Academia.eduAcademia.edu

Figure 8 – uploaded by Viviana Fernandez

See full PDFdownloadDownload figure
Conditional EVT percent error: 1.14; Unconditional EVT percent error: 1.05

Figure 8 Conditional EVT percent error: 1.14; Unconditional EVT percent error: 1.05

Related Figures (18)

Notes: IPSA stands for Price Index of Selective Stocks, and gathers the 40 most traded stocks on the Santiago Stock Exchange, Chile. The sample period for IPSA is January 1990-November 2002 (data source: Bloomberg, Central Bank of Chile); for the Ch$/US$ exchange rate is 1988-2002 (data source: Central Bank of Chile); for copper is 1998-2002 (data source: London Metal Exchange); and, for the proxy of a 1-zero coupon bond is 1993-2001 (data source: Bloomberg). fj represents the autocorrelation coefficient of order j. P-values are between parentheses. DESCRIPTIVE STATISTICS OF DAILY RETURNS
Notes: IPSA stands for Price Index of Selective Stocks, and gathers the 40 most traded stocks on the Santiago Stock Exchange, Chile. The sample period for IPSA is January 1990-November 2002 (data source: Bloomberg, Central Bank of Chile); for the Ch$/US$ exchange rate is 1988-2002 (data source: Central Bank of Chile); for copper is 1998-2002 (data source: London Metal Exchange); and, for the proxy of a 1-zero coupon bond is 1993-2001 (data source: Bloomberg). fj represents the autocorrelation coefficient of order j. P-values are between parentheses. DESCRIPTIVE STATISTICS OF DAILY RETURNS
Note: Histograms are drawn with bin widths chosen by the Scott rule (see Venables and Ripley, 2002, chapter 5).
Note: Histograms are drawn with bin widths chosen by the Scott rule (see Venables and Ripley, 2002, chapter 5).
Notes: The sample periods are the same as those described at the bottom of Table 1. GARCH (1,1) PARAMETER ESTIMATES AND SPECIFICATION TESTS
Notes: The sample periods are the same as those described at the bottom of Table 1. GARCH (1,1) PARAMETER ESTIMATES AND SPECIFICATION TESTS
Notes: the degrees of freedom of the t distribution have been determined in each case by fitting a GARCH(1,1) model to the data assuming t innovations. They are the following: 8.9 for IPSA, 3.6 for the CH$/US$ exchange rate, 9.6 for copper, and 3.4 for the proxy of a 1-year zero coupon bond.
Notes: the degrees of freedom of the t distribution have been determined in each case by fitting a GARCH(1,1) model to the data assuming t innovations. They are the following: 8.9 for IPSA, 3.6 for the CH$/US$ exchange rate, 9.6 for copper, and 3.4 for the proxy of a 1-year zero coupon bond.
In sample (1990-2000) percent error: 0.77; Out of sample: (2001-2002) percent error=0.42.
In sample (1990-2000) percent error: 0.77; Out of sample: (2001-2002) percent error=0.42.
In-sample (1988-2000), percent error: 0.62; Out-of-sample (2001-2002), percent error: 1.61.
In-sample (1988-2000), percent error: 0.62; Out-of-sample (2001-2002), percent error: 1.61.
In sample (1998-2000) percent error: 0.99; Out-of-sample (2001-2002): 0.66%
In sample (1998-2000) percent error: 0.99; Out-of-sample (2001-2002): 0.66%
In sample (1993-1999) percent error: 0.88; Out-of-sample (2000-2001) percent error: 1.4
In sample (1993-1999) percent error: 0.88; Out-of-sample (2000-2001) percent error: 1.4
Conditional EVT percent error: 1.17; Unconditional EVT percent error: 1.31
Conditional EVT percent error: 1.17; Unconditional EVT percent error: 1.31
Conditional EVT percent error: 0.8; Unconditional EVT percent error: 0.53
Conditional EVT percent error: 0.8; Unconditional EVT percent error: 0.53
Conditional EVT percent error: 1.3; Unconditional EVT percent error: 1.62
Conditional EVT percent error: 1.3; Unconditional EVT percent error: 1.62
(a) Population Quantiles
(a) Population Quantiles
Notes: (1) “Cond.” and “Unc.” stand for conditional and unconditional, respectively. (2) Parameter estimates are obtained by the method of maximum likelihood, as described earlier. (3) A p-value less than 5 percent is taken as evidence against the null hypothesis.
Notes: (1) “Cond.” and “Unc.” stand for conditional and unconditional, respectively. (2) Parameter estimates are obtained by the method of maximum likelihood, as described earlier. (3) A p-value less than 5 percent is taken as evidence against the null hypothesis.
Notes: 1 UF was equivalent to Ch$16,022.5, and the exchange rate was Ch$678.8 per US$ on Sep- tember 13, 2001. 99-PERCENT VALUE AT RISK FOR A POSITION ON CH$ 100 MILLION ON SEPTEMBER 13, 2001
Notes: 1 UF was equivalent to Ch$16,022.5, and the exchange rate was Ch$678.8 per US$ on Sep- tember 13, 2001. 99-PERCENT VALUE AT RISK FOR A POSITION ON CH$ 100 MILLION ON SEPTEMBER 13, 2001
TABULATION OF RETURNS ON THE Ch$/US$ EXCHANGE RATE AND THE (PROXY OF A ZERO-COUPON BOND
TABULATION OF RETURNS ON THE Ch$/US$ EXCHANGE RATE AND THE (PROXY OF A ZERO-COUPON BOND
Notes: (1) “Cond.” and “Unc.” stand for conditional and unconditional, respectively. (2) The sample periods for the S&P 500, the Wilshire 5000, and the Russell 3000 are, respectively, January 1980-December 2002, January 1991-December 2002, and January 1988-December 2002 (data source: Bloomberg). All retums series are daily. (3) A p-value less than 5 percent is taken as evidence against the null hypothesis.
Notes: (1) “Cond.” and “Unc.” stand for conditional and unconditional, respectively. (2) The sample periods for the S&P 500, the Wilshire 5000, and the Russell 3000 are, respectively, January 1980-December 2002, January 1991-December 2002, and January 1988-December 2002 (data source: Bloomberg). All retums series are daily. (3) A p-value less than 5 percent is taken as evidence against the null hypothesis.
Related topics:
Extreme Value TheoryValue at RiskMarket Risk
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved