DIRECTORATE GENERAL ECONOMICS, RESEARCH AND STATISTICS Forecasting Tourist Inflows Through Google use Concha Artola Economic Analysis and Forecasting General.

Slides:



Advertisements
Similar presentations
New approaches for data collection and analyses Per Nymand-Andersen European Central Bank, Directorate General Statistics CCSA session on International.
Advertisements

Measuring the US Economy Economic Indicators. Understanding the Lingo Annualized Rates Example: GDP Q3 (Final) = $11,814.9B (5.5%) Q2: GDP = $2,
1 Performance of the Israeli Economy and Bank of Israel Policy Challenges Bank of Israel Annual Report 2010 March 30, 2011.
Chapter 5 Research Design.
A Closer View at Overall Economic Activity FED CHALLENGE THE CURRENT STATE OF THE U.S. ECONOMY Housing Confidence Store Sales Manufacturing GDP Fed Balance.
ECONOMIC STATISTICS AND NATIONAL ACCOUNT IN ETHIOPIA By Sehin Merawi Central Statistical Agency of Ethiopia.
The transmission mechanism of monetary policy Banco Central do Brasil conference: “One year of inflation targeting” 10th July 2000 Alec Chrystal Bank of.
Just Google It: Can Internet Search Terms Help Explain Movements in Retail Sales? Daniel Ayoubkhani (ONS) & Matthew Swannell (ONS)
What We Know of the Youth Market
1. 2 What is EXCELTUR : ¶A non profit association of the 25 biggest & leading Spanish tourist companies ¶Includes airlines, railways, cruise companies,
United Nations Statistics Division Scope and Role of Quarterly National Accounts Training Workshop on the Compilation of Quarterly National Accounts for.
20 Indicators Everyone Should Know C2ER’s 52 nd Annual Conference Oklahoma City June 7, 2012 Jenny Hsu Manager of Economic Research 20 Indicators Everyone.
Japan Industrial Production Published by: Ministry of Economy, Trade and Industry (METI), Japan Frequency: Monthly Period Covered: Prior month Volatility:
Chapter 12 - Forecasting Forecasting is important in the business decision-making process in which a current choice or decision has future implications:
Chapter 5: Monetary Theory and Policy. 1-2 Chapter 5: Monetary Theory and Policy Chapter Outline: Monetary Theory. Economic Indicators Monitored by the.
Economic Indicators. Concepts  Variables that provide information about the state of the economy.  Every economic indicator has a story to tell.  Need.
Session 1 : Ireland’s Society in Economic Downturn Kieran Walsh Central Statistics Office.
Session 2 : The Downturn & Irish Business Richard McMahon Central Statistics Office.
Future Financial Forecast An update on the big economic picture for credit, interest rates and inflation. Paul Cahill Chief Executive Officer.
The Problem of Unemployment in Europe and America A Lecture at the American University of Paris September 23, 2004.
Session 5: Demand forecasting
OECD Short-Term Economic Statistics Working PartyJune Analysis of revisions for short-term economic statistics Richard McKenzie OECD OECD Short.
Slides 13b: Time-Series Models; Measuring Forecast Error
1 © 2009 The Conference Board, Inc. Job information is entered here! Trusted Insights for Business Worldwide
Measuring and Forecasting Demand
Sales Management 8 Estimating Demand. Time Sales 0 Market Potential Industry Forecast Company Potential Company Forecast (Industry Forecast ≤ Market Potential)
Sales Forecasting Introduction Sales forecasting is a difficult area of management. Most managers believe they are good at forecasting. However, forecasts.
Game Plan Real Estate in the News Key Economic Numbers Sales Volume & Inventory Sales Prices Forecast.
Communication and dissemination of indicators Soong Sup Lee, World Bank.
WEBDATANET CONFERENCE May 26-28, 2015 Salamanca Forecasting unemployment with internet search data: does it help to improve predictions when job destruction.
European Commission Directorate General Economic and Financial Affairs Using BCS data for tracking q-o-q GDP growth Andreas Reuter Business and consumer.
Session 5: Flash Estimates of Gross Domestic Product Introduction.
Prague, 10 June The financial / economic crisis and the hospitality industry by Ákos Niklai Member of the HOTREC Executive Committee.
1 DOMESTIC ECONOMIC CONDITIONS Jeff Fuhrer Director of Research Federal Reserve Bank of Boston Equipment Leasing and Finance Association Credit and Collections.
DETERMINANTS OF INFLATION IN ROMANIA Student: COVRIG NICOLAE Supervisor: Prof. MOISĂ ALTĂR.
OECD World Forum “Statistics, Knowledge and Policy,” Palermo, November The Use and Abuse of Real-Time and Anecdotal Information in Monetary.
Copyright 2010, The World Bank Group. All Rights Reserved. Business tendency surveys, part 1 1 Business statistics and registers.
Macroeconomic Data in Real Time Tara M. Sinclair George Washington University Weidenbaum Center Media Retreat Wianno Club in Cape Cod June 28, 2011.
Inflation Report November Output and supply.
Managerial Economics Demand Estimation & Forecasting.
United Nations Economic Commission for Europe Statistical Division UNECE Workshop on Consumer Price Indices Istanbul, Turkey,10-13 October 2011 Session.
7th International Forum on Tourism Statistics Stockholm, Sweden, 9-11 June 2004 NEW RESEARCH LINES ON TOURISM INFORMATION SYSTEMS: POTENTIAL DEMAND ANALYSES.
Flash Estimates of Gross Domestic Product International Seminar on Early Warning and Business Cycle Indicators 15 December 2009.
1 Revisions analysis of OECD composite leading indicators (CLI) Emmanuelle Guidetti Third Joint European Commission OECD Workshop on Business and Consumer.
1 The Decomposition of a House Price index into Land and Structures Components: A Hedonic Regression Approach by W. Erwin Diewert, Jan de Haan and Rens.
Today’s Economics Class : ) (17 April, 2012) 1.Follow-up on Application Activity: “Exploring the Stock Market” and your economic analysis. 2.Introducing.
Answers to Review Questions  1.What are the ultimate targets of monetary policy?  The ultimate targets of monetary policy include stable prices, sustainable.
Sales Management 8 Estimating Demand. Time Sales 0 Market Potential Industry Forecast Company Potential Company Forecast (Industry Forecast ≤ Market Potential)
FISCAL CLIFF & ECONOMIC INDICATOR By: Claire Murray.
Short Term Indicators on African Countries. Regional data needs arising from the Financial crisis Measure the impact of the financial crisis on African.
Business Events Sector UK & Europe. Contents Potential: Business Events 2.European Strategy  European Business Events  Target audiences  Achieving.
Automotive in 2011 Detroit to the Dealer Wards Dealer Magazine Steve Finlay ESA & Company Adam Armbruster.
1 Unit 3 Measurement and Fiscal Policy. 2 Table of Contents.
United Nations Economic Commission for Europe Statistical Division Production and dissemination of short-term economic statistics: the need for long timeseries,
Inflation Report August Demand Chart 2.1 Nominal GDP.
Simon Compton Methodology Directorate Office for National Statistics
Noncompetitive division charts and policy questions The following pages provide a range of indicators (listed in alphabetical order) that you can use to.
Sales Forecasting Sunday 17th, 2016.
Norges Bank 1 Executive Board meeting 27 September 2006.
Creating a Forecast Charles Steindel January 21, 2010 All views expressed are those of the author only and not necessarily those of the Federal Reserve.
Travel and Tourism in Spain to 2020 The report focuses on global major leading industry players with information such as company profiles, product picture.
THE FACTS & FIGURES Government and Industry Committee, London – 20 October 2015 UK PRINTING.
Economics Indicators INFLATION & THE CONSUMER PRICE INDEX (CPI)
New approaches for data collection and analyses
Martine Durand, Director, OECD Statistics Directorate
Carsten Boldsen Hansen Economic Statistics Section, UNECE
Macroeconomic Strategy Primer
The Monetary-Financial Environment
Brexit and the UK Economy
Revision Analysis for General Economic Statistics
Presentation transcript:

DIRECTORATE GENERAL ECONOMICS, RESEARCH AND STATISTICS Forecasting Tourist Inflows Through Google use Concha Artola Economic Analysis and Forecasting General Online Research (GOR 15) Cologne, March 2015

Problems with official statistics: Compilation delay of several weeks Subsequent revisions Sample size may be small Not available at all geographic levels 2. Motivation of this study Central Banks (MP) and Policy Makers in search of timely economic indicators in order to assess the state of economy and implement policies according In order to do so they typically rely on official statistics on employment, retail sales, house sales, etc. As well as soft data as Business or consumer confidence surveys Can Google Trends data help predict current economic activity?  Before release of preliminary statistics  Before release of final revision Google Search Indexes: Real time data No revisions Large samples Available by region or city

OUTLINE 1.Economic indicators based on web searches. 1.How does Google Trends work? 2.In the toolkit of Central Banks 2.Forecasting tourist inflows to Spain using Google searches 1.Results 2.Warnings 3.Other applications 4.Conclusions 3

OUTLINE 1.Economic indicators based on web searches. 1.How does Google Trends work? 2.In the toolkit of Central Banks 2.Forecasting tourist inflows to Spain using Google searches 1.Results 2.Warnings 3.Other applications 4.Conclusions 4

Economic indicators based on internet activity How does Google Trends work? 5 Google Insights for Search

Economic indicators based on internet activity How does Google trends work? 6 Google Insights for Search

Economic indicators based on internet activity How does Google Trends work? 7 Select a search term. (e.g. SPAIN) We control for the geographical area where searches are generated the time period (starting at 2004) and the category (here all categories). Results and meaning:  The chart represents relative interest in search terms including the word SPAIN.  Strong Seasonality  The number in the bar (50%) DOESN´t Mean that 50% of overall searches include the term SPAIN: For a sensible reading one must know that Google indexes are both scaled and normalized. This means: Data are normalized meaning we don´t know the absolute figures for search volume, only the relative frequency. In top of that, data are scaled in a range [0,100], 0 meaning non enough searches, 100 is the top interest Besides these very salient facts, the seasonality of the time series is noteworthy. This is better seen in the second chart where only searches about Spain lying into the TRAVEL category are included Caution should be exercised in some aspects.

Forecasting tourist inflows to Spain using Google searches Warnings Wording is important. Holiday and vacation. Do not use the tool in a pure mechanical way in order to avoid large mistakes. Ex Port of Spain 8

Wording is important 9

10

Wording is important 11

Economic indicators based on internet activity In the toolkit of Central Banks Bank of Israel: launched as soon as 2009 an index based on Google searches which serves as an indicator of demand in the economy, whose results are regularly included in the press release following the Monetary Committee. Bank of England: tracks internet searches in specific fields, namely unemployment trends and housing market developments (QB 2011Q2) New York Fed: applied Google Insights for search to the mortgage markets, finding a sizeable improvement in the outlook for refinancing applications filled when Google searches on mortgage refinancing are included in the model. Bank of Australia: applies Google searches to two specific fields, tracking online shopping and unemployment developments (QB June 2012) Bank of Spain: tracks tourist inflows. DO published in March 2012 and box in QB July Other projects ongoing. 12

OUTLINE 1.Economic indicators based on web searches. 1.How does Google Trends work? 2.In the toolkit of Central Banks 2.Forecasting tourist inflows to Spain using Google searches 1.Results 2.Warnings 3.Other applications 4.Conclusions 13

Forecasting tourist inflows to Spain using Google searches Results Can Google beat the performance of conventional short-term forecasting models? 14  Adjust the best possible forecasting model using the usual statistics, including the lagged endogenous variable (Model 0).  Add the Google Trends index as an additional explanatory variable (Model 1).  Assess the improvement in the predictions. This is typically done through the mean absolute error (MAE) of the out-of-sample predictions using a rolling window forecast. Does Model 1 forecast better than Model 0? Standard procedure as seen in Choi, Varian & others

Forecasting tourist inflows to Spain using Google searches Results The indicator constructed only considers tourists from the UK, Germany and France, the three main clients of the Spanish tourism industry. Overall, these three countries account for 60% of the foreign tourists in Spain, and their tourism pattern is very similar to that of the aggregate of total foreign tourists It looks like good enough to focusing on travellers from these three countries 15

Forecasting tourist inflows to Spain using Google searches Results 16 United Kingdom: 1 to 2 months Correlation (0): 0.51 Correlation (-1): 0.75 Correlation (-2): 0.78 Germany: 2 months Correlation (0): 0.30 Correlation (-2): 0.80 France: 1 to 3 months Correlation (0): 0.26 Correlation (-1): 0.57 Correlation (-2): 0.57 Correlation (-3): 0.56

Forecasting tourist inflows to Spain using Google searches Results 17

Forecasting tourist inflows to Spain using Google searches Results 18 Model 0: MAE = 3.8% Model 1: MAE = 2.2%

Forecasting tourist inflows to Spain using Google searches Results 19 Forecast of Tourist inflows to Spain Origin December 2011Model 0Model 1 Forecast for ,59458,538 Upper i conidence65,15063,652 Lower i. confidence52,03853,423 Magnitude of ic11.2%8.7% -21.9% Observed ,636 error1.66%1.57% -5.9% Forecasts with origin Dec 2011 Model 0 Model 1

CONCLUSIONS The use of Big data for Economic Analysis is still at its infancy 20 “ Your Billion Google Searches Now a Central Bank Tool” (Bloomberg, August 14 th 2012) includes different views on the new tool. Mostly positive, although some more skeptical. San Francisco Fed John Williams referred to what he sees as a “data revolution”: “an enormous amount of information that will better help us understand in very real time what´s going on” Further research is guaranteed, but the field is really promising: In E. Brynjolfsson (MIT) words: When central bankers were looking at traditional data, they were essentially looking out the rear-view mirror”

DG ECONOMICS, STATISTICS AND RESEARCH THANKS FOR YOUR ATTENTION

OUTLINE 1.Economic indicators based on web searches. 1.How does Google Trends work? 2.In the toolkit of Central Banks 2.Forecasting tourist inflows to Spain using Google searches 1.Results 2.Warnings 3.Other applications 4.Conclusions 22

Nowcasting Auto sales Forecasting auto sales using Google searches 1.searching terms: “compra coche” “coche nuevo” … 2.By brand name. 23 Both failed: looks difficult to determine a procedure which adequately capture the intention to buy a car. What are the key words used by potential shoppers? Possible explanation: CONSTANT CONSIDERATION (Google study for US consumers): 63% of final shoppers started their online search looking for a particular brand or model up to six months before the purchase…..but only 20% of shoppers end up buying the car they were initially interested in.

Nowcasting Auto sales Forecasting auto sales using Google searches FAILED 1.searching terms: “compra coche” “coche nuevo” … 2.By brand name. 24 What are the key words used by potential shoppers? CONSTANT CONSIDERATION (Google study for US consumers): 63% of final shoppers started their online search looking for a particular brand or model up to six months before the purchase…..but only 20% of shoppers end up buying the car they were initially interested in.

Nowcasting auto sales using Google searches Simultaneous indicator based on Insurance 25

Forecasting auto sales using Google searches Simultaneous indicator based on Insurance Results 26