Page 396 - Contributed Paper Session (CPS) - Volume 4

CPS2449 Louisa Nolan et al.

What can data science do for economic
statistics?
Louisa Nolan, Jeremy Rowe, Steven Hopkins, Sonia Williams
Data Science Campus, Office for National Statistics, UK

Abstract
The Data Science Campus of the UK’s Office for National Statistics was set up
to explore how data science could change the evidence base for the UK. One
of the key areas for exploration is economics. More than two years after the
Campus was set up, this presentation looks at what has been achieved.
How have we enhanced or supplemented traditional economic statistics?
What are the challenges for incorporating unstructured data, collected by third
parties into official statistics? How can we best share what we have learned,
and what are the challenges for implementing data science prototypes into
production?
Here, we present examples of economic data science from the Campus,
including from our Faster economic indicators project, and use these to
illustrate how we have addressed the challenges described.

Keywords
data science; economics; big data; official statistics

1. Introduction
The appetite for faster, more granular and more comprehensive
information has never been higher. Policymakers and analysts demand faster,
better insights into the state of economies and societies in order to make well-
informed, timely decisions on national and international matters.
With the growing availability of big data and large administrative datasets,
and the tools, technology and skills to understand, process and analyse these,
National Statistics Institutes (NSIs) are being challenged to produce outputs
that meet the growing demand for richer, more timely data.
In this paper, we use three economics projects from the Campus to
illustrate how we have been meeting this challenge:
• initial work from our Faster indicators of UK economic activity
programme (1), which uses three datasets: her Majesty’s Revenue and
Customs (HMRC) UK Value Added Tax (VAT) returns; ship tracking data
from automated identification systems (AIS) for UK waters; and road
traffic sensor data for England
• understanding the characteristics of high growth companies using
non-traditional data sources (2)

385 | I S I W S C 2 0 1 9