival of the rth success (when the parameter r is a positive integer). The waiting time refers to the number of independent Bernoulli trials needed to reach the rth success. This interpretation of the negative binomial distribution gives us a good way of relating it to the binomial distribution. For example, if the rth success takes place after k failed Bernoulli trials (for a total of k+r trials), then there can be at most r-1 successes in the first k+r trials. This tells us that the survival function of the negative binomial distribution is the cumulative distribution function (cdf) of a binomial distribution. In this post, we gives the details behind this observation. A previous post on the negative binomial distribution is found here.

A random experiment resulting in two distinct outcomes (success or failure) is called a Bernoulli trial (e.g. head or tail in a coin toss, whether or not the birthday of a customer is the first of January, whether an insurance claim is above or below a given threshold etc). Suppose a series of independent Bernoulli trials are performed until reaching the rth success where the probability of success in each trial is p. Let X_r be the number of failures before the occurrence of the rth success. The following is the probablity mass function of X_r.

\displaystyle (1) \ \ \ \ P(X_r)=\binom{k+r-1}{k} p^r (1-p)^k \ \ \ \ \ \ k=0,1,2,3,\cdots

Be definition, the survival function and cdf of X_r are:

\displaystyle (2) \ \ \ \ P(X_r > k)=\sum \limits_{j=k+1}^\infty \binom{j+r-1}{j} p^r (1-p)^j \ \ \ \ \ \ k=0,1,2,3,\cdots

\displaystyle (3) \ \ \ \ P(X_r \le k)=\sum \limits_{j=0}^k \binom{j+r-1}{j} p^r (1-p)^j \ \ \ \ \ \ k=0,1,2,3,\cdots

For each positive integer k, let Y_{r+k} be the number of successes in performing a sequence of r+k independent Bernoulli trials where p is the probability of success. In other words, Y_{r+k} has a binomial distribution with parameters r+k and p.

If the random experiment requires more than k failures to reach the rth success, there are at most r-1 successes in the first k+r trails. Thus the survival function of X_r is the same as the cdf of a binomial distribution. Equivalently, the cdf of X_r is the same as the survival function of a binomial distribution. We have the following:

\displaystyle \begin{aligned}(4) \ \ \ \ P(X_r > k)&=P(Y_{k+r} \le r-1) \\&=\sum \limits_{j=0}^{r-1} \binom{k+r}{j} p^j (1-p)^{k+r-j} \ \ \ \ \ \ k=0,1,2,3,\cdots \end{aligned}

\displaystyle \begin{aligned}(5) \ \ \ \ P(X_r \le k)&=P(Y_{k+r} > r-1) \ \ \ \ \ \ k=0,1,2,3,\cdots \end{aligned}

Remark
The relation (4) is analogous to the relationship between the Gamma distribution and the Poisson distribution. Recall that a Gamma distribution with shape parameter \alpha and scale parameter n, where n is a positive integer, can be interpreted as the waiting time until the nth change in a Poisson process. Thus, if the nth change takes place after time t, there can be at most n-1 arrivals in the time interval [0,t]. Thus the survival function of this Gamma distribution is the same as the cdf of a Poisson distribution. The relation (4) is analogous to the following relation.

\displaystyle (5) \ \ \ \ \int_t^\infty \frac{\alpha^n}{(n-1)!} \ x^{n-1} \ e^{-\alpha x} \ dx=\sum \limits_{j=0}^{n-1} \frac{e^{-\alpha t} \ (\alpha t)^j}{j!}

A previous post on the negative binomial distribution is found here.

The Negative Binomial Distribution

A counting distribution is a discrete distribution with probabilities only on the nonnegative integers. Such distributions are important in insurance applications since they can be used to model the number of events such as losses to the insured or claims to the insurer. Though playing a prominent role in statistical theory, the Poisson distribution is not appropriate in all situations, since it requires that the mean and the variance are equaled. Thus the negative binomial distribution is an excellent alternative to the Poisson distribution, especially in the cases where the observed variance is greater than the observed mean.

The negative binomial distribution arises naturally from a probability experiment of performing a series of independent Bernoulli trials until the occurrence of the rth success where r is a positive integer. From this starting point, we discuss three ways to define the distribution. We then discuss several basic properties of the negative binomial distribution. Emphasis is placed on the close connection between the Poisson distribution and the negative binomial distribution.

________________________________________________________________________

Definitions
We define three versions of the negative binomial distribution. The first two versions arise from the view point of performing a series of independent Bernoulli trials until the rth success where r is a positive integer. A Bernoulli trial is a probability experiment whose outcome is random such that there are two possible outcomes (success or failure).

Let X_1 be the number of Bernoulli trials required for the rth success to occur where r is a positive integer. Let p is the probability of success in each trial. The following is the probability function of X_1:

\displaystyle (1) \ \ \ \ \ P(X_1=x)= \binom{x-1}{r-1} p^r (1-p)^{x-r} \ \ \ \ \ \ \ x=r,r+1,r+2,\cdots

The idea for (1) is that for X_1=x to happen, there must be r-1 successes in the first x-1 trials and one additional success occurring in the last trial (the xth trial).

A more common version of the negative binomial distribution is the number of Bernoulli trials in excess of r in order to produce the rth success. In other words, we consider the number of failures before the occurrence of the rth success. Let X_2 be this random variable. The following is the probability function of X_2:

\displaystyle (2) \ \ \ \ \ P(X_2=x)=\binom{x+r-1}{x} p^r (1-p)^x \ \ \ \ \ \ \ x=0,1,2,\cdots

The idea for (2) is that there are x+r trials and in the first x+r-1 trials, there are x failures (or equivalently r-1 successes).

In both (1) and (2), the binomial coefficient is defined by

\displaystyle (3) \ \ \ \ \ \binom{y}{k}=\frac{y!}{k! \ (y-k)!}=\frac{y(y-1) \cdots (y-(k-1))}{k!}

where y is a positive integer and k is a nonnegative integer. However, the right-hand-side of (3) can be calculated even if y is not a positive integer. Thus the binomial coefficient \displaystyle \binom{y}{k} can be expanded to work for all real number y. However k must still be nonnegative integer.

\displaystyle (4) \ \ \ \ \ \binom{y}{k}=\frac{y(y-1) \cdots (y-(k-1))}{k!}

For convenience, we let \displaystyle \binom{y}{0}=1. When the real number y>k-1, the binomial coefficient in (4) can be expressed as:

\displaystyle (5) \ \ \ \ \ \binom{y}{k}=\frac{\Gamma(y+1)}{\Gamma(k+1) \Gamma(y-k+1)}

where \Gamma(\cdot) is the gamma function.

With the more relaxed notion of binomial coefficient, the probability function in (2) above can be defined for all real number r. Thus the general version of the negative binomial distribution has two parameters r and p, both real numbers, such that 0<p<1. The following is its probability function.

\displaystyle (6) \ \ \ \ \ P(X=x)=\binom{x+r-1}{x} p^r (1-p)^x \ \ \ \ \ \ \ x=0,1,2,\cdots

Whenever r in (6) is a real number that is not a positive integer, the interpretation of counting the number of failures until the occurrence of the rth success is no longer important. Instead we can think of it simply as a count distribution.

The following alternative parametrization of the negative binomial distribution is also useful.

\displaystyle (6a) \ \ \ \ \ P(X=x)=\binom{x+r-1}{x} \biggl(\frac{\alpha}{\alpha+1}\biggr)^r \biggl(\frac{1}{\alpha+1}\biggr)^x \ \ \ \ \ \ \ x=0,1,2,\cdots

The parameters in this alternative parametrization are r and \alpha>0. Clearly, the ratio \frac{\alpha}{\alpha+1} takes the place of p in (6). Unless stated otherwise, we use the parametrization of (6).
________________________________________________________________________

What is negative about the negative binomial distribution?
What is negative about this distribution? What is binomial about this distribution? The name is suggested by the fact that the binomial coefficient in (6) can be rearranged as follows:

\displaystyle \begin{aligned}(7) \ \ \ \ \ \binom{x+r-1}{x}&=\frac{(x+r-1)(x+r-2) \cdots r}{x!} \\&=(-1)^x \frac{(-r-(x-1))(-r-(x-2)) \cdots (-r)}{x!} \\&=(-1)^x \frac{(-r)(-r-1) \cdots (-r-(x-1))}{x!} \\&=(-1)^x \binom{-r}{x} \end{aligned}

The calculation in (7) can be used to verify that (6) is indeed a probability function, that is, all the probabilities sum to 1.

\displaystyle \begin{aligned}(8) \ \ \ \ \ 1&=p^r p^{-r}\\&=p^r (1-q)^{-r} \\&=p^r \sum \limits_{x=0}^\infty \binom{-r}{x} (-q)^x \ \ \ \ \ \ \ \ (8.1) \\&=p^r \sum \limits_{x=0}^\infty (-1)^x \binom{-r}{x} q^x \\&=\sum \limits_{x=0}^\infty \binom{x+r-1}{x} p^r q^x \end{aligned}

In (8), we take q=1-p. The step (8.1) above uses the following formula known as the Newton’s binomial formula.

\displaystyle (9) \ \ \ \ \ (1+t)^w=\sum \limits_{k=0}^\infty \binom{w}{k} t^k

________________________________________________________________________

The Generating Function
By definition, the following is the generating function of the negative binomial distribution, using :

\displaystyle (10) \ \ \ \ \ g(z)=\sum \limits_{x=0}^\infty \binom{r+x-1}{x} p^r q^x z^x

where q=1-p. Using a similar calculation as in (8), the generating function can be simplified as:

\displaystyle (11) \ \ \ \ \ g(z)=p^r (1-q z)^{-r}=\frac{p^r}{(1-q z)^r}=\frac{p^r}{(1-(1-p) z)^r}; \ \ \ \ \ z<\frac{1}{1-p}

As a result, the moment generating function of the negative binomial distribution is:

\displaystyle (12) \ \ \ \ \ M(t)=\frac{p^r}{(1-(1-p) e^t)^r}; \ \ \ \ \ \ \ t<-ln(1-p)

________________________________________________________________________

Independent Sum

One useful property of the negative binomial distribution is that the independent sum of negative binomial random variables, all with the same parameter p, also has a negative binomial distribution. Let Y=Y_1+Y_2+\cdots+Y_n be an independent sum such that each X_i has a negative binomial distribution with parameters r_i and p. Then the sum Y=Y_1+Y_2+\cdots+Y_n has a negative binomial distribution with parameters r=r_1+\cdots+r_n and p.

Note that the generating function of an independent sum is the product of the individual generating functions. The following shows that the product of the individual generating functions is of the same form as (11), thus proving the above assertion.

\displaystyle (13) \ \ \ \ \ h(z)=\frac{p^{\sum \limits_{i=1}^n r_i}}{(1-(1-p) z)^{\sum \limits_{i=1}^n r_i}}
________________________________________________________________________

Mean and Variance
The mean and variance can be obtained from the generating function. From E(X)=g'(1) and E(X^2)=g'(1)+g^{(2)}(1), we have:

\displaystyle (14) \ \ \ \ \ E(X)=\frac{r(1-p)}{p} \ \ \ \ \ \ \ \ \ \ \ \ \ Var(X)=\frac{r(1-p)}{p^2}

Note that Var(X)=\frac{1}{p} E(X)>E(X). Thus when the sample data suggest that the variance is greater than the mean, the negative binomial distribution is an excellent alternative to the Poisson distribution. For example, suppose that the sample mean and the sample variance are 3.6 and 7.1. In exploring the possibility of fitting the data using the negative binomial distribution, we would be interested in the negative binomial distribution with this mean and variance. Then plugging these into (14) produces the negative binomial distribution with r=3.7 and p=0.507.
________________________________________________________________________

The Poisson-Gamma Mixture
One important application of the negative binomial distribution is that it is a mixture of a family of Poisson distributions with Gamma mixing weights. Thus the negative binomial distribution can be viewed as a generalization of the Poisson distribution. The negative binomial distribution can be viewed as a Poisson distribution where the Poisson parameter is itself a random variable, distributed according to a Gamma distribution. Thus the negative binomial distribution is known as a Poisson-Gamma mixture.

In an insurance application, the negative binomial distribution can be used as a model for claim frequency when the risks are not homogeneous. Let N has a Poisson distribution with parameter \theta, which can be interpreted as the number of claims in a fixed period of time from an insured in a large pool of insureds. There is uncertainty in the parameter \theta, reflecting the risk characteristic of the insured. Some insureds are poor risks (with large \theta) and some are good risks (with small \theta). Thus the parameter \theta should be regarded as a random variable \Theta. The following is the conditional distribution of N (conditional on \Theta=\theta):

\displaystyle (15) \ \ \ \ \ P(N=n \lvert \Theta=\theta)=\frac{e^{-\theta} \ \theta^n}{n!} \ \ \ \ \ \ \ \ \ \ n=0,1,2,\cdots

Suppose that \Theta has a Gamma distribution with scale parameter \alpha and shape parameter \beta. The following is the probability density function of \Theta.

\displaystyle (16) \ \ \ \ \ g(\theta)=\frac{\alpha^\beta}{\Gamma(\beta)} \theta^{\beta-1} e^{-\alpha \theta} \ \ \ \ \ \ \ \ \ \ \theta>0

Then the joint density of N and \Theta is:

\displaystyle (17) \ \ \ \ \ P(N=n \lvert \Theta=\theta) \ g(\theta)=\frac{e^{-\theta} \ \theta^n}{n!} \ \frac{\alpha^\beta}{\Gamma(\beta)} \theta^{\beta-1} e^{-\alpha \theta}

The unconditional distribution of N is obtained by summing out \theta in (17).

\displaystyle \begin{aligned}(18) \ \ \ \ \ P(N=n)&=\int_0^\infty P(N=n \lvert \Theta=\theta) \ g(\theta) \ d \theta \\&=\int_0^\infty \frac{e^{-\theta} \ \theta^n}{n!} \ \frac{\alpha^\beta}{\Gamma(\beta)} \ \theta^{\beta-1} \ e^{-\alpha \theta} \ d \theta \\&=\int_0^\infty \frac{\alpha^\beta}{n! \ \Gamma(\beta)} \ \theta^{n+\beta-1} \ e^{-(\alpha+1) \theta} d \theta \\&=\frac{\alpha^\beta}{n! \ \Gamma(\beta)} \ \frac{\Gamma(n+\beta)}{(\alpha+1)^{n+\beta}} \int_0^\infty \frac{(\alpha+1)^{n+\beta}}{\Gamma(n+\beta)} \theta^{n+\beta-1} \ e^{-(\alpha+1) \theta} d \theta \\&=\frac{\alpha^\beta}{n! \ \Gamma(\beta)} \ \frac{\Gamma(n+\beta)}{(\alpha+1)^{n+\beta}} \\&=\frac{\Gamma(n+\beta)}{\Gamma(n+1) \ \Gamma(\beta)} \ \biggl( \frac{\alpha}{\alpha+1}\biggr)^\beta \ \biggl(\frac{1}{\alpha+1}\biggr)^n \\&=\binom{n+\beta-1}{n} \ \biggl( \frac{\alpha}{\alpha+1}\biggr)^\beta \ \biggl(\frac{1}{\alpha+1}\biggr)^n \ \ \ \ \ \ \ \ \ n=0,1,2,\cdots \end{aligned}

Note that the integral in the fourth step in (18) is 1.0 since the integrand is the pdf of a Gamma distribution. The above probability function is that of a negative binomial distribution. It is of the same form as (6a). Equivalently, it is also of the form (6) with parameter r=\beta and p=\frac{\alpha}{\alpha+1}.

The variance of the negative binomial distribution is greater than the mean. In a Poisson distribution, the mean equals the variance. Thus the unconditional distribution of N is more dispersed than its conditional distributions. This is a characteristic of mixture distributions. The uncertainty in the parameter variable \Theta has the effect of increasing the unconditional variance of the mixture distribution of N. The variance of a mixture distribution has two components, the weighted average of the conditional variances and the variance of the conditional means. The second component represents the additional variance introduced by the uncertainty in the parameter \Theta (see The variance of a mixture).

________________________________________________________________________

The Poisson Distribution as Limit of Negative Binomial
There is another connection to the Poisson distribution, that is, the Poisson distribution is a limiting case of the negative binomial distribution. We show that the generating function of the Poisson distribution can be obtained by taking the limit of the negative binomial generating function as r \rightarrow \infty. Interestingly, the Poisson distribution is also the limit of the binomial distribution.

In this section, we use the negative binomial parametrization of (6a). By replacing \frac{\alpha}{\alpha+1} for p, the following are the mean, variance, and the generating function for the probability function in (6a):

\displaystyle \begin{aligned}(19) \ \ \ \ \ \ &E(X)=\frac{r}{\alpha} \\&\text{ }\\&Var(X)=\frac{\alpha+1}{\alpha} \ \frac{r}{\alpha}=\frac{r(\alpha+1)}{\alpha^2} \\&\text{ } \\&g(z)=\frac{1}{[1-\frac{1}{\alpha}(z-1)]^r} \ \ \ \ \ \ \ z<\alpha+1 \end{aligned}

Let r goes to infinity and \displaystyle \frac{1}{\alpha} goes to zero and at the same time keeping their product constant. Thus \displaystyle \mu=\frac{r}{\alpha} is constant (this is the mean of the negative binomial distribution). We show the following:

\displaystyle (20) \ \ \ \ \ \lim \limits_{r \rightarrow \infty} [1-\frac{\mu}{r}(z-1)]^{-r}=e^{\mu (z-1)}

The right-hand side of (20) is the generating function of the Poisson distribution with mean \mu. The generating function in the left-hand side is that of a negative binomial distribution with mean \displaystyle \mu=\frac{r}{\alpha}. The following is the derivation of (20).

\displaystyle \begin{aligned}(21) \ \ \ \ \ \lim \limits_{r \rightarrow \infty} [1-\frac{\mu}{r}(z-1)]^{-r}&=\lim \limits_{r \rightarrow \infty} e^{\displaystyle \biggl(ln[1-\frac{\mu}{r}(z-1)]^{-r}\biggr)} \\&=\lim \limits_{r \rightarrow \infty} e^{\displaystyle \biggl(-r \ ln[1-\frac{\mu}{r}(z-1)]\biggr)} \\&=e^{\displaystyle \biggl(\lim \limits_{r \rightarrow \infty} -r \ ln[1-\frac{\mu}{r}(z-1)]\biggr)} \end{aligned}

We now focus on the limit in the exponent.

\displaystyle \begin{aligned}(22) \ \ \ \ \ \lim \limits_{r \rightarrow \infty} -r \ ln[1-\frac{\mu}{r}(z-1)]&=\lim \limits_{r \rightarrow \infty} \frac{ln(1-\frac{\mu}{r} (z-1))^{-1}}{r^{-1}} \\&=\lim \limits_{r \rightarrow \infty} \frac{(1-\frac{\mu}{r} (z-1)) \ \mu (z-1) r^{-2}}{r^{-2}} \\&=\mu (z-1) \end{aligned}

The middle step in (22) uses the L’Hopital’s Rule. The result in (20) is obtained by combining (21) and (22).

________________________________________________________________________

Reference

  1. Klugman S.A., Panjer H. H., Wilmot G. E. Loss Models, From Data to Decisions, Second Edition., Wiley-Interscience, a John Wiley & Sons, Inc., New York, 2004

Splitting a Poisson Distribution

We consider a remarkable property of the Poisson distribution that has a connection to the multinomial distribution. We start with the following examples.

Example 1
Suppose that the arrivals of customers in a gift shop at an airport follow a Poisson distribution with a mean of \alpha=5 per 10 minutes. Furthermore, suppose that each arrival can be classified into one of three distinct types – type 1 (no purchase), type 2 (purchase under $20), and type 3 (purchase over $20). Records show that about 25% of the customers are of type 1. The percentages of type 2 and type 3 are 60% and 15%, respectively. What is the probability distribution of the number of customers per hour of each type?

Example 2
Roll a fair die N times where N is random and follows a Poisson distribution with parameter \alpha. For each i=1,2,3,4,5,6, let N_i be the number of times the upside of the die is i. What is the probability distribution of each N_i? What is the joint distribution of N_1,N_2,N_3,N_4,N_5,N_6?

In Example 1, the stream of customers arrive according to a Poisson distribution. It can be shown that the stream of each type of customers also has a Poisson distribution. One way to view this example is that we can split the Poisson distribution into three Poisson distributions.

Example 2 also describes a splitting process, i.e. splitting a Poisson variable into 6 different Poisson variables. We can also view Example 2 as a multinomial distribution where the number of trials is not fixed but is random and follows a Poisson distribution. If the number of rolls of the die is fixed in Example 2 (say 10), then each N_i would be a binomial distribution. Yet, with the number of trials being Poisson, each N_i has a Poisson distribution with mean \displaystyle \frac{\alpha}{6}. In this post, we describe this Poisson splitting process in terms of a “random” multinomial distribution (the view point of Example 2).

________________________________________________________________________

Suppose we have a multinomial experiment with parameters N, r, p_1, \cdots, p_r, where

  • N is the number of multinomial trials,
  • r is the number of distinct possible outcomes in each trial (type 1 through type r),
  • the p_i are the probabilities of the r possible outcomes in each trial.

Suppose that N follows a Poisson distribution with parameter \alpha. For each i=1, \cdots, r, let N_i be the number of occurrences of the i^{th} type of outcomes in the N trials. Then N_1,N_2,\cdots,N_r are mutually independent Poisson random variables with parameters \alpha p_1,\alpha p_2,\cdots,\alpha p_r, respectively.

The variables N_1,N_2,\cdots,N_r have a multinomial distribution and their joint probability function is:

\displaystyle (1) \ \ \ \ P(N_1=n_1,N_2=n_2,\cdots,N_r=n_r)=\frac{N!}{n_1! n_2! \cdots n_r!} \ p_1^{n_1} p_2^{n_2} \cdots p_r^{n_r}

where n_i are nonnegative integers such that N=n_1+n_2+\cdots+n_r.

Since the total number of multinomial trials N is not fixed and is random, (1) is not the end of the story. The following is the joint probability function of N_1,N_2,\cdots,N_r:

\displaystyle \begin{aligned}(2) \ \ \ \ P(N_1=n_1,N_2=n_2,\cdots,N_r=n_r)&=P(N_1=n_1,N_2=n_2,\cdots,N_r=n_r \lvert N=\sum \limits_{k=0}^r n_k) \\&\ \ \ \ \ \times P(N=\sum \limits_{k=0}^r n_k) \\&\text{ } \\&=\frac{(\sum \limits_{k=0}^r n_k)!}{n_1! \ n_2! \ \cdots \ n_r!} \ p_1^{n_1} \ p_2^{n_2} \ \cdots \ p_r^{n_r} \ \times \frac{e^{-\alpha} \alpha^{\sum \limits_{k=0}^r n_k}}{(\sum \limits_{k=0}^r n_k)!} \\&\text{ } \\&=\frac{e^{-\alpha p_1} \ (\alpha p_1)^{n_1}}{n_1!} \ \frac{e^{-\alpha p_2} \ (\alpha p_2)^{n_2}}{n_2!} \ \cdots \ \frac{e^{-\alpha p_r} \ (\alpha p_r)^{n_r}}{n_r!} \end{aligned}

To obtain the marginal probability function of N_j, j=1,2,\cdots,r, we sum out the other variables N_k=n_k (k \ne j) in (2) and obtain the following:

\displaystyle (3) \ \ \ \ P(N_j=n_j)=\frac{e^{-\alpha p_j} \ (\alpha p_j)^{n_j}}{n_j!}

Thus we can conclude that N_j, j=1,2,\cdots,r, has a Poisson distribution with parameter \alpha p_j. Furrthermore, the joint probability function of N_1,N_2,\cdots,N_r is the product of the marginal probability functions. Thus we can conclude that N_1,N_2,\cdots,N_r are mutually independent.

________________________________________________________________________
Example 1
Let N_1,N_2,N_3 be the number of customers per hour of type 1, type 2, and type 3, respectively. Here, we attempt to split a Poisson distribution with mean 30 per hour (based on 5 per 10 minutes). Thus N_1,N_2,N_3 are mutually independent Poisson variables with means 30 \times 0.25=7.5, 30 \times 0.60=18, 30 \times 0.15=4.5, respectively.

Example 2
As indicated earlier, each N_i, i=1,2,3,4,5,6, has a Poisson distribution with mean \frac{\alpha}{6}. According to (2), the joint probability function of N_1,N_2,N_3,N_4,N_5,N_6 is simply the product of the six marginal Poisson probability functions.

The Poisson Distribution

Let \alpha be a positive constant. Consider the following probability distribution:

\displaystyle (1) \ \ \ \ \ P(X=j)=\frac{e^{-\alpha} \alpha^j}{j!} \ \ \ \ \ j=0,1,2,\cdots

The above distribution is said to be a Poisson distribution with parameter \alpha. The Poisson distribution is usually used to model the random number of events occurring in a fixed time interval. As will be shown below, E(X)=\alpha. Thus the parameter \alpha is the rate of occurrence of the random events; it indicates on average how many events occur per unit of time. Examples of random events that may be modeled by the Poisson distribution include the number of alpha particles emitted by a radioactive substance counted in a prescribed area during a fixed period of time, the number of auto accidents in a fixed period of time or the number of losses arising from a group of insureds during a policy period.

Each of the above examples can be thought of as a process that generates a number of arrivals or changes in a fixed period of time. If such a counting process leads to a Poisson distribution, then the process is said to be a Poisson process.

We now discuss some basic properties of the Poisson distribution. Using the Taylor series expansion of e^{\alpha}, the following shows that (1) is indeed a probability distribution.

\displaystyle . \ \ \ \ \ \ \ \sum \limits_{j=0}^\infty \frac{e^{-\alpha} \alpha^j}{j!}=e^{-\alpha} \sum \limits_{j=0}^\infty \frac{\alpha^j}{j!}=e^{-\alpha} e^{\alpha}=1

The generating function of the Poisson distribution is g(z)=e^{\alpha (z-1)} (see The generating function). The mean and variance can be calculated using the generating function.

\displaystyle \begin{aligned}(2) \ \ \ \ \ &E(X)=g'(1)=\alpha \\&\text{ } \\&E[X(X-1)]=g^{(2)}(1)=\alpha^2 \\&\text{ } \\&Var(X)=E[X(X-1)]+E(X)-E(X)^2=\alpha^2+\alpha-\alpha^2=\alpha \end{aligned}

The Poisson distribution can also be interpreted as an approximation to the binomial distribution. It is well known that the Poisson distribution is the limiting case of binomial distributions (see [1] or this post).

\displaystyle (3) \ \ \ \ \ \lim \limits_{n \rightarrow \infty} \binom{n}{j} \biggl(\frac{\alpha}{n}\biggr)^j \biggl(1-\frac{\alpha}{n}\biggr)^{n-j}=\frac{e^{-\alpha} \alpha^j}{j!}

One application of (3) is that we can use Poisson probabilities to approximate Binomial probabilities. The approximation is reasonably good when the number of trials n in a binomial distribution is large and the probability of success p is small. The binomial mean is n p and the variance is n p (1-p). When p is small, 1-p is close to 1 and the binomial variance is approximately np \approx n p (1-p). Whenever the mean of a discrete distribution is approximately equaled to the mean, the Poisson approximation is quite good. As a rule of thumb, we can use Poisson to approximate binomial if n \le 100 and p \le 0.01.

As an example, we use the Poisson distribution to estimate the probability that at most 1 person out of 1000 will have a birthday on the New Year Day. Let n=1000 and p=365^{-1}. So we use the Poisson distribution with \alpha=1000 \times 365^{-1}. The following is an estimate using the Poisson distribution.

\displaystyle . \ \ \ \ \ \ \ P(X \le 1)=e^{-\alpha}+\alpha e^{-\alpha}=(1+\alpha) e^{-\alpha}=0.2415

Another useful property is that the independent sum of Poisson distributions also has a Poisson distribution. Specifically, if each X_i has a Poisson distribution with parameter \alpha_i, then the independent sum X=X_1+\cdots+X_n has a Poisson distribution with parameter \alpha=\alpha_1+\cdots+\alpha_n. One way to see this is that the product of Poisson generating functions has the same general form as g(z)=e^{\alpha (z-1)} (see The generating function). One interpretation of this property is that when merging several arrival processes, each of which follow a Poisson distribution, the result is still a Poisson distribution.

For example, suppose that in an airline ticket counter, the arrival of first class customers follows a Poisson process with a mean arrival rate of 8 per 15 minutes and the arrival of customers flying coach follows a Poisson distribution with a mean rate of 12 per 15 minutes. Then the arrival of customers of either types has a Poisson distribution with a mean rate of 20 per 15 minutes or 80 per hour.

A Poisson distribution with a large mean can be thought of as an independent sum of Poisson distributions. For example, a Poisson distribution with a mean of 50 is the independent sum of 50 Poisson distributions each with mean 1. Because of the central limit theorem, when the mean is large, we can approximate the Poisson using the normal distribution.

In addition to merging several Poisson distributions into one combined Poisson distribution, we can also split a Poisson into several Poisson distributions. For example, suppose that a stream of customers arrives according to a Poisson distribution with parameter \alpha and each customer can be classified into one of two types (e.g. no purchase vs. purchase) with probabilities p_1 and p_2, respectively. Then the number of “no purchase” customers and the number of “purchase” customers are independent Poisson random variables with parameters \alpha p_1 and \alpha p_2, respectively. For more details on the splitting of Poisson, see Splitting a Poisson Distribution.

Reference

  1. Feller W. An Introduction to Probability Theory and Its Applications, Third Edition, John Wiley & Sons, New York, 1968

The hazard rate function

In this post, we introduce the hazard rate function using the notions of non-homogeneous Poisson process.

In a Poisson process, changes occur at a constant rate \lambda per unit time. Suppose that we interpret the changes in a Poisson process from a mortality point of view, i.e. a change in the Poisson process mean a termination of a system, be it biological or manufactured, and this Poisson process counts the number of of terminations as they occur. Then the rate of change \lambda is interpreted as a hazard rate (or failure rate or force of mortality). With a constant force of mortality, the time until the next change is exponentially distributed. In this post, we discuss the hazard rate function in a more general setting. The process that counts of the number of terminations will no longer have a constant hazard rate, and instead will have a hazard rate function \lambda(t), a function of time t. Such a counting process is called a non-homogeneous Poisson process. We discuss the survival probability models (the time to the next termination) associated with a non-homogeneous Poisson process. We then discuss several important examples of survival probability models, including the Weibull distribution, the Gompertz distribution and the model based on the Makeham’s law. We aso comment briefly the connection between the hazard rate function and the tail weight of a distribution.

\text{ }

The Poisson Process
We start with the three postulates of a Poisson process. Consider an experiment in which the occurrences of a certain type of events are counted during a given time interval. We call the occurrence of the type of events in question a change. We assume the following three conditions:

  1. The numbers of changes occurring in nonoverlapping intervals are independent.
  2. The probability of two or more changes taking place in a sufficiently small interval is essentially zero.
  3. The probability of exactly one change in the short interval (t,t+\delta) is approximately \lambda \delta where \delta is sufficiently small and \lambda is a positive constant.

\text{ }

When we interpret the Poisson process in a mortality point of view, the constant \lambda is a hazard rate (or force of mortality), which can be interpreted as the rate of failure at the next instant given that the life has survived to time t. With a constant force of mortality, the survival model (the time until the next termination) has an exponential distribution with mean \frac{1}{\lambda}. We wish to relax the constant force of mortality assumption by making \lambda a function of t instead. The remainder of this post is based on the non-homogeneous Poisson process defined below.

\text{ }

The Non-Homogeneous Poisson Process
We modifiy condition 3 above by making \lambda(t) a function of t. We have the following modified counting process.

  1. The numbers of changes occurring in nonoverlapping intervals are independent.
  2. The probability of two or more changes taking place in a sufficiently small interval is essentially zero.
  3. The probability of exactly one change in the short interval (t,t+\delta) is approximately \lambda(t) \delta where \delta is sufficiently small and \lambda(t) is a nonnegative function of t.

\text{ }

We focus on the survival model aspect of such counting processes. Such process can be interpreted as models for the number of changes occurred in a time interval where a change means “termination” or ‘failure” of a system under consideration. The rate of change function \lambda(t) indicated in condition 3 is called the hazard rate function. It is also called the failure rate function in reliability engineering and the force of mortality in life contingency theory.

Based on condition 3 in the non-homogeneous Poisson process, the hazard rate function \lambda(t) can be interpreted as the rate of failure at the next instant given that the life has survived to time t.

Two random variables naturally arise from a non-homogeneous Poisson process are described here. One is the discrete variable N_t, defined as the number of changes in the time interval (0,t). The other is the continuous random variable T, defined as the time until the occurrence of the first change. The probability distribution of T is called a survival model. The following is the link between N_t and T.

\text{ }

\displaystyle \begin{aligned}(1) \ \ \ \ \ \ \ \ \ &P[T > t]=P[N_t=0] \end{aligned}

\text{ }

Note that P[T > t] is the probability that the next change occurs after time t. This means that there is no change within the interval (0,t). We have the following theorems.

\text{ }

Theorem 1.
Let \displaystyle \Lambda(t)=\int_{0}^{t} \lambda(y) dy. Then e^{-\Lambda(t)} is the probability that there is no change in the interval (0,t). That is, \displaystyle P[N_t=0]=e^{-\Lambda(t)}.

Proof. We are interested in finding the probability of zero changes in the interval (0,y+\delta). By condition 1, the numbers of changes in the nonoverlapping intervals (0,y) and (y,y+\delta) are independent. Thus we have:

\text{ }

\displaystyle (2) \ \ \ \ \ \ \ \ P[N_{y+\delta}=0] \approx P[N_y=0] \times [1-\lambda(y) \delta]

\text{ }

Note that by condition 3, the probability of exactly one change in the small interval (y,y+\delta) is \lambda(y) \delta. Thus [1-\lambda(y) \delta] is the probability of no change in the interval (y,y+\delta). Continuing with equation (2), we have the following derivation:

\text{ }

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &\frac{P[N_{y+\delta}=0] - P[N_y=0]}{\delta} \approx -\lambda(y) P[N_y=0] \\&\text{ } \\&\frac{d}{dy} P[N_y=0]=-\lambda(y) P[N_y=0] \\&\text{ } \\&\frac{\frac{d}{dy} P[N_y=0]}{P[N_y=0]}=-\lambda(y) \\&\text{ } \\&\int_0^{t} \frac{\frac{d}{dy} P[N_y=0]}{P[N_y=0]} dy=-\int_0^{t} \lambda(y)dy \end{aligned}

\text{ }

Evaluating the integral on the left hand side with the boundary condition of P[N_0=0]=1 produces the following results:

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &ln P[N_t=0]=-\int_0^{t} \lambda(y)dy \\&\text{ } \\&P[N_t=0]=\displaystyle e^{\displaystyle -\int_0^{t} \lambda(y)dy} \end{aligned}

\text{ }

Theorem 2
As discussed above, let T be the length of the interval that is required to observe the first change. Then the following are the distribution function, survival function and pdf of T:

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &F_T(t)=\displaystyle 1-e^{-\int_0^t \lambda(y) dy} \\&\text{ } \\&S_T(t)=\displaystyle e^{-\int_0^t \lambda(y) dy} \\&\text{ } \\&f_T(t)=\displaystyle \lambda(t) e^{-\int_0^t \lambda(y) dy} \end{aligned}

Proof. In Theorem 1, we derive the probability P[N_y=0] for the discrete variable N_y derived from the non-homogeneous Poisson process. We now consider the continuous random variable T, the time until the first change, which is related to N_t by (1). Thus S_T(t)=P[T > t]=P[N_t=0]=e^{-\int_0^t \lambda(y) dy}. The distribution function and density function can be derived accordingly.

\text{ }

Theorem 3
The hazard rate function \lambda(t) is equivalent to each of the following:

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &\lambda(t)=\frac{f_T(t)}{1-F_T(t)} \\&\text{ } \\&\lambda(t)=\frac{-S_T^{'}(t)}{S_T(t)} \end{aligned}

\text{ }

Remark
Theorem 1 and Theorem 2 show that in a non-homogeneous Poisson process as described above, the hazard rate function \lambda(t) completely specifies the probability distribution of the survival model T (the time until the first change) . Once the rate of change function \lambda(t) is known in the non-homogeneous Poisson process, we can use it to generate the survival function S_T(t). All of the examples of survival models given below are derived by assuming the functional form of the hazard rate function. The result in Theorem 2 holds even outside the context of a non-homogeneous Poisson process, that is, given the hazard rate function \lambda(t), we can derive the three distributional items S_T(t), F_T(t), f_T(t).

The ratio in Theorem 3 indicates that the probability distribution determines the hazard rate function. In fact, the ratio in Theorem 3 is the usual definition of the hazard rate function. That is, the hazard rate function can be defined as the ratio of the density and the survival function (one minus the cdf). With this definition, we can also recover the survival function. Whenever \displaystyle \lambda(x)=\frac{f_X(x)}{1-F_X(x)}, we can derive:

\text{ }

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &S_X(x)=\displaystyle e^{-\int_0^t \lambda(y) dy} \end{aligned}

\text{ }

As indicated above, the hazard rate function can be interpreted as the failure rate at time t given that the life in question has survived to time t. It is the rate of failure at the next instant given that the life or system being studied has survived up to time t.

It is interesting to note that the function \Lambda(t)=\int_0^t \lambda(y) dy defined in Theorem 1 is called the cumulative hazard rate function. Thus the cumulative hazard rate function is an alternative way of representing the hazard rate function (see the discussion on Weibull distribution below).

——————————————————————————————————————
Examples of Survival Models

–Exponential Distribution–
In many applications, especially those for biological organisms and mechanical systems that wear out over time, the hazard rate \lambda(t) is an increasing function of t. In other words, the older the life in question (the larger the t), the higher chance of failure at the next instant. For humans, the probability of a 85 years old dying in the next year is clearly higher than for a 20 years old. In a Poisson process, the rate of change \lambda(t)=\lambda indicated in condition 3 is a constant. As a result, the time T until the first change derived in Theorem 2 has an exponential distribution with parameter \lambda. In terms of mortality study or reliability study of machines that wear out over time, this is not a realistic model. However, if the mortality or failure is caused by random external events, this could be an appropriate model.

–Weibull Distribution–
This distribution is an excellent model choice for describing the life of manufactured objects. It is defined by the following cumulative hazard rate function:

\text{ }

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &\Lambda(t)=\biggl(\frac{t}{\beta}\biggr)^{\alpha} \end{aligned} where \alpha > 0 and \beta>0

\text{ }

As a result, the hazard rate function, the density function and the survival function for the lifetime distribution are:

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &\lambda(t)=\frac{\alpha}{\beta} \biggl(\frac{t}{\beta}\biggr)^{\alpha-1} \\&\text{ } \\&f_T(t)=\frac{\alpha}{\beta} \biggl(\frac{t}{\beta}\biggr)^{\alpha-1} \displaystyle e^{\displaystyle -\biggl[\frac{t}{\beta}\biggr]^{\alpha}} \\&\text{ } \\&S_T(t)=\displaystyle e^{\displaystyle -\biggl[\frac{t}{\beta}\biggr]^{\alpha}} \end{aligned}

\text{ }

The parameter \alpha is the shape parameter and \beta is the scale parameter. When \alpha=1, the hazard rate becomes a constant and the Weibull distribution becomes an exponential distribution.

When the parameter \alpha<1, the failure rate decreases over time. One interpretation is that most of the defective items fail early on in the life cycle. Once they they are removed from the population, failure rate decreases over time.

When the parameter 1<\alpha, the failure rate increases with time. This is a good candidate for a model to describe the lifetime of machines or systems that wear out over time.

–The Gompertz Distribution–
The Gompertz law states that the force of mortality or failure rate increases exponentially over time. It describe human mortality quite accurately. The following is the hazard rate function:

\text{ }

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &\lambda(t)=\alpha e^{\beta t} \end{aligned} where \alpha>0 and \beta>0.

\text{ }

The following are the cumulative hazard rate function as well as the survival function, distribution function and the pdf of the lifetime distribution T.

\text{ }

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &\Lambda(t)=\int_0^t \alpha e^{\beta y} dy=\frac{\alpha}{\beta} e^{\beta t}-\frac{\alpha}{\beta} \\&\text{ } \\&S_T(t)=\displaystyle e^{\displaystyle -\biggl(\frac{\alpha}{\beta} e^{\beta t}-\frac{\alpha}{\beta}\biggr)} \\&\text{ } \\&F_T(t)=\displaystyle 1-e^{\displaystyle -\biggl(\frac{\alpha}{\beta} e^{\beta t}-\frac{\alpha}{\beta}\biggr)} \\&\text{ } \\&f_T(t)=\displaystyle \alpha \ e^{\beta t} \ e^{\displaystyle -\biggl(\frac{\alpha}{\beta} e^{\beta t}-\frac{\alpha}{\beta}\biggr)} \end{aligned}

\text{ }

–Makeham’s Law–
The Makeham’s Law states that the force of mortality is the Gompertz failure rate plus an age-indpendent component that accounts for external causes of mortality. The following is the hazard rate function:

\text{ }

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &\lambda(t)=\alpha e^{\beta t}+\mu \end{aligned} where \alpha>0, \beta>0 and \mu>0.

\text{ }

The following are the cumulative hazard rate function as well as the survival function, distribution function and the pdf of the lifetime distribution T.

\text{ }

\displaystyle \begin{aligned}. \ \ \ \ \ \ \ \ \ &\Lambda(t)=\int_0^t (\alpha e^{\beta y}+\mu) dy=\frac{\alpha}{\beta} e^{\beta t}-\frac{\alpha}{\beta}+\mu t \\&\text{ } \\&S_T(t)=\displaystyle e^{\displaystyle -\biggl(\frac{\alpha}{\beta} e^{\beta t}-\frac{\alpha}{\beta}+\mu t\biggr)} \\&\text{ } \\&F_T(t)=\displaystyle 1-e^{\displaystyle -\biggl(\frac{\alpha}{\beta} e^{\beta t}-\frac{\alpha}{\beta}+\mu t\biggr)} \\&\text{ } \\&f_T(t)=\biggl( \alpha e^{\beta t}+\mu t \biggr) \ e^{\displaystyle -\biggl(\frac{\alpha}{\beta} e^{\beta t}-\frac{\alpha}{\beta}+\mu t\biggr)} \end{aligned}

\text{ }

The Tail Weight of a Distribution
The hazard rate function can provide information about the tail of a distribution. If the hazard rate function is decreasing, it is an indication that the distribution has a heavy tail, i.e., the distribution significantly puts more probability on larger values. Conversely, if the hazard rate function is increasing, it is an indication of a lighter tail. In an insurance context, heavy tailed distributions (e.g. the Pareto distribution) are suitable candidates for modeling large insurance losses (see this previous post or [1]).

\text{ }

Reference

  1. Klugman S.A., Panjer H. H., Wilmot G. E. Loss Models, From Data to Decisions, Second Edition., Wiley-Interscience, a John Wiley & Sons, Inc., New York, 2004

The mean excess loss function

We take a closer look at the mean excess loss function. We start with the following example:

Example
Suppose that an entity is exposed to a random loss X. An insurance policy offers protection against this loss. Under this policy, payment is made to the insured entity subject to a deductible d>0, i.e. when a loss is less than d, no payment is made to the insured entity, and when the loss exceeds d, the insured entity is reimbursed for the amount of the loss in excess of the deductible d. Consider the following two questions:

\text{ }

  1. Of all the losses that are eligible to be reimbursed by the insurer, what is the average payment made by the insurer to the insured?
  2. What is the average payment made by the insurer to the insured entity?

\text{ }

The two questions look similar. The difference between the two questions is subtle but important. In the first question, the average is computed over all losses that are eligible for reimbursement (i.e., the loss exceeds the deductible). This is the average amount the insurer is expected to pay in the event that a payment in excess of the deductible is required to be made. So this average is a per payment average.

In the second question, the average is calculated over all losses (regardless of sizes). When the loss does not reach the deductible, the payment is considered zero and when the loss is in excess of the deductible, the payment is X-d. Thus the average is the average amount the insurer has to pay per loss. So the second question is about a per loss average.

—————————————————————————————————————-

The Mean Excess Loss Function
The average in the first question is called the mean excess loss function. Suppose X is the random loss and d>0. The mean excess loss variable is the conditional variable X-d \ \lvert X>d and the mean excess loss function e_X(d) is defined by:

\text{ }

\displaystyle (1) \ \ \ \ \ e_X(d)=E(X-d \lvert X>d)

\text{ }

In an insurance context, the mean excess loss function is the average payment in excess of a threshold given that the loss exceeds the threshold. In a mortality context, the mean excess loss function is called the mean residual life function and complete expectation of life and can be interpreted as the remaining time until death given that the life in question is alive at age d.

The mean excess loss function is computed by the following depending on whether the loss variable is continuous or discrete.

\text{ }

\displaystyle (2) \ \ \ \ \ e_X(d)=\frac{\int_d^\infty (x-d) \ f_X(x) \ dx}{S_X(d)}

\displaystyle (3) \ \ \ \ \ e_X(d)=\frac{\sum \limits_{x>d} (x-d) \ P(X=x)}{S_X(d)}

\text{ }

The mean excess loss function e_X(d) is defined only when the integral or the sum converges. The following is an equivalent calculation of e_X(d) that may be easier to use in some circumstances.

\text{ }

\displaystyle (4) \ \ \ \ \ e_X(d)=\frac{\int_d^\infty S_X(x) \ dx}{S_X(d)}

\displaystyle (5a) \ \ \ \ \ e_X(d)=\frac{\sum \limits_{x \ge d} S_X(x) }{S_X(d)}

\displaystyle (5b) \ \ \ \ \ e_X(d)=\frac{\biggl(\sum \limits_{x>d} S_X(x)\biggr)+(w+1-d) S_X(w) }{S_X(d)}

\text{ }

In both (5a) and (5b), we assume that the support of X is the set of nonnegative integers. In (5a), we assume that the deductible d is a positive integer. In (5b), the deductible d is free to be any positive number and w is the largest integer such that w \le d. The formulation (4) is obtained by using integration by parts (also see theorem 3.1 in [1]). The formulations of (5a) and (5b) are a result of applying theorem 3.2 in [1].

The mean excess loss function provides information about the tail weight of a distribution, see the previous post The Pareto distribution. Also see Example 3 below.
—————————————————————————————————————-
The Mean in Question 2
The average that we need to compute is the mean of the following random variable. Note that (a)_+ is the function that assigns the value of a whenever a>0 and otherwise assigns the value of zero.

\text{ }

\displaystyle (6) \ \ \ \ \ (X-d)_+=\left\{\begin{matrix}0&\ X<d\\{X-d}&\ X \ge d \end{matrix}\right.

\text{ }

The mean E((X-d)_+) is calculated over all losses. When the loss is less than the deductible d, the insurer has no obligation to make a payment to the insured and the payment is assumed to be zero in the calculation of E[(X-d)_+]. The following is how this expected value is calculated depending on whether the loss X is continuous or discrete.

\text{ }

\displaystyle (7) \ \ \ \ \ E((X-d)_+)=\int_d^\infty (x-d) \ f_X(x) \ dx

\displaystyle (8) \ \ \ \ \ E((X-d)_+)=\sum \limits_{x>d} (x-d) \ P(X=x)

\text{ }

Based on the definitions, the following is how the two averages are related.

\displaystyle E[(X-d)_+]=e_X(d) \ [1-F_X(d)] \ \ \ \text{or} \ \ \ E[(X-d)_+]=e_X(d) \ S_X(d)

—————————————————————————————————————-
The Limited Expected Value
For a given positive constant u, the limited loss variable is defined by

\text{ }

\displaystyle (9) \ \ \ \ \ X \wedge u=\left\{\begin{matrix}X&\ X<u\\{u}&\ X \ge u \end{matrix}\right.

\text{ }

The expected value E(X \wedge u) is called the limited expected value. In an insurance application, the u is a policy limit that sets a maximum on the benefit to be paid. The following is how the limited expected value is calculated depending on whether the loss X is continuous or discrete.

\text{ }

\displaystyle (9) \ \ \ \ \ E(X \wedge u)=\int_{-\infty}^u x \ f_X(x) \ dx+u \ S_X(u)

\displaystyle (10) \ \ \ \ \ E(X \wedge u)=\biggl(\sum \limits_{x < u} x \ P(X=x)\biggr)+u \ S_X(u)

\text{ }

Interestingly, we have the following relation.

\text{ }

\displaystyle (X-d)_+ + (X \wedge d)=X \ \ \ \ \ \ \text{and} \ \ \ \ \ \ E[(X-d)_+] + E(X \wedge d)=E(X)

\text{ }

The above statement indicates that purchasing a policy with a deductible d and another policy with a policy maximum d is equivalent to buying full coverage.

Another way to interpret X \wedge d is that it is the amount of loss that is eliminated by having a deductible in the insurance policy. If the insurance policy pays the loss in full, then the insurance payment is X and the expected amount the insurer is expected to pay is E(X). By having a deductible provision in the policy, the insurer is now only liable for the amount (X-d)_+ and the amount the insurer is expected to pay per loss is E[(X-d)_+]. Consequently E(X \wedge d) is the expected amount of the loss that is eliminated by the deductible provision in the policy. The following summarizes this observation.

\text{ }

\displaystyle (X \wedge d)=X-(X-d)_+ \ \ \ \ \ \ \text{and} \ \ \ \ \ \ E(X \wedge d)=E(X)-E[(X-d)_+]

—————————————————————————————————————-
Example 1
Let the loss random variable X be exponential with pdf f(x)=\alpha e^{-\alpha x}. We have E(X)=\frac{1}{\alpha}. Because of the no memory property of the exponential distribution, given that a loss exceeds the deductible, the mean payment is the same as the original mean. Thus e_X(d)=\frac{1}{\alpha}. Then the per loss average is:

\displaystyle E[(X-d)_+]=e_X(d) \ S(d) = \frac{e^{-\alpha d}}{\alpha}

\text{ }

Thus, with a deductible provision in the policy, the insurer is expected to pay \displaystyle \frac{e^{-\alpha d}}{\alpha} per loss instead of \displaystyle \frac{1}{\alpha}. Thus the expected amount of loss eliminated (from the insurer’s point of view) is \displaystyle E(X \wedge d)=\frac{1-e^{-\alpha d}}{\alpha}.

\text{ }

Example 2
Suppose that the loss variable has a Gamma distribution where the scale parameter is \alpha and the shape parameter is n=2. The pdf is \displaystyle g(x)=\alpha^2 \ x \ e^{-\alpha x}. The insurer’s expected payment without the deductible is E(X)=\frac{2}{\alpha}. The survival function S(x) is:

\displaystyle S(x)=e^{-\alpha x}(1+\alpha x)

\text{ }

For the losses that exceed the deductible, the insurer’s expected payment is:

\displaystyle \begin{aligned}e_X(d)&=\frac{\int_d^{\infty} S(x) \ dx}{S(d)}\\&=\frac{\int_d^{\infty} e^{-\alpha x}(1+\alpha x) \ dx}{e^{-\alpha d}(1+\alpha d)} \\&=\frac{\frac{e^{-\alpha d}}{\alpha}+d e^{-\alpha d}+\frac{e^{-\alpha d}}{\alpha}}{e^{-\alpha d}(1+\alpha d)} \\&=\frac{\frac{2}{\alpha}+d}{1+\alpha d} \end{aligned}

\text{ }

Then the insurer’s expected payment per loss is E[(X-d)_+]:

\displaystyle \begin{aligned}E[(X-d)_+]&=e_X(d) \ S(d) \\&=\frac{\frac{2}{\alpha}+d}{1+\alpha d} \ \ e^{-\alpha d}(1+\alpha d) \\&=e^{-\alpha d} \ \biggl(\frac{2}{\alpha}+d\biggr) \end{aligned}

\text{ }

With a deductible in the policy, the following is the expected amount of loss eliminated (from the insurer’s point of view).

\displaystyle \begin{aligned}E[X \wedge d]&=E(X)-E[(X-d)_+] \\&=\frac{2}{\alpha}-e^{-\alpha d} \ \biggl(\frac{2}{\alpha}+d\biggr) \\&=\frac{2}{\alpha}\biggl(1-e^{-\alpha d}\biggr)-d e^{-\alpha d} \end{aligned}

\text{ }

Example 3
Suppose the loss variable X has a Pareto distribution with the following pdf:

\text{ }

\displaystyle f_X(x)=\frac{\beta \ \alpha^\beta}{(x+\alpha)^{\beta+1}} \ \ \ \ \ x>0

\text{ }

If the insurance policy is to pay the full loss, then the insurer’s expected payment per loss is \displaystyle E(X)=\frac{\alpha}{\beta-1} provided that the shape parameter \beta is larger than one.

The mean excess loss function of the Pareto distribution has a linear form that is increasing (see the previous post The Pareto distribution). The following is the mean excess loss function:

\text{ }

\displaystyle e_X(d)=\frac{1}{\beta-1} \ d +\frac{\alpha}{\beta-1}=\frac{1}{\beta-1} \ d +E(X)

\text{ }

If the loss is modeled by such a distribution, this is an uninsurable risk! First of all, the higher the deductible, the larger the expected payment if such a large loss occurs. The expected payment for large losses is always the unmodified expected E(X) plus a component that is increasing in d.

The increasing mean excess loss function is an indication that the Pareto distribution is a heavy tailed distribution. In general, an increasing mean excess loss function is an indication of a heavy tailed distribution. On the other hand, a decreasing mean excess loss function indicates a light tailed distribution. The exponential distribution has a constant mean excess loss function and is considered a medium tailed distribution.

Reference

  1. Bowers N. L., Gerber H. U., Hickman J. C., Jones D. A., Nesbit C. J. Actuarial Mathematics, First Edition., The Society of Actuaries, Itasca, Illinois, 1986
  2. Klugman S.A., Panjer H. H., Wilmot G. E. Loss Models, From Data to Decisions, Second Edition., Wiley-Interscience, a John Wiley & Sons, Inc., New York, 2004