PensionReforms' summary and comments
When a rational, informed saver decides how much to put aside for retirement, the amount of the state's pension provides the base coverage and private provision will supplement public entitlements to help reach the target.

So when the state decides to replace part of public provision with a compulsory Tier 2 arrangement of "Personal Retirement Accounts" (PRAs), logic suggests that the existing supplementary private savings should not change.  The pre-funded Tier 2 account will substitute the PAYG defined benefit Tier 2 pension, at least as to part but the overall target would be unaffected.  On the other hand, if the reform increases state-delivered retirement pension wealth (or seemingly makes it more 'certain'), we might expect other private savings to fall.

Is that what actually happens?

The report sets the tone from the outset:
"... most PAYG systems, which are supported by younger workers paying into the system, are collapsing as a result of an aging population - given these issues, PRA plans are more financially sustainable than PAYG systems and increase private savings." 

PensionReforms' readers will expect a challenge to this simplistic summary - more on that below.

The report looks at the Mexican experience.  In 1997, part of the state's PAYG Tier 2 was replaced by a system of PRAs.  The report calls this a "natural experiment" because Mexico's 1997 reform affected different groups of employees differently.  Most private sector employees - participating in the Mexican Social Security Institute (IMSS) - were affected and so could be compared as a group with employees who were not forced into the PRA system. The report was able, through looking at income and expenditure data (matched to social security membership information) to see the direct impact of the PRA system on 'private' behaviour.

The PRA-based reform produced a scheme much like the Chilean model and the PRA itself is run by a private AFORE, much like the Chilean AFPs.

However, there is a major difference - as the report notes, 97% of the AFORE's assets are in government bonds - what the report did not make clear is that this is a requirement, not an option - see here for a fuller explanation of the real nature of the Mexican reform.

So whereas previously, a contributor to the 'unfunded' PAYG pension depended on a future government's promise to pay the Defined Benefit pension, under the reformed system, the contributor now depends on a future government's promise to repay debt.  PensionReforms suggests that this is not a change of economic substance for the country though it does shift economic risks from taxpayers to contributors.  Because members at the transition will receive the higher of the two benefits, pension costs to tomorrow's taxpayers must, in fact, be higher.  For new members, it could also be higher if the combined arrangement does not achieve the country's welfare objectives, whatever they may be.

PensionReforms notes that comparative value comparisons over time between Defined Contribution and Defined Benefit systems are driven entirely by the assumptions.  In Mexico, wages are falling relative to inflation while investment returns have outpaced inflation.  Given that the Defined Benefits are related to pay while the PRA benefits are related to investment returns (though the initial contributions are pay-related) the addition of the PRA system will look as though retirement benefits will be higher.  Whether it turns out that way is a rather different issue.  The report doesn't examine the robustness of the key assumptions it uses.

After using a model to explain what the likely effects of the Mexican reforms might be, the report then examines the actual experience based on data from 10,000 randomly chosen households.  In this analysis, "savings" is the macro-economic measure of the difference between income and consumption. "Consumption" includes spending on both durables and nondurables.  The survey data has no information on stocks of wealth.  PensionReforms thinks that all this probably does not give us a complete picture for the report's purposes of what is actually happening in Mexican households.

Notwithstanding this reservation, here is what the report concluded:
"The main findings are that social security reform increased consumption and crowded out saving for low- and lower-middle-income individuals.  The impact is stronger for older cohorts closer to retirement age.  We find no effect for upper-middle and higher-income individuals....

"According to the Life Cycle Model, these findings suggest a permanent decline in household savings for a significant proportion of the labor force in Mexico.  The latter may generate households who are not adequately prepared for retirement."

So, what to do about this?

The report recommends making the voluntary Tier 3 section of the PRA system more attractive, with "fiscal exemptions" targeted to lower income contributors.  More work needs to be done to promote retirement saving and finally, contributors should be encouraged to work longer to build greater PRA savings.

PensionReforms has some problems with the thesis on which the Mexican reforms (and the report) are founded.  PAYG Defined Benefit pension systems are not "collapsing"; PRAs are not necessarily more "financially sustainable" and pre-funding a Tier 2 scheme does not necessarily increase savings, particularly if all the costs, both fiscal and economic, are taken into account.  The government's "social quota" of US$5 a month is an explicit, additional cost of the new system.  

Both PAYG and pre-funded systems depend for their "sustainability" on the strength of the economy on which the claims are eventually made - at retirement, not now.  Unless it can be demonstrated that pre-funded Tier 2 systems make the economy stronger, the Mexican reforms, by turning an explicit pension promise into an explicit debt, will have re-arranged future claims but not made those claims more secure.

The report suggests without evidence that "PRA plans ... increase private savings".  The evidence from Chile is actually that very little of the 18.2 percentage points increase in saving over the period to 1994 can be attributed to the AFP system introduced in 1981 - see here and here for more on this.

The evidence suggests that Mexicans agree the PRA system will increase their retirement incomes and so they have reduced their other savings.  They reached that conclusion because of what the PRA system's promoters told them and because, if their pension under the new system turns out to be less than it would have been under the old system, the Mexican taxpayer will make up the difference.  PensionReforms would have expected such results from the outset.  (File size 260 KB; 39 pp) 251