[Gretl-devel] Re: dbnomics: Dataset duplication in window lists
On Wed, 12 Jan 2022, Sven Schreiber wrote:
I'm looking at the Bundesbank (BUBA) datasets through gretl's
DB.NOMICS
addon. (Snapshot from two weeks ago.)
I'm noticing that some datasets are apparently duplicated or even
quadrupled:
- BBK01: appears 4 times
- BBNZ1 (National Accounts...): appears twice
This is a quirk of the Bundesbank provider. There's no bug in this
respect on the gretl end.
The error on trying to access the list of series under BUBA/BBK01
was due to another Bundesbank quirk, namely that the series under
BBK01 (and maybe others, I don't know) do not have a "series_name"
recorded in the JSON from dbnomics. That's handled in version 0.43
of the gretl dbnomics addon: if there's no series_name we'll use the
series_code instead.
Let me add: in my opinion the dbnomics label "series_name" is
misleading. When this element is present (almost always, except for
BUBA, so far as I can tell) it's not what I would think of as a
name, but rather a (typically quite verbose) description. Here's an
example from a BDF (Banque de France) series, misspelling and all:
"series_code" : "SC1.M.FR.1000.F51100.M.X.EUR.E.S",
"series_name" : "Net issues of quoted shares (including
contributions in kind), ESA 95 valuation, Seasonnally adjusted"
Allin