Abstract
Tax reform proposals in the spirit of the “flat tax” model typically aim to reduce three parameters: the average tax burden, the progressivity of the tax schedule, and the complexity of the tax code. We explore the implications of changes in these parameters for entrepreneurial activity, measured by counts of firm births. The Swiss fiscal system offers sufficient intranational variation in tax codes to allow us to estimate such effects with considerable precision. We find that high average taxes and complicated tax codes depress firm birth rates, while tax progressivity per se promotes firm births. The latter result supports the existence of an insurance effect from progressive corporate income taxes for riskaverse entrepreneurs. However, implied elasticities with respect to the level and complexity of corporate taxes are an order of magnitude larger than elasticities with respect to the progressivity of tax schedules.
Similar content being viewed by others
Notes
See, e.g., Keen et al. (2008) for a general appraisal of recent flattax reforms.
By considering corporate taxes in isolation, we take a narrower view than the most radical flattax model, in which a single tax rate is applied across all tax bases and corporate income may be taxed only when paid out as dividends (see, e.g., Hall and Rabushka, 2007).
This intuitive assertion finds theoretical support in endogenous growth models, where entrepreneurs are primarily cast in the role of conduits between scientific research and marketoriented production (see, e.g., Michelacci, 2003). We can also invoke some relevant empirical evidence. Reynolds et al. (1995) and Audretsch and Fritsch (2002) have found that regions with higher firm formation rates enjoy higher growth, in the United States and Germany, respectively. These results were broadly confirmed by a number of countrylevel studies in the January 2008 special issue of Small Business Economics (see Fritsch, 2008). Employing indirect measures of entrepreneurship, Glaeser et al. (1992) have found industrylevel employment growth to be higher in US with belowaverage firm sizes; and Murphy et al. (1991) have reported positive growth effects of the share of engineering graduates in a large cross section of countries.
A corresponding analysis concerning personal income taxes has been provided by Varian (1980).
Cullen and Gordon (2006b) have put it as follows: “For any given tax treatment of losses, a progressive tax schedule on profits, holding expected taxes constant, should encourage risk taking. With progressive rather than proportional taxes, the owners get to keep a smaller fraction of large profits but a larger fraction of small profits. If expected tax payments are held fixed, this is a tradeoff that any riskaverse individual gains from making.”
Cowell (1975) used the term “compensation” for what we refer to as the “constant expected tax bill” condition.
If entrepreneurial ventures are externally financed and entrepreneurs are subject to moral hazard (i.e. they have an incentive to shirk if their stake in the success of the venture is low), then the riskreducing element implicit in progressive taxation may impede entrepreneurship (see e.g. Keuschnigg and Nielsen, 2004; and Hagen and Sannarnes, 2007). To the extent that the incidence of progressive taxation is felt by financiers rather than by entrepreneurs, however, the findings of the earlier literature on taxation and risk taking still apply.
Gentry and Hubbard (2000, 2005) have regressed the probability that an individual switches from employment to selfemployment on a set of variables including (a) the projected tax rate in case of unchanged employment status and (b) a measure of tax progressivity computed as the difference in tax rates between a “successful” scenario, where taxable income increases by x percent, and an “unsuccessful” scenario, where taxable income decreases by y percent. They did not, however, control for the expected (i.e. probability weighted) tax rate in case of a switch to selfemployment.
To cite Slemrod and Bakija (2004, p. 166), “a graduate taxrate structure does not by itself directly contribute any significant complexity to the taxpaying process. Once taxable income is computed, looking up tax liability in the tax tables is a trivial operation (…).”
Edmiston et al. (2004) explain the apparent positive effect of the length of tax codes by pointing out that more lines could imply greater legal precision—an aspect which might indeed be relevant in transition countries.
The main simplification of our approach compared to existing theory is that we constrain the range of choices to two options. This simplification allows us to posit a general (Bernoulli) utility function, which, unlike those adopted in prior studies, need not exhibit increasing absolute risk aversion (see also Feldstein, 1969). Cullen and Gordon (2006a) propose a similar model, taking utility as the log of income.
This framework also applies to cases where π _{ L }<0. In such cases, the corporate tax rate turns negative, implying a subsidy (e.g. through lossoffset or carryforward provisions). Since taxation in our model does not include a lumpsum tax part (payable independently of the realization of profits), we exclude π _{ L }=0. In our model, if π _{ L }=0, only π _{ H } would be taxed (at the same rate as the flat tax rate).
In what follows, brackets are used for mathematical operations, whereas parentheses are used for functions.
The certainty equivalent of EU(w ^{prog}) is not represented in Fig. 1. From Jensen’s inequality it follows that this point is located to the left of E(w ^{flat, prog}).
The interpretation of the estimated coefficients, can be quite different; with the conditional logit implying a “zerosum” reallocation of a fixed number of firms and the Poisson implying a pure “positivesum” taxinduced generation of new firms (Schmidheiny and Brülhart 2011).
Corporate taxation in Switzerland is based on legally binding statutory rates that depend solely on firms’ profitability and capital base. The definitions of these tax bases have been harmonized countrywide by a federal law that has been in force since 1993 and that foresees no firmspecific or sectorspecific regimes except for some clauses to avoid double taxation of holding companies. Some (mainly industrial) firms can be offered tax rebates for a maximum of ten years after setting up a new operation. Available evidence suggests that they affect less than 4 percent of new firms (Brülhart et al. 2012).
In principle, there could be unobserved exogenous shocks that have direct causal effects on both the progressivity of local tax schedules and on sectorlevel firm births, thus potentially biasing our estimated tax effects. While such a configuration does not strike us as likely, we cannot rule it out categorically, as our data do not allow us to instrument the tax variables. We seek to mitigate this potential problem by including a large set of controls.
Clustering being a variant of the Eicker–White “sandwich” variance estimator, it provides consistent standard error estimates even in the presence of overdispersion (see e.g. Cameron and Trivedi, 1998, p. 65f.). We also estimated our standard errors using oneway clustering by municipality and threeway clustering by municipality, sector and year, but did not find our inference results to be significantly affected.
In 8 of those 21 cantons, municipalities decide on a single multiplier that applies to both personal and corporate taxes. In the remaining 13 cantons, at least some municipalities apply separate multipliers to the two tax bases. See also Brülhart and Jametti (2006).
See Henneberger and Ziegler (2008).
The average population of our sample municipalities was 7,928 in 2001 and 7,243 in 2005. These municipalities were host to 85 (89) percent of all new firms in 2001 (2005). The data cover roughly the upper size quartile of Swiss municipalities. Tax data for smaller municipalities are not collected centrally.
A more sectorally disaggregated approach is not possible since our data on the distribution of profits and capital are available at the twodigit level only. We were forced to omit four sectors, for which no firm births were observed in our sample period: NACE 10 (coal mining), 12 (ore mining), 13 (uranium mining) and 23 (coke, refined petroleum and nuclear fuel). We also had to drop NACE 16 (tobacco) due to missing wage data. We therefore work with 46 sectors throughout.
The first, third and fifth sextiles for pretax rate of returns are 3, 12 and 37 percent (canton of Aargau) against 2, 9 and 32 percent (Switzerland). The quantiles for Aargau are based on firmlevel reported profit data, whereas the national quantiles are calculated using the national profit and capital distributions published by the Federal Tax Administration.
The index of the corporate income tax burden computed by the Federal Tax Administration for the year 2004 has a value of 97.4 for the canton of Aargau. The national average is 100, with values ranging from 57.3 (Schwyz) to 126.7 (Geneva). Aargau levies a minimum corporate tax of 500 Swiss francs (≈500 US dollars) on profits and capital together. Therefore, to calculate sector averages, we excluded all observations with a simple tax of 500 francs, even if they declared positive but very low profits. Furthermore, we considered observations with an implied pretax rate of return of more than 200 percent to be unreliable and excluded them.
The Swiss corporate tax system allows corporations to deduct actual tax payments from their pretax income. Therefore, our EATRs are defined as\(\ \frac{t^{\pi } ( \pi t^{K}K ) }{ ( 1+t^{\pi } ) \pi }\), where π denotes pretax profits, K is own capital, t ^{π} is the statutory corporate income tax rate and t ^{K} is the statutory capital tax rate.
Due to some small cell sizes, the Aargau data do not allow us to calculate sufficiently reliable sectorlevel distributions. We therefore prefer to rely on frequency distributions for Switzerland as a whole (available aggregated across sectors) for the profitability dispersion measure.
The weights applied are 0.375 for the cases of low and high capital and 0.25 for the mediancapital case, thus taking into account that the low and high cases refer to the upper end of the first and third quartile, respectively. The fact that two of our progressivity measures have negative minima (see Table 3) is explained by one canton (Aargau) applying a fixed minimum tax of CHF 500 on all incorporated firms, which implies regressive taxation for certain small firms with low profitability. Furthermore, the definition of EATRs implies that there is some small withincanton variation in progressivity even though municipalities apply a single multiplier to the cantonlevel tax schedule. Eliminating this variation by taking averages of the progressivity measures within each canton and year has no discernible impact on our results.
This measure is known as a “relative share adjustment” (see, e.g., Kesselman and Cheung 2004). It is a weighted average of a local index of tax progressivity, RSA _{ k }, where \(\mathit{RSA}_{k}=\frac{1\mathit{ATR}_{k}}{1\mathit{ATR}}1\). ATR _{ k } is the average tax rate for the kth income group, and ATR is the aggregate average tax rate. RSA _{ k } has an intuitive interpretation, since it can be used to calculate the gain or loss to a specific income group of switching to a fully proportional tax. For example, if RSA _{ k }=0.03, a ktype taxpayer would suffer an income loss of 3 percent if the existing system were replaced by a proportional tax. The global index of progressivity, RSA _{ G }, is then calculated as follows: \(\mathit{RSA}_{G}=\sum_{k=1}^{K}\phi_{k}\mathit{RSA}_{k}\), where \(\phi_{k}=\theta_{k} ( \theta_{k}+2\sum_{l=k+1}^{K}\theta_{l} ) \), and \(\theta_{k}=\frac{w_{k}}{\sum_{k=1}^{K}w_{k}}\ \)is posttax income share of the kth taxpayer (w _{ k } being posttax income of the kth taxpayer).
Word counts are based on the official compendium of cantonal tax laws Steuern der Schweiz. This compendium reproduces the content of all cantonal tax laws in a standardized format. It has the advantage of using harmonized terminology and thus allowing meaningful comparisons of word counts. The fact that three Swiss cantons are officially bilingual and have identical tax codes in both French and German allows us to quantify the “excess words” in tax codes due to the French language. In the canton of Berne, the French version of the tax code is 36 percent longer than the German one, and in the cantons of Fribourg and Valais, these differences correspond to 44 and 29 percent, respectively. Thus, the average “surplus word count” due to the French language is 37 percent. Therefore, we divide the word count for Latin cantons by 1.37 (the tax code for the Italianspeaking canton of Ticino being recorded in French in the compendium).
These results can be obtained from the authors upon request.
We thank an anonymous referee for suggesting us to consider also nonincorporated organizations.
Our estimates suggest a more than proportional reaction of firm births to changes in corporate tax levels. It would of course be erroneous to read into this a potential for revenueincreasing tax cuts, as our model does not capture responses of the entire tax base.
An alternative interpretation could be that new firms prefer more progressive tax schedules, given an expected tax bill, because they are credit constrained: the lower tax liability in case of a bad profit outcome may offer a greater gain in terms of access to external funding than the loss implied by a higher tax liability in case of a good outcome. See Keuschnigg and Ribi (2009) for a model of corporate income taxation with creditconstrained firms.
Firm births in our data could in principle also be the result of “income shifting” from bigger to smaller incorporated units, or through the incorporation of sole proprietorships. Data permitting, it would be interesting in future work to analyze these two margins separately from genuine startup ventures.
The published EATRs correspond to average cantonal, municipal and church tax rates for a representative household (married couple with two children) and for a range of reference incomes.
Annual municipal expenditures are only available for the 26 canton capitals and 16 other municipalities. However, the Swiss Federal Finance Administration publishes overall annual municipal spending for each canton. We compute annual municipal spending for the other municipalities by subtracting the expenditure of the (26+16) municipalities from overall municipal expenditures and then dividing it by the population of the remaining municipalities. Thereby, the remaining municipalities are attributed identical values of publicexp within each canton.
Wage data are compiled by the Swiss Federal Statistical Office for seven Swiss regions, five of which comprise several cantons (the cantons of Zurich and Ticino representing regions on their own), and for sectoral aggregates that correspond roughly to the NACE 1digit level. These data are available for the years 2002 and 2004. We linearly extrapolate wage for the remaining years.
We obtained these data from the consultancy firm Wüest & Partner.
References
Ahsan, S. M. (1974). Progression and risktaking. Oxford Economic Papers, 26, 318–328.
Audretsch, D. B., & Fritsch, M. (2002). Growth regimes over time and space. Regional Studies, 36(2), 113–124.
Becker, R., & Henderson, V. (2000). Effects of air quality regulations on polluting industries. Journal of Political Economy, 108, 379–421.
Brülhart, M., & Jametti, M. (2006). Vertical versus horizontal tax externalities: an empirical test. Journal of Public Economics, 90(10–11), 2027–2062.
Brülhart, M., Jametti, M., & Schmidheiny, K. (2012) Do agglomeration economies reduce the sensitivity of firm location to tax differentials? Economic Journal, forthcoming.
Cameron, A. C., & Trivedi, P. K. (1998). Regression analysis of count data. Cambridge: Cambridge University Press.
Cameron, A. C., Gelbach, J. B., & Miller, D. L. (2011). Robust inference with multiway clustering. Journal of Business & Economic Statistics, 29(2), 238–249.
Carlton, D. (1983). The location and employment choices of new firms: an econometric model with discrete and continuous endogenous variables. Review of Economics and Statistics, 65, 440–449.
Cowell, F. A. (1975). Some notes on progression and risktaking. Economica, 42, 313–318.
Cullen, J. B., & Gordon, R. (2006a). Tax reform and entrepreneurial activity. In J.M. Poterba (Ed.), Tax policy and the economy (Vol. 20, pp. 41–71). Cambridge: MIT Press.
Cullen, J. B., & Gordon, R. (2006b). How do taxes affect entrepreneurial activity? A comparison of U.S. and Swedish law (in Swedish). In P. Braunerhjelm & J. Wiklund (Eds.), Entreprenörskap och tillväxt (pp. 71–93). Stockholm: Swedish Foundation for Small Business Research.
Cullen, J. B., & Gordon, R. H. (2007). Taxes and entrepreneurial risktaking: theory and evidence for the U.S. Journal of Public Economics, 91, 1479–1505.
Da Rin, M., Di Giacomo, M., & Sembenelli, A. (2011). Entrepreneurship, firm entry, and the taxation of corporate income: Evidence from Europe. Journal of Public Economics, 95(9–10), 1048–1066.
de Mooij, R. A., & Ederveen, S. (2003). Taxation and foreign direct investment: a synthesis of empirical research. International Tax and Public Finance, 10, 673–693.
de Mooij, R. A., & Nicodème, G. (2008). Corporate tax policy, entrepreneurship and incorporation in the EU. International Tax and Public Finance, 15, 478–498.
Domar, E. D., & Musgrave, R. A. (1944). Proportional income taxation and risktaking. The Quarterly Journal of Economics, 58, 388–422.
Duranton, G., Gobillon, L., & Overman, H. G. (2011). Assessing the effects of local taxation using microgeographic data. Economic Journal, 121(555), 1017–1046.
Edmiston, K., Mudd, S., & Valev, N. (2004). Tax structures and FDI: the deterrent effects of complexity and uncertainty. Fiscal Studies, 24, 341–359.
Feld, L. P., & Kirchgässner, G. (2003). The impact of corporate and personal income taxes on the location of firms and on employment: some panel evidence for the Swiss cantons. Journal of Public Economics, 87, 129–155.
Feldstein, M. (1969). The effects of taxation on risk taking. Journal of Political Economy, 77, 755–764.
Fritsch, M. (2008). How does new business formation affect regional development? Introduction to the special issue. Small Business Economics, 30(1), 1–14.
Gentry, W. M., & Hubbard, G. (2000). Tax policy and entrepreneurial entry. American Economic Review, 90, 283–287.
Gentry, W. M., & Hubbard, G. (2005). “Success taxes”, entrepreneurial entry and innovation. In A. B. Jaffe, J. Lerner, & S. Stern (Eds.), Innovation policy and the economy (Vol. 5). Cambridge: NBER/MIT Press.
Glaeser, E. L., Kallal, H. D., Scheinkman, J. A., & Shleifer, A. (1992). Growth in cities. Journal of Political Economy, 100(6), 1126–1152.
Goolsbee, A. (2004). The impact of the corporate income tax: evidence from state organizational form data. Journal of Public Economics, 88, 2283–2299.
Gordon, R. H. (1985). Taxation of corporate capital income: tax revenues versus tax distortions. The Quarterly Journal of Economics, 100(1), 1–27.
Gordon, R. H., & Slemrod, J. (2000). Are “real” responses to taxes simply income shifting between corporate and personal tax bases. In J. Slemrod (Ed.), Does atlas shrug? The economic consequences of taxing the rich. Cambridge: Harvard University Press.
Guimaraes, P., Figuieredo, O., & Woodward, D. (2003). A tractable approach to the firm location decision problem. Review of Economics and Statistics, 85, 201–204.
Guimaraes, P., Figueiredo, O., & Woodward, D. (2004). Industrial location modeling: extending the random utility framework. Journal of Regional Science, 44, 1–20.
Hagen, K. P., & Sannarnes, J. G. (2007). Taxation of uncertain business profits, private risk markets and optimal allocation of risk. Journal of Public Economics, 91, 1507–1517.
Hall, R. E., & Rabushka, A. (2007). The flat tax (2nd ed.). Stanford: Hoover Institution, Stanford University.
Henneberger, F., & Ziegler, A. (2008). Eine Methode zur objektiven Bestimmung von Ober und Untergrenzen marktkonformer Aktionä rsgehälter. Der Schweizer Treuhänder, 2008(1–2), 69–79.
Hines, J. R. (2007). Corporate taxation and international competition. In A. J. Auerbach, J. R. Hines, & J. Slemrod (Eds.), Taxing corporate income in the 21st century. Cambridge: Cambridge University Press.
Keen, M., Kim, Y., & Varsano, R. (2008). The flat tax(es): principles and evidence. International Tax and Public Finance, 15, 712–751.
Kesselman, J. R., & Cheung, R. (2004). Tax incidence, progressivity, and inequality in Canada. Canadian Tax Journal, 52, 709–789.
Keuschnigg, C., & Nielsen, S. B. (2004). Progressive taxation, moral hazard, and entrepreneurship. Journal of Public Economic Theory, 6, 471–490.
Keuschnigg, C., & Ribi, E. (2009). Profit taxation and finance constraints (CEPR Discussion Paper#7433). London: Centre for Economic Policy Research.
Michelacci, C. (2003). Low returns in R&D due to the lack of entrepreneurial skills. Economic Journal, 113(484), 207–225.
Moulton, B. R. (1986). Random group effects and the precision of regression estimates. Journal of Econometrics, 32(3), 385–397.
Murphy, K. M., Shleifer, A., & Vishny, R. W. (1991). The allocation of talent: implications for growth. The Quarterly Journal of Economics, 106(2), 503–530.
Rathelot, R., & Sillard, P. (2008). The importance of local corporate taxes in business location decisions: evidence from French micro data. Economic Journal, 118, 499–514.
Reynolds, P. D., Miller, B., & Maki, W. R. (1995). Explaining regional variation in business births and deaths: US 1976–1988. Small Business Economics, 7(5), 389–407.
Schmidheiny, K., & Brülhart, M. (2011). On the equivalence of location choice models: conditional logit, nested logit and Poisson. Journal of Urban Economics, 69(2), 214–222.
Slemrod, J. (2005). The etiology of tax complexity: evidence from U.S. state income tax systems. Public Finance Review, 33, 279–299.
Slemrod, J., & Bakija, J. (2004). Taxing ourselves (3rd ed.). Cambridge: MIT Press.
Varian, H. R. (1980). Redistributive taxation as social insurance. Journal of Public Economics, 14(1), 49–68.
Waterson, M. (1985). On progressive taxation and risktaking. Oxford Economic Papers, 37, 510–519.
Acknowledgements
We thank the editor (Dhammika Dharmapala), two anonymous referees, Thiess Buettner, Mario Jametti and seminar participants at the University of Barcelona (IEB) for helpful comments; and Roland Aregger (Aargau tax authority), Andrea Grossi (Federal Statistical Office), Dieter Marmet (Wüest & Partner), Raphaël Parchet and Werner Tanner (State Secretariat for Economic Affairs) for the generous provision of data. Financial support from the EU’s Sixth Framework Programme (“MicroDyn”project) and from the Swiss National Science Foundation (grants 612065970, CRSI11_130648 and “NCCR Trade”) is gratefully acknowledged.
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix A: Proof of Proposition 2
From (1) and (2) it follows that
and
Then Proposition 1 and (4) and (5) imply
where
Appendix B: Control variables
The list of baseline explanatory tax variables (taxcontrols) is as follows.

Level of capital tax (captaxlevel): We calculate an industryspecific EATR on corporate capital for all municipalities and years.

Provisions to alleviate double taxation of dividends (dividendprovision): Dummy variable which is set equal to 1 if a canton has a reduced tax rate on dividend income and to 0 otherwise.

Level of the personal income tax (incometaxlevel): The Swiss federal tax administration publishes representative EATRs on personal income for all of the municipalities in our sample.^{Footnote 37} As we cannot know what municipality the owners of our sample firms reside in, we have considered two hypotheses for all personal taxes: (a) firm owners live in the municipality their firm is located in, or (b) owners live in the canton their firm is located in. Since the results do not differ significantly, we report results based on the second hypothesis. We thus compute incometaxlevel as the weighted average personal income tax burden, using the published cantonal sample mean of the EATR on low, median and high income households (corresponding to the first, third and fifth sextile of the national household income distribution).

Progressivity of the personal income tax (incometaxprogressivity): Based on the published cantonaverage EATR on low, median and high income, we define incometaxprogressivity1, incometaxprogressivity2 and incometaxprogressivity3 analogously to corptaxprogressivity13.

Level of the wealth tax (wealthtaxlevel): We compute this variable as the cantonalaverage EATRs for a person with taxable wealth of 300,000 Swiss francs (≈300,000 US dollars), which corresponds approximately to the mean wealth level among individuals with nonzero declared wealth over our sample period.

Inheritance tax (inheritancetax): This variable takes the value of 1 if a canton has an inheritance tax for direct descendants in a given year and 0 otherwise.
The list of baseline nontax explanatory variables (othercontrols) is as follows.

Public expenditure (publicexp): Firms not only pay taxes, they may also benefit from public spending. We construct this variable as the sum of municipal and cantonal percapita public spending, excluding social transfers and deflated with the consumer price index. The public spending items included in publicexp are public administration, security, education, culture and sports, roads, and public transport.^{Footnote 38}

Wage level (wage): We control for average monthly wages per sector and region, deflated by the consumer price index.^{Footnote 39}

Property prices (propertyprice): This variable is defined as the unweighted average of median municipality–yearlevel market prices per square meter of retail space, office space and industrial real estate, deflated by the consumer price index.^{Footnote 40}

Geography: To capture accessibility (and thus potentially agglomeration effects), we include three additional control variables: disthighway, the road distance from every municipality to the nearest highway access, distairport, the road distance to the nearest international airport, and distuniversity, the distance to the nearest university.

Culture (latin): We control for potential cultural and attitudinal differences by introducing the dummy variable latin that takes the value of 1 if the main language of a canton is French or Italian and 0 if it is German.

Unemployment (unemploymentrate): We control for the population share of registered unemployed workers by municipality and year.

Size of the municipality (munsize): We use the log of the average resident population per year and municipality as the exposure variable.
Rights and permissions
About this article
Cite this article
Bacher, H.U., Brülhart, M. Progressive taxes and firm births. Int Tax Public Finance 20, 129–168 (2013). https://doi.org/10.1007/s107970129218z
Published:
Issue Date:
DOI: https://doi.org/10.1007/s107970129218z