simdiff: Simulation of diffusion processes.

In ESGtoolkit: Toolkit for Monte Carlo Simulations

Description

This function makes simulations of diffusion processes, that are building blocks for various risk factors' models.

Usage

simdiff(
  n,
  horizon,
  frequency = c("annual", "semi-annual", "quarterly", "monthly", "weekly", "daily"),
  model = c("GBM", "CIR", "OU"),
  x0,
  theta1 = NULL,
  theta2 = NULL,
  theta3 = NULL,
  lambda = NULL,
  mu_z = NULL,
  sigma_z = NULL,
  p = NULL,
  eta_up = NULL,
  eta_down = NULL,
  eps = NULL,
  seed = 123
)

Arguments

n	number of independent observations.
horizon	horizon of projection.
frequency	either "annual", "semi-annual", "quarterly", "monthly", "weekly", or "daily" (1, 1/2, 1/4, 1/12, 1/52, 1/252).
model	either Geometric Brownian motion-like ("GBM"), Cox-Ingersoll-Ross ("CIR"), or Ornstein-Uhlenbeck ("OU"). GBM-like (GBM, Merton, Kou, Heston, Bates) dX_t = θ_1(t) X_t dt + θ_2(t) X_t dW_t + X_t JdN_t CIR dX_t = (θ_1 - θ_2 X_t) dt + θ_3√(X_t) dW_t Ornstein-Uhlenbeck dX_t = (θ_1 - θ_2 X_t)dt + θ_3 dW_t Where (W_t)_t is a standard brownian motion : dW_t ~~ ε √(dt) and ε ~~ N(0, 1) The ε is a gaussian increment that can be an output from simshocks. For 'GBM-like', θ_1 and θ_2 can be held constant, and the jumps part JdN_t is optional. In case the jumps are used, they arise following a Poisson process (N_t), with intensities J drawn either from lognormal or asymmetric double-exponential distribution.
x0	starting value of the process.
theta1	a numeric for model = "GBM", model = "CIR", model = "OU". Can also be a time series object (an output from simdiff with the same number of scenarios, horizon and frequency) for model = "GBM", and time-varying parameters.
theta2	a numeric for model = "GBM", model = "CIR", model = "OU". Can also be a time series object (an output from simdiff with the same number of scenarios, horizon and frequency) for model = "GBM", and time-varying parameters.
theta3	a numeric, volatility for model = "CIR" and model = "OU".
lambda	intensity of the Poisson process counting the jumps. Optional.
mu_z	mean parameter for the lognormal jumps size. Optional.
sigma_z	standard deviation parameter for the lognormal jumps size. Optional.
p	probability of positive jumps. Must belong to ]0, 1[. Optional.
eta_up	mean of positive jumps in Kou's model. Must belong to ]0, 1[. Optional.
eta_down	mean of negative jumps. Must belong to ]0, 1[. Optional.
eps	gaussian shocks. If not provided, independent shocks are generated internally by the function. Otherwise, for custom shocks, must be an output from simshocks.
seed	reproducibility seed

Value

a time series object.

Author(s)

Thierry Moudiki

References

Black, F., Scholes, M.S. (1973) The pricing of options and corporate liabilities, Journal of Political Economy, 81, 637-654.

Cox, J.C., Ingersoll, J.E., Ross, S.A. (1985) A theory of the term structure of interest rates, Econometrica, 53, 385-408.

Iacus, S. M. (2009). Simulation and inference for stochastic differential equations: with R examples (Vol. 1). Springer.

Glasserman, P. (2004). Monte Carlo methods in financial engineering (Vol. 53). Springer.

Kou S, (2002), A jump diffusion model for option pricing, Management Sci- ence Vol. 48, 1086-1101.

Merton, R. C. (1976). Option pricing when underlying stock returns are discontinuous. Journal of financial economics, 3(1), 125-144.

Uhlenbeck, G. E., Ornstein, L. S. (1930) On the theory of Brownian motion, Phys. Rev., 36, 823-841.

Vasicek, O. (1977) An Equilibrium Characterization of the Term Structure, Journal of Financial Economics, 5, 177-188.

See Also

simshocks, esgplotts

Examples

kappa <- 1.5
V0 <- theta <- 0.04
sigma_v <- 0.2
theta1 <- kappa*theta
theta2 <- kappa
theta3 <- sigma_v

# OU

sim.OU <- simdiff(n = 10, horizon = 5, 
               frequency = "quart",  
               model = "OU", 
               x0 = V0, theta1 = theta1, theta2 = theta2, theta3 = theta3)
head(sim.OU)
par(mfrow=c(2,1))
esgplotbands(sim.OU, xlab = "time", ylab = "values", main = "with esgplotbands")                
matplot(time(sim.OU), sim.OU, type = 'l', main = "with matplot")                


# OU with simulated shocks (check the dimensions)

eps0 <- simshocks(n = 50, horizon = 5, frequency = "quart", method = "anti")
sim.OU <- simdiff(n = 50, horizon = 5, frequency = "quart",   
               model = "OU", 
               x0 = V0, theta1 = theta1, theta2 = theta2, theta3 = theta3, 
               eps = eps0)
par(mfrow=c(2,1))
esgplotbands(sim.OU, xlab = "time", ylab = "values", main = "with esgplotbands")                
matplot(time(sim.OU), sim.OU, type = 'l', main = "with matplot")
# a different plot
esgplotts(sim.OU)


# CIR

sim.CIR <- simdiff(n = 50, horizon = 5, 
               frequency = "quart",  
               model = "CIR", 
               x0 = V0, theta1 = theta1, theta2 = theta2, theta3 = 0.05)
esgplotbands(sim.CIR, xlab = "time", ylab = "values", main = "with esgplotbands")                  
matplot(time(sim.CIR), sim.CIR, type = 'l', main = "with matplot")



# GBM

eps0 <- simshocks(n = 100, horizon = 5, frequency = "quart")
sim.GBM <- simdiff(n = 100, horizon = 5, frequency = "quart",   
               model = "GBM", 
               x0 = 100, theta1 = 0.03, theta2 = 0.1, 
               eps = eps0)
esgplotbands(sim.GBM, xlab = "time", ylab = "values", main = "with esgplotbands")                
matplot(time(sim.GBM), sim.GBM, type = 'l', main = "with matplot")


eps0 <- simshocks(n = 100, horizon = 5, frequency = "quart")
sim.GBM <- simdiff(n = 100, horizon = 5, frequency = "quart",   
               model = "GBM", 
               x0 = 100, theta1 = 0.03, theta2 = 0.1, 
               eps = eps0)                
esgplotbands(sim.GBM, xlab = "time", ylab = "values", main = "with esgplotbands")                
matplot(time(sim.GBM), sim.GBM, type = 'l', main = "with matplot")

ESGtoolkit documentation built on July 1, 2020, 11:24 p.m.