#673 1997 · Mexico's 1997 pension reform (studied by Kumler, Verhoogen & Frías) · Public finance / tax compliance
Mexican employers under-reported wages to shrink their tax bill for years — their own employees knew, and had no reason to say anything, until a pension reform gave them one
the problem
Formal Mexican employers routinely under-reported employee wages to the social-security system to reduce their payroll-tax bill, and the employees who knew their true pay better than any auditor had no personal stake in reporting the fraud
background
Under Mexico's pre-1997 pay-as-you-go pension system, a worker's eventual retirement benefit wasn't closely tied to what an employer actually reported and contributed for that specific individual — benefits were determined more by aggregate formulas than by an individual's own contribution record. That meant an employee who watched their employer under-report their wage to the social-security agency, shrinking the employer's payroll-tax bill, had no direct financial reason to object: the fraud cost the employee nothing personally under the old benefit structure, even though the employee alone had accurate, first-hand knowledge of what their true wage actually was.
Researchers Todd Kumler, Eric Verhoogen and Judith Frías compared two independent sources of individual-level wage data in Mexico — firms' own wage reports to the Mexican social security agency (IMSS) and workers' self-reported wages in an independent household labor-force survey — to measure how much under-reporting was actually occurring and how it changed after Mexico's 1997 pension reform switched younger workers onto individual retirement accounts, where the benefit paid out is directly determined by what gets reported and contributed on that specific worker's behalf.
what everyone would do
The standard response to systematic employer under-reporting of wages to evade payroll tax is more enforcement aimed at the employer — more auditors, stiffer penalties for firms caught misreporting, more compliance paperwork. That approach treats the problem as one of insufficient deterrence against the party doing the misreporting, and requires the government to actively discover and prove fraud that employees, who already know the true numbers, had no reason to help surface.
what they saw
Kumler, Verhoogen and Frías's account of the reform shows the government recognized that the workers watching their own paychecks already possessed exactly the information any auditor would need to catch under-reporting — they simply had no reason to act on it, since the old pay-as-you-go pension system didn't tie their own eventual benefit closely to what got reported for them individually. Rather than build better enforcement to catch employers, the 1997 reform rewired the underlying incentive: switching younger workers to individual retirement accounts where the benefit paid out IS a direct function of reported and contributed wages meant an employer's under-reporting now directly reduced that specific worker's own future retirement money, giving millions of already-informed employees a genuine personal reason to start caring whether their reported wage was accurate.
the move
Rather than adding auditors, penalties, or new compliance paperwork aimed at employers, the 1997 reform changed who had a personal financial stake in accurate wage reporting: for the newly-covered younger cohort, an employer under-reporting wages was no longer a victimless shortcut, it directly reduced the worker's own eventual retirement account balance — turning millions of employees who already possessed the accurate information into people who now had a real reason to care whether that information was reported correctly.
why it works
By making the employee's own future benefit a direct, transparent function of what got reported for them individually, the reform converted an employee who previously watched fraud happen to their paycheck with no stake in stopping it into someone with a genuine financial reason to object, ask questions, or push back — without the government needing to add a single new auditor or penalty. This is precisely why the effect showed up specifically and only among the newly-incentivized younger cohort while older workers under the unchanged pay-as-you-go system saw no comparable decline in under-reporting: the mechanism worked through the individual employee's changed stake in the number, not through any general deterrence effect that would have applied equally regardless of which pension system a given worker was under. Because employees already possess accurate information about their own true wages as a simple byproduct of being paid, this fix required no new monitoring infrastructure at all — the government only needed to make the existing information matter to the person who already held it.
the payoff
Using a difference-in-differences design comparing older workers (who remained under the old pay-as-you-go system) to younger workers (who moved to the new individual-account system), the study found under-reporting fell specifically and only for the newly-incentivized younger cohort, with the decline in under-reporting greatest in employer-sector-region cells that initially employed a younger workforce on average — exactly the pattern predicted if the mechanism worked through employees' individual financial stake rather than through any general deterrence effect that would have applied equally to older and younger workers alike.
where it breaks
This mechanism depends on there being a real, meaningful outcome the informed party can be tied to — a pension system's individual-account structure creates a natural, high-stakes link between accurate reporting and a worker's own future money, but not every misreporting relationship has an equally clean, high-value outcome available to tie to the informed party's stake. It also depends on the informed party actually having some leverage or voice to act on their new incentive — an employee who discovers under-reporting but has no realistic way to raise it with an employer, no protection from retaliation, or no channel to report it, gains a reason to care without necessarily gaining a way to act, limiting how much the incentive alone can move behavior. And this fix specifically targets misreporting that an insider already has accurate knowledge of — it does nothing for fraud or errors that are genuinely hidden from everyone inside the relationship, since there's no informed party to enlist when nobody on the inside actually knows the true number either.
what came after
Published as Kumler, Verhoogen & Frías, 'Enlisting Employees in Improving Payroll-Tax Compliance: Evidence from Mexico' (Review of Economics and Statistics 102(5), 2020; NBER Working Paper 19385, first circulated 2013), the study is cited in public-finance and tax-compliance literature as evidence that giving an already-informed party (an employee who already knows their true wage) a personal financial stake in accurate reporting can improve compliance more durably and cheaply than adding external monitoring or enforcement aimed at the party doing the misreporting.
references
- [1]Enlisting Employees in Improving Payroll-Tax Compliance: Evidence from MexicoNational Bureau of Economic Research (Working Paper 19385; published in Review of Economics and Statistics 102(5), 2020), 2013nber.org
- [2]Enlisting Employees in Improving Payroll-Tax Compliance: Evidence from MexicoIZA Institute of Labor Economics (Discussion Paper 7591), 2013iza.org