evy processes with Parisian reflection below and classical reflection above

Florin Avram, Kazutoshi Yamazaki2016

TL;DR

This paper introduces a new mathematical model for financial risk management where capital injections (to avoid ruin) and dividend payouts are not continuous but occur at discrete, random Poisson observation times. The authors develop a rigorous framework using spectrally negative Lévy processes and scale functions to derive key financial identities, offering a more realistic depiction of corporate finance under uncertainty.

What Problem Does This Paper Address?

Traditional models often assume that interventions like capital injections or dividend payments happen instantaneously and continuously. However, in reality, these actions are often triggered by observations and executed at specific points in time, which are frequently governed by random processes. This paper addresses this gap by modeling a company's financial health as a spectrally negative Lévy process, where "bailouts" (capital injections) occur only at independent Poisson observation times if the company's capital falls below zero (Parisian reflection). Additionally, it considers a scenario with a dividend-paying strategy, modeled by classical reflection above a certain barrier. The core challenge is to derive precise mathematical formulas for crucial financial quantities like ruin probabilities, capital injections, and dividend payouts under these more realistic, discrete observation conditions.

Methods

The paper leverages the powerful framework of spectrally negative Lévy processes. The key mathematical tools employed are:

  • Scale Functions: These are fundamental to the analysis of Lévy processes and allow for the computation of various fluctuation identities. The paper extensively uses and extends existing results on scale functions.
  • Parisian Reflection: This is a novel aspect, where the process is only "reflected" (i.e., pushed up to zero) at specific Poisson observation times when it's below zero, rather than continuously.
  • Classical Reflection: This is used to model dividend payouts, where the process is reflected above a certain barrier.
  • Excursion Theory: For processes with unbounded variation, excursion theory is used to analyze the behavior of the process between jumps and to derive identities without relying on approximation schemes.
  • Strong Markov Property: This property is crucial for linking different stages of the process and for constructing proofs by conditioning on the first passage times.

The paper systematically builds upon existing literature, adapting and extending techniques to handle the discrete observation times and the specific types of reflections.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Realism in Financial Modeling: The paper highlights that discrete, Poisson-driven observations for interventions (bailouts, dividends) are more realistic than continuous models and can lead to significantly different outcomes.
  • Parisian Reflection as a Key Concept: Introducing "Parisian reflection" at Poisson times provides a more nuanced way to model capital injections, acknowledging the time and observation delays inherent in financial operations.
  • Unified Framework: The paper offers a unified mathematical framework to analyze both capital injections (Parisian reflection below) and dividend payouts (classical reflection above) within the context of spectrally negative Lévy processes.
  • Power of Scale Functions: The research demonstrates the continued power and versatility of scale functions in deriving complex financial identities for reflected and refracted Lévy processes.

Key Results

The paper derives a comprehensive set of "fluctuation identities" for two main processes:

  1. Lévy process with Parisian reflection below: This models capital injections at Poisson observation times. The authors compute total discounted values for capital injections under various killing conditions (exiting specific intervals).
  2. Lévy process with Parisian reflection below and classical reflection above: This models both capital injections and dividend payouts. The paper computes total expected discounted values for dividends and capital injections.

Specific theorems (e.g., Theorem 3.1, 3.2, 3.3, 3.4, 3.5) provide precise formulas for these quantities, often involving intricate combinations of scale functions, their derivatives, and related quantities like Z-functions. The results are presented for both bounded and unbounded variation cases of the underlying Lévy process.

Limitations and Future Work

The authors acknowledge that their work builds upon existing theories and extends them to a more complex scenario. While the paper provides a rigorous mathematical framework, its direct applicability might depend on the specific parameters of the Lévy process and the Poisson observation process. Future work could involve:

  • Exploring other types of observation processes (e.g., Erlang-distributed intervals, as mentioned in the introduction).
  • Extending the model to more general Lévy processes or incorporating other financial features like transaction costs or different dividend strategies.
  • Investigating the numerical implications of these formulas for practical risk management and optimal dividend strategies.

Summary updated

Abstract

We consider a company that receives capital injections so as to avoid ruin. Differently from the classical bail-out settings where the underlying process is restrict ed to stay at or above zero, we study the case bail-out can only be made at independent Poisson observation times. Namely, we study a version of the reflected process that is pushed up to zero only on Poisson arrival times at which the process is below zero. We also study the case with additional classical reflection above so as to mode l a company that pays dividends according to a −
BibTeX
@misc{avram2016evy,
  author = {Florin Avram and Kazutoshi Yamazaki},
  title = {{evy processes with Parisian reflection below and classical reflection above}},
  year = {2016},
}
Figure