Page 1 of 1
Dynamic conditional correlation "DCC Garch"
Posted: Sat Jul 24, 2021 12:55 pm
by HamaNadia
Dear members
When I am trying to run a DCC Garch model I received this error message:
garch(p=1,q=1,mv=dcc) / R_SP500 R_WTI R_Gold R_Green
## REG20. GARCH Cannot Be Used with Gaps/Missing Values
However, when I checked my dataset I don't found any missing values.
P.S: You can find attached my database
Please any help !!
Re: Dynamic conditional correlation "DCC Garch"
Posted: Sat Jul 24, 2021 1:34 pm
by PeterF
Hello,
your data set has gaps.It is daily data, but it has no data for market holidays. It depends on the options in the Calendar instruction, how your data is treated. If you use Calendar(D) then you will get missing values in the series loaded. However, if you read the data as undated then you should not get the error message.
Regards, PeterF
Re: Dynamic conditional correlation "DCC Garch"
Posted: Sat Jul 24, 2021 10:51 pm
by TomDoan
As Peter says, the data has gaps in the five-day-a-week scheme, so you can't use a straight daily CALENDAR. The following will allow you to "map" the dates to consecutive entries:
OPEN DATA "Dataset.xlsx"
DATA(FORMAT=XLSX,ORG=COLUMNS,JULIAN=JDATE) 1 324 R_SP500 R_WTI R_Gold R_Green
CAL(JULIAN=JDATE)
However, you probably are going to have difficulty fitting a DCC GARCH to those---they just aren't similar enough for DCC to work properly. DCC was proposed to do a set of (say) stock prices jointly; you have an mix of stocks and commodities.