Cash flow technologies

Past research shows a significant market response to announced increases in R&D expenditures (Woolridge, 1988). However, the reaction depends on the firm's industry. The reaction is significantly positive for firms operating in high-technology industries but significantly negative for firms in low-technology industries (Chan, Martin, and Kensinger, 1990, and Zantout and Tsetsekos, 1994). This suggests the investment opportunities hypothesis that R&D investments by firms with promising growth opportunities are generally worthwhile, whereas other R&D investments may be wasteful.

We provide confirmatory evidence of this market reaction using a different measure of investment opportunities, Tobin's q. Tobin's q is the ratio of the market value of the firm's assets to their replacement cost. Chan, Martin, and Kensinger (1990) and Zantout and Tsetsekos (1994) proxy growth opportunities using a binary (high, low) technology variable. Tobin's q may be a better measure than a binary proxy because many firms operate in multiple industries, which makes industry classification problematic. Also, even if a firm operates in a single mature industry, it may have differentiating competencies to create growth.

We also examine whether free cash flow can explain cross-sectional differences in the market's response to R&D increases. Jensen (1986) argues that managers will invest free cash flow in wasteful investments rather than pay it out to shareholders. The potential agency costs of R&D investments are arguably higher, therefore, for high-free-cash-flow firms. On the other hand, R&D investments by low-free-cash-flow firms increase the chance the firm will seek new external financing. New external financing provides monitoring, and the firm's willingness to undergo such monitoring may be a favorable signal. Therefore, we would expect that announcement-period abnormal Returns for increases in R&D will be inversely related to free cash flow.

Free cash flow agency costs may depend on the firm's investment opportunities. Firms with relatively more growth opportunities are less likely to have free cash flow. Therefore, the potential agency costs of R&D expenditures are lowest for low-free-cash-flow/high-q firms. The converse holds for high-free-cash-flow/low-q firms. Accordingly, the free cash flow hypothesis predicts positive (negative) announcement-period abnormal returns for low-free-cash-flow/high-q (high-free-cash-flow/low-q) firms.

Consistent with past work, we find a positive relation between the market's response to announcements of increases [TABULAR DATA FOR TABLE 1 OMITTED] in R&D and Tobin's q. This relationship holds after taking into account the effect of other potential explanatory variables. We do not find evidence that free cash flow explains cross-sectional differences in abnormal returns. However, we do find a positive relation between the market's response and the announcing firm's debt ratio and level of institutional ownership of its equity. This finding is consistent with a broad interpretation of the free cash flow hypothesis. A high debt ratio implies precommitted future cash flows and greater institutional oversight, both of which can lower the expected agency costs of free cash flow.

I. Sample Selection and Description

We obtained a preliminary sample of R&D increase announcements from the Dow Jones News Retrieval database and applied four screening criteria: 1) The announcement is an initial announcement of a future plan to increase R&D expenditures; 2) the announced plan does not involve funding from customers or from government contracts; 3) the announced plan does not pertain to a joint venture or a cooperative agreement with another firm; and (4) the announcing firm has sufficient data on the Center for Research on Security Prices (CRSP) tape.

Our final sample consists of 252 announcements made between June 1979 and December 1992 by 121 NYSE- and AMEX-listed firms. The announcements are fairly evenly distributed on a per-year basis, with years ranging from 4.4% to 10.7% of the sample. Table 1 reports the sample's industry classification distribution.

Tobin's q is computed using the Lindenberg and Ross (1981) algorithm with the modifications described in Lang, Stulz, and Walkling (1989).(1) The test variable for q is the average q for the three fiscal years prior to the announcement.(2) The cash flow ratio is defined for the fiscal year before the announcement. It is operating income before depreciation minus interest expense, taxes, preferred dividends, and common dividends, divided by the book value of total assets (Lang, Stulz, and Walkling, 1991).

[TABULAR DATA FOR TABLE 2 OMITTED]

Of the 252 announcements, 121 have sufficient COMPUSTAT data to calculate q and cash flow ratios.(3) The mean q is 1.52, with a standard deviation of 1.31 and a range of 0.33 to 6.80. The mean cash flow ratio is 8.67%, with a standard deviation of 3.58% and a range of 1.12% to 18.19%.

II. Empirical Results

This section provides the empirical results, which support the investment opportunities hypothesis but not the free cash flow hypothesis.

A. Average Shareholder Wealth Effects

Table 2 reports the cumulative average abnormal stock returns for sample firms based on standard event-study methodology. Sample firms realize, on average, a significantly positive abnormal two-day announcement-period stock return of 0.477%. This result is not caused by outliers. Of the abnormal returns, 60.7% are positive. No significant abnormal returns are observed preceding and following the announcement period.

B. Abnormal Stock Return Differences Between Subsamples

Following Lang, Stulz, and Walkling (1991), we divide our sample by firms with high and low q (above vs. below one) and by firms with high and low cash flow (above vs. below the sample median, 8.6%). Table 3 reports cumulative average abnormal two-day announcement-period returns for the subsamples. In Panel A, the sample is partitioned by q and the cash flow ratio. In Panel B, the sample is partitioned jointly by q and the cash flow ratio. T-test results and nonparametric Wilcoxon tests for differences between subsamples are also given.

As can be seen in Panel A, high-q firms have a significantly positive average abnormal return of 0.929%. In contrast, low-q firms have an insignificant average abnormal return of -0.16%. The difference is significant and consistent with the investment opportunities hypothesis.

Panel A also reports comparisons based on the cash flow ratio. There is no significant difference between high- and low-cash-flow firms.

The results shown in Panel B provide a direct test of the free cash flow hypothesis, by examining the joint effect of cash flow and q. According to the free cash flow hypothesis, low-cash-flow/high-q firms have the lowest potential agency costs connected with R&D expenditures. This subsample has a significantly positive average abnormal return of 1.11%, the highest among the four subsamples. However, the difference between low-cash-flow/high-q and high-cash-flow/high-q firms is not significant.

Also according to the free cash flow hypothesis, high-cash-flow/low-q firms have the largest potential agency costs connected with R&D expenditures. This subsample has an insignificantly positive average abnormal return. [TABULAR DATA FOR TABLE 3 OMITTED] Moreover, there is again no significant difference between low-cash-flow/low-q and high-cash-flow/low-q firms.

The results in Panel B, then, do not support the free cash flow hypothesis. They do, however, provide some support for the investment opportunities hypothesis. The differences between high- and low-q firms are of the expected sign for both high- and low-cash-flow firms, although the difference is significant only for low-cash-flow firms.

In summary, the evidence in Panels A and B of Table 3 supports the investment opportunities hypothesis. However, consistent with Vogt's (1994) finding, the results do not support the free cash flow hypothesis.

C. Cross-Sectional Regression Analysis of the Abnormal Stock Returns

Table 3 results do not control for other factors that could influence the market's response to announcements of R&D expenditure increases. In this section, we report four cross-sectional regressions, one of which tests the investment opportunities and free cash flow hypotheses while controlling for several other factors.

Table 4 shows the regression results. They are consistent with those of Table 3. The coefficient for q is significantly positive for each model in which that variable is included. The coefficient for the cash flow ratio is insignificant. The intercept in Model 3 represents low-cash-flow/high-q firms and is significantly positive. As in Table 3, firms with a low cash flow ratio differ significantly from those with high versus low q in Model 3.