the effect of defense spending on the output of real gross domestic product in the united states...
TRANSCRIPT
The Effect of National Defense Spending on Real Gross Domestic
Product in the United States 1947-2011
Andrew E. ArterJames J. Jozefowicz
Indiana University of PennsylvaniaDepartment of Economics
Previous Literature
• Smith and Tuttle (2007)
• 1947:Q2 to 2004:Q4
• No Positive Correlation
• Trade-Offs
Previous Literature
• Atesoglu (2007)
• 1948:Q1 to 2007:Q1
• Positive Correlation
• Sensitive to previous literature
Data
• Sources: Federal Reserve of St. Louis (FRED), Bureau of Economic Analysis
• Quarterly Data: 1947:Q2 - 2011:Q2
• AR(1)
• All Real
• OLS
Variables
Variable Explanation Expected Sign
GDPD GDP Deflator (Quarterly % Change)
?
NDEF Real Non-Defense Spending +
DEF Real Defense Spending +
PCON Real Personal Consumption Expenditures
+
INV Real Gross Private Domestic Investment
+
INT Real Aaa Interest Rate +
URATE Unemployment Rate (Quarterly % Change)
-
Variable Explanation
GDP Real Gross Domestic Product
Dependent Variable
Independent Variables
Descriptive StatisticsVariable Mean St. Dev. Max Min
GDPD 50.28674 32.57418 112.1180 12.59800
NDEF 1225.584 392.3076 1849.986 389.8640
DEF 242.4568 213.9279 831.3000 17.50000
PCON 4312.429 2571.264 9392.700
1149.700
INV 886.9012 612.4344 2266.300 167.1000
INT 6.703800 2.833538 15.01000 2.536667
URATE 5.721401 1.641089 10.70000 2.600000
Descriptive Statistics
• Number of Observations: 256• Maximum DEF: $831.3 Billion (2010:Q3)• Minimum DEF: $17.5 Billion (1947:Q3)• St. Deviation DEF: $213.9 Billion • Maximum NDEF: $1.85 Trillion (2002:Q3)• Minimum NDEF: $389.9 Billion (1947:Q4)• St. Deviation NDEF: $392.3 Billion
Descriptive Statistics
• Maximum Real GDP: $13.33 Trillion (2007:Q4)• Minimum Real GDP: $1.77 Trillion (1947:Q3)• Maximum PCON: $9.39 Trillion• Minimum PCON: $1.15 Trillion• Maximum INV: $2.27 Trillion• Minimum INV: $167.1 Billion
Econometric Issues
• Serial Correlation (D.W > 1.789)
• AR1
• Delayed Effects of Spending
• Insignificant Variables
Variable/Stat
Model 1 Model 2 Model 3 Model 4 Model 5
Constant -31.749(-0.145)
39.457(0.225)
71.051(0.565)
383.539(1.499)
262.348(1.512)
GDPD -3.211(-0.654)
-1.862(-0.400)
OMITTED 8.186(2.384)
7.237(2.318)
NDEF 0.658(7.758)
0.657(7.778)
0.662(7.949)
0.699(8.061)
0.718(8.297)
DEF 0.954(6.302)
0.947(6.346)
0.945(6.597)
0.968(5.885)
0.949(5.742)
PCON 0.825(20.735)
0.820(21.969)
0.829(21.971)
0.896(25.229)
0.909(26.399)
INV 0.875(25.765)
0.876(26.069)
0.874(24.791)
0.879(25.383)
0.902(28.704)
INT 2.905(0.815)
OMITTED 2.361(0.676)
2.681(0.729)
3.959(1.096)
URATE -8.367(-2.272)
-8.859(-2.452)
-8.337(-2.275)
-5.805(-1.576)
OMITTED
Durbin-Watson
2.006 1.990 1.996 1.922 1.859
Adjusted R-Squared
0.999973 0.999974 0.999973 0.999972 0.999972
Time Trend
9.676(2.911)
9.059(3.017)
7.744(4.752)
OMITTED 7.924(2.693)
Summary of Results
• Model 2
• Signs of coefficients are robust
• Omitted TIME and URATE show significant changes
• Omitted INT and GDPD show insignificant changes
Summary of Results• All Independent Variables are statistically
significant except for INT
• Durbin-Watson Stat > 1.789 across all models
• PCON and INV are significant contributors
• DEF and NDEF are both significant
• DEF is positively correlated
Conclusion
• Definite Positive Correlation
• Regression is not serially correlated and explains a large part of the story
• Results are coherent with Atesoglu (2007)
• PCON and INV have a significantly bigger effect than Government Spending (NDEF + DEF)
Lessons
• DEF is not a waste
• Relative size of spending
• Stimulative action should revolve around PCON and INV for greatest GDP output
• Trade-Offs always exist