Motivation and Hypotheses
Free cash flows, pension plan contributions and financial slack
The use of pension plans to park excess cash flows and build financial slack is deemed desirable for a number of reasons. An attractive feature of pension plans is that the employer has almost full discretion as to the timing and extent of contributions to the pension plan. Once minimum levels required by ERISA are satisfied, the employer can decide whether to exceed these levels in any given year. Since the time horizon of the pension obligation is long (depending on the average time to retirement of employees), employers have latitude in the exact timing (and extent) of contributions to the plans. Larger (smaller) contributions made earlier will necessitate smaller (larger) contributions in later periods.
Due to the employer’s discretion of timing and extent of pension plan contributions, employers may be likely to make greater contributions in periods where the firm has generated (excess) cash flows beyond its current needs for reinvestment in the business. Furthermore, pension plans are tax advantaged as contributions to pension plans are tax deductible8 and earnings on pension plan assets accumulate tax-free for the use of future retirees. Thus, unlike other (alternative) investment opportunities of the firm, the return on the investment in the pension plan is not taxable.
Consequently, a firm’s larger current contribution will accumulate tax free in the pension plan, guaranteeing a high after-tax rate of return. As a result, the firm may be able to contribute lesser amounts in the future, and use the saved cash flows for future investments in the business as required.
This, therefore, leads to the following testable hypothesis (in the alternate form):
H1: There is a positive relationship between free cash flows and pension plan contribution levels.
Rejection of the null hypothesis and finding a positive relationship between cash flows and contribution levels would be consistent with our premise that managers act in a manner consistent with Myers-Majluf's financial slack theory by using pension plans to store free cash flows.
However, as it is also plausible that finding a positive relationship between contributions and free cash flows may be a result of an “ability to pay” hypothesis; i.e. firms which have more (less or negative) free cash flows are able to make more (less) contributions. To differentiate between the two possibilities we take note that the financial slack argument is asymmetric; i.e. only firms with free cash flows will build up financial slack. Thus, we expect the relationship between contributions and free cash flows to be strong for firms with positive free cash flows, and to not hold for firms with negative free cash flows. In (alternate) hypothesis form we posit
H1a: The relationship between free cash flows and pension plan contributions is positive and significant for firms with positive free cash flows, but not for firms with negative free cash flows.
Utilization of (pension plan) financial slack
In the Myers and Majluf world, firms have positive NPV projects for which they are storing excess cash flows. This means that, eventually, they will use the financial slack to invest in these (profitable) projects. More specifically, if management uses pension plans to store excess cash flows in the manner predicted by Myers and Majluf, then after a period of time (when the investment projects materialize) we would expect those firms to reverse the process, by increasing capital expenditures which would subsequently result in improved profitability. This leads to the following testable hypothesis (in the alternate form):
H2: Firms that reverse their level of pension plan contributions will have greater levels of capital expenditures and profitability as compared to firms that continue to increase the level of their pension plan contributions.
Finally, to the extent that the market is able to anticipate this behavior, we would expect that market returns for "reversing" firms whose NPV projects have materialized would be greater than for firms which as yet had not began to implement the investment plans. In (alternate) hypothesis form, we have
H3: Firms that reverse their level of pension plan contributions will have greater returns than firms that continue to increase the level of their pension plan contributions.
Methodology and Results
Firms in our sample were collected from the 1997 Compustat Annual Industrial file. We selected firms over the six-year period 1991-1996. A firm for a given year was selected if data availability made the following calculations possible.
Pension contributions (CONTA) can be estimated9 from the Compustat database, and are non-negative10. Pension contribution is estimated as the pension expense (data item 295) minus the change in the accrued pension liability or prepaid pension cost (asset) (data item 290 plus data item 300) from the prior year. These changes were examined net of (changes in) the minimum liability adjustment (data item 298).
Free Cash Flow (FCFA) is the net operating cash flow (data item 308) minus the capital expenditures (data item 128).
Return on pension assets (PENROA) (data item 333, which Compustat records as a negative number for gains).
The net funding status of the pension plan (FUNDA) can be calculated from Compustat data as the fair market value of pension assets (sum of data items 287 and 296) minus the projected benefits obligation (sum of data items 286 and 294).
Paid pension benefits (BENA) can be estimated from Compustat, and are non-negative. Pension benefits are estimated (see appendix) as pension contributions (as estimated above) minus the change in the fair market value of pension assets (data item 287 plus data item 296) minus the actual return on assets (data item 333, which Compustat records as a negative number for gains).
The tax rate (TAX) of the firm is defined as the tax payments (data item 317, if available, or tax expense, data item 16, divided by pre-tax income, data item 170). The tax rate is winsorized to be between 0 and 60%.
To make cross-sectional comparisons meaningful, the first five variables were scaled by total assets of the firm (data item 6). Firms whose total assets were less than $0.5 million were deleted from the sample. To eliminate the influence of extreme observations, we deleted firms where any one of the following variables fell into the top or bottom 0.5% of the cross-sectional distribution of firms:
The ratio of contributions to total assets.
The ratio of pension benefits paid to total assets.
The ratio of free cash flow to total assets.
The ratio of net funding status to total assets.
After applying these selection criteria, our sample consists of 8,643 firm-years made up of 1,683 different firms.
The first two variables, contributions (CONTA) and free cash flows (FCFA) were required as it is the relationship between these variables that is being tested for in Hypotheses 1 and 1a. However, testing these relationships requires controlling for other variables which may affect firm contribution levels.
Funding Status (FUNDA)
Contribution levels are likely to be related to the pension plan’s funding status. While ERISA places minimum funding requirements on firms, employees may pressure firms to make larger contributions to the pension plan when the plans are severely underfunded. Finally, it is likely that a firm may wish to make greater contributions to underfunded plans because underfunded plans represent liabilities, albeit (at least to some extent) off-balance-sheet liabilities. Employers may be concerned about the impact of these off-balance-sheet liabilities on future debt or equity issuance, and may choose to reduce these liabilities through pension plan contributions. Thus, it is reasonable to postulate that firms with overfunded pension plans are less likely to contribute to pension plans than firms with underfunded pension plans.
Tax Status (TAX)
The discussion which described the motivation for using pension plans to build financial slack noted the importance of the tax advantaged status of pension plans. Thus, firms with higher tax brackets can better utilize the tax benefits of pension plan contributions and we postulate that firms in higher tax brackets will contribute greater amounts to their pension plans.
Benefits Paid (BENA)
The other cash flow relevant to a firm’s pension plans is the benefits paid by the plan to the retirees. The higher the level of benefits paid, ceteris paribus, plan assets decrease and future returns generated by the plan assets will decrease requiring larger contributions to replenish the coffers. Thus, one would expect higher levels of contributions to be required as benefits paid increase.
Return on Pension Assets (PENROA)
The relation between returns generated by the pension fund and contributions in not unambiguous. On the one hand, one can argue contributions to the pension fund and returns generated by the fund are substitutes for one another. The greater the return earned by the pension plan assets, the less need there is for contributions to cover benefits payable and a negative relationship between returns and contributions is expected.
On the other hand, and in the context of our paper, a more intriguing possibility exists. If, as opposed to paying out free cash flows, management seeks alternative investment possibilities to store the free cash flows, then the more attractive the alternative source the more likely management will invest there. If management uses pension plans as one of its possible avenues to store free cash flows, then the greater the investment returns from the plan assets, the more likely management would be to increase pension plan contributions. This argues for a positive relationship between contributions and pension plan returns.
(Insert Exhibit 1 about here)
Exhibit 1 presents summary statistics about our variables. The (average) pension contributions represent about one third of benefits paid to retirees, indicating that many firms were on pension holiday. Similarly, the funding status of the firm is positive on the average, with more firms having greater pension plan assets than pension obligations (PBO). Most sample firms had positive free cash flows (possibly due to our selection criteria), and positive actual return on pension plan assets, reflecting the favorable market conditions for many of the years included in the sample period.
Exhibit 2 presents the correlation matrix of our variables. The first line of the matrix provides information as to the association between contributions (CONTA) and our proposed explanatory variables. In all cases, but one, the relationships are in the expected direction and highly significant (level of significance at least .0001). The one noteworthy result is the highly significant positive relationship between contributions and returns earned by plan assets (PENROA). This result is contrary to the argument that as returns were higher, there would be less need for firms to contribute to plan assets. On the other hand, it is suggestive of pension plans being viewed as an alternative choice for management to invest the firm’s funds, as the positive correlation is consistent with higher returns leading to higher contributions being made to pension plans to take advantage of those returns.
(Insert Exhibit 2 about here)