Risk Reporting
169
(UL: large, infrequent, severe incidents), the EL distribution is more concentrated,
more symmetric and more suitable for averaging. Large loss events need to be
reported individually, while EL can be the object of classic descriptions, with
minimum, maximum and average.
B e n c h m a r k t o G r o s s I n c o m e
Reporting the losses to a benchmark, typically the
gross income, can be very effective to attract the attention of senior management.
In my experience, firms that have developed a high-performing operational risk
management practice experience a total volume of losses between 1.8% and 2.2% of
operational losses to gross income. Between 2.2% and 3% is common and above 3%
are higher levels of operational losses, often due to environments with many manual
and complex processes or a history of underinvestment in people and systems. A ratio
of operational losses to gross income inferior to 1.5% is more likely to originate
from underreporting rather than from good operational performance. Interestingly,
firms for the food retail and telecommunication sectors told me they are experiencing
operational losses around 2% of gross income as well and working hard to reduce
those losses.
T U R N I N G D A T A I N T O S T O R I E S
In risk reporting, like many other forms of reporting, the value of information lies
in deviations from the norm. For example, credit card frauds are detected by abnor-
mal spending patterns, excellent traders (positive outliers) have unusually long per-
formance records, rogue traders (negative outliers) display unusually low volatility in
performance, the best managers have greater staff retention and the highest produc-
tivity levels, and poor suppliers have the largest number of operational mishaps. The
value of information lies in data patterns, in concentrated parts of the distributions and
in distances between observations. You need to pinpoint where the most or least num-
ber of events are, what the variations are in performance, the record highs and record
lows, and how to interpret everything. To turn data into stories, by all means look at
the individual data points, graphically and numerically, clean obvious entry errors and
missing fields, and then look for outliers, clusters and patterns. Next, seek interpre-
tation for these behaviors before distilling the points into summary statistics. Adjust
your analysis to your data.
Risk reporting is the opportunity to investigate the reality behind numbers and
to determine what is going well and what might go wrong. Remember that positive
outliers carry just as much information as negative ones. Sadly, most firms still dedicate
more attention to reporting problems than highlighting and explaining success stories.
Do'stlaringiz bilan baham: