Cj. Liew et Dp. Robinson, Measuring dynamic economic effects under the constant-technology versus varying-technology assumptions, ANN REG SCI, 35(2), 2001, pp. 299-331
This study measures the time-period-specific industrial price and output ef
fects of cost-related variables (transportation cost, wage rate, and intere
st rate) by utilizing the Dynamic Variable Input-Output (VIO) model. The Dy
namic Variable Input-Output (VIO) model extends the static single regional
version of the MultiRegional Variable Input-Output (MRVIO) model which is a
partial general equilibrium model that incorporates the input-output model
.
By using the 15 sector industrial transaction table derived from the 1987 U
.S. Benchmark input-output table for the constant-technology assumption cas
e, and transaction tables derived from the 1987-1983 U.S. input-output tabl
es for the varying-technology assumption case, we estimate cost-related var
iable effects on industrial price and output that are spread over several y
ears. The dynamic price and output elasticities identify each period's impa
cts, and they add up to the static total price and output elasticities, res
pectively, when we adopt the constant-technology assumption. When we adopt
the varying-technology assumption, each period's impacts do not add up to t
he neat dynamic totals. This study also finds that the initial period's pri
ce and output elasticities of the Dynamic VIO model are exactly the same as
price and output elasticities of the static VIO model, thereby showing tha
t the static VIO model underestimates the price and output impacts.
Empirical results show that price elasticities are all positive for both ow
n and cross impacts. Empirical results also show that output elasticities a
re negative for own impacts but mixed in sign for cross impacts because of
the substituting behavior of firms and consumers. The distributions of both
price and output elasticities reveal that ripple effects vary among differ
ent industries, over different time periods, among the cost-related variabl
es, and between the two different technology assumptions. The distributions
of both price and output impacts are more apparent during the first four o
r five periods.
Hypotheses testings on the differences of mean elasticities between the two
cases of technology assumptions show that under 10% level of significance,
there are almost no differences in elasticities between the two cases of t
echnology assumptions. However, as we increase the significance level, the
total of five year periods' impacts show that they do differ under the two
technology assumptions. Consequently, we recommend the use of constant tech
nology for forecasting time horizons less than five years, and the use of v
arying-technology for forecasting time horizons longer than five years.