Page 433 - Special Topic Session (STS) - Volume 3

STS552 João Falcão Silva et al.
to 61% and 5% respectively. Graph1b shows that there are no unallocated
amounts under the estimated method and that the share of household sector
in total of non-banks represents the highest estimated increase of 12.2%
approximately (from reported 2.7% to estimated 14.9%) whereas the Non-
financial corporations share increase 10.9% (from 15.6% to 26.5%). On the
contrary, the government sector decreases from 1.1% reported share to 0.9%.
We elect Portugal as a domestic country to estimate households’ cross
border positions. We find that as of Q4 2018, 97.7% of deposit liabilities to
non-banks ($12.9 billion) are reported with sector breakdown in to sector F
($2.6 billion, 20.4%) and sector P ($10 billion, 77.3%). In the total reported
amount for sector N, the share of the unallocated amounts (sector X plus
sector K) correspond to 6.9%. In addition, 72.7% of deposit liabilities to sector
23
P is available with subsector breakdown (Graph 1c) . In addition, H sector
comprise nearly $6.9 billion, 53.6% of reported amount for sector P, sector C
represents 18.8% and sector G has almost a null share. The estimated amounts
show that Non-financial corporations share (Graph 1d) increases by 1.6%
(from 18.8% to 20.5%) and Households from a reported share of 53.6% to an
estimated of 58.3%.
Our results show that this estimation method appears consistent not only
for counterparties in all countries but also for individual counterparty countries
such as Portugal. The estimated shares/amounts for Portugal are expected to
be close to the actual amounts because coverage of reporting is as high as
93.1% for non-financial sub-sectors. However, the simple proportional
estimates could be improved by estimating sub-sectors for individual
reporting countries and then using simple/weighted average of the estimated
shares from the reported data to estimate the bilateral positions of countries
that do not report these subsectors. Further refinement of estimates
backwards could be achieved using a 4 or 8 quarter moving average and also
test the robustness of our estimation by comparing with reported data when
available.

5. Discussion and Conclusion
We develop this methodological framework on uses of reported data,
estimation of non-reported data and data gaps, aiming to provide users with
more complete information. The detailed version of this work is in progress (at
advanced stage) and will provide much more additional details that we could
not include in this version for the conference due to limitation of length. It is
needless to mention that the mirror data methodology we propose and have

This unallocated share comes to 5.3% of total liabilities to non-banks (aggregate). As of Q4
23
2018, while 43 countries report liabilities to sector N of Portugal, 28 countries report cross-
border deposit liabilities to sector H and the share of unallocated non-bank is only 7% of non-
bank.
422 | I S I W S C 2 0 1 9