a given sector expresses the total change in income in the region per unit
change in sales to final demand in that sector. As shown in Figure 4, the
direct income per dollar final demand in agriculture is $.04; however, the
total income per dollar final demand in agriculture is $0,133, The compu-
tation of the income multiplier assumes constant income per dollar output
relations in each sector in addition to the other assumptions of input-
output analysis. Similarly, employment multipliers are obtained from direct
employment per unit output in each sector and interdependence coefficients;
these express total employment change per unit sales to final demand, Out-
put, employment, and income multipliers are the traditional tools of input-
output analysis.
 A recent addition to applications of input-output analysis is in the
area of resource use and environmental impacts [8, 9], Resource and pol-
lution multipliers are computed similarly to the traditional economic
multipliers from information on direct resource and pollution per unit
output and interdependence coefficients for sectors,
 Now that the main types of multipliers obtainable from input-output
analysis have now been described, it is important at this point to review
the restrictive assumptions (and therefore limitations) of input-output
analysis. First, the coefficients depend on how the industries in an
economy are divided up among sectors; different sectoring may lead to
different multipliers if aggregated sectors are inhomogeneous, The most
important assumptions have to do with the use of constant coefficients,
For example, the use of constant technical coefficients to express require-
ments per unit of output of one industry from another implies a fixed
production technology with constant returns to scale (i,e,, doubling inputs
doubles output) and no substitution among inputs, This assumption is valid
in the short run close to the time period in which coefficients are measured,
but it becomes less reasonable over a longer period during which changes in
technology may be occurring.
 The use of constant local purchase coefficients in a regional model
requires the assumption of constant purchasing patterns as well as constant
technology, If imports are reduced due to new industries moving into the
region or local industries expanding, this will cause regional purchase
coefficients to change.