Measuring Labour Regulation: Worker protections around the world

6 December 2023

Summary

  • There are long running debates about the economic and social impacts of legislation that aims to improve worker protections. These debates pose questions such as: Do labour laws cause unemployment? Is their impact on productivity positive or negative? Do they materially improve outcomes for workers?
  • Our research, using a statistical measure of labour regulation, can help policy makers to address these questions and better understand the effects of labour laws on employment, productivity and equality.
  • The Centre for Business Research Labour Regulation Index (CBR-LRI), recently updated, provides a measure of the strength of worker protections in law, covering nearly 250,000 observations on labour laws in over 100 countries, over 50 years between 1970-2022.
  • Two key trends emerge in the 2023 update. First, dismissal laws in many countries strengthened in response to the Covid-19 emergency, as governments sought to stabilise employment during the pandemic.
  • Second is the normalisation of platform work, which in many countries is increasingly being assimilated to the rules governing regular employment.
  • Data from the updated CBR-LRI can aid assessment of policy options concerning the legal governance of platform work and of employment relations more generally, at a time when many countries are considering reforms to their labour law systems.

1 Why measure the strength of labour laws?

Laws governing work (‘labour laws’) are to be found in every country. However, until recently we have lacked good measures of the strength of the protection they offer workers. Without such measures it is difficult to get a clear picture of how labour laws are changing over time and how far they differ across countries. Data are not automatically generated for making these historical and geographic comparisons: they must be constructed.

Designing a statistical measure of worker protection is no easy task, but it is one with the potential to throw light on questions which are high on the policy agenda. Do labour laws, as is sometimes argued, cause unemployment? Is their impact on productivity positive or negative? Do they materially improve outcomes for workers?

These questions are more pressing than ever in a period of rapid technological change, with platform and gig work changing the balance of power in the labour market, and putting into question the usefulness of laws built around the so-called ‘standard employment relationship’ of full-time, permanent and regular work.

2 How the Labour Regulation Index was constructed

The task of constructing a global index of labour laws began at the Centre for Business Research in Cambridge University around a decade ago. Similar indices already existed but only for some of the laws we were interested in exploring. The OECD’s Employment Protection Indicators (most recently updated in 2020: see OECD, 2023) covered laws on dismissal but omitted any reference to other areas of interest including the regulation of working time, collective bargaining and codetermination laws, and the law governing industrial action. The World Bank’s Doing Business Reports had a wider remit and an extensive year and country coverage, but did not code for laws before the late 2000s. We wanted to create historical data going back to the 1970s. Without a long time series it is difficult to come to a clear view of the effects of the legal regulation of work. How labour laws are responding now to new forms of work is influenced not just by current conditions but by modes of regulation inherited from earlier periods.

The index we created, known as the Centre for Business Research Labour Regulation Index (‘CBR-LRI’), has recently been updated (Adams et al., 2023), thanks to the support of the ESRC through the Digital Futures at Work Research Centre. The Index now covers 1970 through to the end of 2022, and a total of 117 countries have been coded.

The methodology we employ builds on that set out in the Handbook on Constructing Composite Indicators published under the auspices of the OECD and European Commission in 2008. The key idea is that of ‘construct validity’, which involves a series of elements (see Illustration 1).

A statistical ‘construct’ is a measure of the underlying reality (called a ‘concept’) which the index is seeking to measure. The concept in our case is ‘labour regulation’ meaning the regulatory content of a law: how far does it protect workers or, conversely, bind employers? Labour laws differ, for example, in the limits they set to daily or weekly working time, the strictness of dismissal protection, the scope of the right to strike, and so on. It is important to know not just that a law exists but how protective it is.

  1. Concept
  2. Construct
  3. Indicators
  4. Coding protocols
  5. Measurement scale
  6. Weights
  7. Aggregation

The ‘construct’ contains several elements. Firstly, there are the individual indicators or variables which express different aspects of the construct numerically. We devised 40 such indicators, grouped into five sub-indices representing particular areas of labour regulation: laws on how far different forms of employment (including self-employment, part-time work, fixed-term employment and temporary agency work) are assimilated to the ‘core’ employment contract; laws on working time; laws governing unjust or unfair dismissal; laws on collective employee representation; and laws on strikes and other forms of industrial action. The summary indicators, and the sub-indices into which they are grouped, are set out below.

2.1 CBR-LRI sub-indices and indicators
  1. The law, as opposed to the contracting parties, determines the legal status of the worker
  2. Part-time workers have the right to equal treatment with full-time workers
  3. Part time workers have proportionately equal dismissal rights to those of full-time workers
  4. Fixed-term contracts are allowed only for work of limited duration
  5. Fixed-term workers have the right to equal treatment with permanent workers
  6. Maximum duration of fixed-term contracts
  7. Agency work is prohibited or strictly controlled
  8. Agency workers have the right to equal treatment with permanent workers of the user undertaking
  1. Annual leave entitlements
  2. Public holiday entitlements
  3. Overtime premia
  4. Weekend working
  5. Limits to overtime working
  6. Duration of the normal working week
  7. Maximum daily working time
  1. Legally mandated notice period (all dismissals)
  2. Legally mandated redundancy compensation
  3. Minimum qualifying period of service for normal case of unjust dismissal
  4. Law imposes procedural constraints on dismissal
  5. Law imposes substantive constraints on dismissal
  6. Reinstatement normal remedy for unfair dismissal
  7. Notification of dismissal
  8. Redundancy selection
  9. Priority in re-employment
  1. Right to unionisation
  2. Right to collective bargaining
  3. Duty to bargain
  4. Extension of collective agreements
  5. Closed shops
  6. Codetermination: board membership
  7. Codetermination and information/consultation of workers
  1. Unofficial industrial action
  2. Political industrial action
  3. Secondary industrial action
  4. Lockouts
  5. Right to industrial action
  6. Waiting period prior to industrial action
  7. Peace obligation
  8. Compulsory conciliation or arbitration
  9. Replacement of striking workers

In choosing these particular indicators and grouping them into the five sub-indices we used our judgment as to what were the most important elements of contemporary systems of the laws relating to work. In terms of subject-matter, the sub-indices overlap closely with the content of ILO conventions and recommendations. The individual indicators mostly cover the same issues as those developed by the World Bank, although ours are somewhat more granular. Because of time and resource constraints we do not cover every aspect of labour laws; some selectivity is unavoidable. The methodology we developed could be applied to code these other areas, which include health and safety law and discrimination law, in future.

2.2 Methodology

Having identified the indicators, we then apply a coding protocol (in effect, a verbal algorithm) which instructs the coder on how to assign values to laws of different types in order to arrive at a score for a given indicator. These are protocols we developed ourselves at the outset of the project. Initially applied to a small sample of countries, they have proved to be a good fit for the variety of labour laws we have observed in the course of constructing the dataset. The current 117 countries for which we can access relevant primary sources between them represent over 95% of world GDP. Table 1 gives an example of a coding protocol which relates to the legal approach to employment status, and provides an example of how it was applied in a particular country (here, the UK).

Each of the coding protocols contains a measurement scale which describes the range of values to be applied. We use a 0-1 scale, with higher values indicating more protection.

Lastly, we need to consider how to aggregate individual indicators to arrive at an overall composite score. This involves the question of whether, and if so how, weights should be used when the individual scores are averaged or aggregated.

Our primary sources consist of the laws on work passed in various countries from 1970 to the present day. We source the laws in the form of legal texts: statutes (legislation) and judgments (rulings of courts). Once we have the texts, we can apply our coding protocols to arrive at scores for the indicators, which are reported by country and year. The scores can then be combined (aggregated or averaged) in order to produce the composite score (either by reference to the sub-indices or to the overall index) for a given country-year.

According to the OECD Handbook, composite indices such as the CBR-LRI ‘are much like mathematical or computational models’, the construction of which ‘owes more to the craftsmanship of the modeller than to universally accepted scientific rules for encoding’ (OECD, 2008: 14). There is no one, unique way to code labour laws, and our own approach is not, in itself, more or less correct than alternatives. Rather, ‘the justification for a composite indicator lies in its fitness for the intended purpose and in peer acceptance’ (OECD, 2008: 14).

The protocols we developed for the Cambridge index were constructed for the specific purpose of generating a measure of labour regulation across countries and over time, which could be used to inform econometric analysis to estimate the social and economic effects of labour laws, as well as the wider political factors driving their adoption. Because of their very extensive time and country coverage they should prove useful to the wider social science community and to policy makers.

Acceptance by peers does not in itself guarantee the validity of an index. The widely-used World Bank indicators on labour law, part of its Doing Business Reports since 2004, were discontinued in 2021, because of fears that countries were ‘gaming’ the system of reporting on which the index depended. It would be naïve not to acknowledge that certain governments will have an interest in seeing their laws described in a particular way.

Addressing this, in the case of the CBR-LRI, we rely only on the texts of the laws, which are public documents, and on the coding protocols we ourselves devised, to arrive at the codings. The texts are sourced from the ILO’s NATLEX database and other documentation, much of which is available on the internet, or, if not, in law libraries.

The coding is carried out by members of a relatively small project team (currently, five people), and final scores, based on the application of the relevant algorithm in each case, are arrived at by a process of discussion and deliberation between us. Because we use a detailed coding protocol, it should be possible to arrive at an objective or ‘externally valid’ understanding of the ‘correct’ score for a given law. Although there is a danger of confirmation bias in our approach, we sought to minimise that risk by making our draft codings available to the wider labour law community prior to the finalisation of the latest iteration of the dataset. This was done via early publication of a draft on the Centre for Business Research website in June 2023, and the circulation of a request for feedback via the Labour Law Research Network, which has a global reach.

Where, after checking public sources, we were unsure about the provenance or meaning of a law, we consulted country-level legal experts. Since the dataset was first published in something like its current form in 2013, we have benefited from a steady of stream of correspondence with labour lawyers from the around the world, who have offered comments on it.

If complete objectivity may not be possible, a high degree of accuracy and transparency in the application of the coding protocols should be. Alongside a spreadsheet containing the values, we have published a book of codes and sources which sets out in detail the basis for each of the values. The aim here is not just to report our work but to enable other researchers to see exactly how we arrived at the country scores.

We have by no means solved all the problems inherent in the process of measuring the law. Because our primary sources are legal texts, the dataset does not track how laws are perceived to work by employers or workers; nor do we include information about enforcement in our dataset. We code for ‘jural’ law, not law ‘in practice’. This could be construed as a limitation.

However, the legal content of a rule cannot be dismissed as a matter of no relevance to its practical operation. While there may well be an enforcement gap in even in the most stable and well-functioning legal systems, data on enforcement are available from a number of other sources, and can be used in conjunction with our dataset to get a more fully rounded picture of the law’s operation in practice.

For example, there are lengthy time series, matching our own for year by year coverage, for data on the ‘rule of law’, in the form of the World Bank’s Governance Indicators, and the Freedom House indicator of human rights violations. The ILO collects country-level data on labour inspection, and several countries report on the volume and type of employment disputes going before courts and tribunals.

Perhaps the most difficult issue we have had to deal with is the question of weights. It is intrinsically unlikely that each of the forty indicators in our index has equal significance in all countries; yet this is the effect of simply averaging or aggregating them to arrive at a composite country score. The problem we face is that there is no very obvious way to assign weights a priori. This may be partly overcome through the use of a statistical method for identifying clusters of variables, such as factor analysis or principal component analysis. We report our data without weights so that researchers can make their own decisions on which weighting methods to use, if any; there is an argument to the effect that default of equal weighting may work well enough for a dataset of this scale and scope, containing as it does nearly 250,000 individual observations covering between them over 50 years and more than 100 countries.

3 First results from the 2023 dataset: what has been happening to labour laws around the world?

3.1 Long-term trends in labour protection

Our initial analysis offers some insight into how different countries have approached labour regulation over the last 50 years.

3.1.1 There has been a steady, if modest, improvement in the protective content of labour laws across all countries since the 1970s.

The rate of change has been fairly steady, and has been generally in the direction of regulation favouring workers over employers: shorter working weeks, tighter controls over dismissal, expanded collective bargaining and codetermination rights, and greater protection for the right to strike. There is no evidence here of the ‘end’ of labour law or of its desuetude.

Of course, the image of a steady improvement across all countries and all years is something of a mirage; not just a bird’s eye view but more like that of a distant satellite. When we dive down to the level of individual countries, we see not just constant change, but sharp reversals, with periods of deregulation (or weakening of labour rights) followed by improvements in labour rights, and vice versa. We also see different trends by areas of law.

3.1.2 Worker protections in the UK have been strengthened by EU directives, following a period of deregulation in the 1980s.

Figures 2a-2e chart changes in the UK since 1970. The protective content of the law on industrial action has not recovered from its precipitous fall in the early 1980s, while working time protection, from an even more pronounced decline, has rebounded since the late 1990s, although not to its previous level. The law governing flexible forms of work has shown a marked rise in protection over the same period.

In each of the charts, the average for all European countries is included as a benchmark. This clarifies the influence of European standards on British labour law: the implementation in the UK of the EU’s directives on working time and different forms of employment (part-time, fixed-term and temporary agency work) are largely responsible for the increase in the overall score during the 1990s and 2000s.

3.1.3 Deregulation of the kind the UK experienced in the 1980s, generated by neoliberal policies which treated labour laws as a ‘burden on business’, is not the norm.

Only three other countries – New Zealand in the early 1990s, Georgia in the mid-2000s, and Ukraine since 2022 – show comparably sharp declines to those experienced in the UK in the 1980s (see Figure 3, with the all country average also included). It would seem that deregulatory episodes have lasting effects: the UK recovered largely through the external influence of EU law, while in New Zealand and Georgia, the level of protection revived after a few years had passed, but did not reach its previous level. It remains to be seen what will happen in Ukraine. Its recent deregulatory measures, while introduced under the auspices of the wartime emergency, were planned before the current conflict there began, and might be expected to outlast it, since they reflect a broad cross-party consensus on the desirability of going down a neoliberal or libertarian route to labour market governance.

3.1.4 America is not just an outlier among liberal democracies, but across all countries.

Another small subset of countries have had weak labour laws throughout the period covered by the dataset: as Figure 4 shows (with the all country average again included as a benchmark), Iran, Saudi Arabia and the USA fall into this category. In terms of the low level of protection conferred by law on workers, the United States is not just an outlier among liberal democracies, but across all countries. The causes and consequences of US exceptionalism would repay further study.

In certain regions of the world, stronger labour protections are associated with periods of democratic rule, and weaker labour laws with dictatorships. As we can see from Figure 5, labour law protections declined sharply during periods of authoritarian rule across South America in the 1970s, but this trend was reversed when military dictatorships were ended and labour rights restored, a pattern common to Argentina, Brazil, Chile and Uruguay in this period. The ending of apartheid in South Africa in the 1990s was similarly associated with a strengthening of labour rights.

3.1.5 Labour protection laws are strengthening across East Asia

In other regions, a rising trend of labour regulation can be observed independently of the nature of the political regime, liberal-democratic or otherwise. Across East Asia, we see marked improvements across the board (Figure 6): for example, in Vietnam, following its adoption of a labour code in 1994 and more recently via amendments to its laws on freedom of association, and in China, thanks to its 2007 Labour Contracts Act. Japan’s labour laws, which have been in the middle to upper range of protection for some time, have seen a recent strengthening. While labour protection laws were strengthening across East Asia there was also a decline in informal work. The direction of cause and effect between these trends may become clearer with further research.

3.2 Signs of recent improvements in labour protection

Updating to 2022 has also revealed some more recent trends. After a period of some weakening of labour protections in Europe, following the sovereign debt crisis of 2009, labour rights have seen a rebound, driven by two factors.

3.2.1 Covid 19 led to tightened control over dismissals

One is the Covid 19 pandemic, which prompted several countries to tighten controls over dismissals. In some cases, these changes were combined with a loosening of rules on overtime and the use of fixed-term contracts. A smaller number of countries loosened dismissal controls and introduced restrictions on the right to strike in response to Covid (see Table 2).

3.2.2 In some countries there are signs of increased protections for platform workers

A second trend is the tightening of rules governing the ease with which employers can escape labour law regulations through the use of different employment forms. As platform work has become more salient, not just in practice but as an issue on the agenda for legislators and also for courts via litigation, there has been a tendency to implement interventions. These interventions set basic protections for platform workers regardless of their status, or create a presumption of employee or worker status, granting them access to the protections generally made available by labour laws. The November 2023 decision of the UK Supreme Court in Deliveroo, ruling that delivery riders were not entitled to seek recognition for collective bargaining, is out of step with international trends.

This ‘normalisation’ of platform work follows the pattern established in the 1990s and 2000s with respect to ‘flexible’ or ‘precarious’ forms work: on the one hand, acknowledgement of the legitimacy of these employment forms; on the other hand, a move to align some of the rules governing such flexible forms with the more ‘standard’ forms of employment. Both trends are most clearly visible in EU member states, but if the experience of the 2000s is anything to go by, the European model may well be diffused to other parts of the world over time, an instance of the ‘Brussels effect’ which occurs when countries beyond Europe adopt EU standards. The EU has proved highly active in exporting its laws, including on worker protection through foreign trade agreements, while companies in third countries increasingly follow EU standards in order to access European markets (Bradford, 2020).

4. Understanding the impacts of labour laws, including on platform work

Earlier versions of the CBR-LRI have been used to test some of the key claims made in debates over the economic and social effects of labour laws. Among the emerging findings are the following:

  • Worker-protective labour laws tend to increase the labour share of national income, that is, the proportion of income going to wages and salaries as opposed to dividends and profits (Adams et al., 2019).
  • Labour protection is positively correlated with productivity (Deakin and Sarkar, 2008).
  • Employment protection legislation is positively correlated with innovation, as measured by the number of patents filed by firms (Acharya et al., 2013; Belloc, 2019).
  • The employment effects of labour protection are mostly positive, although the magnitudes of the observed effects are small (Adams et al., 2019).

The updated dataset will enable further analysis of the impacts of labour laws in recent years, which we are currently carrying out. As with previous iterations of the dataset, the 2023 version is available to other researchers to use.

The additional decade or so that we have now coded is a period that saw considerable changes in labour laws around the world during the Covid crisis and in response to the rise of platform work. In particular, the new iteration of the CBR-LRI will facilitate better understanding of how the rules on employment status are impacting platform workers and other precarious workers in the often uncertain ‘grey zone’ between employment and self-employment.

Authors

Bhumika Billa, Researcher, University of Cambridge.

Louise BishopLouise Bishop, Digit Research Fellow, University of Cambridge.

Professor Simon Deakin, Digit Co-Investigator, Professor of Law and Director of the Centre for Business Research, University of Cambridge.

Kamelia Pourkermani, Digit Research Fellow, University of Cambridge.

Tvisha Shroff, Ph.D. student, Faculty of Law, University of Cambridge.

References

Acharya, V., Baghai, R. and Subramanian, K. (2013) ‘Labor laws and innovation’ Journal of Law and Economics, 56: 997-1037.

Adams, Z., Bishop, L., Deakin, S., Fenwick, C., Martinsson-Garzelli, S., and Rusconi, G. (2019) ‘The economic significance of laws relating to employment protection and different forms of employment: analysis of a panel of 117 countries, 1990-2013’ International Labour Review, 158: 1-35.

Adams, Z.,  Billa, B., Bishop, L., Deakin, S. and Shroff, T. (2023) CBR Labour Regulation Index (Dataset of 117 Countries, 1970-2022) Codes and Sources (Cambridge: Centre for Business Research) https://doi.org/10.17863/CAM.9130.2.

Belloc, F. (2019) ‘Institutional complementarities between labour laws and innovation’ Journal of Institutional Economics, 15: 235-258.

Bradford, A. (2020) The Brussels Effect: How the European Union Rules the World (Oxford: OUP).

Deakin, S. and Sarkar, P. (2008) ‘Assessing the long-run economic impact of labour law systems: a theoretical reappraisal and analysis of new time series data’ Industrial Relations Journal, 39: 453-487.

Deakin, S. (2018) ‘The use of quantitative methods in labour law research: a defence and reformulation’ Social and Legal Studies, 27: 456-474.

OECD (2008) Handbook on Constructing Composite Indicators (Paris: OECD).

OECD (2023) Indicators of Employment Protection

World Bank (various years) Doing Business Reports (Washington, DC: World Bank).

About the CBR Labour Regulation Index

The CBR Labour Regulation Index (CBR-LRI) is one of a number of leximetric datasets created in the Centre for Business Research at Cambridge University which make it possible to track changes in labour, company and insolvency laws in many countries over an extended period of time. The datasets employ a novel leximetric coding method which was developed by the project team. The CBR-LRI codes for labour laws in 117 countries (equivalent to 95% of world GDP). It organises the codings into 40 individual indicators and 5 sub-indices, covering laws on different forms of employment, working time, dismissal, employee representation, and industrial action. The 2023 update to the CBR-LRI extends the yearly coverage of the index by a further decade, so that it now codes for the period 1970-2022.

Share this: