tommasobelluzzo / systemicrisk Goto Github PK

Dear Tommaso,

I am trying to run the CrossSectional part of this project. Is it correct to edit run.m like this?(see below) BTW, nothing else changed.

measures_setup = { % NAME ENABLED ANALYZE COMPARE FUNCTION 'Component' false true true @(ds,temp,file,analyze)run_component(ds,temp,file,bw,0.99,0.98,0.05,0.2,0.75,analyze); 'Connectedness' false true true @(ds,temp,file,analyze)run_connectedness(ds,temp,file,bw,0.05,false,0.06,analyze); 'CrossEntropy' false true true @(ds,temp,file,analyze)run_cross_entropy(ds,temp,file,bw,'G',0.4,'W','N',analyze); 'CrossQuantilogram' false true true @(ds,temp,file,analyze)run_cross_quantilogram(ds,temp,file,bw,0.05,60,'SB',0.05,100,analyze); 'CrossSectional' true true true @(ds,temp,file,analyze)run_cross_sectional(ds,temp,file,0.95,0.40,0.08,0.40,3,analyze); 'Default' false true true @(ds,temp,file,analyze)run_default(ds,temp,file,bw,'BSM',3,0.08,0.45,2,0.10,100,5,0.95,analyze); 'Liquidity' false true true @(ds,temp,file,analyze)run_liquidity(ds,temp,file,bw,21,5,'B',500,0.01,0.0004,analyze); 'RegimeSwitching' false true true @(ds,temp,file,analyze)run_regime_switching(ds,temp,file,true,true,true,analyze); 'Spillover' false true true @(ds,temp,file,analyze)run_spillover(ds,temp,file,bw,10,'G',2,4,analyze); };

If the way I set is correct, I'm confused about the error that MATLAB reported.(see below)

bivariate_caviar>bivariate_caviar_internal (line 78) too many outputs error bivariate_caviar (line 49) [caviar,beta,ir_fm,ir_mf,se,stats] = bivariate_caviar_internal(r,a,cir,cse); error run_cross_sectional>run_cross_sectional_internal (line 111) [caviar,~,ir_fm,ir_mf] = bivariate_caviar(r_i,ds.A); error run_cross_sectional (line 48) [result,stopped] = run_cross_sectional_internal(ds,temp,out,k,d,car,sf,fr,analyze); error run>@(ds,temp,file,analyze)run_cross_sectional(ds,temp,file,0.95,0.40,0.08,0.40,3,analyze) error run (line 159) [result,stopped] = run_function(ds,temp,out,analyze);

I'm new to MATLAB, so maybe I'm doing a stupid thing here.
How to fix the issue to get the results as expected?

I‘ve installed MATLAB Version: 9.1.0.441655 (R2016b), and tried the run.m again. Now the "parse_dataset" funtion works fine for the "Example_Large.xlsx" after "Example_Large.mat" being deleted.

When I set 'CrossSectional' ENABLED true, ANALYZE true, and COMPARE true, some errors reported as below:

Error using cellfun
Input #2 expected to be a cell array, was string instead.

Error in safe_plot>safe_plot_internal (line 50)
r = cellfun(@(x)[' ' x],r,'UniformOutput',false);

Error in safe_plot (line 18)
safe_plot_internal(ipr.handle);

Error in run_cross_sectional>analyze_result (line 322)
safe_plot(@(id)plot_correlations(ds,id));

Error in run_cross_sectional>run_cross_sectional_internal (line 164)
analyze_result(ds);

Error in run_cross_sectional (line 51)
[result,stopped] =
run_cross_sectional_internal(ds,sn,temp,out,k,d,car,sf,fr,analyze);

Error in
run>@(temp,out,analyze)run_cross_sectional(ds,sn,temp,out,0.95,0.40,0.08,0.40,3,analyze)

Error in run (line 121)
[result,stopped] = run_function(temp,out,analyze);

In safe_plot.m, the relevant codes about line 50 are as below:

function safe_plot_internal(handle)

persistent ids;

name = func2str(handle);
name = regexprep(name,'^@\([^)]*\)','');
name = regexprep(name,'\([^)]*\)$','');

try
    id = [upper(name) '-' upper(char(java.util.UUID.randomUUID()))];
catch
    id = randi([0 10000000]);
    
    while (ismember(id,ids))
        id = randi([0 100000]);
    end
    
    ids = [ids; id];
    id = [upper(name) '-' sprintf('%08s',num2str(id))];
end

try
    handle(id);
catch e
    delete(findobj('Type','Figure','Tag',id));
    
    r = getReport(e,'Extended','Hyperlinks','off');
    r = split(r,newline());
    r = cellfun(@(x)['  ' x],r,'UniformOutput',false);  %%% line 50 is here
    r = strrep(strjoin(r,newline()),filesep(),[filesep() filesep()]);

    warning('MATLAB:SystemicRisk',['The following exception occurred in the plotting function ''' name ''':' newline() r]);
end

end

I'm new to MATLAB and it's hard for me to understand these codes completely in a short time. Is there any idea to this issue? Thanks a lot.

Originally posted by @DF-18 in #12 (comment)

I have a dataset that is unbalanced panel data, which means each firm may not have the same time span.

For example, in Shares sheet, based on the default dataset format, the time span is fixed. So in my unbalanced panel dataset, some cells for closing price are null, because those firms has not been listed or has ben delisted. Those cells are blank in the excel file.

When I input this dataset, an error occured as below:

misuse parse_dataset>ensure_field_consistency (line 491)
The 'Shares' sheet contains invalid column types.
error parse_dataset>read_table (line 775)
tab = ensure_field_consistency(name,tab,i,output_vars{i},data_types{i},date_format_dt);
error parse_dataset>parse_table_standard (line 694)
tab = read_table(file,file_name,index,name,date_format,data_types);
error parse_dataset>parse_dataset_internal (line 65)
tab_shares = parse_table_standard(file,file_name,1,'Shares',date_format_base,[],[],true);
error parse_dataset (line 47)
ds =
parse_dataset_internal(file,file_sheets,version,date_format_base,date_format_balance,shares_type,crises_type,distress_threshold);
error run (line 71)
ds = parse_dataset(file,ds_version,'dd/mm/yyyy','QQ yyyy','P','R',0.05);

It says there are some columns in 'Shares' sheet are invalid.

I checked the data format, and all cells are numeric, except the header(1st row) and the date(1st column), which is identical to the default dataset.

I can't figure out why this error occured. Would u mind checking the dataset in attachment?

BTW, I've tried to replace blank cells with "0". And the error reported is the same as before.

Thank u very much.
inputdata.zip

Dear Tommaso,

May I ask a question about the function 'read_table'?

I found that the function is nonexistent in my matlab R2020b. I check the guidelines on the main page of the following products and toolboxes, and they are all installed：

• Computer Vision Toolbox
• Curve Fitting Toolbox
• Econometrics Toolbox
• Financial Toolbox
• Image Processing Toolbox
• Optimization Toolbox
• Parallel Computing Toolbox
• Statistics and Machine Learning Toolbox
• System Identification Toolbox

Could you please tell me where is the source of this function? I would appreciate it a lot if you could give me a hand. Thank you very much.

I need Matlab codes to calculate weekly Marginal Expected Shortfall (MES) and SRISK.

Hi, Tommaso,
Thank you for sharing your code. Unfortunately, when I run the code (file run.m), I get some errors. Matlab is not my usual software, I just wanna implement something. I've run the code with different versions of Matlab, including R2016b, R2014b and R2012a, and the error is the same.
At the beginning, I set the path to the folder where are the code files (SR folder): cd 'C:\Matlab\SR'; after that, I run the 'run' file and get the following errors:

Undefined function or variable 'detectImportOptions'.
Error in parse_dataset>parse_dataset_internal (line 73)
opts = detectImportOptions(file,'Sheet',1);
Error in parse_dataset (line 19)
data = parse_dataset_internal(res.file);
Error in run (line 12)
data = parse_dataset(fullfile(path,'\Datasets\Example.xlsx'));

I have read that usually these errors are due to the different versions of Matlab, but unfortunately, I couldn't solve the problem in this way. Could you please take a look over these errors? And, also, what is the Matlab version in which you run your code?
Thank you very much.

Best,
Nicu

          When I run the run.m in the MATLAB2018B of Windows 10,there is an error:  

错误使用 parallel.FevalFuture/fetchNext (line 217)
The function evaluation completed with an error.

出错 run_spillover>run_spillover_internal (line 70)
[future_index,value] = fetchNext(futures);

出错 run_spillover (line 37)
result = run_spillover_internal(data,out_temp,out_file,ipr.bandwidth,ipr.lags,ipr.h,ipr.generalized,ipr.analyze);

出错 run (line 90)
result_spillover = run_spillover(data,out_temp_spillover,out_file_spillover,252,2,4,true,true);

原因:
错误使用 statmvnrobj (line 23)
Covariance matrix SIGMA is not positive-definite.

How to solve it?

Dear Tommaso,

May I ask some questions about the parameters setting?

The description of input parameters in run_cross_sectional.m is as below.

% [INPUT]
% ds = A structure representing the dataset.
% sn = A string representing the serial number of the result file.
% temp = A string representing the full path to the Excel spreadsheet used as template for the result file.
% out = A string representing the full path to the Excel spreadsheet to which the results are written, eventually replacing the previous ones.
% k = A float [0.90,0.99] representing the confidence level (optional, default=0.95).
% d = A float [0.1,0.6] representing the six-month crisis threshold for the market index decline used to calculate the LRMES (optional, default=0.4).
% car = A float [0.03,0.20] representing the capital adequacy ratio used to calculate SES and SRISK (optional, default=0.08).
% sf = A float [0,1] representing the fraction of separate accounts, if available, to include in liabilities and used to calculate SES and SRISK (optional, default=0.40).
% fr = An integer [0,6] representing the number of months of forward-rolling used to calculate the SRISK, simulating the difficulty of renegotiating debt in case of financial distress (optional, default=3).
% analyze = A boolean that indicates whether to analyse the results and display plots (optional, default=false).
%

The two questions related are:

1. Brownlees and Engle(2016) assumed the liabilities of the firm is non-negotiable. Under this kind of set, does it mean to set the 'fr' parameter to zero?

2. Since 'd' represents the six-month crisis threshold, how could I change the horizontal period from 'six-month' to, like 'one-month' or 'three-month'?

Thanks a lot.

Within the calculate_mes function in the run_cross_sectional.m file, the current bandwidth for the kernel density estimator appears to be based on the Silverman "rule-of-thumb bandwidth estimator":

    r0_n = 4 / (3 * length(r0_m));
    r0_s = min([std(r0_m) (iqr(r0_m) / 1.349)]);
    h = r0_s * (r0_n ^ (-0.2));

However, I believe that the exponent is incorrect. I believe the formula for h should actually be:

    h = r0_s * (r0_n ^ (0.2));

See https://link.springer.com/chapter/10.1007/978-3-030-16272-6_3#Sec11 Section 5.5.

In addition I believe that you may want to scale r0_s by the conditonal market volatility, as everything else within the kernel density estimate is scaled. That is I believe this line:

    r0_s = min([std(r0_m) (iqr(r0_m) / 1.349)]);

Should actually be:

    r0_s = min([std(r0_m ./ s_m) (iqr(r0_m ./ s_m) / 1.349)]);

When using actual P&L, if you don't scale r0_s by s_m, the value of h will be proportional to the standard deviation of the P&L. Practically, this will mean that for the value of f will be approximately .5 as (c ./ s_m) - u) divided by some very large number will converge to approximately 0, since (c ./ s_m) - u) should always be on the scale of approximately -10 to 10.

Dear Tommaso,

It is possible to adjust the code to diplay the results in a excel file for CATFIN, Absorbtion Ratio and Turbulence index for every firm in part, not the average for the whole sample?

Thanks!

Hi, Tommaso,

There is a nice application of cross-quantilogram (Han et al., 2016) to systemic risk. You think it's possible to adapt the R code to Matlab? See https://sites.google.com/site/whangyjhomepage/research/software.
Thanks.

Originally posted by @NicuSV in https://github.com/TommasoBelluzzo/SystemicRisk/issues/5#issuecomment-590282384

Hi Tommaso,
In SystemicRisk/ScriptsModels/connectedness_metrics.m, in order to compute the MES, you wrote the following 2 lines of code that I didn't fully understand the reasoning behind it.

r0_n = 4 / (3 * length(rm_0));
...
h = r0_s * r0_n ^0.2;

Is this explained in any research paper/ book?

Moreover, are you somehow replicating the following formula to obtain the ES and then compute the MES?

I'm really confused.
Thank you,
Suzy

Hello! In the Example_Large.xlsx of Datasets,there is "Separate Accounts".Can you tell me what the "Separate Accounts" is?Thank you very much!

Hi, Tommaso,
Thank you for sharing your code.
In SystemicRisk/ScriptsProbabilistic/calculate_mes.m,
I noticed that you have changed the
lrmes = 1 - exp(-18 .* mes);
to
lrmes = 1 - exp(log(1 - d) .* beta_x);
Why?
Is there any literature supporting this ?
I am So confusing.
Thanks a lot.

Best,
Andy

Hi Tommaso,

Thanks for the amazing works, the toolbox is quite useful！
And I‘m really confused about the output of CROSS-SECTIONAL MEASURES. CoVaR computation requires the coefficients estimates through quantile regression, b = quantile_regression(y,x,a), how to get the estimation results of the coefficient "b"?

Best regards,
Guo