Anticollusion Enforcement: Justice for Consumers and Equity for Firms

Abstract

Price-fixing cartels are pervasive. John Connor’s data set of Private International Cartels, comprising 1,014 suspected cartels that were either convicted of price fixing or under investigation from 1990 to 2013, reveals the total affected sales of these cartels to be around $1.5 trillion. In recent years, the number of new identified cartels has been averaging over 70 per year, suggesting that the share of industrial output that is cartelized, even for developed economies, is substantial.¹

However, as recent trends in the number of detected collusion suggest, cartels are also crumbling rapidly. Antitrust enforcement around the world has been steadily picking up speed. Total criminal antitrust fines increased from $107 million in 2003 to $1.1 billion in 2012, and total prison sentences increased from an annual average of 3,313 days in the 1990s to 23,398 days by the end of 2012. Higher fines and new tools such as leniency programs for cartel whistleblowers have led to unprecedented enforcement action not only in the United States and the European Union (EU) but also in the rest of the world.

The breakdown of collusive activities that involve higher prices and restricted output is likely to result in the expansion of production. It might also lead to technological change: colluding firms have fewer incentives to innovate, especially when they face little threat of external competition (Vives 2008). We predict that such expansion in production capacity and investment in technological change when moving from the collusive to the oligopolistic equilibria will cause the financing deficits of firms to increase. That might trigger a change in the means of financing and a likely change in the leverage.

In this paper, we study how firms adjust their financing strategies when the cost of collusion become higher and cartels break up. On the one hand, such increased competition in the product market could make firms raise debt that provides convex incentives to managers (Brander and Lewis 1986; Maksimovic 1988). On the other hand, the firms might be rebalancing their capital structure due to the trade-off considerations as lower profitability and higher ex ante probability of default make debt less appealing.

Yet these direct capital structure considerations of pure debt-equity rebalancing and the strategic use of debt are unlikely to be the only relevant factors in the environment when firms are moving between different competitive equilibria. The firms are likely to adjust their product market strategies and such readjustment would lead to a different asset mix which would mean indirect implications to the capital structure choices. In particular, firms might be increasing financing to sustain financial flexibility (Fresard 2010; Hoberg, Phillips, and Prabhala 2014). While both debt and equity can provide such financial flexibility, the exact mix of debt and equity is likely to be a function of the type of investment that the firm is trying to raise. Indeed, expansion of the physical factories might be easier to finance with debt suited to fund collateralizable assets while intangible investment might be better suited to fund with the equity (Kerr and Nanda 2015). Firms might also raise money to keep as excess funds that can be held in cash balance as a competitive threat to rivals. In that case, equity might also be preferable as debt limits firm’s ability to respond to product market threats by removing excess funds (Phillips 1995; Kovenock and Phillips 1995; Kovenock and Phillips 1997).

Since theory does not provide clear predictions about how firms’ financing behavior responds to such a change in the competitive regime, in this paper we take advantage of the variation in the cost of collusion coming from the staggered passage of leniency legislation in 63 countries from 1993 to 2011 to study how more aggressive cartel enforcement affects firms’ debt-equity choices and financial leverage. Leniency programs have been one of the most important developments for cartel detection and deterrence (Chen and Rey 2013). By allowing reduced fines or even providing immunity to cartel members that collaborate in conviction cases, leniency laws are expected to have increased the costs of forming cartels and the benefits of breaking them up.² Such strengthening of antitrust policy has changed the competitive landscape and thus provides an opportunity to identify the effect of changes in the strategic interactions between firms on their financing choices and capital structure.

Importantly, countries passed the leniency programs at different points in time between 1993 and 2011. Based on this staggered nature of the law passage, we attempt to identify the effect of a less collusive product market environment on firms’ financing decisions by following a difference-in-differences strategy. In other words, controlling for firm and time fixed effects, we compare the change in financing choices of firms that were affected by the law to the contemporaneous change in choices of the control group of firms that were headquartered in the countries that have not yet passed such a law.

We find that following the adoption of a leniency law, firms expand by increasing asset growth, coming from higher cash holdings, capital expenditures, and investment in intangible assets, and they fund this expansion by issuing more equity. In some of our specifications, debt issuance activity also increases, though much more modestly. Consistent with more aggressive equity issuance, with some exceptions, we also find that leverage declines following the adoption of a leniency law.

Our results are robust when we control for industry*year, region*year, and industry*region*year fixed effects to filter out regional or industry-specific trends that could simultaneously affect financing choices of firms and competition policy. We also control for observable variables capturing macroeconomic conditions. To mitigate the concern that the adoption of these laws signaled other events that could affect capital structure through alternative channels, we control for import penetration to capture the effect of trade agreements; industry’s Herfindahl-Hirschman index (HHI); country’s capital account openness; adoption of competition law and other anticartel legislation; corporate governance reforms; stock and credit market development; country’s rule of law, governance, and regulatory quality; and corporate tax changes.

While existing evidence suggests that cartels are quite pervasive, the above results do not distinguish between firms that are members of cartels, and those that are not. We expect our results to be stronger for the former. Using a database on actual cartel convictions, we predict the propensity for a firm to be a member of a convicted cartel based on its industry and country, and other firm characteristics. We find that our results are significantly stronger for firms with higher predicted probability of being part of a cartel. We also explore cross-industry heterogeneity and find that our results are stronger in the cases where collusion is more stable according to the theory predictions: for firms in the more concentrated industries, firms in the industries that use less patent protection, and firms in low-growth industries.

As an additional source of variation, we look at how the financing choices of firms in a country respond not only to that country’s adoption of leniency laws but also to the adoptions in other countries where these firms are likely to experience product market interactions. In particular, we look at how a firm is affected by the passage of leniency laws in countries that are major export destinations of the firm’s industry and in countries where the firm’s subsidiaries are located. An appealing feature of this setting is that the passage of a leniency law in another country is likely to be even more exogenous to any other factors in the home country that might simultaneously affect financing decisions of firms and antitrust policy. Indeed, we find consistent results that asset growth and equity issuance also increases for firms in an industry when other countries that are important export destinations for that industry adopt a leniency law, or when the laws are adopted in countries where the firms’ subsidiaries are located.

Finally, we look at another case of strengthening anticartel legislation, the enactment of Antitrust Criminal Penalty Enhancement and Reform Act (ACPERA) in the United States in 2004. The act increased the fines and provided more incentives for leniency law applicants, and thus has been lauded as one of the most significant revisions in the history of the U.S. antitrust system. We again predict the propensity for a firm to be a member of a convicted cartel based on its industry and firm characteristics. We find that U.S. firms with higher predicted probability of being part of a cartel increased asset growth, financing deficit, and equity issuance.

What we demonstrate in this paper is consistent with the interpretation that when the nature of equilibrium in the product market changes from collusion to oligopolistic competition, firms increase asset growth, by raising cash holdings and investment, and finance such investment with equity to retain financial flexibility. Given that all former cartel members are expected to expand investment in production capacity and increase output, financing the expansion with debt would make firms vulnerable to rival firms changing their strategies and expanding production capacity even more aggressively with equity financing.³

At first glance, our results are also consistent with the trade-off theory, which suggests that, when profits are expected to fall, firms will reduce leverage. Indeed, Xu (2012) examines the effect of higher import penetration (instrumented by tariff cuts and exchange rate changes) on leverage, and finds that leverage drops even controlling for current profitability. She attributes this to expected lower future profitability following a reduction of barriers to entry into the industry. An important difference between Xu’s (2012) setting and ours is that unlike firms exposed to greater import competition who experience decrease in asset growth, firms newly exposed to leniency law increase asset growth. Byoun (2008) finds that firms with above-target debt ratios are much slower to adjust capital structure toward the target when they have a financing deficit, as opposed to when they have a financing surplus.⁴ Therefore, since firms newly exposed to a leniency law increase their asset growth and financing deficits as they expand output and capacity, it is unlikely that the significant increase in equity issuance activity is purely driven by debt-ratio rebalancing motives. For these firms, equity issuance serves the dual roles of maintaining financial flexibility in the face of competition and financing asset growth.

Finally, a contemporaneous paper by Ferrés, Ormazabal, and Sertsios (2016) examines capital structure choices of U.S firms that were discovered to have participated in a cartel. They find that these firms reduce leverage during the collusion period, which is seemingly at odds with our findings. A few differences in setup might explain this. First, Ferrés, Ormazabal, and Sertsios (2016) only consider U.S. firms that have been convicted, while we examine all U.S. and international firms that have been exposed to leniency law. Second, the behavior of firms that recognize that collusion is no longer feasible could be dissimilar from that of firms prior to entering a period of collusion. Interestingly, the results on equity issuance and repurchase activities are quite symmetric, – while we find that firms step up equity issuance after the introduction of leniency law, Ferrés, Ormazabal, and Sertsios (2016) discover that firms repurchase equity during collusion period (but presumably also reduce debt as they become more profitable).

1. Relation to Literature

We first provide a brief background of the theoretical literature on product market collusion that motivates our empirical enquiry. Two parts of this literature are relevant for our study. The first concerns the implications of a change in the nature of competition from collusion to oligopolistic competition on firms’ investment behavior. The second studies the implications of such a change (and the associated investment behavior) on firms’ financing choices, especially, the choice between debt and equity.

1.1 Collusion versus oligopolistic competition

Firms collude to maximize joint profits. They do so by restricting output and charging higher prices than what would prevail if they did not coordinate their output and pricing decisions.

Collusion is typically distinguished by being explicit or being tacit. Explicit collusion occurs when participating firms communicate with each other and fix prices, production capacities, or input costs such as wages. Colluding firms might also divide geographic or product markets. In most jurisdictions, such cartels which result in lower consumer welfare are illegal and are subject to the antitrust enforcement. When a cartel arrangement is no longer sustainable (e.g., because of more stringent antitrust enforcement), the participating firms become competitors, and the market structure becomes an oligopoly. Regardless of whether the strategic variables chosen by the firms are prices or quantities (and the resultant competition is characterized by Cournot or Bertrand competition), prices are lower and qualities are higher compared to the collusive outcome.

Collusion also can be a tacit when it involves no direct communication between firms. Even such collusion can be stable if it involves a repeated interaction between firms over many periods. Firms coordinate their prices to levels that are above those that would prevail if they were interacting only once. This is possible under the threat of “retaliation” should any firm deviate and undercut the rivals. To be effective, the retaliation must imply sufficient loss of long-term profits so that the loss outweighs the gain from a one-time deviation, and it must also be in the best interests of the retaliating firms to continue with the retaliation. In a repeated game setting, a multiplicity of equilibria are possible under collusion, but all of them result in higher joint profits, higher prices, and more restrictive output compared to the absence of collusion.

Unlike explicit collusion, tacit collusion is difficult to prosecute.⁵ However, because of the absence of direct communication, it may be difficult for firms to coordinate their prices, especially when firms are asymmetric because of the multiplicity of equilibria. Many collusive arrangements will have both tacit and explicit components. Grout and Sonderegger (2005) argue that “parties might be more likely to engage in overtly collusive practices specifically in those circumstances that are predicted by the theory as being adverse to collusion” because “the need for cartel members to communicate intensifies precisely when collusion is harder to sustain.” More stringent antitrust enforcement targeting explicit collusion may therefore cause even tacit collusion more difficult to sustain.

When collusion (either explicit or tacit) breaks down and firms have to compete with each other for market share, new investment may be necessary. One simple reason for this is that each firm will produce more in the new equilibrium, and this might require investment in additional production capacity. Firms may also have to step up investment in research and development (R&D), accelerate new product development, and hold more cash to finance such expenditures in the future and deal with predatory tactics by rivals. All this is likely to generate a demand for financing. We next discuss the financing choices firms face.

1.2 Implications for financing

In contrast to the implications for investment and R&D spending when the equilibrium switches from collusion to oligopolistic competition, the financing choices that firms will make are far from clear.

Debt financing makes equityholders’ payoff convex in cash flows. The literature on the strategic role of debt (Brander and Lewis 1986; Maksimovic 1988; Dasgupta and Titman 1998) is based on the premise that limited liability could affect the product market choices of firms. Brander and Lewis (1986) show that firms in an oligopoly have an incentive to choose debt in equilibrium. This is the case regardless of whether firms compete in prices or quantities. However, there is no role for debt when firms are colluding and maximizing shareholder value, regardless of whether firms are colluding tacitly in a repeated game setting (Maksimovic 1988) or whether they are colluding explicitly and setting prices to maximize joint profit each period (Brander and Lewis 1986).⁶ That said, it is difficult to infer directly from these models what one should expect regarding financing or capital structure choice if, for exogenous reasons, the nature of competition changes from collusion to oligopolistic competition. One reason for this is that investment decisions are typically not considered in these models, and as noted, the transition to a new product market equilibrium may well entail new investment in tangible and intangible assets, which in turn might affect financing choice and affect subsequent capital structure.

Another line of research emphasizes the need to maintain financial flexibility under competitive pressure. Kovenock and Phillips (1995, 1997) explicitly consider the role of capital structure in financing investment in oligopolistic product markets. The authors argue that firms will be constrained in expanding production capacity once internal funds are exhausted. Given rival firms’ choices, this will happen at lower levels of production capacity when the debt ratio is higher. When firms compete in prices, higher debt ratios can increase both own and rival firms’ prices and profits.⁷ This reasoning suggests that when collusion gives way to oligopolistic competition and firms need to expand investment, they would prefer to maintain lower debt ratios and preserve internal financing.

Fairly robust empirical support has been provided for such a “financial flexibility” argument. Zingales (1998) examines the effect of deregulation in the trucking industry. He finds that, consistent with Kovenock and Phillips (1995, 1997), debt overhang model (Myers 1977), and models of “deep pockets” and predatory behavior (Telser 1966; Benoit 1984; Bolton and Scharfstein 1990), firms with high leverage prior to deregulation are more likely to exit the industry, due to their inability to compete effectively by increasing investment and charging lower prices.⁸ Subsequent empirical research has also found the importance of financial flexibility in competitive environments.⁹ In our context, this line of research would imply that as collusion ends, firms need to be poised to increase investment and compete aggressively. This requires less reliance on debt as opposed to equity, and building up internal liquidity.

Mainstream theories of capital structure provide additional perspectives on this issue. Firms—especially larger firms that are more likely to be cartel members—generally finance their capital expenditures with debt, rather than equity. When collusion ends, they are in a situation where they need to step up their investment, but at the same time, they experience decrease in their profitability. In this situation, the pecking order hypothesis suggests that the reliance on debt financing would increase. On the other hand, the trade-off theory of capital structure calls for lower optimal leverage as expected bankruptcy costs increase. Firms are therefore expected to rebalance to a lower leverage ratio.¹⁰ However, the literature also finds that trade-off-type rebalancing mostly occurs when firms have a financing surplus, that is, when their growth opportunities slow down and internal funds are used to reduce debt. There is little evidence that highly levered firms rebalance via equity issuance, presumably because such rebalancing entails wealth transfer to bondholders.

Finally, the type of financing that firms seek in the new equilibrium is likely to be sensitive to the type of assets they acquire. Fixed assets that can be collateralized are typically financed with debt, whereas riskier R&D investment, advertising, employee training are easier to finance with equity. Moreover, if firms want to carry a liquidity buffer either to fend off predatory threats or to finance future investment by adding to their cash holdings, equity is likely to be the preferred financing choice.¹¹

The change from a collusive regime to an oligopolistic one thus is very likely to be associated with additional external financing, with both debt and equity being issued, but their relative importance is ultimately an empirical question.

2. Empirical Strategy and Data

2.1 Background on leniency laws

Given the importance of the antiwelfare implications of product market collusion, governments have devoted considerable resources in tackling it. One of the most effective tools has been the introduction of leniency programs (or leniency laws). Leniency laws allow market regulators or the courts to grant full or partial amnesty to those firms that, despite being a part of a collusive agreement, cooperate in providing information about it. In particular, a typical leniency program stipulates that the first firm that provides substantial evidence to the government (if the latter does not yet have sufficient evidence to prosecute the cartel) gets automatic amnesty. In countries where the firm’s managers, employees and directors may face criminal liability for participating in a collusive agreement, as is the case in the United States, amnesty also extends to waiving such criminal liability. As suggested by Hammond (2005), U.S. leniency law, which was strengthened in 1993, proved successful in destabilizing existing cartels and deterring the formation of new cartels and has thus inspired other countries to pass similar laws. Dong, Massa, and Žaldokas (2018) show that the passage of leniency laws significantly harms collusion. In particular, they find that the passage of leniency laws increases conviction rates and generally lowers gross margins of firms, thus also capturing the effect of leniency laws on the unobservable break-ups of cartels. Table 1 reports the list of leniency law passage years around the world.

Although the laws are not passed in a vacuum and are arguably influenced by economic and political conditions in the respective countries, countries do not seem to have followed one particular trend and reason for such law passage. Some countries passed the law after prominent collusion cases. For instance, Hungary did so after it faced significant criticism concerning its competition investigation against mobile telephone operators, while Switzerland strengthened its competition law in 2003, including the passage of leniency laws, after it failed to prosecute firms involved in the vitamin cartel. Taiwan passed the law as a response to general concerns about rising consumer prices.

Other countries passed leniency laws after significant pressures from the United States, the EU, or supranational organizations (Lipsky 2009). For instance, Mexico passed the law in 2006 following general recommendations of an OECD Peers Review in 2004 on Competition Law and Policy in Mexico which reported that its antitrust authority needs better investigative tools, including the ability to give leniency to a whistleblower revealing secret cartel conduct. Similarly, the United States had bargained for strengthening of Singapore’s antitrust law in its negotiations for a bilateral free trade agreement.¹² Moreover, the EU has fostered the adoption of leniency laws by its member states and often seeks similar provisions in its bilateral association and trade agreements. The IMF and the World Bank ask for the overhaul of antitrust laws as a condition for loans and other funding (Bradford 2012).

In some cases, the passage of leniency laws was contentious. For instance, the leniency law met significant opposition in the Swiss Parliament as the law relies on denunciations that run contrary to Swiss legal tradition. Japanese Business Federation (Nippon Keidanren), the most influential industrial organization in Japan, extensively argued against such a law in Japan, claiming that cooperating and informing on fellow participants in exchange for a lower sanction is an affront to Japanese culture, and should only be considered as part of the wider review of the entire criminal law system. In Table IA1, we report the table from Dong, Massa, and Žaldokas (2018) that estimates the Cox proportional hazard model predicting the passage of leniency law in the country. The most consistent predictor is the economic development of the country as proxied by the log gross domestic product (GDP). Right-wing affiliation of the chief executive has a negative effect, as compared to center- or left-wing orientation.

2.2 Identification

Against this background, we posit that no single particular trend has led to leniency law passages. We thus employ a difference-in-differences identification strategy to estimate the effect of competition on financing strategies of firms.

(Leniency Law)|$_{kt}$| equals 0 before the passage of the leniency law in country |$k$|⁠, and 1 afterward. |$X_{ikt}$| is a vector of the different firm, country and industry controls, while |$\gamma $| and |$\tau $| are firm and year fixed effects, respectively. In our baseline specification, our control variables |$X_{ikt}$| include firm size and profitability, the country’s GDP and unemployment rate, imports as a percentage of GDP, and the exchange rate. In a standard difference-in-differences setting with staggered implementation of laws (Bertrand and Mullainathan 2003; Giroud and Mueller 2010), the treated group comprises all firms that are headquartered in countries that have passed a leniency law by year |$t$|⁠. The control group comprises firms in countries that never adopted a leniency law in our sample period and firms headquartered in countries that adopted a leniency law at some later point of time.

In addition to using the passage of leniency laws in a firm’s headquarter country, we implement an alternative identification strategy. We create a treatment variable based on a firm’s exposure to the passage of leniency laws in those countries to which the firm’s industry sends a significant fraction of its exports. By making it more difficult to form international cartels with industry peers in the countries that are likely to be firm’s sales markets, the passage of leniency law in another country also increases the costs of collusion.

Unlike Equation (1), in Equation (2), unless no country to which a firm’s industry is exporting has passed a leniency law, a firm is considered as treated, and the intensity of treatment changes as more of the countries to which this industry exports adopt leniency law.

Finally, our third identification strategy relies even more directly on the international nature of firm operations. In the specifications above, we assign our treatment of leniency law passage based on the firm’s headquarter country, where presumably most firms have most of their sales. Export market leniency laws measure already considers that firms also sell to other countries and are exposed to the other countries’ antitrust codes. However, for a subset of firms we go further and have collected data on their actual international operations. We can thus test whether the passage of laws in other countries where they operate, also has a significant effect. More specifically, we measure a firm’s exposure to leniency laws by looking at the distribution of the firm’s operations around the world in terms of sales as recorded in Lexis-Nexis Corporate Affiliations database. So, we construct a measure of exposure to leniency law changes based on the proportion of firm activity that takes place in the country that experiences the law change. To illustrate, consider two firms, A and B, both headquartered in Germany. Firm A has 75% of its operations in Germany, and 25% in France; firm B has 25% of its operations in Germany and 75% in France. So, when Germany introduced the leniency law in 2000, firm A should have been affected more than firm B.

2.3 Data

In our analysis, we consider all nonfinancial firms in Worldscope, Compustat Global, and Compustat North America data sets from 1990 to 2012. We start with Compustat Global and Compustat North America samples and following the techniques described in Karolyi and Wu (2018) we augment this data set with Worldscope data as this allows us expand the sample of international firms we cover in our analysis and provides additional variables we can use in the analysis. Our initial sample covers 633,400 firm-years. We report some descriptive statistics in Table 1.

We collect information on the passage of leniency laws in 63 large countries from the Cartel Regulation 2013, published by Getting the Deal Through. We manually double check this information and complement it using press releases and news articles in Factiva database. We report the years when leniency laws were passed in Table 2.

Export data used to construct Export market leniency laws measure comes from CEPII TradeProd Database that has bilateral trade flows for more than 200 countries at ISIC industry level over 1980-2006. We match them to the three-digit SIC and average over the respective values within the three-digit SIC in case multiple three-digit ISIC codes match to three-digit SIC codes. The data on firm operations around the world come from the subsidiary data in Lexis Nexis Corporate Affiliations database.

3. Main Results

3.1 Univariate results

We start by plotting asset growth and financing deficit for the affected firms. Figure 1 plots the mean of the mean change in asset growth of treated firms and control firms in the same industry, 2 years before and 5 years after the adoption of a leniency law in a country. Thus, the control firms are all firms in the same industry in countries that had not passed a leniency law in the 7 years surrounding the event date.

Figure 1

Trends

We plot mean change in (a) asset growth, winsorized at 1%, (b) net equity issuance and net debt issuance, winsorized at 1%, (c) net equity issuance (common stock over lagged assets), winsorized at 1%, and (d) net debt issuance (financial debt over lagged assets), winsorized at 1%. As a control sample, we consider firms that were not affected by a leniency law over the same period as the treated firm but were in the same SIC3 industry; that is, control firms did not have a leniency law introduced over 2 years before to 5 years after the introduction of the leniency law for the treated firm.

Open in new tabDownload slide

While we see that the treated and the control firms start from the same level, asset growth starts shooting up for the treated firms in the year after the passage of leniency laws but later trends level off. Similarly, for the financing deficit, we see rather parallel trends 2 years before the law passage and then diverging trends 1 year after the law passage. These results suggest that asset growth and financing deficit increase after the passage of leniency laws. We also report corresponding graphs for equity and debt issuance. Later, we further study these trends in a regression setting.

3.2 Leniency laws and asset growth

In this section, we present regression results to examine the effect of leniency law adoption by a country on the asset growth of firms in that country. We first provide general results on whether firms increase their asset size, and we later deconstruct whether such increase comes from the capital expenditures in expanding capacity, the investment into intangible assets, or the increase in the cash buffer. We report the specifications where we use the passage of the leniency law as our treatment variable in Table 3, panel A, and the results where we study export market leniency law in Table 3, panel B.

3.2.1 Baseline regressions

Panel A of Table 3 presents results on our baseline specifications in the difference-in-differences setting, which incorporates firm and year fixed effects. The dependent variable is the annual asset growth. In Column 1, we only consider the effect of a leniency law without any additional controls, and, in Column 2, we add firm-specific variables and several other variables to capture macroeconomic conditions and import competition. We see that leniency law passage is associated with higher asset growth. We choose the latter specification to be our baseline, and we find that the asset growth increases by as much as 7% in our specification, which corresponds to one-third of the unconditional mean in the sample.

In Column 3, we add the additional control variables: tangibility, sales growth, lagged by 2 years, and industry market-to-book ratio.¹³ Tangibility, estimated as the proportion of fixed assets over total assets, should capture firm’s collateralizable assets while the lagged sales growth should capture growth opportunities. While our data limitations do not allow us to control for individual market-to-book ratios for international firms going back to 1990s, we control for industry market-to-book ratio to capture industry growth opportunities. We calculate the latter based on Compustat North America data as the three-digit SIC industry median, and we control for it for all—even non-North American—countries. These additional controls come at the expense of substantially reducing our sample but we still find that the passage of a leniency law has a significant positive effect on the asset growth.

In Column 4, we control for industry*year fixed effects. This means that we are comparing treated and control firms in the same year in the same industry. As a result, the specification controls for any common industry trends that could be correlated with leniency law adoption. The coefficient of leniency law remains positive and significant. Column 5 includes region*year fixed effects to absorb factors at the regional level, such as those related to multilateral or regional trade agreements, or any regional economic trends that could affect the capital structure and the propensity of leniency law adoption in these regions.¹⁴ The specification thus explores variation within adopting and nonadopting countries within each region and year to examine whether there is any effect of leniency law adoption on asset growth. The coefficient of leniency law adoption is positive and significant.

Finally, in Columns 6 and 7, we perform two important robustness checks that relate to two particular geographic areas. Since the United States adopted a leniency law very early in our sample period, and at the same time U.S. firm-years constitute one-fourth of the regression sample in Column 2, it is possible that our results are driven by a time-trend affecting U.S. firms only. In Column 6, only non-U.S. firm-years are retained, and both the magnitude of the effect decreases but also the significance of the effect of leniency law drops.¹⁵ Finally, in Column 7, we address a possible concern with the determination of the year when a leniency law becomes relevant for firms in the EU. While the EU adopted a leniency law that would become applicable to all EU member countries in 2002, individual countries passed a leniency law that would apply to all firms doing business with these countries in a staggered manner. In Column 7, we assume that the effective date for leniency for an EU member is the later of 2002 and the year the country joined the EU. Our results remain unchanged.

3.2.2 Export market leniency laws

In this section, we repeat the same tests used in Table 3, panel A, but replace the leniency law adoption indicator variable with the Export market leniency laws measure that looks at the leniency law adoption in the countries other than the firm headquarter country. In addition, we study the subsidiary-sales-weighted measure of leniency law adoption in other countries.

Table 3, panel B, reports the results. The coefficient of the Export market leniency laws variable is positive and significant in all specifications. The results suggest that as the exposure of domestic firms to markets under leniency law increases, suggesting greater exposure to markets characterized by noncollusive behavior, asset growth rises. Results for subsidiary exposure to leniency laws reported in Column 7 are similar.

We perform similar sets of robustness tests as before. Columns 2 and 3 control for additional firm variables. Column 4 includes three-digit SIC industry*year fixed effects. Column 5 includes geographic region*year fixed effects. Column 6 restricts the sample to non-U.S. firms. Our results are robust to these specifications.

Using both identification strategies, we find that stronger actions against collusion have led to a faster growth in assets, that is, larger investment. We note that such increase in asset growth and investment comes despite the drop in profitability (see Dong, Massa, and Žaldokas 2018 for this result).

3.2.3 Deconstructing asset growth

As we argue in Section 1, the actual use of funds is likely to determine the type of financing. The expansion of the physical factories might be easier to finance with debt suited to fund collateralizable assets while intangible investment might be better suited to fund with the equity. Firms might also raise money, raised from equity issuances, to keep as excess funds that can be kept as a competitive threat to rivals. We thus next check where the growth in the assets is coming from. We group the change in assets into three categories: (a) capital expenditures; (b) investment in intangible assets; (c) and the change in cash holdings.

We then estimate our specifications like in Table 3, panel A, Column 2, and in Table 3, panel B, Column 2, on these three different components.¹⁶ In Columns 1 and 2 of Table 4, we report the results for capital expenditures, defined as the change in tangible assets, adjusted for depreciation and scaled by lagged assets. In Columns 3 and 4, we report the results for change in intangible assets, adjusted for amortization and scaled by lagged assets. In Columns 5 and 6, we report the results for change in cash and short-term investments, scaled by lagged assets. These results show that leniency laws have led to the increase in investment into both tangible and intangible assets and most notably in increased holdings of cash. The increase in cash holdings could be a consequence of fixed costs of issuance, because of which firms raise more than they plan to invest immediately and hold cash for future investment, or because of a need to maintain financial flexibility to fend off predatory threats by rival firms.

3.3 Financing deficit and issuance activity

Further, we look at financing deficit. Following Chang and Dasgupta (2009), the financing deficit is defined as the difference between a firm’s requirement for funds (due to investment and dividend payments) and internally generated funds, and is identically equal to the sum of net issue of debt plus net issue of equity. Our results, based on the same specifications as for the asset growth, are reported in Table 5. We find that leniency laws are associated with a higher financing deficit, that is, larger issuance activity.

We further study the sources of financing. Table 6 presents the results for the net equity issuance, and Table 7 presents the results for the net debt issuance. Overall, we find strong effects for equity issuance and weaker effects for the debt issuance. That both debt and equity issuances increase is consistent with the finding that the financing deficit increases after the passage of leniency laws as firms step up investment and experience decrease in profits. However, most of the effect comes from the equity issuance. The economic magnitude of the effect of a leniency law, though positive and often statistically significant, for debt issuance is about one-tenth of that for equity issuance.