Content uploaded by åsa Hansson
Author content
All content in this area was uploaded by åsa Hansson
Content may be subject to copyright.
The Wealth Tax and Entrepreneurial Activity
Åsa Hansson
Department of Economics
Lund University
April, 2006
Abstract:
Entrepreneurship is often credited with generating important positive economic
externalities. For example, entrepreneurs are often credited for promoting innovation,
discovering new markets, and serving as a mechanism for knowledge spillover.
Governments increasingly view encouraging entrepreneurship as an important policy
objective.
Economists have long studied the determinants of entrepreneurship. Taxation has
also been found to be important, in particular income taxes and capital taxes. One form
of taxation that has not been considered so far, however, is the wealth tax. The wealth
tax is likely to influence entrepreneurship negatively, by affecting the pool of capital
available to start up businesses as well as reducing the net return to successful
entrepreneurship.
This paper illustrates the impact of a tax on wealth on entrepreneurship using a
simple model of the choice between becoming an entrepreneur or an employee. Actual
data is then used to crudely investigate whether the wealth tax indeed has a measurable
effect on self-employment in OECD countries, using increasingly sophisticated
techniques. A difference-in-difference type estimator using the abolishment of the wealth
tax as a ”natural experiment” points to a consistent pattern of a perceptible, but small
impact.
Key words: Entrepreneurship; wealth tax; difference-in-difference estimation
JEL classification: H24; H31; J23
1
1. Introduction
Entrepreneurial activity as an engine for economic growth has been recognized since
Schumpeter (1934). It is widely credited with promoting new product innovation,
discovering new markets, and replacing inefficient incumbents in the process coined
“creative destruction”. In the competition between nations, governments have generally
sought to promote entrepreneurship, though with varying degrees of success.
Social scientists have long tried to establish what factors are important
determinants of entrepreneurial activity. For instance, education, risk propensity, assets
as well as economic and social environment have been found to be important.
Strangely, few studies have focused on taxation, even though taxation is
frequently found to be an important explanatory factor. Of the studies that include tax
rates, most have generally found a positive relationship between income tax rates and
entrepreneurship (Long, 1982, Blau, 1987, Parker, 1996, Gordon, 1998, Schuetze, 2000,
Cullen & Gordon, 2002, and Parker & Robson, 2003) -- perhaps because higher income
tax rates give greater incentives to underreport or reclassify taxable income (which is
done more easily for entrepreneurs than employees) -- or a negative relationship between
tax rate progressivity and entrepreneurship (Robson & Wren, 1999, Gentry & Hubbard,
2000, Keuschnigg & Nielsen, 2002, and Gentry & Hubbard, 2004a), plausibly because
tax progressivity reduces financial returns of successful entrepreneurship and, hence, the
probability of entry. In addition, capital gains taxation has been found to retard
entrepreneurship (Poterba, 1989 and Keuschnigg & Nielsen, 2002, and 2004).
Though previously unstudied, entrepreneurial activity may be inversely related to
taxes on individual wealth. Indeed, OECD countries that levied a wealth tax in 2003 had
33 percent lower self-employment rates than those that did not tax wealth. The wealth
tax may influence the probability of becoming an entrepreneur for at least two reasons.
First, because of high risk and asymmetric information it is difficult for small start-up
businesses to obtain external financing, and the importance of own capital for becoming
and staying successful as an entrepreneur is well known (Holtz-Eakin et al. 1994, Lindh
& Ohlsson, 1996, Blanchflower & Oswald, 1998, Davidsson & Henrekson, 2002, and
Gentry & Hubbard, 2004b). By negatively impacting the amount of wealth available,
2
thus, the wealth tax may hinder some potential entrepreneurs from starting their own
business.
Second, expected income is an important driving force for potential entrepreneurs.
By decreasing the expected returns to successful entrepreneurship, the wealth tax may,
hence, negatively impact the probability of seeking to become one.
In this paper, I present a simple model that illustrates how the wealth tax affects
the choice between becoming self-employed or working for another. Specifically, I begin
with the Blanchflower & Oswald (1998) model and extend it to incorporate a wealth tax.
The model shows how a wealth tax impairs self-employment by limiting funds available
and by reducing the incentives to become self-employed.
To examine whether the wealth tax affects self-employment I then analyze actual
data empirically, albeit crudely. A simple comparison of OECD countries that tax wealth
and countries that do not suggests that there is a remarkable difference in self-
employment between the two sets of countries. For instance, during the time period 1980
to 2003 countries that did not tax wealth had on average a self-employment rate that was
2.3 percentage points (24 percent) higher than self-employment in countries that taxed
wealth. More careful analysis, however, reveals that much of the gap can be explained
by other inter-country differences and that the actual effect of the wealth tax is much
smaller. Specifically, by performing simple difference-in-difference estimation using
abolition of the wealth tax in four countries as “natural experiments”, I find that
abolishing a wealth tax increases self-employment by 0.2 to 0.5 percentage points.
2. Reasons why the wealth tax may affect entrepreneurial activity
As mentioned above there are two primary reasons to expect that the wealth tax may
influence the probability of becoming an entrepreneur. First, the wealth tax influences
occupational choice directly by negatively impacting the amount of funds available.
Second, the wealth tax affects the proportion of individuals with entrepreneurial vision by
impacting the expected net profit. A model that captures both mechanisms is developed
as follows.
I start with a simple model based largely on Blanchflower & Oswald (1998).
Individuals choose between working as entrepreneurs or becoming employees. To
3
become an entrepreneur requires both entrepreneurial vision, possessed by β percent of
the population, and capital. There exist a number of potential entrepreneurial projects that
have not yet been developed, each of them requiring a different amount of capital, k. In
addition to capital, each project needs one entrepreneur’s labor. The profit from project k
is π(k), assumed to be strictly increasing. Capital endowment is distributed across the
population with density function, f(k), where k lies between zero and one (normalizing
the wealthiest person’s capital assets at unity). Those individuals lacking required capital
can attempt to borrow with probability z of successfully obtaining a loan.
1
In addition, it is assumed that there is no unemployment. Anyone can find work
in the non-entrepreneurial sector at wage w, which equals the marginal product of labor
in the non-entrepreneurial sector. w is assumed to be declining in N, the number of
employees. Population is normalized at unity and in equilibrium the number of
entrepreneurs is E.
Utility for an entrepreneur is given by
,)(
*
iku +=
π
(1)
where π(k*) is profit from entrepreneurship, i is utility from being independent
2
, and k* is
the amount of capital needed for the marginal entrepreneurial project. For employees the
only source of income is from wage labor and utility, u, is u = w.
Individuals with entrepreneurial vision successively form their own businesses
until, in equilibrium, either capital or vision constraints are binding in aggregate or the
utility from running a business equals that from wage work. In the latter case, w = π(k*)
+ i.
The number of entrepreneurs in the economy, and by choice of units the
probability of self-employment for one individual, is
∫∫
+=
*
0
1
*
)()(
k
k
dkkfzdkkfE
ββ
(2)
.1 N−=
(3)
1
Venture capital is ignored here. The venture capital market is underdeveloped, especially in Europe,
making it unrealistic for many potential entrepreneurs to receive outside financing (Keuschnigg & Nielsen,
2004).
2
Several studies have found that self-employed enjoy higher utility than employees (e.g., Blanchflower &
Oswald, 1998).
4
The total number of entrepreneurs in the economy is, thus, equal to the product of the
probability of having entrepreneurial vision, the number of individuals with required
capital plus the probability of vision, the probability of successfully obtaining a loan and
number of individuals with less than required capital.
It can easily be seen that increases in the proportion of population with vision (β),
the utility of being independent (i), and the probability of obtaining a loan (z) lead to
increases in the equilibrium number of entrepreneurs in the economy.
Assume now that the government taxes wealth at tax rate t and that the tax
revenues obtained are used to finance public goods that neither affect the profit of
entrepreneurship nor occupational choice directly.
3
The wealth tax can affect the number
of entrepreneurs in the economy through several channels. First, it has a direct effect on
the amount of capital available to finance entrepreneurial activity. Even before a wealth
tax was introduced, individuals were constrained from businesses by lack of capital. By
negatively impacting the amount of wealth available, the wealth tax constrains potential
entrepreneurs even further. The marginal project still requires k* but the entrepreneur
now needs k*/(1-t) pre-tax, increasing the number of potential entrepreneurs that are
capital constrained.
The number of entrepreneurs in the economy is now given by
∫∫
−
−
+=
)1/(*
0
1
)1/(*
.)()(
tk
tk
dkkfzdkkfE
ββ
(2’)
To identify how the probability of entrepreneurship is affected by the wealth tax,
equation (2’) is differentiated with respect to t, which gives:
4
).1)(
1
(
)1(
)()(
2
)1/(*
0
1
)1/(*
−
−
−
+
∂
∂
+
∂
∂
=
∫∫
−
−
z
t
k
f
t
k
dkkfz
t
dkkf
tdt
dE
tk
tk
βββ
(3’)
The sign of expression (3’) determines how the wealth tax affects the number of
entrepreneurs. The last term in expression (3’) is negative since z (the probability of
obtaining a loan) is always less than 1, and, hence, the sign is determined by dβ/dt.
3
The government can also redistribute the revenues back to the individuals in form of lump-sum transfers.
4
The distribution of capital endowments is assumed to be unaffected by the wealth tax.
5
Proposition 1. The wealth tax affects the probability of becoming an entrepreneur
negatively if and only if dβ/dt < 0.
Second, the wealth tax influences the proportion of the population with entrepreneurial
vision by reducing the returns to successful entrepreneurship. The proportion of
individuals with entrepreneurial vision can be expressed as,
)),((
π
β
β
E= (4)
where expected profit increases the proportion of entrepreneurial vision and β is
monotonically positive in E(π).
The wealth tax reduces the returns of successful entrepreneurship. Specifically,
since potential profit is assumed to increase the individual’s wealth, the net profit is
reduced by the wealth tax to (1-t)E(π(k)). This, in turn, suppresses the proportion of the
population interested in starting their own business. The wealth tax, thus, negatively
influence entrepreneurial vision, β.
5
Proposition 2. dβ/dt is negative since the wealth tax negatively affects net expected
profit and therefore the incentives to become entrepreneur.
3. An Empirical Analysis
3.1 Data
I study the impact of the wealth tax on entrepreneurship using data for a large sample of
OECD countries. This sample provides both wealth tax and non-wealth tax countries.
Unfortunately, data on entrepreneurial activity are not readily available; self-employment
is therefore used instead as is standard in empirical work. Specifically, I use data from
OECD on the share of self-employed excluding farmers over total non-agricultural
employment for 22 OECD countries.
3.2 Simple Estimates
5
Even if many countries that tax wealth exempt wealth held in firms from taxation, more consumable
wealth as, for instance, property is generally subject to wealth tax which negatively affects the returns of
successful entrepreneurship. In France and Sweden for instance, a shareholder holding more than 25
percent of the shares in a company is exempt from paying wealth taxes on the holding. In Spain, only 15
percent is required to be exempt.
6
Currently, eight OECD countries (Finland, France, Iceland, Luxembourg, Norway, Spain,
Sweden, and Switzerland) tax individual wealth. Figure 1 plots the average fraction of
self-employment (in non-agricultural sectors) between 1980 and 2003 separately for
countries that had a wealth tax the entire period (Finland, Iceland, Luxembourg, Norway,
Spain, Sweden, and Switzerland) and countries that did not tax wealth for the entire
period (Australia, Belgium, Canada, Ireland, Japan, New Zealand, Portugal, Turkey, UK,
and the US).
6
As is evident from Figure 1, the difference in self-employment between the two
groups is sizable. The average self-employment rate during this period was 11.9 percent
in the non-wealth tax countries compared to 9.6 percent in the wealth tax countries, a
difference of 2.3 percentage points (19 percent). Not only does the level of self-
employment differ between wealth and non-wealth tax countries, but the change in self-
employment differs as well. For instance, the average yearly growth in self-employment
was 0.63 percent in non-wealth tax countries compared to 0.39 percent in the wealth tax
countries. Figure 1 also suggests that self-employment declined since the mid 1990’s
both in non-wealth and wealth tax countries, with larger decline in wealth tax countries
(2.6 percentage points) than non-wealth tax countries (0.8 percentage points).
Whether these differences are truly due to the wealth tax or caused by other
confounding factors is impossible to disentangle from Figure 1, however. Additional
information can be obtained, however, by looking at the pattern of self-employment in
the four countries that abolished the wealth tax during this period (Austria, Denmark,
Germany, and the Netherlands). Figure 2 graphs self-employment in Austria (abolished
1994), Denmark (abolished 1997), Germany (abolished 1997), and the Netherlands
(abolished 2000), respectively.
All four countries had average self-employment rates well below the average for
the non-wealth tax countries (11.9 percent). Indeed, the average self-employment rate in
these four countries was fully 4 percentage points (34 percent) lower than the non-wealth
tax countries. Interestingly, however, the downward trend in self-employment seen
generally in Figure 1 is not evident here; instead, self-employment seems to be
6
France taxes wealth as well. However, the tax was lifted between 1987 and 1988. Italy instituted a
temporary wealth tax that was in effect between 1993 and 1997. Note that Turkey is not included in Figure
1.
7
increasing. Austria, for example, witnessed an average yearly growth in self-employment
of 2.3 percent after the wealth tax was removed in 1993 in contrast to an average decline
of 1.5 and 0.6, respectively, in wealth and non-wealth tax countries. In Denmark and
Germany, the average growth rate in self-employment was -0.1 and 0.8 percent,
respectively, after their wealth taxes were eliminated in 1997 compared to -2.8 and -0.8
percent for the wealth and non-wealth tax countries. Similarly, self-employment
increased an average of 2.2 percent after the removal of the wealth tax in 2000 in the
Netherlands compared to declines of 3.4 percent and 1.0 percent in wealth and non-
wealth tax countries during the same period, respectively.
3.3 Difference-in-difference estimation
Even the before-after comparisons of countries that abolished wealth taxes during the
studied period are at best indicative and may be confounded by secular trends, so I also
construct estimates based on the formal difference-in-difference approach. Since the data
are necessarily limited, however, results must be interpreted cautiously. Specifically, I
compare the change in self-employment before and after removal of the wealth tax
between countries that abolished the wealth tax (the treatment group) and countries that
either taxed wealth the whole period or never taxed wealth (the control group). This
strategy ensures that any secular trends that are correlated with the wealth tax and self-
employment do not bias the coefficient estimate.
I begin with a simple estimate where the change in the self-employment rates one
year after and the year before the wealth tax was abolished is compared to the same
difference for the control countries for the four countries individually. Specifically, for
each of countries that abolished the wealth tax the change after the wealth tax was
abolished is compared to the change over the same time period for the control countries.
I also estimate the difference-in-difference by comparing the change two years after the
wealth tax was removed and the year before to allow more time for individuals to
respond to the change. In addition, I use different control groups to see how sensitive the
results are to the choice of control group.
Table 1 reports the results. In the first column, both wealth and non-wealth tax
countries are included in the control group. Self-employment increased after the wealth
8
tax was abolished in each country but the increase is much smaller than Figure 1 suggests
after secular trends were removed by the difference-in-difference estimation. On average,
removing the wealth tax increased self-employment by 0.5 and 0.7 percentage points at
one year and two years after the wealth tax was removed, respectively, though there is
variation across the four countries. In the Netherlands self-employment grow by 1
percentage point, for example, while the increase in Denmark was much smaller.
The estimates are potentially contaminated, however, because the control
countries include wealth tax countries that, while they did not abolish the wealth tax, may
have made other substantial changes to the tax rates and allowances allowed making
them poor control candidates. To avoid this problem, I also present estimates (column 2)
with only countries that never taxed wealth as the control group. While this did not
change the average results appreciably, 0.55 and 0.66, respectively, the effect for
Denmark declined substantially (indeed negative for two years after removal) and the
estimates for Austria increased.
The four countries that abolished the wealth tax may differ systematically from
the other countries; for instance, they are high-tax countries and their low self-
employment rates may be a result of high-tax countries having lower self-employment
(Fölster (2002)). To mitigate this problem, I also estimate the difference-in-difference
using only high-tax countries in the control group (Belgium, Finland, France, Iceland,
Norway, and Sweden) in column 3.
7
Again, the results are generally unaffected.
Finally, I estimated the model using neighboring countries and countries with
similar economic conditions as controls (columns 4 and 5, respectively). Specifically,
France and Germany are Austria’s controls, Finland and Norway are Denmark’s controls,
Austria and France are Germany’s controls, and France and Luxembourg are the
Netherlands’ controls. Again, the results are robust and the average effect of removing
the wealth tax is still approximately a 0.5 percentage point increase in self-employment.
3.4 Difference-in-difference estimation after controlling for other covariates
7
Note that Belgium is the only country that did not tax wealth. Excluding Belgium does not change the
results noticeably, however.
9
These simple difference-in-difference estimates are problematic in several respects,
however. First, they rest on a key assumption that the average change in outcome had
been the same for the treatment and control group had the wealth tax not been abolished.
That is, other factors, such as economic conditions, must affect the treatment and the
control group similarly. If that is not the case additional explanatory variables must be
included. Second, if the treatment group is not truly randomly assigned the difference-in-
difference estimates may be biased. This could happen if the removal of the wealth tax
was instituted to improve the entrepreneurial climate. While it is hard to deal with the
latter problem, performing regression analysis that allows for other control variables can
potentially mitigate the former problem.
Performing a standard difference-in-difference estimation with covariates are
problematic in this setting not only because the countries abolishing the wealth tax are
few but also as the removal of the wealth tax occurred in different years for the various
countries. One way to deal with this is to “rinse out” the effect of other potential
important explanatory variables before performing the difference-in-difference
estimation. Specifically, I rinse out real GDP per capita, unemployment, transfer as share
of GDP, variables that have been found to explain self-employment in other studies (e.g.,
Robson & Wren, 1999, Fölster, 2002, and Parker & Robson, 2003), as well as time and
country specific effects and estimate the difference-in-difference estimates on the
resulting residuals using the same control groups as in Table 1.
Table 2 reports the results. In general, the estimates presented in Table 2 are
considerably smaller than those presented in Table 1 suggesting that other factors
explained part of the difference in self-employment between those that abolished the
wealth tax and the controls found in Table 1. In particular, real GDP per capita,
unemployment, transfers, time and country fixed effects are found to impact self-
employment in the treatment and control countries differently. Self-employment
increased on average with 0.2 percentage points when the wealth tax was abolished using
wealth and non-wealth tax countries as control. When different control groups are used
the increase is similar; although the increase is smaller when non-wealth and high-tax
countries are used as controls. For the individual countries the difference-in-difference
estimates for Austria and the Netherlands are especially lower compared to those in Table
10
1, suggesting that the increase found for Austria in Table 1 is largely driven by other
factors as are a large fraction of the increase for the Netherlands.
Two other countries, France and Italy, both abolished and reintroduced wealth
taxes during the study period and can provide additional information about the effect a
wealth tax has on self-employment. As the changes were temporary, the information that
can be drawn from the experience in these countries is limited, however. Taxpayers in
France were given a temporary relief from the wealth tax between 1986 and 1988. Italy,
on the other hand, introduced a wealth tax in 1993 and abolished the same tax in 1998.
Table 3 reports difference-in-difference estimates on the changes in the wealth tax in
France and Italy using the same control groups as in previous tables when controlling for
real GDP, unemployment, transfers, time and country specific effects. The control
country in the last row in the upper and lower part of the table is Italy for France and
Spain for Italy.
The effect on self-employment from removing the wealth tax in France was
positive, albeit small. Reintroducing the tax a few years later had a negative (also small)
effect on self-employment as expected. For Italy, the effect of introducing the wealth tax
actually had a positive, but small, effect on self-employment, in all rows but the last.
Abolishing the wealth tax a few years later had a positive and large effect, however, at
least when compared to Spain.
A potential problem with these estimates is that the elimination of the wealth tax
was in many cases part of a larger reform that included other tax changes as well that
could affect self-employment. The wealth tax reform in the Netherlands, for instance,
was part of a general tax reform that removed the wealth tax but replaced it with a 30
percent tax on theoretical revenue from capital, assumed to equal 4 percent of net assets
while exempting actual capital income from taxation. In Austria, the wealth tax removal
was also part of a larger tax reform that increased general tax credits for all taxpayers and
to compensate for the revenue loss increased the corporate tax rate. To remove the effect
of other simultaneous tax changes that took place, I control for various tax rates (the tax
gap between labor and capital income taxation, as well as the marginal tax rate on labor
and capital, and the corporate tax rate) in addition to already included covariates before
performing the difference-in-difference estimation. When the tax gap and corporate tax
11
rates are controlled for the effect of the wealth tax removal in the Netherlands is reduced
dramatically to around 0.10. For the three other countries the change is much smaller.
Controlling for the corporate tax rate increases the estimate for Austria from roughly zero
to about 0.25 percentage points. For the other countries the estimates are generally lower
but are less sensitive to the inclusion of the corporate tax rate. Moreover, the estimates
for Italy presented in Table 3 are sensitive to the inclusion of the corporate tax rate as
well as capital gain taxes. In contrast to Table 3, the introduction of a wealth tax in Italy
has a negative effect on self-employment when corporate tax rates are controlled for and
a negative or near zero effect when capital gain taxes are included.
4. Discussion
Empirically, I find consistent indications of a perceptible, but small, negative impact of a
wealth tax on self-employment as predicted by the theoretical model. Other factors such
as perhaps attitudes toward risk, overall business friendliness, and labor market
institutions that vary across countries are likely to be more important than the wealth tax.
These estimates should be interpreted with care, however, since the study suffers
from several shortcomings. First, the number of countries that abolished a wealth tax is
admittedly small severely limiting the econometric techniques available. Moreover, the
removal took place at different times, requiring strong assumptions to pool the four
treatment countries and the control countries.
Second, elimination of the wealth taxes may not have been random. If the wealth
tax reforms sought to boost low self-employment, they may be unfit to use as “natural
experiments”. Even if the reforms were not designed to boost self-employment but to
enhance efficiency and improve economic performance that may indirectly affect self-
employment as well, however. Unfortunately, this problem is hard to overcome with
currently available data and methods.
Third, the wealth tax may have offsetting positive effects on self-employment
making the overall effect small. In many countries, the wealth tax has been criticized for
exempting the very wealthy from the tax. One reason for that is that “working capital”
held in businesses is exempt.
8
This provides incentives for wealthy individuals to
8
See footnote 5.
12
become self-employed in order to escape taxation more easily. If many individuals
become self-employed in order to hide wealth this effect can outweigh the potential
negative effects a wealth tax may have.
9
Unfortunately, I am unable to disentangle this
effect from the total effect.
5. Conclusions
Promoting entrepreneurship is widely recognized as an important policy objective. In
general, income taxation has been found to have a positive impact on self-employment
while the progressivity of the income tax as well as capital gains taxation have been
found to affect self-employment negatively. One form of taxation that has so far not been
considered, but likely affect self-employment, is the wealth tax.
This paper develops a simple model illustrating how the wealth tax can impede
self-employment by reducing the amount of capital available and by reducing the pay-off
to successful entrepreneurship.
Actual data from 22 OECD countries indicate that countries that do not tax wealth
have systematically higher self-employment than countries that do tax individual wealth.
Average self-employment, indeed, was 2.3 percentage points (24 percent) higher in
countries without a wealth tax than in countries that taxed wealth over the time period
1980 to 2003. Difference-in-difference estimation using the elimination of wealth taxes
as a “natural experiment”, however, indicates that the boost in self-employment due to
the removal of the tax is likely much smaller, some 0.2 to 0.5 percentage points,
suggesting that the systematically higher self-employment in countries without taxes on
wealth is likely an artifact of other factors correlated mutually with the wealth tax -
perhaps overall business-friendliness, attitudes toward risks, or social norms.
Nevertheless, it must be cautioned that these estimates suffer a number of serious
limitations. For instance, the data available are limited and the abolishment of the wealth
tax took place different years in the various countries making it difficult to construct
control groups. The main strength of the results, however, lies in the consistency across a
wide number of different specifications.
9
Note that self-employment driven by tax incentives is undesirable from a social perspective. Those
becoming self-employed for tax reasons may do so only part time, however, reducing the problem.
13
This paper should be seen as a first attempt to study the effect a wealth tax has on
entrepreneurship and inspiration for future research, preferably using micro level data to
address the issues discussed here as well as to disentangle how the wealth tax affects self-
employment (by limiting available capital, reducing the net return, or by impacting risk-
taking and effort).
14
References
Blanchflower, D. & A. Oswald, 1998, What Makes an Entrepreneur?, Journal of Labor
Economics, 16(1), 26-60.
Blau, D., 1987, A Time-Series Analysis of Self-Employment in the United States,
Journal of Political Economy, 95(3), 445-467.
Cullen, J. & R. Gordon, 2002, Taxes and Entrepreneurial Activity: Theory and Evidence
for the US, NBER Working Paper, No 9015.
Davidsson P. & M. Henrekson, 2002, Determinants of the Prevalence of Start-ups and
High-Growth Firms, Small Business Economics, 39, 81-104.
Fölster, S., 2002, Do Lower Taxes Stimulate Self-Employment, Small Business
Economics, 19, 135-145.
Gentry, W. & G. Hubbard, 2000, Tax Policy and Entrepreneurial Entry, American
Economic Review, 90, 283-287.
Gentry, W. & G. Hubbard, 2004a, Tax Policy and Entry into Entrepreneurship, Mimeo.
Gentry, W. & G. Hubbard, 2004b, Entrepreneurship and Household Savings, Advances in
Economic Analysis and Policy, 4(1).
Gordon, R., 1998, Can High Personal Tax Rates Encourage Entrepreneurial Activity?,
IMF Staff Paper, Vol 45(1), 49-80.
Hansson, Å., 2002, The wealth tax and economic growth, Scandinavian Working Papers
in Economics, No. 2002:20.
Holtz-Eakin, D., Joulfaian, D., & H. Rosen, 1994, Sticking it Out: Entrepreneurial
Survival and Liquidity Constraints, Journal of Political Economy, 102(1), 53-75.
Keuschnigg, C. & S. B. Nielsen, 2002, Tax policy, venture capital, and entrepreneurship,
Journal of Public Economics, 87, 175-203.
Keuschnigg, C. & S. B. Nielsen, 2004, Start-ups, venture capitalists, and the capital gains
tax, Journal of Public Economics, 88, 1011-1042.
Lindh, T. & H. Ohlsson, 1996, Self-Employment and Windfall Gains: Evidence from the
Swedish Lottery, Economic Journal, 106(439), 1515-1526.
Long, J., Income Taxation and the Allocation of Market Labor, Journal of Labor
Research, 3, 259-276.
15
Parker, S., 1996, A Time Series Model of Self-employment under Uncertainty,
Economica, 63, 459-475.
Parker, S. & M. Robson, 2003, Explaining international variations in entrepreneurship:
evidence from a panel of OECD countries, Mimeo, University of Durham.
Poterba, J., 1989, Capital Gains Tax Policy Toward Entrepreneurship, National Tax
Journal, 42(3), 375-389.
Robson, M. & C. Wren, 1999, Marginal and Average Tax Rates and the Incentives for
Self-Employment, Southern Economic Journal, 65(4), 757-773.
Schuetze, H., 2000, Taxes, Economic Conditions and Recent Trends in Male Self-
employment: A Canada-US Comparison, Labour Economics, 7, 507-544.
Schumpeter, J., 1934, The Theory of Economic Development, Harvard University Press,
Cambridge, Mass.
16
Figure 1. Average self-employment rates in wealth and non-wealth tax countries,
respectively, over 1980 to 2003.
0
2
4
6
8
10
12
14
16
1
9
8
0
1
9
8
1
1982
1983
1
984
1
9
8
5
1986
1987
1988
1
9
8
9
1
9
9
0
1991
1992
1993
1
9
9
4
1
9
95
1996
1997
1
9
9
8
1
9
9
9
2
0
00
2001
2002
2
0
0
3
year
Self-employment in percent
se without wealth tax
se with wealth tax
Figure 2.
Average self-employment in countries that abolished the wealth tax.
0
2
4
6
8
10
12
1980
1981
1
9
82
1
9
83
19
84
1985
1986
1
9
87
19
88
19
89
1990
1991
1
9
92
19
93
19
94
19
9
5
1996
1
9
97
19
98
19
99
2000
2001
2
0
02
2
0
03
Year
Self-employment in percent
Austria94
Denm97
Germ97
Neth00
17
Table 1
. Difference-in-difference estimates of the effect of wealth tax eliminations on
self-employment (in percentage points) in four countries that abolished the wealth tax.
Country
-before, after
Wealth and
non-wealth
tax countries
Non-wealth
tax
countries
High tax
countries
High tax
countries with
wealth tax
Neighboring
and similar
countries
Austria
-1993,1995
-1993,1996
0.61
0.46
0.80
0.63
0.45
0.34
0.40
0.36
0.88
0.55
Denmark
-1996,1998
-1996,1999
0.02
0.42
-0.087
0.26
0.07
0.41
0.07
0.41
0.07
0.45
Germany
-1996,1998
-1996,1999
0.54
0.76
0.43
0.59
0.59
0.74
0.59
0.74
0.31
0.36
Netherlands
-1999,2001
-1999,2002
1.01
1.14
1.07
1.17
1.02
1.14
0.99
1.12
0.61
0.79
Average
-one year after
-two years after
0.54
0.69
0.55
0.66
0.53
0.65
0.51
0.66
0.47
0.54
18
Table 2. Difference-in-difference estimates on the effect of wealth tax eliminations on
self-employment (in percentage points) after controlling for real GDP per capita,
unemployment, transfer, time and country specific effects.
Country
-before, after
Wealth and
non-wealth
tax countries
Non-wealth
tax countries
High tax
countries
High tax
countries with
wealth tax
Neighboring
and similar
countries
Austria
-1993,1995
-1993,1996
0.04
0.03
0.04
0.02
0.04
0.03
0.05
0.05
-0.06
-0.15
Denmark
-1996,1998
-1996,1999
0.19
0.30
0.11
0.20
0.15
0.19
0.17
0.22
0.25
0.26
Germany
-1996,1998
-1996,1999
0.26
0.37
0.17
0.27
0.21
0.27
0.23
0.29
0.10
0.17
Netherlands
-1999,2001
-1999,2002
0.16
0.28
0.12
0.23
0.10
0.18
0.11
0.20
0.44
0.60
Average
-one year after
-two years after
0.19
0.25
0.11
0.18
0.13
0.17
0.14
0.19
0.18
0.22
19
Table 3. Difference-in-difference estimates on the effect of wealth tax introduction and
abolishment on self-employment in Italy and France (in percentage points) after
controlling for real GDP per capita, unemployment, transfer, time and country specific
effects.
Country
-before, after
Wealth and
non-wealth
tax countries
Non-wealth
tax countries
High tax
countries
High tax
countries with
wealth tax
Neighboring
and similar
country
France
Abolishing
-1985,1987
-1985,1988
0.13
0.06
0.09
0.04
0.19
0.12
0.24
0.15
0.13
0.06
France
Reintroducing
-1988,1990
-1988,1991
0.15
0.17
-0.05
-0.07
-0.10
-0.13
-0.12
-0.16
-0.35
-0.14
Italy
Introducing
-1992,1994
-1992,1995
0.14
0.12
0.13
0.12
0.09
0.06
0.11
0.07
-0.27
-0.14
Italy
Abolishing
-1997,1999
-1997, 2000
0.24
0.36
0.18
0.25
0.17
0.24
0.19
0.26
1.45
2.07
20
Appendix
Data description
Variable Definition Source Mean Standard
Deviation
Self-
employment
Share of self-
employed over total
non-agricultural
employment
OECD Economic
Outlook
11.3 4.9
GPD per capita
Real GDP per capita;
PPP adjusted in
constant dollars
OECD Economic
Outlook
20291 6954
Unemployment
Unemployment in
percentage of the
labor force
OECD Economic
Outlook
5.2 3.6
Total transfers
Government transfers
as a share of GDP
OECD, Economic
Outlook
16.1 5.6
Marginal tax
rate on labor
The marginal tax rate
on labor facing an
income earner making
twice the average
production worker
Hansson, 2002 44.5 12.9
Marginal tax
rate on capital
The marginal tax rate
on capital gains facing
an income earner
making twice the
average production
worker
Hansson, 2002 28.3 20.1
Corporate tax
rate
The corporate tax rate Hansson, 2002 37.9 7.4
Tax gap
The difference
between the marginal
tax rate on labor and
capital
Own calculations
based on Hansson,
2002
16.1 17.7