Abstract
This paper combines the Integrated Cyber Risk Assessment Model (ICRAM) with Monte Carlo simulation and vector autoregression (VAR) to assess and forecast cyber risk. In the model, risk is the product of three components: how often threat events occur, how vulnerable the organization is, and how much a successful incident costs. The inputs are calibrated on survey data from 250 public and private organizations in Armenia. We analyse three scenarios (realistic, optimistic and pessimistic) at five levels of cyber defense spending, from 1% to 5% of the budget. For each scenario, 10,000 Monte Carlo runs give the distribution of annual losses, with its mean, percentiles and tail measures (VaR95, ES95), and a VAR(1) model describes how the risk components move together over time. Expected losses fall steadily as the investment share rises. Going from 3% to 4% is associated with a 43–51% drop in expected losses, depending on the scenario, although each additional percentage point saves less in absolute terms. The loss distribution is strongly right-skewed (skewness ≈ 2.1; ES95 is roughly three times the mean), so budget decisions should take the tail into account and not rely on average losses alone. Since the survey is cross-sectional, the link between investment and risk should be read as an association, not a causal effect. The approach offers a risk-based way to justify cybersecurity spending in economic terms.

This publication has 14 references indexed in Scilit: