230
Journal of Applied Economics
US industry in the 1990s reveals major improvements over the RLS estimation
(Figure 2). All variables are significant, have the expected sign and the explanatory
power of the model is very high (89%). As expected, there is a negative and
significant relationship between LSENASA and the relative size of non-competitive
contracts with the coefficient of LSENASA equal to -0.42. The interpretation of this is
that a one US$ change in SENASA leads to a 0.42 cents change in the opposite
direction of NASAnc contracts. As a result, downward trends in the NASA space
expenditure post-Cold War are resulting in upward changes in the non-competitive
contracting value.
Table 2 . Modelling LNASAnc by OLS (1974 to 2003)
Variable |
Coefficient |
Std.Error |
t-value |
r2 |
Constant |
-0.78 |
3.15 |
-0.25 |
0.00 |
LSENASA |
-0.42 |
0.12 |
-3.39 |
0.31 |
LNASAtop10 |
1.92 |
0.58 |
3.31 |
0.30 |
s1994 |
0.58 |
0.05 |
12.41 |
0.86 |
R2 = 0.89, F(3, 26) = 71.12 [0.00] DW = 1.44; RSS = 0.26 for 4 variables and 30 observations.
The diagnostics reveal no problems with autocorrelation, ARCH effects, or normality of the
error term (probabilities in brackets):
AR 1- 2F( 2, 24) = 1.23 [0.31]
ARCH F( 1, 24)= 0.00 [0.95]
Normality Chi(2) = 0.13 [0.94]
Notes: LNASAnc: the logarithm of the value of the NASA non-competitive contracts as a
percentage of the total value awarded; LSENASA: the log of NASA space appropriations at
constant 1999 prices; LNASAtop10: the log of NASA contracted value to the top 10 firms
as a percentage of the total appropriations; s1994: step dummy variable capturing
consolidation-effects of the US space industry in the mid-1990s. Data sources: NASA
(1983) to (2004a) and NASA (2004b).
To measure the impact of the structural break taking place around the mid-
1990s on the level of competition in NASA contracts, we can re-estimate the
relationship of Table 2 without the step dummy variable with sample range from
1974 to 1997 and forecast the period from 1998 to 2003. The numerical difference
between the forecasted and the actual value is an approximation of the impact of
the industrial consolidation on the level of competition in NASA’s contracting
behaviour. The forecasts are thus expected to be significantly lower than the
actual value of LNASAnc throughout the forecast period. The performance of the
relevant estimation was much poorer than that in Table 2 (all right hand side