The European Company Survey (ECS) has substantial information on training and learning decisions made by firms and their background characteristics, such as sector, size, age, hierarchical structure, product market strategy, technology adoption and work practices. This rich dataset allows a detailed analysis of the relationship between firms’ characteristics and their training offer. To this end, the OECD has conducted regression analysis for different training and learning outcomes. Boxes throughout the chapters present summaries of the main findings. This annex provides a technical description of the methodology and the results.
Training in Enterprises
New Evidence from 100 Case Studies
Report
Getting Skills Right
Table of contents
Share
Facebook
Twitter
LinkedIn
Annex B. Econometric analysis
Data
Overview of methodology
The sample used for the econometric analyses includes enterprises from all countries covered in the ECS microdata (the EU-27 plus the UK), which have more than 50 employees and belong to the manufacturing and services sectors. The ECS has several categorical variables, which require managers to select one option among several categories, for instance:
In 2018, how many employees in this establishment have received on-the‑job training or other forms of direct instruction in the workplace from more experienced colleagues?
☐ None at all
☐ Less than 20%
☐ 20% to 39%
☐ 40% to 59%
☐ 60% to 79%
☐ 80% to 99%
☐ All
A binary outcome variable was derived by merging multiple categories, using the sample average as the cut-off point. For example, for the question above, a dummy variable was created by assigning a value 1 if the company reported that more than 40% of employees had received on-the‑job or direct instruction in the workplace. For each binary outcome variable, the OECD estimated a probit model. The list of independent variables includes dummy variables for country, number of employees, age, hierarchy levels, type of market strategy, adoption of several High Performance Workplace Practices (HPWPs), change in employment, profitability, share of permanent contracts and technological change. Standard errors are clustered by country and the industry-size strata for the sample. The OECD also explored the possibility of implementing a multinomial logit approach, but this was ultimately discarded, due to sample size considerations and difficulties in interpreting coefficients.
After estimating the probit model, average marginal effects were computed. Given that all variables are categorical, each marginal effect shows the change in probability of experiencing a particular outcome (e.g. having a comparatively high share of employees participating in training) for a category of enterprises (e.g. financial services sector enterprises) compared to the omitted category (e.g. manufacturing firms). The results should not be interpreted causally, but show how a particular firm characteristic is correlated with a training or learning outcome, while controlling for other factors.
The sections below present the results discussed across different chapters.
Results related to Chapter 2
Table A B.1. Enterprise characteristics and the mode of delivery (Box 2.3)
|
Dimension |
Independent variable |
Dependent variable: High training propensity in courses |
Dependent variable: High training propensity on-the‑job |
|---|---|---|---|
|
Sector (omitted: Manufacturing) |
Wholesale & Retail Trade |
0.120*** (0.0329) |
0.0517** (0.0247) |
|
Transportation & Storage |
0.107** (0.0436) |
0.0332 (0.0370) |
|
|
Accommodation & Food |
0.0140 (0.0443) |
0.183*** (0.0586) |
|
|
Information & Communication |
‑0.0241 (0.0528) |
0.100* (0.0514) |
|
|
Financial & Insurance |
0.328*** (0.0452) |
0.139 (0.0847) |
|
|
Real Estate |
0.198 (0.147) |
0.00928 (0.115) |
|
|
Professional Services |
0.139*** (0.0398) |
0.109** (0.0518) |
|
|
Administrative & Support |
‑0.0226 (0.0743) |
0.0620 (0.0495) |
|
|
Arts & Recreation |
‑0.113* (0.0624) |
0.0535 (0.0749) |
|
|
Other Services |
0.103*** (0.0391) |
0.140*** (0.0418) |
|
|
Size (omitted: 250+ employees) |
50 to 249 employees |
‑0.0462* (0.0264) |
‑0.0524*** (0.0183) |
|
Age (omitted: 30+ years) |
10 years or less |
‑0.0165 (0.0455) |
0.0984*** (0.0240) |
|
11 to 20 years |
0.0297 (0.0208) |
0.101*** (0.0278) |
|
|
21 to 30 years |
0.0291 (0.0223) |
0.0263 (0.0167) |
|
|
Levels of hierarchy (omitted: 3+ levels) |
1 or 2 levels |
‑0.140** (0.0650) |
‑0.0114 (0.0407) |
|
3 levels |
‑0.0289 (0.0250) |
0.0234 (0.0149) |
|
|
Product market strategy (omitted: lower prices) |
Customisation |
0.0530** (0.0244) |
0.0351 (0.0329) |
|
Introducing new products/services |
0.0728 (0.0473) |
0.0155 (0.0543) |
|
|
Better quality |
0.0530** (0.0244) |
0.0305 (0.0426) |
|
|
Technology adoption (omitted: technology not adopted) |
Uses robots |
0.00807 (0.0226) |
0.0820*** (0.0198) |
|
Purchased customised software |
0.0414** (0.0173) |
0.0370 (0.0322) |
|
|
Use data analytics |
0.0648*** (0.0244) |
0.105*** (0.0262) |
|
|
HPWPs (omitted: practice not adopted) |
Workforce autonomy |
0.0536*** (0.0197) |
0.0534** (0.0221) |
|
Performance pay |
0.0735** (0.0301) |
0.0432 (0.0335) |
|
|
Teamwork |
0.0322** (0.0161) |
0.0558*** (0.0204) |
|
|
Observations |
5,701 |
5,683 |
|
Note: The table reports average marginal effects from a probit model. *** reports marginal effects for which p<0.01, ** when p<0.05, and * when p<0.1. The model is estimated separately for training sessions and on-the‑job training. The dependent variable is equal to 1 if at least 40% of employees receive training sessions or on-the‑job training. Clustered standard errors between parentheses.
Source: OECD calculations, using microdata from ECS 2019.
Table A B.2. Enterprise characteristics and intensity of problem solving at work (Box 2.5)
|
Dimension |
Independentent variable |
Dependent variable: High problem solving intensity at work |
|---|---|---|
|
Sector (omitted: Manufacturing) |
Wholesale & Retail Trade |
‑0.0565** (0.0219) |
|
Transportation & Storage |
‑0.0335 (0.0361) |
|
|
Accommodation & Food |
‑0.0506 (0.0494) |
|
|
Information & Communication |
0.189*** (0.0291) |
|
|
Financial & Insurance |
0.0600 (0.0515) |
|
|
Real Estate |
‑0.0416 (0.103) |
|
|
Professional Services |
0.278*** (0.0259) |
|
|
Administrative & Support |
‑0.0303 (0.0834) |
|
|
Arts & Recreation |
‑0.0753 (0.0971) |
|
|
Other Services |
0.0532 (0.0371) |
|
|
Size (omitted: 250+ employees) |
50 to 249 employees |
0.0193 (0.0155) |
|
Age (omitted: 30+ years) |
10 years or less |
0.00551 (0.0287) |
|
11 to 20 years |
0.0610* (0.0348) |
|
|
21 to 30 years |
0.0120 (0.0179) |
|
|
Levels of hierarchy (omitted: 3+ levels) |
1 or 2 levels |
0.0433 (0.0432) |
|
3 levels |
‑0.00862 (0.0160) |
|
|
Product market strategy (omitted: lower prices) |
Customisation |
0.0371** (0.0174) |
|
Introducing new products/services |
0.0841** (0.0331) |
|
|
Better quality |
0.00287 (0.0272) |
|
|
Technology adoption (omitted: technology not adopted) |
Uses robots |
0.00903 (0.0236) |
|
Purchased customised software |
0.0571*** (0.0197) |
|
|
Use data analytics |
0.0529** (0.0210) |
|
|
HPWPs (omitted: practice not adopted) |
Workforce autonomy |
0.109*** (0.0234) |
|
Performance pay |
0.145*** (0.0350) |
|
|
Teamwork |
0.0612** (0.0244) |
|
|
Observations |
5,701 |
|
Note: The table reports average marginal effects from a probit model. *** reports marginal effects for which p<0.01, ** when p<0.05, and * when p<0.1. The dependent variable is equal to 1 if at least 20% of employees are in jobs that require finding solutions to unfamiliar problems. Clustered standard errors between parentheses.
Source: OECD calculations, using microdata from ECS 2019.
Results related to Chapter 3
Table A B.3. Enterprise characteristics and main reasons for providing training (Box 3.1)
|
Dimension |
Independent variable |
Dependent variable: Flexibility |
Dependent variable: Motivation |
Dependent variable: Innovation |
Dependent variable: Skills |
|---|---|---|---|---|---|
|
Sector (omitted: Manufacturing) |
Wholesale & Retail Trade |
‑0.0322 (0.0223) |
0.0541*** (0.0185) |
0.00440 (0.0219) |
‑0.00833 (0.00628) |
|
Transportation & Storage |
‑0.0298 (0.0364) |
‑0.0190 (0.0191) |
‑0.0108 (0.0182) |
0.00228 (0.00655) |
|
|
Accommodation & Food |
0.0623 (0.0467) |
0.0799*** (0.0228) |
0.0701* (0.0355) |
‑0.00917 (0.0158) |
|
|
Information & Communication |
‑0.163*** (0.0444) |
‑0.142* (0.0727) |
0.00992 (0.0427) |
‑0.0583*** (0.0217) |
|
|
Financial & Insurance |
‑0.134* (0.0708) |
‑0.000641 (0.0353) |
0.0298 (0.0525) |
0.0117 (0.0104) |
|
|
Real Estate |
‑0.0112 (0.0774) |
0.0641 (0.0498) |
‑0.104 (0.139) |
0.000990 (0.0215) |
|
|
Professional Services |
‑0.0292 (0.0410) |
0.0283 (0.0236) |
‑0.0250 (0.0288) |
0.00404 (0.0117) |
|
|
Administrative & Support |
0.0236 (0.0414) |
‑0.0403 (0.0250) |
0.0101 (0.0400) |
0.00317 (0.00695) |
|
|
Arts & Recreation |
0.0327 (0.0589) |
‑0.0129 (0.0339) |
‑0.0303 (0.0656) |
‑0.00444 (0.0153) |
|
|
Other Services |
‑0.0723** (0.0314) |
‑0.00797 (0.0300) |
‑0.00563 (0.0238) |
‑0.00742 (0.0108) |
|
|
Size (omitted: 250+ employees) |
50 to 249 employees |
‑0.0683*** (0.0236) |
‑0.00614 (0.0128) |
0.0122 (0.0139) |
‑0.00694 (0.00590) |
|
Age (omitted: 30+ years) |
10 years or less |
0.0123 (0.0500) |
‑0.0214 (0.0533) |
0.0278 (0.0217) |
‑0.0166 (0.0219) |
|
11 to 20 years |
‑0.00109 (0.0218) |
0.0322*** (0.0101) |
0.00845 (0.0196) |
‑0.00175 (0.00930) |
|
|
21 to 30 years |
0.0165 (0.0293) |
0.0410** (0.0187) |
0.0182 (0.0150) |
0.00557 (0.00847) |
|
|
Levels of hierarchy (omitted: 3+ levels) |
1 or 2 levels |
0.0173 (0.0729) |
‑0.0116 (0.0438) |
‑0.00788 (0.0300) |
‑0.0262* (0.0142) |
|
3 levels |
0.0215 (0.0216) |
‑0.00805 (0.0112) |
0.0302* (0.0169) |
‑0.00629 (0.00659) |
|
|
Product market strategy (omitted: lower prices) |
Customisation |
0.0178 (0.0357) |
0.0677** (0.0314) |
0.0942* (0.0492) |
0.00644 (0.0121) |
|
Introducing new products/services |
0.0310 (0.0409) |
0.0483 (0.0328) |
0.106*** (0.0404) |
0.00398 (0.0148) |
|
|
Better quality |
‑0.0108 (0.0187) |
0.0575** (0.0221) |
0.0638* (0.0356) |
0.00596 (0.0118) |
|
|
Technology adoption (omitted: technology not adopted) |
Uses robots |
0.0146 (0.0227) |
‑0.0138 (0.0217) |
0.0214 (0.0176) |
‑0.00433 (0.00938) |
|
Purchased customised software |
0.0313* (0.0172) |
0.00548 (0.0142) |
0.0105 (0.0117) |
0.00234 (0.0101) |
|
|
Use data analytics |
0.0576* (0.0332) |
0.00622 (0.0333) |
0.0258 (0.0202) |
0.0188** (0.00739) |
|
|
HPWPs (ommmitted: practice not adopted) |
Workforce autonomy |
0.0157 (0.0157) |
0.0544*** (0.0197) |
0.0647*** (0.0169) |
0.0221** (0.00985) |
|
Teamwork |
‑0.0509** (0.0244) |
0.0296 (0.0309) |
0.0523* (0.0282) |
0.0125 (0.0149) |
|
|
Performance pay |
0.0795*** (0.0240) |
0.0362*** (0.0120) |
0.0410** (0.0161) |
0.0202*** (0.00641) |
|
|
Observations |
5,679 |
5,681 |
5,676 |
5,319 |
|
Note: The table reports average marginal effects from a probit model. *** reports marginal effects for which p<0.01, ** when p<0.05, and * when p<0.1. The variable Flexibility is equal to 1 if the enterprise declares that ‘allowing employees to acquire skills they need to do other jobs’ is an important factor motivating training provision. The variable Motivation is equal to 1 if the enterprise declares that ‘improving employee morale’ is an important factor motivating training provision. The variable Innovation is equal to 1 if the enterprise declares that ‘increasing the capacity of employees to articulate ideas about improvement’ is an important factor motivating training provision. The variable Skills is equal to 1 if the enterprise declares that ‘ensuring that employees have the skills they need to do their current job’ is an important factor motivating training provision. The model is estimated separately for the four different dependent variables. Clustered standard errors between parentheses.
Source: OECD calculations, using microdata from ECS 2019.
Table A B.4. Enterprise characteristics and time constraints (Box 3.2)
|
Dimension |
Independentent variable |
Dependent variable: Workload is an obstacle to training |
|---|---|---|
|
Sector (omitted: Manufacturing) |
Wholesale & Retail Trade |
0.00831 (0.0254) |
|
Transportation & Storage |
‑0.00143 (0.0209) |
|
|
Accommodation & Food |
‑0.0131 (0.0268) |
|
|
Information & Communication |
‑0.0133 (0.0225) |
|
|
Financial & Insurance |
0.138 (0.125) |
|
|
Real Estate |
0.0498* (0.0280) |
|
|
Professional Services |
‑0.112* (0.0655) |
|
|
Administrative & Support |
‑0.119** (0.0495) |
|
|
Arts & Recreation |
‑0.0914** (0.0359) |
|
|
Other Services |
‑0.00805 (0.0231) |
|
|
Size (omitted: 250+ employees) |
50 to 249 employees |
‑0.0220 (0.0317) |
|
Age (omitted: 30+ years) |
10 years or less |
‑0.0448** (0.0217) |
|
11 to 20 years |
‑0.0880*** (0.0191) |
|
|
21 to 30 years |
0.0195 (0.0245) |
|
|
Levels of hierarchy (omitted: 3+ levels) |
1 or 2 levels |
0.00434 (0.0209) |
|
3 levels |
0.00831 (0.0254) |
|
|
Product market strategy (omitted: lower prices) |
Customisation |
‑0.00143 (0.0209) |
|
Introducing new products/services |
‑0.0131 (0.0268) |
|
|
Better quality |
‑0.0133 (0.0225) |
|
|
Technology adoption (omitted: technology not adopted) |
Uses robots |
0.138 (0.125) |
|
Purchased customised software |
0.0498* (0.0280) |
|
|
Use data analytics |
‑0.112* (0.0655) |
|
|
HPWPs (ommmitted: practice not adopted) |
Workforce autonomy |
‑0.119** (0.0495) |
|
Performance pay |
‑0.0914** (0.0359) |
|
|
Teamwork |
‑0.00805 (0.0231) |
|
|
Observations |
5,659 |
|
Note: The table reports average marginal effects from a probit model. *** reports marginal effects for which p<0.01, ** when p<0.05, and * when p<0.1. The dependent variable is equal to 1 if the enterprise declares that participation in training and professional development activities is only possible if workload and work schedules allow for it, to 0 otherwise. Clustered standard errors between parentheses.
Source: OECD calculations, using microdata from ECS 2019.
Results related to Chapter 4
Table A B.5. Enterprise characteristics and direct employee involvement in decision-making on training (Box 4.1)
|
Dimension |
Independentent variable |
Dependent variable: Direct employee involvement in decision-making on training |
|---|---|---|
|
Sector (omitted: Manufacturing) |
Wholesale & Retail Trade |
‑0.0517* (0.0303) |
|
Transportation & Storage |
‑0.0284 (0.0423) |
|
|
Accommodation & Food |
0.0589 (0.0394) |
|
|
Information & Communication |
0.117*** (0.0284) |
|
|
Financial & Insurance |
0.0634 (0.0747) |
|
|
Real Estate |
0.139 (0.104) |
|
|
Professional Services |
0.0608* (0.0326) |
|
|
Administrative & Support |
0.0826 (0.0679) |
|
|
Arts & Recreation |
0.0658 (0.109) |
|
|
Other Services |
0.0459 (0.0343) |
|
|
Size (omitted: 250+ employees) |
50 to 249 employees |
0.0180 (0.0164) |
|
Age (omitted: 30+ years) |
10 years or less |
‑0.0123 (0.0382) |
|
11 to 20 years |
0.0810*** (0.0296) |
|
|
21 to 30 years |
0.0787** (0.0344) |
|
|
Levels of hierarchy (omitted: 3+ levels) |
1 or 2 levels |
‑0.0554 (0.0334) |
|
3 levels |
‑0.0372** (0.0152) |
|
|
Product market strategy (omitted: lower prices) |
Customisation |
0.0874* (0.0481) |
|
Introducing new products/services |
0.0119 (0.0433) |
|
|
Better quality |
0.0394 (0.0518) |
|
|
Technology adoption (omitted: technology not adopted) |
Uses robots |
0.00232 (0.0358) |
|
Purchased customised software |
0.0442** (0.0186) |
|
|
Use data analytics |
0.0883*** (0.0213) |
|
|
HPWPs (omitted: practice not adopted) |
Workforce autonomy |
0.106*** (0.0161) |
|
Performance pay |
0.0492 (0.0358) |
|
|
Teamwork |
0.0592*** (0.0219) |
|
|
Observations |
5,509 |
|
Note: Note: The table reports average marginal effects from a probit model. *** reports marginal effects for which p<0.01, ** when p<0.05, and * when p<0.1. Direct influence defined as ‘great or moderate extent’ as response to answer to the question ‘Please think of the period since the beginning of 2016. In your opinion, to what extent have employees directly influenced management decisions in the following areas?. Clustered standard errors between parentheses.
Source: OECD calculations, using microdata from ECS 2019.