It is well documented that US defined contribution pension plans are largely invested in the shares of their employer. I argue that when the (single representative) worker holds shares in the firm, he tends to monitor the manager. On the one hand, the manager and shareholders gain from the productivity of the worker. On the other hand, the manager bears the cost of being monitored by the worker, and the shareholders loose part of their ownership power to the worker. Therefore, there is an optimal ownership limit for the worker from the viewpoint of the firm. I derive conditions under which the worker will never invest in the firm's stock if he has the freedom to do so. Nevertheless, under alternative conditions, it will be advantageous for the worker to invest his pension assets heavily in company stock, even if the under-diversification cost is very high.

Among the requirements a pension plan must meet to qualify for tax benefits are the nondiscrimination rules. Nondiscrimination rules are designed to ensure that pension benefits do not disproportionately accrue to highly compensated employees. But the rules are also complex and increase administrative and compliance costs associated with offering a pension plan. Recent pension reform proposals would simplify nondiscrimination rules, reducing administrative and compliance costs and potentially leading to more employers offering pension benefits. However, there are concerns that any loosening of the rules could lead to a drop in participation by low-wage workers. This paper examines the economic incentive that nondiscrimination rules provide to employers to cross subsidize employees; that is, the incentive to increase pension benefits (and total compensation) paid to low-paid workers for the express purpose of enabling high-paid workers to receive a higher proportion of compensation in the form of pension benefits. The study calculates the incentives faced by a hypothetical firm, and then illustrates how those incentives change when assumptions about employee contribution behavior, employee compensation, and employer-matching formulas are allowed to vary. Results show that only firms with a relatively low ratio of low-paid workers to high-paid workers would have an economic incentive under a standard 401(k) plan to cross subsidize employees. Although this incentive may exist in a large number of firms, these firms likely employ only a small portion of the workforce. This is ultimately an empirical question, however, and examining data on the distribution of earnings within pension plans, as well as determining if firms find nondiscrimination rules binding, would be a useful extension of this research.

This paper solves and simulates a stochastic life-cycle model of an economy with 401(k) plans. We use the model to establish a benchmark for patterns of contributions and participation and show how these patterns depend on such features as employer matching, vesting policies, and the ability to make early withdrawals. Consistent with empirical studies, the model predicts relatively low participation rates among younger workers and shows that these rates tend to rise with more generous matching, lower vesting periods, and improved liquidity.

I explore the diversity among pension plans offered to faculty at 52 Canadian universities. Far from displaying homogeneity, I find marked differences between the plans, both in their structure (DB, DC, and hybrid) and in the replacement ratios from those plans. Using financial simulations I construct representative pension payoff profiles for each plan and find expected replacement ratios (the annual annuity stream available at retirement expressed as a percentage of final wages) ranging from 54.3% to 88.2% and internal rates of return on contributions varying from 10.39% to 18%.