×

An asymptotic analysis of the bootstrap bias correction for the empirical CTE. (English) Zbl 1219.62071

Summary: The \(\alpha \)-level Conditional Tail Expectation (CTE) of a continuous random variable \(X\) is defined as its conditional expectation given the event \(\{X > q_\alpha \}\), where \(q_\alpha\) represents its \(\alpha \)-level quantile. It is well known that the empirical CTE (the average of the \(n(1-\alpha)\) largest order statistics in a sample of size \(n\)) is a negatively biased estimator of the CTE. This bias vanishes as the sample size increases but in small samples can be significant. Hence the need for bias correction. Although the bootstrap method has been suggested for correcting the bias of the empirical CTE, recent research shows that alternate kernel-based methods of bias correction perform better in some practical examples. To further understand this phenomenon, we conduct an asymptotic analysis of the exact bootstrap bias correction for the empirical CTE, focusing on its performance as a point estimator of the bias of the empirical CTE.
We provide heuristics suggesting that the exact bootstrap bias correction is approximately a kernel-based estimator, albeit using a bandwidth that converges to zero faster than mean square optimal bandwidths. This approximation provides some insight into why the bootstrap method has markedly less residual bias, but at the cost of having higher variance. We prove a central limit theorem (CLT) for the exact bootstrap bias correction using an alternate representation as an \(l^1\) distance of the sample observations from the \(\alpha \)-level empirical quantile. The CLT, in particular, shows that the bootstrap bias correction has a relative error of \(n^{-1/4}\). In contrast, for any given \(\epsilon > 0\), and under the assumption that the sampling density is sufficiently smooth, the relative error of order \(O(n^{1/2+\epsilon})\) is attainable using kernel-based estimators. Thus, in an asymptotic sense, the bootstrap bias correction as a point estimator of the bias is not optimal in the case of smooth sampling densities. Bootstrapped risk measures have recently found interest as estimators in their own right; as an application we derive the CLT for the bootstrap expectation of the empirical CTE. We also report on a simulation study of the effect of small sample sizes on the quality of the approximation provided by the CLT.
In support of the bootstrap method we show that the bootstrap bias correction is optimal if the sampling density is constrained only to be Lipschitz of order \(1/2\) (or, loosely speaking, to have only half a derivative). Because in practice densities are at least twice differentiable, this optimality result largely fails to make the bootstrap method attractive to practitioners.

MSC:

62G09 Nonparametric statistical resampling methods
62G07 Density estimation
60F05 Central limit and other weak theorems
62G20 Asymptotic properties of nonparametric inference
65C60 Computational problems in statistics (MSC2010)

Software:

R; Snow; snow
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] Artzner P., Mathematics and Finance 9 (3) pp 203– (1999) · Zbl 0980.91042 · doi:10.1111/1467-9965.00068
[2] BloCh D. A., Annals of Mathematics and Statistics 39 pp 1083– (1968) · Zbl 0245.62043 · doi:10.1214/aoms/1177698342
[3] Bofinger E., Australian Journal of Statistics 17 (1) pp 1– (1975) · Zbl 0346.62038 · doi:10.1111/j.1467-842X.1975.tb01366.x
[4] Brazauskas V., Journal of Statistical Planning and Inference 138 (11) pp 3590– (2008) · Zbl 1152.62027 · doi:10.1016/j.jspi.2005.11.011
[5] Cheng C., Journalof Statistical Planning and Inference 59 (2) pp 291– (1997) · Zbl 0900.62209 · doi:10.1016/S0378-3758(96)00110-3
[6] Chung K. L., A Course in Probability Theory., 3. ed. (2001)
[7] CsÖrgą M., Volume 42 of CBMS-NSF Regional Conference Series in Applied Mathematics (1983)
[8] Durrett R., Probability: Theory and Examples., 3. ed. (2005) · Zbl 1202.60002
[9] Falk M., Statistics and Probability Letters 4 (4) pp 217– (1986) · doi:10.1016/0167-7152(86)90070-2
[10] Falk M., Statistics and Probability Letters 4 (2) pp 69– (1986) · Zbl 0585.62076 · doi:10.1016/0167-7152(86)90020-9
[11] Hall P., Annals of Statistics 17 (2) pp 692– (1989) · Zbl 0672.62051 · doi:10.1214/aos/1176347135
[12] Hall P., Probability Theory Related Fields 80 (2) pp 261– (1988) · Zbl 0637.62038 · doi:10.1007/BF00356105
[13] Hardy M. R., Investment Guarantees: Modeling and Risk Management for Equity-Linked Life Insurance (2003) · Zbl 1092.91042
[14] Hoeffding W., Journal of the American Statistical Association 58 pp 13– (1963) · doi:10.1080/01621459.1963.10500830
[15] Hutson A. D., Journal of the Royal Statistical Society Series B Statistical Methodology 62 (1) pp 89– (2000) · Zbl 0941.62048 · doi:10.1111/1467-9868.00221
[16] Jones B. L., North American Actuarial Journal 7 (4) pp 44– (2003) · Zbl 1084.62537 · doi:10.1080/10920277.2003.10596117
[17] Jones M., Annals of the Institute of Statistical Mathematics 44 (4) pp 721– (1992) · Zbl 0772.62022 · doi:10.1007/BF00053400
[18] Kaiser T., North American Actuarial Journal 10 (4) pp 249– (2006) · doi:10.1080/10920277.2006.10597425
[19] Kim J. H. T., ASTIN Bulletin 39 (1) pp 199– (2009) · Zbl 1205.91085 · doi:10.2143/AST.39.1.2038062
[20] Kim J. H. T., ASTIN Bulletin 37 (2) pp 365– (2007) · Zbl 1154.62378 · doi:10.2143/AST.37.2.2024072
[21] Ko B., ASTIN Bulletin 39 (2) pp 717– (2009) · Zbl 1178.62025 · doi:10.2143/AST.39.2.2044655
[22] Manistre B. J., North American Actuarial Journal 9 (2) pp 129– (2005) · Zbl 1085.62511 · doi:10.1080/10920277.2005.10596207
[23] Necir A., The Actuarial CTE Risk Measure for Heavy-Tailed Losses: A New Estimator and Confodence Intervals (2009)
[24] Parzen E., Journal of the American Statistical Association 74 (365) pp 105– (1979) · doi:10.1080/01621459.1979.10481621
[25] R Development Core Team, R: A Language and Environment for Statistical Computing (2008)
[26] Reiss R.-D., Metrika 25 (1) pp 9– (1978) · Zbl 0375.62018 · doi:10.1007/BF02204347
[27] Reiss R.-D., Approximate Distributions of Order Statistics (1989) · Zbl 0682.62009 · doi:10.1007/978-1-4613-9620-8
[28] Serfling R. J., Approximation Theorems of Mathematical Statistics (1980) · Zbl 0538.62002 · doi:10.1002/9780470316481
[29] Sheather S., Statistical Data Analysis Based on the L1 Norm and Related Methods pp 203– (1987)
[30] Sheather S. J., Australian Journal of Statistics 25 (1) pp 109– (1983) · Zbl 0516.62042 · doi:10.1111/j.1467-842X.1983.tb01203.x
[31] Siddiqui M. M., Journal of Research of the National Bureau of Standards Section B 64 pp 145– (1960) · Zbl 0096.13402 · doi:10.6028/jres.064B.017
[32] Silverman B. W., Monographs on Statistics and Applied Probability (1986)
[33] Tierney L., Snow: Simple Network of Workstations (2008)
[34] Tierney L., International Journal of Parallel Progress 37 pp 78– (2009) · Zbl 1191.68146 · doi:10.1007/s10766-008-0077-2
[35] Tukey J. W., Proceedings of the National Academy of Sciences U.S.A. 53 pp 127– (1965) · Zbl 0168.40205 · doi:10.1073/pnas.53.1.127
[36] Welsh A. H., Statistics and Probability Letters 6 (6) pp 427– (1988) · Zbl 0637.62043 · doi:10.1016/0167-7152(88)90103-4
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. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.