The Impact of Organizational Capital on Cost Stickiness: Evidence from Japanese Firms

Abstract

This study examined the impact of organizational capital (OC) on the cost stickiness of Japanese firms and analyzed whether this effect varies with the magnitude of sales changes. Using 12,727 firm-year observations from Tokyo Stock Exchange-listed firms between 2007 and 2024, we estimated the economic value of OC by capitalizing and amortizing selling, general, and administrative (SG&A) expenses, then classified firms into high- and low-OC groups based on the median. Cost stickiness was then compared across groups using the basic, ABJ, and extended models, with robustness checks based on adjusted OC and two-way fixed effects models. The results indicate that high-OC firms exhibit stronger cost stickiness, while low-OC firms display weaker or insignificant stickiness. The effect depends on the magnitude of sales fluctuations: stickiness is pronounced under small changes but diminishes or disappears under larger shocks. Overall, this study contributes by highlighting the role of organizational resources in shaping asymmetric cost behavior, extending explanations beyond adjustment costs or managerial incentives, and providing novel evidence from Japan, where firms generally exhibit cost stickiness regardless of OC level, reflecting institutional and cultural contexts.

1. Introduction

Intangible assets, including organizational values and capabilities, have attracted increasing attention in recent decades due to fundamental changes in industrial structures. In the information and telecommunications industry, particularly among firms such as Google, Apple, Meta, and Amazon (GAMA) that dominate the global IT market, intangibles such as organizational capital (OC) have become a key driver of competitive advantage. Research in management accounting and finance has challenged the traditional assumption that costs change proportionally with activity levels. Instead, costs often behave asymmetrically—rising more when sales increase than they fall when sales decline —a phenomenon known as cost stickiness (Anderson et al., 2003). Cost stickiness has important implications for corporate performance, resource allocation, earnings forecasts, and investor decision-making, which makes it essential to investigate the conditions and determinants of this phenomenon. This asymmetry is typically attributed to resource adjustment and reacquisition costs and managers’ expectations about demand recovery (Banker & Byzalov, 2014).

Research on intangibles has traditionally examined intellectual capital (IC). Early studies emphasized categories such as human capital, organizational capital, customer capital, and elements such as brand value and corporate reputation (Kaplan & Norton, 1992; Edvinsson & Malone, 1997; Teece, 2000). Later research recognized organizational capital as a distinct and particularly important component of IC, reflecting the systems, processes, and structures that support knowledge integration and value creation (Dzinkowski, 2000; Guthrie et al., 2004; Subramaniam & Youndt, 2005; Martínez-Torres, 2006; Hsu & Fang, 2009). Building on this stream of research, the present study focuses on organizational capital and its role in shaping cost behavior. Prior studies have addressed the definition of OC (Lev, 2000), its measurement (Lev & Radhakrishnan, 2005; Lev, 2008), and its relationship with firm performance (Lev et al., 2009). Venieris et al. (2015) further demonstrate that U.S. firms with high OC exhibit stronger cost stickiness, while those with low OC may even display cost anti-stickiness. However, existing research remains largely concentrated in the U.S. and Europe, leaving little systematic investigation into Japanese companies.

Japan provides a distinctive institutional and cultural context in which to study the relationship between OC and cost stickiness. Japanese firms are characterized by long-term employment practices, seniority-based wage systems, and organizational routines that emphasize stability and harmony. Moreover, government policies and Japanese GAAP reporting systems reinforce resource retention during downturns. These features suggest that cost stickiness in Japanese firms may differ from patterns observed in U.S. firms, offering a unique opportunity to investigate how OC shapes asymmetric cost behavior under alternative institutional conditions. Studying Japanese firms, therefore, provides country-specific insights and helps to assess how institutional arrangements moderate the relationship.

This study makes several contributions to the literature. First, it highlights the role of organizational resources in shaping cost behavior, extending explanations of cost stickiness beyond adjustment costs and managerial expectations. Second, it provides novel evidence from Japan, showing that unlike U.S. firms, Japanese companies generally exhibit cost stickiness regardless of OC level, reflecting institutional and cultural characteristics. Third, by applying an adjusted measure of OC that excludes advertising expenses, the study enhances the accuracy of measuring organizational resources and their impact on cost behavior. It also explicitly characterizes how the relationship between OC and stickiness varies with the magnitude of sales changes. In doing so, it contributes to the cross-country literature by distinguishing the Japanese context from prior evidence in other countries and periods, thereby enhancing the generalizability of cost stickiness research.

The remainder of this paper is structured as follows: Section 2 reviews prior studies and develops the hypotheses, Section 3 describes the methodology and data, Section 4 presents results, and Section 5 concludes.

2. Literature Review and Hypotheses Development

2.1. Literature Review

2.1.1. Organizational Capital

Prior studies on organizational capital have explored its definitions, measurement approaches, and implications for corporate performance. OC comprises the knowledge and capabilities that constitute a system that skillfully combines employee capacity with physical capital. Evenson and Westphal (1995) further divide OC into three categories: manufacturing capacity, investment capacity, and inventive capacity.

Lev and Radhakrishnan (2005) quantitatively measure OC using data on firms’ sales and selling, general and administrative (SG&A) expenses, number of employees as a proxy for labor, and tangible fixed assets as a proxy for capital. They find that OC has a positive relationship with sales growth. Building on this, Lev et al. (2009) argue that OC, closely associated with SG&A expenses, is the most important intangible embedded in a firm’s organizational structure and technological infrastructure. It facilitates the flow of knowledge to improve operational efficiency and represents the unique structural and organizational design and business processes that create a sustainable competitive advantage. Investments in such OC are expected to enable firms to achieve greater operational efficiency than their competitors.

Early research on intellectual capital emphasized components such as human capital, organizational capital, customer or relational capital, and elements such as brand value and corporate reputation (Kaplan & Norton, 1992; Edvinsson & Malone, 1997; Teece, 2000). Later studies increasingly recognized organizational capital as a distinct and particularly important dimension of IC, reflecting the systems, processes, and structures that support knowledge integration and value creation (Dzinkowski, 2000; Guthrie et al., 2004; Subramaniam & Youndt, 2005; Martínez-Torres, 2006; Hsu & Fang, 2009).

In Europe, and especially in Scandinavian countries, scholars also made important contributions to the development of IC theory and reporting practices. Sveiby (1997) introduced the intangible assets monitor, which provided one of the earliest systematic frameworks for identifying and measuring knowledge-based assets in organizations. Roos et al. (1997) developed the IC navigator, highlighting the strategic role of IC in creating and sustaining competitive advantage. Petty and Guthrie (2000) offered a comprehensive review of IC research, measurement, and reporting, which helped to consolidate the literature and laid the foundation for further policy and corporate practice in IC disclosure, particularly in Europe and Australasia. These contributions broadened the scope of IC research beyond the Anglo-American context and reinforced the recognition of OC as a key element of IC.

Building on this foundation, Martín-de-Castro et al. (2011) emphasized the role of OC in providing organizational cohesion to business processes and activities. Similarly, Hosomi (2014) highlighted its relevance in the Japanese context, defining OC as IC produced by organizational forms that include a firm’s unique structure, standardized business processes, and organizational culture.

2.1.2. Cost Stickiness

Since OC is closely associated with adjustment costs and managerial decisions, several studies have linked it to cost behavior, particularly cost stickiness. Venieris et al. (2015) measure OC using the methodology of Lev et al. (2009) and find that, in the U.S., firms with high OC exhibit cost stickiness, whereas those with low OC exhibit cost anti-stickiness. When sales decline, firms with more intangibles increase slack in unused resources to a greater extent than firms with fewer intangibles. This suggests that the higher the investment in intangibles, the greater the adjustment costs, and the more optimistic managers become about future sales growth absorbing such slack. It was also noted that the magnitude of sales changes affects the relationship between OC and cost stickiness.

Prior studies on cost stickiness have also examined the phenomenon itself, its formative factors, and its implications for profits. Discussions on cost stickiness can be traced back to the critique of proportional cost behavior. Noreen and Soderstrom (1994, 1997) tested this assumption and rejected the notion of a strictly proportional relationship between costs and cost drivers. Their findings and methodology laid the foundation for subsequent research on asymmetric cost behavior.

Following this line of inquiry, Anderson et al. (2003), using U.S. firm data, provide evidence of cost stickiness, where the absolute rate of decrease in SG&A expenses when sales decrease is smaller than the absolute rate of increase in costs when activity rises. By contrast, Balakrishnan et al. (2004) and Cannon (2014) provide evidence of cost anti-stickiness, where the absolute rate of decrease in SG&A expenses when sales fall exceeds the absolute rate of increase in costs when activity rises. Additionally, Hosomi and Nagasawa (2018) provide empirical evidence of cost anti-stickiness in Japanese local public corporations and show that the degree of cost stickiness varies across industries.

Regarding the factors underlying cost stickiness, managers with optimistic expectations may temporarily retain underutilized management resources (Banker & Byzalov, 2014; Banker et al., 2013). This decision occurs if they expect sales declines to be temporary and estimate that the costs of retaining unused resources are lower than the costs of adjustment (e.g., selling machinery and equipment versus repurchasing them, or laying off employees versus rehiring). These studies highlight that managerial decisions to maintain underutilized resources are shaped by expectations about future sales and by the trade-off between adjustment and reacquisition costs (Banker et al., 2013, 2014, 2025).

Weiss (2010) shows that firms with high-cost stickiness are less accurate in analysts’ earnings forecasts than those with low-cost stickiness. Similarly, Kama and Weiss (2013) find that cost stickiness affects analysts’ coverage and that investors incorporate cost stickiness when reacting to earnings surprises.

Regarding the role of resource adjustment costs in shaping cost behavior, prior studies have provided evidence from both operational and financial perspectives. Dierynck et al. (2012) show that managerial incentives to meet or beat the zero-earnings benchmark lead to more symmetric labor cost adjustments in firms reporting small profits, while firms with healthier profits avoid excessive layoffs and instead adjust working hours. By contrast, He et al. (2020) find that firms with higher resource adjustment costs and stickier costs pay lower dividends, as they are less able to sustain high payout levels in the future. They further demonstrate that this negative relation is driven by labor adjustment costs.

Banker et al. (2025) provide evidence that firms pursuing a differentiation strategy exhibit greater cost stickiness, on average, than firms pursuing a cost leadership strategy. This relationship is moderated by managers’ optimistic or pessimistic expectations about future sales. However, prior studies on cost stickiness have not clarified the role of OC in Japanese firms.

In summary, prior studies have clarified the definitions, measurement, and performance implications of OC and demonstrated its role as an important component of intellectual capital. Studies on cost stickiness have established its determinants and consequences and have begun to link it to OC, showing that firms with higher OC tend to exhibit greater stickiness. Moreover, prior findings on the moderating role of sales-change magnitude are largely drawn from non-Japanese settings. However, systematic evidence on how OC influences cost stickiness in Japanese firms remains scarce. This gap is important because Japan’s institutional and cultural context is characterized by long-term employment practices, seniority-based wage systems, and organizational routines that emphasize stability, which may condition the mechanisms through which OC affects cost behavior. Accordingly, this study aims to extend the cross-country literature by investigating the relationship between OC and cost stickiness in Japan and to clarify how institutional arrangements shape asymmetric cost behavior.

2.2. Theoretical Background and Hypotheses Development

2.2.1. Impact of Organizational Capital on Cost Stickiness

Building on the prior literature reviewed above, this section outlines the theoretical background linking organizational capital (OC) to cost stickiness and develops the study’s hypotheses. Prior research has shown that OC shapes firms’ ability to retain and utilize resources, which in turn affects how costs respond to sales fluctuations. Drawing on these insights, the following subsections present hypotheses regarding the expected effects of OC on cost behavior.

Firms with higher OC are expected to have longer cost adjustment periods, and managers are expected to be more optimistic about future sales projections (Chen et al., 2019). Therefore, when sales decline, firms with higher levels of OC will have a greater slack of underutilized resources than those with lower levels of OC (Venieris et al., 2015). The increase in the slack of underutilized resources affects the behavior of SG&A expenses. Specifically, in firms with higher levels of OC, managers have a longer-term outlook for business performance and, as a result, make decisions to maintain SG&A expenses even in the face of declining sales, which may cause cost stickiness (Balakrishnan & Gruca, 2008).

Additionally, OC is associated with SG&A expenses (Lev et al., 2009). However, since advertising expenses in SG&A are associated with the relationship between the firm and its customers, they are also categorized as relationship capital and customer capital (Venieris et al., 2015). To measure OC more precisely, we obtain adjusted SG&A expenses by excluding advertising expenses from SG&A expenses. Based on adjusted SG&A expenses, we define adjusted organizational capital (AOC) as OC that excludes the bias introduced by relational and customer capital.

Therefore, OC is expected to amplify cost stickiness through the mechanism of resource retention and managerial optimism, making high-OC firms more prone to asymmetric cost behavior. On the basis of the above considerations, this study establishes the following hypotheses:

Hypothesis 1 (H1). 

Firms with high organizational capital exhibit greater cost stickiness than firms with low organizational capital.

2.2.2. Impact of Organizational Capital on Cost Stickiness Depending on the Magnitude of Sales Changes

Prior research also suggests that the impact of sales changes on cost behavior depends on their magnitude. Ciftci and Zoubi (2019) reported that cost stickiness is more pronounced when sales changes are relatively small (≤10%) but diminishes or even disappears when changes are large (10–50%). Similarly, Venieris et al. (2015) found that in the U.S., firms with high OC exhibit cost stickiness, while those with low OC display cost anti-stickiness. When sales fluctuations are modest, OC strengthens cost stickiness, particularly in high capital firms, whereas the effect is weaker in low capital firms. Conversely, when fluctuations are substantial, both groups adjust resources more aggressively, leading to reduced or insignificant stickiness.

Thus, the magnitude of sales shocks conditions the influence of OC on cost behavior: asymmetry is heightened under smaller changes but weakens or vanishes when changes are larger. Accordingly, this study proposes the following hypotheses:

Hypothesis 2 (H2). 

When the absolute rate of sales change is small, firms with high organizational capital exhibit greater cost stickiness than those with low organizational capital; when the rate is large, cost stickiness weakens or disappears.

3. Method

3.1. Model Design

3.1.1. Basic Model

In prior studies on the estimation of cost stickiness, a three-stage approach has been described, consisting of the basic model, the ABJ model (Anderson et al., 2003), and the extended model (Venieris et al., 2015).

The estimation of cost stickiness begins with the following basic model (1). In this specification, SG&A expenses increase by β₁% for a 1% increase in sales of firm i at the end of fiscal year t, while for a 1% decrease in sales, SG&A expenses decrease by (β₁ + β₂)%. Here, $D_{i,t}$ is a dummy variable that equals 1 if sales decreased at the end of fiscal year t, and 0 otherwise. Because the model is specified in a log–log form, the coefficients are interpreted as elasticities. Accordingly, β₂ captures the degree of cost stickiness, reflecting the asymmetry in cost response between sales increases and decreases.

In addition to the basic model, we also estimate using Adjusted SG&A, as specified in model (2), where SG&A expenses are adjusted by excluding advertising expenses. This adjustment is made because advertising expenses primarily reflect the relationship between the firm and its customers and are therefore less directly related to organizational capital. By excluding these expenses, the adjusted measure of SG&A is expected to more accurately reflect the portion of costs associated with internal organizational resources.

$$\log \left({\displaystyle \frac{{SG\&A}_{i,t}}{{SG\&A}_{i,t-1}}}\right)={\beta }_0+{\beta }_1\times \log \left({\displaystyle \frac{{\mathit{Sales}}_{i,t}}{{\mathit{Sales}}_{i,t-1}}}\right)+{\beta }_2\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)+{\epsilon }_{i,t}$$

(1)

$$\log \left({\displaystyle \frac{{Adjusted\_SG\&A}_{i,t}}{Adjusted\_{SG\&A}_{i,t-1}}}\right)={\beta }_0+{\beta }_1\times \log \left({\displaystyle \frac{{\mathit{Sales}}_{i,t}}{{\mathit{Sales}}_{i,t-1}}}\right)+{\beta }_2\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)+{\epsilon }_{i,t}$$

(2)

3.1.2. ABJ Model

The ABJ model extends the basic specification by incorporating additional factors that may influence cost stickiness. Specifically, the model includes physical asset intensity, management expectations about future sales, and macroeconomic conditions. Asset intensity is proxied by the ratio of total assets to sales, and management expectations are proxied by a dummy variable indicating two consecutive periods of sales decline. Following Anderson et al. (2003), who used gross national product (GNP), this study uses gross national income (GNI) instead, because Japanese national accounts report GNI. These two indicators are conceptually similar and highly correlated. ${GNI}_t$ denotes gross national income in year t, which is included to account for macroeconomic conditions and is interacted with sales decreases.

The ABJ specification is estimated using models (3) and (4), which extend the basic specification by incorporating interaction terms that allow the degree of cost stickiness to vary with firm- and macro-level characteristics. As in the basic model, SG&A increases by β₁% for a 1% increase in sales, while for a 1% decrease in sales, it decreases by (β₁ + β₂)%. The additional interaction terms identify how contextual factors moderate the asymmetric cost response. β₃ is the coefficient of the interaction between sales decreases and asset intensity. A positive and significant β₃ would imply that firms with higher asset intensity experience stronger cost stickiness when sales decline. β₄ is the coefficient of the interaction between sales decreases and the dummy variable ${Ds}_{i,t}$. A positive and significant β₄ would suggest that managers’ expectations about the persistence of sales declines influence cost adjustments. β₅ is the coefficient of the interaction between sales decreases and macroeconomic conditions. This captures whether broader economic conditions exacerbate or mitigate cost stickiness in response to sales decreases.

An alternative specification is also estimated using Adjusted SG&A, as shown in model (4), where advertising expenses are excluded to obtain a cost measure more directly related to organizational capital.

$$\begin{array}{cc}\log \left({\displaystyle \frac{{SG\&A}_{i,t}}{{SG\&A}_{i,t-1}}}\right)\hfill & ={\beta }_0+{\beta }_1\times \log \left({\displaystyle \frac{{\mathit{Sales}}_{i,t}}{{\mathit{Sales}}_{i,t-1}}}\right)+{\beta }_2\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\hfill \\ & +{\beta }_3\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)\hfill \\ & +{\beta }_4\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}\hfill \\ & +{\beta }_5\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_t+{\epsilon }_{i,t}\hfill \end{array}$$

(3)

$$\begin{array}{c}\log \left({\displaystyle \frac{Adjusted\_{SG\&A}_{i,t}}{Adjusted\_{SG\&A}_{i,t-1}}}\right)\\ ={\beta }_0+{\beta }_1\times \log \left({\displaystyle \frac{{\mathit{Sales}}_{i,t}}{{\mathit{Sales}}_{i,t-1}}}\right)+{\beta }_2\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\\ +{\beta }_3\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)\\ +{\beta }_4\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}\\ +{\beta }_5\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_t+{\epsilon }_{i,t}\end{array}$$

(4)

3.1.3. Extended Model

Building on the ABJ specification, the extended model further incorporates free cash flow (FCF) as an additional control variable to capture potential agency problems (Chen et al., 2012). Thus, in addition to physical asset intensity, management expectations about future sales, and macroeconomic conditions, this model includes FCF as a proxy for agency costs.

The extended specification is estimated using models (5) and (6). As in the basic and ABJ specifications, SG&A increases by β₁% for a 1% increase in sales, while for a 1% decrease in sales, it decreases by (β₁ + β₂)%. The additional interaction terms allow this asymmetry to vary with contextual factors. In particular, β₆ is the coefficient of the interaction between sales decreases and free cash flow. A positive and significant β₆ would imply that firms with higher levels of free cash flow experience stronger cost stickiness when sales decline, consistent with the agency problem perspective. β₇ to β₁₀ represent the main effects of the control variables, such as asset intensity, sales persistence, macroeconomic conditions, and free cash flow. In addition, β₁₁ and β₁₂ control for financial leverage and firm size, respectively. Leverage captures the effect of capital structure on cost behavior, as more highly leveraged firms may be under pressure to adjust costs more flexibly. Size reflects economies of scale and organizational complexity, which may also influence the degree of cost stickiness (Hosomi & Ge, 2025).

An alternative specification is also estimated using Adjusted SG&A, as shown in model (6), where advertising expenses are excluded to obtain a cost measure more directly related to organizational capital.

$$\begin{array}{cc}\log \left({\displaystyle \frac{{SG\&A}_{i,t}}{{SG\&A}_{i,t-1}}}\right)\hfill & ={\beta }_0+{\beta }_1\times \log \left({\displaystyle \frac{{\mathit{Sales}}_{i,t}}{{\mathit{Sales}}_{i,t-1}}}\right)+{\beta }_2\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\hfill \\ & +{\beta }_3\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)\hfill \\ & +{\beta }_4\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}\hfill \\ & +{\beta }_5\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_t\hfill \\ & +{\beta }_6\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}+{\beta }_7\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)\hfill \\ & +{\beta }_8\times {Ds}_{i,t}+{\beta }_9\times {GNI}_t+{\beta }_{10}\times {FCF}_{i,t}+{\beta }_{11}\times {Leverage}_{i,t}\hfill \\ & +{\beta }_{12}\times {Size}_{i,t}+{\epsilon }_{i,t}\hfill \end{array}$$

(5)

$$\begin{array}{c}\log \left({\displaystyle \frac{{Adjusted\_SG\&A}_{i,t}}{{Adjusted\_SG\&A}_{i,t-1}}}\right)\\ ={\beta }_0+{\beta }_1\times \log \left({\displaystyle \frac{{\mathit{Sales}}_{i,t}}{{\mathit{Sales}}_{i,t-1}}}\right)+{\beta }_2\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\\ +{\beta }_3\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)\\ +{\beta }_4\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}\\ +{\beta }_5\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_t\\ +{\beta }_6\times D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}+{\beta }_7\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)\\ +{\beta }_8\times {Ds}_{i,t}+{\beta }_9\times {GNI}_t+{\beta }_{10}\times {FCF}_{i,t}+{\beta }_{11}\times {Leverage}_{i,t}\\ +{\beta }_{12}\times {Size}_{i,t}+{\epsilon }_{i,t}\end{array}$$

(6)

In summary, the empirical analysis relies on a three-stage modeling framework. The basic model establishes the benchmark estimation of cost stickiness. The ABJ model extends this framework by incorporating firm-level and macroeconomic factors that may affect cost adjustment. The extended model further accounts for the potential agency problem by including free cash flow as an additional control variable. Taken together, these specifications provide a comprehensive basis for examining the determinants of cost stickiness and assessing the moderating role of organizational capital. The variables above, including the interaction variables, are presented in

Table 1. Analysis variables (Estimation of cost stickiness).

Variable	Definition
${SG\&A}_{i,t}$	SG&A expenses of firm i at the end of fiscal year t.
$Adjusted\_{SG\&A}_{i,t}$	SG&A expenses excluding advertising expenses of firm i at the end of fiscal year t.
${Sales}_{i,t}$	Sales of firm i at the end of fiscal year t.
$D_{i,t}$	Dummy variable equal to 1 if sales of firm i decreased at the end of fiscal year t, and 0 otherwise.
${Ds}_{i,t}$	Dummy variable equal to 1 if sales of firm i decreased for two successive years at the end of fiscal year t, and 0 otherwise.
${Assets}_{i,t}$	Total assets of firm i at the end of fiscal year t.
${GNI}_t$	Growth rate of gross national income at the end of fiscal year t.
${FCF}_{i,t}$	Free cash flow of firm i at the end of fiscal year t.
${Leverage}_{i,t}$	Ratio of total debt to shareholders’ equity of firm i at the end of fiscal year t.
${Size}_{i,t}$	Natural logarithm of total assets of firm i at the end of fiscal year t.
Interaction Variable
$D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	Interaction of sales decreases with SG&A expenses. Represents cost stickiness in response to sales decreases.
$D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$	Interaction of sales decreases with asset intensity. Represents the moderating effect of asset intensity on cost stickiness.
$D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$	Interaction of sales decreases with persistent sales decline dummy. Represents the moderating effect of persistent sales declines (management expectations) on cost stickiness.
$D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_t$	Interaction of sales decreases with GNI. Represents the moderating effect of GNI (macroeconomic conditions) on cost stickiness.
$D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$	Interaction of sales decreases with free cash flow. Represents the moderating effect of free cash flow (agency problems) on cost stickiness.

.

3.2. Estimation of Organizational Capital

Prior studies have taken three approaches to estimating organizational capital: input-based, output-based, and a combination of both approaches. Specifically, the input-based approach focuses on SG&A and other costs incurred to develop and maintain OC (Lev, 2008; Lev et al., 2009). The output-based approach captures the contribution of OC to sales and its impact on controlling operating expenses (Martín-de-Castro et al., 2011). The approach of combining both approaches is a more integrated means of estimating the economic value of OC. Following Venieris et al. (2015), this study combines both approaches to estimate the economic value of OC.

Based on the assumption that the SG&A expenses of firms that make inputs to OC are higher than those of firms that do not, we capitalize one year of SG&A expenses as inputs to OC and amortize them over three years from the perspective of capitalization rather than expensing. The capitalization and amortization of SG&A expenses not only allows us to evaluate the time trend of inputs to OC but also reduces the impact of one-time changes in SG&A expenses.

$${SG\&A\_Capitalised}_{i,t}={SG\&A}_{i,t}+{\displaystyle \frac{2}{3}}{SG\&A}_{i,t-1}+{\displaystyle \frac{1}{3}}{SG\&A}_{i,t-2}$$

(7)

Recognizing that OC is a system that efficiently combines physical and human capital to build or maintain a firm’s competitive advantage (Lev et al., 2009), we use SG&A expenses as a proxy variable for OC, the number of employees as a proxy variable for human capital, and tangible fixed assets as a proxy variable for physical capital, with the rate of change in each as the rate of change in sales and coefficients of the sales.

$$\log {\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}={\alpha }_{0,i,t}+{\alpha }_{1,i,t}\log {\displaystyle \frac{{SG\&A\_Capitalised}_{i,t}}{{SG\&A\_Capitalised}_{i,t-1}}}+{\alpha }_{2,i,t}\log {\displaystyle \frac{{Emp}_{i,t}}{{Emp}_{i,t-1}}}+{\alpha }_{3,i,t}\log {\displaystyle \frac{{PPE}_{i,t}}{{PPE}_{i,t-1}}}+\log {\displaystyle \frac{{\epsilon }_{i,t}}{{\epsilon }_{i,t-1}}}$$

(8)

According to the Cobb–Douglas type production function, ${\alpha }_{0,i,t}$ is the scale factor, ${\alpha }_{2,i,t}$ is the labor allocation rate, and ${\alpha }_{3,i,t}$ is the capital allocation rate (Hall, 2001).

$${PreSales}_{i,t}={\alpha }_{0,i,t}{Emp}_{i,t}^{{\alpha }_{2,i,t}}{PPE}_{i,t}^{{\alpha }_{3,i,t}}{\epsilon }_{i,t}$$

(9)

The difference between a firm’s actual and projected sales estimates the contribution of OC to sales.

$${AbSales}_{i,t}={Sales}_{i,t}-{PreSales}_{i,t}$$

(10)

Similarly, to approaches (8), (9), and (10), which estimate the excess contribution of OC to sales, we estimate the degree to which OC is restricted to operating expenses.

$$\log {\displaystyle \frac{{Cost}_{i,t}}{{Cost}_{i,t-1}}}={\alpha }_{4,i,t}+{\alpha }_{5,i,t}\log {\displaystyle \frac{{SG\&A\_Capitalised}_{i,t}}{{SG\&A\_Capitalised}_{i,t-1}}}+{\alpha }_{6,i,t}\log {\displaystyle \frac{{Emp}_{i,t}}{{Emp}_{i,t-1}}}+{\alpha }_{7,i,t}\log {\displaystyle \frac{{PPE}_{i,t}}{{PPE}_{i,t-1}}}+\log {\displaystyle \frac{{\epsilon }_{i,t}}{{\epsilon }_{i,t-1}}}$$

(11)

$${PreCost}_{i,t}={\alpha }_{4,i,t}{Emp}_{i,t}^{{\alpha }_{6,i,t}}{PPE}_{i,t}^{{\alpha }_{7,i,t}}{\epsilon }_{i,t}$$

(12)

$${AbCost}_{i,t}={Cost}_{i,t}-{PreCost}_{i,t}$$

(13)

The economic value of OC is the sum of its contribution to sales and its suppression of operating expenses.

$${AbProfit}_{i,t}={AbSales}_{i,t}+{AbCost}_{i,t}$$

(14)

Since OC is assumed to be a depreciable asset with a useful life of five years, its output is capitalized and amortized over that period. To estimate its economic value, the amount is scaled with the total assets at the end of the fiscal year (Lev et al., 2009).

$${OC}_{i,t}={\displaystyle \frac{{\sum }_4^{k=0}(1-0.2k){AbProfit}_{i,t-k}}{{Assets}_{i,t}}}$$

(15)

We estimate adjusted SG&A expenses because advertising expenses are related to the relationship between the firm and its customers and are not as related to OC. Then, we re-estimate adjusted organizational capital, $Adjusted\_{OC}_{i,t}$ (Venieris et al., 2015).

$$Adjusted\_{SG\&A}_{i,t}={SG\&A}_{i,t}-{Advertise}_{i,t}$$

(16)

The variables above are derived from the capitalization and amortization of SG&A expenses and the Cobb–Douglas-type production function, as specified in Equations (7)–(16), and are presented in

Table 2. Analysis variables (Estimation of organizational capital).

Variable	Definition
${SG\&A\_Capitalised}_{i,t}$	Capitalized and amortized SG&A expenses of firm i at the end of fiscal year t. Represents inputs to organizational capital through accumulated SG&A investments.
${PreSales}_{i,t}$	Projected sales of firm i at the end of fiscal year t.
${AbSales}_{i,t}$	Contribution of OC to sales of firm i at the end of fiscal year t. Represents the excess sales attributable to organizational capital.
${Cost}_{i,t}$	Operating expenses of firm i at the end of fiscal year t.
${PreCost}_{i,t}$	Projected operating expenses of firm i at the end of fiscal year t.
${AbCost}_{i,t}$	Contribution of OC to cost containment of firm i at the end of fiscal year t. Represents the extent to which OC reduces operating expenses.
${AbProfit}_{i,t}$	Aggregate contribution of OC of firm i at the end of fiscal year t. Represents the combined contribution of OC to sales growth and cost savings.
${OC}_{i,t}$	Economic value of OC of firm i at the end of fiscal year t.
$Adjusted\_{OC}_{i,t}$	Economic value of adjusted OC of firm i at the end of fiscal year t.
${Emp}_{i,t}$	Number of employees of firm i at the end of fiscal year t.
${PPE}_{i,t}$	Property, plant, and equipment of firm i at the end of fiscal year t.
${Advertise}_{i,t}$	Advertising expenses of firm i at the end of fiscal year t.

.

To classify firms into high- and low-OC groups, we adopt a median-split approach following Venieris et al. (2015). Firms with OC values above the sample median are categorized as the high-OC group, while those below the median are classified as the low-OC group. This classification provides a clear and consistent benchmark, avoiding the potential ambiguity of quartile-based classifications.

3.3. Samples

3.3.1. Sample Selection

This study tests the hypotheses using financial data from Japanese firms listed on the Prime and Standard sections of the Tokyo Stock Exchange as of the end of March 2025. The period of review ranged from the fiscal year ending 31 March 2008, through the fiscal year ending 31 March 2025. The sample excludes banking, insurance, securities and other financial services industries. For the purposes of this analysis, we used only consolidated financial data for firms whose fiscal year ends in March and ends in 12 months under Japanese Generally Accepted Accounting Principles. Financial data were obtained from the Nikkei NEEDS-Financial QUEST database. The initial conditions for data collection are summarized in

Table 3. Sample selection.

Item	Condition
Target companies	As of the end of March 2025, Tokyo Stock Exchange-listed companies excluding those classified as banks, securities firms, insurance companies, and other financial institutions based on the TSE industry classification.
Periods	18 fiscal years from March 2008 through March 2025 (FY2007 through FY2024)
Data	Consolidated Financial Data (Japanese GAAP)
Settlement month	March
Number of months for settlement of accounts	12 months

.

3.3.2. Sample Processing

This study examined firms meeting the following criteria: (1) listed on the Tokyo Stock Exchange, adopting Japanese accounting standards, with complete financial data from FY2007 to FY2024, a fiscal year ending in March, and a 12-month reporting period; and (2) having sufficient quarterly data to calculate OC using the Lev model (Lev, 2008; Lev et al., 2009).

To improve robustness and limit the influence of extreme values, we identified outliers in continuous variables using the 3-sigma rule, defining them as values more than three standard deviations above or below the mean. These observations were excluded from the dataset, as summarized in

Table 4. Sample processing.

Process	Number
① 3239 Tokyo Stock Exchange-listed companies as of 31 March 2025, excluding banking, securities, insurance, and other financial industries classified according to the Tokyo Stock Exchange industry classification for the 18 fiscal years from March 2008 to March 2025.	58,032
② After ${SG\&A\_Capitalised}_{i,t}$ calculation, the period under analysis is 16 fiscal years from March 2010 to March 2025.	51,824
③ After $\log {\displaystyle \frac{{SG\&A\_Cap}_{i,t}}{{SG\&A\_Cap}_{i,t-1}}}$ calculation, the period under analysis is 15 fiscal years from March 2011 to March 2025.	48,585
④ After ${OC}_{i,t}$ calculation, the period under analysis is 11 fiscal years from March 2015 to March 2025.	35,629
⑤ After meeting the prescribed conditions for analysis and eliminating missing and abnormal values.	12,727

.

Descriptive statistics of the sample are presented in

Table 5. Descriptive statistics.

Variable	n	Mean	sd	Min	Median	Max
$\log \left({\displaystyle \frac{{SG\&A}_{i,t}}{{SG\&A}_{i,t-1}}}\right)$	12,727	0.019	0.030	−0.128	0.017	0.160
2. $\log \left({\displaystyle \frac{{Adjusted\_SG\&A}_{i,t}}{Adjusted\_{SG\&A}_{i,t-1}}}\right)$	2468	0.018	0.029	−0.093	0.016	0.124
3. $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	12,727	0.023	0.035	−0.076	0.018	0.173
4. $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	12,727	−0.004	0.009	−0.076	0	0
5. $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$	12,727	0.026	0.197	−0.582	0.035	0.658
6. ${Ds}_{i,t}$	12,727	0.089	0.285	0	0	1
7. ${GNI}_{i,t}$	12,727	0.996	1.528	−3.2	1.3	3.3
8. ${FCF}_{i,t}$	12,727	3124.895	11,999.325	−84,183	1011	91,190
9. ${Leverage}_{i,t}$	12,727	2.166	1.070	1.05	1.86	14.21
10. ${Size}_{i,t}$	12,727	4.804	0.601	3.112	4.753	6.634
11. ${OC}_{i,t}$	12,727	−0.400	6.234	−42.885	0.874	17.782
12. $Adjusted\_{OC}_{i,t}$	12,727	1.808	4.609	−15.520	2.315	18.326

. Comparison of sample by year and industry is presented in Appendix A,

Table A1. Comparison of sample by year.

Fiscal Year	Number of Data	Composition Ratio (%)
2015	1247	9.80
2016	1323	10.40
2017	1410	11.08
2018	1420	11.16
2019	1294	10.17
2020	694	5.45
2021	1261	9.91
2022	1377	10.82
2023	1329	10.44
2024	1372	10.78
Total amount	12727	100.00

and

Table A2. Comparison of sample by industry.

Industry	Number of Data	Composition Ratio (%)
Fishery, Agriculture & Forestry	6	0.35
Mining	4	0.24
Construction	99	5.85
Foods	71	4.20
Textile & Apparels	27	1.60
Pulp & Paper	17	1.01
Chemicals	143	8.46
Pharmaceutical	18	1.06
Oil & Coal Products	6	0.35
Rubber Products	13	0.77
Glass & Ceramics Products	34	2.01
Iron & Steel	32	1.89
Nonferrous Metals	23	1.36
Metal Products	51	3.02
Machinery	142	8.40
Electric Appliances	150	8.87
Transport Equipment	68	4.02
Precision Instruments	27	1.60
Other Products	44	2.60
Electric Power & Gas	12	0.71
Land Transportation	51	3.02
Marine Transportation	10	0.59
Air Transportation	2	0.12
Warehousing and Harbor transportation	26	1.54
Information & Communication	137	8.10
Wholesale Trade	178	10.53
Retail Trade	110	6.51
Real Estate	38	2.25
Services	152	8.99
Total amount	1691	100.00

.

4. Findings and Discussion

4.1. The Impact of Organizational Capital on Cost Stickiness

To examine the relationship between OC and cost stickiness, we first calculated OC from SG&A expenses (Lev et al., 2009). To address potential endogeneity concerns, particularly reverse causality, firms were classified into high- and low-OC groups based on the median value of lagged OC (t − 1). The median-split approach follows Venieris et al. (2015) (see Section 3.2). Specifically, firms with lagged OC values above the sample median were assigned to the high-OC group, while those below the median were assigned to the low-OC group. Cost stickiness for each group was subsequently estimated and compared based on Anderson et al. (2003).

Before conducting the estimations, we performed the RESET test to verify the linearity of the pooled model specification (see

Table A3. Linearity test (RESET test).

		Model	Statistic	p-Value	df1	df2
SG&A cost stickiness by organizational capital	Low-OC group	Basic (1)	36.66	1.47 × 10−16	2	6358
		ABJ (3)	34.15	1.77 × 10−15	2	6355
		Extended (5)	23.34	7.93 × 10−11	2	6288
	High-OC group	Basic (1)	12.61	3.43 × 10−6	2	6359
		ABJ (3)	12.85	2.70 × 10−6	2	6356
		Extended (5)	13.32	1.69 × 10−6	2	6199
Adjusted SG&A cost stickiness by organizational capital	Low-OC group	Basic (2)	8.61	1.95 × 10−4	2	1181
		ABJ (4)	8.22	2.85 × 10−4	2	1178
		Extended (6)	6.38	1.75 × 10−3	2	1154
	High-OC group	Basic (2)	7.05	8.98 × 10−4	2	1282
		ABJ (4)	5.96	2.65 × 10−3	2	1279
		Extended (6)	3.59	2.78 × 10−2	2	1260
SG&A cost stickiness by adjusted organizational capital	Low-OC group	Basic (1)	27.99	7.87 × 10−13	2	6358
		ABJ (3)	26.29	4.27 × 10−12	2	6355
		Extended (5)	15.17	2.68 × 10−7	2	6287
	High-OC group	Basic (1)	20.33	1.58 × 10−9	2	6359
		ABJ (3)	19.77	2.77 × 10−9	2	6356
		Extended (5)	18.17	1.35 × 10−8	2	6200
Adjusted SG&A cost stickiness by adjusted organizational capital	Low-OC group	Basic (2)	6.74	1.23 × 10−3	2	1216
		ABJ (4)	6.76	1.21 × 10−3	2	1213
		Extended (6)	3.76	2.36 × 10−2	2	1188
	High-OC group	Basic (2)	8.83	1.56 × 10−4	2	1247
		ABJ (4)	8.49	2.18 × 10−4	2	1244
		Extended (6)	5.99	2.58 × 10−3	2	1226
SG&A cost stickiness by organizational capital under small sales changes	Low-OC group	Basic (1)	1.02	3.62 × 10−1	2	4374
		ABJ (3)	1.68	1.86 × 10−1	2	4371
		Extended (5)	1.84	1.58 × 10−1	2	4334
	High-OC group	Basic (1)	0.44	6.47 × 10−1	2	4375
		ABJ (3)	0.13	8.76 × 10−1	2	4372
		Extended (5)	0.66	5.16 × 10−1	2	4269
SG&A cost stickiness by organizational capital under large sales changes	Low-OC group	Basic (1)	3.99	1.9 × 10−2	2	1936
		ABJ (3)	1.51	2.2 × 10−1	2	1933
		Extended (5)	0.56	5.7 × 10−1	2	1897
	High-OC group	Basic (1)	6.28	1.9 × 10−3	2	1937
		ABJ (3)	1.66	1.9 × 10−1	2	1934
		Extended (5)	2.29	1.0 × 10−1	2	1890
Adjusted SG&A cost stickiness by organizational capital under small sales changes	Low-OC group	Basic (2)	3.03	4.88 × 10−2	2	806
		ABJ (4)	1.09	3.37 × 10−1	2	803
		Extended (6)	0.46	6.32 × 10−1	2	787
	High-OC group	Basic (2)	0.06	9.45 × 10−1	2	920
		ABJ (4)	0.07	9.30 × 10−1	2	917
		Extended (6)	0.12	8.85 × 10−1	2	899
Adjusted SG&A cost stickiness by organizational capital under large sales changes	Low-OC group	Basic (2)	0.16	8.5 × 10−1	2	372
		ABJ (4)	0.45	6.4 × 10−1	2	369
		Extended (6)	1.40	2.5 × 10−1	2	357
	High-OC group	Basic (2)	1.01	3.7 × 10−1	2	341
		ABJ (4)	0.74	4.8 × 10−1	2	338
		Extended (6)	0.15	8.6 × 10−1	2	330
SG&A cost stickiness by adjusted organizational capital under small sales changes	Low-OC group	Basic (1)	0.81	4.47 × 10−1	2	4374
		ABJ (3)	1.78	1.69 × 10−1	2	4371
		Extended (5)	0.76	4.69 × 10−1	2	4332
	High-OC group	Basic (1)	1.03	3.57 × 10−1	2	4375
		ABJ (3)	0.38	6.83 × 10−1	2	4372
		Extended (5)	0.53	5.89 × 10−1	2	4271
SG&A cost stickiness by adjusted organizational capital under large sales changes	Low-OC group	Basic (1)	2.80	6.1 × 10−2	2	1936
		ABJ (3)	0.57	5.6 × 10−1	2	1933
		Extended (5)	0.95	3.9 × 10−1	2	1899
	High-OC group	Basic (1)	9.22	1.0 × 10−4	2	1937
		ABJ (3)	4.17	1.6 × 10−2	2	1934
		Extended (5)	2.48	8.4 × 10−2	2	1888
Adjusted SG&A cost stickiness by adjusted organizational capital under small sales changes	Low-OC group	Basic (2)	4.02	1.84 × 10−2	2	823
		ABJ (4)	2.14	1.18 × 10−1	2	820
		Extended (6)	2.04	1.31 × 10−1	2	803
	High-OC group	Basic (2)	0.26	7.73 × 10−1	2	903
		ABJ (4)	0.86	4.24 × 10−1	2	900
		Extended (6)	0.13	8.75 × 10−1	2	883
Adjusted SG&A cost stickiness by adjusted organizational capital under large sales changes	Low-OC group	Basic (2)	0.05	9.6 × 10−1	2	382
		ABJ (4)	0.01	9.9 × 10−1	2	379
		Extended (6)	1.22	3.0 × 10−1	2	367
	High-OC group	Basic (2)	0.09	9.1 × 10−1	2	331
		ABJ (4)	0.10	9.0 × 10−1	2	328
		Extended (6)	0.74	4.8 × 10−1	2	320

Note: The RESET test suggests some misspecification in the full-sample regressions, but subsample analyses generally do not reject linearity, supporting the use of a linear specification.

). Whereas the test indicated potential misspecification in the full-sample regressions, the results for subsample analyses did not reject the linear specification. Overall, these findings suggest that the linearity assumption is broadly reasonable for the purposes of our analysis. The estimations employ the basic, ABJ, and extended models (see Section 3.1). The estimation results are presented in

Table 6. SG&A cost stickiness by organizational capital.

	Low OC Group			High OC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β0: Constant	0.0075 ***	0.0079 ***	−0.0124 **	0.0084 ***	0.0086 ***	−0.0141 ***
	(0.0005)	(0.0005)	(0.0055)	(0.0005)	(0.0005)	(0.0041)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.4687 ***	0.4635 ***	0.4471 ***	0.4647 ***	0.4627 ***	0.4405 ***
	(0.0160)	(0.0161)	(0.0165)	(0.0160)	(0.0161)	(0.0165)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.0844 *	−0.1199 **	−0.1271 **	−0.0981 **	−0.0933 **	−0.1603 ***
	(0.0488)	(0.0564)	(0.0613)	(0.0457)	(0.0476)	(0.0508)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.0095	−0.2675		−0.4213 **	−0.4962 **
		(0.2232)	(0.2518)		(0.2115)	(0.2318)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.5569 ***	0.2880 ***		0.2844 ***	−0.0392
		(0.0664)	(0.1051)		(0.0722)	(0.1345)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.1465 ***	−0.0891 ***		−0.1339 ***	−0.0754 ***
		(0.0234)	(0.0269)		(0.0192)	(0.0230)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			1.66 × 10−5			7.47 × 10−6 *
			(1.79 × 10−5)			(3.94 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0120 ***			−0.0073 ***
			(0.0026)			(0.0020)
β8: ${Ds}_{i,t}$			−0.0046 ***			−0.0049 **
			(0.0016)			(0.0020)
β9: ${GNI}_{i,t}$			0.0012 ***			0.0013 ***
			(0.0003)			(0.0002)
β10: ${FCF}_{i,t}$			−3.74 × 10−7 ***			−6.83 × 10−8 ***
			(1.16 × 10−7)			(2.28 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0010 ***			−0.0009 ***
			(0.0003)			(0.0003)
β12: ${Size}_{i,t}$			0.0052 ***			0.0046 ***
			(0.0013)			(0.0008)
No. of observations	6363	6363	6303	6364	6364	6214
Adj. R-squared	0.28890	0.29732	0.30264	0.28470	0.29082	0.29521

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 4, indicating that multicollinearity is not a significant concern.

,

Table 7. Adjusted SG&A cost stickiness by organizational capital.

	Low OC Group			High OC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β0: Constant	0.0070 ***	0.0073 ***	−0.0158	0.0074 ***	0.0076 ***	−0.0248 ***
	(0.0012)	(0.0012)	(0.0114)	(0.0009)	(0.0009)	(0.0078)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.5009 ***	0.4965 ***	0.4843 ***	0.4858 ***	0.4808 ***	0.4772 ***
	(0.0332)	(0.0331)	(0.0329)	(0.0301)	(0.0301)	(0.0309)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.0400	0.0507	0.0753	−0.0673	−0.0546	−0.0946
	(0.1244)	(0.1542)	(0.1667)	(0.0903)	(0.0963)	(0.1009)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.9319	−0.9343		−0.6038	−0.8805 *
		(0.6305)	(0.7106)		(0.4354)	(0.4578)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.6083 ***	0.0586		0.4384 ***	0.3567 *
		(0.1733)	(0.2774)		(0.1346)	(0.1898)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0907	−0.0939		−0.1265 ***	−0.1036 **
		(0.0659)	(0.0735)		(0.0435)	(0.0507)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			3.85 × 10−5			3.45 × 10−6
			(4.55 × 10−5)			(6.59 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0075			−0.0139 ***
			(0.0052)			(0.0040)
β8: ${Ds}_{i,t}$			−0.0102 **			−0.0001
			(0.0047)			(0.0028)
β9: ${GNI}_{i,t}$			0.0003			0.0006
			(0.0006)			(0.0004)
β10: ${FCF}_{i,t}$			−4.81 × 10−7 **			−1.03 × 10−7 ***
			(2.28 × 10−7)			(3.95 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0011			0.0005
			(0.0007)			(0.0007)
β12: ${Size}_{i,t}$			0.0061 **			0.0060 ***
			(0.0027)			(0.0015)
No. of observations	1186	1186	1169	1287	1287	1275
Adj. R-squared	0.32195	0.32985	0.34202	0.32453	0.33120	0.34191

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 4, indicating that multicollinearity is not a significant concern.

,

Table 8. SG&A cost stickiness by adjusted organizational capital.

	Low AOC Group			High AOC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β0: Constant	0.0078 ***	0.0080 ***	−0.0139 ***	0.0082 ***	0.0084 ***	−0.0115 ***
	(0.0005)	(0.0005)	(0.0046)	(0.0005)	(0.0005)	(0.0038)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.4649 ***	0.4610 ***	0.4432 ***	0.4691 ***	0.4660 ***	0.4395 ***
	(0.0153)	(0.0154)	(0.0159)	(0.0171)	(0.0172)	(0.0177)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.0964 *	−0.1142 *	−0.1183 *	−0.0797 *	−0.0880 *	−0.1400 ***
	(0.0493)	(0.0619)	(0.0662)	(0.0458)	(0.0470)	(0.0502)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.1500	−0.4608		−0.2400	−0.3353
		(0.2591)	(0.2877)		(0.2091)	(0.2334)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.5537 ***	0.2702 **		0.3053 ***	0.0061
		(0.0689)	(0.1099)		(0.0691)	(0.1282)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.1481 ***	−0.0940 ***		−0.1322 ***	−0.0694 ***
		(0.0243)	(0.0280)		(0.0186)	(0.0222)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			2.15 × 10−5 *			5.44 × 10−6
			(1.21 × 10−5)			(4.04 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0147 ***			−0.0084 ***
			(0.0029)			(0.0023)
β8: ${Ds}_{i,t}$			−0.0048 ***			−0.0045 **
			(0.0017)			(0.0019)
β9: ${GNI}_{i,t}$			0.0011 ***			0.0014 ***
			(0.0003)			(0.0002)
β10: ${FCF}_{i,t}$			−1.27 × 10−7 **			−7.22 × 10−8 ***
			(5.94 × 10−8)			(2.3 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0010 ***			−0.0008 **
			(0.0003)			(0.0003)
β12: ${Size}_{i,t}$			0.0056 ***			0.0041 ***
			(0.0011)			(0.0007)
No. of observations	6363	6363	6302	6364	6364	6215
Adj. R-squared	0.28363	0.29149	0.29703	0.28960	0.29597	0.30042

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 4, indicating that multicollinearity is not a significant concern.

and

Table 9. Adjusted SG&A cost stickiness by adjusted organizational capital.

	Low AOC Group			High AOC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β0: Constant	0.0074 ***	0.0074 ***	−0.0066	0.0070 ***	0.0073 ***	−0.0200 ***
	(0.0012)	(0.0012)	(0.0084)	(0.0009)	(0.0009)	(0.0075)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.4872 ***	0.4869 ***	0.4710 ***	0.5041 ***	0.4976 ***	0.4921 ***
	(0.0311)	(0.0311)	(0.0310)	(0.0330)	(0.0330)	(0.0338)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.0738	0.1741	0.2145	−0.1054	−0.1387	−0.1969 *
	(0.1178)	(0.1385)	(0.1498)	(0.1026)	(0.1065)	(0.1118)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−1.630 ***	−1.855 ***		0.0273	−0.0465
		(0.5601)	(0.6361)		(0.5290)	(0.5629)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.5072 ***	−0.0415		0.5657 ***	0.4816 **
		(0.1672)	(0.2726)		(0.1481)	(0.2098)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0656	−0.0676		−0.1321 ***	−0.1024 **
		(0.0674)	(0.0741)		(0.0433)	(0.0503)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			1.16 × 10−5			8.6 × 10−6
			(1.92 × 10−5)			(7.58 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0091			−0.0131 ***
			(0.0056)			(0.0043)
β8: ${Ds}_{i,t}$			−0.0110 **			0.0002
			(0.0048)			(0.0029)
β9: ${GNI}_{i,t}$			0.0005			0.0005
			(0.0006)			(0.0004)
β10: ${FCF}_{i,t}$			−2.59 × 10−7 **			−8.72 × 10−8 **
			(1.06 × 10−7)			(4.37 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0005			−8.03 × 10−6
			(0.0007)			(0.0007)
β12: ${Size}_{i,t}$			0.0038 *			0.0052 ***
			(0.0020)			(0.0014)
No. of observations	1221	1221	1203	1252	1252	1241
Adj. R-squared	0.31710	0.32635	0.33445	0.33099	0.34025	0.34757

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 4, indicating that multicollinearity is not a significant concern.

.

In addition, the Studentized Breusch–Pagan tests (see

Table A4. Heteroskedasticity test (Studentized Breusch–Pagan test).

		Model	Statistic	p-Value	df
SG&A cost stickiness by organizational capital	Low-OC group	Basic (1)	329.16	3.35 × 10−72	2
		ABJ (3)	338.13	6.28 × 10−71	5
		Extended (5)	412.38	9.04 × 10−81	12
	High-OC group	Basic (1)	173.96	1.68 × 10−38	2
		ABJ (3)	186.65	2.03 × 10−38	5
		Extended (5)	220.37	1.99 × 10−40	12
Adjusted SG&A cost stickiness by organizational capital	Low-OC group	Basic (2)	37.23	8.23 × 10−9	2
		ABJ (4)	37.58	4.57 × 10−7	5
		Extended (6)	59.64	2.62 × 10−8	12
	High-OC group	Basic (2)	33.93	4.29 × 10−8	2
		ABJ (4)	38.71	2.72 × 10−7	5
		Extended (6)	54.35	2.37 × 10−7	12
SG&A cost stickiness by adjusted organizational capital	Low-OC group	Basic (1)	267.48	8.29 × 10−59	2
		ABJ (3)	277.01	8.75 × 10−58	5
		Extended (5)	329.02	3.71 × 10−63	12
	High-OC group	Basic (1)	222.20	5.62 × 10−49	2
		ABJ (3)	235.18	8.29 × 10−49	5
		Extended (5)	254.05	1.95 × 10−47	12
Adjusted SG&A cost stickiness by adjusted organizational capital	Low-OC group	Basic (2)	27.59	1.02 × 10−6	2
		ABJ (4)	27.80	3.99 × 10−5	5
		Extended (6)	41.51	4.03 × 10−5	12
	High-OC group	Basic (2)	44.65	2.01 × 10−10	2
		ABJ (4)	49.54	1.72 × 10−9	5
		Extended (6)	62.54	7.76 × 10−9	12
SG&A cost stickiness by organizational capital under small sales changes	Low-OC group	Basic (1)	9.09	1.06 × 10−2	2
		ABJ (3)	11.51	4.21 × 10−2	5
		Extended (5)	72.18	1.25 × 10−10	12
	High-OC group	Basic (1)	2.20	3.34 × 10−1	2
		ABJ (3)	9.98	7.59 × 10−2	5
		Extended (5)	33.20	9.01 × 10−4	12
SG&A cost stickiness by organizational capital under large sales changes	Low-OC group	Basic (1)	102.61	5.22 × 10−23	2
		ABJ (3)	109.67	4.81 × 10−22	5
		Extended (5)	155.28	4.80 × 10−27	12
	High-OC group	Basic (1)	89.07	4.55 × 10−20	2
		ABJ (3)	95.27	5.23 × 10−19	5
		Extended (5)	109.36	8.02 × 10−18	12
Adjusted SG&A cost stickiness by organizational capital under small sales changes	Low-OC group	Basic (2)	1.37	5.03 × 10−1	2
		ABJ (4)	2.14	8.29 × 10−1	5
		Extended (6)	17.17	1.43 × 10−1	12
	High-OC group	Basic (2)	6.93	3.13 × 10−2	2
		ABJ (4)	12.48	2.88 × 10−2	5
		Extended (6)	37.93	1.57 × 10−4	12
Adjusted SG&A cost stickiness by organizational capital under large sales changes	Low-OC group	Basic (2)	2.61	2.71 × 10−1	2
		ABJ (4)	3.91	5.62 × 10−1	5
		Extended (6)	18.36	1.05 × 10−1	12
	High-OC group	Basic (2)	16.73	2.32 × 10−4	2
		ABJ (4)	19.73	1.40 × 10−3	5
		Extended (6)	19.76	7.17 × 10−2	12
SG&A cost stickiness by adjusted organizational capital under small sales changes	Low-OC group	Basic (1)	8.29	1.58 × 10−2	2
		ABJ (3)	11.39	4.42 × 10−2	5
		Extended (5)	53.91	2.83 × 10−7	12
	High-OC group	Basic (1)	2.28	3.20 × 10−1	2
		ABJ (3)	9.82	8.05 × 10−2	5
		Extended (5)	23.66	2.26 × 10−2	12
SG&A cost stickiness by adjusted organizational capital under large sales changes	Low-OC group	Basic (1)	84.21	5.18 × 10−19	2
		ABJ (3)	91.94	2.63 × 10−18	5
		Extended (5)	136.35	3.26 × 10−23	12
	High-OC group	Basic (1)	111.82	5.24 × 10−25	2
		ABJ (3)	118.02	8.23 × 10−24	5
		Extended (5)	127.92	1.62 × 10−21	12
Adjusted SG&A cost stickiness by adjusted organizational capital under small sales changes	Low-OC group	Basic (2)	1.38	5.03 × 10−1	2
		ABJ (4)	2.80	7.31 × 10−1	5
		Extended (6)	13.96	3.03 × 10−1	12
	High-OC group	Basic (2)	8.37	1.52 × 10−2	2
		ABJ (4)	16.70	5.10 × 10−3	5
		Extended (6)	28.74	4.31 × 10−3	12
Adjusted SG&A cost stickiness by adjusted organizational capital under large sales changes	Low-OC group	Basic (2)	3.48	1.76 × 10−1	2
		ABJ (4)	4.63	4.63 × 10−1	5
		Extended (6)	16.98	1.50 × 10−1	12
	High-OC group	Basic (2)	14.77	6.22 × 10−4	2
		ABJ (4)	18.48	2.40 × 10−3	5
		Extended (6)	21.71	4.09 × 10−2	12

Note: The Breusch–Pagan test indicates heteroskedasticity in most specifications; therefore, all pooled OLS results are reported with robust standard errors.

) revealed heteroskedasticity in most full-sample regressions. Accordingly, all pooled OLS results are reported with heteroskedasticity-robust standard errors to ensure reliable inference.

4.1.1. Organizational Capital and SG&A Cost Stickiness

Table 6 reports regression results with the logarithmic annual change rate of SG&A expenses as the dependent variable. Firms are divided into high and low OC groups.

For the low capital group, the estimated coefficient β₁ is 0.4687 *** (0.4635 ***; 0.4471 ***), indicating that a 1% sales increase corresponds to a 0.4687% (0.4635%; 0.4471%) rise in SG&A expenses. This implies that SG&A costs rise less than proportionally with sales, reflecting managerial efforts to achieve scale efficiency when sales expand. The coefficient β₂ is −0.0844 * (−0.1199 **; −0.1271 **), indicating cost stickiness, though significance is weak in the basic model (10% level) and stronger in the ABJ and extended models (5% level). This means that when sales fall, cost reductions are smaller, suggesting firms are reluctant to immediately cut resources such as administrative staff or overhead, which may involve adjustment costs. The combined β₁ + β₂ = 0.3843 (0.3436; 0.3200), meaning a 1% sales decline results in a smaller reduction in SG&A expenses.

For the high capital group, β₁ is 0.4647 *** (0.4627 ***; 0.4405 ***), meaning a 1% sales increase leads to a 0.4647% (0.4627%; 0.4405%) rise in SG&A expenses. The magnitude is similar to the low OC group, suggesting that both groups exhibit proportional SG&A growth during sales expansion. The estimated β₂ is −0.0981 ** (−0.0933 **; −0.1603 ***), indicating stickiness, consistently significant across all model specifications. This suggests that firms with stronger organizational capital are more constrained in adjusting SG&A downward, likely because accumulated human capital, internal routines, and organizational processes cannot be easily reduced without harming long-term capability. The combined β₁ + β₂ = 0.3666 (0.3694; 0.2802) suggests that a 1% sales decline results in a smaller expense reduction. Since β₂ is more negative for the high capital group, SG&A costs are stickier in these firms.

The extended models provide additional insights into factors driving stickiness by incorporating firm-level and macroeconomic variables. For low capital firms, the asset-to-sales ratio (β₇ = −0.0120 ***) has a significant dampening effect, suggesting firms with larger asset bases relative to sales show smaller SG&A increases. This indicates that asset-intensive firms tend to rely more on fixed investments and less on discretionary SG&A spending, making SG&A less sensitive to sales increases. β₃ is insignificant in both the ABJ model (−0.0095) and the extended model (−0.2675), implying asset intensity does not materially alter stickiness when OC is low. In other words, low-OC firms already exhibit relatively flexible cost structures, and asset intensity adds little explanatory power. By contrast, in high-OC firms, β₃ (−0.4213 ** in ABJ, −0.4962 ** in extended) and β₇ (−0.0073 ***) are significantly negative, indicating asset intensity strengthens cost reductions during sales declines, thereby weakening stickiness. This suggests that firms with high organizational capital and high asset intensity can still reduce costs more aggressively, potentially by leveraging economies of scale or reallocating resources more effectively.

Sales decline persistence also affects stickiness. In both groups, β₈ is significantly negative (Low: −0.0046 ***; High: −0.0049 **), suggesting that firms experiencing prolonged sales declines reduce SG&A costs more aggressively. This finding reflects adaptive behavior: persistent downturns push firms to eventually cut SG&A despite initial reluctance. However, β₄ shows opposite patterns: in low capital firms, it is significantly positive in both ABJ (0.5569 ***) and extended (0.2880 ***) models, meaning successive declines partially offset expected expense reductions, reinforcing stickiness. This implies that managers in low-OC firms may hold back on cutting costs too quickly, anticipating that declines may reverse. In high capital firms, β₄ is significant in the ABJ model (0.2844 ***) but becomes insignificant in the extended model (−0.0392). This suggests that, once other firm-level controls are considered, sales persistence has a limited additional effect on cost behavior in high-OC firms.

Macroeconomic conditions influence behavior similarly across groups. β₉ is significantly positive (Low: 0.0012 ***; High: 0.0013 ***), suggesting that stronger macroeconomic conditions are associated with higher SG&A expenditures. This result is consistent with the view that during favorable macroeconomic periods, firms expand discretionary spending to support growth. Meanwhile, β₅ is significantly negative (Low: −0.1465 ***; High: −0.1339 ***), suggesting that, during economic expansion, reductions in SG&A costs are mitigated when sales fall, thereby increasing stickiness. This indicates that firms are less willing to cut SG&A during downturns if the broader economy is performing well, possibly reflecting optimism about recovery.

Finally, financial flexibility as measured by free cash flow reveals contrasting effects. For the low OC group, β₁₀ is significantly negative (−3.74 × 10⁻⁷ ***), implying that firms with greater free cash flow cut SG&A costs more readily when sales decline, reducing stickiness. This supports the view that financially flexible firms adjust spending more efficiently in response to downturns. By contrast, for the high OC group, β₆ is significantly positive (7.47 × 10⁻⁶ *) while β₁₀ is also significantly negative (−6.83 × 10⁻⁸ ***). This indicates that although abundant cash flow encourages SG&A maintenance in the short term, overall, firms with higher OC still reduce SG&A when cash flow is high, although to a lesser extent than low-OC firms. This result suggests that although abundant cash flow initially encourages managers in high-OC firms to maintain SG&A spending, overall, these firms still reduce costs when financial slack is high, but the adjustment is weaker than in low-OC firms.

In addition, firm characteristics play a consistent role across both groups. Leverage (Low: −0.0010 ***; High: −0.0009 ***) exerts a negative effect, suggesting that more highly leveraged firms cut SG&A costs more aggressively during sales declines, thereby reducing stickiness. This is consistent with the disciplining role of debt, which forces managers to tighten cost control under financial pressure. By contrast, firm size (Low: 0.0052 ***; High: 0.0046 ***) is significantly positive, implying that larger firms maintain relatively higher SG&A expenditures, which contribute to greater cost stickiness. This aligns with the idea that larger organizations have more administrative complexity and overhead, making rapid cost adjustments more difficult.

In summary, both low- and high-OC firms exhibit SG&A cost stickiness, but the effect is stronger for firms with greater organizational capital. These results support Hypothesis H1, confirming that OC level significantly influences the degree of cost stickiness.

4.1.2. Organizational Capital and Adjusted SG&A Cost Stickiness

Table 7 reports the regression results using the logarithmic annual change in adjusted SG&A expenses as the dependent variable, dividing firms into low OC and high OC groups.

For low-OC firms, β₁ is 0.5009 *** (0.4965 ***; 0.4843 ***), indicating that a 1% increase in sales is associated with a 0.5009% (0.4965%; 0.4843%) increase in adjusted SG&A expenses. This suggests that adjusted SG&A rises roughly in proportion to sales growth, consistent with flexible cost behavior. The coefficient β₂ is 0.0400 (0.0507; 0.0753) and is not statistically significant in any specification, implying no clear stickiness. This indicates that when sales fall, low-OC firms reduce adjusted SG&A almost symmetrically, showing limited organizational rigidity. The sum β₁ + β₂ = 0.5409 (0.5472; 0.5596) shows that a 1% sales decline yields a near-symmetric reduction.

For high-OC firms, β₁ is 0.4858 *** (0.4808 ***; 0.4772 ***), meaning that a 1% sales increase leads to a 0.4858% (0.4808%; 0.4772%) increase in adjusted SG&A expenses. This again reflects proportional cost behavior under expansion. The estimated β₂ is −0.0673 (−0.0546; −0.0946), but it is not statistically significant in any specification. Thus, adjusted SG&A expenses in high-OC firms do not show robust stickiness once adjustments are made, suggesting that organizational capital’s influence is less pronounced when non-discretionary items are removed. The sum β₁ + β₂ = 0.4185 (0.4262; 0.3826) indicates that although reductions are smaller in magnitude, they are not statistically distinguishable from symmetry.

In summary, both low- and high-OC firms show little evidence of stickiness in adjusted SG&A expenses. This result implies that once SG&A is adjusted to exclude non-discretionary components, the role of organizational capital in amplifying stickiness weakens considerably. These results provide a nuanced view of H1, which posits that OC influences the extent of cost stickiness in SG&A expenses.

4.1.3. Adjusted Organizational Capital and SG&A Cost Stickiness

Table 8 reports results using the logarithmic annual change in SG&A expenses as the dependent variable, separating firms into low AOC and high AOC groups.

For the low AOC group, the estimated β₁ is 0.4649 *** (0.4610 ***; 0.4432 ***), indicating that a 1% increase in sales is associated with a 0.4649% (0.4610%; 0.4432%) increase in SG&A expenses. This shows that SG&A costs rise less than proportionally with sales, reflecting partial scale efficiency. The coefficient β₂ is −0.0964 * (−0.1142 *; −0.1183 *), suggesting the presence of cost stickiness. The effect is weakly significant in the basic model (10% level) but becomes stronger in the ABJ and extended models. This indicates that when sales fall, low AOC firms are reluctant to reduce SG&A proportionally, possibly due to adjustment costs or managerial inertia. The combined coefficient β₁ + β₂ = 0.3685 (0.3468; 0.3249) indicates that a 1% decrease in sales leads to a 0.3685% (0.3468%; 0.3249%) decline in SG&A expenses.

For the high AOC group, the estimated β₁ is 0.4691 *** (0.4660 ***; 0.4395 ***), indicating that a 1% increase in sales results in a 0.4691% (0.4660%; 0.4395%) increase in SG&A expenses. This again reflects the partial proportionality of SG&A to sales increases. The estimated β₂ is −0.0797 * (−0.0880 *; −0.1400 ***), consistently negative and statistically significant, confirming stickiness. This suggests that high AOC firms face stronger organizational rigidity, as accumulated routines and capabilities make cost reductions harder when sales fall. The sum β₁ + β₂ = 0.3894 (0.3780; 0.2995) likewise indicates a muted reduction during sales declines. The more negative β₂ values in the high AOC group imply stronger stickiness compared to the low AOC group.

Overall, both groups display SG&A cost stickiness, with the effect more pronounced in high AOC firms. This supports H1, showing that adjusted organizational capital strengthens the asymmetry in SG&A cost behavior, particularly under sales declines.

4.1.4. Adjusted Organizational Capital and Adjusted SG&A Cost Stickiness

Table 9 reports the regression results using the logarithmic annual change in adjusted SG&A expenses as the dependent variable, comparing low AOC and high AOC firms.

For low AOC firms, β₁ is 0.4872 *** (0.4869 ***; 0.4710 ***), indicating that a 1% increase in sales is associated with a 0.4872% (0.4869%; 0.4710 ***) increase in adjusted SG&A expenses. This suggests that adjusted SG&A responds almost proportionally to sales growth, reflecting flexible cost behavior. The coefficient β₂ is 0.0738 (0.1741; 0.2145) and is not significant, indicating no clear stickiness. This means that when sales fall, low AOC firms reduce adjusted SG&A almost symmetrically, showing little evidence of rigid cost structures. The sum β₁ + β₂ = 0.5610 (0.6610; 0.6855) suggests a near-symmetric response to sales declines.

For high AOC firms, β₁ is 0.5041 *** (0.4976 ***; 0.4921 ***), meaning that a 1% increase in sales leads to a 0.5041% (0.4976%; 0.4921%) increase in adjusted SG&A expenses. This indicates that SG&A scales proportionally with sales expansion. The estimated β₂ is −0.1054 (−0.1387; −0.1969 *), negative and significant only in the extended model, suggesting partial evidence of cost stickiness. This implies that high AOC firms are slower in reducing adjusted SG&A during downturns, likely because organizational routines and embedded capabilities create resistance to cost cuts. The sum β₁ + β₂ = 0.3987 (0.3589; 0.2952) indicates a smaller decline in adjusted SG&A expenses when sales fall, especially when controlling for firm characteristics.

In summary, low AOC firms show no significant stickiness in adjusted SG&A expenses, while high AOC firms exhibit some stickiness, particularly in extended specifications. These results support H1, suggesting that adjusted organizational capital is associated with greater asymmetry in SG&A adjustments, even after excluding non-discretionary components.

Across specifications, high OC and high AOC firms generally exhibit stronger SG&A cost stickiness, reflected in more negative β₂ coefficients. In other words, when sales decline, these firms reduce SG&A expenses less, sustaining higher expenditure levels. By contrast, low OC and low AOC firms show weaker or insignificant stickiness, indicating more symmetric cost adjustments. Overall, these findings robustly support H1. These results align with Anderson et al. (2003), who documented asymmetric SG&A adjustments, and extend their evidence by showing that organizational capital amplifies the degree of stickiness. Unlike Venieris et al. (2015), however, low-OC firms in Japan also display cost stickiness, reflecting institutional differences.

4.2. The Impact of Organizational Capital on Cost Stickiness Based on the Magnitude of Sales Changes

To test the hypothesis that the effect of organizational capital on cost stickiness depends on the magnitude of sales changes, this study adopts a comparative approach by dividing the sample into two groups: firms experiencing small sales changes and those experiencing large sales changes.

Sales changes of up to and including 10% are classified as “small changes,” while those between 10% and 50% are classified as “large changes” (Venieris et al., 2015). The estimation results for these groups are presented in

Table 10. SG&A cost stickiness by organizational capital under small sales changes.

	Low OC Group			High OC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β0: Constant	0.0055 ***	0.0057 ***	0.0001	0.0077 ***	0.0077 ***	−0.0183 ***
	(0.0006)	(0.0006)	(0.0053)	(0.0005)	(0.0005)	(0.0040)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.5583 ***	0.5503 ***	0.5265 ***	0.5124 ***	0.5129 ***	0.4795 ***
	(0.0292)	(0.0294)	(0.0307)	(0.0264)	(0.0265)	(0.0273)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.2576 ***	−0.2574 ***	−0.2879 ***	−0.1263 *	−0.0832	−0.1653 **
	(0.0728)	(0.0788)	(0.0844)	(0.0681)	(0.0688)	(0.0740)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.0493	−0.0558		−0.2527	−0.3218
		(0.3187)	(0.3557)		(0.2865)	(0.3030)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.4986 ***	0.2482		0.2054 *	−0.0808
		(0.0950)	(0.1614)		(0.1060)	(0.1773)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.1592 ***	−0.0805 **		−0.1820 ***	−0.0868 **
		(0.0334)	(0.0382)		(0.0343)	(0.0384)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			2.17 × 10−5			8.71 × 10−6
			(2.34 × 10−5)			(5.85 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0034			−0.0070 ***
			(0.0024)			(0.0020)
β8: ${Ds}_{i,t}$			−0.0031 *			−0.0027
			(0.0018)			(0.0021)
β9: ${GNI}_{i,t}$			0.0010 ***			0.0014 ***
			(0.0003)			(0.0002)
β10: ${FCF}_{i,t}$			−3.44 × 10−7 ***			−1.23 × 10−7 ***
			(1.03 × 10−7)			(2.26 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0014 ***			−0.0008 ***
			(0.0004)			(0.0003)
β12: ${Size}_{i,t}$			0.0019			0.0053 ***
			(0.0012)			(0.0008)
No. of observations	4379	4379	4349	4380	4380	4284
Adj. R-squared	0.13852	0.14507	0.15381	0.14901	0.15496	0.17200

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 4, indicating that multicollinearity is not a significant concern.

,

Table 11. SG&A cost stickiness by organizational capital under large sales changes.

	Low OC Group			High OC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β0: Constant	0.0125 ***	0.0136 ***	−0.0076	0.0104 ***	0.0116 ***	0.0054
	(0.0029)	(0.0029)	(0.0137)	(0.0025)	(0.0025)	(0.0094)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.4290 ***	0.4155 ***	0.4016 ***	0.4536 ***	0.4386 ***	0.4167 ***
	(0.0418)	(0.0418)	(0.0423)	(0.0382)	(0.0382)	(0.0387)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.0568	0.0462	0.0486	−0.0677	−0.0969	−0.0917
	(0.0922)	(0.0975)	(0.0992)	(0.0819)	(0.0845)	(0.0873)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.1884	−0.4500 **		0.0574	0.0335
		(0.1778)	(0.2005)		(0.1410)	(0.1601)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.2030 ***	−0.4328 **		0.1499 ***	−0.3389
		(0.0583)	(0.2089)		(0.0566)	(0.2647)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0699 ***	−0.0537 ***		−0.0699 ***	−0.0524 ***
		(0.0143)	(0.0178)		(0.0114)	(0.0153)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			2.02 × 10−5			1.4 × 10−6
			(2.08 × 10−5)			(2.78 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0254 ***			−0.0064
			(0.0065)			(0.0047)
β8: ${Ds}_{i,t}$			−0.0383 ***			−0.0302 **
			(0.0124)			(0.0153)
β9: ${GNI}_{i,t}$			0.0008			0.0010
			(0.0007)			(0.0006)
β10: ${FCF}_{i,t}$			7.2 × 10−8			2.91 × 10−8
			(3.15 × 10−7)			(4.23 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0020 **			−0.0009
			(0.0008)			(0.0009)
β12: ${Size}_{i,t}$			0.0061 **			0.0015
			(0.0031)			(0.0018)
No. of observations	1941	1941	1912	1942	1942	1905
Adj. R-squared	0.34790	0.35874	0.36771	0.37899	0.39488	0.39396

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 6, indicating that multicollinearity is not a significant concern.

,

Table 12. Adjusted SG&A cost stickiness by organizational capital under small sales changes.

	Low OC Group			High OC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β0: Constant	0.0037 ***	0.0040 ***	0.0097	0.0046 ***	0.0049 ***	−0.0219 ***
	(0.0013)	(0.0013)	(0.0115)	(0.0010)	(0.0010)	(0.0083)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.6493 ***	0.6343 ***	0.6159 ***	0.6575 ***	0.6475 ***	0.6437 ***
	(0.0649)	(0.0644)	(0.0669)	(0.0540)	(0.0541)	(0.0568)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.2699	−0.4198 **	−0.4471 *	−0.3772 **	−0.3794 **	−0.4029 **
	(0.1849)	(0.2072)	(0.2295)	(0.1641)	(0.1711)	(0.1799)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.1209	0.1671		−0.4737	−0.5984
		(0.6064)	(0.6631)		(0.6353)	(0.6282)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.9426 ***	0.7535 *		0.4561 *	0.5218
		(0.2132)	(0.4218)		(0.2492)	(0.3530)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0627	−0.0657		−0.0928	−0.0840
		(0.0871)	(0.0959)		(0.0805)	(0.0923)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			2.53 × 10−5			9.14 × 10−7
			(3.92 × 10−5)			(9.87 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			0.0026			−0.0096 **
			(0.0047)			(0.0043)
β8: ${Ds}_{i,t}$			−0.0029			0.0021
			(0.0052)			(0.0032)
β9: ${GNI}_{i,t}$			1.91 × 10−5			0.0004
			(0.0005)			(0.0005)
β10: ${FCF}_{i,t}$			−8.85 × 10−8			−1.3 × 10−7 ***
			(2.19 × 10−7)			(4.3 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0014 *			−0.0002
			(0.0007)			(0.0008)
β12: ${Size}_{i,t}$			−0.0006			0.0052 ***
			(0.0026)			(0.0016)
No. of observations	811	811	802	925	925	914
Adj. R-squared	0.19235	0.20455	0.19587	0.20796	0.21021	0.21587

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 4, indicating that multicollinearity is not a significant concern.

,

Table 13. Adjusted SG&A cost stickiness by organizational capital under large sales changes.

	Low OC Group			High OC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β0: Constant	0.0102	0.0118 *	−0.0404	0.0143 ***	0.0148 ***	−0.0243
	(0.0062)	(0.0063)	(0.0334)	(0.0052)	(0.0051)	(0.0208)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.5251 ***	0.5066 ***	0.5098 ***	0.4167 ***	0.4096 ***	0.4155 ***
	(0.0822)	(0.0830)	(0.0829)	(0.0811)	(0.0811)	(0.0790)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.0774	0.1376	0.1135	0.1170	−0.0741	−0.1199
	(0.1938)	(0.2064)	(0.1996)	(0.1761)	(0.1784)	(0.1807)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.7901 *	−1.019 **		0.2143	−0.1605
		(0.4195)	(0.4404)		(0.2422)	(0.3190)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.2531	−0.2048		0.4854 ***	0.6777
		(0.1569)	(0.6573)		(0.1174)	(0.4809)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.1174 **	−0.1411 ***		−0.0279	−0.0338
		(0.0492)	(0.0532)		(0.0328)	(0.0373)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			5.52 × 10−5 ***			1.95 × 10−6
			(1.87 × 10−5)			(4.45 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0292 **			−0.0249 ***
			(0.0135)			(0.0096)
β8: ${Ds}_{i,t}$			−0.0278			0.0120
			(0.0379)			(0.0256)
β9: ${GNI}_{i,t}$			−0.0019			8.65 × 10−5
			(0.0016)			(0.0010)
β10: ${FCF}_{i,t}$			−6.52 × 10−7			−3.17 × 10−8
			(5.68 × 10−7)			(8.43 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0020			0.0024
			(0.0016)			(0.0016)
β12: ${Size}_{i,t}$			0.0142 *
			(0.0075)
No. of observations	377	377	372	346	346	345
Adj. R-squared	0.38696	0.40288	0.42134	0.38734	0.40507	0.41360

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 6, indicating that multicollinearity is not a significant concern.

,

Table 14. SG&A cost stickiness by adjusted organizational capital under small sales changes.

	Low AOC Group			High AOC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β0: Constant	0.0058 ***	0.0059 ***	−0.0031	0.0075 ***	0.0075 ***	−0.0160 ***
	(0.0006)	(0.0006)	(0.0042)	(0.0005)	(0.0005)	(0.0036)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.5490 ***	0.5437 ***	0.5094 ***	0.5214 ***	0.5193 ***	0.4931 ***
	(0.0298)	(0.0300)	(0.0310)	(0.0257)	(0.0258)	(0.0268)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.2481 ***	−0.2459 ***	−0.2491 ***	−0.1328 **	−0.1058	−0.1988 ***
	(0.0746)	(0.0853)	(0.0903)	(0.0662)	(0.0664)	(0.0721)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.0138	−0.0379		−0.3070	−0.2786
		(0.3560)	(0.4032)		(0.2984)	(0.3175)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.4938 ***	0.1127		0.2317 **	0.0612
		(0.0966)	(0.1677)		(0.1055)	(0.1723)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.1754 ***	−0.1011 ***		−0.1622 ***	−0.0596
		(0.0338)	(0.0388)		(0.0341)	(0.0380)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			9.53 × 10−6			9.2 × 10−6
			(1.73 × 10−5)			(5.96 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0044 *			−0.0065 ***
			(0.0026)			(0.0022)
β8: ${Ds}_{i,t}$			−0.0049 **			−0.0010
			(0.0020)			(0.0019)
β9: ${GNI}_{i,t}$			0.0010 ***			0.0014 ***
			(0.0003)			(0.0002)
β10: ${FCF}_{i,t}$			−2.37 × 10−7 ***			−1.1 × 10−7 ***
			(6.07 × 10−8)			(2.32 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0013 ***			−0.0009 ***
			(0.0003)			(0.0003)
β12: ${Size}_{i,t}$			0.0027 ***			0.0048 ***
			(0.0010)			(0.0007)
No. of observations	4379	4379	4347	4380	4380	4286
Adj. R-squared	0.13112	0.13812	0.14765	0.15889	0.16405	0.18025

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 4, indicating that multicollinearity is not a significant concern.

,

Table 15. SG&A cost stickiness by adjusted organizational capital under large sales changes.

	Low AOC Group			High AOC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β0: Constant	0.0125 ***	0.0137 ***	−0.0129	0.0106 ***	0.0118 ***	−0.0047
	(0.0028)	(0.0028)	(0.0128)	(0.0026)	(0.0026)	(0.0088)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.4349 ***	0.4214 ***	0.4063 ***	0.4435 ***	0.4289 ***	0.3951 ***
	(0.0401)	(0.0400)	(0.0405)	(0.0405)	(0.0405)	(0.0410)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.0305	0.0193	0.0350	−0.0336	−0.0555	−0.0401
	(0.0895)	(0.0970)	(0.0989)	(0.0858)	(0.0880)	(0.0910)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.2018	−0.4630 **		0.0749	−0.0514
		(0.1866)	(0.2086)		(0.1545)	(0.1751)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.2488 ***	−0.4709 **		0.1126 **	−0.3420
		(0.0605)	(0.2218)		(0.0546)	(0.2426)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0682 ***	−0.0539 ***		−0.0725 ***	−0.0501 ***
		(0.0152)	(0.0185)		(0.0111)	(0.0149)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			1.36 × 10−5			7.69 × 10−7
			(1.01 × 10−5)			(2.92 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0274 ***			−0.0160 ***
			(0.0071)			(0.0052)
β8: ${Ds}_{i,t}$			−0.0426 ***			−0.0280 **
			(0.0132)			(0.0139)
β9: ${GNI}_{i,t}$			0.0006			0.0012 **
			(0.0007)			(0.0006)
β10: ${FCF}_{i,t}$			6.12 × 10−8			1.78 × 10−8
			(1.44 × 10−7)			(4.21 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0018 **			−0.0012
			(0.0008)			(0.0009)
β12: ${Size}_{i,t}$			0.0074 **			0.0037 **
			(0.0030)			(0.0017)
No. of observations	1941	1941	1914	1942	1942	1903
Adj. R-squared	0.33671	0.34817	0.35728	0.38911	0.40526	0.40791

** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 6, indicating that multicollinearity is not a significant concern.

,

Table 16. Adjusted SG&A cost stickiness by adjusted organizational capital under small sales changes.

	Low AOC Group			High AOC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β0: Constant	0.0042 ***	0.0045 ***	0.0038	0.0042 ***	0.0044 ***	−0.0198 **
	(0.0014)	(0.0013)	(0.0089)	(0.0010)	(0.0010)	(0.0077)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.6277 ***	0.6145 ***	0.5744 ***	0.6790 ***	0.6672 ***	0.6700 ***
	(0.0670)	(0.0665)	(0.0691)	(0.0511)	(0.0511)	(0.0544)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.2260	−0.3370	−0.3172	−0.4224 ***	−0.4702 ***	−0.5041 ***
	(0.1873)	(0.2191)	(0.2369)	(0.1602)	(0.1656)	(0.1721)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.1821	−0.0652		−0.5399	−0.5391
		(0.6455)	(0.7077)		(0.6561)	(0.6395)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		1.002 ***	0.7090		0.4596 *	0.5655 *
		(0.2193)	(0.4336)		(0.2412)	(0.3368)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0951	−0.1116		−0.0580	−0.0280
		(0.0865)	(0.0966)		(0.0816)	(0.0911)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			2.31 × 10−5			−3.51 × 10−7
			(2.6 × 10−5)			(1 × 10−5)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			0.0029			−0.0126 ***
			(0.0049)			(0.0043)
β8: ${Ds}_{i,t}$			−0.0047			0.0032
			(0.0054)			(0.0032)
β9: ${GNI}_{i,t}$			7.76 × 10−5			0.0004
			(0.0006)			(0.0004)
β10: ${FCF}_{i,t}$			−2.4 × 10−7 **			−1.06 × 10−7 **
			(1.19 × 10−7)			(4.44 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0005			−0.0013 *
			(0.0008)			(0.0008)
β12: ${Size}_{i,t}$			0.0005			0.0051 ***
			(0.0021)			(0.0015)
No. of observations	828	828	818	908	908	898
Adj. R-squared	0.17854	0.19135	0.18503	0.22714	0.22925	0.23618

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 4, indicating that multicollinearity is not a significant concern.

and

Table 17. Adjusted SG&A cost stickiness by adjusted organizational capital under large sales changes.

	Low AOC Group			High AOC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β0: Constant	0.0126 **	0.0136 **	−0.0325	0.0118 **	0.0122 **	−0.0426 **
	(0.0060)	(0.0061)	(0.0244)	(0.0053)	(0.0053)	(0.0217)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.5000 ***	0.4877 ***	0.4802 ***	0.4437 ***	0.4383 ***	0.4450 ***
	(0.0805)	(0.0810)	(0.0797)	(0.0836)	(0.0838)	(0.0832)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.1336	0.2140	0.2229	0.0599	−0.1424	−0.2074
	(0.1899)	(0.2018)	(0.1914)	(0.1804)	(0.1878)	(0.1940)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.9871 **	−1.354 ***		0.4482	0.0902
		(0.3819)	(0.4196)		(0.2968)	(0.3840)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.2931 *	−0.3895		0.4333 ***	0.8178 *
		(0.1552)	(0.6585)		(0.1223)	(0.4793)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.1197 **	−0.1344 **		−0.0401	−0.0581
		(0.0502)	(0.0530)		(0.0333)	(0.0383)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			4.35 × 10−5 **			6.66 × 10−5
			(1.99 × 10−5)			(5.23 × 10−5)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0418 ***			−0.0316 ***
			(0.0143)			(0.0103)
β8: ${Ds}_{i,t}$			−0.0419			0.0236
			(0.0388)			(0.0256)
β9: ${GNI}_{i,t}$			−0.0018			−0.0004
			(0.0015)			(0.0010)
β10: ${FCF}_{i,t}$			−5.86 × 10−7 **			−2.03 × 10−8
			(2.53 × 10−7)			(8.97 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0021			0.0019
			(0.0015)			(0.0017)
β12: ${Size}_{i,t}$			0.0134 **			0.0096 **
			(0.0054)			(0.0039)
No. of observations	387	387	382	336	336	335
Adj. R-squared	0.38364	0.40384	0.42916	0.38974	0.40691	0.41837

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: VIF values are all below 6, indicating that multicollinearity is not a significant concern.

.

4.2.1. Organizational Capital and SG&A Cost Stickiness Across Sales Change Magnitudes

Table 10 and Table 11 report the regression results using the logarithmic annual change in SG&A expenses as the dependent variable, classifying firms by OC level and sales-change magnitude (Venieris et al., 2015).

For small sales changes (see Table 10), stickiness is clear. For the low OC group, β₁ is 0.5583 *** (0.5503 ***; 0.5265 ***), indicating that a 1% sales increase leads to about a 0.56% rise in SG&A expenses. This reflects partial proportionality of SG&A with sales expansion. The coefficient β₂ is −0.2576 *** (−0.2574 ***; −0.2879 ***), strongly significant and negative, confirming asymmetry. This means that when sales fall, SG&A costs decrease less, consistent with cost stickiness. The combined β₁ + β₂ = 0.3007 (0.2929; 0.2386) confirms that a 1% sales decline results in a much smaller reduction in SG&A. In the high OC group, β₁ is 0.5124 *** (0.5129 ***; 0.4795 ***), indicating SG&A grows by 0.5124% (0.5129%; 0.4795%) when sales rise 1%. This suggests that SG&A growth remains proportional under sales expansion. The estimated β₂ is −0.1263 * (−0.0832; −0.1653 **), confirming asymmetry but with weaker significance than in the low OC group. This implies that high-OC firms still show some stickiness, although the evidence is less consistent across models. The combined β₁ + β₂ = 0.3861 (0.4297; 0.3142) likewise indicates smaller reductions during sales declines.

For large sales changes (see Table 11), the evidence of cost stickiness weakens considerably. For the low OC group, β₁ is 0.4290 *** (0.4155 ***; 0.4016 ***), showing SG&A increases by 0.4290% (0.4155%; 0.4016%) for a 1% sales rise. This reflects proportional adjustment under growth. The coefficient β₂ is 0.0568 (0.0462; 0.0486), all insignificant, indicating no asymmetry. This suggests that under larger shocks, SG&A cuts are more symmetric. For the high OC group, β₁ is 0.4536 *** (0.4386 ***; 0.4167 ***), likewise confirming proportional SG&A growth. This indicates that cost structures adjust in line with sales expansion. The estimated β₂ is −0.0677 (−0.0969; −0.0917), negative but not significant. Thus, even high-OC firms show little evidence of stickiness under large shocks, suggesting managers adjust SG&A more flexibly when sales swings are large.

Comparing across sales-change magnitudes, cost stickiness is statistically significant only for smaller changes (up to 10%) and largely disappears when changes fall between 10% and 50%. This pattern applies to both low- and high-OC firms, consistent with the idea that stickiness is more relevant under marginal adjustments but not under major shifts.

In summary, SG&A cost stickiness is evident under smaller sales changes but weakens considerably as the magnitude of changes increases. These findings support Hypothesis H2, which posits that the impact of organizational capital on cost stickiness varies with the magnitude of sales changes. When sales changes are modest, firms tend to preserve organizational capacity and resources, leading to asymmetric cost behavior. In contrast, larger sales changes compel substantial cost realignment, reducing stickiness.

4.2.2. Organizational Capital and Adjusted SG&A Cost Stickiness Across Sales Change Magnitudes

Table 12 and Table 13 report regression results using the logarithmic annual change in adjusted SG&A expenses as the dependent variable, dividing firms by their level of OC and by the magnitude of sales changes.

For small sales changes (see Table 12), the results show strong evidence of cost stickiness. For low-OC firms, β₁ is 0.6493 *** (0.6343 ***; 0.6159 ***), indicating that a 1% increase in sales is associated with a 0.6493% (0.6343%; 0.6159%) increase in adjusted SG&A expenses. This suggests that SG&A rises slightly more than proportionally during expansions. The coefficient β₂ is −0.2699 (−0.4198 **; −0.4471 *), consistently negative and significant in two specifications. This implies that when sales fall, cost reductions are smaller, reflecting asymmetric adjustment. The combined β₁ + β₂ = 0.3794 (0.2145; 0.1688) further confirms that SG&A declines much less than sales, consistent with stickiness. In high-OC firms, β₁ is 0.6575 *** (0.6475 ***; 0.6437 ***), showing that SG&A grows by about 0.65% when sales increase by 1%. This indicates a strong proportional relationship in expansions. The estimated β₂ is also negative and significant (−0.3772 **; −0.3794 **; −0.4029 **). This confirms stronger stickiness in high-OC firms, consistent with higher organizational rigidity. The combined β₁ + β₂ = 0.2803 (0.2681; 0.2408) confirms stronger stickiness in high-OC firms, as evidenced by the more negative β₂.

For large sales changes (see Table 13), the evidence of cost stickiness weakens considerably. In low-OC firms, β₁ is 0.5251 *** (0.5066 ***; 0.5098 ***), indicating proportional SG&A growth with sales. This suggests flexible scaling during larger expansions. The β₂ values (0.0774; 0.1376; 0.1135) are small and insignificant, indicating no clear cost asymmetry. This implies that large sales declines trigger symmetric cost adjustments. In high-OC firms, β₁ is 0.4167 *** (0.4096 ***; 0.4155 ***), meaning SG&A rises 0.4167% (0.4096%; 0.4155%) when sales increase 1%. This reflects a more moderate proportional response compared to smaller changes. The β₂ values (0.1170; −0.0741; −0.1199) vary in sign but remain insignificant. Thus, under large sales shifts, even high-OC firms adjust SG&A more symmetrically, with little evidence of stickiness.

Comparing across sales-change magnitudes, adjusted SG&A cost stickiness is statistically significant only for smaller changes (up to 10%) and largely disappears when changes range between 10% and 50%. This pattern holds for both low- and high-OC firms, but the evidence is more consistent and pronounced in the high-OC group.

In summary, adjusted SG&A cost stickiness is evident under smaller sales changes but weakens considerably as the magnitude of changes increases. These findings support Hypothesis H2, which posits that the impact of organizational capital on cost stickiness varies with the magnitude of sales changes. Under small fluctuations, both low and high-OC firms maintain SG&A expenditures asymmetrically, with the effect stronger in high-OC firms. In contrast, larger shifts require unavoidable cost adjustments that diminish stickiness.

4.2.3. Adjusted Organizational Capital and SG&A Cost Stickiness Across Sales Change Magnitudes

Table 14 and Table 15 report regression results using the logarithmic annual change in SG&A expenses as the dependent variable, dividing firms by their level of AOC and the magnitude of sales changes.

For small sales changes (see Table 14), the results provide strong evidence of cost stickiness. In low AOC firms, β₁ is 0.5490 *** (0.5437 ***; 0.5094 ***), suggesting that a 1% increase in sales is associated with a 0.5490% (0.5437%; 0.5094%) increase in SG&A expenses. This indicates that SG&A grows somewhat less than proportionally with sales, reflecting partial scale efficiency during expansions. The coefficient β₂ is −0.2481 *** (−0.2459 ***; −0.2491 ***), negative and significant across all specifications. This means that when sales fall, cost reductions are smaller than increases, consistent with sticky SG&A. The combined β₁ + β₂ = 0.3009 (0.2978; 0.2603) indicates that a 1% decline in sales results in a smaller reduction in SG&A expenses, confirming asymmetric cost behavior. In high AOC firms, β₁ is 0.5214 *** (0.5193 ***; 0.4931 ***), implying that a 1% sales increase leads to a 0.5214% (0.5193%; 0.4931%) rise in SG&A. This reflects a proportional but slightly smaller expansion response than low AOC firms. The coefficient β₂ is −0.1328 ** (−0.1058; −0.1988 ***), negative and significant in two of the three models. This points to some stickiness, though weaker than in the low AOC group. The combined β₁ + β₂ = 0.3886 (0.4135; 0.2943) likewise indicates smaller reductions during declines, but less pronounced than in low AOC firms.

For large sales changes (see Table 15), evidence of cost stickiness weakens considerably. For the low AOC group, β₁ is 0.4349 *** (0.4214 ***; 0.4063 ***), showing that SG&A rises by 0.4349% (0.4214%; 0.4063%) for a 1% sales increase. This suggests broadly proportional scaling in expansions. The β₂ values (0.0305; 0.0193; 0.0350) are small and statistically insignificant, indicating no strong asymmetry in cost adjustment. This means that large shocks appear to prompt more symmetric SG&A adjustments. For the high AOC group, β₁ is 0.4435 *** (0.4289 ***; 0.3951 ***), again indicating proportional SG&A growth with sales. This suggests managers adjust SG&A in line with larger sales movements. The β₂ values (−0.0336; −0.0555; −0.0401) are negative but not significant. Thus, neither group exhibits clear cost stickiness under larger sales changes.

Comparing across sales-change magnitudes, SG&A cost stickiness is statistically significant only for small changes (up to 10%) and largely disappears when changes range between 10% and 50%. This pattern holds for both low and high AOC firms. However, under small changes, the stickiness is more pronounced in the low AOC group, whereas it is weaker in the high AOC group.

In summary, SG&A cost stickiness is evident under smaller sales changes but weakens substantially when the magnitude of changes increases. The effect is stronger in the low AOC group when sales changes are small, while both low and high AOC firms show little evidence of stickiness under larger changes. These findings support Hypothesis H2, which posits that the impact of organizational capital on cost stickiness varies with the magnitude of sales changes. The results imply that cost asymmetry is more relevant when adjustments are marginal, while major shocks leave managers with few options but to adjust costs symmetrically.

4.2.4. Adjusted Organizational Capital and Adjusted SG&A Cost Stickiness Across Sales Change Magnitudes

Table 16 and Table 17 report regression results using the logarithmic annual change in adjusted SG&A expenses as the dependent variable, dividing firms by their level of AOC and the magnitude of sales changes.

For small sales changes (see Table 16), the results show strong evidence of cost stickiness. For the low AOC group, the estimated coefficient β₁ is 0.6277 *** (0.6145 ***; 0.5744 ***), suggesting that a 1% increase in sales is associated with a 0.6277% (0.6145%; 0.5744%) increase in adjusted SG&A expenses. This indicates SG&A grows proportionally with revenue in expansions. The coefficient β₂ is −0.2260 (−0.3370; −0.3172), negative but not significant, implying that cost reductions during declines are smaller, though statistical support is weak. This points to possible but not robust stickiness. The combined β₁ + β₂ = 0.4017 (0.2775; 0.2572) indicates that a 1% decline in sales results in a much smaller reduction in adjusted SG&A expenses, suggesting asymmetry in magnitude, but this is not statistically robust. For the high AOC group, β₁ is 0.6790 *** (0.6672 ***; 0.6700 ***), showing that a 1% increase in sales leads to 0.6790% (0.6672%; 0.6700%) rise in SG&A. This suggests stronger proportional growth during expansions. The coefficient β₂ is also negative and significant (−0.4224 ***; −0.4702 ***; −0.5041 ***). This confirms pronounced cost stickiness when sales decline. The combined β₁ + β₂ = 0.1973 (0.2055; 0.1715) confirms stronger stickiness for high AOC firms.

For large sales changes (see Table 17), evidence of stickiness weakens considerably. For the low AOC group, β₁ is 0.5000 *** (0.4877 ***; 0.4802 ***), indicating that SG&A rises by 0.5000% (0.4877%; 0.4802%) for a 1% increase in sales. This shows proportional adjustment in expansions. The β₂ values (0.1336; 0.2140; 0.2229) are small and insignificant, indicating no strong asymmetry in cost adjustment. This means that costs adjust symmetrically under large shocks. For the high AOC group, β₁ is 0.4437 *** (0.4383 ***; 0.4450 ***), suggesting SG&A rises 0.4437% (0.4383%; 0.4450%) when sales increase 1%. This reflects proportional scaling. The β₂ values (0.0599; −0.1424; −0.2074) are not significant. Thus, neither group displays meaningful stickiness when sales changes are relatively large.

Comparing across sales-change magnitudes, adjusted SG&A cost stickiness is statistically significant only when sales changes are small (up to 10%) and largely disappears when sales changes fall between 10% and 50%. This result holds for both low and high AOC firms, although stickiness appears stronger in high AOC firms when present.

In summary, adjusted SG&A cost stickiness is evident under smaller sales changes but weakens substantially as the magnitude of sales changes increases. These findings support Hypothesis H2, which posits that the impact of OC on cost stickiness varies with the magnitude of sales changes. Firms with higher AOC appear more inclined to retain resources and sustain spending under small sales shifts, but when sales shocks are large, cost flexibility dominates, and stickiness dissipates.

This section examined whether the impact of OC on cost stickiness varies with the magnitude of sales changes. Across all model specifications, the evidence indicates that cost stickiness in both SG&A and adjusted SG&A expenses is statistically significant when sales changes are relatively small (up to 10%), but it largely disappears when sales changes are more substantial (10–50%).

Specifically, both low- and high-OC firms exhibit asymmetric cost behavior under small sales changes. In most cases, the effect is generally stronger for firms with higher OC or AOC, as reflected by more negative β₂ coefficients, although in some adjusted SG&A specifications (Table 14) stickiness appears more pronounced in low AOC firms. This suggests that when sales fluctuations are modest, managers prefer to retain resources and maintain organizational capacity, especially in firms with higher OC, thereby sustaining higher SG&A. For larger sales changes, β₂ becomes statistically insignificant, indicating that costs adjust more symmetrically to sales increases and decreases regardless of OC, as firms are compelled to realign their cost structures under major shocks.

In summary, these results provide strong support for Hypothesis H2, which posits that the effect of OC on cost stickiness depends on the magnitude of sales changes. The findings suggest that OC reinforces cost asymmetry primarily during periods of small sales changes, whereas its influence weakens considerably or even diminishes when sales shocks are larger. This pattern supports the argument of Balakrishnan et al. (2004) that cost behavior is contingent on the scale of sales changes. Similarly to Venieris et al. (2015), we find stronger stickiness in high-OC firms, but unlike the U.S. context, Japanese firms with low OC also exhibit asymmetry, highlighting the role of institutional and cultural settings.

4.3. Robustness Checks

The main analyses are conducted using pooled OLS regressions, which implicitly assume the absence of unobserved firm- and year-specific effects. However, if such unobserved heterogeneity exists and is correlated with key variables, the pooled OLS estimates may be biased and inconsistent. F-test (see

Table A5. F test.

		Model	Comparison	Statistic	df1	df2	p-Value
SG&A cost stickiness by organizational capital	Low-OC group	Basic (1)	One-way vs. Pooled OLS	1.79	941	5419	2.69 × 10−36
			Two-way vs. One-way	24.83	9	5410	3.17 × 10−42
		ABJ (3)	One-way vs. Pooled OLS	1.77	934	5386	4.06 × 10−34
			Two-way vs. One-way	19.11	9	5377	8.14 × 10−32
		Extended (5)	One-way vs. Pooled OLS	1.77	934	5382	1.15 × 10−34
			Two-way vs. One-way	16.74	8	5374	9.11 × 10−25
	High-OC group	Basic (1)	One-way vs. Pooled OLS	1.63	891	5470	9.42 × 10−25
			Two-way vs. One-way	32.38	9	5461	6.15 × 10−56
		ABJ (3)	One-way vs. Pooled OLS	1.62	864	5369	1.65 × 10−23
			Two-way vs. One-way	25.57	9	5360	1.45 × 10−43
		Extended (5)	One-way vs. Pooled OLS	1.64	864	5365	1.80 × 10−24
			Two-way vs. One-way	24.06	8	5357	1.20 × 10−36
Adjusted SG&A cost stickiness by organizational capital	Low-OC group	Basic (2)	One-way vs. Pooled OLS	1.73	203	980	4.86 × 10−8
			Two-way vs. One-way	3.68	9	971	1.50 × 10−4
		ABJ (4)	One-way vs. Pooled OLS	1.74	201	970	4.40 × 10−8
			Two-way vs. One-way	3.40	9	961	4.03 × 10−4
		Extended (6)	One-way vs. Pooled OLS	1.76	201	966	1.96 × 10−8
			Two-way vs. One-way	3.54	8	958	4.69 × 10−4
	High-OC group	Basic (2)	One-way vs. Pooled OLS	1.65	188	1096	9.10 × 10−7
			Two-way vs. One-way	9.20	9	1087	1.55 × 10−13
		ABJ (4)	One-way vs. Pooled OLS	1.63	186	1084	1.69 × 10−6
			Two-way vs. One-way	8.15	9	1075	9.03 × 10−12
		Extended (6)	One-way vs. Pooled OLS	1.62	186	1080	2.62 × 10−6
			Two-way vs. One-way	8.52	8	1072	2.68 × 10−11
SG&A cost stickiness by adjusted organizational capital	Low-OC group	Basic (1)	One-way vs. Pooled OLS	1.76	982	5378	2.93 × 10−35
			Two-way vs. One-way	25.88	9	5369	4.10 × 10−44
		ABJ (3)	One-way vs. Pooled OLS	1.74	972	5345	1.93 × 10−33
			Two-way vs. One-way	20.45	9	5336	3.07 × 10−34
		Extended (5)	One-way vs. Pooled OLS	1.75	972	5341	1.06 × 10−33
			Two-way vs. One-way	18.69	8	5333	6.35 × 10−28
	High-OC group	Basic (1)	One-way vs. Pooled OLS	1.74	912	5449	3.92 × 10−32
			Two-way vs. One-way	28.73	9	5440	2.60 × 10−49
		ABJ (3)	One-way vs. Pooled OLS	1.72	884	5352	6.03 × 10−30
			Two-way vs. One-way	22.22	9	5343	1.86 × 10−37
		Extended (5)	One-way vs. Pooled OLS	1.75	884	5348	1.74 × 10−31
			Two-way vs. One-way	19.18	8	5340	1.04 × 10−28
Adjusted SG&A cost stickiness by adjusted organizational capital	Low-OC group	Basic (2)	One-way vs. Pooled OLS	1.60	213	1005	1.46 × 10−6
			Two-way vs. One-way	4.28	9	996	1.82 × 10−5
		ABJ (4)	One-way vs. Pooled OLS	1.54	210	995	1.14 × 10−5
			Two-way vs. One-way	4.11	9	986	3.32 × 10−5
		Extended (6)	One-way vs. Pooled OLS	1.57	210	991	5.51 × 10−6
			Two-way vs. One-way	4.16	8	983	6.68 × 10−5
	High-OC group	Basic (2)	One-way vs. Pooled OLS	1.80	188	1061	9.56 × 10−9
			Two-way vs. One-way	9.54	9	1052	4.54 × 10−14
		ABJ (4)	One-way vs. Pooled OLS	1.73	187	1049	1.02 × 10−7
			Two-way vs. One-way	8.53	9	1040	2.12 × 10−12
		Extended (6)	One-way vs. Pooled OLS	1.74	187	1045	6.52 × 10−8
			Two-way vs. One-way	8.84	8	1037	9.33 × 10−12
SG&A cost stickiness by organizational capital under small sales changes	Low-OC group	Basic (1)	One-way vs. Pooled OLS	1.43	892	3484	1.67 × 10−12
			Two-way vs. One-way	20.31	9	3475	1.14 × 10−33
		ABJ (3)	One-way vs. Pooled OLS	1.42	888	3468	3.19 × 10−12
			Two-way vs. One-way	17.72	9	3459	4.80 × 10−29
		Extended (5)	One-way vs. Pooled OLS	1.44	888	3464	8.67 × 10−13
			Two-way vs. One-way	16.05	8	3456	1.76 × 10−23
	High-OC group	Basic (1)	One-way vs. Pooled OLS	1.51	811	3566	3.19 × 10−15
			Two-way vs. One-way	22.96	9	3557	2.00 × 10−38
		ABJ (3)	One-way vs. Pooled OLS	1.54	796	3498	2.06 × 10−16
			Two-way vs. One-way	20.58	9	3489	3.78 × 10−34
		Extended (5)	One-way vs. Pooled OLS	1.57	796	3494	1.12 × 10−17
			Two-way vs. One-way	17.66	8	3486	4.63 × 10−26
SG&A cost stickiness by organizational capital under large sales changes	Low-OC group	Basic (1)	One-way vs. Pooled OLS	1.39	704	1234	3.02 × 10−7
			Two-way vs. One-way	7.24	9	1225	2.61 × 10−10
		ABJ (3)	One-way vs. Pooled OLS	1.37	697	1218	9.45 × 10−7
			Two-way vs. One-way	5.31	9	1209	3.92 × 10−7
		Extended (5)	One-way vs. Pooled OLS	1.35	697	1214	2.55 × 10−6
			Two-way vs. One-way	5.41	8	1206	1.02 × 10−6
	High-OC group	Basic (1)	One-way vs. Pooled OLS	1.26	763	1176	2.46 × 10−4
			Two-way vs. One-way	12.09	9	1167	2.08 × 10−18
		ABJ (3)	One-way vs. Pooled OLS	1.27	748	1158	1.37 × 10−4
			Two-way vs. One-way	6.18	9	1149	1.53 × 10−8
		Extended (5)	One-way vs. Pooled OLS	1.26	748	1154	1.82 × 10−4
			Two-way vs. One-way	6.88	8	1146	7.13 × 10−9
Adjusted SG&A cost stickiness by organizational capital under small sales changes	Low-OC group	Basic (2)	One-way vs. Pooled OLS	1.41	179	629	1.35 × 10−3
			Two-way vs. One-way	4.91	9	620	2.26 × 10−6
		ABJ (4)	One-way vs. Pooled OLS	1.38	178	623	2.85 × 10−3
			Two-way vs. One-way	4.58	9	614	7.10 × 10−6
		Extended (6)	One-way vs. Pooled OLS	1.41	178	619	1.60 × 10−3
			Two-way vs. One-way	4.77	8	611	1.07 × 10−5
	High-OC group	Basic (2)	One-way vs. Pooled OLS	1.72	172	750	8.55 × 10−7
			Two-way vs. One-way	9.07	9	741	4.49 × 10−13
		ABJ (4)	One-way vs. Pooled OLS	1.65	169	740	5.34 × 10−6
			Two-way vs. One-way	8.42	9	731	5.08 × 10−12
		Extended (6)	One-way vs. Pooled OLS	1.67	169	736	3.74 × 10−6
			Two-way vs. One-way	9.23	8	728	3.66 × 10−12
Adjusted SG&A cost stickiness by organizational capital under large sales changes	Low-OC group	Basic (2)	One-way vs. Pooled OLS	1.38	139	235	1.60 × 10−2
			Two-way vs. One-way	0.86	9	226	5.58 × 10−1
		ABJ (4)	One-way vs. Pooled OLS	1.33	138	229	2.77 × 10−2
			Two-way vs. One-way	0.72	9	220	6.87 × 10−1
		Extended (6)	One-way vs. Pooled OLS	1.28	138	225	4.87 × 10−2
			Two-way vs. One-way	0.81	8	217	5.93 × 10−1
	High-OC group	Basic (2)	One-way vs. Pooled OLS	1.50	137	206	4.07 × 10−3
			Two-way vs. One-way	2.29	9	197	1.85 × 10−2
		ABJ (4)	One-way vs. Pooled OLS	1.49	137	201	4.74 × 10−3
			Two-way vs. One-way	1.90	9	192	5.37 × 10−2
		Extended (6)	One-way vs. Pooled OLS	1.48	137	197	5.97 × 10−3
			Two-way vs. One-way	2.04	8	189	4.34 × 10−2
SG&A cost stickiness by adjusted organizational capital under small sales changes	Low-OC group	Basic (1)	One-way vs. Pooled OLS	1.46	927	3449	2.15 × 10−14
			Two-way vs. One-way	19.88	9	3440	6.71 × 10−33
		ABJ (3)	One-way vs. Pooled OLS	1.46	921	3430	6.17 × 10−14
			Two-way vs. One-way	17.10	9	3421	6.22 × 10−28
		Extended (5)	One-way vs. Pooled OLS	1.45	921	3426	1.02 × 10−13
			Two-way vs. One-way	16.15	8	3418	1.23 × 10−23
	High-OC group	Basic (1)	One-way vs. Pooled OLS	1.55	818	3559	5.07 × 10−17
			Two-way vs. One-way	22.11	9	3550	6.60 × 10−37
		ABJ (3)	One-way vs. Pooled OLS	1.59	802	3497	1.02 × 10−18
			Two-way vs. One-way	20.18	9	3488	1.95 × 10−33
		Extended (5)	One-way vs. Pooled OLS	1.63	802	3493	1.31 × 10−20
			Two-way vs. One-way	16.64	8	3485	2.00 × 10−24
SG&A cost stickiness by adjusted organizational capital under large sales changes	Low-OC group	Basic (1)	One-way vs. Pooled OLS	1.51	714	1224	2.26 × 10−10
			Two-way vs. One-way	7.21	9	1215	2.96 × 10−10
		ABJ (3)	One-way vs. Pooled OLS	1.48	708	1209	1.62 × 10−9
			Two-way vs. One-way	5.66	9	1200	1.05 × 10−7
		Extended (5)	One-way vs. Pooled OLS	1.46	708	1205	6.00 × 10−9
			Two-way vs. One-way	5.88	8	1197	2.13 × 10−7
	High-OC group	Basic (1)	One-way vs. Pooled OLS	1.30	769	1170	2.56 × 10−5
			Two-way vs. One-way	12.97	9	1161	7.14 × 10−20
		ABJ (3)	One-way vs. Pooled OLS	1.32	753	1151	1.09 × 10−5
			Two-way vs. One-way	6.66	9	1142	2.47 × 10−9
		Extended (5)	One-way vs. Pooled OLS	1.31	753	1147	2.12 × 10−5
			Two-way vs. One-way	6.97	8	1139	5.24 × 10−9
Adjusted SG&A cost stickiness by adjusted organizational capital under small sales changes	Low-OC group	Basic (2)	One-way vs. Pooled OLS	1.31	188	637	9.41 × 10−3
			Two-way vs. One-way	4.81	9	628	3.11 × 10−6
		ABJ (4)	One-way vs. Pooled OLS	1.28	186	631	1.49 × 10−2
			Two-way vs. One-way	4.49	9	622	9.97 × 10−6
		Extended (6)	One-way vs. Pooled OLS	1.29	186	627	1.32 × 10−2
			Two-way vs. One-way	4.85	8	619	8.23 × 10−6
	High-OC group	Basic (2)	One-way vs. Pooled OLS	1.74	171	734	5.00 × 10−7
			Two-way vs. One-way	9.83	9	725	2.76 × 10−14
		ABJ (4)	One-way vs. Pooled OLS	1.66	169	724	5.03 × 10−6
			Two-way vs. One-way	9.24	9	715	2.49 × 10−13
		Extended (6)	One-way vs. Pooled OLS	1.68	169	720	3.07 × 10−6
			Two-way vs. One-way	10.15	8	712	1.78 × 10−13
Adjusted SG&A cost stickiness by adjusted organizational capital under large sales changes	Low-OC group	Basic (2)	One-way vs. Pooled OLS	1.50	153	231	2.67 × 10−3
			Two-way vs. One-way	1.24	9	222	2.70 × 10−1
		ABJ (4)	One-way vs. Pooled OLS	1.46	152	225	4.86 × 10−3
			Two-way vs. One-way	1.03	9	216	4.16 × 10−1
		Extended (6)	One-way vs. Pooled OLS	1.39	152	221	1.33 × 10−2
			Two-way vs. One-way	1.03	8	213	4.12 × 10−1
	High-OC group	Basic (2)	One-way vs. Pooled OLS	1.49	136	197	5.18 × 10−3
			Two-way vs. One-way	2.35	9	188	1.57 × 10−2
		ABJ (4)	One-way vs. Pooled OLS	1.46	136	192	7.79 × 10−3
			Two-way vs. One-way	1.96	9	183	4.57 × 10−2
		Extended (6)	One-way vs. Pooled OLS	1.44	136	188	1.06 × 10−2
			Two-way vs. One-way	2.13	8	180	3.54 × 10−2

Note: The F test confirms the appropriateness of the fixed effects specification.

) further rejects the null hypothesis that pooled OLS is sufficient, supporting the use of fixed effects. To address this concern and enhance the robustness and causal interpretation of the results, we therefore estimate two-way fixed effects models, which absorb both firm- and year-specific unobservables.

4.3.1. The Impact of Organizational Capital on Cost Stickiness (Fixed Effects Models)

Table 18. SG&A cost stickiness by organizational capital (fixed effects models).

	Low OC Group			High OC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.3770 ***	0.3772 ***	0.3732 ***	0.4158 ***	0.4176 ***	0.4116 ***
	(0.0187)	(0.0187)	(0.0198)	(0.0205)	(0.0207)	(0.0208)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.0534	−0.0577	−0.0797	−0.1345 **	−0.1239 **	−0.1669 ***
	(0.0580)	(0.0685)	(0.0737)	(0.0548)	(0.0572)	(0.0593)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.2550	−0.2131		−0.3177	−0.2775
		(0.2853)	(0.2891)		(0.2453)	(0.2462)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.2823 ***	0.2412 **		0.0566	0.0441
		(0.0798)	(0.1201)		(0.0778)	(0.1219)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0529 *	−0.0517 *		−0.0455 *	−0.0475 *
		(0.0300)	(0.0304)		(0.0248)	(0.0251)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			2.74 × 10−5			6.82 × 10−6 *
			(1.95 × 10−5)			(4.09 × 10−5)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0085			0.0115
			(0.0098)			(0.0101)
β8: ${Ds}_{i,t}$			−0.0009			−0.0003
			(0.0019)			(0.0018)
β10: ${FCF}_{i,t}$			−4.8 × 10−7 ***			−4.57 × 10−8 *
			(1.79 × 10−7)			(2.5 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0036 ***			−0.0015 **
			(0.0010)			(0.0008)
β12: ${Size}_{i,t}$			0.0289 ***			0.0383 ***
			(0.0079)			(0.0088)
R-squared (within)	0.38793	0.38942	0.39113	0.37536	0.37568	0.37263
S.E. of regression	0.02255	0.02251	0.02233	0.02025	0.02024	0.02008
F-statistic	78.813 ***	62.311 ***	43.490 ***	70.093 ***	55.199 ***	36.908 ***
Observations	6363	6363	6303	6364	6364	6214

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

,

Table 19. Adjusted SG&A cost stickiness by organizational capital (fixed effects models).

	Low OC Group			High OC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.3684 ***	0.3725 ***	0.3713 ***	0.4171 ***	0.4156 ***	0.4169 ***
	(0.0394)	(0.0399)	(0.0361)	(0.0431)	(0.0432)	(0.0448)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.2062	0.2330	0.2201	−0.0875	−0.0511	−0.0452
	(0.1388)	(0.1746)	(0.2277)	(0.1086)	(0.1186)	(0.1249)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.9724	−0.8829		−0.8828	−0.7347
		(0.8163)	(0.9083)		(0.5693)	(0.5635)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.3668 *	−0.0374		0.3272 **	0.2697
		(0.2038)	(0.3039)		(0.1483)	(0.1923)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0425	−0.1094		−0.1011 **	−0.1113 **
		(0.0844)	(0.0855)		(0.0498)	(0.0521)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			0.0001 *			−2.47 × 10−6
			(6.03 × 10−5)			(7.69 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			0.0313			−0.0057
			(0.0327)			(0.0190)
β8: ${Ds}_{i,t}$			−0.0096 *			−0.0003
			(0.0050)			(0.0026)
β10: ${FCF}_{i,t}$			−2.8 × 10−7			−6.84 × 10−8
			(3.13 × 10−7)			(4.56 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0084 ***			−0.0032 **
			(0.0024)			(0.0014)
β12: ${Size}_{i,t}$			0.0165			0.0282 *
			(0.0117)			(0.0165)
R-squared (within)	0.41172	0.41420	0.43639	0.42211	0.42611	0.42664
S.E. of regression	0.02225	0.02217	0.02150	0.01803	0.01795	0.01780
F-statistic	19.830 ***	15.801 ***	11.860 ***	17.898 ***	14.331 ***	9.9769 ***
Observations	1186	1186	1169	1287	1287	1275

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

,

Table 20. SG&A cost stickiness by adjusted organizational capital (fixed effects models).

	Low AOC Group			High AOC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.3840 ***	0.3845 ***	0.3747 ***	0.4217 ***	0.4226 ***	0.4179 ***
	(0.0185)	(0.0186)	(0.0195)	(0.0214)	(0.0215)	(0.0218)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.0830	−0.0748	−0.0765	−0.1350 **	−0.1323 **	−0.1676 ***
	(0.0585)	(0.0746)	(0.0779)	(0.0556)	(0.0582)	(0.0604)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.3082	−0.3297		−0.1243	−0.0768
		(0.3193)	(0.3242)		(0.2570)	(0.2630)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.2895 ***	0.1849		0.0366	0.0362
		(0.0799)	(0.1224)		(0.0715)	(0.1237)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0569 *	−0.0572 *		−0.0276	−0.0298
		(0.0308)	(0.0312)		(0.0244)	(0.0246)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			2.45 × 10−5			5.65 × 10−6
			(1.49 × 10−5)			(3.85 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0120			0.0119
			(0.0095)			(0.0100)
β8: ${Ds}_{i,t}$			−0.0022			−0.0002
			(0.0020)			(0.0019)
β10: ${FCF}_{i,t}$			−9.15 × 10−8			−5.09 × 10−8 **
			(8.38 × 10−8)			(2.45 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0037 ***			−0.0016 **
			(0.0010)			(0.0007)
β12: ${Size}_{i,t}$			0.0338 ***			0.0329 ***
			(0.0066)			(0.0084)
R-squared (within)	0.38485	0.38636	0.38760	0.38622	0.38608	0.38267
S.E. of regression	0.02284	0.02281	0.02265	0.01964	0.01963	0.01946
F-statistic	82.692 ***	65.381 ***	45.394 ***	75.089 ***	59.039 ***	39.407 ***
Observations	6363	6363	6302	6364	6364	6215

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

and

Table 21. Adjusted SG&A cost stickiness by adjusted organizational capital (fixed effects models).

	Low AOC Group			High AOC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.3727 ***	0.3812 ***	0.3842 ***	0.4365 ***	0.4345 ***	0.4351 ***
	(0.0383)	(0.0388)	(0.0346)	(0.0443)	(0.0445)	(0.0472)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.2087	0.3519 *	0.3276	−0.1234	−0.1581	−0.1765
	(0.1439)	(0.1856)	(0.2191)	(0.1183)	(0.1245)	(0.1283)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−1.959 **	−1.626 *		0.3396	0.5583
		(0.7958)	(0.9159)		(0.6997)	(0.6676)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.2800	−0.1342		0.3547 **	0.2605
		(0.2020)	(0.3086)		(0.1504)	(0.2038)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0107	−0.0595		−0.0512	−0.0538
		(0.0829)	(0.0857)		(0.0469)	(0.0489)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			8.16 × 10−6			8.29 × 10−6
			(2.54 × 10−5)			(7.82 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			0.0322			0.0015
			(0.0351)			(0.0200)
β8: ${Ds}_{i,t}$			−0.0103 **			−0.0005
			(0.0052)			(0.0026)
β10: ${FCF}_{i,t}$			−2.08 × 10−7 *			−6.07 × 10−8
			(1.09 × 10−7)			(5.15 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0074 ***			−0.0031 **
			(0.0027)			(0.0014)
β12: ${Size}_{i,t}$			0.0099			0.0216
			(0.0127)			(0.0170)
R-squared (within)	0.39997	0.40634	0.41458	0.44299	0.44471	0.44572
S.E. of regression	0.02243	0.02228	0.02189	0.01750	0.01745	0.01730
F-statistic	20.165 ***	16.272 ***	11.602 ***	19.396 ***	15.411 ***	10.821 ***
Observations	1221	1221	1203	1252	1252	1241

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

present the results of the two-way fixed effects estimations. Across all specifications, the sales elasticity β₁ is positive and highly significant for both low- and high-OC (or AOC) firms (approximately 0.37 to 0.42), confirming that SG&A expenses co-move with sales even after controlling for unobserved firm and year effects. The stickiness coefficient β₂, defined as the interaction for sales declines, shows clear group differences. In the high-OC and high-AOC groups, β₂ is negative and statistically significant in most specifications (approximately −0.12 to −0.17 in Table 18 and Table 20). This implies that when sales fall by 10 percent, SG&A expenses in high-OC firms decline by about 1.2 to 1.7 percentage points less than in comparable low-OC firms, reflecting stronger downward rigidity. By contrast, in the low-OC and low-AOC groups, β₂ is generally small and insignificant, suggesting more symmetric adjustments.

Compared with the pooling models, the overall pattern remains consistent: high OC (or AOC) firms exhibit systematically stronger cost stickiness, while low OC (or AOC) firms display weaker or insignificant asymmetry. Although the magnitudes of β₂ are somewhat smaller than in the pooling regressions, their signs and statistical significance remain broadly stable after accounting for firm and year fixed effects. Control variables behave largely as expected, leverage is typically negative and significant, size is often positive and significant, and free cash flow is negative in several specifications. Hausman tests are consistently significant at the 1 percent level, confirming the appropriateness of the fixed effects specification.

Overall, these results confirm that the effect of organizational capital on cost stickiness is robust, providing further support for Hypothesis H1. These findings remain robust after controlling for firm and year fixed effects, consistent with prior studies emphasizing the persistence of stickiness (Banker & Chen, 2006; Banker & Byzalov, 2014). Moreover, they provide new evidence from the Japanese context, where employment protection and long-term orientation further reinforce the effect of organizational capital.

4.3.2. The Impact of Organizational Capital on Cost Stickiness Based on the Magnitude of Sales Changes (Fixed Effects Models)

Table 22. SG&A cost stickiness by organizational capital under small sales changes (fixed effects models).

	Low OC Group			High OC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.4280 ***	0.4282 ***	0.4323 ***	0.4198 ***	0.4238 ***	0.4168 ***
	(0.0364)	(0.0364)	(0.0380)	(0.0301)	(0.0305)	(0.0310)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.1021	−0.0924	−0.1350	−0.0551	−0.0285	−0.1027
	(0.0841)	(0.0902)	(0.0963)	(0.0773)	(0.0799)	(0.0844)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.2011	−0.2386		0.0882	0.1016
		(0.3703)	(0.3853)		(0.3641)	(0.3566)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.1517	0.2148		−0.0994	−0.2435
		(0.1161)	(0.1851)		(0.1150)	(0.1683)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0389	−0.0424		−0.0310	−0.0155
		(0.0447)	(0.0445)		(0.0395)	(0.0396)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			3.65 × 10−5			1.55 × 10−5 **
			(2.43 × 10−5)			(6.03 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0094			0.0009
			(0.0107)			(0.0104)
β8: ${Ds}_{i,t}$			0.0005			−0.0015
			(0.0020)			(0.0019)
β10: ${FCF}_{i,t}$			−2.94 × 10−7 **			−7.19 × 10−8 ***
			(1.19 × 10−7)			(2.28 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0031 **			−0.0023 ***
			(0.0013)			(0.0009)
β12: ${Size}_{i,t}$			0.0265 ***			0.0460 ***
			(0.0079)			(0.0086)
R-squared (within)	0.24537	0.24529	0.25090	0.26308	0.26281	0.27899
S.E. of regression	0.01782	0.01781	0.01774	0.01620	0.01620	0.01590
F-statistic	54.291 ***	42.737 ***	30.347 ***	49.441 ***	38.892 ***	28.227 ***
Observations	4379	4379	4349	4380	4380	4284

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

,

Table 23. SG&A cost stickiness by organizational capital under large sales changes (fixed effects models).

	Low OC Group			High OC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.3964 ***	0.3936 ***	0.3811 ***	0.4486 ***	0.4510 ***	0.4445 ***
	(0.0466)	(0.0462)	(0.0469)	(0.0418)	(0.0423)	(0.0439)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.1030	−0.0866	−0.0842	−0.2890 ***	−0.2921 ***	−0.2812 ***
	(0.1085)	(0.1206)	(0.1219)	(0.0946)	(0.0984)	(0.1040)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.3618	−0.2570		0.3198	0.4412 *
		(0.2929)	(0.2825)		(0.2173)	(0.2400)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.1351 *	−0.6300 **		0.0857	−0.2419
		(0.0736)	(0.2879)		(0.0742)	(0.3176)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0116	−0.0089		−0.0544 ***	−0.0588 ***
		(0.0270)	(0.0260)		(0.0208)	(0.0214)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			1.54 × 10−5			2.47 × 10−6
			(2.53 × 10−5)			(3.32 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0214			0.0245
			(0.0211)			(0.0244)
β8: ${Ds}_{i,t}$			−0.0450 ***			−0.0210
			(0.0173)			(0.0196)
β10: ${FCF}_{i,t}$			−8.95 × 10−7			1.54 × 10−8
			(5.52 × 10−7)			(5.91 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0026			−0.0012
			(0.0024)			(0.0022)
β12: ${Size}_{i,t}$			0.0334 **			0.0207
			(0.0145)			(0.0171)
R-squared (within)	0.45354	0.45465	0.45882	0.47979	0.48443	0.47150
S.E. of regression	0.02645	0.02639	0.02606	0.02226	0.02213	0.02214
F-statistic	121.48 ***	96.101 ***	67.665 ***	152.41 ***	121.77 ***	80.783 ***
Observations	1941	1941	1912	1942	1942	1905

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

,

Table 24. Adjusted SG&A cost stickiness by organizational capital under small sales changes (fixed effects models).

	Low OC Group			High OC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.5194 ***	0.5194 ***	0.5174 ***	0.5943 ***	0.5878 ***	0.5767 ***
	(0.0767)	(0.0763)	(0.0778)	(0.0626)	(0.0623)	(0.0643)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.0886	−0.2560	−0.2890	−0.3938 **	−0.3883 *	−0.3306
	(0.2097)	(0.2669)	(0.2766)	(0.1918)	(0.2125)	(0.2251)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.0336	0.1154		−0.8090	−0.5751
		(0.8142)	(0.7795)		(0.8208)	(0.7556)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.6906 ***	0.7779 *		0.3286	0.1656
		(0.2484)	(0.4458)		(0.2975)	(0.3940)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		0.0109	0.0081		−0.0651	−0.0902
		(0.1258)	(0.1240)		(0.0840)	(0.0874)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			1.2 × 10−5			−1.41 × 10−8
			(5.02 × 10−5)			(9.9 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			0.0292			0.0046
			(0.0257)			(0.0159)
β8: ${Ds}_{i,t}$			3.78 × 10−5			−0.0013
			(0.0051)			(0.0032)
β10: ${FCF}_{i,t}$			−1.62 × 10−7			−1.13 × 10−7 ***
			(2.29 × 10−7)			(4.28 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0080 ***			−0.0037 ***
			(0.0029)			(0.0014)
β12: ${Size}_{i,t}$			0.0167			0.0244 *
			(0.0191)			(0.0140)
R-squared (within)	0.29933	0.30425	0.30761	0.36289	0.36372	0.36145
S.E. of regression	0.01657	0.01647	0.01626	0.01429	0.01425	0.01404
F-statistic	14.069 ***	11.340 ***	8.1006 ***	14.968 ***	11.889 ***	8.1660 ***
Observations	811	811	802	925	925	914

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

,

Table 25. Adjusted SG&A cost stickiness by organizational capital under large sales changes (fixed effects models).

	Low OC Group			High OC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.3094 ***	0.2937 ***	0.2969 ***	0.4647 ***	0.4660 ***	0.4620 ***
	(0.0956)	(0.0944)	(0.0943)	(0.1056)	(0.1053)	(0.1167)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.3300	0.2760	0.3183	−0.3334	−0.4394 *	−0.4644 *
	(0.2121)	(0.2399)	(0.2482)	(0.2291)	(0.2487)	(0.2602)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.2118	−0.1087		−0.0888	−0.0694
		(0.6292)	(0.6226)		(0.3536)	(0.4085)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.4623 **	−0.6023		0.3381 **	0.5055
		(0.2223)	(0.8542)		(0.1639)	(0.3913)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.1058	−0.0758		−0.0608	−0.0628
		(0.0719)	(0.0713)		(0.0551)	(0.0527)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			6.96 × 10−5			−4.67 × 10−6
			(5.79 × 10−5)			(7.61 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			0.0042			−0.0567
			(0.0488)			(0.0436)
β8: ${Ds}_{i,t}$			−0.0529			0.0113
			(0.0498)			(0.0235)
β10: ${FCF}_{i,t}$			−5.31 × 10−7			−1.81 × 10−7
			(8.96 × 10−7)			(1.32 × 10−7)
β11: ${Leverage}_{i,t}$			−0.0021			0.0032
			(0.0043)			(0.0079)
β12: ${Size}_{i,t}$			0.0218			0.0165
			(0.0329)			(0.0489)
R-squared (within)	0.45932	0.46800	0.46551	0.51694	0.52581	0.52112
S.E. of regression	0.02915	0.02872	0.02844	0.01911	0.01879	0.01857
F-statistic	26.247 ***	21.538 ***	15.368 ***	32.698 ***	26.913 ***	19.463 ***
Observations	377	377	372	346	346	345

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

,

Table 26. SG&A cost stickiness by adjusted organizational capital under small sales changes (fixed effects models).

	Low AOC Group			High AOC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.4274 ***	0.4289 ***	0.4227 ***	0.4185 ***	0.4204 ***	0.4222 ***
	(0.0368)	(0.0368)	(0.0383)	(0.0296)	(0.0299)	(0.0307)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.1222	−0.1245	−0.1253	−0.0563	−0.0429	−0.1402 *
	(0.0854)	(0.0948)	(0.0996)	(0.0754)	(0.0776)	(0.0825)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		0.0646	−0.0098		−0.0761	−0.0617
		(0.3930)	(0.4271)		(0.3762)	(0.3642)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.1675	0.0761		−0.1181	−0.1224
		(0.1196)	(0.1911)		(0.1113)	(0.1631)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0658	−0.0680		0.0018	0.0141
		(0.0446)	(0.0445)		(0.0391)	(0.0393)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			1.34 × 10−5			1.95 × 10−5 ***
			(1.95 × 10−5)			(5.94 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0135			0.0041
			(0.0106)			(0.0102)
β8: ${Ds}_{i,t}$			−0.0013			8.85 × 10−5
			(0.0021)			(0.0017)
β10: ${FCF}_{i,t}$			−1.25 × 10−7 **			−6.53 × 10−8 ***
			(5.83 × 10−8)			(2.45 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0043 ***			−0.0018 **
			(0.0014)			(0.0008)
β12: ${Size}_{i,t}$			0.0301 ***			0.0389 ***
			(0.0081)			(0.0079)
R-squared (within)	0.24601	0.24621	0.25154	0.27537	0.27502	0.29051
S.E. of regression	0.01794	0.01793	0.01786	0.01582	0.01582	0.01553
F-statistic	57.972 ***	45.678 ***	32.305 ***	52.224 ***	41.070 ***	29.663 ***
Observations	4379	4379	4347	4380	4380	4286

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

,

Table 27. SG&A cost stickiness by adjusted organizational capital under large sales changes (fixed effects models).

	Low AOC Group			High AOC Group
Variable	Basic (1)	ABJ (3)	Extended (5)	Basic (1)	ABJ (3)	Extended (5)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.3881 ***	0.3853 ***	0.3657 ***	0.4175 ***	0.4247 ***	0.4199 ***
	(0.0445)	(0.0440)	(0.0450)	(0.0420)	(0.0424)	(0.0440)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.1067	−0.1217	−0.0942	−0.2292 **	−0.2236 **	−0.2216 **
	(0.1058)	(0.1203)	(0.1194)	(0.0936)	(0.0974)	(0.1021)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.1539	−0.1245		0.2129	0.3089
		(0.2779)	(0.2685)		(0.2376)	(0.2639)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.1669 **	−0.6655 **		0.0631	−0.1173
		(0.0771)	(0.3068)		(0.0724)	(0.2840)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0077	−0.0035		−0.0624 ***	−0.0665 ***
		(0.0276)	(0.0268)		(0.0196)	(0.0201)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			2.05 × 10−5 *			9.69 × 10−7
			(1.05 × 10−5)			(3.51 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0151			0.0092
			(0.0218)			(0.0257)
β8: ${Ds}_{i,t}$			−0.0487 ***			−0.0114
			(0.0183)			(0.0175)
β10: ${FCF}_{i,t}$			3.32 × 10−8			−1.73 × 10−8
			(2.47 × 10−7)			(5.76 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0036			0.0004
			(0.0025)			(0.0022)
β12: ${Size}_{i,t}$			0.0301 *			0.0296 *
			(0.0154)			(0.0171)
R-squared (within)	0.46521	0.46625	0.46860	0.50039	0.50484	0.49259
S.E. of regression	0.02606	0.02600	0.02572	0.02170	0.02158	0.02157
F-statistic	128.89 ***	101.94 ***	71.566 ***	164.03 ***	131.01 ***	86.901 ***
Observations	1941	1941	1914	1942	1942	1903

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

,

Table 28. Adjusted SG&A cost stickiness by adjusted organizational capital under small sales changes (fixed effects models).

	Low AOC Group			High AOC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.4837 ***	0.4932 ***	0.4828 ***	0.6165 ***	0.6064 ***	0.6044 ***
	(0.0828)	(0.0816)	(0.0835)	(0.0560)	(0.0566)	(0.0591)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	−0.0258	−0.1529	−0.1635	−0.4230 **	−0.4702 **	−0.4335 **
	(0.2253)	(0.2937)	(0.2917)	(0.1802)	(0.1997)	(0.2120)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.2735	−0.0826		−0.5741	−0.3280
		(0.8843)	(0.8579)		(0.8802)	(0.8151)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.8574 ***	0.8408 *		0.2592	0.2028
		(0.2735)	(0.4665)		(0.2722)	(0.3759)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0761	−0.0841		0.0132	−0.0049
		(0.1285)	(0.1264)		(0.0750)	(0.0794)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			4 × 10−6			2.92 × 10−6
			(2.91 × 10−5)			(1.01 × 10−5)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			0.0303			−0.0010
			(0.0256)			(0.0183)
β8: ${Ds}_{i,t}$			−0.0013			0.0003
			(0.0054)			(0.0033)
β10: ${FCF}_{i,t}$			−1.68 × 10−7			−1.23 × 10−7 ***
			(1.24 × 10−7)			(4.61 × 10−8)
β11: ${Leverage}_{i,t}$			−0.0092 ***			−0.0031 ***
			(0.0025)			(0.0011)
β12: ${Size}_{i,t}$			0.0212			0.0203
			(0.0185)			(0.0144)
R-squared (within)	0.27139	0.27887	0.28814	0.38816	0.38773	0.38329
S.E. of regression	0.01713	0.01700	0.01673	0.01365	0.01363	0.01344
F-statistic	13.799 ***	11.211 ***	8.0968 ***	16.241 ***	12.847 ***	8.8324 ***
Observations	828	828	818	908	908	898

* Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

and

Table 29. Adjusted SG&A cost stickiness by adjusted organizational capital under large sales changes (fixed effects models).

	Low AOC Group			High AOC Group
Variable	Basic (2)	ABJ (4)	Extended (6)	Basic (2)	ABJ (4)	Extended (6)
β1: $\log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.3308 ***	0.3225 ***	0.3193 ***	0.4676 ***	0.4660 ***	0.4621 ***
	(0.0895)	(0.0864)	(0.0922)	(0.1118)	(0.1140)	(0.1210)
β2: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)$	0.2830	0.2752	0.2952	−0.2883	−0.3817	−0.4038
	(0.2182)	(0.2583)	(0.2717)	(0.2458)	(0.2714)	(0.2653)
β3: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times \log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$		−0.5562	−0.5555		0.0546	0.0907
		(0.5418)	(0.5417)		(0.3846)	(0.4531)
β4: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {Ds}_{i,t}$		0.4762 **	−0.5967		0.2291	0.4039
		(0.2142)	(0.8676)		(0.1832)	(0.4747)
β5: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {GNI}_{i,t}$		−0.0927	−0.0674		−0.0770	−0.0821
		(0.0739)	(0.0747)		(0.0567)	(0.0559)
β6: $D_{i,t}\times \log \left({\displaystyle \frac{{Sales}_{i,t}}{{Sales}_{i,t-1}}}\right)\times {FCF}_{i,t}$			4.09 × 10−5			−6.59 × 10−6
			(4.48 × 10−5)			(8.47 × 10−6)
β7: $\log \left({\displaystyle \frac{{Assets}_{i,t}}{{Sales}_{i,t}}}\right)$			−0.0192			−0.0501
			(0.0422)			(0.0540)
β8: ${Ds}_{i,t}$			−0.0578			0.0115
			(0.0513)			(0.0283)
β10: ${FCF}_{i,t}$			−7.81 × 10−8			−1.73 × 10−7
			(3.71 × 10−7)			(1.3 × 10−7)
β11: ${Leverage}_{i,t}$			−0.0032			0.0042
			(0.0041)			(0.0077)
β12: ${Size}_{i,t}$			0.0166			−0.0034
			(0.0308)			(0.0586)
R-squared (within)	0.49084	0.50161	0.49708	0.52117	0.52547	0.51762
S.E. of regression	0.02712	0.02666	0.02644	0.01933	0.01909	0.01892
F-statistic	33.589 ***	27.721 ***	19.687 ***	33.646 ***	27.356 ***	19.651 ***
Observations	387	387	382	336	336	335

** Significance at the 5% level. *** Significance at the 1% level. Note: Hausman test results (p-value < 0.001 for all models) consistently support the fixed effects specification.

present the two-way fixed effects results when firms are divided by both organizational capital (OC or AOC) and the magnitude of sales changes. Across specifications, the sales elasticity β₁ remains positive and highly significant for all groups, confirming that SG&A expenses consistently co-move with sales after controlling for firm and year effects.

For small sales changes (up to 10%), the stickiness coefficient β₂ is negative in both low- and high-OC (or AOC) groups. In high-OC and high-AOC firms, β₂ is statistically significant in most specifications, with magnitudes around −0.12 to −0.17 (Table 22, Table 24 and Table 26). This implies that when sales decline by 10 percent, SG&A falls by 1.2 to 1.7 percent less than in comparable low-OC firms, indicating stronger downward rigidity. In low-OC groups, β₂ is generally smaller and often insignificant, suggesting weaker asymmetry. These findings mirror the pooling models and remain robust once firm- and year-specific effects are controlled.

For large sales changes (10–50%), however, the evidence of stickiness weakens substantially. In most cases, β₂ turns statistically insignificant, regardless of OC or AOC levels (Table 23, Table 25, Table 27 and Table 29). This indicates that under major sales shocks, SG&A adjusts more symmetrically, as managers are compelled to realign costs regardless of organizational capital intensity.

Relative to the pooling regressions, the fixed effects estimates show a broadly consistent pattern: cost stickiness appears only under small sales fluctuations, especially in high-OC or high-AOC firms, but it diminishes under large changes. Hausman tests support the fixed effects specification at the 1 percent level, reinforcing robustness.

In summary, these results provide strong support for Hypothesis H2. They show that organizational capital reinforces asymmetric cost behavior under small sales changes, but its effect vanishes under large shocks, where cost flexibility dominates. This pattern is consistent with Balakrishnan et al. (2004), who emphasized the role of adjustment scale, but our results differ from Venieris et al. (2015) by showing that in Japan, even low-OC firms exhibit some degree of stickiness, reflecting institutional and cultural characteristics.

The fixed effects estimates are consistent with the earlier OLS results, but they offer greater robustness by absorbing firm- and year-fixed effects. This design reduces concerns of omitted variable bias and enhances the credibility of the interpretation of organizational capital’s impact on cost stickiness. In addition, the results show that stickiness is most evident under small sales changes but weakens or disappears when sales shifts are large. This pattern is consistent with the mechanism of resource specificity and adjustment frictions: firms with higher organizational capital rely on specialized routines and relationship-specific resources, which encourages managers to maintain SG&A during modest fluctuations, but when sales shocks are substantial, the pressure to realign costs overrides these frictions, leading to more symmetric adjustments.

5. Conclusions

5.1. Summary of Findings

This study investigates the relationship between organizational capital and cost stickiness among Japanese firms. The analysis focuses on companies listed on the Prime and Standard sections of the Tokyo Stock Exchange that adopt Japanese GAAP as of the end of March 2025. It covers 18 fiscal periods from FY2007 to FY2024, following the introduction of Japan’s quarterly reporting system.

The regression results across Table 6, Table 7, Table 8, Table 9, Table 10, Table 11, Table 12, Table 13, Table 14, Table 15, Table 16 and Table 17 reveal consistent patterns in SG&A and adjusted SG&A expenses relative to OC and AOC.

First, the coefficients on sales increases (β₁) are uniformly positive and statistically significant at the 1% level across all specifications, ranging from approximately 0.45 to 0.67. This indicates that both SG&A and adjusted SG&A expenses scale proportionally with sales growth, regardless of OC.

Second, the asymmetry coefficients (β₂) provide robust evidence of cost stickiness under small sales changes (≤10%). For both SG&A and adjusted SG&A, β₂ is negative and statistically significant in most specifications, particularly for firms with high OC or AOC. For instance, β₂ is approximately −0.49 and highly significant for SG&A against OC, while for adjusted SG&A under both OC and AOC specifications, β₂ remains consistently negative, around −0.47, and significant at the 1% level. These results indicate that costs decline less during sales reductions than they rise during expansions of similar magnitude. The combined coefficient (β₁ + β₂) is therefore substantially smaller, confirming asymmetric cost behavior.

By contrast, when sales changes are larger (10–50%), β₂ becomes statistically insignificant across all models, showing that cost asymmetry largely disappears under substantial economic fluctuations.

Comparisons between low- and high-capital firms reinforce this pattern. β₂ is more negative for high-OC and high-AOC firms, suggesting stronger cost stickiness, thus supporting Hypothesis H1, which posits that higher OC constrains cost reductions during downturns.

The effect of OC is also contingent on sales-change magnitude. Both low- and high-capital firms display cost asymmetry under small sales changes, but the effect is consistently stronger in high capital firms. Once sales changes are larger, the asymmetry disappears regardless of capital intensity. This provides strong support for Hypothesis H2, which posits that the influence of OC on cost stickiness depends on the size of sales shocks.

Additional analyses show that asset intensity, persistence of sales declines, macroeconomic conditions, and financial flexibility modulate cost stickiness depending on OC. Taken together, these findings demonstrate that OC amplifies cost stickiness while shaping its sensitivity to the scale of sales changes, highlighting its central role in explaining heterogeneous cost behavior across firms and economic conditions.

The two-way fixed effects results confirm that SG&A costs in high-OC and high-AOC firms exhibit stronger downward rigidity when sales decline. This effect is significant during small sales changes but becomes weak or disappears during larger sales contractions. The findings enhance the credibility of the interpretation of the link between organizational capital and cost stickiness, supporting the theoretical view that the adjustment costs embedded in organizational capital are most binding under modest fluctuations, but less so under major shocks.

By focusing on Japanese listed firms, this study provides novel evidence from an institutional and cultural context characterized by long-term employment practices, organizational routines, and distinctive reporting systems. These features offer a unique lens to understand how OC operates, thereby extending the generalizability of cost stickiness research.

5.2. Implications

The findings have significant implications for research and managerial practice.

First, the empirical results provide strong support for Hypothesis H1: firms with higher OC and AOC exhibit greater stickiness in (adjusted) SG&A expenses than low capital firms. This suggests that cost behavior depends on whether managers view intangibles, such as OC, as investments or expenses (Banker & Byzalov, 2014). Under the rational decision-making explanation for cost stickiness (Banker & Chen, 2006), managers may anticipate temporary sales declines and retain resources if the cost of reduction now exceeds the cost of reacquisition later. Managers focused on short-term performance treat OC as an expense, resulting in lower stickiness, whereas long-term-oriented managers treat it as an investment, increasing stickiness.

Second, this study highlights an institutional contrast between Japan and the U.S. In U.S. firms, low-OC firms may display cost anti-stickiness, while high-OC firms exhibit stickiness (Venieris et al., 2015). In Japan, cost stickiness occurs in both low and high-OC firms, although consistently stronger in high-OC firms. This reflects Japan’s institutional and cultural context, where lifetime employment and seniority-based wages prevail. For example, during the COVID-19 pandemic, U.S. unemployment peaked at 14.7% in April 2020, whereas Japanese unemployment peaked at only 3.1% in October 2020 due to employee retention supported by government policies. This evidence is consistent with macro-level findings that aggregate cost stickiness predicts lower future unemployment rates (Rouxelin et al., 2018), suggesting that firm-level cost retention decisions can translate into broader labor market stability. This observation is also consistent with prior evidence indicating that employee orientation increases SG&A cost stickiness (Liu et al., 2019). Given the prevalence of strong employee orientation in Japan, reinforced by both culture and institutional arrangements, Japanese firms are structurally inclined to maintain higher levels of cost stickiness than U.S. firms, regardless of capital levels.

Third, the effect of sales-change magnitude supports Hypothesis H2. High-OC firms display stronger stickiness under small sales changes, but the effect diminishes under larger fluctuations. This underscores that cost behavior is conditional on the scale of economic shocks (Balakrishnan et al., 2004).

Descriptive evidence of negative average OC offers additional institutional insights (see Table 5). OC measures the economic value of organizational capital, indicating that Japanese firms, on average, show negative abnormal profits. This reflects institutional, organizational, and psychological priorities rather than inefficiency. Japanese firms prioritize employment stability, seniority-based wage systems, and labor–management coordination, while keiretsu structures (inter-firm business groups formed through long-term trading relationships and cross-shareholdings) reduce external capital market pressure for short-term profitability (McGuire & Dow, 2009). From an organizational psychology perspective, managers value organizational harmony, employee well-being, and social stability over short-term profits (Haga et al., 2019; Inaba, 2024). Consequently, negative OC may represent a deliberate trade-off: sacrificing short-term abnormal returns to preserve long-term organizational value, skill accumulation, and corporate responsibility, offering a distinctive interpretation of the OC–cost behavior link and highlighting the institutional features of Japanese firms.

Overall, OC functions as a double-edged sword: it strengthens competitive advantage but constrains cost flexibility. Managers must balance sustaining organizational resources with adaptability to market changes. Investments in OC enhance long-term competitiveness but may slow SG&A adjustments during downturns. High-OC firms should anticipate amplified cost stickiness, affecting budgeting, performance evaluation, and risk management, especially during recessions or persistent demand declines. Conversely, low-OC firms enjoy greater cost flexibility. For investors, high stickiness signals long-term capability but may reduce short-term profitability. Accordingly, valuation models should account for the asymmetric cost behavior of high-OC firms when projecting earnings and assessing firm value.

5.3. Contributions and Limitations

This study makes several significant contributions to research on cost behavior and organizational capital. First, it provides robust empirical evidence that OC is a key determinant of SG&A cost stickiness. While prior studies have largely emphasized structural factors or managerial incentives, this study highlights the role of intangible organizational resources in shaping asymmetric cost behavior, extending theoretical understanding beyond traditional explanations such as adjustment costs, agency problems, or managerial optimism biases.

Second, the findings contribute to cross-country literature by revealing both similarities and differences with prior evidence from U.S. firms. Consistent with U.S. results, Japanese firms with higher OC exhibit greater cost stickiness. However, unlike in the U.S., Japanese firms generally display cost stickiness regardless of their OC level. This suggests that institutional and cultural factors, including lifetime employment practices, seniority-based wage systems, and reduced external capital market pressures, significantly shape cost behavior. Thus, the study underscores the importance of considering institutional context in comparative analyses of cost behavior.

Third, by analyzing a unique dataset of 12,727 firm-year observations of TSE-listed firms over 18 fiscal years (2007–2024), the study provides long-term, large-scale evidence within Japan’s institutional setting, which remains largely underexplored. This enhances both the robustness and generalizability of the findings.

Finally, the study offers the insight that OC functions as a double-edged sword. While it strengthens firms’ stability, competitiveness, and long-term capability building, it also imposes rigidity in cost adjustments, amplifying cost stickiness. This dual effect carries important implications for managerial decision-making in budgeting, resource allocation, and risk management.

Second, the study does not explicitly control for industry effects, in line with Venieris et al. (2015). However, incorporating industry-specific analyses or controlling for sectoral heterogeneity could yield more nuanced insights and enhance the robustness of the results.

Third, the study excludes the variable $\log \left({\displaystyle \frac{{Emp}_{i,t}}{{Sales}_{i,t}}}\right)$, initially considered as a proxy for human capital, due to severe multicollinearity (VIF values exceeding 10). In addition, prior empirical evidence also indicates that $\log \left({\displaystyle \frac{{Emp}_{i,t}}{{Sales}_{i,t}}}\right)$ is statistically insignificant at the 10% level (Hirai & Shiiba, 2006). Future research could explore alternative proxies for human capital that avoid multicollinearity, providing a more comprehensive understanding of labour-related organisational resources in explaining cost stickiness.

Despite these contributions, the study has some limitations, which suggest directions for future research. First, organizational capital is inherently difficult to observe directly, and its measurement still leaves substantial room for improvement. In this study, the measurement of OC is based on abnormal profits linked to SG&A expenses. Although consistent with established literature, this approach may not fully capture the multifaceted nature. Future studies could refine its quantification by incorporating additional factors such as IT investment, data assets, human-capital density, and training expenditures, as suggested by recent work (Theodorou et al., 2024; Jacobs, 2024). Including such dimensions would provide a more comprehensive representation of organizational resources that shape cost behavior.

Second, the research perspective warrants further expansion. He and Li (2024) construct a “labor-mobility network” to capture the fluidity of human or knowledge capital. However, under Japan’s lifetime-employment regime, organizational capital is more likely to be “locked in” within firms, which reduces resource reversibility and potentially intensifies cost stickiness. This institutional contrast highlights the need for future studies to incorporate labor-mobility dynamics when examining how OC interacts with labor-related resources in shaping asymmetric cost behavior.

Third, the study excludes the variable $\log \left({\displaystyle \frac{{Emp}_{i,t}}{{Sales}_{i,t}}}\right)$, initially considered as a proxy for human capital, due to severe multicollinearity (VIF values exceeding 10). In addition, prior empirical evidence also indicates that $\log \left({\displaystyle \frac{{Emp}_{i,t}}{{Sales}_{i,t}}}\right)$ is statistically insignificant at the 10% level (Hirai & Shiiba, 2006). Future research could explore alternative proxies for human capital that avoid multicollinearity, providing a more comprehensive understanding of labour-related organisational resources in explaining cost stickiness. In addition, Li et al. (2024) conceptualize cross-shareholding networks as a form of “relational organizational capital” with important corporate-governance implications. Dense cross-shareholdings may increase cost stickiness by providing resource slack, yet they may also attenuate it through enhanced monitoring. Future studies could therefore incorporate cross-shareholding intensity as a moderating variable to examine its influence on the mechanisms underlying cost stickiness.