Chapter 3 Multipliers and Leakages

Published by

Bartik, Timothy J.
Making Sense of Incentives: Taming Business Incentives to Promote Prosperity.
W.E. Upjohn Institute, 2019.
Project MUSE. https://muse.jhu.edu/book/82051.

For additional information about this book

https://muse.jhu.edu/book/82051

[41.59.199.65] Project MUSE (2025-04-12 12:39 GMT)

 Chapter 3
Multipliers and Leakages
How to Think about Incentives
How should we evaluate incentives’ benefits and costs? 22
That
depends on incentives’ goals.
Let’s start with a state perspective. Chapter 7 will consider a
national perspective. But states are the main actors in the current
incentive wars. From an individual state’s perspective, what should
governors and state legislators be trying to do with incentives?
Presumably, helping state residents. Incentive costs and incentive-
induced jobs affect state residents in many ways. Job growth may
sometimes harm the environment. More and different people in a state
may affect the quality of life, in ways both good and bad.
But incentives’ most important effects are on state residents’ per
capita incomes—both average per capita incomes and the distribution
of income across different income groups. I don’t mean to sound like
an economist obsessed with money, but higher per capita incomes are
important to peoples’ well-being.
Furthermore, we know a lot about how job growth affects income;
how growth affects the local quality of life is harder to measure. Qual-
ity of life effects also may vary quite a bit with the state’s circum-
[41.59.199.65] Project MUSE (2025-04-12 12:39 GMT)

stances and the particular job creation project. Generalizing about

quality of life effects may not be feasible.
Therefore, I focus on how the level and distribution of state res-
idents’ per capita incomes are affected. The reader should keep in
mind that job growth may also affect a state’s quality of life in other
ways. Ideally, we would adjust our estimates of income effects for
these harder-to-measure effects, whether positive or negative.
So, what is the logic for how incentives affect state residents’ per
capita incomes?

17
18 Bartik

MULTIPLIERS AND SPILLOVERS

Let’s start with the positive. Why might incentives increase state
residents’ incomes? (Negative costs for state residents’ incomes will
be considered later in this chapter.)
To see the effects of incentives, envision the state economy as
a machine. Incentives are a fuel that can set the machine going. The
machine of the state economy has various mechanisms that transmit
the incentives’ fuel into the machine’s output: various goodies for
the state’s residents. Some of these transmission mechanisms may
even multiply the effects of the incentive fuel on the machine’s ability
to produce goodies. Later on, we consider leakages in the machine,
which slow down the machine and reduce its production of goodies.
Incentives have several multipliers and spillovers that increase
state per capita income (Figure 3.1).23 Some incentives will directly
benefit state residents who own businesses receiving incentives, but
most incentives go to out-of-state corporations.24 How can helping
out-of-state corporations ever benefit a state’s residents?
For most incentives, benefiting state residents requires that the
incentive induce some local job creation. This job creation could be
induced in several ways: because of the incentive, a business chooses
to locate in the state rather than some other state. Or because of the
incentive, a new business is created that would not have occurred
otherwise. Or because of the incentive, a business chooses to expand
in the state, and otherwise that expansion would not have taken place.
Or because of the incentive, a business retains its current jobs in the
state, rather than downsizing or closing. In each case, the incentive
leads to some additional jobs in the state that would not have existed
in the state “but for” the incentive.
An economic developer might say: “What I do must have some
impact. Surely you have to admit that providing tax breaks must have
some impact on the probability of favorable location and expansion
decisions made by business executives. The “but for” percentage for
incentives must be greater than zero.” The “but for” percentage also
Figure 3.1 A Model of Incentives’ Benefits


     

 
 
 
  
    
  
    

 
    
       
 
      
  
   
   




 
 
   

19
 20 Bartik

may not be 100 percent, but it is realistic to acknowledge the eco-

nomic developer’s point that it is greater than zero.
Job creation in the incented firm will have some multiplier effects.
The incented firm may buy more from local suppliers, which will in
turn hire more. Workers at the incented firm, and its local suppliers,
will have more earnings to spend. Some of these workers’ increased
earnings may be spent at local companies: grocery stores, hardware
stores, restaurants, gift shops, etc. These local companies may hire
more employees to distribute their goods and services. If any of these
purchased goods and services are produced locally (craft beer? local
farmers? local jewelry makers?), then such local production jobs
increase.
If the incented firm is high-tech, and there are many other nearby
high-tech firms, there may be additional multiplier effects. High-
tech firms often cluster. Why this clustering? High-tech firms such
as Google steal ideas from one another. Google does so in part by
stealing workers from other high-tech firms.25 If an incentive induces
local job creation in a high-tech firm, this additional high-tech firm
can contribute more workers and ideas to the local high-tech cluster.
This adds to the productivity of the local high-tech cluster, which will
encourage other local high-tech firms to start up or expand.26
The total boost to local jobs, from both incented firms and their
multiplier jobs, will make it easier for state residents to find jobs. The
local employment-to-population ratio—the “employment rate”—will
go up in the short run. Local workers who otherwise would not have
[41.59.199.65] Project MUSE (2025-04-12 12:39 GMT)

a job end up with additional job experience. A worker who gains job
experience will gain job skills. Those added job skills are of many
types: from my job experience, I learn how to work with industrial
machinery, I gain self-confidence, I am less likely to get involved
with alcohol and drugs, and so on. All these added job skills make me
more employable in the long run. My long-run employment rate and
wage rate will be higher.
This job growth will also boost wages for workers in general,
with more labor demand relative to labor supply. This boost in wages
 Multipliers and Leakages 21

would exceed any increases in local consumer prices; an economist

would say that local “real wages,” wages adjusted for local prices,
will increase. Because of both higher employment rates and wages,
state residents will experience an increase in per capita earnings.
This increase in per capita earnings is likely to be a higher per-
centage boost in income for state residents who otherwise would have
low and moderate incomes. Low- and moderate-income Americans
are more likely to be unemployed, or so discouraged by poor job pros-
pects that they give up seeking jobs. They are also more likely to be
underemployed, working in lower-wage jobs than they are capable
of doing. Job creation, by increasing employment rates and wages,
particularly helps these nonemployed or underemployed Americans.
An economist would say that the distribution of earnings gains from
local job growth is likely to be “progressive,” which is simply jargon
for the income boost being a higher percentage of income for those
lower in the income distribution.
Growth will increase housing demand. Higher housing demand
will boost property values. State residents who own real estate will
experience a capital gain.
These capital gains from higher property values are likely to be a
higher percentage of income for state residents who are higher in the
income distribution, as higher-income Americans are more likely to
be homeowners. An economist would say that the distribution of local
real estate capital gains from job growth is “regressive,” which is sim-
ply another way of saying that the boost to economic well-being is a
greater percentage of income for those higher in the income distribu-
tion. This regressivity is moderate, as home ownership does extend,
at lower rates, into lower-income groups.
(What dominates, the progressivity of earnings gains, or the
regressivity of property value gains? Stay tuned until the next chapter!)
The increase in business activity and profits, sales, worker earn-
ings, and property values all will increase various tax bases. Revenue
for state and local governments will increase, even with no change
22 Bartik

in tax rates. If nothing else changes, this provides a fiscal benefit for
state and local governments.
This fiscal benefit in turn helps state residents. State and local
governments have more or less binding balanced budget require-
ments—they will use this fiscal benefit to either lower residents’ tax
rates or increase spending on public services or welfare programs
that transfer income to state residents. Either way, one could argue
that state residents’ true incomes per capita, adjusted for govern-
ment policy, will increase. Lower tax rates obviously increase state
residents’ net incomes after subtracting taxes. Higher welfare ben-
efits also increase state residents’ incomes. Higher public services
increase state residents’ well-being—which we could view as their
“real income”—in that they provide state residents with some service
they value: the roads have fewer potholes, the state parks have better
and more accessible trails, more policemen and firemen make life a
little safer.
In sum, state residents’ incomes increase because of higher
employment rates, higher wage rates, higher property values, and
either lower tax rates or higher public services.

LEAKAGES AND NEGATIVE FEEDBACKS

But the machine of the state economy has various leakages.

Incentives’ boost to state per capita incomes is reduced or eliminated
by several leakages or negative feedbacks (Figure 3.2).
The “but for” percentage is less than 100 percent. Not all firms
receiving incentives were induced to locate, expand, or retain jobs
because of the incentive. Even without any incentives, many of these
firms would have chosen this state anyway. Many other state and
local characteristics and national forces affect business decisions—
local wage rates, skills of local workers, the area’s access to national
markets, and so on. A typical incentive, which is equivalent to a 3
percent wage subsidy, will not overwhelm these other forces.
 Figure 3.2 Adding Incentive Costs to the Model


     

 
 
 
  
    
  
    

   

         
[41.59.199.65] Project MUSE (2025-04-12 12:39 GMT)


          
  
          Š
    †‡ˆ     
         
      ‰  
 
  

  
   
  
      

       „
­     
  ­ 
 €‚ƒ    

23
24 Bartik

As mentioned previously, incentives to nontradable firms will

displace jobs at other local nontradable firms. For firms that sell to
a local market, their expansion takes away sales and jobs from other
firms selling to the local market. The new local hardware store cuts
into the sales of nearby hardware stores.
One might think that wise policymakers would never provide
cash incentives to nontradable firms. But incentives are often pro-
vided to business activities that are in part nontradable. Witness gov-
ernment subsidies for sports teams. If we’re trying to increase total
local jobs, such subsidies make little sense.27 Any expanded jobs at
the sports stadium in large part come from local residents spending
more at the stadium. This leaves less local disposable income for
other local goods and services. Jobs at the stadium and at nearby bars
and restaurants may go up, but jobs in other neighborhoods’ restau-
rants and bars, and other local retailers, will go down.
Even for tradable firms, multiplier effects may be less than
expected. Increased local wages and property prices will increase
business costs. These increased business costs will cause other busi-
nesses not to locate or expand in the area. For example, high costs in
Silicon Valley and other high-tech areas may drive away many non-
tech businesses. In addition, these increased business costs may make
local suppliers less competitive and result in increased use of suppli-
ers outside the local area.
Increased business costs reduce profits for some businesses
owned by local residents. Businesses that sell locally may be able
to pass on increased costs to local consumers.28 But businesses that
sell nationally have their prices set nationally, so they must absorb
the business cost increases. A manufacturer in Grand Rapids, Michi-
gan, selling to customers in Chicago or Germany will simply have
to absorb any higher costs if Grand Rapids happens to experience a
growth boom. These losses to local business owners tend to be much
higher for upper-income residents, who are far more likely to own a
business.
 Multipliers and Leakages 25

Not all the jobs that are created in the state will go to residents.
Job growth increases population growth. In-migrants will take some
of the new jobs.
This in-migration also reduces upward wage pressures. Labor
supply increases will to some extent offset the increase in labor
demand.
Increased population also raises the demand for government-
provided goods and services. In-migrants want good schools for their
children. New schools will have to be built and additional teachers
hired to avoid increases in class size. New population creates conges-
tion on roads and other local public infrastructure. If we want to avoid
increased congestion, we have to spend more on infrastructure. All of
this reduces fiscal benefits from incentives.
If state and local governments refuse to increase spending, then
state residents are hurt in another way—by reduced quality of public
services through more crowded schools and roads. Increased popula-
tion has costs regardless of state and local policy: policymakers can
just choose what form those costs take, either increased needs for
spending or reduced local public service quality. Pick your poison.
If the combination of incentive costs plus increased tax revenue,
minus increased spending needs, is on net a negative cost for state and
local governments, then policymakers will have to keep the budget
balanced and pay for these costs. Taxes must go higher, or public
spending lower. Either option may have negative effects on the state
economy. On the demand side of the economy, higher taxes or lower
public spending will reduce state residents’ disposable incomes.
Lower disposable incomes of state residents will reduce demand for
goods and services, which will hurt some businesses in the state and
hence hurt their workers.
Higher taxes or lower public spending may also affect the state
economy on the supply side. Higher business tax rates may cause
other firms to avoid locating or expanding in the state. Cuts in some
types of public spending may reduce the productivity of the state
economy, and thus reduce state residents’ wages and earnings. Cuts in
 26 Bartik

education and job training may reduce skills of state residents, which
will reduce their future wages.

KEY FACTORS AFFECTING INCENTIVE BENEFITS

What is the likely magnitude of these multipliers and leakages,

these spillovers and negative feedbacks? How are these magnitudes
affected by policy?

“But For” Percentage

Research suggests the “but for” percentage is usually less than 25

percent.29 What does this mean? For a new facility location or expan-
sion decision, this means that the incentive tipped the location deci-
sion toward this state only 25 percent of the time or less. The other 75
percent of the time, the firm would have made the same new facility
location decision, or same expansion decision, even if no incentive
had been provided. Another way to put it: even if the state had not
provided incentives to a firm providing a “Big Number” of jobs, 75
percent of those jobs would have been created in the state anyway.30
This low “but for” percentage greatly increases the costs of incen-
tives relative to their benefits. At least 75 percent of the time, incen-
tives are all costs, with no job creation benefits.31 And these incentive
costs will require increased state and local taxes, or higher state and
[41.59.199.65] Project MUSE (2025-04-12 12:39 GMT)

local public spending, either of which will have negative economic

effects.
Incentive effectiveness can be increased by making incentives
more up front.32 Business executives making investment decisions
have an exaggerated focus on the short run, on what happens to prof-
its in the next few years.33 The incentive provided 10 years from now
has little effect on tipping the location or expansion decision. By then,
the executive making the location or expansion decision may be long
gone. Compared to our typical incentive structure, which continues
 Multipliers and Leakages 27

incentives at a high level through at least 10 years, up-front incentives

would be over one-third more effective: per dollar of incentive, the
“but for” percentage would be over one-third higher.34
Customized services to smaller businesses tend to be far cheaper
per assisted job. It would be difficult or impossible to provide useful
services to smaller businesses as big as Foxconn-level wage subsidies
of 30 percent, or even as big as average business tax incentives. Yet
even at much lower costs, customized services to smaller businesses
can have considerable effects. Advice to small business is cheap,
and—if high-quality—can be effective if used as directed. For these
customized services, research shows that the ratio of job creation
effects to costs is plausibly 10 times as great as for cash incentives.35
In other words, if we reallocated a given dollar amount of our
incentive budget away from up-front tax incentives to large corpora-
tions, and toward provided customized job training or manufactur-
ing extension services to smaller businesses, we would induce the
creation of at least 10 times as many jobs. However, at some point
we might face limitations on how much can usefully be spent on cus-
tomized services to smaller businesses. It might not be as great as
the total $50 billion incentive budget, although it is plausibly much
greater than the current $3 billion devoted to such services. Advice is
useful, but spending 17 times as much on advice is not necessarily 17
times more useful.

Multipliers

Many impact studies done for state economic development agen-

cies estimate multiplier effects of 2.5–4.0. For every job created in
incented firms, 1.5–3.0 other jobs are created in other state firms; the
multiplier, or ratio of total jobs created to jobs created in incented
firms, would then be 2.5–4.0.
Recent research shows that these multipliers are overstated.36
These studies’ multipliers typically focus only on the positive impacts
from increased demand for local suppliers and retailers. These studies
28 Bartik

overlook negative feedback effects from higher costs: higher costs in

the state will reduce the state’s competitiveness in attracting business
growth.37 Negative feedback effects reduce demand-oriented multi-
pliers by at least one-third.
Plausible state multipliers for most incented firms are around
1.7–2.0.38 For every 10 jobs directly created by incentives, 7–10 addi-
tional jobs will be indirectly created in the state, after allowing for
both demand effects, and negative effects of higher costs.
Multipliers may be higher for high-tech firms in high-tech-
oriented local economies because of cluster effects. Recent research
suggests that such extra cluster effects are concentrated in the rela-
tively few local economies where high-tech clusters are significantly
above average size.39 The existing high-tech cluster need not be as
large as Silicon Valley’s—a high-tech concentration like Denver or
the Twin Cities is enough to cause higher multipliers. On the other
hand, out of the approximately 700 local labor markets in the United
States, only 60 or so probably have sufficiently large concentrations
of high tech to expect significantly higher high-tech multipliers.
These are the local labor markets in which the share of employment
in high-tech industries is at least one-third greater than in the average
community. Most of these high-tech cluster communities are above
average in population.40
In such high-tech cities, the job multiplier for a new high-tech
firm may be as large as 3.0.41 In contrast, in cities with an average or
below high-tech sector, high-tech multipliers are no larger than for
other firms. States may make incentives more efficient by targeting
high tech, but only for the relatively few high-tech cities—targeting
high tech does not work if one is building an area’s high-tech sector
from scratch, or even if one’s high-tech sector is at the U.S. average.
Not everywhere can be Silicon Valley or a biotech center.
Multipliers are higher on average if the incentives go to locally
owned firms. Locally owned firms, compared to otherwise similar
non–locally owned firms, are more likely to use local suppliers.42
A locally owned bookstore is more likely than a Barnes and Noble
 Multipliers and Leakages 29

bookstore to use a local accountant. In addition, the incentives will

increase the profits of local owners, who will spend some of these
profits locally, further increasing local jobs. The net effect: for two
otherwise similar projects, the project with incentives going to local
owners might have a multiplier higher by 5–10 percent, or would be
2.0 rather than 1.85.
Multipliers are also higher if the local economy has high rates of
unemployment or nonemployment. With more available local labor,
employers are better able to hire locally, which increases the number
of new jobs created. An economically depressed area, compared to
a booming area,43 might have a multiplier higher by around 15 per-
cent—in other words, rather than being 1.75, it might be 2.0.

Who Gets the Jobs?

In the short run, in the average state’s economy, for every 10 new
jobs created, 6 jobs increase the employment-to-population ratio and
4 jobs go to in-migrants. But after five or six years, the proportion
increasing the employment-to-population ratio declines to 10–30
percent—for every 10 new jobs, 1–3 jobs increase the employment
rate for state residents, and the other 7–9 jobs go to in-migrants. This
change over time is due to increased in-migration effects over time.
But these effects after 5–6 years seem quite persistent, through at least
15–20 years.44
On the one hand, it is surprising to many economists that these
[41.59.199.65] Project MUSE (2025-04-12 12:39 GMT)

labor market effects of more jobs are so persistent. A once-and-for-all

increase in a state’s jobs has long-term effects in increasing state resi-
dents’ employability, which will boost long-run earnings. Good eco-
nomic fortune in the short run alters economic fates in the long run.
If you add up the increased earnings from a higher employment-to-
population ratio over many years, you end up with a sizable increase
in lifetime earnings.
On the other hand, only a minority of new jobs benefit a state’s
residents. Most new jobs go to in-migrants. These in-migrants are fine
30 Bartik

people, but they were not the people who paid to finance the state’s
incentive program. In addition, it is likely that the gain to in-migrants
is minor—if these new jobs had not been created in this state or local
economy, the in-migrants could have found jobs somewhere else.45
It might seem surprising that so few jobs go to the original resi-
dents and so many to in-migrants. What if a new incented firm hires
all local residents? What needs to be understood is that ultimately,
new jobs in a state economy must go to either state residents who
would otherwise not be employed, or to new residents—there is no
other choice. Mathematically, if one increases state employment,
either the employment-to-population ratio goes up, or the population
goes up—or, more likely, some combination of the two.
In the real world, the way this plays out is that if a firm hires
state residents who are already employed, this creates a job vacancy
chain. The new hires create job vacancies at the firms where the
residents were previously employed. In turn, these firms make job
hires from some combination of three groups: 1) state residents who
are already employed, 2) state residents who are not employed, 3)
in-migrants. Any subsequent hire of state residents who are already
employed creates another vacancy, which is filled in the same three
ways. The resulting vacancy chain is terminated only by the hiring
of either a nonemployed state resident or an in-migrant. Because of
these vacancy chains, the ultimate impact on state residents’ employ-
ment rates, versus in-migration, depends on many factors in addition
to whom the incented firm hires.46 We need to consider the hiring
practices of all the firms that are part of the job vacancy chain.
How can policy increase the proportion of new jobs going to
state residents?47 One way is by targeting incentives at places with an
excess supply of labor: economically depressed areas.48 When a local
economy’s unemployment rate is high, or its labor force participation
rates are low, a higher proportion of new jobs go to local residents
and a lower proportion to in-migrants. In local economies that are
persistently depressed, compared to local economies that are doing
well, the proportion of new jobs going to local residents goes up about
 Multipliers and Leakages 31

two-thirds. If the proportion of new jobs going to local residents was

15 percent in the economically booming local economy, it would be
25 percent, two-thirds higher, in the economically depressed local
economy. Why are the local employment rate effects higher? Because
with more available local residents who are nonemployed, many of
whom will have competitive job skills, firms all along the job vacancy
chain become more likely to hire a local, nonemployed resident rather
than a local resident who is employed or someone from out of state.
Because of both higher local hiring and higher multipliers in dis-
tressed areas, targeting distressed areas not only has more progressive
benefits, by helping the nonemployed more, but also has greater eco-
nomic benefits for a state’s economy. (Chapter 4 will have some bot-
tom line numbers.) A state targeting needy areas is like the joke about
the early Quakers in Philadelphia: they came to do good but ended up
doing very well indeed.
Effects on state residents’ employment rates can also be increased
by policies that encourage incented firms and other firms to hire more
state residents who lack jobs. State and local governments can use
carrots or sticks to do this. One possible carrot is providing high-
quality customized job training services. The local community col-
lege can provide training programs customized to firms’ skill needs.
Those programs can include local residents who lack jobs. Business
participation in such programs is likely to increase the proportion of
new hires who are local residents who lack jobs. The ultimate impact
will be to reduce the hiring of in-migrants, and to make local resi-
dents more competitive in the labor market because of their greater
job experience.
Another carrot is offering some incentives that are specifically
tied to hiring the nonemployed. A problem is that some employ-
ers may be reluctant to hire the unemployed. These employers may
stigmatize the unemployed as being less-productive workers. For
example, one experiment showed that if welfare recipients informed
prospective employers that the employer would receive a subsidy for
hiring them—which thereby informed the employer that they were on
 32 Bartik

welfare—then this hiring subsidy actually reduced the chances of the

welfare recipient being hired.49
Therefore, incentives to hire the nonemployed are likely to be
more effective if integrated into job training programs that help screen
and train the nonemployed, which will help reduce stigma effects.
Only the unemployed who go through this screening and training
would be eligible for a hiring subsidy, thereby reducing employer
fears about the unemployed being unproductive. These job training
programs also can screen employers. Employers who do not stigma-
tize the unemployed can be recruited and therefore are more likely to
respond positively to a subsidy for such hiring. Only such recruited
employers would be offered a hiring subsidy. In the 1980s, Minnesota
ran such a program on a large scale, with some evidence of success.50
A stick is accompanying all incentives with a local hiring require-
ment. Because of fears of driving away businesses, state and local
policymakers tend to use a light tap with these sticks. Some local
areas, such as Berkeley, have “first source” requirements for firms
receiving incentives: such firms are required, for entry-level job
openings, to consider as a first source referrals from local job training
agencies. Local hiring is not required for all job openings; it is only
encouraged for some job openings.

What Is a Local Economy?

As discussed above, when a local economy is depressed, multi-

[41.59.199.65] Project MUSE (2025-04-12 12:39 GMT)

plier effects are larger, and a higher proportion of the new jobs go to
local residents. But what is the size of a meaningful local economy?
A metropolitan area? A neighborhood?
The best answer is somewhere between a neighborhood and a
metropolitan area. A typical U.S. county might be around the right
size, in terms of land area, to make a meaningful local labor mar-
ket. Research suggests that a significant proportion of the increased
earnings per capita from increases in job growth within a county will
occur within that county.51 Some labor market benefits will spill over
 Multipliers and Leakages 33

into surrounding counties in the metropolitan area, but enough bene-

fits will remain in the county for the county’s labor market conditions
to be meaningfully affected. In contrast, for a city neighborhood, most
of the benefits of new jobs spill over into surrounding neighborhoods.
Targeting jobs at a high-unemployment neighborhood does not tar-
get jobs at the neighborhood’s residents. Such targeting might mainly
increase neighborhood rents, encouraging the neighborhood to gen-
trify and turn over to higher-income residents.
What if policymakers want to target within a county? They might
want to target more new jobs for disadvantaged individuals in dis-
advantaged neighborhoods. Such targeting is better accomplished
through local job training programs or hiring incentives. Just plop-
ping down jobs in a neighborhood is ineffective.52

Fiscal Benefits

Incentives may provide state and local governments with some

fiscal benefits: gains in tax revenue, due to an expanded tax base,
that exceed the increase in public service needs, due to an expanded
population. However, these fiscal benefits are usually small enough
that incentives are unlikely to pay for themselves: the fiscal benefits
will be less than the incentive costs.
Fiscal benefits from incentives will vary greatly, based on local
circumstances. If local roads and schools are already congested, then
growth will require expensive new infrastructure. If existing infra-
structure has plenty of capacity, these infrastructure costs will be less.
But on average, increased public spending needs are likely to use
up at least 90 percent of increased state and local tax revenue. As
we have discussed, a state’s population growth in the long run will
likely be 70–90 percent as great as a state’s job growth. State and
local spending needs in the long run, other things equal, will tend to
track population growth. (Whether there is excess capacity in cur-
rent infrastructure may push spending needs up or down relative to
population growth.) Total income in the state will grow along with
34 Bartik

job growth. State and local tax revenue grows somewhat slower than
overall state income: state income tax rates don’t go up much with
income; consumption of goods covered by the sales tax lags income
growth; property tax revenue increases are often restricted by state
property tax limitations. Therefore, state and local tax revenue is
likely to grow somewhat slower than job growth. The net effect is that
public spending needs will tend to be at least 90 percent of revenue
growth and may exceed revenue growth if expensive new infrastruc-
ture is needed.
The upshot of all this is that incentives are not a free lunch—they
must be paid for. Incentives must be financed in some way, by raising
state and local taxes, or by cutting state and local public spending.

Opportunity Costs

As already mentioned, the higher taxes or lower public spending

used to pay for the net costs of incentives will affect a state’s econ-
omy in two ways: demand-side effects and supply-side effects. The
demand-side effects are the effects of higher taxes or higher spending
on demand by consumers for local goods and services. The supply-
side effects are the effects of higher business taxes or productive pub-
lic services on the supply of capital or labor to the state economy. Of
all these opportunity costs of paying for incentives, the most impor-
tant, because they are the largest and most persistent, are the opportu-
nity costs of reducing spending on productive public services.
Demand-side financing effects of incentives are moderate. For
example, in the baseline model that will be presented in the next
chapter, paying for the net cost of incentives, after allowing for fiscal
benefits, via household tax increases offsets about 7 percent of the net
job creation induced by incentives.53 The intuition is that demand-side
effects are only temporary. While the incentive is being paid for, the
added taxes reduce demand for goods and services. But this effect
is temporary; it goes away after the higher taxes used to pay for the
incentive are over, whether that takes 5 years or 10 years. And it takes
 Multipliers and Leakages 35

at least tens of thousands in added taxes to reduce demand sufficiently

to destroy one job.54 In contrast, the induced jobs from incentives, and
their multiplier effects, are more permanent. A similar argument can
be made for the demand-side effects of reduced consumer spending
due to lower public spending: the demand-side effects on the state’s
economy are temporary; these temporary effects do not count for as
much as the more persistent effects from incentives.
Supply-side effects from higher business taxes are more persis-
tent but are modest in size relative to the potential effects of busi-
ness incentives. Most firms paying higher business taxes are produc-
ing nontradable goods and services. Firms in nontradable industries
base their business activity not on local business taxes but on local
demand. The number of jobs in fast-food restaurants depends on
the local population size and its income, not on how much in taxes
McDonalds pays.
In addition, many firms are not currently considering any invest-
ment decisions. The average business is satisfied with its status quo
scale of operations. For those firms, moderately higher business taxes
will not have the potential for tipping many investment decisions. In
contrast, incentives target tradable-industry firms that are currently
making location or expansion decisions. As a result, a given dollar
amount of incentives has higher job creation effects than a given dol-
lar amount of reductions in general business taxes.
Furthermore, incentives target firms in tradable industries, which
have much higher multiplier effects than firms in nontradable indus-
[41.59.199.65] Project MUSE (2025-04-12 12:39 GMT)

tries. An auto company locating or expanding a plant will create many

jobs in suppliers and local retailers. Incentivizing a local McDonalds
to expand or start up a new facility will destroy as many jobs as it cre-
ates by reducing sales at the nearby Burger King.
But as we will see, the opportunity costs of incentives are large
if the incentives are financed by cutting productive public spending.
Axing needed roads or education has large and extremely persistent
effects in damaging the state economy. This will be explored in more
detail in Chapter 4.
36 Bartik

DIFFERENCES FROM USUAL INCENTIVE MODELS

This model of incentives’ benefits and costs differs greatly from

commonly used state models. Most states only look at the revenue
benefits of incentives for state governments. This is incomplete
even for an analysis restricted to fiscal benefits. It overlooks revenue
effects on local governments. More importantly, commonly used state
models overlook effects on spending needs for state and local public
services.
For example, the Legislative Fiscal Bureau for the state of Wis-
consin, in evaluating the Foxconn deal, made headlines with its con-
clusion that the deal wouldn’t fiscally break even for the state until
2042.55 This news was a big political negative for the Foxconn deal.
But even this negative fiscal analysis was overly optimistic. The
Fiscal Bureau ignored that the additional jobs and people from the
Foxconn deal will require additional public services and hence public
spending. Yet even if we ignore local schools, counties, and other
local governments, this makes no sense. The state of Wisconsin has
revenue-sharing programs for local governments. The state pays a
substantial share of the K–12 school bill. It shares in infrastructure
costs. A sensible fiscal benefits analysis cannot act as if growth is all
revenue gain and no expenditure costs.
Even more importantly, common state models of incentives over-
look economic benefits. States are not businesses seeking to make
a profit. States do not exist to make money; they exist to serve the
interests of state residents. These residents are interested in more and
better job opportunities. It is bizarre to analyze any program aimed
at increasing job growth, which is pursued to improve residents’
job opportunities, and then to ignore such effects in the benefit-cost
analysis.
Commonly used models also explicitly or implicitly assume
that the “but for” for any incented firm is 100 percent—none of the
incented jobs, or any substitutes for those jobs, would have occurred
in the state without the incentives. If one assumes incentives are 100
 Multipliers and Leakages 37

percent successful, is it any wonder that many evaluations of incen-

tives are positive?56
Economic developers sometimes argue that tax incentives have
zero costs for the government and state residents. As one economic
developer said, “It is important to understand that incentives . . . are
not ‘giving away’ tax dollars, they are just taking away fewer of the
company’s dollars by reducing some of their taxes for a period of
time.”57 So, rather than losing tax dollars, we are actually gaining tax
dollars from the taxes we still collect on this new firm: “For every
dollar we ‘did not take’ from this company in taxes to incentivize
them to come to our community we will collect $22 in new taxes—
new tax dollars that would not exist if they choose to go somewhere
else.” But that’s the key issue: Is it true that none of these jobs would
have existed in the state but for the incentives? If three-fourths of the
incented firms would have located in the state anyway, then those
incentives are giving away tax dollars that state and local govern-
ments would have otherwise received.

THE DEVIL IS IN THE DETAILS

As discussion in this chapter shows, the benefits and costs of

incentives for a state’s residents are affected by many factors. In the
next chapter, I report how these factors add up in determining bene-
fits—and what difference is made by specific policies.