arXiv subjects

Aleksandar Arandjelović

Publications and source records attributed to Aleksandar Arandjelović.

6 recordsLinked to original sources

arXiv2025

Solving stochastic climate-economy models: A deep least-squares Monte Carlo approach

Stochastic versions of recursive integrated climate-economy assessment models are essential for studying and quantifying policy decisions under uncertainty. However, as the number of state variables and stochastic shocks increases, solving these models via deterministic grid-based dynamic programming (e.g., value-function iteration / projection on a discretized grid over continuous state variables, typically coupled with discretized shocks) becomes computationally infeasible, and simulation-based methods are needed. The least-squares Monte Carlo (LSMC) method has become popular for solving optimal stochastic control problems in quantitative finance. In this paper, we extend the application of the LSMC method to stochastic climate-economy models. We exemplify this approach using a stochastic version of the DICE model with five key uncertainty sources highlighted in the literature. To address the complexity and high dimensionality of these models, we incorporate deep neural network approximations in place of standard regression techniques within the LSMC framework. Our results demonstrate that the deep LSMC method can be used to efficiently derive optimal policies for climate-economy models in the presence of uncertainty.

econ.GN↗

arXiv2025

Optimal Investment, Consumption, and Insurance with Durable Goods under Stochastic Depreciation Risk

We study an infinite-horizon optimal investment, consumption and insurance problem for an economic agent who consumes a perishable and a durable good. The agent trades in a risk-free asset, a risky asset, and a durable good whose price follows a correlated diffusion, while the stock of the durable good depreciates deterministically and is subject to insurable Poisson loss shocks. The agent can partially hedge these shocks via an insurance contract with loading and chooses optimal perishable consumption, portfolio holdings, and insurance coverage to maximise expected discounted CRRA utility. Exploiting the homogeneity of the problem, we reduce the Hamilton--Jacobi--Bellman equation to a static one-dimensional optimisation over constant portfolio shares and derive a semi-explicit optimal strategy. We then prove a verification theorem for the associated jump-diffusion wealth process with insurance, establishing the existence and optimality of this constant-fraction strategy under explicit transversality conditions for both risk-aversion regimes $0<γ<1$ and $γ>1$. Numerical experiments illustrate the impact of stochastic depreciation risk and insurance loading on the optimal allocation to financial assets, durable goods, and insurance coverage.

econ.GN↗

arXiv2025

Optimal risk mitigation by deep reinsurance

We consider an insurance company which faces financial risk in the form of insurance claims and market-dependent surplus fluctuations. The company aims to simultaneously control its terminal wealth (e.g. at the end of an accounting period) and the ruin probability in a finite time interval by purchasing reinsurance. The target functional is given by the expected utility of terminal wealth perturbed by a modified Gerber-Shiu penalty function. We solve the problem of finding the optimal reinsurance strategy and the corresponding maximal target functional via neural networks. The procedure is illustrated by a numerical example, where the surplus process is given by a Cramér-Lundberg model perturbed by a mean-reverting Ornstein-Uhlenbeck process.

q-fin.RM↗

arXiv2025

Scenario-based actuarial climate risk assessment via calibration of the DICE model to the shared socioeconomic pathways

Accounting for climate-related risks is an emerging problem for life insurers around the world. In this paper, we demonstrate how scenario trajectories for global temperature can be obtained using the cost-benefit Dynamic Integrated Climate-Economy (DICE) model calibrated to the five Shared Socioeconomic Pathways (SSPs). These scenarios can also be calculated under different carbon emission mitigation targets such as achieving net-zero carbon emissions by a specific year. We show how to calibrate the DICE model to align industrial and land-use carbon emissions with projections from six leading process-based integrated assessment models (IAMs): IMAGE, MESSAGE--GLOBIOM, AIM/CGE, GCAM, REMIND--MAgPIE and WITCH--GLOBIOM. The obtained scenario trajectories of global temperature can be linked to the climate-change induced excess mortality in various regions that, in turn, can be used for stress testing of life insurance portfolios. We illustrate this using synthetic portfolios of life insurance and annuity products.

econ.GN↗

arXiv2023

Life cycle insurance, bequest motives and annuity loads

We investigate insurance purchases when bequest motives are age-varying and life insurance and life annuities both carry loads. The existing life cycle literature assumes bequests are normal goods without being either necessities or luxuries. Much of the literature also assumes implicitly that life annuity loads are negative. A key finding of the literature is that the demand for life insurance and the demand for life annuities are symmetrical. It is optimal to buy life-contingent insurance throughout life, even under loads. A life annuity phase backs directly onto a life insurance phase. We find that realistic examples with positive loads on both products reveal up to two distinct periods of non-participation, one in midlife and the other adjoining the maximum age. We highlight examples with necessity bequests during child-rearing years and luxury bequests thereafter. This set of assumptions explains why a substantial demand for life insurance during child-rearing years can co-exist with negligible demand for life annuities later on. A realistic 18% load on both products generates this outcome.

econ.GN↗

arXiv2023

Importance sampling for option pricing with feedforward neural networks

We study the problem of reducing the variance of Monte Carlo estimators through performing suitable changes of the sampling measure which are induced by feedforward neural networks. To this end, building on the concept of vector stochastic integration, we characterize the Cameron-Martin spaces of a large class of Gaussian measures which are induced by vector-valued continuous local martingales with deterministic covariation. We prove that feedforward neural networks enjoy, up to an isometry, the universal approximation property in these topological spaces. We then prove that sampling measures which are generated by feedforward neural networks can approximate the optimal sampling measure arbitrarily well. We conclude with a comprehensive numerical study pricing path-dependent European options for asset price models that incorporate factors such as changing business activity, knock-out barriers, dynamic correlations, and high-dimensional baskets.

q-fin.CP↗