PensionReforms' summary and comments
Many think that employees aren't saving enough for retirement.  Supporters of the principles of 'behavioural economics' think that employees should be forced to address the issue of retirement saving by being auto-enrolled into the employer's scheme when first starting work.  Then they can have a choice about whether to continue as members by opting-out.

Auto-enrolment will certainly increase the number of members but whether it means more retirement savings depends on at least two considerations:
-   Do the employees who remain members reduce other savings they might be making outside the employer-sponsored scheme?  We should expect at least some substitution.
-   Is there a finite budget for the total amount the employer is prepared to spend on the retirement saving schemes of all its employees?  If there is, the more employees who join, the less will be spent on each employee.

This report looks at the second of these propositions.  Do employers with auto-enrolment saving schemes subsidise those savings on a different basis from employers with voluntary schemes?  The answer to that seems to be 'yes' for the sample of employers measured.

"We use data from the Form 5500 and the Pensions & Investment top 1,000 pension funds.  The 5500 data include the full universe of private pension plans but does not include information on automatic enrollment; the P&I 1,000 data represent a limited number of plan sponsors but include an automatic enrollment indicator.  Our analysis sample resulting from merging the Form 5500 data for 2007 with the P&I 1,000 is limited to 826 plans from 532 employers. Yet, these plans hold about half of the total 401(k) assets and accounts for about 30 percent of all participants in the system."

The report acknowledges the limitations of the sample but finds that "...firms with automatic enrollment have employer match rates that are about 7 percentage points lower than those without automatic enrollment, even after controlling for firm characteristics."

The report notes that the regressions they carried out suggest a relationship between automatic enrollment and match rates but don't necessarily mean that auto enrollment causes lower match rates.

Even with the reduced match, the auto-enrolment schemes still cost the employers more than the alternative:
"...our calculations suggest that a 7 percentage point reduction in match rates would offset at least 42 percent of the increase in costs for firms with participation rates of 60 percent or more before automatic enrollment."

Presumably, the reduction in costs will be somewhat less if the participation rates, before auto-enrolment, were less than 60%.

"The findings of this paper indicate that while automatic enrollment is likely to achieve the goal of increasing pension coverage, it might also work against the principal goal of increasing retirement savings."

PensionReforms doesn't follow this conclusion.  Based on the average results found in the research, the reduced match rates still left the average auto-enrolment scheme 58% more expensive (the reduced rate offset only 42% of the increased cost).  That means there is more money coming from employers than under voluntary schemes.  On the other hand, the average contributed for each employee member will be less and perhaps that is what the report means.

The report also notes that, with a lower match in auto-enrolment schemes. Members might be encouraged to contribute less themselves, so reducing the total amount saved in respect of each member.

The report's findings have been challenged by one other organisation - the Employee Benefit Research Institute (EBRI).  Its press release is here.  The problems, in brief, seemed to be the report's constructed match rates and the failure in the report's database to identify the date when auto-enrolment was introduced.  EBRI will apparently produce its own report on this topic - PensionReforms will cover that.  It isn't easy to see from the EBRI press release why the report's findings might be wrong.

In PensionReforms' opinion, the report's findings seem like common sense and suggest that employers might have a total budget for remuneration (including subsidised benefits).  The more employees who join, the less there is available for each.  Auto-enrolment might get employees thinking about retirement saving issues but any decision they make (or don't make in an auto-enrolment environment) can't be costless.  Someone has to pay for it and the choices are limited: shareholders, taxpayers, ratepayers, customers and now, according to the report, fellow employee-members. (File size 156 KB; 36 pp) 375