bbva-google economic indicators · pdf filebbva-google economic indicators . ... people living...
TRANSCRIPT
BBVA-Google economic indicators
Madrid, 16 January 2012
BBVA-Google economic indicators
Page 2
Introduction
• Objective: to present BBVA-Google economic indicators, on this occasion relating to the Spanish Tourism sector
• These indicators improve the accuracy and speed at which tourist inflows and overnight stays are forecasted
• Searches allow new and extremely useful indicators to be obtained to estimate tourism trends in Spain
• High added value in social and economic terms, given their contribution to corporate decision-making
• Upholding excellence and technology in the pursuit of knowledge
BBVA-Google economic indicators
Page 3
Motive behind the project
We know past data in retrospect (backcasting). Examples: GDP, LFS, … human capital
We intuit the present (nowcasting)
We do not know future (forecasting)
Information is key to taking the right economic decisions
BBVA-Google economic indicators
Page 4
The usual approach to the problem
Univariate ARIMA models (Box and Jenkins, 1976)
Multivariate VARMA models
Long tradition since the 1970s, used widely nowadays
Extrapolating past trend observed to the future
BBVA-Google economic indicators
Page 5
The background in BBVA Research
Dynamic Factor Models: learning process (KF) that updates parameters
Estimates real time GDP growth (nowcasting) and evaluate indicators
Very good results in Spain, EMU, Mexico, USA …
In 2009 we embarked on a large-scale project to apply new methods
Spain: observed GDP growth and forecasts based on the MICA-BBVA model (%qoq)Source: BBVA ResearchCurrent forecast: 13 January 2012
-1,5
-1,0
-0,5
0,0
0,5
1,0
1,5
2Q0
9
3Q0
9
4Q
09
1Q10
2Q10
3Q10
4Q
10
1Q11
2Q11
3Q11
4Q
11
-1,5
-1,0
-0,5
0,0
0,5
1,0
1,5
CI at 20% CI at 40% CI at 60%
Latest estimate Observed
BBVA-Google economic indicators
Page 6
The challenge of improving forecasts
Through indications of the future observed inthe present
Internet: a powerful tool that can be used to find information before taking decisions
Google is the leading Internet search engine
How can forecasts be improved?
Information
WorkConsumption
Investment
Economicactivity
Present Future
BBVA-Google economic indicators
Page 7
The background in Google
AR models with searches to predict the present beat those that do not include them
Examples: retail, vehicle and house sales, and tourism
Can searches help to forecast the future?
Choi and Varian (2009): Predicting the Present with Google Trends
BBVA-Google economic indicators
Page 8
The project
Filtered Google searches+ =Dynamic factor
models Indicators
This collaboration project was developed for application in various countries and sectors considered to be of strategic importance for BBVA and Google
BBVA-Google economic indicators
Page 9
Pilot project: Tourism in Spain
Accounting for more than 10% of GDP and employment,
and 27% of exports,
12% of revenue in the external Balance of Payments
Preliminary data make us optimistic about actual results
Tourism is a strategic sectorPercentage accounted for by tourism
2008 2009 2010
GDP 10.4 9.9 10.2
Exports 26.5 23.9 27.0
Balance of paymentrevenues
11.6 12.2 12.2
Employment 11.0
Source: INE, Cuenta Satélite del Turismo, and BdE
BBVA-Google economic indicators
Page 10
Variables forecastedForeign tourist inflows Overnight stays
People living abroad making one or more overnight stays
Every night spent by a Spanish or foreign visitor in a hotel establishment
Both published on around the 22nd or 23rd of the following month
Foreign tourist inflows Source: INE
Total overnight stays in hotels Source: INE
0
1,000,000
2,000,000
3,000,000
4,000,000
5,000,000
6,000,000
Jan
-08
May
-08
Sep
-08
Jan
-09
May
-09
Sep
-09
Jan
-10
May
-10
Sep
-10
Jan
-11
May
-11
Sep
-11
Foreign visitorsA
pr-
08
Oct
-08
Jan
-09
Ap
r-0
9
Oct
-09
0
5,000,000
10,000,000
15,000,000
20,000,000
25,000,000
30,000,000
35,000,000
40,000,000
45,000,000
Jan
-08
Jul-0
8
Jul-0
9
Jan
-10
Ap
r-10
Jul-1
0
Oct
-10
Jan
-11
Ap
r-11
Jul-1
1
Oct
-11
Total overnight stays
BBVA-Google economic indicators
Page 11
Google searches
Weekly data since 2006 in English, German, French and Italian (+ Spanish for overnight stays)
A similar seasonal pattern to tourist inflows but one month ahead
The correlation is significant and extremely high (0.6) even factoring in yoy growth rates
Foreign tourist inflows and Google searchesSource: INE and Google (Base=100, January 2006, RHS)
0
1000000
2000000
3000000
4000000
5000000
6000000
Jan
-06
Jul-0
6
0
100
200
300
400
500
600
700
800
900
Tourist inflows (lhs) Searches (rhs)
Jan
-07
Jul-0
7
Jan
-08
Jul-0
8
Jan
-09
Jul-0
9
Jan
-10
Jul-1
0
Jan
-11
Jul-1
1
BBVA-Google economic indicators
Page 12
Results: a common factor in these variables
The model allows the non-observable factor tobe forecasted …
clarifying the trend of tourist inflows and searches ….
using a recursive process where the model itself learns from its mistakes
Tourist inflows and BBVA-Google indicator searchesSource: INE, Google and BBVA Research
-0.25
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
Jan
-07
Jul-0
7
-8
-6
-4
-2
0
2
4
6
8
Tourist inflows (lhs) Common factor (rhs)
Jan
-08
Jul-0
8
Jan
-09
Jul-0
9
Jan
-10
Jul-1
0
Jan
-11
Jul-1
1
Jan
-12
BBVA-Google economic indicators
Page 13
Results: improved forecasting capacity
Searches contain useful information that helps forecast the present and the future
Forecasting capacity improves by between 18% and 20% when searches are included
Improving between 24% and 31% when searches are included and dynamic factor models is
specified
Forecasting errors
Foreign tourist inflows Back-casting
Now-casting
Fore-casting
AR model 1.00 1.00 1.00
AR model incl. searches 0.98 0.90 0.82
FD model incl. searches 0.94 0.81 0.76
Mean squared error (AR=1.0, January-December 2011)
BBVA-Google economic indicators
Page 14
Results: forecasts at February 2012
Tourist inflows will reach 1.92 million, 4.4% more than in February 2011
39.8 million visitors are forecasted from February 2011, 10.7% more than in the previous
year
Overnight stays are expected to reach 14.3 million, 0.5% more than in February 2011
Tourist inflows and BBVA-Google indicator forecastsCurrent forecast: 16 January 2012Source: INE, Google and BBVA Research
-12-10-8-6-4-202468
1012141618
20
Jan
-10
Mar
-10
May
-10
Jul-1
0
Sep
-10
No
v-10
Jan
-11
Mar
-11
May
-11
Jul-1
1
Sep
-11
No
v-11
Jan
-12
CI at 20% CI at 40%CI at 60% Latest estimateObserved Cumulative annual figure yoy, %
BBVA-Google economic indicators
Page 15
Results: publications
Flash notes containing monthly forecasts Scientific publications describing technical details of the methodology used and results obtained
All forecasts will have free access to the public, for the benefit of society as a whole, sharing and disseminating the knowledge acquired through this collaboration
BBVA-Google economic indicators
Page 16
Working team contactsRafael Doménech
Chief Economist for Developed [email protected]
Miguel CardosoChief Economist [email protected]
José RuizPrincipal [email protected]
Juan Ramón GarcíaSenior [email protected]
Camilo Andrés UlloaEconomist [email protected]
SpainJuan Ruiz Chief Economist for Economic Scenarios
Jaime Martínez Senior [email protected]
Economic Scenarios Máximo Camacho Professor at the Universidad de Murcia
External partners
Olga San JacintoHead of
Google Spain and Portugal [email protected]
Javier LópezBusiness Analyst [email protected]
Álvaro FerrerTravel Analyst [email protected]
Google Spain
BBVA-Google economic indicators
Madrid, 16 January 2012