Understanding South African Manufacturing Markups

Markups in South African Manufacturing Are they high and what can they tell us? TIPS Conference 20-21 May 2014: Manufacturing Led Growth for Employment and Equality Nimrod Zalk

Overview Critically reviews recent attempts to estimate and interpret manufacturing markups / price-cost margins(PCMs) Particular focus on Aghion, Braun, Fedderke (ABF) (2008) Academic citation Policy influence: ASGISA Harvard Group , NDP, IMF, World Bank, OECD Three core claims re PCMs 1970-2004 (C1) SA PCMs higher than rest of the World (C2) SA PCMs have been non-reducing (C3) Large and negative causal relationship between higher PCMs and lower growth Replicability of results is key to scientific method: ABF did not respond to requests for access to their data

Perspectives on industrial structure, conduct and performance Late apartheid stagnation consensus Domination by handful of large conglomerates Extensive holdings within and across sectors Failure to diversify, add value and in particular to develop manufacturing exports Market distortions perspective Product markets: tariffs; sector concentration Factor markets: financial repression (low or negative real interest rate); overvalued exhange rate Capital misallocation Liberalise product and factor markets

Perspectives on industrial structure, conduct and performance Minerals-Energy-Complex MEC manufacturing fundamentally linked to mining and energy sectors Capital accumulation chiefly through complementary large private and public corporation investments Trade policy played a secondary role Reorient resources and capabilties of conglomerates to diversify through building on economies of scale and scope Market distortions perspective won and fundamentally shaped post-apartheid economic policy ABF and related papers sit firmly within market distortions perspective

ABF: methods and datasets ABF seek to explain SA manufacturing productivity growth as a function of PCMs PCM measures Roeger methodology: difference between (quantity- based) primal and (price-based) dual of Solow Residual / Total Factor Productivity (TFP) as measure of imperfect competiton Lerner Index: (VA W) / Sales or (VA W rK) / Sales PCM estimates heavily sensitive to assumptions Intermediate inputs Rental price of capital Productivity growth Labour productivity TFP

ABF: methods and datasets Pooled Mean Group Estimation (PMGE) Assumes all manufacturing sectors have the same long term equilibrium PCM with only short term variations Appropriateness in light of structural change since 1970? Regressions Neoclassical growth model Explanatory variable: PCMs Dependent variable: productivity growth Solow Residual / TFP based estimates Neoclassical growth model and variants assumes perfect competition, full employment etc. Appropriateness of estimating and interpreting imperfect competition under model which assumes perfect competiton?

PCMs: Cross-country estimates [M]ark-ups are significantly higher in South African manufacturing than they are in corresponding industries worldwide ___________ Consistently over the three datasets, mark-ups are significantly higher in South African industries than they are in corresponding industries worldwide ___________ We find consistent evidence of pricing power in South African industry that is greater than international comparators, and which is non-declining over time. ___________ Consistently across the three datasets, we found that: (i) mark-ups remain significantly higher in South African industries than in corresponding industries worldwide

PCMs: Cross-country estimates Unido Indstat2 Compiled from NSAs dependent on frequency of reporting Structural break in SA data from 1993: shift to conformance with international SNAs Missing or implausible values for 1992, 1994, 1995, 1997, 1998 High data volatility World average against which SA data is compared is not presented No conceptual distinction between Advanced and Developing / Transition

PCMs: Cross-country estimates Manufacturing Price-Cost Margin estimates: for South Africa and Advanced and Developing / Transition economy averages, 1993-2010 0.30? 0.28? 0.26? 0.24? 0.22? 0.20? 0.18? 0.16? 0.14? 0.12? 0.10? 0.08? 1993? 1994? 1995? 1996? 1997? 1998? 1999? 2000? 2001? 2002? 2003? 2004? 2005? 2006? 2007? 2008? 2009? 2010? Mean:? Developing/Transi on? Median:? Developing/Transi on? Mean:? Advanced? Median:? Advanced? Note: SA 1998 value: 0.46, truncated South? Africa? Source: Unido Indstat2

PCMs: Cross-country estimates Worldscope listed firm data SA PCMs 0.12 vs World Average 0.11 But much greater prominence to selected financial ratios (Net Income/Sales, Net Income/Assets, Net Income/Equity), subordination of others (ABF, 2008; Fedderke (2013) Net Income/Sales: SA 50 percent higher than world average Net Income/Assets: SA double world average (Fedderke (2013)) Listed firms as representative of entire population of manufacturing firms (ABF, 2008; Fedderke (2013)) Categorisation of listed firms Apparent reliance on indices such as Industrials , Consumer Goods Indices include many non-manufacturers In-country and foreign operations Financial activities of non-financial corporations (NFCs) Survivorship bias (Gilbert, 2013) Listed firm data virtually meaningless without correcting for these factors

PCMs: SA-specific estimates ABF Present results of Roeger method with PMGE derived period averages Use Lerner index proxy in their regressions, claiming they are consistent with Roeger results Our replication of Lerner index proxy Manufacturing PCMS lower than most other broad sectors Not greater than 10% Have declined dramatically since 2004 Considerable variation amongst manufacturing sectors PCM trends consistent with profitability: Net Operating Surplus

PCMs: SA-specific estimates South African PCMs by broad sector, 1993-2012 ? 0.45? ? ? 0.40? ? ? 0.35? ? ? 0.30? ? ? 0.25? ? ? 0.20? ? ? 0.15? ? ? 0.10? ? ? 0.05? ? ? -? ? ? ? ? -0.05? ? ? -0.10? ? ? -0.15? ? 1993? 1994? 1995? 1996? 1997? 1998? 1999? 2000? 2001? 2002? 2003? 2004? 2005? 2006? 2007? 2008? 2009? 2010? 2011? 2012? Agriculture,? forestry? and? fishing? Coal? mining? Other? mining? Manufacturing? Civil? engineering? and? other? construc on? Wholesale? and? retail? trade? Transport? and? storage? Communica on? Business? services? Medical,? dental? and? veterinary? services? Source: SASID Gold? and? uranium? ore? mining? Building? construc on? Catering? and? accommoda on? services? Finance? and? insurance? Excluding? medical,? dental? and? veterinary? services?

PCMs: SA-specific estimates South African Net Operating Surplus by broad sector, 1993-2012 ? 0.70? ? ? 0.65? ? ? 0.60? ? ? 0.55? ? ? 0.50? ? ? 0.45? ? ? 0.40? ? ? 0.35? ? ? 0.30? ? ? 0.25? ? ? 0.20? ? ? 0.15? ? ? 0.10? ? ? 0.05? ? ? -? ? ? ? ? -0.05? ? ? -0.10? ? ? -0.15? ? 1993? 1994? 1995? 1996? 1997? 1998? 1999? 2000? 2001? 2002? 2003? 2004? 2005? 2006? 2007? 2008? 2009? 2010? 2011? 2012? Agriculture,? forestry? and? fishing? Other? mining? Civil? engineering? and? other? construc on? Transport? and? storage? Business? services? Source: SASID Coal? mining? Manufacturing? Wholesale? and? retail? trade? Communica on? Medical,? dental? and? veterinary? services? Gold? and? uranium? ore? mining? Building? construc on? Catering? and? accommoda on? services? Finance? and? insurance? Excluding? medical,? dental? and? veterinary? services?

Relationship between productivity and PCMs Potential for endogeneity: causality could run from PCMs to productivity Instrumentation: various measures of import penetration Instrumentation unsuccesful with respect to cross-country (sector and firm) PCMs No statistical significance of listed firm PCMs Instrumentation ambiguous in relation to SA-specific sector PCMs: partial and low correlation between PCMs and instrumental variables Agion et al. (2013) General Method of Moments results also ambiguous Selective measures of trade liberalisation as explanatory variables PCMs as control variable PCM statistical significance at expense of significance of trade variable

Conclusions No convincing evidence marshalled by ABF that SA manufacturing PCMS are in general high By international standards Relative to other SA sectors Evidence of any clear relationship between PCMs and productivity is at best tenuous Sectoral PCM measures as constructed by ABF add little value Heavily dependant on assumptions Masks heterogeneity between and within sectors Cannot substitue for the need for Detailed sectoral case studies Broader political economy context

Conclusions More detailed evidence points to some adverse outcomes of reforms to lessen market distortions Capital account liberalisation Exit of long term capital Reliance on short term inflows drive overvaluation and volatility Relative prices turn against manufacturing Corporate restructuring Unbundling lowered cross-sectoral concentration Sector-specific rebundling entrenched dominance in certain sectors e.g. intermediate inputs Competition Act: negotiated outcome with limited powers to act against pre-existing structure A number of value-adding and labour-intensive sectors caught between monopsonistic input providers and intense product market competition under liberalisation