Economic Impact Multipliers for Kansas
"Kansas Business Review" Vol 12, No. 3, Spring 1989
David Burress is an assistant professor in the economics department at the University of Kansas and a research associate
at the Institute for Public Policy and Business Research.
At one time, economic multipliers were arcane technical devices used by academic economists; now they appear frequently in
the business news, in political discourse, and even on the front page of the local newspaper. The new multipliers reported
in this paper are often smaller than previously published multipliers for Kansas, even though the new multipliers account
for more effects. They were derived using the Kansas Long-Term Model (or KLTM), a dynamic input-output model under
development at the Institute for Public Policy and Business Research. In this paper, I will discuss the significance of
the new multipliers, their uses, and limitations.
The range of results is summarized in Table 1. As shown, a typical or median income multiplier is
around 0.46; a typical output multiplier is around 1.6; a typical wage-wage multiplier is around 1.9; and a typical
employment multiplier is around 1.9. As the table shows, however, multipliers for individual industries can vary over a
substantial range. There are several reasons for treating these extreme values with caution.
It sometimes seems that the bigger a multiplier is, the more often it is quoted.(1) In any case, some distinctly
one-sided political and economic motives encourage the propagation of exaggerated multipliers.
In particular, economic multipliers are used to justify public concessions to private industry. Such assistance for
business may include land acquisition, new roads and sewers, job training programs, subsidized loans, and tax
incentives.(2) The extent of public concessions is determined through bargaining between government and industry,
and in the course of the bargaining those who stand to gain most from the new enterprise have a natural tendency to
inflate the relevant multipliers.(3)
The inflation of multipliers may stem less from venality than from an innate optimism, which seems to be necessary in the
risky business of development. Since multipliers are costly to measure, of uncertain accuracy, varied in their meanings,
and multifarious in their origins, a convenient range of multiplier values is always available; discriminating users are
free to choose the best values for their purposes.
This paper will add to the low range of multiplier values, rather than to their usefulness to optimists. That is, these
multipliers are significantly smaller than those that tend to appear in the press. While they do not approach perfect
accuracy, our results do provide good evidence against the routine use of large multipliers.
Multipliers translate a known (or assumed) direct effect into an estimated total impact equal to the direct
effect plus an indirect effect. For example, 100 jobs in a new manufacturing plant (the direct effect) might lead
indirectly to 60 more jobs in the local service sector (the indirect effect). In this case, the total impact would be 160
jobs, and the employment multiplier would be 1.6.
The indirect effects may result from several different kinds of linkages. Some examples would be new sales to the
manufacturing plant, sales to the new workers, additional sales to the sellers, and even expanded sales to an expanded
government made necessary by the expanded economy. The indirect effects also may include investment and the migration of
Using multipliers correctly depends on correctly identifying the direct effects. According to the export-base theory, a
region's economy depends mainly on what its businesses sell to other regions (quantified as the direct effect) and how the
income from sales flows through the regional economy (described by multipliers).(4) The key point here is to distinguish
changes in sales to other regions from economic activity in general. (Other possible applications of multipliers are
For example, suppose a new retail store is built in downtown Wichita. Although Table 2 includes
multipliers for Kansas retailing, it would be wrong to apply them to the new store's sales: the new retail store will
probably sell almost exclusively to Kansans, so it will probably take business from other Wichita retailers. The indirect
effect approximately cancels the direct effect; the true multiplier is very close to zero. In order to correctly apply the
Kansas retailing multipliers in Table 2, one must first identify a bona fide increase in sales to non-Kansans.(5)
In addition to increased sales to non-Kansans, there are two less common situations where these multipliers can be used. A
second use of multipliers has to do with import substitution -- replacing goods purchased from non-Kansans with
goods produced within the state. Under reasonable assumptions(6), the multipliers reported in this paper apply to
an import substitution event as well as to an expansion of the export base.
A third use of multipliers has to with income transfers and taxation across state boundaries. For example, about 13
percent of Kansas personal income comes from Social Security payments. According to the income-income multiplier for
transfer payments in Table 2, each additional dollar of Social Security income received in
Kansas leads to an additional indirect income of fifty cents in Kansas. Consequently, the total effect of Social Security
payments amounts to some 20 percent of Kansas personal income. The same principle applies in reverse to Social Security
taxes and other federal taxes. An increase of one dollar in federal taxes paid by Kansans leads to a total reduction of
around $1.50 in Kansas disposable income.
Thus, multipliers apply to negative changes as well as to positive changes. Reductions in Kansas interstate export sales,
reductions in transfer payments, and the replacement of Kansas products by imports all have negative effects on the Kansas
economy that can be estimated using these multipliers.
Using multipliers correctly, as with everything else in applied economics, depends on taking them with a large dose of
ceteris paribus. For example, to estimate the changes in the Kansas economy resulting from an increase in Kansas aircraft
sales using these multipliers, one must first assume that federal taxes and transfer payments in Kansas and sales of all
other goods to non-Kansans remain constant. One must also assume that spending patterns do not change: the same fraction
of income is spent on imports out of new income as was spent out of old income, and within the state each economic unit
spends the same fraction of its resources on each type of product as it did when the data underlying the model were
What Multipliers Measure
A multiplier is always a ratio between some measure of a total effect and some measure of the direct effect that caused
it. This definition covers many different measures. To explain the multipliers reported below, I will describe them in the
context of five dimensions common to all multipliers.
How Change is Denominated
A change in the size of an industry can be measured in several ways: jobs, income, sales, and output quantities are
examples. Also, the total effect can be denominated differently than the direct effect, leading to a large number of
possible combinations. Starting with the three dimensions most commonly used (jobs, sales, and income), we immediately
have nine possible types of multipliers: job-job, job-sales, job-income, sales-job, and so on.
Three of these, however, are used more frequently than the others, and are often given shorter names. Employment (job-job)
multipliers refer to the total number of new jobs in the region per each direct new job in a given export-base
industry.(7) Output (sales-sales, or output-export) multipliers refer to the total dollar value of new sales
resulting in the region from one dollar in direct new export sales in a given industry. Income (income-sales, or
income-export) multipliers refer to the total amount of additional income in the region that results from one dollar in
direct new export sales in a given industry.
In this paper I report a fourth type, wage-wage multipliers. These multipliers relate the total new wage and salary income
in Kansas to the change in wages and salaries earned in the exporting sector. For farming, proprietors' income was used
instead of wage and salary income. Wage-wage multipliers have no common short names.
It is also possible to define extended multipliers that measure the effect of economic change on other variables such as
population, government expenditures, environmental pollution, or the relative distribution of income.(8)
The Target Region
The multipliers reported in this paper are specific to Kansas. This simple point has several implications.
First, the indirect effects measured by the multiplier may take place anywhere in Kansas, and not necessarily in the same
city or county as the direct effect. For example, some part of the total effect is due to increased state government
purchases, which tend to be centered in Topeka. Also, the effects of increased demand tend to trickle up from smaller
rural areas to the larger urban centers.
Second and consequently, the multipliers for individual cities and counties in Kansas will almost always be smaller than
those for Kansas as a whole. To put this in other terms, smaller areas purchase a larger fraction of their needs from
outside the region than do larger areas; these larger leakages leave fewer dollars to recirculate in the local economy, so
multipliers are smaller. Therefore, for measuring local impacts in Kansas, the multipliers reported in this paper should
be treated as upper bounds.
Third, the multiplier for each industry is based on average spending patterns of all Kansas firms in that industry. When
firms closer than average to the border make an export sale, more of the money from the sale will leak across the border,
so the multiplier will be smaller than average. Conversely, transactions of firms closer than average to the center of the
state will tend to have an above-average multiplier.
Level of Aggregation
The multipliers reported here are broken out by the forty-eight sectors or general types of industry defined in
Table 3. In principle it is possible to break these down further, but in practice the cost of
obtaining data rises rapidly as the number of sectors increases, and the cost of disaggregation quickly becomes
prohibitive. Therefore any multiplier must be viewed as an average over all the detailed types of firms in that sector in
In the Kansas Long-Term Model (along with most other input-output models) sectoral definitions obey some accounting
definitions that create pitfalls for the unwary. First, output is defined in terms of producer prices, meaning
prices prior to the costs of transportation, wholesaling, and retailing. Second, output in the trade and transportation
sectors is measured by the markup or increase in cost, and not by the total value of sales or shipments.
Short-Run and Long-Run Models
An economic event may have a complex time path that is hard to summarize in a single multiplier. Many authors distinguish
between the short-run impact of a change and the long-run effect after the regional economy returns to
equilibrium.(9) In this paper I report only the long-run multipliers. They represent the total increase in the
measured variable resulting from a permanent one-unit increase in exports in a given sector.(10)
In the long run, multipliers change. New technology and changes in taste lead to changes in expenditure patterns; changes
in transportation costs and in regional comparative advantages lead to changes in the geographical pattern of trade and
leakage. Changes in expenditure patterns caused by price changes and the business cycle may also change multipliers in the
short run.(11) Generally the multipliers I report in this paper are based on data from 1977 to 1981. How
applicable these multipliers are to the present and the future is a subject for future research.
How Multipliers are Constructed
Even multipliers that measure the same thing may vary in magnitude when different data sources or measurement techniques
are used in their construction. The choice of measurement technique affects the accuracy(12) and completeness of
the multipliers and the cost of their construction.(13)
Econometric versus Input-Output Models(14)
Econometric models use a regression technique to relate changes in total economic activity to changes in exports of
different types of goods. The regression is on a sample of observations, which may include a single region at many points
in time; many regions at a single point in time; or in a few studies, pooled regions across time. At its very best, this
approach is more direct and more credible than any alternative.
In most econometric applications, however, there are some severe limitations. First, by their nature the resulting
multipliers are averages over time or over regions or both; the multipliers are usually not localized in time or space.
Second, various econometric difficulties tend to limit accuracy and the feasible extent of disaggregation. Third, and most
problematic, the method works well only in those very rare cases where extensive and accurate data on regional exports are
available by sector.
Input-output models such as the Kansas Long-Term Model can provide a more detailed disaggregation than econometric models,
but they also cost more. Input-output models are constructed by collecting extremely detailed information on the sales
between each pair of sectors for one year. With forty-eight sectors, that amounts to some 2,304 (forty-eight times
forty-eight) groups of data. Some models require much more data to measure imports, exports, sales to households, and
purchases and sales for investment purposes. The data are arranged in matrices that describe the flow of economic activity
between sectors. Thus, input-output models assume that inputs will bear the same proportion to outputs or to household
income in the future as they did when the data were collected. The multipliers are then inferred from the data using
Since input-output multipliers are based on indirect reasoning and depend on a number of assumptions, it is very hard to
measure their accuracy. A few attempts have been made to compare input-output multipliers directly with corresponding
econometric multipliers;(15) although the results of these comparisons have been inconclusive, the two types of
multipliers do seem to give similar results.
In particular, when the input-output multipliers reported in this paper were compared recently to multipliers from an
independent econometric model of Kansas,(16) the average mean errors were typically 25 to 30 percent. Neither
model gave systematically larger multipliers, but for some sectors the two multipliers varied by as much as 50 percent
depending on which model generated them.
Input-output models are classed as nonsurvey, partial survey (or hybrid), and full survey models. The classification
describes where they lie on the continuum between models based on data from published sources (with or without adjustments
and modifications) and models based on direct original surveys of firms and households in the region.
Full survey models are believed to be substantially more accurate than nonsurvey models. Even the best full survey model,
however, depends on piecing together fragmentary information from many sources, including government records and
statistics collected for other purposes; even the worst nonsurvey model ultimately depends on government statistics
derived from surveys of firms and households. The accuracy of a model depends finally on how suitable the underlying
source surveys are to the task at hand: whether the surveys covered the actual target region or some larger region;
whether the sectoral and accounting conventions were consistent with the assumed sectoral scheme; and whether the data had
to be extensively adjusted or modified.
Nonsurvey input-output models have been constructed for most states of the United States(17). Because of their
cost, full survey models have been constructed in only a few states. Kansas is one of these states; Professor Jarvin
Emerson at Kansas State University has constructed full survey models of Kansas for several different years. For technical
reasons,(18) the Emerson data were not used in the Kansas Long-Term Model. Instead, the model uses published
data. For a fuller description of the data, see the Appendix, and also the following discussion of trade data.
Measurement of Trade Data
The accuracy of multipliers is especially sensitive to the accuracy of data on sales across state or regional
boundaries.(19) Unfortunately, these data are not routinely published by government agencies; also, they are
especially hard to survey. Consequently, most multiplier measurements rely on rather crude approximations for exports and
Methods based on location quotients are among the least bad of these approximations. With this approach, one assumes that
a product is imported only when all the local production of that product is used up. Any surplus production is exported.
This approach understates the leakages caused by imports and, consequently, overstates the multipliers. Compared with
multipliers from full survey models, multipliers from location quotient models are typically about 20 percent bigger, but
in a few sectors they may be much larger.(20)
The multipliers reported here from the Kansas Long-Term Model rely on actual survey data from the Census of Transportation
for interstate shipments of material goods, but interstate trade in services was estimated using location quotients.
Static and Dynamic Multipliers
Econometric multipliers compare the total effect of an economic event with the direct effect, using straightforward
statistical averages based on regression. In contrast, input-output multipliers measure the total effect by accounting
piecemeal for a number of different indirect effects. Consequently, input-output multipliers generally omit some of the
indirect effects, in particular the ones that are either hard to measure or expected to be small.
Input-output multipliers are classified according to which indirect effects have been accounted for. Type I multipliers
only account for one kind of indirect effect: induced sales from one firm to another firm within the region. Type II
multipliers, like those reported in this paper, also account for induced household demands, so they are larger and more
Input-output multipliers are also classed according to how much time is included in the model that generated them: a
static model describes only the first ripple in the economy after the initial transaction. In contrast, a dynamic model
like the KLTM can describe the detailed time-path of indirect effects of a transaction. The multipliers reported here take
into account several long run effects, not only intermediate product demands and household demands, but also the effects
of state and local government expenditures that are induced by an increase in state income. The effects of state and
private capital investment needed to replace the depreciation associated with a permanent expansion in output are also
An unfortunate feature of current multiplier research is that estimates of accuracy are rarely published. This shortcoming
is understandable: it is hard to measure the accuracy of multipliers, especially hard for the first researcher, who has no
standard of comparison. The ultimate user of multipliers is left to guess at the reliability of the multipliers he or she
uses, and the user is in a worse position to make these guesses than was the original producer.
In that light, and taking into account the spotty available information,(22) I estimate the error in the
individual multipliers reported here is within 25 to 35 percent of the true value on average, though some individual
multipliers may be worse.(23) Note that even these accuracy figures are based partly on the assumptions that the
KLTM model theory corresponds to reality and that we have not made numerical blunders.
In cases where there is no detailed sectoral information (for example, for evaluating the future effects of a development
incentive that applies to all sectors), it is reasonable to use generalized Kansas multipliers based on the median
multipliers in Table 1. That is, a generalized Kansas output-export multiplier is around 1.9; a generalized Kansas
export-income multiplier is around 0.46; a generalized Kansas wage-wage multiplier is around 1.9; and a generalized Kansas
job-job multiplier is around 2.0. I expect that the level of these generalized multipliers is biased by less than 30
percent, and possibly less than 20 percent. In other words, improved information in the future is unlikely to change the
measured level of these generalized multipliers by as much as 30 percent.
The work reported here was performed under the general direction of Mohamed El-Hodiri and with the able assistance of Tony
Firner, Bob Glass, Lori Munsch, Pat Oslund, David Reardon, John Thissen, and Rita Thissen.
The following gives a description of how the I-O parameters were estimated. A list of data sources is given at the end of
The matrix formula used for constructing the export-output multipliers given in equation (1) below. Similar equations
apply to the other multipliers; for details see Burress and Clifford .
is the identity matrix.
is a non-negative matrix which translates output quantities into induced Kansas
demands, assuming constant wage rates, returns to capital, and prices (which are determined outside the region), and also
constant ownership shares of property income by Kansans and non-Kansans. These induced demands include not only household
consumption purchases, but also state and local government demands.
is a non-negative matrix of intermediate input demand coefficients.
is a non-negative matrix of investment coefficients.
is the set of desired ratios of capital to capacity.
is a non-negative matrix of normal import requirements (i.e., imports required to meet the domestic demand under
the usual condition of excess capacity).
is a matrix of capacity depreciation rates.
is the set of desired ratios of capacity to output.
The matrix shows flows of goods between sectors. It was first inferred for
eighty-six sectors using the 1981 BEA Make and Use Tables for the nation under commodity-based technology assumptions
(i.e. ). For consistency with theory, negative coefficients were eliminated as
follows. Large negative coefficients in two sectors were removed by pre-aggregating communications with business services.
A few very small negative coefficients in other sectors were simply set to zero. Then the
matrix was aggregated to forty-eight sectors. The aggregation was weighted using County Business Pattern data on Kansas
output, so as to alter the national flows to reflect the structure of the Kansas economy.
The matrix describes how economic activity affects investment in capital
equipment. Investment coefficients were inferred for eighty-six sectors from the 1981 BEA capital flows table.
Capital to capacity ratios were calculated for manufacturing sectors by averaging
six years of data by sector. Capital was taken from the BEA U.S. Tangible Reproducible Wealth series; annual output
figures and the ratio of output to preferred capacity were also from the BEA. The capital stock was then corrected to
agree with the concepts used in the capital flow table by prorating on 1981 sectoral investment from the two sources. A
similar procedure was used in nonmanufacturing sectors, except that capacity was assumed to equal output for each year.
In calculating , the critical or desired level of capacity utilization was
inferred for each manufacturing sector by regressing net investment on capacity utilization; the intercept (point of zero
net investment) was then taken to be the critical level of utilization.
Depreciation rates were inferred from the 1981 Tangible Reproducible Wealth series.
The capital coefficients (,,
) were then aggregated to forty-eight sectors, prorating on 1981 Kansas outputs.
Import coefficients were estimated using the 1977 MRIO for material flows, and
1981 location quotients for flows of services. For inferring regional income coefficients φ, property income was
assumed exported in all sectors except agriculture, where all income was assumed received by Kansans. Labor income was
assumed received by Kansans. Bureau of Labor Statistics wage bill data were used to estimate 1981 wage coefficients.
For inferring consumption coefficients , the 1981 BEA Use Tables were used for
household demands; Census of government data were used to infer government demands in Kansas.
Kansas Department of Human Resources, Research and Analysis Section. Labor Market Summary. March 1988. Photocopy.
Transportation in America: A Statistical Analysis of Transportation in the United States. Washington, D.C.:
Transportation Policy Associates, March 1983.
U.S. Bureau of the Census. County Business Patterns 1981: Kansas. Series CBP-81, no. 18. Washington, D.C.:
Government Printing Office, 1983.
U.S. Bureau of the Census. County Business Patterns 1981: United States. Series CBP-81, no. 1. Washington, D.C.:
Government Printing Office, 1982.
U.S. Bureau of the Census. Government Finances in 1981-82. Series GF82, no. 5. Washington, D.C.: Government
Printing Office, 1983.
U.S. Bureau of the Census. Government Finances in 1980-81. Series GF81, no. 5. Washington, D.C.: Government
Printing Office, 1982.
U.S. Bureau of the Census. 1982 Census of Governments: Compendium of Government Finances. Series GC82 no. 4.
Washington, D.C.: Government Printing Office, December 1984.
U.S. Bureau of Economic Analysis. Regional Economic Information System. Farm Income and Expenses. April 1988.
Table CA45. Computer Printout.
U.S. Bureau of Economic Analysis. Regional Economic Information System. Full-Time and Part-Time Employees by Industry,
1969-1987. September 1988. Table SA25. Diskette.
U.S. Bureau of Economic Analysis. "The Input-Output Accounts of the U.S. Economy, 1981." Survey of Current Business
67, no. 1 (January 1987), pp. 42-58.
U.S. Bureau of Economic Analysis. The National Income and Product Accounts of the United States, 1929-82.
Washington, D.C.: Government Printing Office, September 1986.
U.S. Bureau of Economic Analysis. "New Structures and Equipment by Using Industries, 1977." Survey of Current Business
65, no. 11 (November 1985), pp. 26-35.
U.S. Bureau of Economic Analysis. Personal Income by Major Source and Earnings by Major Industry. Table CA5.
Washington, D.C.: Bureau of Economic Analysis, April 1988. Computer Printout.
U.S. Bureau of Economic Analysis. Quarterly State Personal Income, 1981:I-1984-IV. Table SQ7. Washington, D. C.:
Bureau of Economic Analysis, October 15, 1987. Diskette.
U.S. Bureau of Economic Analysis. Regional Economic Information System. Wage and Salary Disbursements, 1958 - 1987.
September 1988. Table SA25. Diskette.
U.S. Bureau of Labor Statistics. Employment and Wages, Annual Averages 1981. Washington, D.C.: Government Printing
Office, November 1983. Microfiche.
U.S. Department of Health and Human Services. National Archives and Records Service. The Multiregional Input-Output
Accounts, 1977, by Jack G. Faucett, Linda K. Lent, and Harry J. Chmelynski. Contract Number HHS-100-81-0057. August
1983. Computer Tape.
- For example, large multipliers are reported in Berry ; and the Kansas Business News . Both of these
articles rely on a study [Heins 1982] contracted by the Institute for Economic and Business Research (or IEBR, the precursor to IPPBR), on behalf of the Kansas Chamber of Commerce and Industry. The Heins study was released but not
endorsed by IEBR; indeed, an internal review criticized the study both for its implausible methodology and for its
implausible results [Sexton, undated]. In particular, the study used a regression model that implied, contrary to
most theory and evidence, that small towns supply a higher fraction of their needs locally than do larger cities. It
may not be surprising that the results of the Heins study have been circulated very widely in Kansas.
- For some discussion and citations on the effects of intergovernmental competition on economic development, see my
earlier article in this review .
- Some members of the community are likely to be harmed rather than helped by aggressive concessions for development;
retired people, for example, are unlikely to benefit from local economic development, but might be subject to higher
taxes and crime and congestion resulting from development. These interest groups would have an incentive to
understate rather than overstate multipliers.
At the same time, these interest groups are unlikely to be actively involved in discussions of development policy,
for two reasons. First, they have a diffuse rather than a concentrated interest (many individuals stand to lose a
small amount, but none stand to lose a large amount). Second, they typically have limited knowledge and experience
about development, and their access to discussions of development policy is limited. A similar argument is made in
the public choice literature, for example, Olsen .
- See Burress  for a more detailed discussion of export base theory.
- For a discussion of actual retail sales from non-Kansans from Johnson County, Kansas, see Sanborn et al. .
- Burress  gives conditions under which import multipliers equal the corresponding export multipliers.
- Because of the nature of available data, the job-job multipliers reported in the paper count part-time jobs with the
same weight as full-time jobs.
- We have not undertaken this type of analysis. For a survey of extended environmental multipliers, see Briassoulis
- In econometric modeling, the short-run multiplier is referred to as an impact multiplier, and the long-run
multiplier is referred to as a total multiplier. Note that the total multiplier in this sense should not be confused
with the total effect. In particular, the short-run multiplier is determined by the total short-run effect.
- McNutlty  and others cited therein have argued that export multipliers make the most sense as long-run
concepts, because of the time required for investment activities as well as for population movements. Although
McNulty's empirical work was flawed (see Gerking and Isserman ), his theoretical argument remains persuasive.
Indeed, Andrew and Tate  and others cited therein give evidence that export multipliers de require several
years to take their full effect. Moreover, the multipliers reported in this paper are based on a dynamic
input-output model with some rather arbitrary assumptions about short-run lags, but these assumptions do not affect
the long-run multipliers. [See Burress and Clifford, 1988.]
- For a discussion of short-run changes in multipliers, see the two 1984 articles by Pickerill.
- In this paper I take an aggressively positive and pro-user view of accuracy. That is, I define an accurate
multiplier as one which, under ceteris paribus conditions, measure as nearly as possible what an ordinary user
expects it to measure. In particular, I doubt whether an ordinary user had any interest in Type I or Type II
multipliers on their own merits; instead, he is looking for a measure of the total effect of an exogenous change.
Many researchers on multipliers have argued for a narrower and more defensive view. For example, Giarratani [1980, p.
193] states"[A] distinction between forecasting and impact analysis must be drawn because there is simply no
empirical possibility of measuring the extent of error in the latter instance." This seems unnecessarily to defend
impact multipliers against dangerous confrontations with reality. For an example of an attempt to measure errors in
an quasi-impact analysis, see Conway.
- An extensive review of techniques for construction multipliers is given by Richardson .
- I have omitted the simplest and least expensive way to measure multipliers, which is an economic base model. This
model divides the economy into only two sectors: industries that mainly export (basic industries) and industries
that mainly produce for local consumption (non-basic industries). The multiplier represents the ratio of total
economic activity to activity in the basic industries. This approach has fallen into disfavor because it is believed
to be inaccurate. In particular, it is highly aggregative (there is only one exporting sector), and exports are
poorly measured. In the real world, many industries produce partly for local consumption and partly for export.
- For citations and a discussion of attempts to compare multipliers, see Burress and Clifford .
- The comparison of multipliers is in Burress and Clifford .
- For a review of state input-output models, see Englinski, Oslund, and Burress .
- See M. Jarvin Emerson [1969; 1974]. Professor Emerson has completed but not published a 1985 model. The model most
recently published by Emerson was for 1971, about a decade older than the most recent national model. Also, like all
full-survey models, Emerson's model is static rather than dynamic. Emerson's sector scheme, which heavily emphasizes
agriculture, does not reconcile with standard published data like the national capital flow matrix. Finally, one
purpose of the present research was to establish a method for updating Kansas multipliers with a greater frequency
than full-survey methods would allow.
- Garhart and Giarratani  discuss the critical role of trade and give citations.
- See Bourque  and citations therein for comparisons of location quotients to full survey models.
- The multiplier derivation and formula is given in Burress and Clifford . More detailed documentation on the
model and data are given in Oslund, Thissen, and Thissen ; and Burress, Oslund, and Thissen .
- As discussed above, Burress and Clifford  found that the KLTM multipliers presented here differ by some 30
percent from corresponding econometric multipliers. Using a model with a structure similar to the Kansas Long-Term
Model but with much superior data, Conway  found that long-run income multipliers appeared to be more
accurate, perhaps within 10 percent. Conway  found that Type I multipliers may drift by 10 percent over time.
Bourque  found that a top-down model similar to the Kansas Long-Term Model may have average errors of 15 to 20
percent as compared to a direct survey model. See also citations therein.
- In other words, I believe that multipliers measured according to the methodology used here have the following
property: if a large sample of post-impact studies were carried out, where each study attempted to measure the
actual total effect of a substantial change in the exports of one industry, then the mean absolute percent error
between the study results and predictions based on these multipliers would be less than 35 percent.
Berry, Mike. "Dormant Factory Bought; Sale Recharges Hays Outlook." Wichita Eagle-Beacon, June 14, 1988, pp. 1A,2A.
Bourque, Philip J. "Regional I/O Modeling (Some Reflections about the Survey Approach and a Comparison of WAIO and RIMS
Multipliers)." Paper presented at the International Conference on Construction and Use of Input-Output Models, Terra Alta,
West Virginia, 1988.
Briassoulis, Helen. "Integrated Economic-Environmental-Policy Modeling at the Regional and Multiregional Level:
Methodological Characteristics and Issues." Growth and Change, July 1986, pp. 22-34.
Burress, David. "Import Substitution Impact Multipliers." Economics Research Technical Note, no. 36. Lawrence, Kansas:
Institute for Public Policy and Business Research, University of Kansas, 1987.
Burress, David. "Targeting New Interstate Trade: A Proposal for Reforming Kansas Development Tax Policy." Kansas Business
Review, 11, no. 2 (Winter 1988), pp. 21-29.
Burress, David; and Norman Clifford. A Comparison of Dynamic I-O Multipliers for Kansas with Parallel Econometric
Multipliers. Discussion Paper Series, no. 1988.2. Lawrence, Kansas: Institute for Public Policy and Business Research,
University of Kansas, 1988.
Burress, David; Michael Eglinski; and Pat Oslund. A Survey of Static and Dynamic State-level Input-Output Models.
Discussion Paper Series, no. 1988.1. Lawrence, Kansas: University of Kansas, Institute for Public Policy and Business
Research, University of Kansas, 1988.
Burress, David; Mohamed El Hodiri; et al. Final Report: Research Improvement Award for Economic Development. Monograph
Series, no. 124. Lawrence, Kansas: Institute for Public Policy and Business Research, University of Kansas, 1987.
Burress, David; Pat Oslund, and John Thissen. Revised Sector Definition for KLTM. Economics Research Technical Note, no.
63. Lawrence, Kansas: Institute for Public Policy and Business Research, University of Kansas, 1988.
Conway, Richard S., Jr. "The Stability of Regional Input-Output Multipliers." Environment and Planning 9, no. 2
(February 1977), pp. 197-214.
Conway, Richard S., Jr. "The Washington Projection and Simulation Model: Ten Years of Experience with a Regional
Interindustry Econometric Model." Paper presented at the International Conference on Construction and Use of Input-Output
Models, Terra Alta, West Virginia, 1988.
Emerson, M. Jarvin. The Kansas Input-Output Study, report no. 33, Topeka, Kansas: State of Kansas Department of Economic
Emerson, Jarvin M. "Interregional Trade Effects in Static and Dynamic Input-Output Models." In Proceedings of the Sixth
International Conference on Input-Output Techniques, Vienna, edited by Karen R. Polenske and Jiri V. Skolka. Cambridge,
Massachusetts: Ballinger, 1974, pp. 263-277.
Garhart, Robert E.; and Frank Giarratani. "Non-survey Input-Output Estimation Techniques: Evidence on the Structure of
Errors." Journal of Regional Science 27(2), 1987, pp. 245-253.
Gerking, Shelby D.; and Andrew M. Isserman. "Bifurcation and the Time Pattern of Impacts in the Economic Base Model."
Journal of Regional Science 21(4), 1981, pp. 451-467.
Giarratani, Frank. "The Scientific Basis for Explanation in Regional Analysis." The Regional Science Association Papers
45, 1980, pp. 185-196.
Heins, James. "The Impact of 100 New Manufacturing Jobs on an Average County in Kansas." Report prepared for the Kansas
Association of Commerce and Industry, and the Kansas Department of Development, under contract to the Institute for
Economic and Business Research, University of Kansas, 1982.
Kansas Business News. "New Jobs Create Other Jobs." April 1988, p. 4.
Miller, Ronald E.; and Peter D. Blair. Input-Output Analysis: Foundations and Extensions. Engelwood Cliffs, NJ:
Prentice-Hall, Inc., 1985.
McNulty, James E. 1977. "A Test of the Time Dimension in Economic Base Analysis." Land Economics 53, no. 3 (August 1977)
Olson, Mancur. The Logic of Collective Action, revised, Cambridge: Harvard University Press, 1971.
Oslund, Pat; John Thissen; and Rita Thissen. Data Sets for the 48- and 11- Sector Steady State I-O Multipliers. Economics
Research Technical Note, no. 62. Lawrence, Kansas: Institute for Public Policy and Business Research, University of
Pickerill, Karen L. "A Theoretical Justification for a Variable Export Base Multiplier." Western Kentucky University
College of Business Administration Working Paper Series 3, no. 12, February 1984a.
Pickerill, Karen L. "An Empirical Investigation into the Stability of Export Base Multipliers: The Kentucky Case." Western
Kentucky University College of Business Administration Working Paper Series 3, no. 13, February 1984b.
Richardson, Harry W. "Input-Output and Economic Base Multipliers: Looking Backward and Forward." Journal of Regional
Science 25, no. 4 (1985), pp. 607-659.
Round, Jeffrey I. "Nonsurvey Techniques: A Critical Review of the Theory and the Evidence." International Regional Science
Review 8, no. 3 (1983) pp. 189-212.
Sanborn, Gina. 1988. The Nature and Significance of the Overland Park/Johnson County Economy. Monograph Series, no. 155.
Lawrence, Kansas: Institute for Public Policy and Business Research, University of Kansas, 1988.
Sexton, Richard, to Anthony Redwood, undated. Internal report of the Institute for Public Policy and Business Research
regarding 1982 study by James Heins.
Back to Publications
Back to Home Page