## Sums of pairwise quasi-asymptotically independent random variables with consistent variation.(English)Zbl 1181.62011

Let $$X_1\dots X_n$$ be dependent real-valued random variables with distribution functions $$F_1, \dots F_n$$, respectively. They are pairwise quasi-asymptotically dependent and heavy-tailed, i.e., $Ee^{rX_i}=\int^{\infty}_{-\infty}e^{rX}dF_i(x)=\infty\quad \text{for\;any}\;r>0.$ The heavy-tailed distributions are subexponential distributions. Let us denote this class by $$\varphi$$. In the paper, the tail asymptotic behavior of the sum $$S_n=X_1+\dots +X_n$$ is considered. The main statements are as follows. Theorem 3.1 states that for $$F_i\in\varphi, 1\leq i\leq n$$, $P\{S_n>x\}\sim \sum_{n=1}^n \big(1-Fi(x)\big) \quad \text{as}\;x\rightarrow\infty.$ A similar statement holds for $P\bigg(\sum_{k=1}^nW_kX_k>x\bigg) \quad \text{if}\quad W_k,\;1\leq k\leq n,$ are independent of $$\{X_i,\;1\leq i\leq n\}$$ and integer-valued, with $$EW^p_k$$ for some $$p$$ and $$F_i\in\varphi$$.
If the random variables $$\{X_i, 1\leq i\leq n\}$$ are identically distributed with distribution $$F$$ with extended regular variation, conditions for the relation $P\bigg\{\sum^\infty_{k=1}W_kX_k>x\bigg\}\sim \sum^\infty_{n=1}P\{W_kX_k>x\}\quad \text{as}\;\;x\rightarrow\infty$ are obtained as well. Finally, if $$F\in\varphi$$ and $$\tau$$ is independent of the sequence $$\{X_1, X_2,\dots\}$$, integer-valued and such that $$0<E\tau^{p+1}<\infty$$ for some $$p$$, then $P\{S_\tau >x\}\sim E\tau\big( 1-F(x)\big), \quad \text{as}\quad x\rightarrow\infty.$ The Introduction and Chapter 2 contain references on the tail behavior of sums of dependent random variables as well as a discussion on heavy-tailed distributions. The list of references contains 33 positions.

### MSC:

 62E20 Asymptotic distribution theory in statistics 60G50 Sums of independent random variables; random walks 62G32 Statistics of extreme values; tail inference 62H20 Measures of association (correlation, canonical correlation, etc.)

QRM
Full Text:

### References:

  DOI: 10.1007/s10687-006-0011-1 · Zbl 1142.60009  Bingham N.H., Regular Variation (1987)  DOI: 10.1239/jap/1077134672 · Zbl 1054.60012  DOI: 10.1016/j.insmatheco.2006.06.004 · Zbl 1183.60033  DOI: 10.1016/0304-4149(94)90113-9 · Zbl 0799.60015  DOI: 10.1007/s10687-007-0033-3 · Zbl 1150.60029  DOI: 10.1214/aoap/1035463328 · Zbl 0879.60053  de Haan L., Extreme Value Theory: An Introduction (2006) · Zbl 1101.62002  DOI: 10.1080/15326340600878016 · Zbl 1159.91403  Embrechts P., Modelling Extremal Events for Insurance and Finance (1997) · Zbl 0873.62116  DOI: 10.1239/jap/1152413743 · Zbl 1104.60313  DOI: 10.1080/03461230500361943 · Zbl 1144.91026  DOI: 10.2298/PIM0694171J · Zbl 1164.60301  DOI: 10.1006/jmva.1993.1061 · Zbl 0778.62045  DOI: 10.1239/jap/1208358953 · Zbl 1137.62310  Kortschak D., Methodol. Comput. Appl. Probab.  DOI: 10.1002/0471722065  DOI: 10.1016/j.insmatheco.2005.05.009 · Zbl 1099.91068  McNeil A.J., Quantitative Risk Management: Concepts, Techniques and Tools (2005) · Zbl 1089.91037  DOI: 10.1239/jap/1077134670 · Zbl 1051.60032  Resnick S.I., Extreme Values, Regular Variation, and Point Processes (1987) · Zbl 0633.60001  DOI: 10.1023/A:1025148622954 · Zbl 1035.60053  Resnick S.I., Heavy-Tail Phenomena: Probabilistic and Statistical Modeling (2007) · Zbl 1152.62029  DOI: 10.1080/15326349108807204 · Zbl 0747.60062  DOI: 10.1016/j.insmatheco.2007.10.001 · Zbl 1154.60032  DOI: 10.1023/B:SIMJ.0000048931.70386.68  Tang Q., Electron. J. Probab. 11 pp 107– (2006)  DOI: 10.1080/07362990802006964 · Zbl 1141.62036  Tang Q., Stochastic Process. Appl. 108 pp 299– (2003) · Zbl 1075.91563  DOI: 10.1023/B:EXTR.0000031178.19509.57 · Zbl 1049.62017  DOI: 10.1016/j.spl.2007.03.042 · Zbl 1210.91061  DOI: 10.1080/15326340600649029 · Zbl 1095.60008  DOI: 10.1239/jap/1197908812 · Zbl 1134.60322
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.