PensionReforms' summary and comments
Governments have found it quite difficult to force their citizens to save for retirement, even when the law says they must (because there is a compulsory Tier 2 scheme).  The latest World Bank publication (reviewed here) says, for example, that in Latin America and the Caribbean, "nearly half the countries have coverage rates below 30 percent".

This report assembles household survey data to analyse Tier 2 participation rates for salaried and self-employed workers in seven countries of Latin America: Brazil, Chile, Colombia, Costa Rica, El Salvador, Nicaragua and Peru.  There are fundamental differences in the pension schemes of these countries, and wide variation in participation.  The data cover the period 1990-2002, but coverage varies by country.

Participation is legally mandatory in each of these countries for all workers except the self-employed.  For sampled salaried workers aged 15 to 64 years working more than 5 hours a week, average participation rates vary from a low of 32% in Nicaragua to a high of 78% in Chile.  Average participation rates for the other counties, in ascending order, are Peru (33%), Colombia (53%), El Salvador (54%), Brazil (71%) and Costa Rica (76%).  Participation of the self-employed-which is voluntary-is less than 5% in El Salvador, Nicaragua and Peru, 8% in Colombia, 22% in Chile, and 42% in Costa Rica.

The report acknowledges that the household survey data are less than perfect for its purposes, as the recorded participation rates are biased upwards in every country other than Chile.  In Brazil, Costa Rica and El Salvador, the survey information refers to contributions to social security in general, which includes health care and other programmes as well as pensions.  In Columbia and Peru the survey questions relate to a worker's affiliation rather than contribution status.  It is very possible for a worker to be affiliated to social security-for example, with a tier 2 personal account- while not actively contributing.  The data of El Salvador suffer from both of these biases.

Fitting a probit model to individual-level data yields expected and unsurprising results:
"In all countries studied, old age pension participation strongly increases with the education and the age of a worker.  Women tend to contribute more than males, while being married and head of the household increases an individual's probability of contributing, particularly for males.  Individuals in households with a higher share of non-earners are more likely to contribute, while the size of the household is negatively correlated with the probability of contribution.  Individuals working in urban areas at firms with more than five employees, employed full time, in the public sector and in manufacturing are more likely to contribute than other workers.  Workers in households where other members are already contributing and workers with higher earnings are also more likely to contribute.  Our results indicate that factors related to the demand for insurance (captured with individual and household variables) may account for at least 20-30 percent of the explained variance."

More surprisingly, "despite the lower contribution rates for self-employed workers," the authors "find strong commonalities in the contribution patterns of wage employees and self-employed workers." This suggests that "the low contribution rates observed in Latin America are driven by a combination of certain types of workers' low willingness to participate in social security programs and the State's inability to enforce firms' contributions for workers not willing to participate."

The report concludes, ".if the problem lies in the low willingness or ability to participate of a large number of less-advantaged workers, policies intended to increase the coverage of social security programs should alter the current equation of benefits and contributions.  This may imply subsidizing workers with low willingness or ability to contribute, improving information about the benefits of planning for old age, or better targeting the package of benefits to the needs and risks of people with low willingness to contribute."

PensionReforms observes that forcing citizens to save, even with significant tax breaks for those savings seems generally not to be working in the seven Latin American countries covered.  The report's recommendations will deliver more subsidies to the poorer and the self-employed and will presumably need some form of resource-testing to ensure that the extra spending goes to the intended recipients. That will make an already complex environment even more complicated.

PensionReforms thinks that a simpler solution might be for the government to provide, from general tax revenue, a basic non-contributory Universal Pension for every elderly resident, or at least for those who are unwilling or unable to contribute enough to the Tier 2 scheme to qualify for a minimum pension. (File size 563 KB;  46 pp) 329