Page 430 - Special Topic Session (STS) - Volume 3

STS552 João Falcão Silva et al.
of sector N, and 38% of sector P (i.e. at the aggregate level 38% of liabilities
12
to sector P are not classified to sectors C, G and H) .
The interlinkages with other data sources using the mirror data is crucial
when analysing the households sector. It offers the use of debtor banks’
liabilities to derive assets of the households with banks. We consider BIS
reporting banks’ cross-border liabilities to the households sector as the
measure of assets of this sector with banks abroad in two alternative methods:
using aggregate and bilateral data.
We noted that the coverage of reported data differs across counterparty
countries. In the overall aggregate as of Q4 2018, 38 of 47 countries reported
subsectors F (54%) and P (41%) of sector N with coverage of 95% and
remaining 5%, accounted by 9 countries, do not report any sub-sectors of
13
sector N . While the estimation of the remaining 5% into subsectors F and P
could be done by various methods, a simple way is to use the proportional
14
approach . If this approach is adopted, the share of F and P would be 57%
and 43% respectively. As shown in Graph 1 we estimate amounts from Q4
2011 (i.e. for period when no information on subsectors was available) for all
subsectors of non-banks in all counterparty countries (top right panel) and of
which those in Portugal (bottom right panel). These estimates are based on
aggregated data of all reporting countries using simple proportional
approach, both for the aggregate of all counterparty countries and of which
15
those in Portugal .
I – Aggregate level estimations
When confidentiality for bilateral data restrictions arises, mirror data
exercises can only be applied at an aggregate level. According to the LBS
information each country can estimate its deposits abroad by using the
reported information for the aggregate of ‘all reporting countries’ vis-à-vis
sectors F, P and subsectors of P and in particular sector H, with possible
16
alternatives . This procedure provides estimated amounts for a given country

12 Seventeen countries do not report subsectors of P: Bahrain, Brazil, Chile, China, Curacao,
Finland, Greece, Hong Kong SAR, Japan, Jersey, Macao SAR, Mexico, Panama, Philippines,
Singapore, Turkey and United States.
13 Of these 9 countries, Singapore comprises about 60%, Jersey and Bahrain each comprise
about 15% and rest by another 6 countries (Brazil, Chile, Curacao, Greece, Mexico and Panama).
14 I.e. allocate 54/95 of 5% to sector F and 41/95 of 5% to sector P.
15 This is done in two steps: (1) Sector F and P first estimated by proportional allocation of Sector
X amounts (see footnote 13). (2) New unallocated sector K amounts, after estimating P, were
allocated in the same way proportionally to sectors C, G and H.
16 All countries do not yet report full breakdown of non-bank subsectors. Thus the sector-
breakdown of aggregate positions by banks in ‘all reporting countries’ vis-à-vis individual
counterparty countries are incomplete. We propose to proportionally allocate residual amounts
to reportable subsectors (see footnotes 13 and 14). The average share for each reportable sub-
sectors (i.e. after reallocating residual amounts) from the latest quarters are applied to sector N
419 | I S I W S C 2 0 1 9