What Unbiased Estimates for Time- Invariant Variables Reveal about the Effect of Distance on Foreign Direct Investment NOITS Stockholm University May 2007.

Slides:



Advertisements
Similar presentations
Introduction Describe what panel data is and the reasons for using it in this format Assess the importance of fixed and random effects Examine the Hausman.
Advertisements

An Application of Time Series Panel Econometrics to Gravity Models By David R. Thibodeau Stephen Clark and Pascal L. Ghazalian.
Lecture 29 Summary of previous lecture LPM LOGIT PROBIT ORDINAL LOGIT AND PROBIT TOBIT MULTINOMIAL LOGIT AN PROBIT DURATION.
Random Assignment Experiments
Economics 20 - Prof. Anderson1 Multiple Regression Analysis y =  0 +  1 x 1 +  2 x  k x k + u 7. Specification and Data Problems.
PSME M1 Economic Growth Tutorial.  Introduction ◦ Review of Classic Solow Model ◦ Shortfalls of Solow ◦ Human Capital Accumulation ◦ Convergence Theory.
THE CROSS-COUNTRY DETERMINANTS OF POTENTIAL AND ACTUAL MIGRATION Frederic Docquier, Giovanni Peri and Ilse Ruyssen International Migration Scholarship.
Technology Diffusion and Income Inequality: how augmented Kuznets hypothesis could explain ICT diffusion? 30 May 2005 Miguel Torres Preto.
Turun kauppakorkeakoulu  Turku School of Economics REGIONAL DIFFERENCES IN HOUSING PRICE DYNAMICS: PANEL DATA EVIDENCE European Real Estate Society 19th.
1 Equity Price and Equity Flows: Testing Theory of the Information-Efficiency Tradeoff Assaf Razin Anuk Serechetapongse Cornell University June 14 th,
Exports x FDI in Heterogenous Firms
Openness, Economic Growth, and Human Development: Evidence from South Asian countries from Middlesex University Department of Economics and.
1Prof. Dr. Rainer Stachuletz Multiple Regression Analysis y =  0 +  1 x 1 +  2 x  k x k + u 7. Specification and Data Problems.
Simple Linear Regression
4. Multiple Regression Analysis: Estimation -Most econometric regressions are motivated by a question -ie: Do Canadian Heritage commercials have a positive.
The Simple Regression Model
Financial Development and International Agricultural Trade D. Susanto and C. P. Rosson Department of Agricultural Economics, Texas A&M University College.
Objectives Understanding the impact of distance and economy size on trade using the gravity model Apply the gravity model for the cases of FDI, and migration.
Controlling for Transactions Bias in Regional House Price Indices Gwilym Pryce & Philip Mason (Conference in Honour of Pat Hendershott, Ohio, July 2006)
Explaining the performance of firms and countries: what role does the business environment play? Simon Commander Katrin Tinn Dubrovnik 26 June, 2008.
Sandy Lai Hong Kong University 1 Asset Allocation and Monetary Policy: Evidence from the Eurozone Harald Hau University.
Descriptive Methods in Regression and Correlation
Correlation and Regression
Hypothesis Testing in Linear Regression Analysis
THE MIGRATION-TRADE LINK: IMMIGRANTS OR EMIGRANTS? Francisco Requena Guadalupe Serrano (Universitat de Valencia)
5.1 Basic Estimation Techniques  The relationships we theoretically develop in the text can be estimated statistically using regression analysis,  Regression.
Investment Analysis and Portfolio management Lecture: 24 Course Code: MBF702.
Herding in the UK Real Estate Market Stephen L. Lee Cass Business School, City University London, 106 Bunhill Row, London.
Andrew k. Rose and T.D. Stanley Presented by: María del Carmen Ramos Herrera.
Think of the North Atlantic country Iceland with FDI in recent years resembling the pattern before the economic crash, so the dataset pre-crash can be.
Determinants of Credit Default Swap Spread: Evidence from the Japanese Credit Derivative Market.
Mergers, Acquisitions and Export Competitiveness: Experience of Indian Manufacturing Sector Researcher: Mishra Pulak, Jaiswal Neha Publishing Year: 2012.
Discussion of: M&A Operations and Performance in Banking by Beccalli and Frantz Emilia Bonaccorsi di Patti Bank of Italy Structural Economic Analysis Dept.
HAOMING LIU JINLI ZENG KENAN ERTUNC GENETIC ABILITY AND INTERGENERATIONAL EARNINGS MOBILITY 1.
Collateral Damage: Exchange Restrictions and Trade Flows Shang-Jin Wei Zhiwei Zhang.
The changing geography of banking – Ancona, Sept. 23 rd 2006 Discussion of: “Cross border M&As in the financial sector: is banking different from insurance?”
Does Trade Cause Growth? JEFFREY A. FRANKEL AND DAVID ROMER*
Chapter 16 Data Analysis: Testing for Associations.
Chapter 10 For Explaining Psychological Statistics, 4th ed. by B. Cohen 1 A perfect correlation implies the ability to predict one score from another perfectly.
MODELS FOR PANEL DATA. PANEL DATA REGRESSION Double subscript on variables (observations) i… households, individuals, firms, countries t… period (time-series.
ANOVA, Regression and Multiple Regression March
Aid, policies and Growth
Endogenous deregulation: evidence from OECD countries Duo and Roller, Economics Letters, 2003,
Ice Breaking FDI Ministry of Commerce December 20th 2007 Dr. Helga Kristjánsdóttir.
1 How Did Employee Ownership Firms Weather The Last Two Recessions? Employee Ownership and Employment Stability in the US : by Fidan Ana Kurtulus.
Substitutability Between FDI and Aid NOITS 9th Annual Workshop Substitutability Between FDI and Aid NOITS 9th Annual Workshop May 13 th 2006 Reykjavík.
Home bias and international risk sharing: Twin puzzles separated at birth Bent E. Sørensen, Yi-Tsung Wu, Oved Yosha, Yu Zhu Presneted by Marek Hauzr, Jan.
Employment, skill structure and international trade: firm- level evidence for France Pierre Biscourp – Francis Kramarz (2007)
Pantelis Pantelidis, University of Piraeus Dimitrios Kyrkilis, University of Macedonia Efthymios Nikolopoulos, University of Macedonia February 2011 The.
Fixed Effects Model (FEM)
Chapter 15 Panel Data Models.
Þjóðarspegillinn 2006 Sjöunda ráðstefna um rannsóknir í félagsvísindum
Vertical Specialization in Multinational Firms
More on Specification and Data Issues
For the World Economy Availability of business services and outward investment: Evidence from French firms Holger Görg Kiel Institute for the World Economy,
More on Specification and Data Issues
An Investigation into the Impact of the Medium-Term Expenditure Framework on Budgetary Outcomes Jim Brumby World Bank.
Sven Blank (University of Tübingen)
Helga Kristjánsdóttir, 15. maí 2007
T test.
Deild Titill rannsóknar Höfundur
By OMER GOKCEKUS and YUI SUZUKI
Flow of Foreign Direct Investment into Iceland
19. október, 2005 Helga Kristjánsdóttir
PhD defense June 16th 2004 Helga Kristjánsdóttir
Understanding the determinants of managerial ownership and the link between ownership and performance CharlesP.Himmelberga R.GlennHubbardab DariusPaliaac.
The Determinants of FDI Inflows to Greece
The Simple Regression Model
Foreign Direct Investment, Culture and Fixed Effects
Presentation transcript:

What Unbiased Estimates for Time- Invariant Variables Reveal about the Effect of Distance on Foreign Direct Investment NOITS Stockholm University May 2007 Ronald Davies, University of Oregon Delia Ionascu, Copenhagen Business School Helga Kristjánsdóttir, University of Iceland

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Introduction This paper applies the XTFEVD procedure on panel data. Three procedures, to three FDI datasets, sometime using difference estimates for time invariant variables.

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Introduction This allows us to estimate the impact of time-invariant variables on FDI while including fixed effects. We find that the omission of fixed effects, when using OLS, significantly biases several of these variables, for example when proxying for culture.

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Introduction We also find that by eliminating estimation biases, the differences across datasets largely disappear. Thus, controversies in the literature that are driven by differences in data sets may be resolved by using this estimation technique.

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Introduction The primary goal of this paper is to apply this panel fixed effects with vector decomposition (XTFEVD) method –fevd stands for ”fixed effects vector decomposition” The idea has been around for some time but not been applied to FDI Applied to three commonly used datasets

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Literature Plümper and Troeger (forthcoming) Developed a three-stage estimation procedure –They find it to efficiently estimate the impact of time-invariant variables - while also controlling for fixed effects –1. Provides estimates for time-variant variables –2. Takes out the unexplainable part –3. Runs again, using the unexplainable part

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Literature Hofstede (2001) 5 cultural distances –Measures do not vary over time –1. Power Distance, high=>less eq society –2. Individualism, high=>indiv appreciated –3. Masculinity, high =>mas appreciated –4. Uncertainty tolerance, high=>avoid uncer –5. Long-term orientation,high=>willing to wait

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Model Setup The basic equation estimated can be described as:

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Data We use FDI data BEA, Bureau of Economic Analysis –Bilateral US inbound and outbound, combined, stocks, affiliate sales OECD –Bilateral inbound FDI stocks –Parent or host country belongs to OECD World Bank, WDI –Total inbound and total outbound FDI flows

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Estimation Results Tables 1 & 2 US stocks, sales OECD –OLS and XTFEVD –Time invariant Dis - colony - language - area

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Table 1. US ResultsTable 2. OECD StockSalesStock OLSXTFEVDOLSXTFEVDOLSXTFEVD Parent GDP2.444***1.624***3.021***3.143***2.432***2.242*** Parent Pop-0.767***-6.803***-1.103***-4.010***-1.083***-3.567*** Parent TC-8.389*** *** *** ***-6.862***8.909*** Host GDP1.536***1.684***2.030***2.146***0.709***1.674*** Host Pop-0.652***-1.871***-1.007***-2.785*** *** Host TC-5.309*** ***-7.963*** ***2.182 Host Inv Cost **0.766* ***-1.109*** Estimation Results, Tables 1 & 2 Time variant variables OLS estimates for parent and host population biased towards zero

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Estimation Results, Tables 1 & 2 Time invariant variables sometimes change sign, implying biased OLS estimates More significant results More consistent estimates Table 1. US StockTable 1. US SalesTable 2. OECD Stock OLSXTFEVDOLSXTFEVDOLSXTFEVD Parent Area-0.318***2.378***-0.105*1.014***-0.039*1.114*** Host Area *** ***0.331***1.437*** Distance10.332*** ***3.893** ***-2.573***-5.229*** Distance Sq ***1.268***-0.241**0.927***0.115***0.258*** Colony-1.070***10.599***-1.063***5.230***0.640***3.870*** Common Lang0.824***-0.878***1.236***0.391***1.273***-0.750*** Island0.653***-1.952*** ***-0.250**-1.307*** Landlock *** ***-0.334***-1.420***

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Estimation Results, Table 3 World Bank data Not bilateral data Not info on parent countries for inbound data Not info on host countries for outbound data

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration InboundOutbound OLSXTFEVDOLSXTFEVD GDP0.977***1.357***2.254***2.467*** Pop-0.193*-3.703***-0.933***-5.412*** TC *** *** *** School-0.768*** ** PD0.240*3.723***0.405***4.449*** Estimation Results, Table 3 Estimation Results, World Bank –Omission of fixed effects leads to biased estimates for trade cost and culture etc.

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration InboundOutbound OLSXTFEVDOLSXTFEVD Independence0.819***-0.733*** *** Masculinity ***-0.174*4.304*** Uncertainty ***-0.759***-5.012*** Time ***0.341**2.189*** Area0.431***1.161***0.143***0.952*** Island ***0.408***-4.234*** Landlock *** *** Estimation Results, T3 cont.

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Estimation Results, Table 4 Differences presented as parent-host –If negative, cannot take log of differences –Carr, Markusen & Maskus (2001) absolute value restricts slope of difference variable to be the same on each side of zero –Therefore, estimate two separate variables, positive differences and negative differences –Find whether the deviation from zero that

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Estimation Results, Table 4 Table 4 presents results not using differences Find to be driven by magnitude, rather than differences to be important predictor of FDI

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Estimation Results Hofstede cultural index: We find that FDI comes from and goes to countries with high –Masculine values –Long term goals –Power Distance FDI outflows from countries with high –Independence (but less into these)

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Estimation Results The XTFEVD procedure is suitable here –When apply conventional fixed effects, then cannot estimate time invariant variables –GDP POP do change if XTFEVD, but not much change if conventional fixed effects –We find culture to be a significant predictor of FDI, and that the cultural attributes of the parent and the host separately that matter, rather than the differences between them

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Overall Estimation Results Time invariant variables: Hofstede index, common language, landlocked, distance When use XTFEVD, then less biased estimates between datasets OECD, World Bank and US data Robust results between datasets if XTFEVD

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Overall Estimation Results We need to be careful in applying the conventional fixed effects estimator to the time invariant variables The OLS estimator provides us with biased results for these When use OLS, then experience unobserved country heterogeneity

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Overall Estimation Results The XTFEVD procedure is suitable here –There have been some different algorithms –But XTFEVD has turned out to be efficient and good for small samples –XTFEVD appears to provide consistent estimates across different datasets

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Conclusions The goal of this paper has been to employ the panel fixed effects with vector decomposition method developed by Plümper and Troeger (forthcoming) Examine the extent to which the omission of fixed effects gives misleading coefficients in FDI regressions. We find that the problems created by omission of fixed effects to be quite significant But that they vary across the three most commonly used FDI datasets. As a result, when including fixed effects, we find much more consistency in the estimates of time- invariant variables across data sets.

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration Conclusions In addition to our main results, we find that the cultural variables constructed by Hofstede are significant determinants of FDI. Since XTFEVD yields far more comparable results across our datasets, using fixed effects may be very useful in removing the problems this variation causes when industry level data are not available. May resolve controversies in empirical literature

UNIVERSITY OF ICELAND Faculty of Economics and Business Administration