The evolution of owner-entrepreneurs’ taxation: five tax regimes over a 160-year period

The institutional literature suggests that long-term tax incentives are crucial for entrepreneurs, but studies on this topic are hampered by problems related to how to define and measure entrepreneurial income. We resolve these problems by drawing on a theoretical definition of the entrepreneur as an owner, which enables us to identify entrepreneurship empirically by means of investments made by active owners of closely held corporations. Using detailed Swedish tax data, we analyze the tax incentives for such owner-entrepreneur investments from 1862 to 2018, thereby highlighting the evolution of a general institutional phenomenon through a long-run, in-depth, country-specific analysis. We calculate the annual marginal effective tax rate (METR) on capital income for investments, distinguishing between average- and top-income entrepreneurs, and between three sources of finance. We identify five tax regimes that indicate substantial differences in institutional quality over time according to the magnitude of the METR and METR differences between average- and top-income entrepreneurs and across sources of finance. Growth-conducive tax incentives shed light on why so many successful entrepreneurial firms were founded in Sweden around 1900, whereas increased taxation helps explain the absence of new large entrepreneurial firms in Sweden after World War II. Improved incentives can be associated with Sweden’s recent entrepreneurial renaissance.


Introduction
Whether people are able and willing to engage in productive entrepreneurship depends largely on the institutional environment (Baumol 1990;Stenholm et al. 2013), of which the tax system forms a crucial part (Bjørnskov and Foss 2013;Hall and Jones 1999;Henrekson and Johansson 2009). However, while researchers understand the tax system's impact on economic behavior and performance in the short run, few have studied the long-term evolution of the tax incentives that entrepreneurs face. A long-run perspective is informative because formal institutions such as tax systems generally change slowly and entrepreneurs need time to adjust.
That said, three related problems -one theoretical, one empirical, and one data-related -complicate the examination of entrepreneurs' long-term tax incentives. First, there is no clear theoretical consensus on how entrepreneurship should be defined (Hébert and Link 2006). Second, there is an empirical problem related to the fact that tax codes never recognize entrepreneurial income as a distinct tax category. Third, there is a general lack of sufficiently long and detailed time series. By addressing these three problems, we can examine the long-term evolution of the taxation of Swedish owner-entrepreneurs, adding to the literature on institutions and entrepreneurship in several ways.
First, we theoretically associate entrepreneurship with profit-oriented ownership following the argument of Knight (1921) that the pursuit of profit is entrepreneurs' major motivation in introducing innovations and that they can only exercise entrepreneurial judgment when they own productive resources. The judgment-based definition suggests that firm ownership should be concentrated in entrepreneurial firms because entrepreneurs prefer to appropriate as much of the potential profit possible and because entrepreneurs and external investors may disagree about the value of entrepreneurs' investment projects, especially in the early phases. Moreover, scholars holding this view generally argue that the unit of analysis in entrepreneurship should be investments rather than, e.g., opportunities (Foss and Klein 2012: 102).
The solution to the second (empirical) problem follows from our solution to the first. Here, we treat owners of closely held corporations as a relevant pool of potential entrepreneurs. Swedish tax law also differentiates between active owners, i.e., owners who take part in corporate governance and development, and passive owners, who merely provide capital. Thus, the theoretical definition enables us to resolve the problem of the lack of a tax category for entrepreneurial income by capturing entrepreneurship as investments made by active owners of closely held corporations, which serve as an empirical approximation that is theoretically congruent with the judgment-based entrepreneurship perspective.
Third, we resolve the time-series problem by exploiting Swedish tax data from 1862 (when Sweden implemented a major new state central government tax system) until 2018. This period encompasses Sweden's industrialization, urbanization, and democratization, as well as the more recent turn from a managed to an entrepreneurial economy (Audretsch and Thurik 2000). While Sweden is often 1 3 The evolution of owner-entrepreneurs' taxation: five tax… considered something of an institutional outlier, we argue that the country's institutional evolution has been similar to other OECD countries: the country was in general characterized by high-quality institutions for the entire period, and only really diverged in terms of tax policy for a few decades in the second half of the twentieth century (de la Escosura 2016). We therefore deem the results of our study to be generalizable beyond the Swedish case.
More specifically, we analyze the evolution of capital income taxation of investments faced by owner-entrepreneurs, defined as active owners of closely held corporations. We calculate the marginal effective tax rate (METR) on capital income by means of the King and Fullerton (1984) method, developed with the explicit purpose of comparing tax rates across countries and investment projects. A recent extension enables us to include the tax rules for closely held corporations for the whole period based on the King and Fullerton framework (Wykman 2022).
To further understand how the evolution of the tax system shapes entrepreneurial incentives and distortions, we distinguish the tax effect on investments when entrepreneurial incomes and sources of finance differ, updating and complementing earlier data with recently compiled tax data on closely held corporations (Wykman 2022). Except for Swedish studies (e.g., Henrekson and Stenkula 2015), we are unaware of any other taxation analysis offering a similar level of detail or coverage. In addition to producing long-term tax series for active owners of closely held corporations as a proxy for entrepreneurs' taxation, our focus is on examining whether the tax system has been characterized by distinct periods with different entrepreneurship incentives.
The analysis reveals that the Swedish tax system experienced dramatic changes from 1862 to 2018; the results suggest the emergence (and disappearance) of several tax regimes with distinct conditions for entrepreneurship, a fact that we confirm econometrically by means of structural break analysis. We identify five tax regimes and characterize them according to three metrics indicative of each regime's entrepreneurial incentives: the magnitude of the METR, the difference in the METR between average-and top-income entrepreneurs, and the difference in the METR according to the source of finance.
Regime I corresponds to the period until the middle of World War I. The regime stands out for its particularly favorable entrepreneurial incentives, with a low METR irrespective of income and source of finance. The METR grew during regime II (lasting until late World War II) and III (ending around the mid-1960s) to peak during regime IV (ending around 1990). The increasingly higher general taxation level likely had adverse effects on entrepreneurial incentives to establish and grow firms, as did the pronounced differences in treatment between top-and average-income entrepreneurs. The favorable treatment of retained earnings over new share issues during regimes II-IV likely favored well-established incumbents with prior profits at the expense of new entrants lacking retained earnings. During regime V, which may be labeled corrective, the METR decreased substantially. Differences between average-and top-income taxation virtually vanished, while taxation differences related to the source of finance decreased.
Our analysis reveals considerable differences in how the institutions of the tax system have affected entrepreneurial investment incentives in Sweden. When we tie these findings to long-term indicators of entrepreneurship, the results suggest that growth-conducive tax incentives for high-impact entrepreneurship help explain the establishment and success of most of Sweden's largest entrepreneurial firms during the first regime. Likewise, the relative dearth of high-impact entrepreneurship during most of the twentieth century can be partly explained by the meager tax incentives that evolved during regimes II, III, and IV. 1 Conversely, regime V's improved incentives likely help explain why Sweden has experienced an entrepreneurial renaissance.
We contribute to the previous literature in two core ways. First, we use a clear theoretical definition of entrepreneurship (the investment activities of the ownerentrepreneur) to arrive at a legal definition (investments made by active owners of closely held corporations) that is sufficiently discriminant to allow empirical study. Second, while elucidating the tax system's long-term effect on incentives to establish and build entrepreneurial firms in Sweden, the study also has a bearing on the broader literature on institutional evolution and quality, highlighting a general phenomenon through an in-depth analysis of a specific country. The analysis is transparent and potentially applicable to other countries.

The owner-entrepreneur
The tax system is one of the fundamental institutions shaping a society's entrepreneurial profile (Bjørnskov and Foss 2013;Hall and Jones 1999;), a fact that Knight (1921) saw clearly, e.g., when arguing that the entrepreneur's primary motivation is to "get rich" and when warning that an excess-profit tax risked discouraging production because the anticipation of unusual profits was "a vital element in the incentive to business activity" (p. 332). Along similar lines, Schumpeter (1918, p. 22), stressed that if the entrepreneurial and competitive process, "baited by profit … were taxed away, that element of the economic process would be lacking which at present is by far the most important individual motive towards economic progress. Even if taxation merely reduced this profit, industrial development would progress considerably more slowly" (cf. Mair and Laramie 2000;Musgrave 1992).
However, Schumpeter argued that the functions of the entrepreneur and the capitalist owner of the means of production could be separated (Henrekson and Jakobsson 2001). By contrast, Knight and his modern followers who see entrepreneurship as a domain for exercising individual judgment argue that entrepreneurship presupposes ownership of a business firm (Bylund 2021; 1 3 The evolution of owner-entrepreneurs' taxation: five tax… Klein 2012), modeling "entrepreneurs as owning, controlling, and combining heterogeneous assets … and deploying these assets within a firm to produce goods and services in anticipation of economic profit" (Foss and Klein 2015: 585;cf. McMullen 2015). As Wennekers et al. (2007, p. 138) state, "(t)here is agreement that entrepreneurs (in the sense of business owners) make judgmental decisions in the face of uncertainty, reap the rewards of perceiving and utilizing opportunities and in the process also run the risk of losing their money and their reputation." Moreover, these scholars argue that the unit of analysis in entrepreneurship should be "the assembly of resources in the present in anticipation of (uncertain) receipts in the future, in other words, investments" (Foss and Klein 2012: 102). However, while scholars holding this view increasingly emphasize and examine how institutional contexts shape entrepreneurship (e.g., Audretsch and Belitski 2021;Foss and Klein 2012), we are unaware of any study that examines how taxation affects owner-entrepreneurs' investment activities in the long run.

Active owners of closely held corporations as owner-entrepreneurs
In the judgment-based view, entrepreneurship is "the act of committing resources in realizing the plan, that is, investing resources and executing the entrepreneurial plan or project" (Foss et al. 2019(Foss et al. : 1204. This definition implies that to capture what entrepreneurship is in a theoretically congruent manner, we should look to active firm owners who make investments. As Brouwer (2002, p. 92), puts it, "Knight's theory portrays investment as a discovery process. Many new ventures will be launched, but only a few will survive and prosper. Such a sketch of events fits actual developments." Thus, because successful entrepreneurs cannot be discernible in advance, identifying the relevant pool of potential entrepreneurs is what matters. Crucially, entrepreneurs generally organize their business activities within limited companies, the organizational form that was likely the most appropriate for firm growth, large-scale activities, and uncertainty management at the beginning of our period of interest (Andersson-Skog and Magnusson 2014). The remuneration accrues to the entrepreneur as an owner, i.e., as the residual once nonowners have been paid. Because sole proprietorships and partnerships generally engage in routine small-business activities (and are usually converted to limited companies if owners wish to expand), we exclude these types of ventures from the analysis.
Closely held corporations make up the lion's share of Sweden's incorporated firms (Andersson et al. 2018), and a major tax reform in 1990-1991 introduced specific tax rates and rules for such firms. In the Swedish tax code (SFS No. 1999: 1229, a closely held corporation is a limited company where four or fewer owners own stock corresponding to more than half of the votes. An owner is active if (s)he or a close family member is, or during the past five years has been, active to a "considerable extent" in the corporation's income generation. Because passive owners are not active in the corporation's income generation, this study does not regard them as entrepreneurs.

Taxation of owner-entrepreneurs
Because investments are central to an entrepreneur's exercise of judgment, the evolution of capital income taxation among owner-entrepreneurs who make investments should offer a consistent view on how entrepreneurial tax incentives change over time. The capital income taxation effect depends on three sets of taxes -personal capital income taxation, corporate income taxation, and wealth taxation.
First, personal capital income taxation is an owner-level category that includes taxes on physical individuals' income from dividends, capital gains, and interest. Generally, such taxes reduce the returns on the cooperative efforts of entrepreneurs and external financiers while also affecting the occupational choice margin, making it less lucrative to leave a salaried position in pursuit of a new business idea. Analyses taking principal-agent problems into account reveal that dividend taxation and capital gains taxation are distortionary for both mature companies and startups (cf. Chetty and Saez 2010;Henrekson and Sanandaji 2016).
Second, corporate income taxation is levied at the firm level. A consistent finding is that corporate taxes reduce investments, discourage equity financing, and encourage debt financing if interest payments are tax deductible, which increases the debt-equity ratio (Huizinga et al. 2008). The disagreements in the tax literature mainly involve the size of the effect and the optimal design of the corporate tax system (e.g., Auerbach et al. 2010). Taxing profits can be expected to negatively affect growth, especially in new and small firms (Michaelas et al. 1999).
Finally, wealth taxation matters for entrepreneurship for several reasons. First, the founder's equity often finances a firm's early phase, although external equity financing is usually necessary if a firm is to grow into a significant industry player (Henrekson and Jakobsson 2001). In contexts where debt finance plays a role, founders frequently pledge personal assets and wealth as collateral to obtain loans (Held et al. 2020). Arguably, more private wealth would enable more entrepreneurial venturing by increasing the supply of informal finance. Additionally, wealth taxation directly affects entrepreneurs' incentives when a tax is levied on the value of their stockholdings. High wealth taxes may even make it difficult for successful entrepreneurs to maintain ownership and control.

Metrics to evaluate owner-entrepreneurs' tax incentives
We combine information on the three components of capital income taxation into a single measure and calculate the METR based on the King and Fullerton method (described in Sect. 3). To understand the incentives, we evaluate the METR based on three metrics impacting the entrepreneurial investment decision.
First, we consider the magnitude of the METR. When the tax components underlying capital income taxation are high, the METR is generally higher, indicating meager entrepreneurial incentives that make entrepreneurial investments less profitable. Second, taxation differences based on income should highlight the incentives to make entrepreneurial investments yielding substantial income and 1 3 The evolution of owner-entrepreneurs' taxation: five tax… wealth. We therefore examine the difference in METR magnitude between top-and average-income entrepreneurs. We classify an average-income entrepreneur as an active owner who pays the same marginal income tax rate as the average production worker (as defined by, e.g., OECD 2010). Similarly, a top-income entrepreneur is an active owner paying the top marginal income tax rate. A large difference reveals the entrepreneur's incentives to seek profit by expanding the firm and is mainly reflected in changes in income tax progressivity.
The third metric is the difference in METR magnitude depending on the source of finance -new share issues, retained earnings, or debt. This difference is crucial, e.g., because new entrepreneurial firms rely more on new share issues and less on retained earnings than mature firms (Gompers and Lerner 2001). While active owners would rely on their own equity rather than jeopardize their independence, larger firms have easier access to debt financing. This discrepancy means that high corporate tax rates coupled with tax-deductible interest payments put smaller firms and potential entrepreneurs at a disadvantage (Davis and Henrekson 1999) while also reducing the retained earnings that can be used to expand ventures after start-up. Thus, a higher corporate income tax rate increases the METR for investments financed with retained earnings and new share issues but decreases the METR for debt, as interest payments are deductible expenses at the corporate level. The personal capital income taxation effect is more ambiguous; dividend and capital gains taxes affect new share issues and retained earnings, and the tax on interest income affects debt-financed investments. Wealth taxes increase the METR in all three cases.

Case selection
A paragon of interventionist policies in the 1970s, Sweden has morphed into a pioneer of deregulation in recent times. Consequently, the country is often treated as an outlier in political economy debates. This outlier status is not warranted for the entire period of 1862-2018, however, as underscored by an examination of de la Escosura's (2016) reconstruction of the Fraser Institute's Economic Freedom Index (excluding the size of government component, for reasons of data availability) for the period 1850-2007 for countries that were OECD members in 1994. Sweden's economic freedom follows the four overall trends of the sample for the period 1850-2007.
Nor was Sweden a high-tax outlier for the first 100 years under study: as late as 1960, "the relative size of the public sector was only marginally above the OECD average" (Henrekson 2005: 441). The subsequent expansion of the welfare state, although "most pronounced in Sweden … was a salient feature of almost all industrialized countries, in particular during the 1960s and 1970s" (Henrekson 2005: 441). Thus, while Sweden went further than other OECD countries, it did not go in a different direction. Because government spending has declined markedly in recent decades, Sweden today "merely" ranks among the top one-third of OECD countries in terms of general government spending (OECD 2021). It is thus reasonable to assume that our analysis is relevant for other countries and that Sweden's divergence in the 1960s and 1970s illustrates what happens when a "normal" country, characterized by high institutional quality, pursues a not-so-normal tax policy path for a few decades, e.g., in terms of taxation of owner-entrepreneurs.
In conclusion, Sweden was a reasonably representative country in terms of tax policy before the 1960s. Hence, it is plausible to assume that entrepreneurs in other countries with a high-quality institutional environment would react to changed tax incentives in similar ways as Swedish entrepreneurs. We therefore deem the results of our study to be generalizable beyond the Swedish case, and that foreign observers can learn from the Swedish experience, e.g., when designing economic policy, cf. Lindbeck (1997).

METR: The King and Fullerton method and structural breaks
The METR is a common summary statistic of investment incentives enabling broad interaction of tax rules, deductions, and credits. By permitting us to consider the Knightian notion of the owner-entrepreneur who invests, the METR for active owners of closely held corporations serves as a relevant proxy for taxation of entrepreneurial activities, enabling us to characterize tax regimes according to the three metrics described earlier (Sect. 2.4). To calculate the METR, we use the King and Fullerton (1984) method, a framework developed to compare tax rates across countries and investment projects (see, e.g., OECD 2007). We focus on investments in machinery to ensure tractability. Because Sweden's tax system is nominal-based, we need to interact the three component taxes of capital income taxation with inflation. 2 Formally, the METR is the difference between the pretax and posttax real rates of return of a marginal investment project divided by the pretax real rate of return. However, the METR is not merely an addition of taxes adjusted for inflation but an equilibrium model that is solved when the present discounted value of the investment profits equals the investment cost and the potential investor is indifferent between the after-tax revenue from the investment project and the after-tax market interest rate.
The model assumes that no (further) tax changes will occur, no repurchase of shares, and that investors can use all tax allowances for investments. A model extension (Wykman 2022) also makes it possible to include specific rules for closely held corporations for the whole period. Moreover, whether the investment is financed with new share issues, retained earnings, or debt alters the equilibrium conditions and affects the results. In summary, we end up with six METR time series (based on two income levels and three sources of finance).
While there have been informal attempts to identify tax regimes (e.g., Du Rietz et al. 2015), we are the first to do so in a formal manner, testing the six METR time series for 1862-2018 for structural breaks individually and jointly. For each series, the purpose is to econometrically examine whether it is possible to identify shorter 1 3 The evolution of owner-entrepreneurs' taxation: five tax… periods within the whole period, i.e., tax regimes, during which the annual METR is similar (in a statistical sense). We do this by calculating the average METR for numerous shorter periods and testing whether the average METR statistically differs between the periods. Structural breaks mark the beginning and end of a period.
We stipulate that the METR is a function of time with a shift parameter β and error term μ. Hence, for each year t, where we estimate β using a standard linear regression model. The main idea behind the analysis of structural breaks is to determine whether a segmentation of t ∈ [1862,2018] exists that significantly improves the model fit (see, Bai and Perron 2003). Assuming structural breaks, we rewrite Eq. (1) as allowing for m breaks in the time interval [1862,2018] and dividing the METR time series into m + 1 segments. The underlying hypothesis is If H 0 holds, no partition (sequence of regressions) explains the METR significantly better over time than Eq. (1). If we reject H 0 , there are between 1 and m structural breaks. To test H 0 , we must both choose the optimal number of breaks and calculate when they occur. The method that we use minimizes the residual sum of squares using a dynamic programming algorithm (Bai and Perron 2003).

Data description: Swedish capital income taxation 1862-2018
Publicly available tax law and tax schedules compiled over several years form the basis for the analysis; see Stenkula et al. (2014, and Wykman (2022) for details. Inevitably, the analysis relies on simplifying assumptions, e.g., concerning income levels and tax allowances, which may affect the magnitude of the METR of any single year when the tax system is progressive. That said, the tax rates differ so much across longer periods that they dominate other effects, meaning that the results mainly reflect the differences in magnitude of the METR between the tax regimes. The tax system's general structure also makes the calculations rather insensitive to different assumptions; for instance, most deductions and allowances are too small to impact the marginal tax rate. The advantage of the King and Fullerton framework is that it illustrates in a straightforward way how capital income is taxed without delimiting the results to a highly specific case with less applicability to the tax system at large (cf. Devereux 2004).
Because earlier studies describe Sweden's corporate income taxation, personal capital income taxation, and wealth taxation at length (Henrekson and Du Rietz 2014;Johansson et al. 2015;Stenkula et al. 2014;Wykman 2022), we only briefly describe the evolution of each category below. (1)

Sweden's personal income taxation
Personal capital income was jointly taxed with other personal income (labor and business income) until the 1990-1991 tax reform, meaning that total income determined the marginal tax rate. However, dividends were tax exempt until 1903 and long-term capital gains until 1965. The tax rate depended on the holding period for capital gains, with a longer holding period corresponding to a smaller proportion of taxable gain. In 1903, a progressive state income tax was implemented, but the local tax system remained proportional. At this time, the top marginal income tax rate was below 10 percent, compared to its peak at approximately 90 percent in the late 1970s. The 1990-1991 tax reform implemented a dual-income tax system and introduced specific rules for closely held corporations. Thereafter, the marginal tax rate on dividends and capital gains from closely held corporations depended only on total income when exceeding a certain amount (the dividend allowance). In practice, this entailed a flat tax rate, generally of 30 percent; however, rules were introduced to prevent the possibility of shifting progressively taxed labor income to capital income, which was taxed at a lower rate. In summary, although the three components of the marginal tax rates of personal capital income followed different trajectories, each increased substantially during the twentieth century until the 1990-1991 tax reform.

Sweden's corporate income taxation
Corporate incomes were taxed according to the same tax tables as personal income before 1911, when personal and corporate income taxation were separated. The corporate tax was progressive, and the top marginal tax rate hovered around 10 percent. When the corporate tax became proportional again in 1939, the rate was approximately 40 percent. After World War II, tax rates increased slowly but consistently, peaking at 52 percent in the late 1980s. Moreover, between 1984 and 1990, the government added an additional profit-sharing tax on corporations to finance so-called wage-earner funds (löntagarfonder), increasing the statutory corporate tax rate by approximately five percentage points. However, possibilities abounded to reduce the statutory corporate tax through allowances and grants, meaning the effective corporate tax rate could be substantially lower, especially for large incumbent firms (Heshmati et al. 2010). While the 1990-1991 tax reform removed most of these options, the statutory tax rate was also substantially cut in steps to 22 percent. 3

Sweden's wealth taxation
Sweden did not tax wealth before 1911. Between 1911 and 1947, wealth taxation was part of the ordinary income tax system, as 1-10 percent (depending on the year) of the taxpayer's wealth was added to his or her taxable income. There were also 1 3 The evolution of owner-entrepreneurs' taxation: five tax… additional income and wealth taxes during and between the world wars and a separate wealth tax on assessed net wealth initiated in 1934. The separate wealth tax increased stepwise from approximately 0.5 percent and peaked at four percent in the early 1980s. Valuation relief and average tax caps occasionally limited the total tax on income and wealth. The tax rate was reduced from the mid-1980s and abolished in 2006. Figure 1 shows how the METR has differed by the source of finance for an averageincome entrepreneur. As seen, the METR for new share issues and retained earnings began to increase in the early 1900s. Following World War I, new share issues became the least favorable option by far, with a METR above 100 percent from 1956 until the 1990-1991 tax reform. 4 Debt finance was generally more favorable than  1902 1907 1912 1917 1922 1927 1932 1937 1942 1947 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 2002 2007 2012 2017 New share issues Retained earnings Debt The METR decreased substantially after the 1990-1991 tax reform, especially for new share issues and debt. Retained earnings became the least favorable financing option, but both the magnitudes and differences between sources of finance were smaller than before. Overall, new share issues were on the same level as debt, arguably because the tax rules for closely held corporations implied a permanent increase in the dividend allowance when investments were financed with new shares issues (in contrast to retained earnings; cf. Wykman 2022). That said, the annual variation was considerable.

The METR: An overview
The general picture from Fig. 1 is echoed in Fig. 2, where we recalculate the METR for top-income entrepreneurs. The METR was basically the same irrespective of the source of finance until the 1903 tax reform made new share issues the least favorable alternative -a situation that persisted until the 1990-1991 tax reform. However, the METR for debt financing fluctuated substantially. At its peak, the top-income METR was 200 percent for new share issues and debt financing but never exceeded 100 percent for retained earnings. After the 1990-1991 tax reform, the top-income METR for all three sources of finance fell, making retained earnings the least favorable option.
Thus far, the analysis highlights several important aspects of the tax incentives facing active owners of closely held corporations. First, in the first fifty years, taxes were low, with negligible differences by entrepreneurial income and source of finance. Second, from then until the 1990-1991 tax reform, new share issues  1902 1907 1912 1917 1922 1927 1932 1937 1942 1947 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 2002 2007 2012 2017 New share issues Retained earnings Debt The evolution of owner-entrepreneurs' taxation: five tax… usually received the least favorable treatment. Third, retained earnings were consistently taxed at lower rates than newly issued equity, which favored incumbent firms relative to entrants. Fourth, top-income entrepreneurs received a less favorable tax treatment for debt financing than for retained earnings from the end of World War I until the 1990-1991 tax reform. Fifth, regime V leveled the playing field, and if anything, retained earnings are the most disfavored source of finance today. As new share issues are considered the most important source of finance for new ventures, this development suggests that the tax conditions for novel entrepreneurship have improved.

Structural breaks and tax regimes
We use OLS models to estimate all regressions. We analyze intercept differences since we are interested in changes in the level of the METR rather than changes in its development over time. The number of structural breaks corresponds to the partition associated with the overall lowest Bayesian information criterion (BIC). First, we calculate structural breaks for the series based on the source of finance presented in Figs. 1 and 2, i.e., three series for both average-income entrepreneurs and topincome entrepreneurs. Table 1 reveals that five out of six series have four structural breaks occurring at approximately the same time, i.e., during or after World War I, during or after World War II, during the late 1960s/early 1970s, and in the early/mid-1990s. The averageincome entrepreneur's METR for debt has only two breaks, in 1963 and 1990.
For tractability, we proceed by creating a new series that is an equally weighted average of the six METR series. 5 As Fig. 3 shows, this merged time series has structural breaks in 1917, 1944, 1967, and 1990, roughly corresponding to those in the individual series.
Four structural breaks imply five tax regimes between 1862 and 2018. The first regime stretched until the middle of World War I, the second until the end of World War II, the third until the late 1960s, and the fourth until the 1990-1991 tax reform,  1916, 1939, 1967, 19901921, 1944, 1967, 1990Retained earnings 1923, 1949, 1972, 19951923, 1949, 1972, 1995Debt 1916, 1939, 1967, 19901963, 1990 when the current regime began. The regimes largely coincide with those Du  propose concerning the evolution of the Swedish tax system and economic policy regimes, based on a graphical analysis of household income taxation.

Robustness checks
We undertook several robustness checks to ensure the validity of our results, as there are several ways to implement a structural change analysis in time series data. To begin with, the results rely on an underlying OLS regression minimizing the residual sum of squares. The regression is on a constant, i.e., only the intercept is analyzed. An immediate extension is to control for a time trend in the data, but this does not significantly affect where the breakpoints take place. The same is the case when we change the penalty criterion (such as BIC, MBIC) and the imposed cost. Additionally, we examine methods that rely on optimizing Gaussian negative log-likelihood instead of minimizing the residual sum of squares. Applying such methods does not significantly change the results either.

Regime characteristics
The first metric by which we characterize the tax regimes (Sect. 2.4) is the magnitude of the METR. Table 2 shows that the METR was low for top-and average-income The second relevant characteristic is the difference in METR magnitude between top-and average-income entrepreneurs. Table 2 shows that the magnitude difference was negligible in the first regime and high in the intervening period (especially during regime III), before being reduced to virtually zero during regime V. This evolution suggests that a substantial disincentive to expand firms appeared and then disappeared in the twentieth century.
The third characteristic is the difference in METR magnitude depending on the source of finance. As seen in Fig. 2, issuing new shares was the least favorable source of finance for top-income entrepreneurs during regimes II-IV. Whereas debt occupied an intermediate position, the treatment of retained earnings was the most favorable. The differences between sources of finance were reduced during regime V to the point that debt and new share issues became more favorable than retained earnings. Corresponding data for the average-income entrepreneur in Fig. 1 convey a roughly similar picture for new share issues and retained earnings, although the differences were smaller in magnitude.
The three metrics highlight the impact that each tax regime had on entrepreneurial activities. First, the generally high METR during regimes III and IV -with levels above 100 percent -likely dampened economic aspirations and incentives to invest. Second, the pronounced differences between top-income and average-income entrepreneurs during regimes II-IV probably disincentivized firms to expand and strive for higher income through high-impact entrepreneurship. Third, the favorable treatment of retained earnings over new share issues as a source of finance during regimes II-IV likely benefited well-established incumbents with prior profits at the expense of new entrants lacking retained earnings. The favorable treatment of debt financing over new share issues during regimes III and IV further favored large, capital-intensive incumbent firms with readily available collateral over new, small firms with less capital.
It is also useful to consider the regimes in a historical perspective. Such an account suggests that the institutional framework of the polity and economy is "a function of the shared mental models and ideologies of the actors" (Denzau and    Regime II  Regime III  Regime IV  Regime V   North 1994, p. 27). For starters, regime I, with its low METR irrespective of income and source of finance, overlaps with a period of liberalization of the Swedish economy in which the liberal minister of finance Carl Johan Gripenstedt played a pivotal role. He was heavily influenced by the French liberal economist Frédéric Bastiat. The second tax regime coincides with a shift in the balance of power to the Social Democratic Party. The party would govern Sweden for most of regimes II, III, and IV, during which the METR increased and diverged considerably depending on income and source of finance. The Social Democrats had a unique opportunity to form society according to their largely egalitarian ideology, and economists favoring a big state and government interventionism had considerable influence on economic policy, e.g., Nobel laurate in economics Gunnar Myrdal. Tax policy was deliberately designed to convert companies to 'social enterprises without owners' and to create an economic system of 'capitalism without capitalists' (Henrekson et al. 2020). This policy became most pronounced throughout tax regime IV, during which the Swedish economy started showing several shortcomings, e.g., lagged economic growth.
Kjell-Olof Feldt, the Social Democratic minister of finance during the 1980s, identified the tax system as a major cause of the poor economic performance. He therefore initiated the 1990-1991 tax reform together with the Liberal Party. This marked the beginning of the fifth tax regime, which may be labeled corrective. The objective was to increase efficiency without lowering the tax burden, for instance by financing reduced marginal tax rates by broadened tax bases.

Discussion: Tax regimes and high-impact entrepreneurship
Our results suggest that the increased taxation of active owners' investments made high-impact entrepreneurial initiatives less favorable after regime I and particularly during regimes III and IV. The 1990-1991 tax reform made the tax system friendlier to entrepreneurship by sharply reducing the magnitude and differences in the METR depending on owners' income and source of finance. Table 3 summarizes the tax regimes' characteristics and entrepreneurial incentives while also including three entrepreneurial proxies gathered and employed in previous research that cover this extensive period. The proxies suggest that regimes characterized by favorable entrepreneurial incentives exhibit a greater degree of entrepreneurial and innovative activity than regimes characterized by meager incentives.
The first and second measures center on the establishment year of Sweden's largest entrepreneurial firms in terms of turnover and employment. Both measures are based on Bornefalk's (2017) systematic analysis of Sweden's 100 largest corporations in 2013 (an update of Axelsson 2006). Bornefalk (2017) classifies 36 of the largest firms by turnover and 31 of the largest firms by employment as genuinely entrepreneurial. The definition means that these firms were founded on individual entrepreneurs' intention to commercialize radically new innovations, with firm growth emanating from one core company centered on one innovation. The firms originate and operate within a diverse set of industries, encompassing traditional sectors like forestry (Holmen), building (Skanska), and manufacturing (Atlas Copco), to name a few, but also industries like 1 3 The evolution of owner-entrepreneurs' taxation: five tax… Table 3 Tax regimes characteristics, entrepreneurial incentives, and three entrepreneurship indicators

Note:
Regime break years are based on the merged series displayed in Fig. 3 Largest entrepreneurial firms(turnover) shows how many of the 36 largest entrepreneurial turnover firms in 2014 were founded during each tax regime. Largest entrepreneurial firms (employment) shows how many of the 31 largest entrepreneurial employment firms were founded during each tax regime. Important entrepreneurial innovations shows how many of the 59 greatest innovations in 2017 were entrepreneurially generated during the different tax regimes Source: Bornefalk (2017) Focusing on the period when firms were founded is warranted given that early growth is an important characteristic of large entrepreneurial firms (e.g., Coad and Karlsson 2022). That said, comparisons over time using this data risk being complicated by the fact that firms founded during the earlier regimes have had more time to grow, but also potentially experienced more economic downturns and periods of economic distress. If the growth effect dominates, which seems reasonable as firms that have already grown large should be more robust to downturns, there is a risk that comparisons over time underestimate entrepreneurship in the latter periods relative to the earlier periods.
As many as nineteen of the largest entrepreneurial firms in terms of turnover started during tax regime I. Another six started during regime II, and another six started during regime III. Regime IV, however, produced only one entrepreneurial firm, while regime V had already produced four such firms by 2013. The pattern is similar when we consider the largest entrepreneurial firms in employment terms. Fifteen started during regime I, five during regime II, and eight during regime III. Again, regime IV scores worst, producing zero such firms, whereas regime V had produced three such firms by 2013. These indicators suggest that more successful entrepreneurial firms were founded during regime I, when tax incentives were beneficial, and probably somewhat understate the positive change during regime V, as firms founded during this period have had less time to grow.
The third measure is based on Sandström's (2014) analysis of Sweden's 100 most prominent innovations. He classifies 59 of them as originating with individual inventor entrepreneurs or with established entrepreneurial firms from four innovative sectors (this includes obvious top innovations like spherical ball bearings, the milk separator, and the tetrahedron). As seen, most top innovations were generated during regime I. Even though the number of innovations was relatively high in the postwar period during regime III, it was markedly lower during regimes II and IV but increased again during regime V. Again, the numbers probably understate the trend during regime V, as innovations in this group have had less time to have a worldwide impact. 6 Other (more fragmented) evidence also suggest that regime V has ushered in something of an entrepreneurial renaissance. Notably, while data for the period 1920-1991 suggest that the average startup rate in the manufacturing industry declined considerably during regimes II-IV (Braunerhjelm and Carlsson 1993), microdata analysis on the share of all young firms for the period 1990-2013 shows "fewer signs of a declining share of entrants. This evolution thus seems to have been halted in the early 1990s" (Heyman et al. 2019(Heyman et al. , p. 1811. 7 Regime V also saw a substantial increase in the number of new jobs in the private sector (Bjuggren and Johansson 2009) and an increase in the share of employees in medium-sized Swedish firms , which had decreased substantially during regime IV (Henrekson and Johansson 1999). In contrast, the number of new and young firms and their level of employment deteriorated during the post-1990 period in the United States (Heyman et al. 2019). Likewise, following a relatively stable period during the twentieth century, stock market capitalization skyrocketed during regime V (Henrekson and Jakobsson 2012) with a record number of newly listed firms (Holmén and Högfeldt 2005). Sweden has also come to have one of Europe's largest buyout sectors, enabling successful spin-outs of numerous divisions from old incumbents (Tåg 2012). In fact, Sweden is presently second only to Silicon Valley in spawning multibillion-dollar tech companies per capita (Frier 2018). The improved conditions of owner-entrepreneur taxation during regime V likely helped pave the way for this increased diversity.

Conclusion
The tax system affects the structure of payoffs for society's economic actors, notably entrepreneurs. Presently, however, there are few rigorous studies tracing the long-run evolution of tax system incentives for entrepreneurship, possibly because real entrepreneurship taxation is a complicated combination of several taxes, financing options, and inflation. Specifically, problems arise from the lack of a generally accepted definition of entrepreneurship, the fact that tax codes do not recognize entrepreneurial income as a distinct tax category, and the absence of detailed, long-term data. We addressed these problems as follows. First, we theoretically emphasized the importance of ownership for entrepreneurship and investment as its most important manifestation. Second, because it is reasonable to assume that a subset of the active owners of closely held corporations have entrepreneurial intentions, we identified the study of such owners' tax incentives when they make investments as highly relevant from an entrepreneurial perspective. Third, the compilation of detailed Swedish data enabled us to study these issues by examining Sweden's capital income taxation of entrepreneurial owners between 1862 and 2018. Thus, the study highlighted a general phenomenon through an in-depth, long-run analysis of a specific country.
We made calculations for average-and top-income entrepreneurs who make an investment financed with new share issues, retained earnings, or debt and included the effects of corporate income taxation, capital income taxation, and wealth taxation (and the interactions of these taxes with inflation). The exercise enabled us to econometrically identify five tax regimes, which we characterized in terms of the magnitude of the METR and differences in the METR depending on owners' income and source of finance. Regime I provided entrepreneurs with incentives conducive to the establishment and growth of firms: income taxes were low, in principle flat, stable, and relatively neutral regarding the source of finance. These incentives weakened during the second regime and more still during regimes III and IV, a development that likely resulted in significant impediments to high-impact entrepreneurship in the postwar era. Notably, progressivity made it challenging to increase personal income by expanding firms, a tendency strengthened by the high taxation of investments financed with new share issues, the preferable financing source for novel entrepreneurship. The relatively more favorable treatment of retained earnings (on which mature firms mainly relied for financing) also impeded industry renewal. These differences help explain why so many successful entrepreneurial firms were founded in Sweden around 1900, and why few such firms were founded after World War II. Arguably, the fifth regime has played a corrective role by strengthening entrepreneurial incentives, as the evidence points to an entrepreneurial renaissance.

Contributions
This study makes two key contributions. First, we demonstrate how a clear theoretical definition of entrepreneurship (the investment activities of the owner-entrepreneur) facilitates the empirical study of entrepreneurial taxation by focusing on the taxation of investments made by active owners of closely held corporations. Because we can follow the taxation of this group for almost 160 years and distinguish between different income and financing categories, we add scope and detail to a literature that has previously focused on short periods. We find evidence of substantial institutional change, as the period saw the emergence of five tax regimes with distinct entrepreneurial incentives. Thus, this long-run perspective on institutional evolution sheds additional light on our understanding of economic performance. The framework makes the analysis transparent and potentially applicable to other countries, offering a promising way to understand observed intra-and intercountry variances in entrepreneurship activity over time. Future studies could also consider the heterogeneous effects of taxation on investments depending on the firm's financial situation and ownership structure.
Second, while our study elucidates the tax system's strong effect on the incentives to establish and build entrepreneurial firms, it also has a bearing on the broader literature on institutional quality and institutional change, highlighting a general phenomenon through an in-depth analysis of a specific country. Connecting our results with the economy's entrepreneurial activity -such as the prevalence of large entrepreneurfounded firms -suggests that the institutions surrounding the tax system critically influence entrepreneurial behavior.

Limitations
Like any study, ours has its limitations. Notably, we do not econometrically establish a link between the identified tax incentives and empirical manifestations of entrepreneurship. This problem arises, first, because we only have one long time series, and second, because there is, to our knowledge, no single satisfying entrepreneurial proxy covering the entire period that offers sufficient annual variation to be meaningfully included in a regression framework. This is the cost of undertaking a long-term study of one country instead of a cross-country analysis covering several economies for a limited period (cf. Stenholm et al. 2013). However, it seems beyond doubt that Sweden's postwar period, and especially regime IV, was less entrepreneurial than regimes I and V (in terms of, e.g., the founding of new, successful entrepreneurial enterprises). Our analysis suggests that these entrepreneurial ebbs and flows depended, at least in part, on entrepreneurship incentives embedded in the tax system. Future studies exploring these issues should try to apply econometric but also qualitative approaches, e.g., a processtracing approach, which would make it possible to formally examine the strength of evidence linking potential causes to consequences.
Second, other tax-related and broader institutional conditions, such as inheritance taxation  or credit market regulations (Henrekson and Jakobsson 2005), may also have affected entrepreneurial activity during the examined period. Taking such changes into account is an important issue for future research delving deeper into how the institutions of the tax system evolved to constrain or facilitate entrepreneurship.
Third, the King and Fullerton method is, admittedly, based on a simple framework incorporating only the most basic elements of the tax system and the costs involved in an incremental investment. In principle, a more detailed analysis of a hypothetical firm's tax behavior could be done based on a mix of different forms of (intangible, fixed and financial) assets, including more complex effects from accounting items from the balance sheet and profit and loss accounts. That said, the relative simplicity of the King and Fullerton framework is also an advantage, and the more detailed and specific the analysis is, the less general the results will be.