27 us gdp forecast
The US GDP will be released tomorrow. Previously it was 2.1% but the consensus forecast now sits at negative 4%.
Is this realistic or even extreme? In this video we find out and we also look at what that might imply for South Africa.
Code used in this video
SAGDP = C00221;
USGDP = GDPC1;
SApa = 100 * pa( SAGDP );
USpa = 100 * pa( USGDP );
SApal = SApa[ lastDate( SApa ) ];
USpal = USpa[ lastDate( USpa ) ];
SAdpa = d( SApa );
USdpa = d( USpa );
SAm = average( SAdpa );
SAs = std( SAdpa );
USm = average( USdpa );
USs = std( USdpa );
cMat = matrix( 11, 3 );
USf = 0;
while( USf >= -10,
USd = USf - USpal,
USz = ( USd - USm ) / USs,
SAz = USz,
SAd = SAz * SAs + SAm,
SAf = SAd + SApal,
r = 1 - USf,
cMat[ r; 1 ] = USz,
cMat[ r; 2 ] = USf,
cMat[ r; 3 ] = SAf,
USf = USf - 1 );
print( cMat );
# Test correlations
# These lines will only work in V2.6.2
# or above
# If you get an error here please upgrade
# to the latest version
ols( z( SAdpa ), 1, z( USdpa ) );
ols( z( SAdpa ), 1, z( USdpa(-1) ) );
ols( z( SAdpa ), 1, z( USdpa(-2) ) );
ols( z( SAdpa ), 1, z( USdpa ), z( SAdpa(-1) ) );
ols( z( SAdpa ), 1, z( USdpa(-1) ), z( SAdpa(-1) ) );
ols( z( SAdpa ), 1, z( USdpa(-2) ), z( SAdpa(-1) ) );
# Make a report
reportDelete( "r" );
reportAdd( "r", "GDP comparison report", "summary", "plain", USpa, SApa );
Note this code assumes that the US and SA GDPs have been
downloaded from FRED and STATSSA already. It should be
available as GDPC1 and C00221 for the code to work.
Output
Calculating 4/28/20 6:42 AM
Loading series for frequency Quarterly
Adding new series GDPC1
Adding new series C00221
11x3 matrix
-0.4926 0.0000 -2.7374
-0.7231 -1.0000 -3.3241
-0.9535 -2.0000 -3.9108
-1.1840 -3.0000 -4.4974
-1.4144 -4.0000 -5.0841
-1.6449 -5.0000 -5.6708
-1.8753 -6.0000 -6.2575
-2.1058 -7.0000 -6.8442
-2.3362 -8.0000 -7.4308
-2.5667 -9.0000 -8.0175
-2.7971 -10.0000 -8.6042
OLS model
---------
z(SAdpa) = f( 1, z(USdpa) )
Sample 1993Q3 - 2019Q4
Overview
NObs +106.0000 DF1 +1.0000 DF2 +104.0000
Y Avg -8.7736E-51 MSE +0.9810 RMSE +0.9905
Variance
Total +105.0000 Explained +2.9722 Unexplained +102.0278
Goodness of fit
R2 +0.0283 R2 Adjust +0.0190 F Prob +0.0847
LHood +1.4485E+84 LogLHood -148.3928 F Value +3.0296
Information criteria
AIC +300.7856 AICc +300.9021 BIC +306.1125
Residuals
Skewness -0.0205 Kurtosis (0) +0.1737
Autocorr -37.4084 Durbin-Watson +2.7228
Jarque-Bera (normal dist) +0.1380 Prob +0.9333
Breusch-Godfrey (autocorr) +14.2951 Prob +0.0002
Breusch-Pagan (homoscedast) +1.0028 Prob +0.3166
ARCH effects +2.3698 Prob +0.1237
Coefficients
z(SAdpa) Coefficient Std Dev t Value p Value
1 +0.0008 +0.0962 +0.0087 +0.9930
z(USdpa) +0.2784 +0.1599 +1.7406 +0.0847
OLS model
---------
z(SAdpa) = f( 1, z(USdpa(-1)) )
Sample 1993Q3 - 2019Q4
Overview
NObs +106.0000 DF1 +1.0000 DF2 +104.0000
Y Avg -8.7736E-51 MSE +1.0031 RMSE +1.0016
Variance
Total +105.0000 Explained +0.6735 Unexplained +104.3265
Goodness of fit
R2 +0.0064 R2 Adjust -0.0031 F Prob +0.4144
LHood +1.4696E+85 LogLHood -149.5736 F Value +0.6714
Information criteria
AIC +303.1472 AICc +303.2637 BIC +308.4741
Residuals
Skewness -0.1163 Kurtosis (0) +0.3916
Autocorr -39.3545 Durbin-Watson +2.7456
Jarque-Bera (normal dist) +0.8991 Prob +0.6379
Breusch-Godfrey (autocorr) +15.6609 Prob +0.0001
Breusch-Pagan (homoscedast) +0.3122 Prob +0.5763
ARCH effects +2.3762 Prob +0.1232
Coefficients
z(SAdpa) Coefficient Std Dev t Value p Value
1 -0.0001 +0.0973 -0.0008 +0.9994
z(USdpa(-1)) +0.1323 +0.1614 +0.8194 +0.4144
OLS model
---------
z(SAdpa) = f( 1, z(USdpa(-2)) )
Sample 1993Q3 - 2019Q4
Overview
NObs +106.0000 DF1 +1.0000 DF2 +104.0000
Y Avg -8.7736E-51 MSE +1.0051 RMSE +1.0025
Variance
Total +105.0000 Explained +0.4698 Unexplained +104.5302
Goodness of fit
R2 +0.0045 R2 Adjust -0.0051 F Prob +0.4957
LHood +1.8001E+85 LogLHood -149.6770 F Value +0.4675
Information criteria
AIC +303.3539 AICc +303.4705 BIC +308.6808
Residuals
Skewness -0.0926 Kurtosis (0) +0.3684
Autocorr -36.8493 Durbin-Watson +2.6937
Jarque-Bera (normal dist) +0.7365 Prob +0.6919
Breusch-Godfrey (autocorr) +13.5127 Prob +0.0002
Breusch-Pagan (homoscedast) +0.0591 Prob +0.8079
ARCH effects +2.8418 Prob +0.0918
Coefficients
z(SAdpa) Coefficient Std Dev t Value p Value
1 +0.0008 +0.0974 +0.0083 +0.9934
z(USdpa(-2)) +0.1095 +0.1602 +0.6837 +0.4957
OLS model
---------
z(SAdpa) = f( 1, z(USdpa), z(SAdpa(-1)) )
Sample 1993Q4 - 2019Q4
Overview
NObs +105.0000 DF1 +2.0000 DF2 +102.0000
Y Avg -0.0094 MSE +0.8761 RMSE +0.9360
Variance
Total +104.0190 Explained +14.6520 Unexplained +89.3670
Goodness of fit
R2 +0.1409 R2 Adjust +0.1240 F Prob +0.0004
LHood +2.0157E+78 LogLHood -140.5469 F Value +8.3616
Information criteria
AIC +287.0938 AICc +287.3315 BIC +295.0557
Residuals
Skewness -0.1008 Kurtosis (0) +0.1078
Autocorr -7.2717 Durbin-Watson +2.1533
Jarque-Bera (normal dist) +0.2223 Prob +0.8948
Breusch-Godfrey (autocorr) +5.7298 Prob +0.0167
Breusch-Pagan (homoscedast) +2.8489 Prob +0.2406
ARCH effects +0.0445 Prob +0.8328
Coefficients
z(SAdpa) Coefficient Std Dev t Value p Value
1 -0.0082 +0.0913 -0.0902 +0.9283
z(USdpa) +0.2091 +0.1524 +1.3719 +0.1731
z(SAdpa(-1)) -0.3356 +0.0921 -3.6425 +0.0004
OLS model
---------
z(SAdpa) = f( 1, z(USdpa(-1)), z(SAdpa(-1)) )
Sample 1993Q4 - 2019Q4
Overview
NObs +105.0000 DF1 +2.0000 DF2 +102.0000
Y Avg -0.0094 MSE +0.8734 RMSE +0.9346
Variance
Total +104.0190 Explained +14.9276 Unexplained +89.0914
Goodness of fit
R2 +0.1435 R2 Adjust +0.1267 F Prob +0.0004
LHood +1.4665E+78 LogLHood -140.3848 F Value +8.5452
Information criteria
AIC +286.7696 AICc +287.0072 BIC +294.7315
Residuals
Skewness -0.2446 Kurtosis (0) +0.1972
Autocorr -8.2739 Durbin-Watson +2.1772
Jarque-Bera (normal dist) +1.1828 Prob +0.5535
Breusch-Godfrey (autocorr) +6.9472 Prob +0.0084
Breusch-Pagan (homoscedast) +1.0252 Prob +0.5989
ARCH effects +0.0004 Prob +0.9847
Coefficients
z(SAdpa) Coefficient Std Dev t Value p Value
1 -0.0079 +0.0912 -0.0865 +0.9312
z(USdpa(-1)) +0.2272 +0.1531 +1.4844 +0.1408
z(SAdpa(-1)) -0.3751 +0.0925 -4.0532 +0.0001
OLS model
---------
z(SAdpa) = f( 1, z(USdpa(-2)), z(SAdpa(-1)) )
Sample 1993Q4 - 2019Q4
Overview
NObs +105.0000 DF1 +2.0000 DF2 +102.0000
Y Avg -0.0094 MSE +0.8801 RMSE +0.9381
Variance
Total +104.0190 Explained +14.2466 Unexplained +89.7724
Goodness of fit
R2 +0.1370 R2 Adjust +0.1200 F Prob +0.0005
LHood +3.2129E+78 LogLHood -140.7846 F Value +8.0935
Information criteria
AIC +287.5691 AICc +287.8067 BIC +295.5310
Residuals
Skewness -0.2039 Kurtosis (0) +0.1530
Autocorr -5.9246 Durbin-Watson +2.1240
Jarque-Bera (normal dist) +0.8063 Prob +0.6682
Breusch-Godfrey (autocorr) +4.4277 Prob +0.0354
Breusch-Pagan (homoscedast) +0.6947 Prob +0.7066
ARCH effects +0.3123 Prob +0.5763
Coefficients
z(SAdpa) Coefficient Std Dev t Value p Value
1 -0.0087 +0.0916 -0.0950 +0.9245
z(USdpa(-2)) +0.1803 +0.1517 +1.1887 +0.2373
z(SAdpa(-1)) -0.3607 +0.0919 -3.9266 +0.0002
Document changes
Deleted r [report]
Added r [report]
Calculation done 4/28/20 6:42 AM
The output remarkably shows that the US GDP is much more volatile than that of South Africa. Looking at the report one notes that US GDP has been more stable since the mid 1980s. If the US GDP is restricted to this time period, replacing the second line
USGDP = GDPC1;
with
USGDP = limit( dateIndex( 1985, 1, 1 ), dateIndex( 2100, 1, 1 ), GDPC1 );
then the US and SA standard deviations are more or less the same and the table becomes
11x3 matrix
-0.8143 0.0000 -3.5562
-1.1990 -1.0000 -4.5358
-1.5838 -2.0000 -5.5154
-1.9686 -3.0000 -6.4949
-2.3534 -4.0000 -7.4745
-2.7382 -5.0000 -8.4541
-3.1229 -6.0000 -9.4336
-3.5077 -7.0000 -10.4132
-3.8925 -8.0000 -11.3928
-4.2773 -9.0000 -12.3723
-4.6621 -10.0000 -13.3519
Here the -4% consensus figure is much more extreme as it weighs in at a standardised value of -2.4 and suggests a SA value of -7.5%.