Political incentives for corporate taxation and competition

In this section, we derive several theoretical insights from a simple model. We model the economy as a Cournot model of quantity competition.Footnote ³ The government in the model decides how to regulate the economy, selecting the rates of taxation and antitrust. We assume that authoritarian leaders value consumer welfare less than democratic leaders, and antitrust enters the model by affecting the number of firms operating in the market. As is always the case in these models, when the number of firms increases, consumer welfare increases, but firm profits decrease. This reduces the corporate revenue available for taxation. Thus, increasing antitrust has diverging effects from a public-minded perspective: it reduces corporate tax revenue but increases consumer surplus. The model makes these assumptions transparent and helps us adjudicate competing effects of taxation and competition policy and their relative value for different governments.

The game is between the government and several firms. The government moves first and selects the tax rate, $t$ , and the level of antitrust. Antitrust affects the number of firms, $n$ , operating in the economy. Although other factors also affect whether a firm enters or exits the market, the model focuses on the government’s policy incentives. Stronger antitrust results in a larger $n$ , while weaker antitrust reduces $n$ .Footnote ⁴ The $n$ firms then engage in economic activity, which we model as Cournot competition over quantity. The firms simultaneously select the quantity, ${q_i}$ , that each firm $i$ produces to maximize its profits. The firms are assumed to be identical.

Each firm’s maximization problem is:Footnote ⁵

$$ma{x_{{q_i}}}\left\{ {{\pi _i} = \left( {1 - \mathop \sum \nolimits_{j = 1}^n {q_j} - c} \right){q_i}} \right\}$$

Which pins down the equilibrium quantities, prices, and per-firm profits:

$${q_i} = {{1 - c} \over {n + 1}} \hskip5pc{p_i} = {{1 + nc} \over {n + 1}} \hskip5pc {\pi _i} = {{{{(1 - c)}^2}} \over {{{(n + 1)}^2}}}$$

Profits in the full economy then are:

$${\rm{\Pi }} = \mathop \sum \nolimits_{i = 1}^n {\pi _i} = {{n{{(1 - c)}^2}} \over {{{(n + 1)}^2}}}$$

And the consumer surplus is:

$$CS = {1 \over 2}\left( {1 - p} \right)q = {{{n^2}{{(1 - c)}^2}} \over {2{{(n + 1)}^2}}}$$

The government selects $n$ and $t$ to maximize a combination of consumer surplus and tax revenue.

$${V_g}\left( {t,n} \right) = BCS + t{\rm{\Pi }} - k{t^2}$$

$k$ is a cost of taxation, which might deter future investment. This functional form ensures an interior solution for taxation. $B$ is the weight that the government attaches to consumer surplus, $CS$ . We assume that more democratic countries, and within authoritarian countries those with more widespread political participation, have a larger weight on consumer surplus.

The government’s first order conditions are:

(1) $${{\partial {V_g}\left( {t,n} \right)} \over {\partial t}} = \Pi - 2kt = 0$$

(2) $${{\partial {V_g}\left( {t,n} \right)} \over {\partial n}} = B{{\partial CS} \over {\partial n}} + t{{\partial \Pi } \over {\partial n}} = 0$$

In the taxation equation (1), the first term represents the marginal benefit of a tax increase, $t$ , in that it increases government revenue proportional to firm profits. The second term represents the marginal costs of the tax increase, in that it discourages future investments. In the antitrust equation (2), the first term again represents the marginal benefit of increasing the number of firms, $n$ , because consumer surplus is increasing in $n$ . The second term represents the marginal cost of increasing the number of firms, because profits are decreasing as the number of firms increases.Footnote ⁶

To see the full equation for antitrust, we can plug equation (1), $t = {\Pi \over {2k}}$ , into equation (2).Footnote ⁷

(3) $${{\partial {V_g}\left( {t,n} \right)} \over {\partial n}} = B+ {{{{(1 - c)}^2}\left( {1 - n} \right)} \over {{{(n + 1)}^2}}} = 0$$

The second order condition is:Footnote ⁸

(4) $${{{\partial ^2}{V_g}\left( {t,n} \right)} \over {\partial {n^2}}} = {{ - {{(1 - c)}^2}} \over {{{(n + 1)}^2}}} + {{2{{(1 - c)}^2}\left( {n - 1} \right)} \over {{{(n + 1)}^3}}}$$

To compute the effect of political institutions, $B$ , on antitrust policy, $n$ , we use the implicit function theorem.

Proof. For 1, use the implicit function theorem. The derivative of the government’s antitrust FOC is ${{{\partial ^2}{V_g}\left( {t,n} \right)} \over {\partial n\partial B}} = 1$ . The SOC is negative, so the overall effect is positive. For 2, see the effect of an increase in $n$ on total profits. Profits are ${{n{{(1 - c)}^2}} \over {{{(n + 1)}^2}}}$ . The derivative of profits with respect to $n$ is negative, ${{\left( {1 - n} \right){{(1 - c)}^2}} \over {{{(n + 1)}^3}}}$ , as long as $n \gt1$ . For 3, the optimal tax policy is $t = {\Pi \over {2k}}$ , which decreases when ${\rm{\Pi }}$ decreases.

In sum, we should observe a weakening of antitrust policy accompanied by an increase in corporate taxation as countries become more authoritarian. Put simply, authoritarian governments are more willing than democratic governments to offer weak antitrust, especially as it allows them to raise revenue through taxation.

Limitations. The model is highly stylized and simplified, which induces several limitations. First, there are markets where firms form cartels and are able to elevate prices. Adding an additional firm to the cartel may not have competitive effects. Absent cartels or coordination of market strategies however, the expectation is that adding more firms to a market will make the marketplace more competitive.

Second, we model autocracies and democracies as differing only in their weights on consumer welfare. There are plausibly other differences that could be relevant here. Democracies may also place higher value on liberal market competition, and authoritarian governments may prefer to use intervention in the economy to pick winners and support their regime. Democratic governments may also have effective institutional protections in place that prevent them from implementing biased policies, which would otherwise privilege individual firms with political connections. These alternatives are largely consistent with the logic of the formal model and would motivate similar insights. We retain the Cournot model in the text, as we want to display the microfoundations for competition, prices, and consumer effects.

In the Appendix, we formalize an alternative model that emphasizes the government’s preference for biased policies. We assume that increasing antitrust policy improves economic performance overall but harms politically connected firms, as they would typically benefit from biased policies, which are ruled out by unbiased interpretations of antitrust law (which yields competition for all). In this model, authoritarian leaders value these politically connected firms, relative to overall economic performance, more than democratic leaders do. The model in the Appendix emphasizes the trade-off between revenue from growing markets versus from taxing politically connected firms, which is an alternative mechanism that provides an explicit connection to the seminal literatures on revenue and political institutions (Olson, Reference Olson1965; Levi, Reference Levi1988) and on particularist policy benefits for politically connected firms in autocracy (Menaldo, Reference Menaldo2016; Betz and Pond, Reference Betz and Pond2023b).

Historical examples. The historical record provides many examples of governments offering weak competition policies or even monopoly rights in exchange for revenue, especially in undemocratic settings. In 1600, the British government granted the East India Company monopoly rights over trade between England and modern India and China. When the monopoly was set to expire, the Company secured an extension by providing revenue to the government, as well as lobbying and bribes: “To seal the deal and prevent future competitive challenges, the East India Company extended a 3.2 million pound loan to the British Treasury, which, in exchange, again granted the monopoly of trade” (Zingales, Reference Zingales2017, 117).

Similar deals were struck in the banking sector in Mexico. Under President and General Porfirio Díaz, bankers were offered high returns in exchange for agreeing to operate a bank and to provide loans to the government.Footnote ⁹ The bankers benefited from the large profits they enjoyed, politicians benefited from having a capital pool they could tap into when needed, and citizens paid the cost. In sum:

These examples are consistent with several assumptions and findings in the model. First, the absence of antitrust policy—and indeed the active creation of monopolies—increased profits for those with political connections, while consumers and the aggregate economy suffered. Second, these heightened profits benefited both the politically connected bankers and the politicians, who were able to finance their expenditures. Third, the policy exchange, competition for transfers, took place under authoritarian institutions.

While historically these sorts of agreements are common, they are hard to locate in the contemporary period. This is plausibly because politicians have an incentive to hide unpopular and inefficient policies. Citizens widely recognize the benefits of a competitive market, so politicians must obscure their support for non-competitive policies (Kono, Reference Kono2006). Accordingly, although we do not expect to observe politicians loudly proclaiming their support for un-competitive policies, their support should nevertheless be captured in contemporary market outcomes. In the next section, we examine the extent to which this pattern endures in the present era and the extent to which it is reflected beyond trade and financial sector policies. We assess whether politicians, especially those who come to power under undemocratic political institutions, create barriers to competition and in exchange receive increased revenues.

Cross-national Data

The theory anticipates that governments trade-off transfers for market competition. They are more willing to make this trade in authoritarian countries, as governments are thought to value consumer welfare and competitive markets less. To assess the theoretical predictions, we first draw on firm-level, cross-national, time-series data to capture competition, corporate tax revenue, and the level of representation. We expect to observe competitive markets where governments introduce and enforce pro-competition policies.

Our dependent variable is the extent of competition and antitrust in the country. We draw on two different measures. We first use the Herfindahl–Hirschman index (HHI) of market concentration, calculated for each industry. The index is calculated by summing up the value of the market share for each firm operating in each industry, at the 4-digit NACE level.Footnote ¹⁰ We log the HHI to reduce the effect of extreme values. We expect that industries will be allowed to grow more concentrated when firms, operating in the industry, provide substantial tax revenue—particularly in more authoritarian countries. The HHI and all firm-level variables detailed below are calculated using data from the Orbis Historical Database.Footnote ¹¹ Concentration varies at the industry level.

Our second dependent variable is logged firm size, as captured by the firm’s total assets. We expect that firms, which provide substantial revenue, will be allowed to grow larger, especially in more authoritarian countries. This variable varies at the firm-level.

Our independent variables of interest must capture corporate taxation and the extent of autocracy. To measure corporate taxation, we use the logged total taxes paid by businesses. The variable is larger for firms that are more important sources of government revenue. Because we expect the effect of corporate taxation on competition and antitrust to depend on political institutions, we also require a measure of authoritarianism in each country. We draw on the Polity2 data from Marshall, Jaggers and Gurr (Reference Marshall, Jaggers and Gurr2017). Since our theory is about less representative institutions, we use the inverted Polity2 score, the autocracy score. We interact the autocracy score with tax revenue, as we expect to observe a stronger, negative association between tax revenue and competition in more authoritarian countries.

We control for several, potentially confounding factors. First, we include the profit margin and the logged number of employees of the firms, as larger firms plausibly earn higher profits and pay more tax revenue. We also control for state capacity, using the Quality of Government indicator from the International Country Risk Guide (ICRG), as a stronger state has a greater ability to tax firms and to enforce regulations, and democracies frequently have more state capacity. Third, we control for domestic economic conditions: market size (logged GDP), wealth (GDP per capita), the annual growth rate, and unemployment (all economic variables are from the World Bank, World Development Indicators). We expect that a wealthier state has more resources to enforce regulations. We also expect that economic conditions affect tax revenue and antitrust, as governments may approve mergers or refrain from antitrust enforcement to stave off business failure and stimulate economic activity (Hylton and Lin, Reference Hylton and Lin2010).

We also include country fixed effects, which account for time-invariant, country specific effects, and industry fixed effects at the 4-digit level (NACE). Industry fixed effects account for differences in industry, like returns to scale, which could be related to firm size, market concentration, profits, and tax revenue. Moreover, if democracy affects industrial development (Nunn, Reference Nunn2007), industry could be related to political institutions. We also include year fixed effects, which provide for a flexible time trend and control for shared shocks to antitrust, political institutions, and taxation over time. We cluster standard errors by country. Once merged, our data includes over 22 million observations from 127 countries from 1992 to 2017. Our theory anticipates that governments may be more willing to accept weak competition if they are compensated through revenue. This trade-off should be more pronounced in countries with less representative institutions. To evaluate our theoretical expectations, we conduct linear regression and interact corporate taxation with autocracy.

Results

Table 1 reports the estimates of the relationship between taxation and antitrust conditional on regime type. Odd columns include firm-level controls, while even columns add the set of country-level controls. In columns (1) and (2), the dependent variable is log industrial concentration. In (3) and (4), the dependent variable is log firm size (total assets).

Table 1. Corporate taxation and competition policy enforcement

Robust standard errors in parentheses, clustered by country: * $p \lt 0.05$ , ** $p\lt 0.01$ , *** $p \lt 0.001$ . The level of analysis is the firm-year. In columns (1) and (2), the dependent variable is log industrial concentration at the 4-digit level. In (3) and (4), it is log firm size (total assets).

From columns (1) to (4), the taxation measure and its interaction with the autocracy variable are statistically significant and positive. The results suggest that an increase in corporate tax revenue is associated with an increase in industrial concentration and an increase in firm size and the effects increase in countries with more authoritarian political institutions. Drawing on column (2) and (4), we illustrate the marginal effects of corporate tax revenue across the autocracy scores in Figure 1. Consistent with our expectations, the marginal effects on industrial concentration and firm size increase in more authoritarian countries.

Figure 1. Marginal effects of corporate tax revenue from Table 1.

In democratic countries where the autocracy score is smaller than -8.5, an increase in corporate tax revenue is associated with a decrease in industrial concentration. This effect reverses as political institutions become more autocratic, and corporate tax revenue is associated with more concentrated industries. In countries with the autocracy score larger than 0, an increase in corporate tax revenue significantly correlates with an increase in industrial concentration. The results in more authoritarian countries accord with our theoretical expectations, but our theory did not predict the negative association between tax revenue and concentration in democratic countries. This negative association could be explained by policy motives in democracy: In democracy, the political opposition has an incentive to identify unpopular policies (like policy privileges for large firms), and democracies may also be better at preventing biased policies and extracting revenue. The result is theoretically and empirically related to state capacity, as democratic countries may be more capable of extracting revenue and limiting market concentration. Indeed the size of the coefficient on tax revenue decreases by over 40 percent when we add the control for state capacity to the model.

The results for firm size accord more closely with our theoretical expectations. Across countries with democratic and authoritarian political institutions, firms that provide more tax revenue are larger on-average, and the association between revenue and size becomes larger still in more authoritarian countries.

Within-country analysis from Turkey

We now turn to a within-country analysis that allows us to explore variation in antitrust and competition, as it relates to political institutions. Looking at competition within a single country holds many potentially confounding factors constant. This is important, as enforcement can vary dramatically across countries, even among those with similar legal institutions (Arslan, Reference Arslan2021).

We select Turkey as our country of interest for various reasons. First, Turkey has substantial variation in its political institutions in recent years. Its authoritarianism score (the reversed Polity2 score) decreased from negative seven to negative nine in 2011, indicating movement towards greater democratization. Then, Turkey experienced a dramatic increase in authoritarianism moving from negative nine to negative three in 2014 and then moving further from negative three to positive four in 2016, as Turkey’s President Recep Tayyip Erdogan suppressed dissent. In terms of the binary concept, these authoritarianism scores correspond to movement from democracy to autocracy in 2014. Figure 2 displays Turkey’s authoritarianism score over time.

Figure 2. Authoritarianism score in Turkey.

Second, recent reforms in Turkey demonstrate the political importance of antitrust enforcement (Arslan, Reference Arslan2021, 265–66). Although the competition authority had previously been relatively independent of political oversight, President Erdogan has reformed the structure of the agency to increase his influence on policy. In 2005, he reduced the number of commissioners. In 2011, he increased the share of commissioners that he appoints, and in 2018, he took the power to appoint all commissioners. He also increased the oversight of the authority by the central ministries (2011), and reduced the independence of the authority by leveling salaries (2012) and allowing less autonomy in hiring (2013). Looking for evidence of political influence in Turkey’s antitrust enforcement therefore seems plausible. Increasing political influence with the competition authority may allow politicians to interfere in competition policy on behalf of politically connected firms, and interference may be facilitated by authoritarian institutions, where there are fewer institutional limits on government authority.

Third, Turkey makes data on the enforcement of antitrust publicly available at the sector level. Turkey reports the number of concluded antitrust investigations in each sector of the economy. Finally, Turkey also has a relatively large economy with detailed firm-level information—including tax data and data on mergers and acquisitions (M&As)—available from Bureau van Dijk’s Orbis Database.Footnote ¹³ Looking at M&As is appropriate in Turkey, because the competition authority is explicitly tasked with investigating and approving mergers (Bradford et al., Reference Bradford, Chilton, Megaw and Sokol2019). The law states that firms must notify the competition authority of planned mergers prior to the sale. The law also directs the agency to prohibit mergers that would create a dominant position (which is relatively broad and does not require that the prohibition be based on abusive acts by the firm) or that would restrict competition. Based on these legal provisions, we expect that—if the competition authority sought to enforce antitrust laws—stricter enforcement of antitrust law would result in fewer M&As and in M&As that produce less concentration. Turkey thus allows us to assess how antitrust and taxation respond to changes in political institutions, holding many other factors constant.

We collect detailed data from Orbis reporting the completed number of M&A deals, the characteristics of the M&A, and the participating firm characteristics for each firm from 2010 to 2018. We merge the M&A data with the Turkish firm-level database using company name and year as an identifier.Footnote ¹⁴ We construct two distinct datasets: one at the firm-level and another at the M&A deal-level. The firm-level dataset includes Turkish firms, while the deal-level dataset compiles information on M&A transactions. For the firm-level data, the unit of observation is a firm in a given year, and it is a deal in a given year for the deal-level data. To measure the tax revenue, profits, and employees at stake in the M&A deals for each firm, we take an average of the values reported by the acquirer and target firms. Averaging the firm data helps improve missingness (for example, we have data on the operating revenue of both acquirer and target firms for only 167 firms—see column (6) in Table 3).

We draw on several measures of antitrust and the competitiveness of the market. First, we conduct a firm-level analysis in Turkey. We code a count variable that captures the number of M&A deals that each firm participated in during each year. Second, we look at the character of M&A deals in the deal-level analysis. For each deal, we code a set of variables capturing characteristics of the M&A transaction. Our first dependent variable is the deal size. We then code dummy variables capturing whether the acquiring and target firms are located in the same industry and whether the acquiring firm gains at least 50 percent ownership. We also code the average level of industrial concentration of the acquiring and target firms’ industries, and the size difference, which we square, between the acquiring and target firm. We expect that as Turkey became more authoritarian, we should observe weaker antitrust and thus a stronger association between tax revenue and the number of M&As. We should also observe larger M&As, M&As where the acquiring firm gains majority ownership, and M&As in more concentrated industries. We might also observe more M&As with larger size differentials, as regulators do not prevent [hostile] takeovers.

Figure 3 presents the average number of completed M&As (as a share of the total number of firms) and the average number of investigations (as a share of all divisions).Footnote ¹⁵ Figure 4 presents the share of competition cases by each NACE Section.Footnote ¹⁶ Our data is at the division level, which is more fine-grained than the section but would be difficult to include in a single figure.

Figure 3. Antitrust enforcement and firm activity in Turkey.

Figure 4. Share of competition cases by NACE Section.

Figures 3 and 4 demonstrate that M&A activity fluctuates considerably but has probably declined on-average, over time in Turkey. The decline in M&As as Turkey became more authoritarian is on its face inconsistent with a weakening of antitrust. However, the decline could plausibly be explained by the increased difficulty of access to financing. Table 2 reports indicators capturing the availability of finance in Turkey over the period between 2008 and 2019. The table shows that fewer firms are taking out bank loans and using banks to finance investments, and that the value of collateral needed to access loans increased since 2008. In addition, our theory makes nuanced predictions about M&As. The theory anticipates more M&As between firms that provide substantial revenue, as well as differences in the character of M&As. We expect regulators to be more permissive toward firms with substantial revenue benefits, especially allowing M&As that increase market concentration or where the firms are from the same industry. Evaluating these more nuanced predictions allows us to assess whether the character of competition enforcement has changed as Erdogan strengthened his control over politics, over regulators and over the economy.

Table 2. Access to financing in Turkey

Data are from the Global Financial Development Database.

Table 3. Corporate tax revenue and M&As

Standard errors in parentheses: * $p \lt 0.05$ , ** $p\lt 0.01$ , *** $p \lt 0.001$ . In column (1), the data is run at the firm-level and the dependent variable is the number of M&As. In all other columns, the data is at the M&A deal-level, and the dependent variables by column are: (2) the deal size, (3) a dummy indicating that the acquirer and target are in the same industry, (4) a dummy indicating that the deal was an acquisition leading to a new majority owner, (5) the average industrial concentration of the acquirer and target, and (6) the operating revenue of the acquirer minus the target, squared.

In a second analysis, we measure the extent of antitrust using the number of investigations into antitrust infringements reported by the Turkish Competition Authority. The Authority provides the sectoral information about investigations from 2010 to 2017.Footnote ¹⁷ We hand-code the sectoral information from the Authority to match the widely used 2-digit NACE division code, available from Orbis. We expect industries that are more important sources of tax revenue to have fewer investigations, especially as Turkey’s political institutions became more authoritarian. Importantly, we assume here that the investigations represent genuine competition investigations, and we expect political influence to result in fewer investigations of those firms that are important tax revenue sources. If Erdogan is able to wield policy to target his political opposition, we might actually observe more investigations as Turkey became more authoritarian. We thus interpret these regressions as largely suggestive. That said, we do observe a positive association between concentration and competition cases (see column 2 of Table 4), suggesting that on-average more concentrated industries are more likely to be investigated.

Table 4. Corporate tax revenue and competition cases

Standard errors in parentheses: * $p\lt 0.05$ , ** $p\lt 0.01$ , *** $p\lt 0.001$ . The dependent variable is the number of competition investigations in each industry-year. Tax, profit, and employee variables are industry averages.

Our independent variables of interest again measure the level of tax revenue and the extent of representation. We use taxation, logged (all data from Orbis). In the industry-level analysis of competition authority investigations, we use the average tax revenue from each industry. As before, we use the authoritarianism score (the inverted Polity2 score) to measure the extent of representation.

To test our theoretical expectations, we again interact the measures of corporate taxation with the authoritarianism score variable. We anticipate that Turkey supervised more competitive markets in years when its political institutions were more democratic. As Turkey’s institutions became more authoritarian, we expect to observe that antitrust weakens and competition is less pronounced for firms and industries that provide larger tax payments, indicated by more and larger M&A deal completions in more concentrated industries and fewer antitrust investigations.

We control for potential confounders, including the average profits of the firms that participated in M&AsFootnote ¹⁸ and the average number of employees, logged. We also include year and industry fixed effects to account for trends over time and stable industry effects (e.g., returns to scale).

Two examples provide an illustration of the sorts of activities captured in the data. In the case of the TAV Havalimanlari (airport) Holding Company, we observe high levels of tax revenue and elevated numbers of acquisitions, both relative to the industry average for holding companies.Footnote ¹⁹ Figure 5 plots the tax revenue and acquisitions of TAV over time. TAV is a Turkish aviation company. The aviation, transportation, and construction industries are thought to be closely connected to Erdogan and have been “mired in corruption” in recent years.Footnote ²⁰ TAV acquired several transport and airport development companies between 2012 and 2019. Importantly, aviation is a sector with high fixed costs, which could pose entry barriers for competitor firms. Traditionally, M&As in industries with high fixed costs are closely scrutinized by regulators. TAV does not seem to have been subject to this scrutiny.

Figure 5. TAV Havalimanlari tax and acquisition activities.

Another firm in our sample that engaged in a large number of M&As is Sise ve Cam Fabrikalari (or simply Sisecam). Sisecam is a glass manufacturer and an important source of tax revenue. Sisecam also increasingly pursued M&A deals over the course of Erdogan’s rise. Sisecam is owned by Isbank,Footnote ²¹ one of the largest banks operating in Turkey. Isbank in turn is owned about a third by the main opposition party.Footnote ²² Erdogan has tried to takeover ownership, but has not yet been successful.Footnote ²³ The size and importance of Isbank and Sisecam plausibly make them attractive targets. At the same time, Erdogan would not want to jeopardize their functioning—and the tax revenue they provide—by doing anything too disruptive. Tax revenue may thus provide some cover to firms, even those who are not owned by Erdogan supporters.

These examples are merely illustrative of how incentives could play out with respect to a specific firm. We now turn to the data to examine whether these firm’s experiences show up in a broader cross-section of firms and industries.

Results

Table 3 reports the estimates of the relationship between corporate taxation and antitrust conditional on regime type in Turkey. Column (1) reports the results of the firm-level analysis, and columns (2) to (6) report the deal-level analysis results. The dependent variable in column (1) is the number of M&As that each firm participated in; in column (2), it is the deal size; in column (3), it is the dummy capturing whether the acquirer and target are in the same industry; in column (4), the dependent variable is the dummy for the transfer of majority ownership; in column (5), it is the average industrial concentration of the acquiring and target firms, and in column (6), the dependent variable is the size difference between the acquirer and the target, squared. Note that the constitutive effect of the autocracy score is absorbed by the year fixed effects, as there is no variation in the autocracy score in each year. We can nevertheless estimate the interactive effect of autocracy and tax revenue.

Although the effects of tax revenue are insignificant on the M&As within the same industry and acquisitions leading to new majority ownership (columns 3 and 4), the coefficients are statistically significant for the number of M&As that the firm engages in (at the 10 percent level), the value of the M&A transaction, the average industrial concentration, and the average size difference (at the five percent level). Furthermore, the estimates of the interaction term are statistically significant in columns 2, 4, and 5. For the remaining results, despite their insignificance, the directions of the coefficients align with our theoretical expectations.

Drawing on these results, we plot the marginal effects of tax revenue in Figure 6. The associations between corporate taxation and different M&A activities become larger as the regime becomes more authoritarian. Especially, in more authoritarian years, the marginal effects of tax revenue become significantly positive for the deal value (column 2), the average industrial concentration (column 5), and the average size difference between acquiring and target firms (column 6). The results suggest that, as Erdogan consolidated his power, firms that provided more tax revenue engaged in higher value transactions, in M&As in concentrated industries, and in deals with counterparts with larger size differences. Additionally, the marginal effects of tax revenue on the acquisition leading to new majority ownership are statistically significant and negative under democratic institutions (column 5): Thus, before Erdogan’s presidency, firms with higher tax payments were less likely to make acquisitions to gain new majority ownership. Taken together the results help corroborate the idea that antitrust authorities scrutinized M&As from firms that are important sources of tax revenue less, as Turkey became more authoritarian. The results are consistent with politicians foregoing the benefits of market competition in exchange for tax revenue in countries with less representative political institutions.

Figure 6. Marginal effects of corporate tax revenue from Table 3.

Table 4 presents the results for antitrust investigations. The level of analysis is the industry-year level. While the coefficient for tax revenue is negatively correlated with investigations, this association does not reach statistical significance. However, the interaction effect is statistically significant and in the expected direction, and its significance persists even after controlling for industrial concentration. This pattern is consistent with tax revenue preventing Turkey’s authorities from investigating specific firms. Additionally, the positive association between concentration and investigations suggests that at least on-average the investigations are being pursued for competitive purposes. To further explore this relationship, we graph the marginal effects of corporate tax revenue on competition cases in Figure 7. The marginal effect of tax revenue becomes statistically significant when the authoritarianism score exceeds 1. This implies that as Turkey’s political institutions grew more authoritarian, authorities were less inclined to investigate industries that generated higher tax revenues. Taken together, the results suggest that Turkey’s authority is neither entirely politicized nor entirely motivated by programmatic objectives: Turkey pursues cases on-average in more concentrated industries. At the same time, it is less likely to investigate firms that provide important sources of tax revenue, and this pattern has become more pronounced as Turkey became more authoritarian.

Figure 7. Marginal effects of corporate tax revenue on cases from Table 4, Column 2.