# A New Heavy Tailed Class of Distributions Which Includes the Pareto

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. The $\mathcal{MPLG}$ Distribution and Its Properties

**Remark**

**1.**

- Generate two random numbers ${u}_{1}$ and ${u}_{2}$ from the standard uniform distribution, $U(0,1)$.
- Generate random variates ${\tilde{a}}_{1}$ from the exponential distribution with mean $1/\theta $ and ${\tilde{a}}_{2}$ from the Erlang distribution with shape parameter 2 and rate parameter $\theta $ by using ${u}_{1}$.
- If ${u}_{2}\le p$, then set $y={\tilde{a}}_{1}$; otherwise, set $y={\tilde{a}}_{2}$.
- Generate $x={x}_{0}\phantom{\rule{0.166667em}{0ex}}exp\left\{y\right\}$.

#### 2.1. Moments and Log-Moments

#### 2.2. Unimodality

#### 2.3. Stochastic Ordering

**Definition**

**1.**

- (i)
- Stochastic order $\left({X}_{1}{\le}_{st}{X}_{2}\right)$ if ${F}_{{X}_{1}}\left(x\right)\ge {F}_{{X}_{2}}\left(x\right)$ for all x.
- (ii)
- Hazard rate order $\left({X}_{1}{\le}_{hr}{X}_{2}\right)$ if ${h}_{{X}_{1}}\left(x\right)\ge {h}_{{X}_{2}}\left(x\right)$ for all x.
- (iii)
- Mean excess order $\left({X}_{1}{\le}_{me}{X}_{2}\right)$ if ${e}_{{X}_{1}}\left(x\right)\ge {e}_{{X}_{2}}\left(x\right)$ for all x, where ${e}_{X}(\xb7)$ is mean excess function given in expression (17).

**Proposition**

**1.**

- i.
- If${x}_{01}={x}_{02}$,${\theta}_{1}\ge {\theta}_{2}$and${\lambda}_{1}\le {\lambda}_{2}$, then${X}_{1}{\le}_{lr}{X}_{2}$,${X}_{1}{\le}_{hr}{X}_{2}$and${X}_{1}{\le}_{st}{X}_{2}$.
- ii.
- If${x}_{01}\le {x}_{02}$,${\theta}_{1}={\theta}_{2}$and${\lambda}_{1}={\lambda}_{2}$, then${X}_{1}{\le}_{lr}{X}_{2}$,${X}_{1}{\le}_{hr}{X}_{2}$and${X}_{1}{\le}_{st}{X}_{2}$.

**Proof.**

#### 2.4. Integrated Tail Distribution and Equilibrium Hazard Rate

#### 2.5. Estimation

#### 2.5.1. Method of Log-Moments

#### 2.5.2. Maximum Likelihood Estimation

#### 2.6. Composite Models

#### 2.6.1. Composite Lognormal-$\mathcal{MPLG}$ Model

#### 2.6.2. Composite Weibull-$\mathcal{MPLG}$ Model

#### 2.6.3. Composite Paralogistic-$\mathcal{MPLG}$ Model

## 3. Insurance Results

#### 3.1. Mean Excess Function

#### 3.2. Excess of Loss Reinsurance

#### 3.3. Value-at-Risk and Tail Value-at-Risk

**Definition**

**2.**

_{δ}(X), is the δ-quantile of X. That is

**Proposition**

**2.**

_{δ}(X) of the$\mathcal{MPLG}(\theta ,\lambda ,{x}_{0})$distribution is

**Proof.**

_{δ}(X). As in general, the loss distributions are typically skewed, the VaR is a non-coherent risk measure due to the lack of subadditivity (see Klugman et al. 2012), for that reason, the Tail Value-at-Risk (TVaR) of X (see Acerbi and Tasche 2002) is usually considered as a more informative and more useful risk measure. The TVaR is given by

**Proposition**

**3.**

_{δ}(X) for$\mathcal{MPLG}(\theta ,\lambda ,{x}_{0})$distribution is given by

**Proof.**

#### 3.4. Limited Loss Variable for Composite Models

**Definition**

**3.**

**Definition**

**4.**

**Proposition**

**4.**

**Proof.**

## 4. Case Studies

#### 4.1. An Application to Automobile Claims Data

#### 4.2. An Application to Fire Insurance Claims

`SMPrcacticals`’ add-on package for

`R`, available from the CRAN website http://cran.r-project.org/. Parameter estimation for all the models considered has been completed by the method of maximum likelihood (which is implemented using the function ‘

`mle`’/‘

`mle2`’ in

`R`). In Table 2, parameter estimates and standard errors (S.E.) for all the composite distributions and the three measures of model validation considered in our first example are exhibited. Among the composite models, we can see that the Weibull-$\mathcal{MPLG}$ composite model gives the best fit overall for this dataset in terms of NLL, AIC and BIC values.

#### 4.3. Theil’s Income Indices

**Proposition**

**5.**

**Proof.**

## 5. Conclusions

## Author Contributions

## Funding

## Conflicts of Interest

## Appendix A

**Expression (1) is a genuine pdf:**

**Proof of Proposition**

**1.**

**Proof of Proposition**

**2.**

**Proof of Proposition**

**3.**

## References

- Acerbi, Carlo, and Dirk Tasche. 2002. On the coherence of expected shortfall. Journal of Banking & Finance 26: 1487–503. [Google Scholar]
- Bakar, Shaiful A., Nor A. Hamzah, Mastoureh Maghsoudi, and Saralees Nadarajah. 2015. Modeling loss data using composite models. Insurance: Mathematics and Economics 61: 146–54. [Google Scholar]
- Calderín-Ojeda, Enrique, and Chun F. Kwok. 2016. Modeling Claims Data with Composite Stoppa Models. Scandinavian Actuarial Journal 9: 817–36. [Google Scholar] [CrossRef]
- Cooray, Kahadawala, and Malwane M. A. Ananda. 2005. Modeling actuarial data with a composite lognormal–Pareto model. Scandinavian Actuarial Journal 5: 321–34. [Google Scholar] [CrossRef]
- De Jong, Piet, and Heller Gillian H. 2008. Generalized Linear Models for Insurance Data. International Series on Actuarial Science; Cambridge: Cambridge University Press. [Google Scholar]
- Ghitany, Mohamed E., Barbra Atieh, and Saralees Nadarajah. 2008. Lindley distribution and its application. Mathematics and Computers in Simulation 78: 493–506. [Google Scholar] [CrossRef]
- Ghitany, Mohamed E., Emilio Gómez-Déniz, and Saralees Nadarajah. 2018. A new generalization of the Pareto distribution and its application to insurance data. Journal of Risk and Financial Management 11: 10. [Google Scholar] [CrossRef]
- Gómez-Déniz, Emilio, and Enrique Calderín-Ojeda. 2014. A Suitable Alternative to the Pareto Distribution. Hacettepe Journal of Mathematics and Statistics 43: 843–60. [Google Scholar]
- Gómez-Déniz, Emilio, and Enrique Calderín-Ojeda. 2015. Modeling insurance data with the Pareto ArcTan distribution. Astin Bulletin 45: 639–60. [Google Scholar] [CrossRef]
- Hogg, Robert V., and Stuart A. Klugman. 2009. Loss Distributions. Vol. 249 of Wiley Series in Probability and Statistics; New York: Wiley. [Google Scholar]
- Klugman, Stuart A., Harry H. Panjer, and Gordon E. Willmot. 2012. Loss Models: From Data to Decisions, 4th ed. Wiley Series in Probability and Statistics; Hoboken: John Wiley and Sons, Inc. [Google Scholar]
- Sarabia, José M., and Enrique Castillo. 2005. About a class of max–stable families with applications to income distributions. Metron LXIII: 505–27. [Google Scholar]
- Sarabia, José M., and Faustino Prieto. 2009. The Pareto-positive stable distribution: A new descriptive model for city size data. Physica A 388: 4179–91. [Google Scholar] [CrossRef]
- Scollnik, David P. M. 2007. On Composite Lognormal-Pareto Models. Scandinavian Actuarial Journal 1: 20–33. [Google Scholar] [CrossRef]
- Shaked, Moshe, and J. George Shanthikumar. 2007. Stochastic Orders. New York: Springer Science and Business Media. [Google Scholar]
- Stoppa, Gabriele. 1990. Proprietà campionarie di un nuovo modello Pareto generalizzato. In Atti XXXV Riunione Scientifica della Società Italiana di Statistica. Padova: Cedam, pp. 137–44. [Google Scholar]
- Su, Chun, and Qihe H. Tang. 2003. Characterizations on heavy-tailed distributions by means of hazard rate. Acta Mathematicae Applicatae Sinica 19: 135–42. [Google Scholar] [CrossRef]
- Yang, Hailiang. 2004. Cramér-Lundberg asymptotics. In Encyclopedia of Actuarial Science. New York: Wiley. [Google Scholar]
- Yitzhaki, Shlomo. 1983. On an Extension of the Gini Inequality Index. International Economic Review 24: 617–28. [Google Scholar]

**Figure 1.**Graphs of the pdf (1) for different values of parameter $\theta $, $\lambda $ and fixed value of ${x}_{0}$.

**Figure 2.**Graphs of the hazard rate function (4) for different values of parameter $\theta $, $\lambda $ and for fixed value of ${x}_{0}$.

**Table 1.**Estimated values of different heavy-tailed models, corresponding standard errors (S.E.), negative of the log-likelihood function (NLL), Akaike’s information criterion (AIC) and Bayesian information criterion (BIC) for automobile claims dataset (De Jong and Heller 2008).

Distribution | Parameter Estimates (S.E.) | NLL | AIC | BIC |
---|---|---|---|---|

Pareto | $\widehat{\theta}=0.661\phantom{\rule{0.166667em}{0ex}}\left(0.010\right)$ | 38,024.80 | 76,051.61 | 76,058.05 |

lognormal | $\widehat{\mu}=6.810\phantom{\rule{0.166667em}{0ex}}\left(0.017\right),\phantom{\rule{0.277778em}{0ex}}\widehat{\sigma}=1.189\phantom{\rule{0.166667em}{0ex}}\left(0.012\right)$ | 38,852.15 | 77,708.31 | 77,721.19 |

loggamma | $\widehat{\alpha}=1.115\phantom{\rule{0.166667em}{0ex}}\left(0.021\right),\phantom{\rule{0.277778em}{0ex}}\widehat{\beta}=5.109\phantom{\rule{0.166667em}{0ex}}\left(0.118\right)$ | 38,998.18 | 78,000.36 | 78,013.23 |

Frećhet | $\widehat{\alpha}=0.659\phantom{\rule{0.166667em}{0ex}}\left(0.019\right),\phantom{\rule{0.277778em}{0ex}}\widehat{s}=254.0\phantom{\rule{0.166667em}{0ex}}\left(11.87\right)$ | 38,408.30 | 76,822.60 | 76,841.92 |

$\widehat{b}=157.1\phantom{\rule{0.166667em}{0ex}}\left(4.279\right)$ | ||||

Weibull | $\widehat{\mu}=0.786\phantom{\rule{0.166667em}{0ex}}\left(0.008\right),\phantom{\rule{0.277778em}{0ex}}\widehat{\sigma}=1690.8\phantom{\rule{0.166667em}{0ex}}\left(33.64\right)$ | 39,491.60 | 78,987.19 | 79,000.07 |

Lomax | $\widehat{\alpha}=2.047\phantom{\rule{0.166667em}{0ex}}\left(0.088\right),\phantom{\rule{0.277778em}{0ex}}\widehat{\lambda}=2205.1\phantom{\rule{0.166667em}{0ex}}\left(133.1\right)$ | 39,169.85 | 78,343.70 | 78,356.58 |

PAT | $\widehat{\alpha}=0.895\phantom{\rule{0.166667em}{0ex}}\left(0.095\right),\phantom{\rule{0.277778em}{0ex}}\widehat{\theta}=0.740\phantom{\rule{0.166667em}{0ex}}\left(0.017\right)$ | 38,006.30 | 76,016.61 | 76,029.49 |

Inverse Weibull | $\widehat{\alpha}=1.053\phantom{\rule{0.166667em}{0ex}}\left(0.012\right),\phantom{\rule{0.277778em}{0ex}}\widehat{\sigma}=518.8\phantom{\rule{0.166667em}{0ex}}\left(7.636\right)$ | 38,595.61 | 77,195.22 | 77,208.09 |

$\mathcal{MPLG}$ | $\widehat{\theta}=0.943\phantom{\rule{0.166667em}{0ex}}\left(0.018\right),\phantom{\rule{0.277778em}{0ex}}\widehat{\lambda}=0.698\phantom{\rule{0.166667em}{0ex}}\left(0.073\right)$ | 37,965.99 | 75,935.98 | 75,948.86 |

**Table 2.**Estimated values of different heavy-tailed models, corresponding standard errors (S.E.), NLL, AIC, and BIC of different composite models for the Danish fire losses dataset.

Distribution | Parameter Estimates (S.E.) | NLL | AIC | BIC | |
---|---|---|---|---|---|

lognormal-Lomax | $\widehat{\mu}=0.104\phantom{\rule{0.166667em}{0ex}}\left(0.020\right)$ | $\widehat{\sigma}=0.182\phantom{\rule{0.166667em}{0ex}}\left(0.011\right)$ | 3860.47 | 7728.94 | 7752.223 |

$\widehat{\lambda}=0.365\phantom{\rule{0.166667em}{0ex}}\left(0.123\right)$ | $\widehat{\theta}=1.144\phantom{\rule{0.166667em}{0ex}}\left(0.029\right)$ | ||||

lognormal-Pareto | $\widehat{\mu}=0.137\phantom{\rule{0.166667em}{0ex}}\left(0.019\right)$ | $\widehat{\sigma}=0.197\phantom{\rule{0.166667em}{0ex}}\left(0.012\right)$ | 3865.86 | 7737.72 | 7755.182 |

$\widehat{\theta}=1.208\phantom{\rule{0.166667em}{0ex}}\left(0.030\right)$ | |||||

lognormal-$\mathcal{MPLG}$ | $\widehat{\mu}=0.045\phantom{\rule{0.166667em}{0ex}}\left(0.017\right)$ | $\widehat{\theta}=2.060\phantom{\rule{0.166667em}{0ex}}\left(0.039\right)$ | 3872.40 | 7752.81 | 7776.09 |

${\widehat{x}}_{0}=0.745\phantom{\rule{0.166667em}{0ex}}\left(0.070\right)$ | $\widehat{\lambda}=65.804\phantom{\rule{0.166667em}{0ex}}\left(377.425\right)$ | ||||

Weibull-Lomax | $\widehat{\tau}=15.345\phantom{\rule{0.166667em}{0ex}}\left(0.671\right)$ | $\widehat{\varphi}=0.969\phantom{\rule{0.166667em}{0ex}}\left(0.007\right)$ | 3823.70 | 7655.40 | 7678.68 |

$\widehat{\lambda}=0.561\phantom{\rule{0.166667em}{0ex}}\left(0.127\right)$ | $\widehat{\theta}=0.971\phantom{\rule{0.166667em}{0ex}}\left(0.007\right)$ | ||||

Weibull-Pareto | $\widehat{\tau}=14.048\phantom{\rule{0.166667em}{0ex}}\left(0.502\right)$ | $\widehat{\psi}=0.997\phantom{\rule{0.166667em}{0ex}}\left(0.008\right)$ | 3840.38 | 7686.76 | 7704.22 |

$\widehat{\theta}=1.003\phantom{\rule{0.166667em}{0ex}}\left(0.008\right)$ | |||||

Weibull-$\mathcal{MPLG}$ | $\widehat{\tau}=18.763\phantom{\rule{0.166667em}{0ex}}\left(0.986\right)$ | ${\widehat{x}}_{0}=0.787\phantom{\rule{0.166667em}{0ex}}\left(0.006\right)$ | 3823.30 | 7654.60 | 7677.88 |

$\widehat{\theta}=1.938\phantom{\rule{0.166667em}{0ex}}\left(0.034\right)$ | $\widehat{\lambda}=4.614\phantom{\rule{0.166667em}{0ex}}\left(0.159\right)$ | ||||

paralogistic-$\mathcal{MPLG}$ | $\widehat{\alpha}=16.719\phantom{\rule{0.166667em}{0ex}}\left(1.311\right)$ | ${\widehat{x}}_{0}=0.901\phantom{\rule{0.166667em}{0ex}}\left(0.020\right)$ | 3824.482 | 7656.96 | 7680.25 |

$\widehat{\theta}=1.989\phantom{\rule{0.166667em}{0ex}}\left(0.035\right)$ | $\widehat{\lambda}=3.379\phantom{\rule{0.166667em}{0ex}}\left(0.266\right)$ |

**Table 3.**Limited expected value for the distributions considered and different values of the policy limit u.

Composite Models | ||||||||
---|---|---|---|---|---|---|---|---|

Policy | Empirical | lognormal | lognormal | Weibull | Weibull | lognormal | Weibull | Paralogistic |

Limit u | Pareto | Lomax | Pareto | Lomax | $\mathcal{MPLG}$ | $\mathcal{MPLG}$ | $\mathcal{MPLG}$ | |

1 | 0.989 | 0.998 | 0.985 | 0.989 | 0.989 | 0.987 | 0.986 | 0.989 |

2 | 1.565 | 1.726 | 1.525 | 1.560 | 1.548 | 1.333 | 1.266 | 1.467 |

5 | 2.138 | 2.489 | 2.122 | 2.155 | 2.143 | 2.015 | 1.943 | 2.093 |

10 | 2.447 | 2.932 | 2.362 | 2.482 | 2.464 | 2.341 | 2.292 | 2.398 |

20 | 2.707 | 3.285 | 2.566 | 2.781 | 2.674 | 2.542 | 2.522 | 2.591 |

100 | 2.958 | 3.851 | 2.826 | 3.299 | 2.923 | 2.740 | 2.778 | 2.793 |

200 | 3.037 | 4.017 | 2.884 | 3.463 | 2.972 | 2.770 | 2.823 | 2.826 |

∞ | 3.063 | 4.665 | 3.005 | 4.289 | 3.060 | 2.804 | 2.885 | 2.868 |

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Bhati, D.; Calderín-Ojeda, E.; Meenakshi, M.
A New Heavy Tailed Class of Distributions Which Includes the Pareto. *Risks* **2019**, *7*, 99.
https://doi.org/10.3390/risks7040099

**AMA Style**

Bhati D, Calderín-Ojeda E, Meenakshi M.
A New Heavy Tailed Class of Distributions Which Includes the Pareto. *Risks*. 2019; 7(4):99.
https://doi.org/10.3390/risks7040099

**Chicago/Turabian Style**

Bhati, Deepesh, Enrique Calderín-Ojeda, and Mareeswaran Meenakshi.
2019. "A New Heavy Tailed Class of Distributions Which Includes the Pareto" *Risks* 7, no. 4: 99.
https://doi.org/10.3390/risks7040099