2657 Views
1099 Downloads |
Public policy reforms and their impact on productivity, investment and employment: new evidence from OECD and non-OECD countries
Balázs Égert
Balázs Égert
Affiliation: OECD Economics Department, Paris, France; CESifo Munich Poschingerstrasse 5, Muenchen, Germany; EconomiX at the University of Paris X Nanterre, Nanterre, France
0000-0002-9540-4205
Correspondence
Balazs.Egert@oecd.org
Article | Year: 2022 | Pages: 179 - 205 | Volume: 46 | Issue: 2 Received: June 1, 2021 | Accepted: February 7, 2022 | Published online: June 1, 2022
|
FULL ARTICLE
FIGURES & DATA
REFERENCES
CROSSMARK POLICY
METRICS
LICENCING
PDF
|
Source
|
Country coverage
|
Time coverage
|
Product market regulation
|
Overall
|
OECD Product market regulation Indicators database
|
Around 60
|
Every five years, only one observation for about 15 countries
|
Barriers to entry
|
Barriers to trade & investment
|
Scope of state control
|
General business sector regulation
|
Business regulation
|
Fraser Institute
|
More than 100 countries
|
Annual, about 10 years
|
Cost of contract enforcement
|
World Bank Doing Business Indicators
|
More than 100 countries
|
Annual, about 10 years
|
Time of contract enforcement
|
Cost of insolvency procedures
|
Time of insolvency procedures
|
Cost of starting a business
|
Time of starting a business
|
Labour market regulation
|
EPL regular contracts
|
OECD
|
Around 60 countries, 10 countries different than for PMR
|
Annual, 30 years, only one observation for about 15 countries
|
Labour market regulation
|
Fraser Institute
|
More than 100 countries
|
Annual, about 10 years
|
EPL regular contracts
|
Cambridge
|
117 countries
|
Annual, 40 years
|
Institutions
|
Legal system
|
Fraser Institute
|
Around 100 countries
|
Annual, about 10 years
|
Legal system – enforcement
|
Legal system – judicial independence
|
Rule of law
|
WB's World Governance Indicators
|
Around 100 countries
|
|
Political stability
|
|
Corruption
|
|
Government effectiveness
|
|
Financial development
|
Financial liberalisation – EFW
|
Fraser Institute
|
Around 100 countries
|
Annual, until 2005
|
Domestic credit % GDP
|
World Bank's World Development Indicators database
|
Around 100 countries
|
Annual, about 30 years
|
Domestic private credit % GDP
|
Bank branches per capita
|
Stock market capitalisation % GDP
|
Stock market turnover % GDP
|
Trade openness
|
Openness
|
World Bank's World Development Indicators database
|
Around 100 countries
|
Annual, about 30 years
|
Log openness
|
Log openness – size adjusted
|
Own calculation based on WDI
|
Trade liberalisation – EFW
|
Fraser Institute
|
Around 100 countries
|
Annual, until 2005
|
Innovation intensity
|
R&D spending % GDP
|
World Bank's World Development Indicators database
|
Around 100 countries
|
Annual, about 30 years
|
Patents/capita
|
Source: Author.
Note: LCAP, on the vertical axis, denotes log per capita income (USD, constant PPP). On the horizontal axes are displayed the policies and institutions. For the rule of law, corruption and government effectiveness, higher numbers show a stronger rule of law, less corruption and a more effective government. START_COST, CONTRACT_COST and INSOLV_COST refer to the cost of starting a business, the time required for contract enforcement and insolvency procedures. REG_BUS and REG_LM_EFW are the EFW’s business regulation and labour market regulation indicators: higher values indicate more business-friendly regulation. EPL_CBR is the Cambridge Labour Regulation Indicator relating to regular contract: higher numbers indicate more stringent regulation.Source: Author.
Note: A_LCAP, on the vertical axis, denotes log per capita income (USD, constant PPP, country averages). On the horizontal axes are displayed the policies and institutions. For the rule of law, corruption and government effectiveness, higher numbers show a stronger rule of low, less corruption and a more effective government. For the OECD’s PMR indicator, its sub-components and the OECD and Cambridge EPL indicators, higher figures reflect more stringent regulation. For the EFW’s labour market regulation indicator, higher values indicate less stringent regulation. Source: Author
|
MFP
|
Capital deepening
|
Employment rate
|
Per capita income
|
Linear relationships
|
|
|
|
|
Within dimension
|
|
|
|
|
Institutions
|
Yes
|
No
|
Yes
|
Yes
|
Business regulation
|
Yes
|
No
|
No
|
No
|
Product market regulation
|
–
|
–
|
–
|
–
|
Labour market regulation
|
–
|
Yes
|
Yes
|
–
|
Financial system development
|
Yes
|
No
|
–
|
Yes
|
Between dimension
|
|
|
|
|
Institutions
|
Yes
|
No
|
Yes
|
Yes
|
Business regulation
|
?
|
No
|
No
|
No
|
Product market regulation
|
BTI
|
BTE, SSC
|
BTE, SSC
|
BTI
|
Labour market regulation
|
Yes??
|
No
|
Yes??
|
No
|
Financial system development
|
Yes
|
Yes
|
–
|
Yes
|
Non-linear relationships
|
conditional on
|
|
per capita income
|
Business regulation
|
Yes
|
No
|
Yes
|
Yes
|
Product market regulation
|
BTE,BTI,SSC
|
NO
|
BTE, SSC
|
BTE,BTI,SSC
|
Labour market regulation
|
No
|
No
|
Yes
|
No
|
|
institutions
|
Business regulation
|
Yes
|
No
|
Yes
|
Yes
|
Product market regulation
|
BTE,BTI,SSC
|
BTE, SSC
|
BTE, SSC
|
BTE,BTI,SSC
|
Labour market regulation
|
No
|
No
|
Yes
|
No
|
|
labour market regulations
|
Business regulation
|
No
|
No
|
No
|
No
|
Product market regulation
|
BTE,BTI,SSC
|
NO
|
BTE,SSC
|
BTE,BTI,SSC
|
Labour market regulation
|
No
|
No
|
No
|
No
|
Notes: Results on the linear relationship are split into two main parts: within dimension (coefficient estimates identified from the time variation in the data); and between dimension (coefficient estimates obtained on cross-sectional data). Non-linear relationships are estimated only on cross-section data (because no time series are available for PMR). The column ‘non-linear variables’ lists the variables, which take different coefficients, depending on the level of other variables. These ‘other variables’ are named in the rows “conditional on …” and are per capita income, institutions and labour market regulations. ‘YES’ implies a statistically significant relationship. ‘?’ implies that the estimated relationship is not very robust. ‘NO’ indicates the absence of a statistically significant relationship. ‘--’ indicates that the variable could not be included in the regressions. BTE, BTI and SSC indicate that there is a statistically significant relationship between the PMR sub-components barriers to entry (BTE), barriers to trade and investment (BTI) and the scope of state control (SSC) on the one hand and economic outcomes (MFP, capital deepening, the employment rate and per capita income) on the other hand.
Note: The ratio displayed above is the ratio between the standard deviation calculated on crosssection observations (averages for individual countries, the pure between effect) and the standard deviation of the series stripped of country means and common time trends (pure within effect). Source: Author’s calculations.
|
Impact through
|
Total impact
|
|
MFP
|
K/Y
|
L
|
Per capita income: aggregated from MFP,
K/Y and L
|
|
Policy measured as one standard
deviation
|
|
Within
|
Between
|
Within
|
Between
|
Within
|
Between
|
Within
|
Between
|
Institutions
|
Government effectiveness
|
7.4
|
50.0
|
|
|
0.8
|
5.2
|
8.2
|
55.2
|
Rule of law
|
5.0
|
42.9
|
|
|
0.5
|
4.5
|
5.5
|
47.4
|
Political stability
|
5.7
|
24.0
|
|
|
1.0
|
4.3
|
6.7
|
28.3
|
Corruption
|
5.9
|
39.8
|
|
|
0.9
|
6.0
|
6.8
|
45.8
|
Business regulation
|
Cost of starting a business
|
0.8
|
1.3
|
9.0
|
15.6
|
|
|
9.8
|
16.9
|
Cost of contract enforcement
|
1.4
|
13.5
|
|
|
|
|
1.4
|
13.5
|
Time of insolvency procedures
|
5.6
|
14.6
|
|
|
1.1
|
2.8
|
6.6
|
17.4
|
Product market regulation
|
PMR – overall
|
–
|
|
–
|
8.9
|
–
|
1.5
|
–
|
10.4
|
PMR – barriers to entry
|
–
|
17.3
|
–
|
5.2
|
–
|
2.0
|
–
|
24.5
|
PMR – barriers to trade & investment
|
–
|
8.3
|
–
|
|
–
|
|
–
|
8.3
|
PMR – scope of state control
|
–
|
|
–
|
6.4
|
–
|
4.1
|
–
|
10.5
|
Labour market regulation
|
EPL – OECD regular contracts
|
|
|
|
|
|
0.9
|
|
0.9
|
EPL – Cambridge indicator
|
|
|
|
|
0.8
|
3.1
|
0.8
|
3.1
|
Labour market regulation (EFW)
|
|
|
2.1
|
5.5
|
0.8
|
2.0
|
2.9
|
7.5
|
Financial development
|
Banking sector
|
4.9
|
12.4
|
4.2
|
10.7
|
|
|
9.1
|
23.0
|
Financial markets
|
8.1
|
17.2
|
|
|
|
|
8.1
|
17.2
|
Note: MFP, K/Y and L indicate by how much per capita income would increase due to policy changes affecting the three supply-side channels. The change in the indicators is defined as one standard deviation in the data. Columns named ‘within’ show that the change in the policies are based on the within dimension (variation over time). Columns named ‘between’ show that the changes in the policies are obtained from the between (cross-section) dimension. The effects are calculated following the methodology set out in box 1 in Égert and Gal (2016). Empty cells indicate the absence of robust empirical relationships. Cells filled with “--“ indicate that regression analysis was not possible for the particular variable and dimension (PMR indicator over time). The coefficient estimates used to calculate the effect are the average of the minimum and maximum coefficient estimates. Table C11 summarises from which particular regressions the coefficient estimates are used. Source: Author’s calculations.
|
If per
capita income is
|
If rule of
law is
|
If OECD's
EPL on regular contracts is
|
|
Below
|
Above
|
Below
|
Above
|
Below
|
Above
|
|
The estimated
threshold
|
The estimated
threshold
|
The estimated
threshold
|
Effects on MFP of
|
|
|
|
|
|
|
PMR – overall
|
40.4
|
17.4
|
28.2
|
12.6
|
30.4
|
25.3
|
PMR – barriers to entry
|
24.5
|
1.5
|
19.4
|
2.8
|
19.4
|
13.0
|
PMR – barriers to trade & investment
|
53.1
|
15.8
|
35.5
|
11.0
|
27.7
|
41.0
|
PMR – scope of state controll
|
27.1
|
5.3
|
18.1
|
2.8
|
16.9
|
11.0
|
Note: Numbers in bold indicate that the calculations are based on coefficient estimates that were statistically not significant at the conventional level of 10%. Source: Author’s calculations
Variables
|
Min
|
Max
|
Mean
|
St. dev.
|
|
Worldwide sample
|
Per capita income
|
5.29
|
11.62
|
8.75
|
1.31
|
Openness
|
0.31
|
449.99
|
91.94
|
51.78
|
Log Openness
|
-1.18
|
6.11
|
4.40
|
0.52
|
Log Openness (size adjusted)
|
-4.93
|
1.99
|
0.24
|
0.49
|
R&D spending %GDP
|
0.01
|
4.52
|
1.00
|
0.99
|
Rule of law
|
-2.67
|
2.00
|
0.00
|
0.99
|
Cost of starting a business
|
0.00
|
1,540.00
|
67.00
|
143.00
|
Time of starting a business
|
1.00
|
687.00
|
41.00
|
59.00
|
Labour market regulation – EFW
|
2.34
|
9.73
|
11475.00
|
1.49
|
Business regulation – EFW
|
2.86
|
8.89
|
6.01
|
1.04
|
Finance – bank branches per capita
|
0.13
|
237.07
|
19.38
|
23.96
|
Stock market capitalisation % GDP
|
0.04
|
606.00
|
54.91
|
62.54
|
|
OECD sample
|
Openness
|
5.73
|
371.44
|
68.15
|
43.87
|
Log Openness
|
1.75
|
5.92
|
4.04
|
0.61
|
Log Openness (size adjusted)
|
-2.12
|
1.32
|
0.03
|
0.49
|
Business spending on R&D % GDP – OECD
|
0.01
|
3.76
|
1.05
|
0.73
|
General spending on R&D % GDP – OECD
|
0.15
|
4.48
|
1.68
|
0.88
|
General spending on basic R&D % GDP – OECD
|
0.05
|
0.90
|
0.31
|
0.16
|
ETCR – overall
|
0.79
|
6.00
|
4.08
|
1.47
|
ETCR – entry barriers
|
0.43
|
6.00
|
3.77
|
1.84
|
ETCR – public ownership
|
0.83
|
6.00
|
4.29
|
1.43
|
EPL regular contracts
|
0.26
|
5.00
|
2.18
|
0.83
|
ALMP spending
|
0.45
|
22.00
|
22.00
|
21.53
|
Variables
|
Worldwide sample
|
OECD sample
|
|
Min
|
Max
|
Mean
|
St. dev.
|
Min
|
Max
|
Mean
|
St. dev.
|
Institutions
|
Legal system
|
2.23
|
8.93
|
5.53
|
1.61
|
4.86
|
8.54
|
7,27
|
1,07
|
Legal system – enforcement
|
0.00
|
8.11
|
4.46
|
1.75
|
3.22
|
8.11
|
5,75
|
1,29
|
Legal system – judicial independence
|
0.60
|
9.15
|
4.80
|
2.15
|
3.85
|
9.17
|
7,06
|
1,74
|
Rule of law
|
-2.38
|
1.94
|
0.01
|
0.99
|
-0.53
|
1.94
|
1,27
|
0,60
|
Civil liberties
|
1.00
|
7.00
|
4.78
|
1.81
|
4.02
|
7.00
|
6,29
|
0,86
|
Polity2 – political stability
|
-10.00
|
10.00
|
3.78
|
6.17
|
27.00
|
10.00
|
8,01
|
3,17
|
Corruption
|
-1.71
|
2.45
|
0.01
|
0.98
|
-0.31
|
2.44
|
1,33
|
0,80
|
Government effectiveness
|
-2.18
|
2.18
|
0.00
|
0.99
|
0.16
|
2.14
|
1,36
|
0,56
|
Business
environment
|
Cost of contract enforcement
|
8.00
|
163.00
|
35.00
|
26.00
|
8.00
|
39.00
|
21,00
|
8,00
|
Time of contract enforcement
|
133.00
|
1,715.00
|
628.00
|
305.00
|
216.00
|
1,332.00
|
517,00
|
260,00
|
Cost of insolvency procedures
|
1
|
76.00
|
17.00
|
12
|
1.00
|
23.00
|
10,00
|
6,00
|
Time of insolvency procedures
|
0.40
|
6.97
|
2.75
|
1.21
|
0.40
|
5.84
|
1,92
|
1,16
|
Cost of starting a business
|
0.00
|
991.49
|
66.32
|
121.4
|
0.05
|
20.69
|
6,80
|
6,42
|
Time of starting a business
|
2.78
|
690.71
|
41.86
|
60.34
|
2.71
|
61.08
|
16,83
|
11,68
|
OECD Product Market Regulation Indicator
|
PMR – overall
|
|
|
|
|
1.18
|
2.8
|
1,73
|
0,35
|
PMR – barriers to entry
|
|
|
|
|
1.49
|
3.07
|
2,06
|
0,37
|
PMR – barriers to trade & investment
|
|
|
|
|
0.20
|
2.09
|
0,74
|
0,41
|
PMR – scope of state control
|
|
|
|
|
1.51
|
3.92
|
2,41
|
0,54
|
Table 1Overview of indicators used in the regression analysis by main policy and outcome areas DISPLAY Table
Figure 1Stylised facts – per capita income, regulation and institutions, annual data DISPLAY Figure
Figure 2Stylised facts – per capita income, regulation and institutions, cross-section data (country averages) DISPLAY Figure
Table 2Summary of estimation results DISPLAY Table
Figure 3The ratio of standard deviation of the pure cross-section to standard deviation over time DISPLAY Figure
Table 3AQuantification results – linear regressions, per capita effects due to the three supply-side channels (in percent) DISPLAY Table
Table 3BQuantification – non-linear regressions (in percent) DISPLAY Table
Table A1Descriptive statistics: time varying variables DISPLAY Table
Table A2Descriptive statistics: time-invariant variables (period averages) DISPLAY Table
* The views expressed in this paper do not necessarily reflect the views of the OECD or any other institution the authors is affiliated with. The author would like to thank to two anonymous referees for their very useful and constructive comments and suggestions. This paper is a considerably revised version of the paper Égert (2018a).
1 Another challenge, mentioned earlier and difficult to tackle here is the widespread informality and the larger difference between de jure and de facto measures of indicators in less-developed countries.
2 It would be interesting to use the sub-indicators. Nevertheless, they are strongly correlated with each other both along the within (variation over time) and between (cross-country variation) dimensions. Hence, they could not be included in the regressions at the same time.
3 SPIDER is a compilation of data from 43 existing data sources. It draws heavily on a large number of existing OECD databases. It includes a number of non-OECD databases such as the World Bank’s Doing Business and World Development Indicators databases of the Penn World Table 8.0. The final source of data in SPIDER is individual research papers, either academically published articles or working papers (for more details, see Égert, Gal and Wanner, 2017).
4 The full set including countries for which a small combination of variables is available comprises 149 countries. The ISO codes of the countries are given as follows: ago alb are arg arm aus aut aze bdi bel ben bfa bgd bgr bhr bhs bih blz bol bra brb brn bwa caf can che chl chn civ cmr cog col cpv cri cyp cze deu dnk dom dza ecu egy esp est eth fin fji fra gab gbr geo gha gmb gnb grc gtm guy hkg hnd hrv hti hun idn ind irl irn isl isr ita jam jor jpn kaz ken kgz khm kor kwt lbn lka lso ltu lux lva mar mda mdg mex mkd mli mlt mmr mne mng moz mrt mus mwi mys nam ner nga nic nld nor npl nzl omn pak pan per phl png pol prt pry qat rus rwa sau sen sgp sle slv srb sur svk svn swe swz syr tcd tgo tha tjk tto tun tur tza uga ukr ury usa ven vnm yem zaf zmb zwe
5 But the scatterplots shown in figures 2 and 3 and in the annex A reported in Égert ( 2018a) do not reveal any apparent link between the two other indicators and economic outcomes.
6 Some of the explanatory variables used in the analysis are strongly correlated with each other. To avoid the problem of multi-collinearity in the regressions, the variables are grouped in the regressions so that strongly correlated variables are not used at the same time. The correlation analysis indicates no major problem of correlation for the variables once country and time fixed effects are purged from the data (for the country/time panel regressions). However, there is clearly a problem of correlation for the cross-section dimension. The institutional variables are strongly correlated with one another but also with the OECD’s PME indicator and sub-components, and the EFW business regulation index. The three labour market regulation indicators are also correlated with each other. There is also a strong correlation between various measures of trade openness. The two measures of innovation intensity also exhibit a high correlation coefficient. Furthermore, R&D spending as a % of GDP is correlated with other covariates as well. Against this background, only variables will be included in the same regression, which are not correlated with each other.
7 Further analysis would be needed to confirm this result.
8 Business sector regulation refers to the World Bank’s Doing Business indicators. Product market regulation indicators refer to the OECD’s PMR indicator.
9 It could be argued that more restrictive labour market regulation would lead to a greater capital deepening as businesses would reduce labour intensity. Empirical results are mixed on this effect. Égert (2018b) provides an overview of the empirical literature on this issue and reports results, using country-level data for OECD countries, according to which more stringent labour market regulation reduces capita deepening.
10 For MFP, the non-linear regressions contain the following linear control variables: human capital, openness, innovation intensity (patents per capita) and financial development (banking sector and stock markets). PMR, labour market regulations and institutions were included if these variables were not the non-linear variables in the regressions.
11 Table C10 in annex C in Égert ( 2018a) provides descriptive statistics of the threshold variables.
12 We also experimented by imposing per capita income threshold of 5,000 and 10,000 USD. Coefficient estimates
are less precisely estimated in these cases (suggesting that it is better to estimate the thresholds rather
than to impose them).
13 Regression were also run to see whether the coefficient estimates on trade openness, innovation intensity and
human capital differ as a function of per capita income levels. Results indicate, especially when only these
three variables are used as explanatory variables, that openness starts to have a positive coefficient if per capita
income is higher than USD 10,000 for time series panel regressions and above USD 6,000 for cross-section
regressions. Similarly, the coefficient estimate on human capital is more positive above comparable thresholds.
No non-linear effect can be identified for innovation intensity.
|
|
June, 2022 II/2022 |