REGULATION, INSTITUTIONS AND PRODUCTIVITY: NEW MACROECONOMIC EVIDENCE FROM OECD COUNTRIES

Transcription

1 REGULATION, INSTITUTIONS AND PRODUCTIVITY: NEW MACROECONOMIC EVIDENCE FROM OECD COUNTRIES Balázs Égert, OECD, Economics Department The usual disclaimer applies

2 Renewed interest in quantifying the impact of reforms on growth OECD s quantification effort: a production function approach to assess policy impacts: GDP per capita labour productivity employment rate investment in physical capital multi-factor productivity labour force participation rate unemployment rate

3 Regulation and MFP in the literature A number of studies using Firm-level data Sectoral-level data Few studies looking at the aggregate level

4 Micro vs. macro estimates Micro approaches: Better identification of policy effects Difficulties in estimating country-level effects diff-in-diff approach, several assumption are needed eg on base effect zero policy effect on worst performing firms/sectors effect is channel specific Macro approaches: Better at providing macroeconomic effects Better at incorporating more policy channels Better at covering a larger number of countries

5 New set of panel estimates

6 SELECTING MFP DRIVERS AT THE AGGREGATE LEVEL

7 MFP drivers Innovation intensity and trade openness - MFP depends on the creation, transmission and absorption of knowledge (Isaksson, 2007) - Innovation intensity and trade openness capture the above phenomena - Various measures of innovation intensity - Trade openness adjusted for country size

8 Innovation intensity MFP drivers - Spending on R&D by industry - Spending on R&D by government - Spending on basic R&D - Nobel Prize winners - Various measures of patens (per capita)

9 MFP drivers Trade openness - Adjusted for country size - Trade openness regressed on population, without country and time fixed effects - Residual gives the part of trade openness not explained by country size

10 Regulations Product and labour market regulations - More stringent regulation can impede the efficient allocation of capital and labour within and across firms and industries. - This can have a negative impact on MFP

11 Regulations Product labour market regulations - OECD s PMR indicator, WB s Doing Business indicator => at low frequency / short timespan - OECD s ETCR indicator: subset of the wide PMR indicator (network industries) - Annual observations ETCR - Overall ETCR, entry barriers, public ownership

12 Regulations Labour labour market regulations i.) per capita spending on active labour market policies (ALMP), ii.) the employment protection legislation (EPL) indicator (for permanent contracts), iii.) the gross unemployment benefit replacement rate (UBRR).

13 Time-varying variables Openness NAME DEFINITION SOURCE MAX TIME COVERAGE open exports and imports of goods and services over GDP divided by two World Bank WDI open_adj openness adjusted for country size, obtained as the residual from a regression of openness on a constant and total population World Bank WDI Innovation intensity berd business expenditures on R&D, % of GDP OECD Main STI indicators berdpriv business expenditures on R&D, financed by industry, % of GDP OECD Main STI indicators berdpub business expenditures on R&D, financed by government, % of GDP OECD Main STI indicators gerd general expenditures on R&D, % of GDP OECD Main STI indicators gerdpriv general expenditures on R&D, financed by industry, % of GDP OECD Main STI indicators gerdpub general expenditures on R&D, financed by government, % of GDP OECD Main STI indicators gerdbasic general expenditures on basic R&D, % of GDP OECD Main STI indicators nobel Nobel prize winners in hard sciences (economics excluded), 1985 and 2data collected from Product market regulations ETCR all Regulation in Electricity, Transport and Communication (ETCR) OECD PMR database ETCR en entry barriers in ETCR OECD PMR database ETCR po public ownership in ETCR OECD PMR database Labour market regulations ALMP per capita spending on active labour market policies OECD ELS statistics EPL the employment protection legislation (EPL) indicator for permanent contracts OECD ELS statistics gross unemployment benefit replacement rate OECD ELS statistics Controls hcap og human capital, calculated using a Mincer equation and mean years of schooling output gap: difference between the level real GDP and a trend estimated using the HP filter OECD EO database OECD EO database

14 Time-invariant variables Institutions - rule of law - quality of legal system - corruption (three alternative measures) Regulations / business environment - PMR and subcomponents - cost and time of starting a business/insolvency - as conditioning variables for the time varying variables (threshold regressions or interactions) - replacing country fixed effects

15 Time-invariant variables NAME SOURCE AVERAGED FOR Product market regulation (PMR) indicator overall PMR indicator State control barriers to entrepreneurship barriers to trade&investment Business environemnt time of insolvency procedures resolving insolvency, years World Bank WDI cost of insolvency procedures resolving insolvency, % of estate World Bank WDI Institutions OECD, Product Market Regulations database , 5-year intervals recovery rate - insolvency procedure as cents on the dollar recovered by secured creditors time of starting a business days World Bank WDI cost of starting a business % of GNI per capita World Bank WDI time of contract enforcemet days cost of contract enforcement % of the claim rule of law World Bank WDI quality of legal system corruption: corruption percention index Fraser Institute's Economic Transparency International corruption: corruption percention index World Bank World Governance Indicators corruption: corruption index Dreher et al. (2007)

16 PMR sub-components Product market regulation State control Barriers to entrepreneurship Barriers to trade and investment Public ownership Involvement in business operations Complexity of regulatory procedures Administrative burdens on start-ups Regulatory protection of incumbents Explicit barriers to trade and investment Other barriers to trade and investment Scope of SOEs Gov t involvement in network sectors Direct control over enterprises Governance of SOEs Price controls Command and control regulation Licenses and permits system Communication and simplification of rules and procedures Admin. burdens for corporations Admin. burdens for sole proprietor firms Barriers in services sectors Legal barriers to entry Antitrust exemptions Barriers in network sectors Barriers to FDI Tariff barriers Differential treatment of foreign suppliers Barriers to trade facilitation

17 MFP MFP MFP MFP Stylised facts: regulation business spending on R&D by industry (% of GDP) (log) size-adjusted trade openness OECD's ETCR indicator spending on ALMP OECD's EPL indicator (regular contracts)

18 MFP MFP MFP Stylised facts: institutions rule of law legal system corruption (WB WGI) higher values = less corruption

19 Descriptive stats (1) time-varying variables Standard deviation is much larger for the between dimension than for the within dimension with CFE & TFE without CFE without CFE and TFE Min Max Mean Sdev Min Max Mean Sdev Min Max Mean Sdev MFP mfp etcr overall etcr_new etcr entry barriers etcr_en etcr public ownership etcr_po openness size adjusted open openness, size adjusted open_adj business exp. on R&D berd business exp. on R&D by industry berdpriv business exp. on R&D by government berdpub general exp. on R&D gerd general exp. on R&D by industry gerdpriv general exp. on R&D by government gerdpub general exp. on basic R&D gerdbasic Nobel Prize winners nobel EPL - permanent contracts epl spending on active labour market policies almp unemployment benefit replacement rate ubrr output gap oghp human capital hcap log(human capital) _lhcap

20 ESTIMATION ISSUES: FINDING THE RIGHT TOOL

21 Estimation strategy Our main interest: long-run effect MFP f ( OPEN j, t, INNOVATION j, t, PMR j, j, t t ) MFP f ( OPEN j, t, INNOVATION j, t, PMR j, t, LMR j, j, t t ) Long-term relationship: Dynamic OLS (DOLS) corrects for endogeneity and serial correlation 1 Y i,t = β i X i,j,t + δ i ΔX i,j,t+l + CFE i + ε c,t j j l= 1

22 Policy interactions Threshold regressions : the impact of one policy on MFP could depend on the level of another policy T LMR if PMR INNOVATION OPEN T LMR if PMR INNOVATION OPEN MFP t t j t j t j t t j t j t j t j, 2, 3, 2 1, 1, 3, 2 1,

23 Data Annual panel data covering All (or almost all) 34 OECD countries Period from 1985 to 2011 (constrained by the MFP measure)

24 Estimation results

25 MFP and innovation intensity - strong positive link bw. R&D spending & MFP - positive link for overall R&D spending - positive link for private R&D spending - but missing or negative link for public R&D spending (whether with long lags or together with private R&D) - positive link for spending on basic research coeff estimate much larger - positive link for our Nobel Prize variable

26 Linear models (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) constant c 9.891** 9.702** 9.904** ** ** ** ** ** ** ** ** ** ** 11.39** ** etcr overall etcr_new ** etcr entry barriers etcr_en ** etcr public ownership etcr_po ** ** ** ** ** ** ** ** -0.02** * ** log (etcr public ownership) l_etcr_po openness size adjusted open_adj 0.007** 0.007** 0.008** 0.008** 0.008** 0.008** 0.008** 0.009** 0.009** 0.008** 0.008** 0.008** 0.008** 0.011** 0.002** log(openness size adjusted) l_open_adj business exp. on R&D berd 0.041** 0.038** 0.053** 0.031** 0.032** 0.038** business exp. on R&D by industry berdpriv 0.059** business exp. on R&D by government berdpub ** business exp. on R&D by government (-5) berdpub ** general exp. on R&D gerd 0.031** general exp. on R&D by industry gerdpriv 0.054** general exp. on R&D by government gerdpub general exp. on R&D by government (-5) gerdpub general exp. on basic R&D gerdbasic 0.349** Nobel Prize winners nobel_new 0.003** cointegration test, H0: no cointegration (p-value) error correction term ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** adjusted R-squared No. of observations No. of countries country fixed effects YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES time fixed effects NO NO NO YES YES YES YES YES YES YES YES YES YES YES YES

27 MFP and trade openness - strong positive link bw. openness & MFP - very robust position relationship - only instance when it becomes weaker => for smaller sample - less developed countries may benefit more from technology diffusion and adoption via trade

28 Policy interactions some degree of substitution bw. R&D spending and openness The positive openness effect is smaller if private R&D spending is high The positive link bw R&D and MFP becomes small at very high levels of openness

29 Regulations Product market regulation: negative relation bw. ETCR & MFP - overall ETCR: only if no year fixed effects used - public ownership part of ETCR if both country and year fixed effects included Labour market regulations: - Results sensitive to sample size and specification - If used in logs: ALMP + sign - If used in logs: EPL + sign => composition effect if tighter EPL destroys low-skill (low-productivity) jobs and leaves high-skill jobs broadly unaffected OR business can compensate through a better organisation of production processes

30 Linear models (1) (2) (3) (4) (5) (6) constant c ** ** ** ** ** ** etcr public ownership etcr_po_ ** -0.02** ** ** ** -0.02** openness size adjusted open_adj_ 0.007** 0.006** 0.01** 0.006** 0.006** 0.01** business exp. on R&D by industry berdpriv_ ** ** ALMP almp_ 4.0E-04 unemployment benefit replacement rate ubrr_ 2.0E-05 EPL epl_ log(almp) l_almp_ 0.033** log(unemployment benefit replacement rate) l_ubrr_ log(epl) l_epl_ 0.123** cointegration test, H0: no cointegration (p-value) ** ** ** ** ** ** error correction term adjusted R-squared No. of observations No. of countries country fixed effects YES YES YES YES YES YES time fixed effects YES YES YES YES YES YES

31 Policy interactions two-way interaction between product and labor market regulations ETCR s negative impact on MFP is at work only if EPL is low (not very restrictive) There is a strong negative relation linking EPL and MFP only if ETCR indicates less stringent product market regulation the negative impact of ETCR doubles in size for very high levels of trade openness labor market regulations interact with each other. ALMP s positive influence on MFP only works if EPL is low

32 Policy interactions (1) (2) (3) (4) (5) (6) (7) threshold variables open berdpriv almp epl etcr_po F-test of nonlinearity (p-value) threshold percentile threshold value c ** ** ** ** ** ** ** etcr public ownership etcr_po ** ** ** openness size adjusted open_adj 0.009** 0.008** 0.006** 0.010** 0.008** 0.010** business exp. on R&D by industry berdpriv 0.064** 0.058** 0.042** ** non-linear variables business exp. on R&D by industry threshold var low 0.066** when threshold var high openness size adjusted when etcr public ownership when ALMP when EPL when threshold var low threshold var low threshold var low threshold var low 0.010** ** 0.007** * ** ** threshold var high threshold var high threshold var high threshold var high 0.004** ** ** 0.031* ** cointegration test, H0: no cointegration (p-value) error correction term ** ** ** ** ** ** ** adjusted R-squared No. of observations No. of countries country fixed effects YES YES YES YES YES YES YES time fixed effects YES YES YES YES YES YES YES

33 Policies and institutions Interactions between time-varying policies and timeinvariant institutions/regulations Most of the benefits of better institutions transit through the channel of R&D spending - firmer rule of law and better law enforcement amplify the positive effect of R&D spending - more costly and lengthy contract enforcement offsets some of the benefits of higher R&D spending - Longer and more costly insolvency procedures attenuate the positive impact of R&D on MFP. - the benefits of R&D will be reduced if it takes longer to start a business.

34 Level effects of institutions and regulations Previous results include country fixed effects How changes in policies are related to average changes over time in MFP But they do not tell why MFP levels differ across countries Remember, cross-country variation is much larger than within variation

35 Replacing CFE by time invariant variables four sets of variables with cross-section variation the country averages of two labor market indicators (EPL and ALMP) barriers to trade and investment institutional quality (the rule of law and the time of insolvency procedures they enter the regressions one by one) corruption

36 Level effects (1) (1) (2) (3) (4) (5) (6) constant 9.331** 9.899** ** 9.461** ** ** etcr public ownership ** ** ** ** openness size adjusted 0.005** 0.008** 0.008** 0.005** 0.002** 0.003** business exp. on R&D by industry 0.055** 0.082** 0.059** 0.050** 0.051** rule of law - constant 0.225** 0.351** PMR barriers to trade&investment - constant ** ** EPL - country averages ** ALMP - country averages 0.006** EPL ** ALMP 0.004** cointegration test, H0: no cointegration (p-value) ** ** ** -0.04** E-05 error correction term adjusted R-squared No. of observations No. of countries country fixed effects NO YES YES NO NO NO time fixed effects NO NO YES YES NO NO

37 NEGATIVE sign on EPL (consistent with earlier findings, e.g Andrews and Cingano, 2014) EPL varies on average much more across countries than over time. Over time, composition effects & compensation through a more efficient organisation of the production process Across countries, the differences are larger and these large differences may translate into considerably different MFP levels by prohibiting a

38 Level effects (2) - corruption (1) (2) (3) constant ** ** 10.59** product market regulation (ETCR public ownership) ** ** trade openness (size adjusted) 0.002** 0.001* 0.003** business expenditure on R&D (% of GDP) 0.035** ** output gap log human capital employment protection legislation (EPL) ** ** -0.18** active labour market policy (ALMP) 0.004** 0.005** 0.003** PMR barriess to trade & investment (cross country) ** ** ** cross country variables corruption perceptions index (Transp. Inter) 0.005** corruption index ** control of corruption (WGI) 0.207** adjusted R-squared No. of observations No. of countries

39 Conclusions Anticompetitive product market regulations are associated with lower MFP levels This negative link dependent on the level of labor market regulations Better institutions and lower regulation amplify the positive impact of R&D spending on MFP Cross-country MFP variations => cross-country variation in LMR, barriers to trade and investment; and institutions (including corruption)

40 Thank you

41 Firm-level evidence - sizeable and persistent differences in firm-level productivity in the same sector (Bartelsman et al., 2004; Syverson, 2011) - due to the distortion in the allocation of capital and labour (Bartelsman et al., 2013) - labour and product market regulations and FDI restrictions penalise allocative efficiency (Andrews and Cingalo, 2014)

42 Sector-level evidence - negative relationship between regulation and multifactor productivity growth (Nicoletti and Scarpetta, 2003) - ICT-intensive sectors particularly penalised (Arnold et al. 2008) - tighter PMR in services and network industries hampers MFP growth in sectors using output of the regulated industries as inputs (Bourlès et al. 2012) - Negative impact of regulation via the negative impact on R&D (Cette et al. 2013) - Labour market regulations also matter for MFP at the sector level (Cette et al. 2014)

43 Sector-level evidence - negative relationship between regulation and multifactor productivity growth (Nicoletti and Scarpetta, 2003) - ICT-intensive sectors particularly penalised (Arnold et al. 2008) - tighter PMR in services and network industries hampers MFP growth in sectors using output of the regulated industries as inputs (Bourlès et al. 2012) - Negative impact of regulation via the negative impact on R&D (Cette et al. 2013) - Labour market regulations also matter for MFP at the sector level (Cette et al. 2014)

44 the MFP frontier: who is leading the pack

45 MFP drivers The productivity frontier - commonly used in firm- and sector-level analysis - more complicated at the aggregate level - (sectoral) composition effects at play - Small countries (LUX, NOR) at the frontier unlikely to force convergence => using MFP level in the USA - modelling the frontier - MFP gap (coeff on frontier =1) - MFP frontier directly, if coeff<1, incomplete convergence - Year fixed effects => more flexible, capturing common trend in the data

46 Descriptive stats (2) cross-section variables Min Max Mean Sdev PMR & sub-components aggregate PMR indicator State control barriers to entrepreneurship barriers to trade and investment Doing business contract enforcement - cost contract enforcement - time insolvency - cost insolvency - time insolvency - recovery rate starting a business - cost starting a business - time Institutions rule of law legal system corruption (Dreher et al.) corruption perception index - TI control of corruption - WB WGI