To test hypotheses 1 and 1a, we examine the following relationship between contributions and our control variables.
The General Linear Model (GLM) applied to this model recognizes that each year may be different and allows the intercept and slope coefficients to vary from year to year.
Consistent with our hypotheses, we expect the coefficient b1y to be positive. Additionally, we expect that the positive relationship will be (driven by) greater for firms with positive free cash flows relative to those with negative free cash flows. As such, we have run the GLM above for our total sample as well as for our sample of firms partitioned on the basis of free cash flows being greater/less than zero. The results are contained in Exhibit 3. Column A presents results for the full sample and Columns B and C show results for the sample partitioned on the basis of positive and negative FCFA, respectively.
(Insert Exhibit 3 about here)
Looking at the “control variables” first, we find that the coefficients on FUNDA and BENA are in the expected direction and highly significant with significance levels below 0.01 for each year and for each partitioning of the data. For the TAX variable, however, the significance found in the univariate results (Exhibit 2) does not carry forward in the multivariate contest of equation (1), as the b4y TAX coefficients are not significantly different from zero12.
Moving to the variable(s) of interest in testing hypotheses 1 and 1a, we start with the overall sample (Column A). We find that the coefficients on FCFA are in the predicted direction (positive) for each of the six years and are significant at (at least) the 5% level of significance in three of the six years. In two of the other three years (1995 and 1996), the coefficients, while not significant at conventional levels, have t-values which are “respectable” falling in the 10% to 15% range.
When the sample is partitioned on the basis of positive/negative FCFA we find, as predicted, stronger results. Column B presents results for firms with positive FCFA. Again, the coefficients are positive in each year. However, we now find the coefficients to be significant at the 5% level in four of the six years and at the 10% level in five of the six years.
For firms with negative FCFA, (Column C), the b1y coefficients are smaller in each year. More importantly, the relationship between FCFA and CONTA falls apart as in four of the six years, the coefficient is negative and when it is positive the t-value is not significant.
The overall positive relationship between CONTA and FCFA as well as the fact that the relationship is driven by those firms for which FCFA > 0 provides evidence consistent with Myers and Majluf's financial slack theory. Our hypothesis that excess free cash flows are “stored” in pension plans as managers use pension plans to build financial slack was confirmed
Moving to the return on plan assets, PENROA, we find that the motivations for storing free cash flows in pension plans to be somewhat more complex. The univariate results found in Exhibit 2 are confirmed in the multivariate context. For the overall sample, the coefficients on PENROA are significantly positive in every year of the six years examined and for five of the six years when the sample is partitioned on the basis of positive/negative FCFA.
These results are suggestive of pension plans being used as more than just a place to store “excess” free cash flows. Rather, they imply that pension plans compete with other investment projects for a firm’s funds and that plans with higher returns get more of the funds. This view is strengthened by the fact that the relationship between CONTA and PENROA holds true even when FCFA are negative (Column C); i.e. firms will contribute to pension plans which perform well even when they do not have excess free cash flows.
Further evidence of this phenomenon is provided in Exhibit 4. In this exhibit, the sample is partitioned into four quartiles on the basis of FCFA and equation (1) is run for each quartile separately. The results13 indicate that, irrespective of the level of FCFA, contributions are positively related to pension plan returns. For each quartile, the coefficients on PENROA are generally significantly greater than zero. Interestingly, the weakest results hold for the quartile with the highest FCFA, as only three of the six years have coefficients that are significantly positive. As high FCFA is consistent with strong profitability, the weak relationship between CONTA and PENROA for this quartile may be indicative of firms who enjoy superior investment choices outside the pension plan and do not “need” the pension plan as an investment opportunity.
(Insert Exhibit 4 about here)
Utilization of financial slack: Research design and results
To test whether changes in pension plan contribution levels can provide information as to whether firms are utilizing the accumulated financial slack as predicted by Myers and Majluf, we first classify firms into two main groups. The first group consists of firms that reduced their pension plan contributions, following increases in contributions in the past three years. The second group consists of firms that continued to increase pension plan contributions, following such increases in the previous three years. If a firm does not have desirable NPV projects currently, but expects to have them in the future, it may continue to increase the contributions to the pension plan as a way to store cash. However, if the firm finds good investment projects that it wishes to undertake, it will reduce the current pension contribution levels, and will undertake those projects.
It is therefore likely that firms in the first group, i.e., those that began to decrease contribution levels after increasing them in the past, are firms whose desirable investment projects materialized. Thus, as compared to firms in the second group, i.e., those that continued to increase pension contribution levels, we would expect that firms in the first group will have greater current and near-future capital expenditures, and greater future accounting profits than firms in the second group (Hypothesis 2). Also, to the extent that market participants realize that firms in the first group have desirable investment opportunities, the future stock returns of firms in the first group should be higher than those of the second group (Hypothesis 3).
To test the above conjectures, we use the following variables:
INCCONTt = 1 if CONTt/OCFt>[SUM3(CONTt-1)/SUM3(OCFt-1)], 0 otherwise.
DECCONTt = 1 if CONTt/OCFt<[SUM3(CONTt-1)/SUM3(OCFt-1)], 0 otherwise.
where CONTt is the cash contribution to the pension plan in year t, OCFt is the operating cash flow in year t, and SUM3(Xt) is the sum of variable X for the most recent three years.
The variable INCCONTt is an indicator variable that takes the value of one when the firm’s current ratio of cash pension contributions to operating cash flows exceeds this average ratio in the prior three years. It indicates that current cash contribution levels have increased as compared to the recent past. In contrast, DECCONTt is an indicator variable that will have the value of one if the firm decreased current contribution levels as compared to the prior three years.
We can now define our two groups as ID (reversing) and II (maintaining), as follows:
ID = Firms where INCCONTt-1=1, and DECCONTt=1.
II = Firms where INCCONTt-1=1, and INCCONTt=1.
Firms in the ID group are those that had increased cash contributions levels in the past, but decreased them currently because of desirable investment projects. We posit that these firms are Myers-Majluf firms who are now reversing course and are beginning to invest. These firms are expected to have higher capital expenditures, profits and returns than firms in the II group, which are Myers-Majluf firms that still do not have desirable investment projects.
We further define the following variables:
GCAPEXt = The growth of capital expenditures in year t, i.e. capital expenditures in year t divided by the average capital expenditures in years t-3 through t-1.
ROAt = The return on assets in year t, defined as income before extraordinary items during year t, divided by total assets at the end of year t.
INDROAt = Industry-adjusted ROA at period t, estimated by ROAt minus the median ROAt in the same 4-digit SIC industry.
AROAt = The average of ROAt over the three year period t to t+2.
AINDROAt = The average of INDROAt over the three year period t to t+2.
ANNRETt = Annual stock return in year t, beginning April 1,year t through March 31, year t+1.
CUMRETt = Cumulative return (ANNRET) during the period April 1, year t, through March 31, year t+2.
SIZERETt = Size-adjusted annual return in year t. It is defined as ANNRETt minus the equally-weighted average return (ANNRETt) of all stocks that belong to the same size (market value of equity) decile at the beginning of year t.
CUMSIZERETt = Cumulative SIZERET during the period April 1, year t, through March 31, year t+2.
Based on Hypothesis 2, we expect to find that firms in the ID group will have higher capital expenditures (GCAPEX) and profitability (ROA and INDROA) than firms in the II group. Similarly, based on Hypothesis 3, we expect to find that the various return measures for the ID group to be higher than the corresponding measures for firms in the II group.
To test these hypotheses, we use all firms with non-negative pension plan contributions during the period 1991-1996. We further restricted firms to have positive operating cash flows at the year of analysis, because the increases or decreases of pension contribution levels were only relevant if the firms generated cash from operations, and had to make a decision on how it should be utilized.
The means and medians of the variables postulated by Hypothesis 2 are provided in Exhibit 5 Panel A for each of the two groups ID and II. The table also reports the results of two-sample t-tests and the Wilcoxon non-parametric tests for equality of means (medians) among the two groups. In a similar fashion, the return results for Hypothesis 3 are presented in Exhibit 5 Panel B.
(Insert Exhibit 5 about here)
Consistent with our hypotheses, the mean and median values for all of our variables are larger for the ID group than the II group. Because firms that decreased pension contributions are likely to have identified desirable investment projects, we would expect the growth of capital expenditures for both the current and following-year to be larger for the ID firms. Note that the differences are statistically significant for the current and following year at the 10% and 5% levels, respectively, for the non-parametric tests and at 10% using the t-test for GCAPEXt+1.
Also, consistent with Hypothesis 3, the means and medians of the ROA and the industry-adjusted ROA, both in the following year and over the following three years, are larger for the ID group than the II group. This is to be expected since the ID firms are likely to have found desirable investment projects that increase future profitability. Note that these differences are always statistically significant on the basis of the non-parametric Wilcoxon statistics.
Panel B presents results for Hypothesis 3. As expected, the ID firms have a higher average (median) annual return and a higher average (median) cumulative return in the years following decreases of pension plan contributions than the firms that continued to increase their pension contributions. Presumably, these results are driven by market participants, who realize that the ID firms began investing in desirable projects, having favorable future cash flows and security price implications. Note that the returns differences of the two groups are statistically significant in all cases, except for the cumulative size-adjusted returns.
Finally, we present one more variation of our results. In the Myers-Majluf world, the assumption of information asymmetry implies that managers possess superior knowledge (relative to outsiders) as to future positive NPV investment opportunities. If this is the case, then the market would (prior to utilization of financial slack) assess such companies as having zero (or low) growth opportunities. Firms with low growth opportunities are characterized as having Market to Book value of equity ratios (M/B) below 1 (see Penman (1992)).
Thus, to make the analysis more compatible with a Myers- Majluf world, we repeated the analysis for firms with M/B ratios below 1. The results are presented in Exhibit 6. As can be seen, there are much fewer observations for hypothesis testing, and as a result, most of the previous findings of Exhibit 5 are not as statistically significant. Still, we observe the same general phenomena; the ID firms tend to have greater investments in capital expenditures, greater future profitability, and better stock returns than the II firms. These results seem to be further confirmation of Hypotheses 2 and 3.
(Insert Exhibit 6 about here)
Summary and Conclusions
This study examined the corporate pension funding decision to see whether it was used by managers as a means of storing cash flows in a manner consistent with Myers and Majluf financial slack theory. The results indicate this to be the case both initially as pension plan contributions are being increased and subsequently when the growth in contribution levels abate.
The study finds a significantly positive relationship between free cash flows and pension plan contributions for firms with positive free cash flows, but not for those with negative cash flows. As pension plans provide high (after-tax) returns and are accessible, the results are indicative of managers using pension plans to build financial slack.
Furthermore, the results indicate that firms who “reverse” and begin to draw down their financial slack have characteristics that are consistent with the Myers and Majluf world. Such firms make relatively more investments that prove to be profitable and enjoy favorable stock returns. These characteristics indicate that the positive NPV investments, which the financial slack was being accumulated for, eventually materialized.
Ballester, Marta, Dov Fried and Joshua Livnat, “Pension Plan Cash Flows: Estimation and Motives”, Journal of Financial Statement Analysis, (Spring 1998), pp. 41-50.
Bodie, Zvi, J. O. Light, R. Morck and R. A. Taggart Jr., “Corporate Pension Policy: An Empirical Investigation”, Financial Analysts Journal, (September-October 1985), pp. 10-16
Bodie, Zvi, J. O. Light, R. Morck and R. A. Taggart Jr., “Funding and Asset Alklocation in Corporate Pension Plans: An Empirical Investigation”, Issues in Pension Economics, edited by Bodie, Zvi, Shoven, J. B. and D. A. Wise, University of Chicago Press, Chicago 1987.
Datta, Sudep, M. E. Iskandar-Datta and E. Zychowicz, “Managerial Self-interest, Pension Financial Slack and Corporate Pension Funding”, The Financial Review (November 1996), pp. 695-720
Jensen, Michael C. "Agency Costs Of Free Cash Flow, Corporate Finance, And Takeovers," American Economic Review, 1986, v76(2), 323-329.
Kallapur, Sanjay; “Dividend Payout Ratios As Determinants Of Earnings Response Coefficients : A Test Of The Free Cash Flow Theory”, Journal of Accounting and Economics (May 1994), pp. 359-375
Lang, Larry, Rene Stulz, Ralph A. Walkling, “A Test of the Free Cash Flow Hypothesis: The Case of Bidder Returns”, Journal of Financial Economics, (October 1991), pp. 315-35
McCabe, George and Ken C. Yook, “Jensen, Myers-Majluf, Free Cash Flow And The Returns To Bidders”, Quarterly Review of Economics and Finance” (Fall 1997), pp. 697-707
Mittelstaedt, H.F., “An Empirical Analysis of the Factors Underlying the Decision to Remove Excess Assets From Overfunded Pension Plans”, Journal of Accounting and Economics (June 1989), pp. 399-418
Myers, Stewart C., and Nicholas S. Majluf, “Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have,” Journal of Financial Economics (June 1984), pp. 187-221
Penman, Stephen H., “Return to Fundamentals”, Journal of Accounting, Auditing and Finance, (19992), pp. 465-483
Smith, Richard L. and Joo-Hyun Kim; “The Combined Effects Of Free Cash Flows And Financial Slack On Bidder And Target Stock Returns”,Journal of Business (April 1994), pp. 281-310
Stickney , Clyde P. .“Analyzing Postretirement Benefit Disclosures” , The Journal of Financial Statement Analysis (Fall 1995).
Szewcyk, S.H., Tsetsekos, G.P., Zantout, Z.; “The Valuation Of Corporate R&D Expenditures: Evidence From Investment Opportunities And Free Cash Flows”, Financial Management, (Spring 1996), pp. 105-110
Thomas, Jacob K., “Why Do Firms Terminate Their Overfunded Pension Plans”, Journal of Accounting and Economics (November 1989), pp. 361-398
White, Gerald I. , Ashwinpaul C. Sondhi and Dov Fried , The Analysis and Use of Financial Statements, John Wiley and Sons 1997.