Finance and Factories Warren Bailey Cornell University European
Finance and Factories Warren Bailey Cornell University European Financial Management Symposium 2017 Finance and Real Economy 7 th to 9 th April 2017 Xiamen University 6 th April 2017 1
Why do factories interest me? Looking at lower Manhattan from Brooklyn Heights, 1960 s 2
A career in the stock market 3
A management job in a bank 4
Goal: A better standard-of-living 5
Now, imagine going up the East River, not across to Manhattan 6
Work in the brewery? No! 7
Work in the sugar refinery? No! 8
My career and stock markets (the first event study) Fama, Fisher, Jensen, and Roll, 1969, The Adjustment Of Stock Prices To New Information, International Economic 9 Review
Study stock market returns Brown and Warner, 1985, Using daily stock returns: The case of event studies, Journal of Financial Economics 10
Study stock market values Tobin’s q ≈ (market value of equity + book value of debt) / book value of assets 11
Study stock trading volume Kandel and Pearson, 1995, Differential Interpretation of Public Signals and Trade in Speculative Markets, Journal of 12 Political Economy
Study stock market investor behavior 13
2017: Reconsider the factory 14
Productivity: Glimmers of a promising direction for finance Hsieh and Klenow, 2009, Misallocation and Manufacturing TFP in China and India, Quarterly Journal of Economics • Productivity should be identical across firms • Re-allocate capital to efficient firms so marginal productivity equals the market price of capital • State-subsidized finance causes firms to accept a lower marginal productivity of capital 15
Productivity: How to measure it? • Sales/Capital? Physical output? Value-added? • Sales/Employee hours ? • Above adjusted for industry average? 16
Cobb-Douglas production function Yi =A K βK L βL M βM • Output “Y”; inputs “K” (capital), “L” (labor), “M” (materials) • Output elasticity βK , βL , βM • Total Factor Productivity (TFP): “A” 17
Classic empirical application Solow, 1957, Technical Change and the Aggregate Production Function, Review of Economics and Statistics • Measure aggregate US production function, 1909 to 1949, annual data • Take natural log of Cobb-Douglas, use logdifferences to estimate “A” (TFP) 18
Classic empirical application A(t) (TFP) is “a rough profile of technical change” across US economy 19
Estimating firm/plant level TFP ln(Yit) =αjt + βKjt ln(Kijt) + βLjt ln(Lijt) + βMjt ln(Mijt) + εijt • i = firm/plant, j = industry, t = time period • αjt = average industry-period TFP • εijt = firm-period TFP relative to industry 20
A simultaneity problem Olley and Pakes, 1996, The Dynamics of Productivity in the Telecommunications Equipment Industry, Econometrica 64 Levinsohn and Petrin, 2003, Estimating Production Functions Using Inputs to Control for Unobservables, Review of Economic Studies 70 • “Firms that have a large positive productivity shock may respond by using more inputs. ” • Use proxies (investment, materials) for unobserved TFP and recover TFP with fitted values (multi-stage process). 21
Survey of economics literature Syverson, Chad, 2011, What Determines Productivity? , Journal of Economic Literature • Survey, 120 references, discussions • Enormous, persistent productivity differences across firms in same industry • Competition, technology spillover, international trade, human capital, management practices, IT and R&D, regulation 22
Example of TFP use in economics Hsieh and Klenow, 2009, Misallocation and Manufacturing TFP in China and India, Quarterly Journal of Economics • Compare China (1998 -2005) , India (1987 -1994) to US • Larger TFP dispersion by industry in China and India • US levels of productivity boost TFP by 30%– 50% in China and by 40% to 60% in India • China closing the gap, India increasingly diverges 23
Example of TFP use in economics Bloom, Draca, Van Reenen, 2016, Trade Induced Technical Change? The Impact of Chinese Imports on Innovation, IT and Productivity, Review of Economic Studies • 12 European countries; patents, IT, TFP; 1996 to 2007 • Import competition enhances technological change and shifts employment to advanced firms • Employment declines, particularly for unskilled 24
Example of TFP use in economics Ge, Lai, and Zhu, 2015, Multinational price premium, Journal of Development Economics • Chinese manufacturing firms, 2000 to 2006, 77, 087 exporters (70% of exports) matched at 8 -digit level to international trade transactions from China Customs • foreign firms charge higher prices than local exporters • Relates to “knowledge-based intangible assets”, more innovative home country, TFP, seconded foreign employees, majority foreign owned 25
TFP use in empirical finance Most obvious uses are in empirical corporate finance. Examine the effects of: • changes in ownership • corporate strategy • managerial issues • trade-offs resulting from active investors 26
Important for finance? Lichtenberg, 1992, Corporate takeovers and productivity. Cambridge, MA: The MIT Press • “Many studies had been made of the purely financial ramifications of changes in corporate control, in particular their effect on stock prices, but little was known about the impact of takeovers on real variables such as output, productivity, employment, R&D, and fixed investment. ” • Quoting Malkiel (1990), “…stock prices are in a sense anchored to certain ‘fundamentals’ but the anchor is easily pulled up and then dropped in another place. ” 27
Leveraged Buyouts and TPF Lichtenberg and Siegel, 1990, The effects of leveraged buyouts on productivity and related aspects of firm behavior, Journal of Financial Economics • About 5% (1100) US plants in LBOs 1981 to 1986 • LBO associated with higher TFP growth compared to industry • Management buyouts have particularly strong positive effect on TFP 28
Lichtenberg and Siegel 1990 • Use of labor and capital declines but more slowly • Administrative staff to factory workers declines • Factory worker wages rise • Plants less likely to subsequently close than other plants • R&D keeps pace with non LBO plants 29
Corporate control and TPF Lichtenberg, 1992, Corporate takeovers and productivity. Cambridge, MA: The MIT Press • About 20, 000 US plants, 1972 to 1981 • Ownership changes associated with inefficient plants, they catch up somewhat • Greatest impact on non production jobs and wages • LBO associated with efficient plants, large improvements, and reduced industrial diversification 30
Mergers, Acquisitions, and TPF Maksimovic and Phillips, 2001, The market for corporate assets: Who engages in mergers and asset sales and are there efficiency gains? , Journal of Finance • About 50, 000 plants per year, 1974 to 1992 • Average 3. 89% change ownership each year • Close to 7% in In expansion years 31
Maksimovic and Phillips 2001 • TFP gain highest when former owner is relatively inefficient or new owner is relatively productive • Plant sold from peripheral division show TFP gains • In M&A, plants placed in buyer’s main division show TFP gains • After controlling for seller and buyer characteristics, TFP gains to M&A are insignificant 32
Maksimovic and Phillips 2001 • Buyer’s comparative advantage is in main industries • M&A premiums relate to gains in productivity • Well-run buyers apply their knowledge and improve plants they acquire • Empire building? Low-productivity buyers reduce productivity of newly acquired plants. 33
Diversification and TFP Schoar, 2002, Effects of corporate diversification on productivity, Journal of Finance • 20366 US plant-years, 1977 to 1995 • Conglomerates relatively more productive but decline in productivity over time • Acquired plants increase productivity, existing plants suffer • Productivity is tracked by stock prices 34
Law, management, and TFP Bertrand Mullainathan, 2003, Enjoying the quiet life? Corporate governance and managerial preferences, Journal of Political Economy • 1976 to 1995, 224188 plant-years • When state law protects managers from takeovers, wages rise, scrapping of old plants declines, opening of new plants declines, and productivity and performance decline • Slow decay suggests preference for “quiet life” 35
Managerial proximity and TFP Giroud, 2013, Proximity and investment: Evidence from plant-level data, Quarterly Journal of Economics • 1972 to 2005, US, 1332824 plant-years; • Airline routes and actual flights connecting US cities • When new airline route shortens time between HQ and plant, investment rises 8% to 9% and TFP rises 1. 3% to 1. 4% 36
Takeovers and TFP Li, 2013, Productivity, restructuring, and the gains from takeovers, Journal of Financial Economics • 1, 430 mergers, 1981 to 2002 • More efficient use of labor and capital raises productivity • Reduced capex, wages, and jobs for same output • Event study responses relate to size of productivity gains 37
Anecdote: The Twinkie trade-off between efficiency and jobs 38
Twinkies, profits, and jobs Bankrupt in 2012, hedge fund “rescues” Twinkies, owners benefit, but jobs are lost. 39
Activist investors and TFP Brav, Jiang, and Kim, 2015, The Real Effects of Hedge Fund Activism: Productivity, Asset Allocation, and Labor Outcomes, Review of Financial Studies • 787, 758 US plant-years, 1994 to 2007 • 2, 000 hedge fund activism events • Comparison to other hedge fund investments 40
Brav, Jiang, and Kim 2015 Plant performance rebounds with activist intervention 41
Brav, Jiang, and Kim, 2015 Cumulative TFP relative to year of activist intervention: • Declines t-3 to t • Rises t to t+3 42
Brav, Jiang, and Kim, 2015 Pre-intervention job losses vanish and labor productivity rises with activist intervention 43
Brav, Jiang, and Kim, 2015 Confirm the idea that activist interventions are key: when investors switch declared status to activist, some evidence that TFP and ROA rise. 44
Other recent work Kim, 2015, How Does Labor Market Size Affect Firm Capital Structure? Evidence from Large Plant Openings, Cornell University working paper • Large US plant openings • 2. 6% to 3. 9% increase in debt-to-capital ratio of local firms relative to others • Labor market deepening makes job loss less costly and reduces indirect costs of financial distress 45
Sources of firm or plant level manufacturing data Examples I will discuss next: • US Census • China National Bureau of Statistics Another: Levinsohn and Petrin (2003) use 1979 to 1986 data on 6665 plants in Chile 46
US Census plant-level data 1. Census of Manufacturers (CMF): all US manufacturing plants for years ending in 2 or 7 , ~300, 000 plants 2. Annual Survey of Manufacturers (ASM) : 50, 000 plants for “non-Census years”, always include plants > threshold number of employees, plus random plants with less • total value of shipments, capital stock and investment, labor hours, and material and energy costs 47
US Census plant-level data • Both publicly and privately owned plants • More accurate estimation of TFP with plant-level data rather than firm-level data 3. Longitudinal Business Database (LBD): annual plant identifiers and ownership changes > 5 million manufacturing and other businesses • include the number of employees, wages, industry SIC • Problem: restricted access 48
China firm level data Ge, Lai, and Zhu, 2015, Multinational price premium, Journal of Development Economics • annual surveys by National Bureau of Statistics (NBS) • financial and nonfinancial variables for all manufacturing firms with sales > RMB 5 million • ownership, employment, capital stock, gross output, value added, wages, industry • Problem: post 2007 lacks several key variables, notably 49 value-added
Why is NBS data so important? We can get Y, K, L, and M from annual reports or stock market databases. However: • NBS covers hundreds of thousands of firms large and small, not just a few thousand stock market listed firms • Firm TFP is computed as a residual relative to industry peers so we need a sufficient number of firms in each of approximately 400 industries • NSB has many useful variables (such as value-added for TFP calculation) as we detail later 50
Chinese manufacturing productivity Brandt, Tombe, and Zhu, 2013, Factor market distortions across time, space and sectors in China, Review of Economic Dynamics • Factor mis-allocation by province-sector shows greater distortions for state-owned sectors Song, and Wu, 2013, Structural estimation on capital market distortions in Chinese manufacturing, NTU w. p. • Productivity (revenue/assets) depends on state ownership and province of location 51
Chinese manufacturing productivity Campello, Ribas, and Wang, 2014, Is the Stock Market Just a Side Show? Evidence from a Structural Reform, Review of Corporate Finance Studies • productivity (sales to assets) increases for listed firms as split share reform increases the liquidity of equity Chen, Jiang, Ljungqvist, Lu, and Zhou, 2016, State Capitalism vs. Private Enterprise, working paper • capital allocated to profitable units only if group is privately owned 52
From the Stock Market to the Factory Floor: The Real Effects of Financial Dysfunctions Warren Bailey, Zhaoran Gong, Huiwen Lai, and Wilson Tong 31 st March 2017 Cornell University and Hong Kong Polytechnic University 53
Our paper • Theoretical model where manager chooses level of effort and degree of expropriation separately • Measure associations between distortions and firm level productivity and employment outcomes • Government ownership, political influences, subsidies, earnings management, tunneling, excess cash holdings 54
Our paper Does the grabbing hand simply steal from shareholders or does it extend into the firm’s operations? Can we learn more by looking beyond stock market data? ? 55
China: recent rapid economic growth and reform • Recent papers document problems, dysfunctions, and other limitations • Political influences • Expropriation and weak disclosure by management • Subsidies and other propping 56
Are we picking on China? • China is a unique and important case • Many capital market problems can be measured readily • Other countries suffer capital market dysfunctions too Accounting fraud Theft by management Securities fraud Incompetent politicians 57
Existing research approach How has the impact of dysfunctions in governance, disclosure, and political relationships been measured? • Stock market values and accounting disclosures • Observable evidence of the impact of the “grabbing hand” of government, managers, and other influential stakeholders on ordinary shareholders We can look at this from a different perspective. 58
Empirical specifications • Outcome variable = dysfunction proxies + controls + ε TFP Labor/Assets Labor productivity Total wage/labor Sales/Labor State ownership Political influences Disclosure quality Tunneling Excess cash holding • Pooled cross section time series regressions • Fixed effects • Two versions of TFP 59
State ownership State Controlled = 1 if plurality of ownership is government, else 0 Also, plurality dummies for: • • • Collective Controlled Individual Controlled Hong Kong, Macau, or Taiwan Controlled Foreign Controlled Other Legal Persons Controlled (benchmark) 60
Political influences Official turnover • “Pre-turnover” is dummy equal to 1 if current year or year prior has provincial promotion event (governor or party secretary) following Piotroski and Zhang (2014). State Subsidy • Annual amount of government subsidy to firm from NBS (Piotroski, Wong, and Zhang, 2015). “Subsidy Dummy” equals 1 if firm receives a subsidy that year. 61
Political influences Political Connections • “From Central ” is dummy variable equal to 1 if current CEO or chairman currently or previously served in a central government administrative or legislative position • “From Provincial” is similarly defined for provincial servie • “To Central” is dummy variable equal to 1 if current CEO or chairman subsequently serves in central government administrative or legislative position 62
Disclosure quality • “Earnings Aggressiveness” = Accruals/Total Assets, Accruals equal total profit minus net cash flow. Firmlevel, not time-varying • “Earnings Smoothing” is ratio of standard deviation of operating cash flows from assets to standard deviation of total profits from assets. Firm-level, not time varying • Loss avoidance” is captured with dummy variables for small versus large and positive versus negative net income. Small positive in particular can be evidence of manipulation. “Small” = less than 1% of assets. 63
Tunneling • “Related Lending” is loans to parent, subsidiary, or other group member divided by total assets (Jian and Wong, 2010). • “Related Party Transactions” is annual change in other accounts receivables (scaled by total assets) from related-party transactions (Bailey, Huang, and Yang, 2011). CSMAR database so listed firms only. • Other receivables/Assets (Jiang, Lee, and Yue, 2010) CSMAR database so listed firms only. 64
Excess cash holdings • Can be diverted by insiders to private benefits • Estimated by residual of a pooled time series cross sectional regression (Fresard and Salva, 2010). • Cash and marketable securities, total assets, net working capital, market value, capex, leverage, R&D, dividends 65
Related empirical research on PRC capital markets Liao, Liu, and Wang, 2014, China’s secondary privatization: Perspectives from the split-share structure reform, Journal of Financial Economics • Less pessimistic SOE performance evidence from more recent years 66
Related empirical research on PRC capital markets Cai, Fang, and Xu, 2011, Eat, Drink, Firms, Government: An Investigation of Corruption from the Entertainment and Travel Costs of Chinese Firms, Journal of Law and Economics • Entertainment and Travel Costs reported in three surveys • Correlated with estimated government services received, government expropriation, relationship capital, top management compensation 67
Related empirical research on PRC capital markets Fan, Wong, and Zhang, 2007, Politically connected CEOs, corporate governance, and Post-IPO performance of China's newly partially privatized firms, Journal of Financial Economics • Substantial cumulative stock return underperformance of IPOs with politically connected CEO 68
Related empirical research on PRC capital markets Berkman, Cole, and Fu, 2011, Political connections and minority shareholder protection: Evidence from securitiesmarket regulation in China, Journal of Financial and Quantitative Analysis • Protection for minority shareholders less effective for state owned firms 69
Disclosure quality and value Fernandes and Ferreira, 2007, The Evolution of Earnings Management and Firm Valuation: A Cross-Country Analysis (October) • 24, 000 firms across 43 countries 1990 to 2003 • Low earnings management is associated with high Tobin’s q 70
Disclosure quality and value: complicated Xie, 2001, The Mispricing of Abnormal Accruals, The Accounting Review Marciukaityte and Varma, 2008, Consequences of overvalued equity: Evidence from earnings manipulation, Journal of Corporate Finance Badertscher, 2011, Overvaluation and the Choice of Alternative Earnings Management Mechanisms. The Accounting Review • Complex feedback between accounting performance, stock valuation, and earnings management 71
Disclosure quality and value: complicated Julio and Yook, 2016, Earnings Management and Corporate Investment Decisions, BGFRB working paper • “Taken together, these results support the view that a moderate amount of earnings management helps improve corporate investment decisions while an excessive amount undoes the benefit of earnings management. “ 72
Our findings (tentative) Political influences • State ownership: strong negative association with TFP (foreign is positive) • Provincial official turnover: strong positive association with TFP when interacted with State Controlled dummy 73
Our findings (tentative) Political influences • Subsidies: strongly positive association with Subsidy Dummy x State Controlled dummy, but strongly negative association with Subsidy/Assets x State Controlled dummy (don’t take too much? ) • Political connections: strongly positive association with From Central, none for From Provincial, and note sample is small (CSMAR listed only) 74
Our findings (tentative) Disclosure quality • Earnings aggressiveness and Accruals: tangled with ROA (our measure is not unexpected) so hard to interpret. Seems negatively correlated with TFP. • Earnings smoothing, Small Positive ROA (loss avoidance): Positive correlation with TFP is surprising: “good” firms manipulate earnings? • Speculation: hide money from the “grabbing hand”? 75
Our findings (tentative) Tunneling • Other Receivables and Related Lending: negatively correlated with TFP, which makes sense • Related Party Transactions: marginally negative or insignificant correlation with TFP, which is okay 76
Our findings (tentative) Excess cash holdings • Strong positive correlation with TFP. Surprising since excess cash holdings should be associated with a “bad” company • Speculation: good companies hold excess cash to fund expansion, rather than borrow or raise additional equity 77
Other findings Include more variables in specifications • Many of the associations, including anomalous ones, hold up to including more variables in each specification Specifications for labor related measures • Sales/Labor is easiest to interpret as a productivity measure, and it is broadly similar to TFP 78
Planned extensions • Extend data to more recent times using more primitive productivity measures • Interactive effects: test out theories for anomalies such as “good” firms manipulating earnings or holding excess cash • Difference-in-difference around key events that can alter relationships (identification) in predicted ways • Use accounting reforms to test if certain firms (SOE, not foreign owned) fake their data 79
Tentative conclusions • Typically, dysfunctions are associated with lower TFP • Why, however, would a “good” (that is, high TFP) company perform earnings management or hold high excess cash balances? • Do anomalies illustrate further unique features of the Chinese financial market environment? 80
Expanding the intersection of finance and productivity • What is the performance of different types of ownership? • What are implications for stock holders, workers, taxpayers? • What is the impact of the political, legal, and regulatory environment? • Culture and finance? 81
Idea: Chinese multi-nationals 2016 (25%) 2014 2010 (failed) 2010 Judge by stock market and by operating performance 82
Idea: Do SWFs identify, add, or destroy value? 83
Idea: What is the effect of restructuring? 3 G Capital • • • Kraft Heinz (2015) = Kraft Food Group + Heinz Burger King (2012) partial spin-off Anheuser-Busch In. Bev (2008) SABMiller (2016) Burger King (2010) + Tim Horton’s (2014) = Restaurant Brands International 84
Idea: how deep is corruption? Dass, Nanda, and Xiao, 2016, Public Corruption in the United States: Implications for Local Firms, Review of Corporate Finance Studies • US state level corruption (corruption convictions) • Lower Tobin’s q and poorer transparency (earnings guidance, number of analysts, stock liquidity) • Less effect for companies that deal extensively with government 85
Idea: how deep is corruption? Smith, 2016, US political corruption and firm financial policies, Journal of Financial Economics • US DOJ data on local political corruption (corruption convictions) • Lower cash holdings and higher leverage for firms located in corrupt areas • Reduce assets available for politicians to expropriate 86
Idea: how deep is corruption? Nanda and Pareek, 2016, Do Criminal Politicians Affect Firm Investment and Value? Evidence from a Regression Discontinuity Approach, https: //ssrn. com/abstract=2782580 • Criminal backgrounds of Indian politicians • Closely contested elections (regression discontinuity) • Lower stock returns and capex in districts that return a criminal politician • SOEs offset • Criminal-politician is less destructive if party in power or if holding a ministerial level post 87
Idea: Using survey data on management style US Census Bureau 2010 Management and Organizational Practices Survey (MOPS) Bloom, Brynjolfsson, Foster, Jarmin, Saporta, and Van Reenen, 2013, Management in America, US Census Bureau Center for Economic Studies Paper No. CES-WP 13 -01. • > 30, 000 manufacturing plants across the US • impact of “structured management practices” 88
Bloom, Brynjolfsson, Foster, Jarmin, Saporta, and Van Reenen, 2013 • Performance monitoring, target setting and incentives enjoy greater productivity and profitability, higher rates of innovation and faster employment growth • more likely in exporters, large establishments, educated employees • increased between 2005 and 2010 for surviving • establishments, especially involving data collection and analysis 89
Idea: More evidence on who gains or loses from capitalism “Detroit Industry Murals”, Diego Rivera, 1932 - 1933 90
Conclusion • It seems that firm or plant level “real” data is underused, though interest seems to be growing. • Acquiring and deploying the data can be annoying but seems promising. • I am pleased that my career now takes me into factories. 91
- Slides: 91