Page 300 - Special Topic Session (STS) - Volume 3
P. 300
STS544 Paolo F. et al.
Nowcast second month Nowcast third month Nowcasts 16 days after StatFi Flash
ME 0.21 -0.12 0.04 0.01
MAE 1.04 1.04 0.79 0.78
RMSE 1.31 1.29 0.98 0.92
MaxE 3.21 3.28 2.15 1.77
Table 4: ME, MAE, RMSE and MaxE for GDO nowcasts, using nowcasts combinations.
The set of predictors is based on trucks' traffic volumes.
The results of Table 4 confirm that the nowcasts produced using traffic
date have a satisfactory predictive performance, very similar to the one based
on firm-level sales. Overall, it is interesting to see that traffic data are allowing
us to create fairly precise estimates of GDP growth well before the official
publication by Statistics Finland. Given the potentially real-time availability of
traffic volumes' measurements, these results indicate the need to further
explore the nowcasting ability of models based on these data.
5. Conclusions
We have examined the potential of large micro-level datasets, in
combination with statis-tical models and machine learning techniques that are
able to handle high-dimensional information sets, for the production of faster
estimates of real economic activity in-dicators, both at the monthly and at the
quarterly frequency. In particular, we have examined the nowcasting
performance of firm-level data, and of trucks' traffic volumes measurements.
We find that a simple combination of the nowcasts obtained from a large
set of machine learning techniques and large dimensional statistical models is
able to produce accurate estimates of monthly real economic activity, or at
least estimates that do not lead to a much larger revision error compared to
the current official publications. While the revision errors do not increase
substantially, our approach allows for a reduction in the publication lag of
roughly 30 days, when considering the monthly indicator. Turning to the
results related to quarterly GDP, we find that our nowcasts would produce
fairly accurate estimates of GDP growth during the third months of the
reference quarter, even though there are few large errors. On the other hand,
the nowcasts computed at t + 16 are accurate and do not show large revisions,
or at least revisions that are compatible with the ones of Statistics Finland.
Even though these estimates would be produced after the end of the quarter,
they would still allow for more than a month reduction of the publication lag.
Finally, it is important to underline the satisfactory performance of traffic
measurements data. The potential of this source of information should be
explored further, given its real-time availability.
289 | I S I W S C 2 0 1 9