Functional limit theorems for discounted exponential functional of random walk and discounted convergent perpetuity
Introduction
Let , be independent copies of a real-valued random variable . Denote by , where , the zero-delayed standard random walk with jumps , that is, and for . For each , put . Plainly, a.s. for all/some if, and only if, a.s. Under this condition we call discounted exponential functional of random walk. Our terminology is inherited from a continuous-time counterpart , where is a Lévy process diverging to , known in the literature as ‘exponential functional of Lévy process’, see, for instance, (Bertoin and Yor, 2005, Carmona et al., 2001). Distributional properties of , with fixed, were investigated in Szabados and Székely (2003). The latter authors call exponential functional of random walk. There are many works investigating the asymptotics of as for various functions (for instance, exponential) in the situation that the corresponding series is divergent. This circle of problems which is very different from that related to discounted exponential functional is discussed in the books Borodin and Ibragimov (1995) and Skorokhod and Slobodenjuk (1970).
As usual, we write to denote convergence in probability, and and to denote weak convergence in a function space and weak convergence of one-dimensional distributions, respectively. We prefer to use the notation as in place of . Also, we denote by () the Skorokhod space of right-continuous functions defined on (on ) with finite limits from the left at positive points.
Here is our first result.
Theorem 1.1 Assume that and . Then, as , in the -topology on , where is a standard Brownian motion.
Remark 1.2 Under the assumptions a.s. and for all a one-dimensional central limit theorem for as was proved in Theorem 2 of Dall’ Aglio (1964) with the help of the method of moments.
Remark 1.3 The limit process in Theorem 1.1 is an a.s. continuous centered Gaussian process on with Integration by parts yields an alternative representation
Let , be independent copies of an -valued random vector with arbitrarily dependent components. Whenever the random series is a.s. convergent for each , following Iksanov et al. (2021), we call its sum discounted convergent perpetuity. We refer to the cited article for the justification of the term and standard limit theorems: a strong law of large numbers, a functional central limit theorem and a law of the iterated logarithm. In particular, Proposition 1.4 given next is Theorem 1.2 in Iksanov et al. (2021), up to the change of variable.
Proposition 1.4 Assume that , and . Then, as , in the -topology on , where is a standard Brownian motion.
A combination of Theorem 1.1 and Proposition 1.4 plus some additional work enables us to treat the case , thereby leading to an ultimate version of the functional central limit theorem for discounted convergent perpetuities.
Corollary 1.5 Assume that , , , , . Then in the -topology on . Here, is a standard Brownian motion, The constant is positive unless for some .
Remark 1.6 Putting in Corollary 1.5 we obtain the one-dimensional central limit theorem where is a random variable with the standard normal distribution. This result was proved in Theorem 6.1 of Vervaat (1979). Our argument is different from that exploited in Vervaat (1979).
A perusal of the proof of Corollary 1.5 reveals that the distributional asymptotic behavior of the process on the left-hand side of (3) is driven by the competitive distributional fluctuations of the functional versions of the random walks and . The latter are represented in the limit by the same integral functional of (generally) dependent Brownian motions and . Because of the minus sign and the dependence the contributions of the two random walks compensate each other. When and are linearly dependent, the Brownian motion is a multiple of . As a result, the contributions of the random walks get mutually neutralized, so that the limit variance is equal to . This means that is not a proper normalization in this case. According to Theorem 1.7, when for nonzero real , no normalization is needed at all, and concentrates tightly around the point in a deterministic manner for large enough.
Theorem 1.7 Assume that and . Then
The rest of the paper is structured as follows. We prove Theorem 1.1, Corollary 1.5 and Theorem 1.7 in Sections 2, 3 and 4 respectively. The appendix collects some auxiliary facts.
Section snippets
Proof of Theorem 1.1
For , put . Since a.s., we have a.s. Write, for , The first summand does not contribute to the limit. Indeed, for any , Further, integration by parts followed by change of variable yields, for any , By Proposition A.1, as ,
Proof of Corollary 1.5
In view of Theorem 1.1 and Proposition 1.4 the most important thing that remains to be done is determining the dependence between the Brownian motions appearing in (1), (2).
Let denote a two-dimensional Wiener process such that has the covariance matrix ( is a two-dimensional standard Brownian motion whenever and are uncorrelated). Assume that we can prove that
Proof of Theorem 1.7
Using , we write With the help of we infer and respectively. We intend to use Lemma A.3(a) with particular and . By the strong law of large numbers for random walks, for each from some set of probability measure
Acknowledgments
We thank an anonymous referee for very detailed and useful comments concerning both mathematics and presentation. The present proof of Lemma A.3 was suggested by the referee. This work was supported by the National Research Foundation of Ukraine (project 2020.02/0014 “Asymptotic regimes of perturbed random walks: on the edge of modern and classical probability”).
References (12)
- et al.
Exponential functionals of Lévy processes
Probab. Surv.
(2005) Convergence of Probability Measures
(1999)- et al.
Limit theorems for functionals of random walks
Proc. Steklov Inst. Math.
(1995) - et al.
Exponential functionals of Lévy processes
Present value of a renewal process
Ann. Math. Stat.
(1964)Stopped Random Walks. Limit Theorems and Applications
(2009)
Cited by (2)
Limit theorems for discounted convergent perpetuities II
2023, Electronic Journal of Probability