FAccT WP 14 2014

Do Corporate Tax Cuts Increase Investments?
∗
Laura Dobbins Martin Jacob†
Freie Universität Berlin WHU – Otto Beisheim School of Management
laura.dobbins@fu-berlin.de martin.jacob@whu.edu
April 20, 2016
Abstract
This paper studies the effect of corporate taxes on investment. Since firms with a foreign
parent have more cross-country profit shifting opportunities than domestically owned firms
do, their effective tax rate and, consequently, their tax-induced costs to investment are
lower. We therefore expect capital investment responses to a corporate tax cut to be
heterogeneous across firms. Using firm-level data on German corporations, we exploit the
2008 tax reform, which substantially cut corporate taxes as an exogenous policy shock and
expect domestically owned firms’ investments to be more responsive to the reform. We show
exactly this in a difference-in-differences setting. We find that the reduction in corporate
tax payments led to a one-to-one increase in the real investments of domestic firms. The
effect is stronger for domestic firms relying more on internal funds. Correspondingly, labor
investment increased more for domestic firms, ensuring a constant mix of input factors. In
addition, we show that domestic firms’ sales grew faster after the tax cut than the sales of
foreign-owned firms. Our results imply that corporate tax changes can increase corporate
investment but that domestic firms benefit more than foreign-owned firms from a tax cut
through higher investment responses resulting in greater sales growth.
Keywords: Corporate taxation, Investment
JEL Classification: G31, H24, H25
∗
We are grateful to an anonymous referee, Juan Manuel Garcia Lara (the editor), Annette Alstadsæter,
Jochen Hundsdoerfer, Maximilian Müller, Michael Overesch, Martin Ruf, Kelly Wentland, and participants
of the 2014 ATA Midyear Meeting in San Antonio (TX), USA, the 2014 Annual Congress of the European
Accounting Association in Tallinn, Estonia, and the 2014 Annual Meeting of the German Academic Associ-
ation for Business Research in Leipzig, Germany, for helpful comments and suggestions. We thank Antonio
de Vito and Maximilian Jensen for excellent research assistance. Laura Dobbins gratefully acknowledges
financial support by PricewaterhouseCoopers.
†
Corresponding author at: WHU – Otto Beisheim School of Management, Burgplatz 2, 56179 Vallendar,
Germany. Tel: +49 261 6509 350; fax: +49 261 6509 359; E-mail address: martin.jacob@whu.edu.
Electronic copy available at: http://ssrn.com/abstract=2362258

1 Introduction
The effect of taxes on corporate investment is a key motivator for governmental reform strategies
to boost domestic economy. For example, in April 2016, Barack Obama published an update to
the earlier President’s Framework for Business Tax Reform from 2012, which suggests to “lower
the corporate tax rate to 28 percent, putting the United States in line with major competitor
countries and encouraging greater investment in America.”1 In 2013, the Swedish government
reduced corporate taxes to 22%, since a “lower corporate tax rate will provide strong incentives
for the business sector to increase its investments.”2 The German tax reform of 2008 that, among
other reform elements, cut the corporate tax rate from about 39% to 29%, which we exploit in
this paper, was implemented to “[increase] the German tax attractiveness for investments.”3
This paper aims to test how corporate investment responds to such a large corporate tax cut.
Empirical evidence from firm-level data on the direct effect of the statutory corporate tax
rate on corporate investment is surprisingly rare (see, e.g., Ljungqvist and Smolyansky, 2014).4
Djankov, Ganser, McLiesh, Ramalho, and Shleifer (2010) analyze macro data from a cross
section of 85 countries in 2004. They document a negative association of effective corporate tax
rates with aggregate investment. However, they do not find robust evidence of an association
of statutory corporate tax rates and aggregate investments. Auerbach (1983) uses macro data
from the United States. Some studies (see, e.g., Summers, 1981; Feldstein, Dicks-Mireaux, and
Poterba, 1983; Cummins, Hassett, and Hubbard, 1996; Auerbach, 2002; Devereux, Griffith, and
Klemm, 2002) apply tax policy measures that combine tax rate and tax base elements using the
q-approach. Other studies (e.g., Jorgenson, 1963; Hall and Jorgenson, 1967; Chirinko, Fazzari,
and Meyer, 1999; Dwenger, 2014) implement the user cost of capital approach. One weakness
1
See page 4 of The President’s Framework for Business Tax Reform: An Update, A Joint Report by The White
House and the Department of the Treasury, available at https://www.treasury.gov/resource-center/tax-
policy/Documents/The-Presidents-Framework-for-Business-Tax-Reform-An-Update-04-04-2016.pdf (last
accessed April 19, 2016).
2
See pages 20 and 21 of the Budget Statement from the Budget Bill 2013 in Sweden, avail-
able at http://www.government.se/contentassets/24e79b514b5b4474aa3d6f5eadb738a4/from-the-budget-
bill-for-2013-budget-statement (last accessed April 19, 2016).
3
This quote is from the legislation draft of the tax reform act from March 27, 2007, available at
dip21.bundestag.de/dip21/btd/16/048/1604841.pdf (also, Brandstetter, 2014, last accessed April 19, 2016).
4
There are several papers on the effect of corporate tax rates on foreign direct investments (see, e.g., de Mooij
and Ederveen, 2003, for a literature review). This paper centers on all — domestic and foreign — invest-
ments. Hence, this paper is related to this literature but goes beyond studies on FDI flows since we
specifically examine corporate investments by domestic firms which are a key factor of the economy in many
countries.
Electronic copy available at: http://ssrn.com/abstract=2362258

of these studies is that tax rate and tax base elements such as depreciation allowances cannot
be separated and that these studies typically do not exploit exogenous variation around large
corporate tax rate cuts.

We argue that corporate taxes can have heterogeneous investment effects across firms in an
economy. Corporate taxes are costs to investment that differ across firms. For example, some
firms can take advantage of lower corporate tax rates abroad through profit shifting, and thereby
reduce tax-induced investment costs. The literature documents that tax avoidance and profit
shifting activities vary considerably across firms (see, e.g., Huizinga and Laeven, 2008; Dischinger
and Riedel, 2011; Dharmapala and Riedel, 2013) and, consequently, so do tax-induced investment
costs. We show empirically that investment responses to a reduced corporate tax burden differ
between domestic and foreign firms. This holds not only for capital investment, but also for labor
investment. These changes in input factors also lead to higher sales growth among domestic firms
relative to foreign firms. Our study builds on the results of the profit shifting literature and adds
more insight into how profit shifting affects firms’ investment decisions.
There are three main challenges to studying the effect of corporate taxes on investment.
First, one needs firm-level data on listed and unlisted firms, since the majority of firms in an
economy are typically unlisted.5 We use a large panel of over 36,000 listed and unlisted firms
from Germany over the period 2004–2011 with information on the location of the ultimate
shareholder. The second requirement relates to identifying tax rate variation that stems from
a large policy change. Our identification strategy exploits a large policy shock in Germany.
The most significant reform element was the corporate tax rate cut from 39% to 29%. Third,
since this tax cut applies to all German corporations, our identification of the tax effect is based
on a difference-in-differences setting exploiting variation in the tax sensitivity within Germany.

One caveat of this identification strategy is, however, that we cannot estimate elasticities for
domestic or foreign firms. We compare domestically owned corporations to corporations where
the ultimate owner is a foreign corporation and examine the difference in their responses to the
reduction in the tax burden.
The intuition behind our empirical identification strategy relates to differences in profit shift-
ing opportunities across firms. Consider two firms with similar economic activity but different
5
According to the OECD, over 95% of firms are small and medium sized enterprises and represent over half
of the employment (http://www.oecd.org/regional/leed/1918307.pdf, last accessed October 28, 2014).
2
ownership structures. In contrast to a firm with a foreign parent, a domestically owned corpo-
ration has no, or fewer, opportunities to shift income across borders. Even if domestic firms
have subsidiaries abroad, multinationals shift profits toward the parent company rather than in
the opposite direction (e.g., Dischinger, Knoll, and Riedel, 2014). Therefore, we argue that do-
mestically owned firms are less engaged in international profit shifting than firms with a foreign
parent.
Empirical evidence suggests that, due to cross-border profit shifting, firms with foreign oper-
ations have lower effective tax rates on corporate profits than firms without foreign operations.
For example, (Rego, 2003; Dyreng and Lindsey, 2009; Dyreng, Hanlon, and Maydew, 2010) doc-
ument profit shifting but they base their results on U.S. data. Despite institutional differences
between the United States and Germany, we transfer these findings to the German setting,
since there is comparable and ample empirical evidence of profit shifting for German and other
European firms (e.g., Weichenrieder, 2009; Dischinger and Riedel, 2011; Dischinger, Knoll, and
Riedel, 2014). Based on this research, we assume a nominal tax cut to be effectively smaller for
multinational firms than for domestic firms.
Thus, in our setting, we assume that foreign firms have an effective tax rate below the
statutory tax rate and below the tax rate of domestically owned firms. Since taxes impose
significant costs to investments, the tax costs of foreign firms are lower than those of domestic
firms. We argue that a cut in the statutory tax rate for all firms has heterogeneous effects across
firms. Assuming that the effective cut in corporate tax rates, and thus the cost reduction, is
larger for domestically owned firms than for foreign-owned firms, we expect their investments to
respond more than the investments of corporations with a foreign parent.6 In addition to reduced
costs of capital, we identify a second channel through which the reduced tax burden triggers
higher investments. Lower taxes increase after-tax cash flows, which is important for firms that
rely more heavily on internal financing, such as small firms (e.g., Carpenter and Petersen, 2002).
We hence additionally expect investment reactions to be stronger for domestic firms that are
sensitive to both channels, lower costs of capital and higher after-tax cash flows.
6
In contrast to Overesch (2009), who examines the effect of cross-country corporate tax rate differences on
German inbound investments, we analyze heterogeneous investment responses across firms with domestic
and foreign shareholders. The data employed by Overesch include only the inbound investments of foreign
firms, whereas we empirically study the effect of corporate tax rates on the allocation of investment across
all firms in Germany, namely domestic and foreign investment. This makes our paper distinct from the
literature on taxation and the location of foreign direct investment.
3
We first present graphical evidence on the difference in investments between domestically
and foreign-owned firms. We observe a parallel trend in investments before the 2008 tax reform.
In each pre-reform year, the difference in investments between domestically owned firms and cor-
porations with a foreign parent is insignificant. Put differently, domestically and foreign-owned
firms have similar investments before the tax reform. After the reform, domestically owned
firms invest significantly more than firms with a foreign parent. The difference in investment is
statistically significantly different from zero and persists over time. The effect is significant in
each year following the tax rate cut. Our results suggest that corporate investments responded
immediately to the tax cut.7
We then test our hypothesis in a matching difference-in-differences setting in which we com-

pare the investment responses of domestically owned and foreign-owned German corporations
around the 2008 reform. Our estimation includes firm fixed effects, year fixed effects, and firm-
level control variables. After profitability, debt, size, sales, wages, and loss firms are controlled
for, the difference-in-differences estimate is significant and positive. We also obtain similar effects
when using an alternative measure of investment that accounts for potential scaling differences.
That is, the 2008 tax reform increased the investment of firms whose ultimate owner is domestic
relative to the investment of foreign-owned firms. A corporate tax cut of 10 percentage points has
large investment effects on domestically owned firms. We find an increase of 5.9% in investment,
equivalent to an average increase in investments in fixed assets of e3.0m (or USD 4.2m) for each
domestic firm. To put this into perspective, the magnitude of this investment increase is similar
to the decrease in corporate tax payments around the tax reform. In other words, the reduction
in corporate tax payments leads to a one-to-one increase in real investments. However, this is
just an average effect that comprises both channels, reduced costs of capital and higher after-tax
cash flows. Overall, this shows that corporate taxes are a considerable cost to investments.
Our approach identifies tax effects through heterogeneous investment responses of domesti-
cally and foreign-owned firms. If the financial crisis affects the investment of foreign and domestic
firms differently and in the same direction as the corporate tax reduction, we would potentially
interpret crisis effects as tax effects. We therefore use several robustness tests to rule out the
7
In contrast, Dwenger (2014) focuses on the long-term effects of user costs on the capital stock. That model
includes a long-term specification of a firm’s demand for capital. Thus, it only allows an interpretation of
long-run effects while misestimating short-run effects.
4
potential impact of the financial crisis: First, we analyze other EU member states without a
tax reform as a counterfactual in a difference-in-difference-in-differences setting and show that
the financial crisis does not explain our findings. The increase in investment of domestic firms
is concentrated in Germany. Additionally, we confirm that heterogeneous investment responses
are driven by the tax cut when analyzing nine of the EU-15 member states separately.8 There
is a similar trend in the investments of domestic and foreign firms around the financial crisis
in eight of these nine countries. The exception, Italy, cut its corporate tax by 5.5 percentage
points in 2008. Finding a significant effect in Italy and Germany but in none of the other sample
countries indicates that heterogeneous investment responses are driven by large corporate tax
reductions and not by the financial crisis.

We next document that the effect of the tax reform differs as expected among domestic firms.
Since an increase in investments can result from two channels—a lower required rate of return
for profitable investments and higher after-tax cash flow due to lower tax payments—we assume
that firms that depend on internal financing, for example, small firms (Carpenter and Petersen,
2002), react more strongly among the group of domestic firms. These firms benefit from both
channels of reduced taxes triggering investment. Using a difference-in-difference-in-differences
approach, we find that both small and large firms increase investments compared to foreign firms
after the reform, but that the effect is stronger for those firms that rely on internal financing.
This allows us to relate the investment response more precisely to firm characteristics.
We also examine whether there is a concurrent increase in labor investments that would
ensure that firms maintain their mix of input factors. We use our baseline difference-in-differences
setting, but with labor expenses as the dependent variable. We find that, for domestic firms,
not only capital investments but also labor expenses increase compared to foreign firms after the
reform. Finally, we analyze the consequences of a change in input factors on the sales growth
of domestic versus foreign-owned firms. Assuming that investments are profitable and result in
higher outputs, we expect higher sales growth for the group of domestic firms that increased
investments. We again use our difference-in-differences approach and find that the change in
investment is accompanied by changes in sales growth. The sales growth of domestic firms is
2.75 percentage points higher than the sales growth of foreign firms after the tax cut. We obtain
8
We do not conduct the test for Denmark (data availability), Belgium (large tax reform in the pre-pseudo
reform period), and countries affected by the sovereign debt crisis (Greece, Spain, and Portugal).
5
similar results for labor expenses and sales growth in a triple difference setting using the sample
of European firms to address concerns that, for example, the crisis led to increased labor input
or growth in the sales of domestic firms. Overall, our results indicate that a reduced tax burden
increases capital and labor investments as well as the sales growth of domestically owned firms
relative to foreign-owned firms.
One main implication is that corporate taxes have large effects on domestic investment
decisions. We show that the tax burden reduction translates on average into a one-to-one increase
in real investments. Effects of a tax cut on investment costs are heterogeneous across domestic
and multinational firms and correspondingly their allocation of investments and, ultimately, sales
and revenues. In large economies with high tax rates, such as Germany, Japan, and the United
States, there are many firms without foreign operations, since the domestic market is sufficiently
large. These countries may benefit from an increase in domestic investments by cutting corporate
tax rates to foster investment and revenues of domestically active firms. Conversely, in countries
with many internationally active firms and a small domestic market, investment responses to
corporate tax changes may be lower than expected. Our results also imply that a high corporate
tax rate comes at the cost of lower investments by domestic firms. The ongoing “race to the
bottom” in corporate tax rates therefore affects not only profit shifting activities of multinational
firms (e.g., Clausing, 2003; Huizinga and Laeven, 2008; Dharmapala and Riedel, 2013), but also
decisions on domestic investment of both domestically owned and foreign-owned firms.

Our paper also contributes to the literature on the effectiveness of tax policy changes on
investment. Recent stimulus packages and large tax reforms in OECD countries include corporate
tax cuts, dividend tax cuts, and bonus depreciation allowances.9 Previous literature on the
effectiveness of such provisions shows mixed results, however. While there is empirical evidence
that payout taxes affect the allocation of investments across listed firms (Becker, Jacob, and
Jacob, 2013), ambiguous findings are obtained for unlisted firms (e.g., Yagan, 2015; Alstadsæter,
9
Over the past decades, many OECD countries have substantially reduced the corporate tax rate: for exam-
ple, the United States around the Tax Reform Act in 1986, Australia in 1987, Austria in 1988 and 2004,
Denmark in 1990, Germany in 2001, Italy in 1997, Japan around 1998, New Zealand in 1987, Norway and
Sweden in 1991, and the United Kingdom between 1983 and 1986. Large dividend tax cuts (more than 10
percentage points) were, for example, observed in the United States in 2003, in Belgium in 1995, in Japan
in 2004, in the Netherlands in 2001, in Spain in 1999, and in Sweden in 2006 for closely held, unlisted cor-
porations. In response to the recent financial crisis, many European Union countries (e.g., Austria, Finland,
France, Ireland, and the United Kingdom) implemented bonus depreciation schemes to increase corporate
investment.
6
Jacob, and Michaely, 2016). In addition, evidence on the impact of bonus depreciation on
corporate investment is mixed (e.g., House and Shapiro, 2008; Hulse and Livingstone, 2010).
Our results show that corporate tax rate changes have heterogeneous effects across domestic
and foreign firms and that investment and revenues by domestic firms respond strongly.
The remainder of the paper is organized as follows: Section 2 presents the institutional
background and a simple investment model. Section 3 presents the data and our empirical
estimation strategy. The empirical results and robustness tests are discussed in Section 4. Section
5 sets forth our conclusions.
2 Institutional Background and Hypothesis Development
2.1 2008 Tax Reform
The German Business Tax Reform Act of 2008 reduced the statutory corporate tax rate from
25% to 15%. In combination with the local business tax rate on corporations, which was also
changed by the reform, this resulted in a decrease of the corporate tax rate by 10 percentage
points, from about 39% until 2007 to 29% as of 2008. However, these provisions did not reduce
overall levels of profit shifting to low-tax countries within German multinationals (Brandstetter,
2014).
The reform also introduced several elements to broaden the corporate tax base, for example,
the interest barrier rule or stricter transfer pricing regulations. Given the considerable reduc-
tion of the nominal tax rate, the decrease in the tax burden was, even combined with other
provisions, mostly a consequence of the tax rate change. The interest barrier rule, which limits
the deductibility of interest payments from taxable profits, was established as an instrument
to constrain financing structures within multinational firms. This thin capitalization rule could
have potential investment effects if the rule were binding and if it thereby could reduce access
to debt to finance new investment.
However, the German interest barrier with its exceptions and escape clauses is relevant only
for very few firms (Blaufus and Lorenz, 2009; Buslei and Simmler, 2012).10 Further, the interest
10
Based on the database dafne (the same database as used for this paper), Blaufus and Lorenz (2009) find that
fewer than 200 firms are potentially affected by the German thin capitalization rule, Buslei and Simmler
(2012) identify between 76 and 564 firms to be potentially threatened.
7
barrier concerns both domestic and foreign firms, and therefore affects treatment and control
group in our difference-in-differences approach similarly. Moreover, in an untabulated test, we
exclude all companies with net interest payments over e3.0m that exceed 30% of the EBIT DA
in any sample year. We use these criteria as a rough approximation for firms that may be
threatened by the interest barrier. Our results remain robust and our main coefficient is very
similar. In fact, the vast majority (98%) of firms in our sample do not even exceed these criteria.
Hence, it is unlikely that the thin capitalization rule biases our results.
Another change within the 2008 reform was a slight increase in the dividend tax rate for
individual, non-incorporated shareholders. According to the partial income method, 60% of
dividend income—compared to 50% before the reform—is taxed at the shareholder’s personal
income tax rate, which ranges from 0% to 45%. This small increase in the dividend tax rate of
zero to 4.5 percentage points for non-corporate shareholders could have, apart from any level
effects, potential effects on the allocation of investments across firms (Becker, Jacob, and Jacob,
2013); however, evidence on this effect for unlisted corporations is mixed and uses much larger
dividend tax cuts (e.g., Yagan, 2015; Alstadsæter, Jacob, and Michaely, 2016).11
Eventually, due to the financial crisis, the German government allowed bonus depreciations
for investments in fixed assets acquired in 2009 and 2010. This provision could have influenced
the corporate investments of German firms in those two years (e.g., House and Shapiro, 2008, for
the United States). However, in Table A.2 of the Online Appendix, we show that the investment
effect is not limited to 2009 and 2010; thus, our results cannot be attributed to the temporary
introduction of bonus depreciation. The persisting investment effect also contradicts a pure crisis
effect, which would reverse for foreign firms in 2010 and 2011.
2.2 Hypothesis Development
We start by formulating a simple model that isolates the effect of a corporate tax rate reduction
on investment followed by implications for other dimensions. We argue that a corporate tax
cut has heterogeneous investment effects across firms. Empirical evidence (e.g., Rego, 2003;
Dyreng and Lindsey, 2009; Dyreng, Hanlon, and Maydew, 2010) shows that firms with foreign
11
Since higher payout taxes increase the costs of external equity, we test if investment effects are concentrated
in firms with high profitability, which would bias our results. Table A.1 of the Online Appendix shows that
growth in investment is not limited to firms with high earnings before interest and taxes (EBIT). Our results
are not driven by the increase in the dividend tax rate.
8
operations have lower effective tax rates on corporate profits, and thus lower costs to investments.
The degree to which firms have access to profit shifting across borders can therefore affect
responsiveness to corporate tax changes. To illustrate this effect, we first consider a firm without
the opportunity to shift profits. We then relax this assumption and allow profit shifting across
borders.
We assume a one-period investment that yields a pre-tax return r. The return is subject to
corporate taxation τ Corp . This is the corporate tax rate in the country where the firm is located.
The net of corporate tax return is distributed to the shareholder who is subject to dividend
taxation τ Div . We follow King (1977), Auerbach (1979), and Bradford (1981) and assume that
the investment is financed with internal funds.12 In sum, the net cash flow at t + 1 equals
(1 + r(1 − τ Corp ))(1 − τ Div ).
Alternatively, the firm could pay out the cash flow immediately. In this case, shareholders
pay dividend taxes at a rate of τ Div . The remaining cash flow is invested at an after-tax return
of i. Comparing these two investment alternatives, the firm will invest in the project if
1
r∗ = i · (1)
1 − τ Corp
Equation (1) shows that the required rate of return for a corporate investment, r∗ , is sensitive
to changes in the corporate tax:
∂r∗ 1
= i · >0 (2)
∂τ Corp (1 − τ Corp )2
Equation (2) implies that a corporate tax cut will reduce the required rate of return of an
investment (r∗ ). Thus, we expect corporate investment to increase following a corporate tax
cut. The underlying assumption of Equation (2) is that the firm has no access to multinational
profit shifting and is subject to the domestic marginal tax rate on corporate profits of τ Corp .
We next relax the assumption that a firm has no access to profit shifting. We again assume
a one-period investment that yields a pre-tax return r. The return is subject to an effective
Corp
corporate tax rate, τEf f . This effective tax rate depends on the percentage of profits, α, shifted
12
The implications of our simple model are similar when this assumption is relaxed. If we assume that the
investment is financed with new equity (e.g., Harberger, 1962, 1966; Feldstein, 1970), the relevant required
rates of return increase by 1−τ1Div in both cases. As the dividend tax also changes slightly around the reform,
we show in the robustness section that this rate change has no effect.
9
to a foreign country where profits are taxed at a rate τFCorp
or . The remaining part of the profits,
(1 − α), is taxed at the domestic corporate tax rate τDom

Corp
. We simplify the determination of the
percentage of profits α shifted abroad and assume that α is exogenous. In the Appendix, we
endogenize α and allow α(.) to be a function of τFCorp Corp
or and τDom . Since the implications of our
model are similar in this case, we use the simplified version of the model to illustrate the effects.
In sum, a firm with access to profit shifting invests in the project if
∗ 1
rShif ting = i · (3)
1 − τEf
Corp
f
with
f = α · τF or + (1 − α) · τDom
Corp Corp Corp
τEf
Corp
From Equation (3), we can derive the effect of a corporate tax change (τ Corp = τDom ) on the
required rate of return of a firm with the opportunity to shift income:
∗
∂rShif ting 1
= i · (1 − α) · >0 (4)
∂τ Corp
(1 − τ Corp )2
If the domestic tax rate is reduced, the effect on the required rate of return of a firm with
∗
profit shifting opportunities, rShif ting , is mitigated by the fraction of income shifted abroad.
Hence, as long as the firm does not shift any profits abroad, α equals zero and both firms with
and without access to profit shifting will respond the same way. However, empirical analysis
Corp
suggests that α > 0 (e.g., Weichenrieder, 2009, for Germany). In this case, τEf f is smaller than
τ Corp if the foreign tax is below the domestic tax rate. Consequently, firms with the opportunity
to shift income abroad are less responsive to changes in the corporate tax rate.
Equation 4 further implies that the effect of a change in the (domestic) corporate tax rate
decreases with a higher degree of profit shifting (α close to one). In the extreme case that all
profits are shifted abroad, firm investment will not respond to changes in the corporate tax rate
at all. Generally, a difference in the investment response to the tax cut depends on α. Given
similar economic activity, as long as domestically owned firms shift fewer profits abroad than
firms with a foreign parent, investment responses to a tax cut are greater for domestic firms than
10
for foreign firms. Profit shifting is biased towards shifting to the parent firm (Dischinger, Knoll,
and Riedel, 2014), and therefore more relevant for foreign-owned than domestically owned firms.
These empirical observations translate into a lower α for domestic than for foreign firms. We
thus formulate our main hypothesis as follows.
Hypothesis 1: Following a corporate tax cut, firms with limited profit shifting opportunities
(domestic firms) will increase investments more than firms with more profit shifting opportunities
(foreign firms).
The heterogeneous investment response of domestic and foreign firms can be explained by
two channels. First, as shown in Equation (4), the required rate of return decreases around
a reduction in the tax burden and, hence, firms find more profitable investments. We denote
this the cost of capital channel, through which the corporate tax burden reduction can affect
investment. Second, the reduction in corporate taxes increases after-tax cash flows. Since some
firms rely more heavily on internal financing, their investment is sensitive to the level of internal
cash flow (e.g., Fazzari, Hubbard, and Petersen, 1988; Faulkender and Petersen, 2012). This is
the cash flow channel, through which the corporate tax burden reduction can affect investment.
We expect the response to the tax reform to be stronger among firms that rely more on internal
cash, since they are responsive through both channels. For those firms that are most affected
by the tax cut (domestic firms), we thus formulate the following hypothesis.
will increase investments more if they rely heavily on internal cash flow to fund investments.
Building on Hypothesis 1, we can also derive predictions on other forms of investments. A

firm’s production function has two input factors, capital and labor. When capital investments
increase, as predicted in Hypothesis 1, firms may need to increase labor input as well. Otherwise,
the relation of input factors changes and the firm could move to a suboptimal mix of input factors.
Hence, firms are expected to increase labor input as well. This increase is again expected
to be more pronounced among firms with limited profit shifting opportunities, because they
11
experience the largest reduction in the required rate of return. Since we have only data on labor
expenditures,13 we formulate our hypothesis as follows.
Hypothesis 3: Following a corporate tax cut, firms with limited profit shifting opportu-
nities (domestic firms) will increase labor expenses more than firms with more profit shifting
opportunities (foreign firms).
The first three hypotheses focus on the heterogeneous effect of corporate taxation on the
capital or labor investments of domestic and foreign-owned firms. If the additional investments
are profitable, one would expect domestic firms to also experience more growth in sales and
revenues than foreign firms. To be more precise, we expect firms’ incremental investments to
translate into sales. While this appears to be intuitive at first glance, there are several reasons
why these investments may not necessarily increase sales. In firms with separation between
ownership and control, managers could use the increase in after-tax cash flows to invest in
perks (Jensen and Meckling, 1976).14 If, on average, managers invest in productive investments,
then we would expect an increase in capital and labor investment (Hypotheses 1 and 3) to also
translate into higher sales and revenues. We thus formulate our fourth hypothesis, as follows.
(domestic firms) will experience more growth in sales and revenues than firms with more profit
shifting opportunities (foreign firms).
3 Estimation Strategy and Data
3.1 Empirical Identification
We test our hypotheses in a difference-in-differences framework that compares the investments of

domestic and foreign-owned German firms around the 2008 tax reform. The resulting estimation
13
There is also data on the number of employees but this data item is not sufficiently reliable.
14
Even in firms that are managed by the owner, there can be unproductive investments, such as consumer
durables that create utility but no output. The result of Alstadsæter, Jacob, and Michaely (2016), that
cash-rich closely held private firms reduce investments following a dividend tax cut, indicates that even
closely held firms invest in unproductive investments.
12
equation for Hypothesis 1 is
Invi,t = α0 + β1 Domestic × Ref orm + β2 EBITi,t−1 + β3 Salesi,t−1 (5)
+ β4 Labori,t−1 + β5 Debti,t−1 + β6 Lossi,t + β7 Ln(T A)i,t−1 + αi + αt + ϵi,t
where the investment of firm i in year t (Invi,t ) is the dependent variable. To test Hypothesis 3
(Hypothesis 4) we use labor expenses over total assets (growth in sales) as dependent variables.
The independent variable of interest is the interaction between Domestic and Ref orm, where
Ref orm is a dummy variable equal to one for all the years after the tax reform (2008 to 2011).
We define a German corporation as domestic if the enterprise’s overall shareholder resides in

Germany. The overall shareholder is defined as having a direct or indirect interest in the German
affiliate of more than 50%. The country of the foreign overall shareholder is known; however,
this more detailed information is not needed for the analysis.
This difference-in-differences approach identifies the investment effects of the reform on do-
mestic firms vis-à-vis foreign firms. Thus, the interaction of Domestic and Ref orm captures
differences in the level of investment between domestic and foreign companies after the 2008 tax
reform. According to our hypothesis, corporate investment increases after the tax cut, but the
effect is larger for domestic firms than for foreign firms. Thus, the β1 coefficient is predicted to
be positive. Our approach captures a causal effect of the corporate tax cut in 2008 on corporate
investment in Germany. However, we are not able to identify the tax elasticity of investment
of domestic and foreign firms as we cannot measure the extent of income shifting of firms, and
as a consequence we do not have sufficient information on the effective tax rate of the ultimate
owner.
There could be concerns that, besides ownership structure, potential differences in economic
activities between the treatment group (domestic firms) and the control group (foreign firms) bias
our results. Clearly, multinational corporations and purely domestic firms differ in many aspects,
for example, in their asset and financing structure or their size. To account for these differences,
we use an exact one-on-one propensity score matching procedure without replacement before
estimating Equation (5). Each foreign firm is matched to a domestic firm according to the
natural logarithms of sales, wages, liabilities, and fixed and total assets of each pre-reform
13
sample year. We additionally match on the industry code.15 This approach has two advantages.
First, the reform does not affect assignment to the treatment or control group, since sorting and
matching are based on pre-reform characteristics. Second, firms cannot enter the treatment or
control group after the reform. This ensures that our results are not driven by new firms and
their investments after the 2008 tax reform. We thus obtain two groups with the same number
of firms prior to the reform that are comparable in firm size, asset structure, and leverage but
that differ in ownership structure. We use the matched sample in addition to the full sample
when estimating Equation (5). The matching strategy eliminates differences in observable firm
characteristics. Additionally, the difference-in-differences approach accounts for time-invariant
differences between the treatment and control groups. Consequently, we are confident that our
analysis with a large tax change effectively captures the investment effect and is not driven by
general differences between the compared corporations.
The regression model controls for several firm-variables that affect corporate investment de-
cisions following the prior literature (e.g., Cummins, Hassett, and Hubbard, 1996; Baker, Stein,
and Wurgler, 2003). We include the ratio of EBIT over the prior year’s total assets and expect
that more profitable firms invest more because of higher availability to fund investments inter-
nally (see, e.g., Fazzari, Hubbard, and Petersen, 1988; Lamont, 1997; Faulkender and Petersen,
2012).16 Likewise, we use sales over the prior year’s total assets to proxy for growth opportu-
nities (e.g., Abel and Blanchard, 1986) and, therefore, expect firms with higher sales to invest
more. We additionally include the ratio of labor costs to the prior year’s total assets as well as
the ratio of total debt to the prior year’s total assets to control for leverage (e.g., Lang, Ofek,
and Stulz, 1996; Ahn, Denis, and Denis, 2006). Further, we include a dummy for experiencing
losses (EBIT<0) and expect loss-incurring firms to invest less. Finally, the variable Ln(T A),
the natural logarithm of total assets, accounts for the size of the firm, since smaller firms have
15
The propensity score results from a probit model. We match on firm-level variables from 2005, 2006, and
2007 simultaneously. Specifically, we match foreign and domestic firms according to the nearest-neighbor
propensity score based on firm characteristics for all three years. We thus obtain pairs of nearest neighbors
for three years and not only one year.
16
Due to data limitations, we do not include cash as a control variable. Moreover, since we use a difference-in-
differences approach, including cash would only change our main coefficient to the extent that cash changes
disproportionally around 2008 for our treatment and control groups and that this disproportional change is
not correlated with the individual firm’s earnings or its debt financing. We are confident that the variables
EBITi,t−1 and Debti,t−1 capture the potential effect. Moreover, in an untabulated test, we estimated the
difference-in-differences coefficient for specifications with and without cash holdings and the change in cash
holdings. The results do not change if we include cash.
14
better investment opportunities (e.g., Carpenter and Petersen, 2002). We lag all firm-level con-
trol variables once to avoid an endogeneity bias. All variables are described in Table 1. The
model contains firm fixed effects to capture time-invariant firm-specific influences on levels of
investment. Firm-fixed effects also control for differences in investments across industries. We
include year fixed effects that account for the business cycle and other macroeconomic effects.17
Note that the inclusion of firm-fixed effects captures the main effect of Domestic, which is time-
invariant. Likewise, the main effect of Ref orm is captured by year-fixed effects. Therefore, we
cannot report Domestic and Ref orm in our main specification. We base our statistical inference
on robust standard errors clustered at the firm level.
The identification of a causal effect of corporate taxes on investment is based on the assump-
tion that the tax reform is the only event affecting relative investment of domestic and foreign
firms around 2008. However, the financial crisis could potentially have an impact on the differ-
ence between domestic and foreign firms. Compared to other European countries, Germany’s
economy has not been hit as hard by the crisis.18 However, the effect on investments in Germany
is ambiguous. If subsidiaries with a foreign parent had less investment opportunities than their
domestic counterparts due to a worse economic situation in the parent country, funds may flow
back to the parent and investments in Germany decrease. In contrast to this argument, Becker
and Riedel (2012) find that, in case of a slower economy in the parent’s country, investments
are rather taken out in the affiliate’s country in expectation of higher returns, which would act
against finding our effect. Importantly, our results are robust to the inclusion of yearly gross
domestic product (GDP) growth rates of the parent countries to account for their economic
19
circumstances. We also address the concern regarding a potential crisis effect by comparing
investment behavior of German firms to investment in nine of the EU-15 member states after
excluding Belgium due to a large tax reform before 2008 and countries that were affected by
the sovereign debt crisis. The remaining countries suffered from the financial crisis similarly
to Germany, but were not subject to a tax reform in 2008. Therefore, these countries should
17
Our results are robust to the inclusion of industry-year-fixed effects. The estimated coefficient of Domestic×
Ref orm is then 0.0610 and statistically significant (p < 0.01), and thus very close to our preferred baseline
estimate from Table 3, Column (4).
18
We control for potential exit effects and bankruptcies during the financial crisis by restricting the sample to
firms that survive at least seven of eight sample years. Table A.3 of the Online Appendix shows that our
results are robust to the restriction.
19
As presented in Table A.4 of the Online Appendix, the estimated coefficient of GDP growth is negative,
which is in line with Becker and Riedel (2012); however, it is non-significant.
15
allow the identification of a potential crisis-related investment effect. We compare investment
by domestic and foreign firms in Germany and other EU-15 member states by using a difference-
in-difference-in-differences approach, which rejects that the influence of the crisis is different on
domestic firms than on foreign firms. Respective tests are presented in Section 4.3.
3.2 Data Sample and Summary Statistics
We use firm-level panel data from Bureau van Dijk’s dafne database. The database contains
the data of German companies,20 including unconsolidated financial data, as well as information
about the company activity, ownership structure, and branch. Industries are divided into 21
categories according to the EU NACE Rev. 2 code. Companies offering financial or insurance
services are excluded. Our sample consists of data from 36,072 corporations over the period
2004–2011, a total of 93,856 firm–year observations. All the financial data we use are based on
German accounting rules.21
The data also include information on the location of the ultimate owner.22 We differentiate
between domestic and foreign companies according to the primary place of business of the
overall shareholder. We define the dummy variable Domestic as equal to one for companies
with a German overall shareholder and zero otherwise.23 The mean of Domestic equals 0.829
(see the summary statistics in Table 1). That is, the vast majority of enterprises in the sample
are domestic. In the matched sample, 51% of firm–year observations stem from domestic firms.
Table 2 provides information on the quality of our matching procedure. In the matched sample,
there are no differences between matched domestic and foreign firms with respect to observable
firm characteristics.
20
These cover limited liability companies with the German legal forms Aktiengesellschaft, Gesellschaft mit
beschränkter Haftung, and Kommanditgesellschaft auf Aktien (KGaA, GmbH & Co. KGaA, and AG & Co.
KGaA), and other corporations. Small corporations are not included in our sample due to the lack of filing
requirements.
21
We provide more detailed information on the data selection process in Table A.5 of the Online Appendix.
22
Note that this information is updated by dafne based on the most recent information available for the
corporation. The database does not track changes in the ultimate owner; therefore, this information is time
invariant. We use data available as of August 2013, when we first collected the data. However, the ownership
information of an individual firm could be older, depending on data availability.
23
We have basic information about the ultimate owner, but we cannot identify the exact legal status. We
assume profit shifting opportunities to be limited in the case of a foreign individual shareholder. This would
drive the β1 coefficient towards zero, since we would then expect similar reactions for both foreign and
domestic firms. In an untabulated test, we restrict the sample to industrial firms. Results remain significant
and of similar size. We are thus confident that this data limitation cannot drive any positive findings on
relative investments of domestically and foreign owned firms.
16
Table 1 also presents the summary statistics and variable definitions of independent vari-
ables.24 Extreme observations in the upper and lower percentiles of the variables are truncated.
Our investment variable Inv is defined as the difference in fixed assets and intangible assets from
t to t − 1 relative to the prior year’s fixed and intangible assets.25 The average growth of fixed
and intangible assets compared to the prior year’s amounts to 9%. We further use information
on EBIT (EBIT ), turnover (Sales), wages to employees (Labor ), and debt (Debt). These four
variables are scaled by the prior year’s total assets. On average, firms in the full sample (matched
sample) have an EBIT-to-assets ratio of 9.4% (9.6%), a turnover-to-assets ratio of 242% (223%),
a wages-to-assets ratio of 64% (45%), and a debt-to-assets ratio of 61% (52%). We further in-
clude the natural logarithm of total assets (average 8.4 and 10.2, respectively) as a measure of
firm size, and a dummy variable for loss firms (average 12.3% and 12.4%, respectively).
[Insert Table 1 about here]
4 Empirical Results
4.1 Graphical Evidence
The simplest way of testing our hypothesis is to track the investments of domestic and foreign-
owned firms over time. Figure 1 uses the matched sample of firms and plots the investment
(Inv) for each group over 2005–2011. We observe a parallel trend in the investments of domestic
and foreign firms prior to the reform. Both sets of firms invest about 5% to 10% of their
fixed and intangible assets each year. Following the 2008 tax reform, the investments of these
two groups diverge. The investments of domestically owned firms exceed those of firms with
a foreign parent each sample year. This effect is visible immediately after the reform, since
reduced tax prepayments increased cash available for investments already in the first year after
24
Tables A.6 and A.7 of the Online Appendix further present descriptive statistics for all our variables for the
pre-reform and post-reform years and for domestic and foreign firms, respectively.
25
Our data do not include capital expenditures. Therefore, we use the change in fixed and intangible assets.
Our preferred measure Inv is highly correlated with capital expenditures (using the same scaling variable)
for listed U.S. firms as well as for listed German firms. Using Compustat (Compustat Global) data over the
period 1977–2013 (1999–2014), we find a correlation coeffiecient between Inv and CapEx of 0.44 (0.45). Our
results also hold if we define investment only as change in fixed assets compared to prior year fixed assets.
The estimated coefficient of Domestic × Ref orm is then 0.0544 and statistically significant (p < 0.01), and
thus very close to our preferred baseline estimate from Table 3, Column (4). As further robustness test, we
account for depreciation and scaling effects (see Section 4.2).
17
the implementation. More importantly, lower taxes resulted in lower required rates of return
for investments. The general trend of both sub-groups reflects the economic downturn. Most
important to our study is, however, the relative investment of domestically owned firms vis-à-vis
foreign owned firms.
[Insert Figure 1 about here]
Figure 2 plots the corresponding difference in investment between these two groups (black
line). The gray lines indicate the upper and lower 95% confidence intervals. Prior to the reform,
the difference is below zero indicating that foreign-owned firms invest more. This difference is,
however, statistically not different from zero. After the reform, domestically owned firms invest
about 5% of their fixed and intangible assets more than foreign-owned firms. This difference
is significant at the 1% level every sample year. That is, domestically owned firms with fewer
opportunities for international profit shifting respond much more strongly to the 2008 tax reform
than firms with a foreign owner and access to profit shifting.
The advantage of the graphical illustration in Figure 2 is that the effect is independent
from the business cycle in Germany, since all firms are affected similarly. However, we also
consider concerns that the financial crisis may have different effects on the two groups of firms.
Figure 2 shows that the effect is persistent over time. Differences in post-reform investment are
statistically not different from each other over time but are different from zero in each post-
reform year. This indicates that the financial crisis (alone) cannot explain our result. If foreign
firms are hit to a larger extent by the crisis in 2008 and 2009 than domestic firms, we would
have seen a reversal in the difference in 2010 and 2011. As the effect is persistent, this is a first
indication that the effect is driven by the tax reform.
As the next step, we compare the German case to investment in seven EU-15 member coun-
tries without a tax reform in the observation period.26 These countries were hit by the financial
crisis and the global recession comparably to Germany but did not change their corporate tax
26
These countries comprise Austria, Finland, France, Ireland, Luxembourg, Netherlands, and the United
Kingdom. We exclude Sweden and Italy from this analysis because the corporate tax rates changed in these
two countries in 2009 and 2008, respectively.
18
rate. Figure 3 and Figure 4 replicate Figure 1 and Figure 2 using data on foreign and domestic
firms in these other EU states.
Figure 3 shows that investment of domestic and investment of foreign firms in the compared
countries pursue a similar trend before and after the pseudo-reform. According to Figure 4, the
differences in investment between the groups are close to zero in every sample year. In other
words, in EU countries without a tax cut, investment of foreign and domestic firms follows a
common trend over the entire sample period. This supports the assumption that the heteroge-
neous investment responses across firms in Germany are not a result of the financial crisis, but
a consequence of the reduced tax burden.
4.2 Difference-in-Differences Results on the Investment Effect
We next test Hypothesis 1 using the difference-in-differences approach.27 Table 3 presents the
coefficient estimates from testing Equation (5). For our baseline tests, we regress investments on
the interaction of Domestic and Ref orm. Columns (1) and (2) of Table 3 use the full sample.
In Columns (3) and (4), we use the matched sample of domestic and foreign firms. We present
the results with and without firm-level control variables. We include year fixed effects and firm
fixed effects in all four specifications.
The estimated coefficient of Domestic × Ref orm is positive and significant in all specifica-
tions. The coefficient estimates are very similar when excluding or including firm-level control
variables. This shows that firm-level characteristics do not affect our results. For the matched
sample, which accounts for differences in the economic activities of foreign and domestic firms,
27
Difference-in-differences estimations could suffer from serial correlation (e.g., Bertrand, Duflo, and Mul-
lainathan, 2004). We address this concern by using standard errors clustered at the firm level. To addi-
tionally address this comment, we collapse pre-reform and post-reform sample years. This results in one
pre-reform average for the dependent variable and each independent variable and one post-reform average for
each variable per firm. Table A.8 of the Online Appendix shows that the difference-in-differences estimate
is still significant and close to the baseline estimate.
19
the coefficient is 0.0592 and is statistically significant (p < 0.01). The economic magnitudes
are large: The increase of 5.9% in investments of domestic firms compared to foreign firms is
equivalent to an average increase in investments in fixed assets by e3.0m (or USD 4.2m) for
each domestic firm. Let us put this into perspective. Firms pay about 4% of their total assets as
corporate income tax before the reform, and the reform-induced tax burden reduction amounts
to about 1% of total assets (going from 39% to 29%, or a decrease by 25.7%). Relative to total
assets, the 5.9% investment increase translates to an increase by 1% of total assets (see also
Table 4 below). Firms thus appear to substitute one Euro of tax payments with a Euro of real
investments. In other words, corporate taxes are a considerable cost to investments and a reduc-
tion in corporate tax payments leads, on average, to a one-to-one increase in real investments in
domestic firms. In sum, the estimates indicate that a corporate tax cut of 10 percentage points
can have large investment effects on domestically owned firms.
The results for our firm-level control variables show that investments increase for firms that
are more profitable (EBIT ), that have higher turnover (Sales) or that are smaller (Ln(TA)).
More precisely, for the full sample (matched sample), a one standard deviation increase in EBIT
results in 4.0% (3.2%) more investment, a one standard deviation increase in Sales results in
8.0% (7.9%) more investment, and a one standard deviation decrease in Ln(TA) results in 63.1%
(35.3%) more investment. The latter can be explained by the better growth and investment op-
portunities of smaller firms.28 We measure investment relative to existing fixed assets. Therefore,
small and high-growth firms have higher investment rates. Investments decrease by about 1.9%
(3.5%) if the firm has negative income (Loss). We find no significant effect for leverage, and
only a weakly significant effect for labor costs.
We test the robustness of our results using an alternative definition of our dependent variable
that accounts for potential scaling effects.29 Instead of using fixed and intangible assets as
scaling variable, we scale changes in fixed and intangible assets by total assets. Table 4 presents
the coefficient estimates for the difference-in-differences estimator (Domestic × Ref orm) for
28
Typically, smaller firms have higher market-to-book ratios, which is a common proxy for growth oppor-
tunities. Using the Datastream sample of firms around the world, the correlation between Ln(TA) and
market-to-book ratios is -0.2940. For German firms, the correlation is -0.2709. This suggests that smaller
firms have higher market-to-book ratios and, thus, also higher growth opportunities.
29
We test the robustness of our firms to the inclusion of small firms and excluded firms below the median
of total assets distribution. The results are robust and similar to the baseline estimates. For the matched
sample, the coefficient of Domestic × Ref orm is 0.0429 and significant (t-stat = 2.38).
20
the full and matched samples. In line with our hypothesis, we find a positive and significant
coefficient estimate for β1 . For the full sample with controls, the coefficient estimate is 0.0074
(t-stat = 2.06). In the matched sample, β1 is 0.0098 and significant (t-stat = 2.57). Panel B uses
an investment measure where we additionally account for depreciation. We define investment as
the difference in fixed assets and intangible assets plus depreciation from t to t−1 relative to prior
year’s fixed and intangible assets. Using this alternative dependent variable does not change our
results. For the full sample with controls, the coefficient estimate is 0.0378 (t-stat = 2.15). In
the matched sample, β1 is 0.0555 and significant (t-stat = 2.77).
Our differentiation between domestic and foreign firms is based on the location of the ultimate
owner as profits are shifted toward the parent company rather than toward a foreign subsidiary
(e.g., Dischinger, Knoll, and Riedel, 2014). However, we cannot rule out that domestic firms
have foreign subsidiaries and, therefore, have access to international profit shifting to some
extent. This is due to limited data on the structure of the enterprise. In another robustness
test, we divide the group of domestic firms into companies with a German parent, but with other
European subsidiaries (GermanM ultinationals) and corporations with a German parent and no
subsidiaries in other European countries (DomesticOnly).If corporations use foreign subsidiaries
for profit shifting activities, purely domestic firms without foreign subsidiaries (DomesticOnly)
should react more strongly to the tax cut. Due to data availability, the location of the subsidiary
is based on information from 2014 only. Moreover, we have information only about European
subsidiaries and, unfortunately, no information about subsidiaries in other countries. Thus, we
decided not to use this test as our main specification. Instead of Domestic × Ref orm, we now
include two interaction terms for DomesticOnly and Ref orm, and for GermanM ultinationals
and Ref orm in our regression; i.e., we now have two treatment groups. Both treatment groups
together represent the group of domestic firms in our baseline specification. Foreign-owned
corporations are still the control group.
Table 5 presents results for the matched sample without control variables in Column (1) and
with control variables in Column (2). The estimated coefficient for DomesticOnly × Ref orm
is positive and significant (p < 0.01) in both specifications. The estimated coefficient for
21
GermanM ultinationals × Ref orm is not significant. While German corporations with sub-
sidiaries abroad in a European country do not invest more than the control group of foreign firms,
we find that the purely German firms increase their investments by more than 7% after the re-
form compared to foreign firms. Moreover, firms without European subsidiaries (DomesticOnly)
invest significantly more than firms with European subsidiaries (GermanM ultinationals) (t-
stat = 1.82 for Column (1) and t-stat = 1.88 for Column (2)). This is in line with our rationale
that firms with better opportunities for international profit shifting (in this specification, Ger-
man multinationals and foreign firms) react less strongly to the tax cut. The results suggest that
German firms engage similarly in profit shifting, whether they have a foreign ultimate owner or
a foreign subsidiary. The investment effect as obtained from the baseline regression consequently
results from those subsidiaries that do not have international profit shifting opportunities for
either an ultimate owner or a subsidiary abroad. We therefore believe that, while we cannot
exclude some profit shifting opportunities for the domestic firms in our original treatment group,
the potential misclassification works against finding an effect in our baseline specification. Rel-
ative to an estimated coefficient of 0.0592 in the baseline specification, the estimated coefficient
for purely domestic firms is 0.0730. The results in Table 5 suggest that the investment response
mainly stems from purely domestic firms.
4.3 Robustness to the Financial Crisis
One main concern about our setting is potential influence of the financial crisis in 2008 and
2009. If, compared to domestic firms, foreign firms invest less in Germany due to the economic
crisis, we may misinterpret crisis effects as tax responses. To test to what extent our results
are driven by the financial crisis, we employ our difference-in-differences setting around pseudo-
reforms in EU-15 member countries. From these countries we exclude Denmark (due to lack of
data availability), Belgium, Italy and Sweden (due to tax reforms in the observation period) and
countries affected by the sovereign debt crisis (Greece, Spain, and Portugal). For the remaining
seven countries, we obtain data from amadeus, a database similar to dafne expanded to all
European countries.
22
As the first step, we extend our main approach and estimate a triple difference model.
To be more precise, we examine differences in corporate investment before to after 2008 (first
difference), between domestic and foreign firms (second difference), and between Germany and
other EU countries without a tax rate change (third difference). If the financial crisis explains
our result, the difference-in-differences coefficient Domestic × Ref orm is positive. That is, the
financial crisis increases the difference between domestic and foreign firms. If, however, the
German corporate tax cut explains the investment responses documented above, Domestic ×
Ref orm is insignificant and the triple difference coefficient Domestic × Ref orm × Germany
becomes positive.
The regressions contain all three double interaction terms, but we list only the interaction
terms of interest. Table 6 presents the results for the other EU countries in Columns (1) and (2),
and results for the triple difference setting in Columns (3) and (4). The estimated coefficients for
Domestic × Ref orm are insignificant in Columns (1) and (2). That is, in countries that suffer
from the financial crisis but that did not change the statutory corporate tax rate, investments
of domestic relative to those of foreign-owned firms do not change around 2008. In the triple
difference setting, we again find insignificant Domestic × Ref orm coefficients. This shows that
the financial crisis does not have heterogeneous investment effects across domestic and foreign
firms. Moreover, the triple difference coefficient is positive and significant in both specifications.
The economic magnitude is comparable to the baseline results. Accordingly, German domestic
firms invest more after 2008 compared to foreign-owned firms in Germany as well as relative to
domestic vis-à- foreign-owned firms in other EU countries.

As another robustness test to the triple difference analysis, we follow the same approach as in
Table 3 and estimate the pseudo-reform effects separately for each EU sample country analyzed in
the previous test. If the financial crisis explains our findings, we would find a positive coefficient
for Domestic × P seudoRef orm when comparing domestic and foreign firms in all seven EU
sample countries observed. We additionally include Italy and Sweden where corporate taxes
were reduced in 2008 and 2009, respectively. Table 7 presents the coefficient estimates for the
nine European sample countries. With the exception of Italy, we observe insignificant coefficients
for Domestic × P seudoRef orm across samples. Two estimated coefficients are positive and
23
borderline significant (p < 0.1) and two other coefficients are significantly negative. From the
results presented in Table 7, we conclude that domestic and foreign firms in eight EU-15 member
states responded similarly to the crisis in terms of investment. This result is independent of size,
industry composition, and the institutional setting of the country.
The fact that the coefficient of Domestic × P seudoRef orm is positive and highly significant
in the case of Italy supports our hypothesis and our findings, since Italy reduced the statutory
corporate tax rate in 2008 by 5.5 percentage points.30 Therefore, our model implications also
hold in a different market during exactly the same global macroeconomic conditions. We are
confident that the documented effect for German corporations is driven by the reduced corporate
tax burden and not by the financial crisis.
4.4 Cash Flow versus Cost of Capital Channel
In a second step, we analyze heterogeneity in investment reactions within the group of domestic
firms (Hypothesis 2). We argue that there are two channels that are responsible for firms’
increased investments: First, a lower corporate tax burden reduces the required rate of return
of investments. Consequently, more investment options are profitable for firms (Cost of Capital
Channel ). This channel applies to all firms. A second reason for higher investment activity
is the higher after-tax cash flow (Cash Flow Channel ). This channel is most important for
corporations that rely more heavily on internal funds to invest. When looking at domestic firms
only, we expect firms that depend more on internal funds to react more strongly to the tax cut,
since they are affected through both channels, whereas the other firms are only affected by the
cost of capital channel. Prior literature has shown that the growth of small firms is limited by
the quantity of internal finance (e.g., Carpenter and Petersen, 2002). Hence, small firms are
more likely to be affected by the cash flow channel. Using the empirical observation that small
firms are more likely to rely on internal cash, we create the dummy variable Small for firms with
total assets in the lowest quartile of the sample. We use a difference-in-difference-in-differences
approach that analyzes if, in the group of domestically owned firms, small corporations increased
30
Note that Sweden reduced its corporate tax rate by 1.7 percentage points in 2009. The reform coefficient
partially picks up this tax rate cut; however, the results are insignificant. One possible explanation for this
finding is that the tax rate cut was too small to have a material influence on investment decisions.
24
investments comparably more than bigger corporations did. We additionally include all other
interactions so that we also compare our treated group (small and domestic) to all other groups,
for example, small and foreign owned firms. Table 8 presents the results for the matched sample
without control variables in Column (1) and with control variables in Column (2).
The estimated coefficients of the triple interaction term are positive and significant in both
specifications (p < 0.056 and p < 0.071, respectively). Accordingly, the investment effect is
higher for small firms than for larger firms. Moreover, the estimated coefficients of the interaction
term between Domestic and Ref orm are also positive and significant (p < 0.05), that is, larger
firms increase investments compared to foreign firms as well. Consequently, domestic firms
increased their investments after the reform whether they depend on internal funds or not. The
joint effects of both interaction terms are 0.1467 and 0.1421, respectively (p < 0.01). Hence, the
effect is stronger for firms that rely more on internal funds, indicating that they take advantage
of the lower tax burden through both channels. The test shows that the higher investment of
domestic firms results from two channels and that, in particular, firms relying on internal cash
flow benefit from the tax reduction.
4.5 Effect on Labor Input
We next test if labor input increased correspondingly with higher investments (Hypothesis 3).
We therefore use our main regression but replace Inv by the change in labor expenses relative
to the prior year. Table 9 presents the results for the full sample in Columns (1) and (2) and
for the matched sample in Columns (3) and (4).
The positive and significant estimated coefficient (p < 0.01) is in line with our expecta-
tion that those firms with higher increases in investments, that is, domestic firms with limited
profit shifting opportunities, also increase labor expenses more than firms with lower investment
responses, that is, foreign firms with more profit shifting opportunities.
In addition, we apply a difference-in-difference-in-differences analysis that compares the in-
crease in the labor expenses of German domestic firms relative to those of foreign firms with
those of domestic and foreign firms in the other EU countries (comparable to Table 6, but with
25
the increase in labor expenses as the dependent variable). The results in Columns (5) and (6)
of Table 9 show that the effect is only visible for German domestic firms and not for domestic
firms in the other EU countries, contradicting concerns that the increase in labor expenses is a
result of the financial crisis.
4.6 Effect of the 2008 Tax Reform on Sales and Revenues
Having established that domestic firms increased investments relative to foreign-owned firms, we
next examine output responses (Hypothesis 4). We test whether the changes in input factors—
investments—also lead to changes in output, namely sales and revenues. To examine this em-
pirically, we rerun our difference-in-differences model from Equation (5) and use sales growth
relative to the prior year as the dependent variable. Note that this does not contradict our main
model, where we use the lagged ratio of sales to total assets as the independent variable. We
now use the current growth in sales, which is conceptually different from the sales-to-assets ratio,
as the dependent variable. Columns (1) to (4) of Table 10 replicate Table 3 using the new de-
pendent variable. We find that, relative to foreign firms, domestic firms’ sales grow significantly
faster. After the reform, the sales growth of domestic firms is 2.75 percentage points higher
than the sales growth of foreign firms. Consequently, the tax reform does not only increase
the investments of domestic firms. Additional input translates into an increase in the sales of
domestic firms.
To address concerns that the differences in sales growth are just an artifact of the financial
and economic crisis around 2008, we repeat the triple difference analysis of Table 6. We compare
the change in sales growth between domestic and foreign-owned firms in Germany to the change
in sales growth between domestic and foreign-owned firms in other European countries. Again,
if our result is driven by the crisis, we should find a significant change in sales growth between
domestic and foreign firms in other European countries and an insignificant coefficient for the
triple interaction (Domestic×Reform×Germany).
26
Columns (5) and (6) of Table 10 present the regression results from the triple difference
analysis. Consistent with the argument that the difference in sales growth is driven by the tax
cut and not by the financial crisis, we find that German domestic firms show higher growth in
sales than German foreign-owned firms (Domestic×Reform×Germany>0). Importantly, there is
no significant difference in sales growth between domestic and foreign firms in countries without
a tax cut. Taken together, our results suggest that a corporate tax cut can trigger investments
in capital and labor that ultimately translate into higher growth in sales.
5 Conclusion
This paper tests the effect of the 2008 corporate tax cut on firm investment in Germany. Our
simple investment model suggests that firms with only limited access to international profit
shifting respond more strongly to a corporate tax cut than firms with foreign operations and the
opportunity to shift income across borders. We test this expectation in a difference-in-differences

setting, which has the advantage of being independent of macroeconomic effects. We find that
domestically owned firms respond much more strongly to the tax cut than firms with an ultimate
owner in a foreign country. Using other EU countries as a counterfactual in a triple difference
setting, we can rule out that this effect is driven by the financial crisis. Thus, we conclude
that firms’ investment strategies are sensitive to corporate tax changes and that the corporate
tax rate affects the allocation of investments across domestically and foreign-owned firms. We
find stronger investment effects for firms that depend more on internal financing. These firms
benefit not only from reduced costs of capital, but also from higher after-tax cash flows to fund
investments. Corresponding to the increase in capital investments, we identify an increase in

the labor investments of domestic firms after the reform, which ensures that firms maintain
their relation of input factors. Finally, we empirically show that the increase in investments is
accompanied by the greater sales growth of domestic firms.
Our results have several policy implications. Large economies with high tax rates, such as
Germany, Japan, and the United States, have both a strong domestic market and large multina-
tional corporations. Countries can increase the investments of domestically active corporations
by reducing corporate tax rates and thus investment costs. Conversely, the reactions of firms
with foreign operations may be weaker than expected, since they are less responsive to changes
27
in the local statutory tax rate. These heterogeneous investment responses are also relevant for
countries with many internationally active firms and a small domestic market. The effect of a
corporate tax cut may be weaker than expected in these countries.

The results also point toward differences in the tax sensitivity of firm policies across cor-
porations, which may affect not only investment strategies, but also financing decisions (e.g.,
MacKie-Mason, 1990; Graham, 1996a,b). Differences in tax sensitivity are closely related to a
firm’s effective tax rate, which decreases, for example, for companies with foreign operations
(e.g., Rego, 2003; Dyreng and Lindsey, 2009; Dyreng, Hanlon, and Maydew, 2010). Testing the
tax responses of domestic versus foreign-owned firms on other business decisions is therefore an
interesting avenue for future research.

Finally, the ongoing race to the bottom of corporate tax rates affects not only profit shifting
activities of multinational firms (e.g., Clausing, 2003; Huizinga and Laeven, 2008; Dharmapala
and Riedel, 2013), but also the allocation of investments and market share across domestically
and foreign-owned corporations. Given that the governments of European Union member coun-
tries have recently cut corporate tax rates to reduce profit shifting (e.g., Sweden cut the corporate
tax rate from 26.3% to 22% in 2013), the effect of these reforms on corporate investment is a
highly relevant empirical question. Our results suggest that firms without access to international
income shifting will increase their domestic investment activity and revenues of domestic firms
following the recent wave of tax rate reductions in the race to the bottom.
28
References
Abel, Andrew B. and Olivier J. Blanchard (1986): The Present Value of Profits and Cyclical
Movements in Investment, Econometrica, 54: 249–273.
Ahn, Seoungpil, David J. Denis, and Diane K. Denis (2006): Leverage and investment in diver-
sified firms, Journal of Financial Economics, 79: 317–337.
Alstadsæter, Annette, Martin Jacob, and Roni Michaely (2016): Do dividend taxes affect cor-
porate investment?, Journal of Public Economics, forthcoming: forthcoming.
Auerbach, Alan J. (1979): Wealth Maximization and the Cost of Capital, Quarterly Journal of
Economics, 93: 433–446.
Auerbach, Alan J. (1983): Corporate Taxation in the United States, Brookings Papers on Eco-
nomic Activity, 2: 451–513.
Auerbach, Alan J. (2002): Taxation and Corporate Financial Policy. In Alan J. Auerbach and
Martin Feldstein, editors, Handbook of Public Economics, volume 3, pages 1251–1292.
Baker, Malcolm P., Jeremy C. Stein, and Jeffrey A. Wurgler (2003): When Does the Market
Matter? Stock Prices and the Investment of Equity-Dependent Firms, Quarterly Journal of
Economics, 118: 969–1005.
Becker, Bo, Marcus Jacob, and Martin Jacob (2013): Payout taxes and the allocation of invest-
ment, Journal of Financial Economics, 107: 1–24.
Becker, Johannes and Nadine Riedel (2012): Cross-border tax effects on affiliate investment –
Evidence from European multinationals, European Economic Review, 56: 436–450.
Bertrand, Marianne, Esther Duflo, and Sendhil Mullainathan (2004): How Much Should We
Trust Differences-In-Differences Estimates?, Quarterly Journal of Economics, 119: 249–275.
Blaufus, Kay and Daniela Lorenz (2009): Wem droht die Zinsschranke? Eine empirische Unter-
suchung zur Identifikation der Einflussfaktoren, Zeitschrift für Betriebswirtschaft, 79: 503–523.
Bradford, David F. (1981): The incidence and allocation effects of a tax on corporate distribu-
tions, Journal of Public Economics, 15: 1–22.
Brandstetter, Laura (2014): Do Corporate Tax Cuts Reduce International Profit Shifting?, FU
Berlin School of Business & Economics Discussion Paper 2014/10.
Buslei, Hermann and Martin Simmler (2012): Impact of introducing an interest barrier: Evi-
dence from the German corporate tax reform 2008, DIW Discussion Paper 1215.
Carpenter, Robert E. and Bruce C. Petersen (2002): Is the Growth of Small Firms Constrained
by Internal Finance?, Review of Economics and Statistics, 84: 298–309.
29
Chirinko, Robert S., Steven M. Fazzari, and Andrew P. Meyer (1999): How responsive is busi-
ness capital formation to its user cost? An exploration with micro data, Journal of Public
Economics, 74: 53–80.
Clausing, Kimberly (2003): Tax-motivated transfer pricing and US intrafirm trade prices, Jour-
nal of Public Economics, 87: 2207–2223.
Cummins, Jason G., Kevin A. Hassett, and R. Glenn Hubbard (1996): Tax reforms and invest-
ment: A cross-country comparison, Journal of Public Economics, 62: 237–273.
de Mooij, Ruud A. and Sjef Ederveen (2003): Taxation and Foreign Direct Investment: A
Synthesis of Empirical Research, International Tax and Public Finance, 10: 673–693.
Devereux, Michael P., Rachel Griffith, and Alexander Klemm (2002): Corporate income tax
reform and international tax competition, Economic Policy, 17: 449–495.
Dharmapala, Dhammika and Nadine Riedel (2013): Earnings shocks and tax-motivated income-
shifting: Evidence from European multinationals, Journal of Public Economics, 97: 95–107.
Dischinger, Matthias, Bodo Knoll, and Nadine Riedel (2014): The role of headquarters in multi-
national profit shifting strategies, International Tax and Public Finance, 21: 248–271.
Dischinger, Matthias and Nadine Riedel (2011): Corporate taxes and the location of intangible
assets within multinational firms, Journal of Public Economics, 95: 691–707.
Djankov, Simeon, Tim Ganser, Caralee McLiesh, Rita Ramalho, and Andrei Shleifer (2010): The
Effect of Corporate Taxes on Investment and Entrepreneurship, American Economic Journal:
Macroeconomics, 2 (3): 31–64.
Dwenger, Nadja (2014): User Cost Elasticity of Capital Revisited, Economica, 81: 161–186.
Dyreng, Scott D., Michelle Hanlon, and Edward L. Maydew (2010): The Effects of Executives
on Corporate Tax Avoidance, The Accounting Review, 85: 1163–1189.
Dyreng, Scott D. and Bradley P. Lindsey (2009): Using Financial Accounting Data to Exam-
ine the Effect of Foreign Operations Located in Tax Havens and Other Countries on U.S.
Multinational Firms’ Tax Rates, Journal of Accounting Research, 47 (5): 1283–1316.
Faulkender, Michael and Mitchell A. Petersen (2012): Investment and Capital Constraints:
Repatriations Under the American Jobs Creation Act, Review of Financial Studies, 25:
3351–3388.
Fazzari, Steven M., R. Glenn Hubbard, and Bruce C. Petersen (1988): Financing constraints
and corporate investment, Brookings Papers on Economic Activity, 1: 141–206.
Feldstein, Martin, Louis Dicks-Mireaux, and James M. Poterba (1983): The effective tax rate
and the pretax rate of return, Journal of Public Economics, 21: 129–158.
30
Feldstein, Martin S. (1970): Corporate Taxation and Dividend Behaviour, Review of Economic
Studies, 37: 57–72.
Graham, John R. (1996a): Debt and the marginal tax rate, Journal of Financial Economics,
41: 41–73.
Graham, John R. (1996b): Proxies for the corporate marginal tax rate, Journal of Financial
Economics, 42: 187–221.
Hall, Robert E. and Dale W. Jorgenson (1967): Tax Policy and Investment Behavior, American
Economic Review, 57: 391–414.
Harberger, Arnold C. (1962): The Incidence of the Corporation Income Tax, Journal of Political
Economy, 70: 215–240.
Harberger, Arnold C. (1966): Efficiency Effects of Taxes on Income from Capital. In M. Krzyza-
niak, editor, Effects of corporate income tax: papers presented at the symposium on business
taxation, Wayne State University. Wayne State University Press, Detroit.
House, Christopher L. and Matthew D. Shapiro (2008): Temporary Investment Tax Incentives:
Theory with Evidence from Bonus Depreciation, American Economic Review, 98: 737–768.
Huizinga, Harry and Luc Laeven (2008): International profit shifting within multinationals: A
multi-country perspective, Journal of Public Economics, 92: 1164–1182.
Hulse, David S. and Jane R. Livingstone (2010): Incentive effects of bonus depreciation, Journal
of Accounting and Public Policy, 29: 578–603.
Jensen, Michael C. and William H. Meckling (1976): Theory of the firm: Managerial behavior,
agency costs and qwnership structure, Journal of Financial Economics, 3: 305–360.
Jorgenson, Dale W. (1963): Capital Theory and Investment Behavior, American Economic
Review, 53: 305–360.
King, Mervyn A. (1977): Public Policy and the Corporation. Chapman and Hall, London.
Lamont, Owen (1997): Cash Flow and Investment: Evidence from Internal Capital Markets,
Journal of Finance, 52: 83–109.
Lang, Larry, Eli Ofek, and René M. Stulz (1996): Leverage, investment, and firm growth, Journal
of Financial Economics, 40: 3–29.
Ljungqvist, Alexander and Michael Smolyansky (2014): To Cut or Not to Cut? On the Impact
of Corporate Taxes on Employment and Income, Working Paper.
MacKie-Mason, Jason (1990): Do Taxes Affect Corporate Financing Decisions?, Journal of

Finance, 45: 1471–1493.
31
Overesch, Michael (2009): The Effects of Multinationals’ Profit Shifting Activities on Real
Investments, National Tax Journal, 62: 5–23.
Rego, Sonja Olhoft (2003): Tax-Avoidance Activities of U.S. Multinational Corporations, Con-
temporary Accounting Research, 20 (4): 805–833.
Summers, Lawrence H. (1981): Taxation and Corporate Investment: A q-Theory Approach,

Brookings Papers on Economic Activity, 1: 67–127.
Weichenrieder, Alfons (2009): Profit shifting in the EU: evidence from Germany, International
Tax and Public Finance, 16: 281–297.
Yagan, Danny (2015): Capital Tax Reform and the Real Economy: The Effects of the 2003
Dividend Tax Cut, American Economic Review, 105: 3531–3563.
32
Appendix
We extend our main model and allow α(.) to be a function of τFCorp Corp
or and τDom . That is, the level
of profit shifting of a multinational firm depends on the corporate tax rates in the countries of
the subsidiaries. The implications we derived for the simplified model are still valid. Hence, a
firm with access to profit shifting invests in a project if
∗ 1
rShif ting = i · (A.1)
1 − τEf
Corp
f
with
f = α(.) · τF or + (1 − α(.)) · τDom

Corp Corp Corp
τEf
and
( )
α(.) = α τFCorp Corp
or , τDom
From Equation (A.1), we can derive the effect of a corporate tax change on the required rate
of return of a firm with the opportunity to shift income (foreign firm):
∗
∂rShif ting 1 − α′ (.)(α(.) − τFCorp
or )
Corp
=i· ( )2 (A.2)
∂τDom 1 − α(.) · τFCorp + (1 − α(.)) · τ Corp
or Dom
Showing that the effect of a corporate tax change on the required rate of return is still larger
for domestic firms than for foreign firms is straightforward:
1 1 − α′ (.)(α(.) − τFCorp
or )
i· ( )2 >i· ( )2 (A.3)
1− Corp
τDom 1 − α(.) · + (1 − α(.)) ·
τFCorp
or
Corp
τDom
| {z } | {z }
Effect on Domestic Firm Effect on Foreign Firm
The necessary condition is that α′ (.) ≥ 0 and α′ (.) ≤ 1. That is, the proportion of profits
shifted abroad is either unaffected by the domestic tax rate (α′ (.) = 0) or increases with the
domestic tax rate (α′ (.) > 0). If we assume that α(.) is a logistic function that takes on values
between zero and one (0 ≤ α(.) ≤ 1 and 0 ≤ α′ (.) ≤ 1), the effect of a tax cut on domestic firms
is larger than on foreign firms.
33
Figure 1: Average Investment: Domestic versus Foreign-Owned Firms
This figure plots the average investments of domestic firms (black line) and foreign-owned firms (dashed
line). Investment is defined as the percentage change in fixed and intangible assets from t − 1 to t. The
dashed vertical line separates the pre-reform from the post-reform years.
20
Investments (in % of Fixed Assets)
5 100 15
2005 2006 2007 2008 2009 2010 2011

Year
Domestic Firms Foreign Firms
34
Figure 2: Difference in the Investments of Domestic and Foreign-Owned Firms
This figure plots the difference in investments between domestic firms and foreign-owned firms (black line).
Investment is defined as the percentage change in fixed and intangible assets from t − 1 to t. The gray
lines are the upper and lower 95% confidence intervals of the difference. The dashed vertical line separates
the pre-reform from the post-reform years. The dashed horizontal lines indicate the average difference in
investments between domestic firms and foreign-owned firms before and after the reform.
Difference in Investment Domestic versus Foreign (in %)
−5 0 5 10
2005 2006 2007 2008 2009 2010 2011

Year
Figure 3: Average Investment: Domestic versus Foreign-Owned Firms—

Counterfactual Analysis
This figure plots the average investments of domestic firms (black line) and foreign-owned firms
(dashed line) in EU-member states without a corporate tax cut. Investment is defined as the
percentage change in fixed and intangible assets from t − 1 to t. The dashed vertical line separates
the pre-reform from the post-reform years.
20
Investments (in % of Fixed Assets)
5 100 15
2005 2006 2007 2008 2009 2010 2011

Year
Domestic Firms Foreign Firms
35
Figure 4: Difference in the Investments of Domestic and Foreign-Owned Firms—
Counterfactual Analysis
This figure plots the difference in investments between domestic firms and foreign-owned firms (black
line) in EU-member states without a corporate tax cut. Investment is defined as the percentage
change in fixed and intangible assets from t − 1 to t. The gray lines are the upper and lower
95% confidence intervals of the difference. The dashed vertical line separates the pre-reform from
the post-reform years. The dashed horizontal lines indicate the average difference in investments
between domestic firms and foreign-owned firms before and after the reform.
Difference in Investment Domestic versus Foreign (in %)
−.05
.05
0.1
2005 2006 2007 2008 2009 2010 2011

Year
36
Table 1: Descriptive Statistics
This table presents the summary statistics and variable descriptions of our variables. The full sample (Panel
A) comprises 36,072 firms and 93,856 firm–year observations. The matched sample (Panel B) uses 4,000
firms and 18,140 observations If not otherwise indicated, the data source is the dafne database. Inv is our
investment measure. It is defined as the difference in fixed assets and intangible assets from t to t − 1 relative
to the prior year’s fixed and intangible assets. Domestic is a dummy variable equal to 1 if the ultimate
owner is located in Germany. EBIT is the ratio of EBIT relative to the prior year’s total assets. Sales is
turnover relative to the prior year’s total assets. Labor is the ratio of wages to the prior year’s total assets.
Debt is the ratio of long-term and short-term liabilities to the prior year’s total assets. Ln(T A) is the natural
logarithm of total assets. Loss is a dummy variable equal to 1 if income<0. GDP Growth is the percentage
point growth in GDP from t − 1 to t in the country where the ultimate owner is located. Source: World
Bank.
Variable Panel A: Full Sample

Mean Std Dev. p25 Median p75
Dependent variable
Inv 0.125 0.602 -0.110 -0.011 0.152
Independent variables
Domestic 0.829 0.376 1 1 1
EBIT 0.094 0.153 0.020 0.067 0.146
Sales 2.416 1.892 1.123 2.009 3.181
Labor 0.640 0.663 0.230 0.458 0.812
Debt 0.606 0.394 0.323 0.569 0.816
Ln(TA) 8.383 1.976 6.900 8.375 9.838
Loss 0.123 0.328 0 0 0
GDP Growth 1.517 3.234 1.083 3.029 3.700
TA (in e1,000) 34,200 264,602 1,044 4,527 19,328
FA(in e1,000) 17,540 199,047 135 833 5,739
Variable Panel B: Matched Sample
Mean Std Dev. p25 Median p75
Dependent variable
Inv 0.087 0.472 -0.089 -0.009 0.116
Domestic 0.510 0.500 0 1 1
EBIT 0.096 0.141 0.022 0.072 0.152
Sales 2.234 1.669 1.141 1.876 2.875
Labor 0.446 0.454 0.171 0.338 0.567
Debt 0.518 0.326 0.278 0.481 0.705
Ln(TA) 10.150 1.518 9.206 10.128 11.073
Loss 0.124 0.329 0 0 0
GDP Growth 1.417 3.062 0.992 2.285 3.314
TA (in e1,000) 102,850 523,063 10,303 26,049 66,786
FA(in e1,000) 51,414 411,313 1,119 5,542 21,763
37
Table 2: Matching Quality
This table presents descriptive statistics. We present averages of matching variables for the full sample and the matched sample
separately for domestic firms and for foreign firms. We also present the t-statistic ([t-stat]) of the significance of the difference
between domestic and foreign firms.
Full Sample Matched Sample

Variable Year Average [t-stat] Average [t-stat]
Domestic Foreign Difference Domestic Foreign Difference
ln(Sales) 2005 7.788 9.670 54.36 10.252 10.335 1.76
2006 7.901 9.780 54.34 10.364 10.440 1.66
2007 8.050 9.870 51.93 10.415 10.494 1.70
ln(Labor) 2005 6.558 7.923 46.13 8.524 8.568 0.97

2006 6.747 7.983 44.71 8.578 8.625 1.06
2007 6.966 8.094 41.65 8.616 8.667 1.16
ln(Debt) 2005 6.136 7.903 55.19 8.917 8.988 1.42

2006 6.022 7.789 62.99 8.949 9.031 1.62
2007 6.055 7.834 64.32 9.005 9.073 1.31
ln(TA) 2005 6.841 8.643 64.07 9.758 9.827 1.40

2006 6.745 8.543 73.71 9.829 9.902 1.50
2007 6.797 8.597 75.13 9.897 9.969 1.46
ln(FA) 2005 5.285 7.032 49.01 7.888 7.912 0.32

2006 5.203 6.953 56.88 7.921 7.942 0.28
2007 5.253 7.020 58.67 7.977 7.988 0.15
38
Table 3: Investment and the 2008 Tax Reform
This table presents the regression results on firms’ investment policies over 2005–2011. The dependent variable is defined as the
difference in fixed assets and intangible assets from t to t − 1 relative to the prior year’s fixed and intangible assets. The independent
variables are defined in Table 1. We report the regression results for the full sample and a matched sample. We include firm fixed
effects and year fixed effects in all specifications. We report robust standard errors clustered at the firm level in parentheses. ***,
**, and * denote a significant difference at the 1%, 5%, and 10% levels, respectively.

(1) (2) (3) (4)
Domestic×Reform 0.0356** 0.0364** 0.0584*** 0.0592***
(0.017) (0.017) (0.018) (0.018)
EBITt−1 0.2583*** 0.2243***
(0.042) (0.066)
Salest−1 0.0425*** 0.0471***
(0.007) (0.013)
Labort−1 -0.0447* -0.0687
(0.025) (0.051)
Debtt−1 0.0119 0.011
(0.021) (0.037)
Ln(TA)t−1 -0.3192*** -0.2327***
(0.021) (0.034)
Losst -0.0187* -0.0349**
(0.010) (0.017)
Firm FE Yes Yes Yes Yes
Year FE Yes Yes Yes Yes
Observations 93,856 93,856 18,140 18,140
Adj. R-squared 0.193 0.210 0.086 0.102
39
Table 4: Investment and the 2008 Tax Reform—Alternative Dependent Variable
This table replicates Table 3 but uses the difference in fixed assets and intangible assets from t to t−1 relative to the prior year’s total
assets (Panel A) and the difference in fixed assets and intangible assets plus depreciation from t to t − 1 relative to prior year’s fixed
and intangible assets (Panel B) as dependent variable. The independent variables are defined in Table 1. We report the regression
results for the full sample and a matched sample. We include firm fixed effects and year fixed effects in all specifications. We report
robust standard errors clustered at the firm level in parentheses. ***, **, and * denote a significant difference at the 1%, 5%, and
10% levels, respectively.
Panel A: Scaling by Total Assets

(1) (2) (3) (4)
Domestic×Reform 0.0070* 0.0074** 0.0094** 0.0098**
(0.004) (0.004) (0.004) (0.004)
Control Variables No Yes No Yes
Observations 93,856 93,856 18,140 18,140
Adj. R-squared 0.171 0.206 0.129 0.159
Panel B: Accounting for Depreciation
(1) (2) (3) (4)
Domestic×Reform 0.0363** 0.0378** 0.0535*** 0.0555***
(0.018) (0.018) (0.020) (0.020)
Observations 48,294 48,294 15,443 15,443
Adj. R-squared 0.218 0.240 0.156 0.174
Table 5: Investment and the 2008 Tax Reform—Purely Domestic versus German
Multinationals
This table replicates Table 3 but separates the treatment group into domestic firms without
subsidiaries (DomesticOnly × Ref orm) and German-owned firms with subsidiaries in the
EU (GermanM ultinationals). The independent variables are defined in Table 1. We report
the regression results for the matched sample only. We include firm fixed effects and year
fixed effects in both specifications. We report robust standard errors clustered at the firm
level in parentheses. ***, **, and * denote a significant difference at the 1%, 5%, and 10%
levels, respectively.
(1) (2)
DomesticOnly× Reform 0.0708*** 0.0730***
(0.024) (0.024)
GermanMultinationals× Reform -0.0033 -0.0040
(0.026) (0.026)
t-stat of Difference in Coefficients 1.82 1.88
Control Variables No Yes
Firm FE Yes Yes
Year FE Yes Yes
Observations 24,595 24,595
Adj. R-squared 0.038 0.053
40
Table 6: Investment and the 2008 Tax Reform—Triple Difference Analysis
variables are defined in Table 1. We report the regression results for the sample of non-EU countries (Column (1) and (2)) and a
sample including Germany ((Column (3) and (4))). We use the matched sample in all columns. We include firm fixed effects and
year fixed effects in all specifications. We report robust standard errors clustered at the firm level in parentheses. ***, **, and *
denote a significant difference at the 1%, 5%, and 10% levels, respectively.
EU-Member States Triple Difference Sample

(1) (2) (3) (4)
Domestic×Reform 0.0079 0.0120 0.0079 0.0120
(0.012) (0.012) (0.012) (0.012)
Domestic×Reform 0.0516** 0.0483**
× Germany (0.021) (0.021)
Matched Sample Yes Yes Yes Yes
Controls No Yes No Yes
Observations 93,754 93,754 111,894 111,894
Adj. R-squared 0.021 0.034 0.022 0.035
41
Table 7: Investment and Pseudo Reforms in Europe
This Table replicates Table 2 but uses data from nine different EU member states around pseudo tax reforms in 2008. Except for Italy and Sweden, no country changed its corporate tax rate. Italy
(Sweden) decreased corporate tax rates by 5.5 (1.7) percentage points in 2008 (2009). We report robust standard errors clustered at the firm level in parentheses. ***, **, and * denote a significant
difference at the 1%, 5%, and 10% levels, respectively.
Austria (Pseudo Reform) Finland (Pseudo Reform) France (Pseudo Reform)

Full Sample Matched Sample Full Sample Matched Sample Full Sample Matched Sample
Domestic 0.0496 0.0476 0.0544 0.0569 0.0781 0.1036* 0.0849 0.1019 0.0045 0.0171 0.0200 0.0236*
×Pseudo Reform (0.061) (0.061) (0.055) (0.057) (0.054) (0.054) (0.064) (0.063) (0.013) (0.013) (0.013) (0.013)
Controls No Yes No Yes No Yes No Yes No Yes No Yes
Year & Firm FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 8,244 8,244 2,040 2,040 6,283 6,283 3,110 3,110 78,048 78,048 49,824 49,824
Adj. R-squared 0.064 0.073 0.053 0.061 0.044 0.059 0.004 0.017 0.040 0.058 0.023 0.036
Ireland (Pseudo Reform) Italy (Reform in 2008) Luxembourg (Pseudo Reform)
42
Domestic 0.1249 0.0958 0.0502 0.0311 0.0573*** 0.1039*** 0.0746*** 0.0903*** -0.5544 -0.5669 -1.0736** -1.0835*
×Pseudo Reform (0.166) (0.154) (0.178) (0.188) (0.026) (0.028) (0.037) (0.037) (0.595) (0.649) (0.531) (0.602)
Observations 2,064 2,064 558 558 86,561 86,561 25,746 25,746 1,403 1,403 671 671
Adj. R-squared 0.094 0.107 -0.001 0.017 0.065 0.085 0.031 0.047 0.065 0.061 0.040 0.047
Netherlands (Pseudo Reform) Sweden (Reform in 2009) United Kingdom (Pseudo Reform)
Domestic -0.0684 -0.055 -0.0037 -0.0084 -0.0294 -0.0123 0.0666 0.067 0.0011 0.0082 -0.0033 -0.001
×Pseudo Reform (0.134) (0.128) (0.131) (0.127) (0.050) (0.050) (0.058) (0.057) (0.020) (0.020) (0.019) (0.019)
Observations 3,321 3,321 1,067 1,067 17,234 17,234 7,795 7,795 59,664 59,664 46,877 46,877
Adj. R-squared 0.089 0.094 0.035 0.044 0.006 0.019 0.003 0.012 0.038 0.051 0.010 0.023
Table 8: Investment and the 2008 Tax Reform—Cash Flow vs. Cost of Capital
Channel
This table replicates Table 3 but separately identifies the effect for small domestic firms that
rely more on internal financing in the treatment group (Domestic × Ref orm × Small). The
independent variables are defined in Table 1. We report the regression results for the matched
sample only. We include firm fixed effects and year fixed effects in both specifications. We
report robust standard errors clustered at the firm level in parentheses. ***, **, and * denote
a significant difference at the 1%, 5%, and 10% levels, respectively.
(1) (2)
Domestic× Reform 0.0355** 0.0373**
(0.018) (0.018)
Domestic× Reform× Small 0.1112* 0.1048*
(0.058) (0.058)
Joint Effect 0.1467*** 0.1421***
T-statistic 2.66 2.58
Control Variables No Yes
Firm FE Yes Yes
Year FE Yes Yes
Observations 18,140 18,140
Adj. R-squared 0.0865 0.103
43
Table 9: Labor expenses and the 2008 Tax Reform
This table presents the regression results on firms’ growth in labor expenses over 2005–2011. The dependent variable is defined as
growth in labor expenses from t − 1 to t relative to the prior year’s labor expenses. The independent variables are defined in Table 1
but exclude lagged labor expenses. We report the regression results for the full sample and a matched sample in Columns (1) & (2)
and (3) & (4), respectively. Columns (5) & (6) report regression results for the matched sample of EU countries including Germany.
We report robust standard errors clustered at the firm level in parentheses. ***, **, and * denote a significant difference at the 1%,
5%, and 10% levels, respectively
German Sample European Sample

(1) (2) (3) (4) (5) (6)
Domestic×Reform 0.0217*** 0.0216*** 0.0216*** 0.0222*** -0.0017 -0.0009
(0.005) (0.005) (0.005) (0.005) (0.003) (0.003)
Domestic×Reform 0.0122*** 0.0108**
× Germany (0.005) (0.005)
Controls No Yes No Yes No Yes
Matched Sample No No Yes Yes Yes Yes
Firm FE Yes Yes Yes Yes Yes Yes
Year FE Yes Yes Yes Yes Yes Yes
Observations 77,653 77,653 17,694 17,694 93,250 93,250
Adj. R-squared 0.294 0.322 0.148 0.175 0.124 0.154
44
Table 10: Sales Growth and the 2008 Tax Reform
This table presents the regression results on firms’ sales over 2005–2011. The dependent variable is defined as the difference between
sales in t and sales in t − 1 relative to the prior year’s sales. The independent variables are defined in Table 1 but exclude lagged sales.
We report the regression results for the full sample and a matched sample in Columns (1) & (2) and (3) & (4), respectively. Columns
(5) & (6) report regression results for the matched sample of EU countries including Germany. We report robust standard errors
clustered at the firm level in parentheses. ***, **, and * denote a significant difference at the 1%, 5%, and 10% levels, respectively
German Sample European Sample

(1) (2) (3) (4) (5) (6)
Domestic×Reform 0.0426*** 0.0445*** 0.0250*** 0.0275*** 0.0002 0.0008
(0.009) (0.009) (0.010) (0.009) (0.003) (0.0003)
Domestic×Reform 0.0163*** 0.0159***
× Germany (0.005) (0.005)
Controls No Yes No Yes No Yes
Matched Sample No No Yes Yes Yes Yes
Firm FE Yes Yes Yes Yes Yes Yes
Year FE Yes Yes Yes Yes Yes Yes
Observations 73,801 73,801 17,282 17,282 106,083 106,083
Adj. R-squared 0.134 0.217 0.068 0.154 0.104 0.165
45
Online Appendix—Not for publication
Table A.1: Effect on the Allocation of Investments

This table presents the regression results on firm investment policy over 2005–2011. We use domestic firms only. We include firms in
the top and bottom quintiles of the EBIT distribution as a measure of internal resources. The dependent variable is defined as the
difference in fixed assets and intangible assets from t to t − 1 relative to the prior year’s fixed and intangible assets. The dependent
variable of interest is the interaction between high-EBIT firms and the reform dummy (HighEBIT ×Ref orm). The control variables
are defined in Table 1. We report the regression results for the full sample and a matched sample. We include firm fixed effects and
year fixed effects in all specifications. We report robust standard errors clustered at the firm level in parentheses. ***, **, and *
denote a significant difference at the 1%, 5%, and 10% levels, respectively.

(1) (2) (3) (4)
High EBIT×Reform -0.0648** 0.0487 -0.0113 0.0698
(0.031) (0.034) (0.047) (0.053)
Observations 14,804 14,804 2,825 2,825
Adj. R-squared 0.160 0.184 0.096 0.114
46
Table A.2: Investment and the 2008 Tax Reform, by Year
This table replicates Table 2 but presents regressions results, which estimate the reform effect separately for each year. The dependent
variable is defined as the difference in fixed assets and intangible assets from t to t − 1 relative to the prior year’s fixed and intangible
assets. The independent variables are defined in Table 1. We report the regression results for the full sample and a matched sample.
We include firm fixed effects and year fixed effects in all specifications. We report robust standard errors clustered at the firm level
in parentheses. ***, **, and * denote a significant difference at the 1%, 5%, and 10% levels, respectively.

(1) (2) (3) (4)
Domestic×2008 0.0256 0.0261 0.0663*** 0.0676***
(0.021) (0.021) (0.026) (0.025)
Domestic×2009 0.0514** 0.0488** 0.0668*** 0.0650***
(0.020) (0.020) (0.023) (0.022)
Domestic×2010 0.0420** 0.0425** 0.0415* 0.0407*
(0.021) (0.020) (0.024) (0.024)
Domestic×2011 0.0250 0.0304 0.0537** 0.0582**
(0.020) (0.020) (0.024) (0.023)
Observations 93,856 93,856 18,140 18,140
Adj. R-squared 0.193 0.210 0.085 0.102
47
Table A.3: Investment, the 2008 Tax Reform, and Firm Survival
This table replicates Table 2 but restricts the sample to firms that survived at least seven sample years (Panel A) or all eight sample
years (Panel B). The dependent and independent variables are defined in Table 1. We report the regression results for the full sample
and a matched sample. We include firm fixed effects and year fixed effects in all specifications. We report robust standard errors
clustered at the firm level in parentheses. ***, **, and * denote a significant difference at the 1%, 5%, and 10% levels, respectively.
Panel A: Firms that Survived at Least Seven Sample Years

(1) (2) (3) (4)
Domestic×Reform 0.0312* 0.0331** 0.0558*** 0.0572***
(0.017) (0.017) (0.018) (0.018)
Observations 92,002 92,002 17,986 17,986
Adj. R-squared 0.191 0.209 0.080 0.097
Panel B: Firms that Survived All Eight Sample Years
(1) (2) (3) (4)
Domestic×Reform 0.0315* 0.0321* 0.0571*** 0.0563***
(0.017) (0.017) (0.018) (0.018)
Observations 82,592 82,592 16,790 16,790
Adj. R-squared 0.185 0.204 0.078 0.100
48
Table A.4: Investment and the 2008 Tax Reform, Controlling for GDP Growth
difference in fixed assets and intangible assets from t to t − 1 relative to the prior year’s fixed and intangible assets. We add a control
for growth in GDP in the ultimate owner’s country. The control variables are defined in Table 1. We report the regression results
for the full sample and a matched sample. We include firm fixed effects and year fixed effects in all specifications. We report robust
standard errors clustered at the firm level in parentheses. ***, **, and * denote a significant difference at the 1%, 5%, and 10%
levels, respectively.

(1) (2) (3) (4)
Domestic×Reform 0.0368** 0.0375** 0.0590*** 0.0596***
(0.017) (0.017) (0.018) (0.018)
GDP Growtht -0.0035 -0.0031 -0.0031 -0.0025
(0.004) (0.003) (0.004) (0.004)
Observations 93,657 93,657 18,045 18,045
Adj. R-squared 0.193 0.210 0.085 0.102
Table A.5: Data selection

This table describes the data selection in the database dafne, and the continuing selection process in
the statistics program. All financial data used is based on German GAAP and unconsolidated financial
statements. The analysis includes data between 2004 and 2011. Our dataset consists of 36,072 firms and
93,856 firm-year observations.
1 Solvent firms
2 Legal form: corporations
3 Industries: no insurance or financial services
4 Corporations with an ultimate owner and information on the parent
country
5 Truncating: first and last percentile of variables
6 Excluding all observations without variables required for the baseline
specification
49
Table A.6: Descriptive Statistics: Pre- vs. post-reform observations
This table presents the summary statistics separately for the pre-reform and post-reform periods.
The full sample (Panel A) comprises 36,072 firms and 93,856 firm–year observations. The matched
sample (Panel B) uses 4,000 firms and 18,140 observations If not otherwise indicated, the data source
is the dafne database. Inv is our investment measure. It is defined as the difference in fixed assets
and intangible assets from t to t − 1 relative to the prior year’s fixed and intangible assets. Domestic
is a dummy variable equal to 1 if the ultimate owner is located in Germany. EBIT is the ratio
of EBIT relative to the prior year’s total assets. Sales is turnover relative to the prior year’s total
assets. Labor is the ratio of wages to the prior year’s total assets. Debt is the ratio of long-term and
short-term liabilities to the prior year’s total assets. Ln(T A) is the natural logarithm of total assets.
Loss is a dummy variable equal to 1 if income<0.

Pre Post
Mean Std Dev. Mean Std Dev.
Dependent variable
Inv 0.145 0.634 0.119 0.591
Domestic 0.836 0.371 0.827 0.378
EBIT 0.099 0.146 0.092 0.156
Sales 2.461 1.922 2.400 1.881
Labor 0.640 0.643 0.640 0.670
Debt 0.597 0.370 0.609 0.402
Ln(TA) 8.410 1.973 8.374 1.977
Loss 0.091 0.288 0.133 0.340
GDP Growth 3.438 0.655 0.865 3.491
TA (in e1,000) 36,056 278,703 33,571 259,643
FA(in e1,000) 19,682 230,162 16,813 187,316
Pre Post
Dependent variable
Inv 0.098 0.475 0.081 0.470
Domestic 0.511 0.500 0.509 0.500
EBIT 0.101 0.141 0.093 0.141
Sales 2.324 1.737 2.189 1.632
Labor 0.465 0.470 0.437 0.445
Debt 0.548 0.338 0.503 0.319
Ln(TA) 10.063 1.556 10.193 1.496
Loss 0.102 0.303 0.134 0.341
GDP Growth 3.366 1.097 0.433 3.258
TA (in e1,000) 99,718 510,581 104,427 529,252
FA(in e1,000) 51,447 429,506 51,398 401,856
50
Table A.7: Descriptive Statistics: Domestic vs. foreign firms
This table presents the summary statistics separately for domestic and foreign firms. The full sample
comprises 36,072 firms and 93,856 firm–year observations. The matched sample uses 4,000 firms and
18,140 observations If not otherwise indicated, the data source is the dafne database. Inv is our
investment measure. It is defined as the difference in fixed assets and intangible assets from t to t − 1
relative to the prior year’s fixed and intangible assets. Domestic is a dummy variable equal to 1 if
the ultimate owner is located in Germany. EBIT is the ratio of EBIT relative to the prior year’s
total assets. Sales is turnover relative to the prior year’s total assets. Labor is the ratio of wages
to the prior year’s total assets. Debt is the ratio of long-term and short-term liabilities to the prior
year’s total assets. Ln(T A) is the natural logarithm of total assets. Loss is a dummy variable equal
to 1 if income<0.

Domestic Foreign
Dependent variable
Inv 0.134 0.616 0.081 0.526
EBIT 0.094 0.151 0.095 0.167
Sales 2.443 1.924 2.283 1.720
Labor 0.676 0.686 0.465 0.504
Debt 0.615 0.394 0.563 0.392
Ln(TA) 8.101 1.927 9.752 1.605
Loss 0.117 0.322 0.149 0.356
GDP Growth 1.605 3.283 1.085 2.946
TA (in e1,000) 25,372 242,132 77,134 350,956
FA(in e1,000) 14,682 197,765 31,438 204,605
Domestic Foreign
Dependent variable
Inv 0.099 0.460 0.075 0.483
EBIT 0.097 0.129 0.094 0.152
Sales 2.231 1.763 2.237 1.565
Labor 0.462 0.498 0.430 0.402
Debt 0.509 0.312 0.527 0.340
Ln(TA) 10.143 1.557 10.156 1.476
Loss 0.109 0.312 0.139 0.346
GDP Growth 1.585 3.182 1.240 2.919
TA (in e1,000) 103,532 598,667 102,139 430,562
FA(in e1,000) 59,527 517,943 42,973 256,822
51
Table A.8: Investment and the 2008 Tax Reform—Collapsed Model
variables are defined in Table 1. We report the regression results for the full sample and a matched sample. Following Bertrand,
Duflo, and Mullainathan (2004), all variables are included as the pre-reform and post-reform average for each firm to control for
serial correlation. We report robust standard errors clustered at the firm level in parentheses. ***, **, and * denote a significant
difference at the 1%, 5%, and 10% levels, respectively

(1) (2) (3) (4)
Domestic×Reform 0.0387*** 0.0197* 0.0680*** 0.0654***
(0.011) (0.011) (0.014) (0.014)
Domestic 0.0283*** 0.0151 -0.0131 -0.0115
(0.009) (0.009) (0.011) (0.011)
Reform -0.0486*** -0.0290*** -0.0443*** -0.0244**
(0.010) (0.010) (0.011) (0.011)
Mean(EBIT) 0.3645*** 0.3246***
(0.021) (0.038)
Mean(Sales) -0.0003 -0.0023
(0.001) (0.003)
Mean(Labor) 0.0122*** 0.0049
(0.002) (0.003)
Mean(Debt) 0.0295*** 0.0114
(0.004) (0.010)
Mean(Ln(TA)) 0.2387*** 0.1546***
(0.010) (0.018)
Mean(Loss) -0.0623*** -0.0274*
(0.010) (0.017)
Observations 54,976 54,976 7,451 7,451
Adj. R-squared 0.002 0.056 0.004 0.043
52

FAccT WP 14 2014

Uploaded by

Copyright:

Available Formats

FAccT WP 14 2014

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

FAccT WP 14 2014

Uploaded by

Copyright:

Available Formats

Do Corporate Tax Cuts Increase Investments?

April 20, 2016

Keywords: Corporate taxation, Investment

JEL Classification: G31, H24, H25

Electronic copy available at: http://ssrn.com/abstract=2362258

Electronic copy available at: http://ssrn.com/abstract=2362258

corporate tax rate cuts.

on a diﬀerence-in-diﬀerences setting exploiting variation in the tax sensitivity within Germany.

We then test our hypothesis in a matching diﬀerence-in-diﬀerences setting in which we com-

reductions and not by the financial crisis.

channels of reduced taxes triggering investment. Using a difference-in-difference-in-differences

decisions on domestic investment of both domestically owned and foreign-owned firms.

2 Institutional Background and Hypothesis Development

2.1 2008 Tax Reform

2.2 Hypothesis Development

to changes in the corporate tax:

(1 − α), is taxed at the domestic corporate tax rate τDom

required rate of return of a firm with the opportunity to shift income:

Building on Hypothesis 1, we can also derive predictions on other forms of investments. A

3 Estimation Strategy and Data

3.1 Empirical Identification

We test our hypotheses in a diﬀerence-in-diﬀerences framework that compares the investments of

Invi,t = α0 + β1 Domestic × Ref orm + β2 EBITi,t−1 + β3 Salesi,t−1 (5)

+ β4 Labori,t−1 + β5 Debti,t−1 + β6 Lossi,t + β7 Ln(T A)i,t−1 + αi + αt + ϵi,t

We define a German corporation as domestic if the enterprise’s overall shareholder resides in

general diﬀerences between the compared corporations.

3.2 Data Sample and Summary Statistics

German accounting rules.21

[Insert Table 1 about here]

[Insert Table 2 about here]

4.1 Graphical Evidence

[Insert Figure 1 about here]

than firms with a foreign owner and access to profit shifting.

[Insert Figure 2 about here]

[Insert Figure 3 about here]

[Insert Figure 4 about here]

a consequence of the reduced tax burden.

4.2 Difference-in-Differences Results on the Investment Effect

[Insert Table 3 about here]

can have large investment eﬀects on domestically owned firms.

[Insert Table 4 about here]

mainly stems from purely domestic firms.

[Insert Table 5 about here]

4.3 Robustness to the Financial Crisis

[Insert Table 6 about here]

domestic vis-à- foreign-owned firms in other EU countries.

tax burden and not by the financial crisis.

[Insert Table 7 about here]

4.4 Cash Flow versus Cost of Capital Channel

[Insert Table 8 about here]

flow benefit from the tax reduction.

4.5 Eﬀect on Labor Input

[Insert Table 9 about here]

4.6 Eﬀect of the 2008 Tax Reform on Sales and Revenues

[Insert Table 10 about here]

opportunity to shift income across borders. We test this expectation in a diﬀerence-in-diﬀerences

owner in a foreign country. Using other EU countries as a counterfactual in a triple diﬀerence

investments. Corresponding to the increase in capital investments, we identify an increase in

corporate tax cut may be weaker than expected in these countries.