228
Journal of Applied Economics
contracts, the variables that are expected to affect it are the proportion of contracts
awarded to the top contractors through time and NASA space appropriations
(SENASA). Despite NASA’s stated pro-competitive policy, industrial consolidation
could therefore become a key factor in explaining an increasing number of contracted
values on a non-competitive basis as a percentage of the total appropriations
(NASAnc- the time series includes follow-on contracts awarded on a non-competitive
basis). In testing this hypothesis, it is important to consider the impact “mega-
mergers” of the mid-1990s had in the relationship between NASAnc, SENASA and the
value of the contracts awarded to the top 10 NASA contractors as a percentage of
the total (NASAtop10).
The econometric tests were initially performed using recursive least squares
(RLS), a method whose results are similar to OLS, but in addition tests for structural
breaks (see Doornik and Hendry 1995, and Figure 2 for details). The variables used
for the estimation were all in logarithmic form to help reduce heteroskedasticity
and normalise variables with very differently scaled data to obtain meaningful
elasticities from the estimations (LNASAnc, LSENASA and LNASAtop10 are respective
the logs of NASAnc, SENASA and NASAtop10). The data sources used for the
empirical analysis are NASA annual procurement reports (NASA 1983 to 2004a)
and NASA (2004b). The sample was chosen to start from 1974. This was due to the
fact that during the late 1960s massive appropriations to NASA were followed by
sizeable reductions in the early 1970s which meant that this era’s budgetary and
contract behaviour was atypical. This was because NASA’s original purpose of
existence, the Apollo program to send the man to the moon before the end of the
1960s, was successful. This period is characterised by the agency’s set-up costs
and massive, Apollo program-specific budgetary appropriations that were set to
decline post-1969 when the first successful mission to the moon was accomplished
and continued to do so until the mid-1970s when the program terminated. The use
of RLS reveals the presence of a structural break in 1994 (Figure 2), which leads to
the use of a step dummy variable (s1994) and a re-evaluation of the relationship
using OLS in Table 2.
The step dummy variable is used because the size of the consolidation is only
partially captured by LNASAtop10. This is because the list of NASA’s top
contractors does not take into consideration firms, but establishments, which
means that it does not fully capture the concentration of contracts to consolidated
firms with several divisions.
As Table 2 indicates, the performance of the re-estimated model with the
incorporation of the step dummy variable to account for the consolidation of the