David F. Hendry 47transformations were entailed by the pre-analysis, the efficiency gains over my pre-
vious “handcrafted” study are huge. In addition, hypotheses that were previously
imposed or sequentially investigated can be evaluated jointly.
1.7 Revisiting the “experiment in applied econometrics”
“If I wasn’t real,” Alice said – half-laughing though her tears, it all seemed
so ridiculous – “I shouldn’t be able to cry.”
“I hope you don’t suppose those are real tears?” Tweedledum interrupted
in a tone of great contempt. (Lewis Carroll, 1899)The recent huge increases in the prices of many foods makes the exercise of
re-examining US food expenditure over 1931–89 (based on the update of Tobin,
1950, by Magnus and Morgan, 1999) of more than just historical interest. If the
various price and income elasticities estimated below are approximately correct,
then substantial responses can be anticipated (indeed, could this be the long-
sought solution to society’s burgeoning obesity problem?). Hendry (1999) sought
to explain why other contributors to Magnus and Morgan (1999) had found their
models were non-constant over the combined inter-war and post-war samples, so
had eschewed modeling the 1930s data. Impulse dummies for a food program
and post-war de-rationing allowed a constant equation to be developed over the
sample 1931–89. While an indicator variable is a crude level of measurement, the
converse strategy of not modeling major institutional interventions seems even less
attractive. Theory and common sense suggest that food programs and switches in
rationing matter; but few theory models allow for such factors in a way suitable for
empirical implementation (although the original analyst of this data also published
on rationing in Tobin, 1952).
The per capita variables are as follows (lower case denotes logs):
ef: constant-price expenditure on food
pf−p: real price of food
e: total constant-price expenditure
s=logY−logE: (an approximation to the savings ratio)
a: average family size.Figure 1.6 shows the time series, and reveals considerable changes over the
period. After falling sharply at the commencement of the Great Depression, both
ef anderise substantially till World War II, fall after, then resume a gentle rise
(panels (a) and (c)), soef,tis vastly more volatile pre-war (panel (e)) (ethas a
similar but less pronounced pattern). Next,pf−pis quite volatile till after the war,
then is relatively stable (panel (b)), whereas the dramatic rise insfrom “forced
saving” during the war is manifest (panel (d)).
The earlier study of a cointegrated VAR for the system established thate,s, and
pf−pwere weakly exogenous in the food demand equation. Here, the general
conditional model allowed for two lags on each ofef,e,pf−p,sand one lag