1.29 times greater than that of the NSF/Census data in 1978, which is sig-
nificant at the 0.05 level (F-test), while it is significantly less than that of
the Compustat data in 1979. Clearly, the reversal of the direction of the
differences and the level of significance attained (0.05 in 1978 and 0.01 in
1979) suggest that R&D expenditures were reported very differently by
some firms in these two years.^5 The sample means (Table 7.2) further sug-
gest that 1979 was unusual for R&D reporting purposes, for the sample
as a whole. Firms tended to report higher R&D expenditures in the Com-
pustat format than they did in the NSF/Census survey, except for 1979
and 1982. The size of the difference was particularly great in 1982. Since
it is well known (National Science Foundation 1985) that the Securities
and Exchange Commission (SEC) Forms 10-K and 10-Q data used to de-
velop the Compustat data permit certain activities (e.g., engineering and
technical service) to be included in R&D that are excluded under the NSF
definitions, one would expect the Compustat means to be greater than the
NSF/Census means.^6
The industry-level data in Tables 7.1 and 7.2 suggest strong differences
between the two data sources in the petroleum industry from 1975 to
- The differences in R&D reporting in the petroleum industry could
result from the accounting treatment of exploration activities and other
noncash expenditures. The variances are significantly different for these
Financial Decision Estimation Results 187
TABLE 7.1 Comparison of R&D Expenditure Data for Matched Samples of
Firms: Compustat versus NSF/Census Data, 1975–1982. Test of Significance for
Differences in Sample Variances: Fratio
1975 1976 1977 1978 1979 1980 1981 1982Complete Sample F= 1.08 1.09 1.15 1.29 2.55 1.16 1.08 1.04
N= (188) (188) (188) (188) (180) (177) (162) (158)
Industries
Construction F= 1.50 1.50 1.50 1.67 2.00 1.33 1.33 2.00
N= (16) (16) (16) (16) (16) (16) (17) (17)
Petroleum F= 11.5 14.5 16 15 18** 1.50 1.25 1.20
N= (12) (12) (12) (12) (12) (11) (11) (11)
Machinery F= 1.17 1.00 1.08 1.15 1.06 1.06 1.05 1.09
N= (16) (16) (16) (16) (16) (16) (15) (15)
Electronics F= 1.29 1.19 1.18 1.83 1.23 1.06 1.12 1.09
N= (21) (21) (21) (21) (22) (22) (17) (15)
Drugs F= 1.08 1.31 1.40 1.57 1.37 1.47 1.50 1.44
N= (25) (25) (25) (25) (25) (24) (23) (23)
Chemicals F= 1.00 1.09 1.05 1.23 1.18 1.28 1.24 1.02
N= (24) (24) (24) (24) (24) (24) (23) (23)
*Significant at 0.05 level; Compustat variance > Census variance.
**Significant at 0.01 level; Compustat variance < Census variance.