As we described in our last post, our new book leverages several different research methods to evaluate the effectiveness of constitutional rights. The principle one of those methods is using large-N data to examine the relationship between rights included in constitutions and the protection of those rights.
Our large-N analysis is made possible by our global database on constitutional rights provisions. To construct it, Mila, as her dissertation project, hand-coded all national constitutions that were in force in 1946 and all the subsequent amendments or wholesale replacements to those documents. For each constitution, she tracked over a hundred constitutional rights provisions and recorded when these provisions appeared in (and disappeared from) national constitutions.
For our book, the dataset was updated to include more years, cross-checked against the timeline of the Comparative Constitutions Project, and a number of historical constitutions were added. The dataset now includes detailed information on a large number of rights from 1946 to 2016 for all 194 widely recognized countries in the international state system.
Using this data, the above figure documents the spread of rights from 1946 to 2016.[1] It reveals a profound increase in the number of rights in constitutions. It also reveals that the increase in constitutional rights is not confined to a handful of countries. In 1946, a country in the 10% percentile had just 4 rights. But by 2016, that number had risen to 24. Similarly, a country in the 90th percentile had 38 rights in 1946, but 62 rights in 2016. Additionally, the most pronounced jump is in the 1990s, which was a period of unprecedented constitution-making activity, spurred by the end of the cold war.
In our book, we focuses on assessing the effect of eight key rights: (1) the freedom of speech; (2) the prohibition of torture; (3) the freedom of movement; (4) the right to education; (5) the right to healthcare; (6) the freedom of religion; (7) the right to unionize; (8) the right to establish political parties. (Although we focus on these rights, we also examine variants of these rights, and in other work we have studied the effect of other rights, including the right to association, right to housing, right to social security, and the right to gender equality.)
We examine the effect of including these rights in constitutions by using four increasingly demanding comparisons. Our goal when doing so is to be transparent about what the basic patterns are in the data. We do more sophisticated things too, but as a starting point we just want to make it clear whether there are positive correlations between rights and the protection of rights.
To illustrate, the below figure presents these four comparisons for the right to unionize. For this analysis, the dependent variable we use to measure de facto labor rights is from the CIRI project. This measure captures the extent to which workers enjoy the “freedom of association at their workplaces and the right to bargain collectively with their employers.”[2]
The first comparison is between the rights performance of countries with and without a given constitutional right. For example, this figure shows that there has been a gradual decrease in workers’ rights around the world, but that countries with the right to unionize do slightly better protecting this right. Of course, countries with and without the right to unionize are likely different in myriad of important ways, so this simplistic analysis cannot tell us whether constitutional rights make a difference.
The second comparison is the rights performance of countries in the five years before and after they adopt a specific constitutional right.[3] This comparison has the advantage of showing whether there were any within-country changes after adopting a given right. For example, the countries that add the right to unionize do notably do better at protecting de facto workers’ rights starting in the year they add this right to their constitution.
The third comparison is the rights performance of countries before and after they adopt a specific constitutional right, while creating a control group by using a stacked event study research design. The intuition of the stacked event study is that we first define a “treatment event” as a period where a country adopts a constitutional right, and we then create “control events” comprised of countries that did not change that same right during the same eleven-year period. Using this sample of treatment events and control events, the third comparison shows that control countries experienced a gradual decline in workers’ rights at the same time that the treated countries noticed an uptick.
Finally, because a limitation with our third counterfactual is that rights adoptions and rights outcomes could be influenced with other factors occurring contemporaneously, the fourth comparison uses the same sample, but estimates a regression model that controls for factors that could influence the adoption of the rights and rights protection (as well as a series of fixed effects for the countries, years, and events). This regression shows that there is a positive and statistically significant relationship between the de jure right to unionization and protection of workers’ rights.
In addition to these four comparisons, for each of the rights we study, we report 10 regression models that use different approaches to estimate the relationship between constitutional rights and de facto rights protections. Through all these results, we document that adding organizational rights to constitutions is associated with those organizations enjoying better rights, but for the individual rights we look at, these increases in rights protections don’t occur. There is simply little change in rights protections for individual rights after countries add them to their constitutions.
Although these regressions cannot establish causal relationship, they can at least reveal whether there are positive associations between rights and rights protections. Through our case studies, we’re then able to examine the reasons for these patterns. The results of that research suggest that dedicated organizations are able to use constitutions to their advantage in a way that does not occur as consistently for individual rights.
Note: Our book, How Constitutional Rights Matter, was recently published by Oxford University Press. This is the fourth in a series of blog posts on parts of the argument and evidence from the book.
----
[1] The figure uses the coding of 87 unique constitutional rights. The dataset has information on many more rights, but this list of 87 rights are all the rights in the dataset that are unique rights, and do not represent more detailed coding of the same right. For example, the list of 87 rights includes the freedom of expression, but not the variable that captures whether the constitution explicitly enshrines viewpoint restrictions on speech.
[2] CIRI Human Rights Data Project, Short Variable Descriptions for Indicators in the Cingranelli-Richards (CIRI) Human Rights Dataset, Document Version 5.21.14 (2014). We specifically use the variable “worker” from the 2014 release of the CIRI dataset.
[3] We use five year windows in our second, third, and fourth comparisons. We do so because it is consistent with prior research on human rights treaties and constitutional rights. Of course, it could take longer than five years for constitutional rights to have effects, but if there is a long delay in rights improvements, it is difficult to attribute the improvement to the adoption of constitutional rights.