The Return on Invested Capital is an accounting metric that reflects the performance of a firm in a given period, regardless the financial flow linked to the operation (PENMAN, 2010). Prior empirical accounting-based research adopts proxies from the accounting statements to analyze the effects of managers decisions and firms’ characteristics to compare performance and indicates that firms’ specific characteristics have major impact on performance than industry characteristics. Accounting numbers allow the users to recognize the firm performance by means of return indexes, such as the Return over Assets (ROA), Return over Equity (ROE) and the Return over Invested Capital - ROIC.

The ROIC excludes the interests and taxes effects, in order to isolate the operational return of the available operational resources to the firm (CHEN and HUANG, 2006; GOLDSZMIDT, 2010; HOUGH, 2006; MISANGYI, ELMS, GRECKHAMER et al., 2006; SIMONS, 1999). Chen and Huang (2006) argues that such number better reflects the firm operational decision making and, then, it should be preferred in relation to the metrics based on total assets or the equity. Also, this configures as a relation between operational profits and operational revenues, acting as a reliable investment decision indicator.

The CVP analysis deals with the classical economic problem about the optimal level and output mix for the firm, assuming that as long as the firm has a set of resources and, at least, one cost is fixed. The accounting cost structure analysis has the necessary characteristics to be a proxy for the economic short run model, characterized by the emphasis on costs and revenue behavior over a set of variations of mix and outputs levels (DOPUCH and BIRNBERG, 1969). The CVP analysis is a simple analytical tool for management decisions (GUIDRY, HORRIGAN and CRAYCRAFT, 1998) that provides a wide financial overview of firms’ decision process (HORNGREN, FOSTER and DATAR, 1994).

The intersection between the curves of Sales Prices and Total Cost indicates the firm’s accounting break-even point, measured by BE = (FC_t /CM_unit ), where BE means the break-even point, which is represented in terms of volume, since the denominator has unitary volume representation; FC_t total fixed costs and fixes expenses; and, CM - or Contribution Margin - which is represented at unitary terms.

At this point, the revenue is equalized to costs, indicating the minimum volume of revenue not to incur in losses. In turn, the marginal revenue relates to each additional unit of production sold, with a sales price function MR = ?(SP) , where MR is the Marginal Revenue and SP represent the sales prices. When firms operate at full capacity, CM and sales increasing are negatively associated since it indicates the necessity of new investments, and following this argument, idle capacity may increase the profit margin of the firm by an increase in sales (JORGENSEN, SADKA and LI, 2009).

The DOL can be used as a risk metric (HUFFMAN, 1983) since that differences in production process impacts on fixed and variable costs share (LEV, 1974), answering to the firm returns as shown in prior literature (MCDANIEL, 1984; NOVY-MARX, 2011; PERCIVAL, 1974). When markets are further from competition, firm’s tends to be less efficiency-seeking and more effectiveness-seeking when competing in a strong oligopoly market (HUNT and DUHAN, 2002), which leads to greater operational leverage statistical significance and influence over operational performance when markets are near perfect competition and monopoly.

At Brazilian capital market, the relationship between returns and DOL has been studied by Tabak and Guerra (2002), Dantas, Medeiros and Lustosa (2006), Lustosa and França (2011) and França and Lustosa (2012). França and Lustosa (2012)also points that in a near competitive market that is a negative relationship between DOL and returns. Zhang (2016) Considering DOL as a metric of operational risk, as shown by Gahlon (1981), we hypothesize that:

H₁ - There is an association between the Degree of Operation Leverage and the Return on Invested Capital.

The economic theory of the firm analyze the relation between a single firm and its industry, and this states the output result to be the outcome of the market forces, considering market price (ALDRICH and PFEFFER, 1976). Differences in market-structure impact on price and production decisions of competing firms on their industries (MAS-COLELL, WHINSTON and GREEN, 1995; THOMPSON JUNIOR and FORMBY, 2002). Under this concept, Besanko, Dranove, Shanley et al. (2006) argues that the firm relies on its conduct to consider the market competitiveness. Moreover, at perfectly competition, accounting numbers and economic theory are able to converge (BEAVER and DEMSKI, 1979), since there are no opportunity costs and we can identify that firms as price-takers and consider the ratio MR/MC to be equal to 1 (FRANÇA, 2012). Under a management perspective, firms compete for shares and resources with their peers leading to product market interactions that have significant impact on firm policies (ZHANG, 2016).

Competition level and market-share are both broadly studied due to its relevance for the firm management decisions and profitability generation capability (RHOADES, 1993; SCHERER, 1965; SHEPHERD, 1972). In addition, (HALL, 2004) shows that changes in economic conditions imposes significant limitation to firms so that the competition plays a moderating effect between nonfinancial indicators and financial performance (BANKER and MASHRUWALA, 2007). Banker and Mashruwala (2007) also show that nonfinancial performance measures make sense in higher competitive markets, since the market structure empowers consumers and employees to choose between different firms.

Prior research also indicates competition as prominent factor on firm-decisions making indicates that firms from United States under strong competitive pressure from the product market rely less on banking finance (BOUBAKER, SAFFAR and SASSI, 2018); and that intense competition leads firms to lower levels of voluntary disclosure (LIN and WEI, 2014). Zhang (2016) shows a moderating role of product market competition in the strategy-performance link. We follow Besanko, Dranove, Shanley et al. (2006) usage of the HHI as a measure to market competition level.

Mandelker and Rhee (1984) research points toward the existence of a positive relationship between the degree of financial leverage and systematic risk. Financial risks are linked with a company selection of its financial structure and financial leverage increases the likelihood of firm’s inability to service the debts, and a good financial policy aims to harmonize the degree of risk faced by shareholders, the impact of borrowing and the return equity (BARAKAT, 2014). Campanella and Serino (CAMPANELLA and SERINO, 2019) research that shows existence of correlation between firm financial leverage and key firm aspects as age, size and profitability.

Recent research indicates that financial leverage plays a big role in maximization shareholders’ but has different impacts on the value of the firm country to country because of different tax laws and different tax brackets (OBRADOVICH and GILL, 2013). Other studies evidence a positive effect of leverage and firm value (GUPTA, KUMAR and VERMA, 2016; RUSSELL, BRIGGEMAN and FEATHERSTONE, 2017; ZEB and RASHID, 2016) and also between systematic risk and degree of financial leverage (HUFFMAN, 1983). In the same way as expected for DOL, to DFL we expect to find statistical significance and influence over operational performance when markets are near perfect competition and monopoly and considering that market structure impacts on firms’ costs behavior and industry returns. Then, we hypothesize that:

H₂ - There is an association between the Degree of Financial Leverage and the Return on Invested Capital

H₃ - Market competition level exert a moderating effect on the relationship between Degree of Operating Leverage and Return on Invested Capital.

Consider the impact of Size on firms’ performance led to diverse evidences. Literature shows that size plays a major variable for firm’s performance and for the study of performance at industry context, in both microeconomic theory and industrial organizational (PORTER, 1979; BESANKO, DRANOVE, SHANLEY et al., 2006), with firms inserted in a causality chain where firm’s size influences its performance (THOMPONSON and FORMBY, 1996). However, under an operational perspective, Size may play different roles.

Fiegenbaum and Karnani (1991) point the necessity of big firms to operate with intense exploration of economies of scale. Also, the study emphasizes a trade-off between size and volume flexibility, where small firms have the advantage of flexibility on sales volume compared to the biggest firms. Since the smaller firms do not incurs in economies of scale, managers are encouraged to perceive better performance by other means.

Marcus (1969) findings Marcus (1969)’s findings indicate an erratic relationship between firm size and profitability within an industry, with some firms showing a positive relationship and others showing a negative relationship. Due to prior literature, we believe that size acts as a moderator of the moderating role of market structure on the relationship between DOL and operational performance, exerting a double-moderating effect.

Considering such evidences of size’s role on firms’ operational performance, we bring it into our model as a control variable. The following chapter presents the applied methods, in order to define our sample, the variables construction, and the applied econometric issues.

We use the ROIC as a performance measurement in order to capture the operational approach of the firm and its sensitivity or exposure to different cost structure (SIMONS, 1999):

This research aims to study firm operational performance, detached from the exposure to taxes among industries. The Earnings before interests, taxes, depreciation, and amortization (EBITDA) overrides other profit lines on the income statement, such as NOPLAT, NOPAT, EBIT or Net Income, due to its alignment with the research purposes.

The Degree of Operating Leverage means the sensitivity of income to a variation of the revenues. We follow Garrison and Noreen (2001) to calculate the traditional observed DOL:

Where OI_i,t is the operating income of firm i at quarter t; and, NR_i,t is the net revenue of firm i at quarter t. The variable aim to capture how market reacts to results of managerial decisions regarding firm costs structure.

Dhaliwal, Huang, Khurana et al. (2008) and Gjerde, Knivsflå and Sættem (2002) concurs that industry specific facts exert impact on firm performance variability. In this perspective, we use the Herfindhal-Hirshman as a metric of product market competition. Following Besanko, Dranove, Shanley et al. (2006), we adopt the function:

Where Market-share is the net operating revenue of a firm i (or its total assets) scaled by the total of the industry. Kelly (1981) and Rhoads (1993) agree that the Herfindhal Index ought to be carefully interpreted due to its limitations, such as requiring public data of each firm, which is unavailable. Besanko, Dranove, Shanley et al. (2006) consider the relative size of the biggest firms to be a major factor on the management and, consequently, on the performance. We approach the subject through concentration level, when we adopt (market-share)² as measure with the intention to capture firms’ market power.

Following the findings of Campanella and Serino (CAMPANELLA and SERINO, 2019) research that shows existence of correlation between firm financial leverage and key firm aspects as age, size and profitability, which indicates the importance of analyzing the DFL impact on firms’ profitability we include DFL as a moderating variable at our model, following Mandelker and Rhee (1984):

Where EAIT_jt is the earnings after interest and taxes of firm j at time t and Xjt denotes the earnings before interest and taxes of firm j at time t. The DFL captures the sensitivity of a company’s earnings per share to fluctuations in its operating income, as a result of changes in its capital structure and we expect a positive effect on ROIC.

Size has been subject of analysis over the years. Shepherd (1972), Hansen and Wenerfelt (1989) and Leon Li and Huang (2011) points toward a positive relationship between Size and return. However, Lever (1996), Chuang (1999) and Pull (2003) shows a counter-hypothesis, that Size plays a negative role on firm’s performance. Porter (1979) approach the same subject and in recent study adopt size as control variable on a firm performance research (WANG, SHARMA and CAO, 2016). The firm’s size proxy adopted is given by the following equation:

To understand the role of Size at Brazilian market on the relationship between DOL and operational performance we will address the question by observing Size as a moderating variable of the relationship between DOL and ROIC.

Firms listed at the Brazilian Stock Exchange between 1996 and 2016 compose the sample, totalizing 879 firms. The characteristics of our research and characteristics of some sectors demand not to consider all database. As consequence were excluded Finance and Insurance and Funds sectors due to their specific regulations; Others sector due to the difficult to establish firm market competition level; Energy sector exclusion is due to the strong regulation and other industry specific characteristics; and, Software and data sector due to the reduced number of observations. After the exclusions, 419 firms remained, totalizing 57,56% of the original sample.

All data were non-consolidated and obtained at Comdinheiro®, specialized databases for market information. From the original 879 firms, 419 firms remained on the database after sector exclusions, which represents 47.67% of the original sample. In addition, negative results were also excluded from our database. The same treatment was applied to missing values. Table 1 shows the number of excluded observations:

TABLE 1
Sample selection process

From the original 26.571 observations that remained after our first sample selection by exclusion of subsectors, we found 13.497 observations that compose the Full Sample after the second sample selection.

Previous analysis (available upon request) on the dependent variable shows the ROIC mean greater than median, which suggest a significant skewness and a kurtosis. In addition, there is a large distance between median and maximum value, denoting an asymmetry distribution, which leads to a high standard deviation, indicating presence of outliers on the distribution. Outliers may disturb the regression significance, and the selected procedure was the exclusion of outliers.

TABLE 2
Descriptive Statistics

Considering that the variables variance allows to standardize the DOL and Size variables, we were able to put those variables in range. The procedure increases the data quality since it provides a sensible unit scale. In this sense “Scaling should be performed in such a way that the variances of the measurements reflect their relative importance” (KRESTA, MACGREGOR and MARLIN, 1991, p. 44), which is what we aim to capture with those variables in our research. Table 2 shows the final sample descriptive statistics.

Differences between sectors structure lead to differences on firm operational return demands. Centering the variable waives the real and effective number in order to allow a more trusted analysis since the comparability analysis refers to the distance of firms’ ROIC from the sector’s ROIC mean. The outlier exclusions solved the high standard deviation and mean greater than median problems with the dependent variable, allowing considering ROIC in range.

The correlation matrix for the numeric variables used in the models (available upon request) indicates the absence of high correlation between independent variables. We also confirmed this issue by the Variance Inflation Mean of 5.25 denoting that the variables do not show multicollinearity problems.

We categorize the sectors considering the observed mean HHI. Industry and Commerce subsectors of our sample as classified as Oligopolistic markets. The Services sector, however, as monopolistic market. Table 3 summarizes the sample:

TABLE 3
Market Structure

Despite the difficulties when considering data quality for the Brazilian market, such as the low number of observations or the high level of concentration, these results shed lights to the firm’s choices and to the consequent analysis of the its impacts and this also indicates those firms that conducted the best output on the period.

We run an OLS regression for the Full Sample, subdivided in three models that contemplates all variables. investigate the existence of a moderating effect of market strucuture on the relationship between ROIC and DOL. At The model (1), we consider the relation of DOL with firms’ operational performance, adding HHI ­and Size as control variables. Considering DFL correlation with firms’ profitability (CAMPANELLA and SERINO, 2019) and the possible moderating effect on ROIC and DOL relationship, model (2) adds DFL as control variable to measure the impact of firms’ financial leverage on ROIC. Model (3) includes the moderation of HHI and the combined moderation HHI-DOL on the analysis to answer to the second and third hypotheses.

Our models allow us to visualize a number of relations between our leverage variables, HHI and ROIC, as shown on our Figure 1. However, the focus of the present work is to investigate the moderating effect of market structure on the ROIC and DOL relationship. Since our scope is restricted, and we will focus on lines 1, 2 and 3. Line 4 shows the combined moderation and will allows to increase our comprehension of the moderating effect of market structure, thus we will often address to its results.

We find a negative signal for DOL in all models, and compared to the mean, an increase on DOL impacts negatively on firms’ operational performance at 1% of alpha. The negative impact of DOL on ROIC is expected, considering that higher levels of operational leverage suggest that firms are profiting on margin rather than volume, and high levels of operational performance indicates maximization usage of firms’ available resources.

FIGURE 1
Research Design
Source: Elaborated by the authors.

For the HHI­ variable, at 1% level, we find that more competition relates negatively with ROIC at all models, which aligns with expected results at Brazilian market. When market tends to monopoly, firms’ maximization point differs from MR = MC maximization point of perfect competition markets. Including DFL as control variable in our model (2) enhance the quality of our results. The DOL and HHI variables remains significant at 1% level, and DFL shows statistical significance at 1% with the expected positive signal.

The third model includes HHI as proxy for market structure moderating the relationship between DOL and ROIC, and DFL as a moderating variable of the HHI moderation, which indicates that firms’ must consider market structure when making cost-structure related decisions. For our full sample, the moderating role of market structure on the relationship between DOL and ROIC shows statistical significance at 1%, but the combined moderation HHI-DFL do not presents itself as significant.

The positive signal of HHI moderation indicates that market competition exert an positive effect on firms’ performance, which indicates that at mean, firms with lower DOL present shows greater operational performance when compared with firms that possess higher DOL. The model suggest that, for our full sample, when markets walks toward competition, firms perceive more influence of the DOL on firms’ ROIC. Which means that managers tends to reduce DOL when markets becomes more competitive, adjusting production to market structure.

For our Full Sample analysis, the results indicate that we shall accept H₁ and H₂ hypothesis, however H­₃ shall be reject due the lack of statistical significance. To increase our understanding of the moderating effect of market structure on the relationship between ROIC and DOL, we segregate our sample in three major sectors. Literature suggest that the heterogeneity between different sectors and market structure may have a major role in the absence of statistical significance (HANSEN and WERNERFELT, 1989; PORTER, 1979, 1989; SHEPHERD, 1972), especially when we consider the assumption of perfect competition or monopoly where the DOL roots (WICKRAMASINGHE and ALAWATTAGE, 2007).

McGahan and Porter (1997) shows that profitability has a complex relationship with different characteristics, as industry effects, and how those variables impact on profitability depends of firms’ sector. Following the argument, we divide ours sample in three major groups that relies on the difference of structures: Industry, Commerce, and Services. At Table 5 we can visualize our results. The models shows greater adherence to Industry sector, with Commerce sector showing lowest adherence level.

We present the model’s output subdivided by sectors and between our three models. HAIR JUNIOR et al. (2008) states that moderating effect occurs when an independent variable can affect a regression by changing the relation between the independent variable of the regression and a dependent variable when the value of the moderator variable changes.

For the Industry subsector, we can point that DOL are statistically significant at 1% level, with a negative signal, to model (1) and model (2), and 5% significance level at model (3), with reduction on significance level and subtle change on coefficient when compared with our Full Sample, which indicates a reduction of DOL role on performance at Industry sector. The HHI is statistically significant at 1% level with negative signal.

Comparing with Full Sample, we can observe that in the Industry sector the impact of market structure is perceived with higher intensity when compared with our Full Sample, at all models. At model 3, we find that is a moderating impact of market structure on DOL and ROIC relationship, with a positive effect on operational performance of firms’ that, at mean, present lower DOL when compared with firms that possess higher DOL operational performance. The significance of the combine moderation of DFL and HHI, with signal alteration, indicates that at mean firms with lower DOL performs poorly than firms with higher DOL when market structure tends to greater competition and the leverage on firms’ capital structure are observed - with similar result is found for the Services sector.

When we look at our HHI, we find that Industry shows the lowest HHI level, which indicates that it is the most competitive sector of our database, despite of still be classified as a oligopoly. However, it is also our major dataset with 7,435 firms. As we expect, DOL and DFL shows statistical significance and our results indicates that, at Industry sector, competition exert a disciplinary behavior when competition increase with firms’ with lower DOL performing better operationally. Furthermore, when DFL moderates the moderating impact of market structure, the combined moderation, our results indicates that there is an optimal combination of both leverage levels that lead firms to better operational result. The reason behind such results is that, ceteris paribus, firms may profit from positive economies of scale due to increase on firms’ margin by reducing unitary costs and that capital structure play an important role on firms’ performance. At Industry sector, we accept all research hypothesis, which indicates that DOL and DFL exert influence over ROIC and that market structure exert influence on firms’ cost decisions. However, DOL do not presents expected signal.

In Commerce sector, at model (1) and model (2), DOL is the only variable that presents statistical significance, at 1% level. Such results indicates that HHI do not exert an impact on the sector. However, in our model (3), when we aim to capture the moderating effect of HHI, there is a change in our results, with HHI and DFL presenting statistical significance at 1% level. The moderating effect variables and DOL, at model (3), no longer shows statistical significance. When, we consider the HHI level for Commerce sector, we observe a strong oligopolistic market (0.569) and low level of adherence of our models for both sectors. We expected such result especially for DOL, since the CVP approach roots on neoclassical microeconomic theory of markets in equilibrium, where markets are on perfect competition configuration (WICKRAMASINGHE and ALAWATTAGE, 2007). The results supports such claim when we highlight the result of our variable HHI*DFL, which presents statistical significance at 1% level and negative signal. Such result indicates that financial leverage may play important role on commerce firms’ strategy and that DFL may exert major impact on firms of commerce sector, concomitantly with market structure, being the leading leverage that impacts firm performance.

TABLE 5
Outputs subdivided by sectors

For commerce sector, our model (3), which is the most robust model, rejects H₁ and H₃. The second hypothesis is accepted under model (3) approach, which indicates a more prominent force of capital structure on firms’ performance than operational factors and we find a moderating effect of market structure over DFL, which opens a debate over most suitable metrics.

The Services sector show statistical significance for DOL in our model (1), at 1% level with negative signal, and in our model (3), at 5% with negative signal. Market structure do not shows statistical significance and DOL shows at our model (2), at 5% level and positive signal. The results regarding the signs of the variables are within our expectation and presents as the same signals observed for Industry sector and for our Full Sample. The inclusion of DFL in our regression captures the statistical significance of DOL found in model (1). However, the inclusion of our moderating variables in model (3) changes again the statistical significance. This result once again suggest that a composition between DOL and DFL, considering market configuration, is required to firms be able to perform operationally better than the sector mean.

When compared with our Full Sample, we can observe that HHI have more influence on firms’ operational performance due to the decrease of coefficient. In addition, the models output shows for Size statistical significance at 1% level and coefficient decrease, which indicates that larger firms have worse operational performance than mean in Services sector.

Services sector market configuration is classified as monopoly, with a HHI of (0.6690) and greater adherence of model 3 where previously expected in comparison with Commerce sector. The statistical significance level at 1% of both of our moderated variables, and signals that concurs with Industry sector results, indicates an influence of market structure and increase of DOL effectiveness when the metric is applied to markets that are at equilibrium. To services sector, the second hypothesis were rejected. The first and third hypothesis, which composes the main objective of the present research are accepted, with the expected signal.