- sarty06711
**Posts:**7**Joined:**

Please send me a copy also at nyc_big_ben@yahoo.com

It somewhat difficult to connect copula used as mathematical notion with financial counterpart. In mathematical setting one needs joint multivariate distribution. In this case in this case copula is a connection between the multivariate distribution function and its marginal distribution. Here there is only one input: multivariate distribution and marginal distributions could be found from it uniquely by integration. In finance it looks like people are trying to avoid to state that the case they are studied ether very simple,i.e. marginal distributions uniquely definea multivariate distribution function or if a multivariate distribution function is could not be explicitly found there exist a fat chance that the set of marginal distributions are put into copula from another multivariate distribution function. In the later case this will not a distribution function.

- yandong2020
**Posts:**12**Joined:**

Hello Mr. Melchi,Would you bother to send me your spreadsheet? Many thanks!yandong2020@gmail.com

Dear all,I definitely need your advice. i work at the university as a research assistant(department of int. economics), and the initial topic of my dissertation was supposed to be about the current financial crisis. However, i got digging into financial contagion recently and got absorbed in the world of copulas. I am quite surprised by the extent of the applications of copulas and would like to base my research on them. However, I don't have a degree in math or smth (economics and finance instead), so I would like to ask those who intensively deal with copulas and their applications to advise me whether I would cope with it or it is only (financial) mathematicians who can handle it.Thanks in advance!

wow.... I was wondering what brought forth this old thread! errr... give it a try, though with the bad name received by some of the methods which you might find in the pre-2007/08 published textbooks, you might be advised to divert your energies in more productive areas. Also, better advise might come forth from the more experienced members soon!QuoteOriginally posted by: narkarDear all,I definitely need your advice. i work at the university as a research assistant(department of int. economics), and the initial topic of my dissertation was supposed to be about the current financial crisis. However, i got digging into financial contagion recently and got absorbed in the world of copulas. I am quite surprised by the extent of the applications of copulas and would like to base my research on them. However, I don't have a degree in math or smth (economics and finance instead), so I would like to ask those who intensively deal with copulas and their applications to advise me whether I would cope with it or it is only (financial) mathematicians who can handle it.Thanks in advance!

- leocabbage
**Posts:**5**Joined:**

QuoteOriginally posted by: tristanreidA 2 dimensional copula that maps from [0,1]^2->[0,1] has the 3 following properties (from Nelson's introduction):1. For every u in [0,1] C(0,u)=C(u,0)=02. For every u in [0,1] C(u,1)=u and C(1,u)=u3. For every (u1,u2),(v1,v2) in [0,1]x[0,1] with u1<=v1 and u2<=v2 C(v1,v2)-C(v1,u2)-C(u1,v2)+C(u1,u2)>=0Having the first property also means being 'grounded'.Having the 3rd property is also called '2-increasing'.I'm trying to upload a picture of a copula, but I don't see an 'attach' button anywhere.You can see the properties described above pretty clearly from a picture.If you have octave or matlab, here's a function to produce a pretty copula graph (for matlab, take out all the 'g's!):##this is how many steps to put in the function, set it to whatever you want.n=40; ##this is theta. this is a Gumbel-Hougaard copula, if you want details I'll post more## in short, you can adjust theta to see a steeper copulat=2; x = y = linspace (0.01, 1, n)'; [xx, yy] = meshgrid (x, y); z = exp(-(((-log(xx)).^t+(-log(yy)).^t).^(1/t))); gset pm3d gset key below gset border 4095 gset surface gset samples 25 gset isosamples 20 gset ticslevel 0 mesh (x, y, z); gset nohidden3d replot view(15,30)-t.I removed all the 'g's, but Matlab still indicates "??? Error using ==> setVector of handles not permitted for set(h)".

I reply post from narkar. I hope is not too late.Which Archimedean Copula is the right one?This paper presents the concept of copula from a practical standpoint. Given the widened use of the multinormal distribution, we argue its inadequacy, while advocate for using the copula as an alternative and better approach. We examine what the copulas are used for within of risk management. Then we expose a guide to choose both the margins and the Archimedean copula that better fit to data. In addition, we provide an algorithm to simulate random bivariate from Archimedean copula. In order to cover the gap between the theory and its practical implementation VBA codes are provided. They are used in a numerical example that illustrates the use of the copula in the pricing of a first-at-default contract. Two spreadsheets accompany to paper, by presenting step by step all practical applications covered.Tools for sampling Multivariate Archimedean CopulasA hurdle for practical implementation of any multivariate Archimedean copula was the absence of an efficient method for generating them. The most frequently used approach named conditional distribution one, involves differentiation step for each dimension of the problem. For this reason, it is not feasible in higher dimension. Marshall and Olkin proposed an alternative method, which is computationally more straightforward than the conditional distribution approach. We present the tools necessary for understand it and use it. We introduce the Laplace Transform and its role in the generation of multivariate Archimedean copulas. In order to cover the gap between the theory and its practical implementation VBA code and R one are provided.I hope this helps.

Last edited by mrmelchi on April 26th, 2009, 10:00 pm, edited 1 time in total.

I'd appreciated if you could send me a copy of that 'famous' speadsheet. The e-mail address: janneyxu@hotmail.com. Many thanks!

Hi Mario, could you send me the spreadsheet please. My email address is: joo21@cantab.netThanks

I've never been clear exactly what the problem with copulas in the credit mess was. The obvious things are that they like VaR are trying to give one number that management can use and that the Gaussian distribution is assumed. It also appears that the volume of data for defaults, recovery, and credit instrument was much to small and even if historical data had been used it was also so small that it would not be statistically significant. Also assumptions about independence was wrong. Were the other main problems that not only management but the quants, accounting and even risk people believed and relied on what the copula told them ? They knew it could not be 'right' but figured it was close enough that they did not need to worry ? Thought the housing and corporate boom would never end and other credit protection [counterparty, AIG and other insurers] was strong enough that counterparty risk was not significant and relying on the accuracy of the credit rating agencies, esp. with the SPV ratings they could get from the agencies. I.e. "drinking their own cool-aid." Copulas are a well known statistical method but like any use of stats or math, can be mis-used and bad assumptions [Gaussian] and bad/weak data can be used. David Li proposed a method to use copulas for finance but [perhaps much after the fact] warned about what they could and could not be used for. Thus was it really the users rather than the method/assumptions that were the problem ?

Hi Mr Melchiori,I've seen your message in contingencyanalysis.com which says that you have spreadsheets for bootstrapping and interpolating yield curve using linear and spline, and I would like get your spreadsheets about Copulas, too. Could you send me those spreadsheets to julio.reyes@subuschile.cl?Thanks a lot

Can I get a copy of the spreadsheet as well? Many thanks! Henderson.geoffrey@gmail.com

- sachintodkarus
**Posts:**1**Joined:**

I would also like to have the spreadsheets for bootstrapping and interpolating the yield curve. Can you please send it at sachintodkarus@yahoo.comThanks in advance.

Hi Mr MechiI would also like your spreadsheet for copulas and bootstrapping please - ae8812@hotmail.co.ukMany thanks!Anthony

Hi Mr Melchiori,Could you also send me your spreadsheets on copulas and if also available on bootstrapping and interpolating yield curve using linear and spline to this email.Greatly appreciated,Dar.