1. Banking and Web-talk:
how online conversations boost the retail banking
industry, evidence from Webank
Filippo Corbelli | Bocconi University| 12.17.2011 Share
2. Credits
✤ Prof. Stefano Caselli
Full professor at the Department of Finance at Bocconi University, Milan
✤ Mrs. Vittoria La Porta
PR and Brand Reputation Manager at Webank SpA
✤ Mr. Pietro Di Giacomo
Engagement Manager at Nuvó Consulting Srl
Filippo Corbelli | Bocconi University| 12.17.2011 Share
3. First of all …
Financial crisis Social Media revolution Banking & Social Media
Filippo Corbelli | Bocconi University| 12.17.2011 Share
4. As a matter of fact,
How does online word-of-mouth What factors drive banking
influence your banking decisions? decisions?
Past 2 years Past 3-5 years
Age
5%
65+ 26%
Others
31%
Age
7%
45-64 14%
Online WOM
7%
Age
12%
28%
30-44 Face-to-face
27%
Age
26%
32%
18-29 Past experience
35%
Filippo Corbelli | Bocconi University| 12.17.2011 Share
5. Why this research?
To understand whether social media are relevant to the retail banking industry.
In other words, to investigate...
...to which extent online word-of-mouth
impacts the users’ decision to open an online
bank account?
Filippo Corbelli | Bocconi University| 12.17.2011 Share
6. Let’s hypothesise that:
Online word-of-mouth and online accounts are
correlated
WOM volume has a positive impact on online accounts
WOM valence influences online accounts
WOM variance influences online accounts
WOM venue-specificity influences online accounts
The amount of online accounts opened in prior years is
informative of future conversations
Filippo Corbelli | Bocconi University| 12.17.2011 Share
7. The ingredients are:
✤ DAY From Jan 2011 to Jun 2011
Natural logarithm of the number of AOL
✤ LNaol
opened in day t
Number of conversations about Webank
✤ CONV
in day t
Volatility in the number of AOL opened
✤ VOLaol
between t +1 and t
Dummy variable to identify the days t at
✤ POST
higher posting frequency
Dummy variable to identify the days t at
✤ SUBJ
higher brand sentiment
Dummy variable to identify the venue-
✤ VENUE
specificity per day t
Filippo Corbelli | Bocconi University| 12.17.2011 Share
8. 3D Analysis:
1. LINEAR REGRESSION
Forward stepwise approach
2. GRANGER’S CAUSALITY
Vector Autoregressive (VAR) model
3. PREDICTION-FIT
Static and dynamic forecasting
Filippo Corbelli | Bocconi University| 12.17.2011 Share
9. Linear Regression
R-squared Breusch-Pagan Engle’s Lagrange Durbin Alternative Breusch-Godfrey
41.63% 0.3109 0.06 0.9617 0.9608
partial NO NO NO 1-order NO higher-order
correlation heteroscedasticity ARCH effects serial correlation serial correlation
Stationary variables
IS IT
NO!
SPURIOUS?!
Augmented Dickey-Fuller test
Filippo Corbelli | Bocconi University| 12.17.2011 Share
10. How to interpret the results:
CONV: +0.8% in AOL (same day)
WOM affects AOL =
POST: +21.39% in AOL (6 days after)
SUBJECTIVITY and SUBJ: +12.08% in AOL (3 days after)
VENUE affect AOL =
VENUE: -19.26% in AOL (3 days after)
270 more AOL today predict on
average 1 more AOL tomorrow
PAST AOL
affect FUTURE AOL =
the larger the fluctuations in AOL,
the higher the impact for Webank
Filippo Corbelli | Bocconi University| 12.17.2011 Share
11. Granger’s causality
M1: 1 lag M2: Lütkepohl (2 lags)
UNIDIRECTIONAL BIDIRECTIONAL
Users talk about Webank as they
WOM can predict future revenue have opened an online account
Filippo Corbelli | Bocconi University| 12.17.2011 Share
12. Does the model predictably fit?
Linear Regression Vector Autoregressive Model
Static Forecasting Dynamic Forecasting
Filippo Corbelli | Bocconi University| 12.17.2011 Share
14. Dynamic
Scenario A Scenario B
forecast June 2011 forecast July 2011
Filippo Corbelli | Bocconi University| 12.17.2011 Share
15. So what?
So What?
“I open a bank …to satisfy a self-
…and I talk about Webank
account as I have presentational need
as I have opened a Webank
heard about Webank and gain social
account…!
online…! currency!”!
Monitoring conversations to predict future AOL;
Subjective or non-informative comments should be favoured; and
Services and content should be easy to be shared.
Filippo Corbelli | Bocconi University| 12.17.2011 Share
16. Q&A
Thanks.
“The best way to predict the future is to create it”
P. Drucker
@filippocorbelli
Filippo Corbelli
Filippo Corbelli | Bocconi University| 12.17.2011 Share