Global Market Power Jan de Loecker (KU Leuven) and Jan Eeckhout (UCL, UPF, GSE) Working Paper, 2018

Transcription

1 Global Market Power Jan de Loecker (KU Leuven) and Jan Eeckhout (UCL, UPF, GSE) Working Paper, 2018 Presented by Sergio Feijoo UC3M Macro Reading Group December 18, 2018

2 Motivation Market power leads to redistribution of resources from consumers to owners of firms.... has welfare effects: Higher prices ùñ Suboptimal number of consumers priced out of the market (Harberger, 1954). Stifles innovation and investment (Aguion et al., 2005).... measurement is crucial for policy design: antitrust, taxation,... Integrated and globalized world ùñ Coordinated policy response. However, we know the evolution for the US and (some) European countries.... little is known about the evolution of market power around the world.

3 This paper Study market power ùñ need to estimate markups. Demand based (Breshanan, 1989), (Berry, Levinsohn and Pakes, 1995): Demanding data requirements: detailed info on transactions (p and q) and assumptions on how and in which markets firms compete (ambiguous). Consequence: measures only for short period and specific industries. Cost-based (Hall, 1998), (De Loecker and Warzynski, 2012): Contribution No need to make assumptions on competitive behaviour nor define a market. All relevant info can be obtained from financial statements of firms. Extend (De Loecker and Eeckhout, 2017) using a richer database that allows to document evolution of markups all around the world.

4 Markup estimation: Methodology De Loecker and Eeckhout (2017) N firms: i 1,..., N. Technology: Q it pv it, K it, Ω it q F it pv it, K it qω it, where V it pv 1,..., V J q are variable inputs (labor, intermediate inputs), K it is the capital stock, Ω it is Hicks-neutral TFP term. Assume competitive input markets and consider V as scalar (composite input). Cost minimization min tv it,k itu s.t. P V it V it ` r it K it Q it p q ě Q it with associated Lagrangian L pv it, K it, Ω it q P V it V it r it K it ` λ it pq it p q Q it q.

5 Markup estimation: Methodology De Loecker and Eeckhout (2017) FOC w.r.t. V it BL p q BV it P V it ` λ it BQ it p q BV it 0. Re-arranging and multyplying both sides by V it {Q it yields where θ V it θ V it BQ it BV it V it Q it P V it λ it V it Q it, is the elasticity ot output w.r.t. input V. λ it is a measure of marginal cost, thus multiplying both sides by P it and further re-arranging yields θ V it P it Q it P V it V it Two key ingredients determine markup µ it : P it λ it µ it. Revenue share of the variable input, P V it V it P it Q it, Output elasticity of the variable input θ V it.

6 Markup estimation: Data Worldscope dataset (Thomson Reuters). From 1980 to 2016: 745,958 firm-year observations, 67,491 different firms in 134 countries. Mostly publicly traded firms, few private firms. Representativeness (for the US): results consistent across sectors; also consistent if using Census weights based on the entire universe of companies. Financial statements: Income Statement, Balance Sheet and Cash Flow. Cost of goods sold (variable cost), Selling, general and administrative costs (overhead cost), capital expenditure and firm s sales. Wage bill («labor share) and market value («profits). Implementation (2-step approach): 1 Estimate output elasticity θ V st for each sector s and year t ùñ all firms in the same sector in a given year have the same technology (but different TFP). 2 Compute firm s markup from information on sales and expenditure on variable inputs.

7 Results Figure: Global Market Power: Baseline.

8 Results Appendix C Figure B.1: Change in Markup Markup weighted by GDP Global Markup (Firm Sales) Global Markup (Continent GDP) Global Markup (Firm Sales) Global Markup (Country GDP) Figure: Global (a) Weight Market= Power: GDP by Robustness. continent (Left) Weight = GDP by (b) continent. Weight = (Right) GDP by Weight country = GDP by Country. Figure C.1: Global Markup: Weighted by GDP and by Firm Sales (Benchmark) One of the concerns about our data is that firms in some countries and continents are disproportionally represented compared to actual economic activity as measured by GDP. If firms from Europe and North America are overrepresented and firms in Africa are underrepresented, then the global markup estimate will disproportionately reflect the markup of Europe and

9 closely in all 6 regions, except for the early 1980s when North America is overrepresented, and Europe. Results.6 EUROPE NORTH AMERICA SOUTH AMERICA ASIA OCEANIA AFRICA Continent Share (Firm Sales) Continent Share (Continent GDP) Figure: Share of each continent: firm sales (baseline) and GDP. Figure C.2: Share of each continent: firm sales (baseline) and GDP.

10 Results Figure: (Left) Distribution of markups: 1980, 2016; (Right) Contour plot of Markups: 1980, Increase both in variance and mode. Flattening of the upper tail. Most firms have relatively low markups, but in 2016 there are more firms with high markups. Mass of distribution disproportionately above 45 o line ùñ on average, firms have higher markups in 2016.

11 Results Figure: Global regions.

12 Results European economies have seen steep increases, particularly Denmark, Switzerland and Italy. Except for Portugal, which had a modest decline in markups, the other European countries all show an increase in markups that is in line with the overall trend. Markup 2016 change Global Average Europe Denmark Switzerland Italy Belgium Greece United Kingdom Norway Ireland France Sweden Netherlands Finland Austria Spain Germany Portugal North America United States Canada Mexico Africa South Africa Markup 2016 change Asia South Korea Hong Kong India Japan Indonesia Thailand Malaysia Pakistan Taiwan Turkey China Philippines Oceania Australia New Zealand South America Argentina Colombia Brazil Peru Venezuela Chile Table 1: of Individual Countries (40 countries out of 134). Countries in each region are ranked Figure: Sample of Individual Countries. *Difference between markup in 2016 and by their change in markup. The Region and Global averages are for all countries in that geographical area, not just those reported in the table.

13 to a decline in markups. Finally, the economies in Oceania experience an increase in line with the global average, Results exemplified by Australia. New Zealand sees less of an increase. In Africa, the overall increase is more moderate, and its largest economy, South Africa, has a modest change. >1.75 (1.5,1.75] (1.25,1.5] [1,1.25] No data Figure Figure: 4: Markup Markup by by Country Country in in In broad lines, markups are increasing across the board, in all continents. It appears that themost more developed economies tendhave to have moderately bigger increases high markups. in whereas some of the emerging market economies see a decline. The world map in Figure 4 summarizes the geographic Emerging dimension economies of theare markup mixed, in with Most bothdeveloped high andeconomies low levels. have moderately high markups, with some high levels in Europe for Denmark, Belgium and Italy for example. Developed economies experience an increase, while the experience for Those in emerging economies in Asia, Africa and Latin America are more mixed, with both emerging economies is also mixed. the highest levels (such as Russia, Bolivia and several African countries), as well as the lowest

14 Results Markup (Inverse) Labor share (KN) Markup Market Value/Sales Markup (Inverse) Labor share (KN).54 1 Markup Market Value/Sales (a) Labor Figure: Share (Left) Labor Share; (Right) Market(b) Values/Sales. Market Value/Sales Figure 5: The Evolution of Labor Share and Market Value/Sales with Global Markups. Labor share data, for the corporate sector, from Karabarbounis and Neiman (2014). Think of labor as V it, declines. As a result, the demand for inputs such as labor decreases. This follows immediately from inspection of our expression for the markup in equation (1), which we can rewrite as and P V it Vit P itq it s L it ω itl it θl it, P it Q it µ it s π it 1 ωitl it P it KK it Fit. P it Q it P it Q it P it Q it = αit µ it. The expenditure share on variable inputs on the left-hand side is inversely related to the markup. One such a variable input is Labor, V it = L it with wage w it = Pit V, and as a 1result, Ò µ the labor share witlit it ùñ Ó quantity P itq the produced ratio of the ùñ expenditure Ó inputs ondemanded labor over sales ùñ isódecreasing s L as it. markups increase. 2 Ò µ it ùñ Ò s π Unfortunately, the itinformation or Ò F it P itq it. on the labor share in our data is not available for all firms to

15 Concluding Remarks Two take-away messages: 1 Increase in global average markups. Heterogeneous evolution among advanced and emerging economies. 2 Tight correlation between increase in markups, labor share decline and increase in market valuation of firms (profits?).