ECB-UNRESTRICTED Paloma Lopez-Garcia Filippo di Mauro Chiara Osbat CompNet approach to competitiveness: Results and policy use ECB Frankfurt 27 January 2014
Prologue Rubric Definition: A competitive economy, in essence, is one in which institutional and macroeconomic conditions allow productive firms to thrive. In turn, the development of these firms supports the expansion of employment, investment and trade (M. Draghi, 30 November 2012) We are seeing the beginning of a recovery that is still weak, still fragile, and still uneven.. some countries, such as Greece, had made meaningful progress on structural reforms while other peripheral nations had not, raising the risk of further instability. (M. Draghi, Davos 24 January 2014) CompNet objective: Provide a robust theoretical and empirical link between the drivers of competitiveness and macroeconomic performance for research and policy analysis 2
The holistic approach to competitiveness of CompNet Rubric 3
WS1 Approach General Rubric Develop indicators capturing more complex dimensions over and above the traditional price/cost based indicators. Value added: Indicators are based on product-level statistics (e.g. about 5,000 products) Apply novel methodologies Examples of indicators (for complete list, check http://www.ecb.europa.eu/home/html/researcher_compnet.en.html): Non-price factors (K.Benkovskis, J.Wörz) Barometer of competitive pressure (K.Benkovskis, M.Silgoner, K.Steiner, J.Wörz) Export sophistication index (E.Bobeica) Shift-share analysis (G.Gaulier, D.Taglioni, S.Zignago and M.Dyadkova, G.Momchilov) Contribution to trade balance (C.Osbat, S.Ozyürt, T.Karlsson) Alternative HCIs based on trade in services (M.Schmitz) Trade-weighted national unit labour costs (M.Silgoner) Measures of integration in Global value Chains (S.Christodoulopoulou and I.Rubene) Use the indicators for policy analysis 4
CompNet: Rubric WS2 working approach country teams micro expertise micro data ECB computational engine Data Dissemination Further Modules (Trade, Financial constraints) Common co authored publication 5 Created a strong ESCB Team, which uses a working procedure with huge potential
WS3 Rubric Approach USITC OECD- WTO TiVA WIOD Importance of Mapping EU GVCs in the European Union WS3: hub for I-O databases Consequences of the financial crisis Effect on overall competitiveness assessment external imbalances 6
Outline Rubric CompNet Micro-founded analysis (PL-G) The potential for research Using the novel firm level data base The proposed use for policy analysis (CO) The drilling down approach Interaction country desks at the ECB with CompNet A few examples 7
CompNet Micro-founded analysis Rubric What data do we already have? What have we already learnt? What new data will be collected in spring? Which new lines of research could we open with your input? How could we use this data for policy and research? 8
Research Rubric time-line: Micro-based Database v1 What is available? Start of firm-level data collection Set-up of a new research infrastructure NCB teams Coordination ECB-NCB Ensuring full comparability Firm-level based database v1: Productivity and cost-related indicators Average, sd, skewness and full distribution of productivity, TFP, ULC, size, capital intensity and labour cost of firms (details) Oct. 2012 Dec 2013 Static and dynamic indicators of allocative efficiency Output: Completion of paper documenting the database Co-authored by 9 all teams Data is aggregated at the 2-digit NACE industries for 11 Countries over 1995-2011 (details) Several research papers from small dedicated teams
Initial findings (1): Huge dispersion and skewness Rubric Firms are very heterogeneous, even within the same 2-digit sector Firm performance distribution is very asymmetric: Few high productive firms and a large mass of low productive ones Important findings from a policy point of view: The impact of macro/policy shocks depends on the underlying distribution Average labour productivity of a sector is not representative Labour Productivity Heterogeneity Labour Productivity Levels 0 20 40 60 80 BE CZ DE EE ES FR HU IT SI SK Within-sector dispersion Across-sector dispersion DE BE FR IT ES SI HU EE CZ SK 0 50 100 150 200 Mean Inter Quartile Range Median p10 - p90 10
Initial findings (2): Different dynamics in tails of distribution Rubric Most productive firms (top 10%) are up to 10 times larger than the median productive firm in the same 2-digit sector ULC of firms at the bottom and the top of the productivity distribution (p10 and p90) have reacted very differently to the crisis Average Size per Lab Prod percentile Period: 2003-2007 ULC in Manufacturing sector Spain 2002-2010 0 5 10 15 BE EE ES FR SI P10 P90 -.4 -.2 0.2.4.6.8 1 2002 2004 2006 2008 2010 year P10 P90 11
Initial findings (3): Allocative efficiency differs Rubric In well functioning markets, the most productive firm should be the largest This is an efficient allocation of resources Allocative efficiency approximated by the covariance (at the 2-digit sector level) between relative productivity and size (details) Improving the allocation of resources fosters aggregate productivity OP Gap 0.05.1.15.2.25 Covariance Productivity and Size, average across years BE CZ DE EE ES FR HU IT PL SI SK Tradables Non tradables Striking differences in tradables vs. non-tradables Highly correlated with sectorspecific market regulation Source: CompNet Dataset 12
Initial findings (4): Correlation with aggregate developments Rubric Trade Balance -.1 -.05 0.05.1 EE HU CZ CZ CZ CZ CZ SI CZ HU HU SK SI CZ HU SI SI SK SI HU ES SI SK SI HU EE EE EEEE EE EE Trade balance vs Lagged P90 Productivity all year averages IT IT IT IT ES ES ES ES IT IT ES ES FR FR FR FR BE FR BE BE BE BE BE BE DE DE DE DE DE Close correlation between the p90 level of the productivity distribution and aggregate trade balance Better fit than with average/median productivity.2.4.6.8 1 1.2 Labour Productivity lagedp90 Regressions 13
January Rubric March 2014: Expansion of the database All these dimensions can be combined Firm-level based database v2: Productivity and costrelated indicators + At least 6 new countries Trade Module Financial Module Mark-up Module Export status of firms and Export volumes Financial position of firms/ credit constraints Sector level mark-ups and bargaining power Banking data (Pilot with IT and NL)? Sector level Sector/size dimension Labour Module Flag HGF/ Employment transition matrixes Labour and product Mkt. structure (WDN3: details) Output Several research papers from small dedicated teams Data available for policy analysis 14
Think Rubric big! CompNet 2.0 Possible policy and research questions What is the productivity premium of exporting firms? Has it changed with the crisis? Are productive firms credit constraint in some countries? Are credit constraints hampering exports? Are sector mark-ups and bargaining power related to sector regulation? Do sector with higher markups distribute less efficiently resources? Are firms growing at a different path within the same sector? Is this related to credit constraints? Is this related to regulation? 15
How can we use all this information for policy work Rubric Zoom out again to the competitiveness toolkit idea: Integrating macro, firm-level and GVC information Model-based benchmarking using Bayesian Model Averaging (BMA) Structured in a way that also facilitates free-style user analysis Objective: User-friendly Incorporates CompNet results Flexible: different degrees of depth depending on time and specific skill 16
Competitiveness Rubric Assessment Toolkit I: econometric analysis Comprehensive crosscountry dataset (27 European countries, period 2002-2012) Econometric analysis BMA output Use of results Macro indicators (>100 indicators and growing) Business conditions, labour market, institutional, financial, infrastructure, human capital, detailed trade Micro indicators (Firm-level information from the entire distribution of firms) TFP, ULC, productivity, cost of employees, BMA Estimate millions of models to identify robust link between trade outcomes and indicators Use estimates to identify set of most relevant indicators for trade performance (but researchers use all, as they are implicitly country weighted) desks Use clustering techniques to identify the relevant peer country group to use for benchmarking Cross-country comparison Identification of robust competitiveness drivers Taylor-made policy advice 17 More details on the variables
Competitiveness Rubric Assessment Toolkit II: Drill-down approach Diagnosis What issue do I want to look at? (e.g. market share, productivity growth ) What will be my LHS measure? (market share, long-term growth, ) What are the potential determinants? Weakness related to e.g. Excessive concentration in slow growing markets Possible ways to collaborate to advance understanding: 1. Without specific intervention of CompNet 2. With specific intervention of CompNet User-friendly interface to be used independently by country desks: (examples in following slides utilise the software Tableau ) researchers Firms size distribution by country sector desks in comparison with EU average Correlation firm size/productivity Measures of allocative efficiency Identify policy questions to study with additional data and models available through CompNet research: Regression of firms productivity vs. financial constraints Add data dimensions Review models to encompass new questions 18
Use Rubric of the data: 4 "Drill-down approach" examples Drill-down example workbook 1) Choose a trade outcome 2) Compare it with relative price growth Latvia Latvia Lithuania Lithuania Malta Malta Slovak Republic Slovak Re.. Greece Greece UK UK Poland Poland Netherlands Netherlands Portugal Portugal Czech rep Czech rep Estonia Estonia Spain Spain Germany Germany Italy Italy France France Bulgaria Bulgaria Denmark Denmark Cyprus Cyprus Hungary Hungary Belgium Belgium Austria Austria Ireland Ireland Slovenia Slovenia Sweden Sweden Finland Finland Croatia Croatia R i 3) Revealed comparative advantage 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Select trade outco.. -0.2238 0.3011 Select trade outco.. Export market share.. Export volume (annu.. Trade balance (% G.. Choose years: 0.2 All Countries in the ful.. 0.1 All In / Out of Outliers.. 0.0 In Out -0.1 Country Austria Belgium -0.2 Bulgaria Croatia Cyprus -0.5-0.4-0.3-0.2-0.1 0.0 0.1 0.2 0.3 0.4-0.45-0.40-0.35-0.30-0.25-0.20-0.15-0.10-0.05 0.00 0.05 0.10 0.15 0.20 Czech rep REER-ULCT [annual growth rate] Relative export prices adjusted for quality [annual growth rate] Denmark Estonia Country regressions: Export market share (annual growth) Finland France Germany Greece Hungary 0.05 Ireland Italy Latvia Lithuania 0.00 Luxemburg Malta Netherlands Poland -0.05 Portugal Romania Slovak Republic Slovenia -0.10 Spain Sweden UK Country -0.15-0.10-0.05 0.00 0.05 0.10 0.15 0.20-0.025-0.020-0.015-0.010-0.005 0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 REER ULCT [annual growth rate] Relative export prices adjusted for quality [annual growth rate] France 4) Use firm-level indicators 1.2 1.2 Multiple Values 2010 1.0 RCA reference 1.0 Measure Names Avg. RCA Exports (.. POLAND 100 Aggregate sector: All 0.8 0.8 Avg. RCA imports (.. BELGIUM 90 Aggregate sector Accommodation Admnistrative and s.. 0.6 0.6 GERMANY 80 Construction Electricity 0.4 0.2 0.4 0.2 SPAIN 70 Information and co.. Manufacturing Mining and quarring Professional, Scien.. 0.0 0.0 2005 2010 2005 2010 2005 2010 2005 2010 2005 2010 2005 2010 2005 2010 2005 2010 2005 2010 2005 2010 Revealed comparative advantage by technology: Medium-high and High-tech (click any line to see tech definition by OECD) Country Finland France Germany Greece Hungary Ireland Italy Portugal Slovak Republic Spain 5 5 SLOVENIA ESTONIA SLOVAKIA 60 50 40 Water and remediat Choose firm-level i.. Labour productivity,.. Labour productivity, l.. Labour productivity, t.. Labour cost, median Labour cost, lowest.. Labour cost, top 90% ULC, median ULC, lowest 10% ULC, top 90% 4 4 HUNGARY 30 Compare with: Labour productivity, median 3 3 20 country BELGIUM 2 2 CZ GERMANY FRANCE 10 POLAND 1 RCA reference 1 ROMANIA ITALY 19
1) Rubric Choose a trade outcome Export market share
2) Rubric Trade outcomes and relative prices adjusting for quality matters
Rubric 3) Revealed comparative advantage
4) Rubric Firm-level labour cost and labour productivity: distributions in levels
Rubric and cumulative growth
The end Rubric THANKS FOR YOUR FEEDBACK Please consult our website http://www.ecb.europa.eu/home/html/researcher_compnet.en.html
Rubric