Undergraduate Honors Thesis

Abstract

• We examine four autoregressive conditional heteroskedasticity (ARCH) type models,
  including one long memory and two asymmetric models, to assess their usefulness in Con-
  ditional Value at Risk and Conditional Expected Shortfall estimation. Alongside the four
  ARCH type models, we consider three additional models: historical simulation, standard
  parametric, and RiskMetrics. Estimation is performed on the five foreign exchange rates
  of the BRICS (Brazil, Russia, India, China, South Africa) emerging economies. We find
  that there is no single best model but that model selection for risk analysis should be done
  on an case by case basis. Furthermore, while the four ARCH type models produce similar
  results when estimating risk measures, we find that the standard GARCH model typically
  outperforms the asymmetric and long memory models when applied to out of sample data.

Creator

• Dyer, Henry

Date Awarded

• 2024-04-09

Academic Affiliation

• Applied Mathematics

Advisor

• Huang, Yu-Jui

Committee Member

• Dougherty, Anne Margaret
• Martins-Filho, Carlos B

Granting Institution

• University of Colorado Boulder

Last Modified

• 2024-04-18

Resource Type

• Undergraduate Honors Thesis

Rights Statement

• In Copyright

Language

• English [eng]

Relationships

In Collection:

• Arts and Sciences Honors Program

Items

Thumbnail	Title	Date Uploaded	Visibility	Actions
	dyer_henry_thesis.pdf	2024-04-16	Public	Download