Revenue Identification in Attaining Consensus Estimates on Income Predictions: The Function of Ownership Concentration and Managerial Ownership Confirmation from Poland

Abstract

Financial reliability, along with clearness of business transactions, is one of the mainstays of sustainability. In this research, I investigate whether enterprises expand discretionary revenue when their income before intentional shaping is marginally under the consensus on the income prediction provided by analysts. The innovation of the paper lies in taking into account the role of managerial ownership, ownership concentration, and higher proportions of institutional investors in this situation. Higher ownership concentration and greater percentage of institutional investors in equity were analysed while considering the expropriation hypothesis. In order to assess the concern of managerial ownership for revenue manipulation, I considered the alignment of interest hypothesis. In this research, I certified that enterprises expand discretionary revenue when their revenue and operating income prior to intentional shaping barely miss the consensus forecast. I found that the existence of managerial ownership curtailed the magnitudes of upward discretionary revenue when revenue prior to intentional shaping was marginally below the consensus on revenue. Greater ownership concentration and higher proportions of institutional investors were on the bound of the statistical trend to expand discretionary revenue when net earnings, before intentional shaping, were marginally below analysts’ forecasts.

1. Introduction

The purpose of this study was to find out if reporting entities increase their discretionary revenue when their income prior to intentional shaping is slightly below analyst consensus for the fourth quarter (income benchmarks) and, above all, to appraise whether elements of ownership structure (such as ownership concentration, a higher proportion of institutional investors in equity, and managerial ownership) influence discretionary revenue in these circumstances. The problem is closely related to the issue of monitoring the activities of enterprises, and thus to the safety of economic transactions and sustainable development. The significance of earnings management with the purpose to attain the net earnings benchmark was proven in the literature [1,2,3,4,5,6]. However, the literature rarely considered the role of other income benchmarks, including revenue and operating profit. Moreover, in the literature, discretionary revenue was seldom analysed as the earnings management instrument to meet the income benchmarks, and if so, the usual income benchmark used is net earnings [7]. Thus, there is a clear need for research in terms of the relevance of various income benchmarks.

This paper most of all examines the importance of ownership structures in shaping discretionary revenue to meet income benchmarks. The literature mentioned a lot of excellent research on the consequences of ownership structure elements in earnings management. They relate in particular to research on the importance of ownership concentration [8,9,10,11,12,13,14], managerial ownership [15,16,17,18,19,20,21,22,23,24,25], and the role of institutional investors [26,27,28,29,30,31]. However, in these studies, general indicators of earnings management were usually considered, both the accrual and real type of earnings management. In particular, the role of the ownership structure elements in shaping discretionary revenue as a chosen earnings management instrument was not analysed in these studies. Moreover, the managing revenue in order to meet various income benchmarks, in particular revenues and operating income, was not analysed. This is relevant since an earnings management tool such as discretionary revenue is closer to the operating activity of the entity than the endmost net earnings. Thus, a research gap was identified. Therefore, I answer the research question on the importance of the ownership structure (ownership concentration, institutional investors and managerial ownership) in shaping discretionary revenue with the intention of accomplishing the benchmarks for various types of income (revenue, operating income, net earnings).

The research was conducted on the basis of companies registered in the Warsaw Stock Exchange (WSE) in Poland, for which quarterly consensus forecasts of income (revenue, operating income and net earnings) in 2016–2020 are prepared. There were several reasons for choosing such a sample of companies. First, the stock prices of the companies listed on the WSE respond positively to meeting the net earnings benchmarks [32]. The second reason is that these companies are relatively diverse in terms of magnitudes of ownership concentration, managerial ownership, and institutional investors in equity [33]. The high variability of the elements of the ownership structure raises a significant question as to its function in earning management, including revenue recognition.

Empirical analysis was carried out on the basis of a final sample of 230 firm-year observations. In the first phase, I estimated discretionary revenues by industry and by year. Then, I described the discretionary revenue as a function of just miss indicator for income prior to intentional shaping and the interaction of just miss indicator with variables describing the ownership structure of enterprises, and control variables.

The paper includes an Introduction (Section 1), Prior Research and Hypothesis Development (Section 2), Data and Methodology (Section 3), Results (Section 4), Discussion (Section 5) and Conclusions (Section 6). In Section 2, I described the previous findings of major investigations on the significance of income benchmarks in intentional formation of earnings, with the special role of ownership structure elements, and I have developed some hypotheses. Section 3 describes the research approach and variables used in the model. In Section 4, I presented an estimation of the model parameters. In Section 5, I discussed the research results and how they relate to those of other authors. Section 6 described the research results and how they relate to the hypotheses, furthermore, I indicated what the added value of the paper is for sustainability.

The novelty and originality of the paper is that the functions of the elements of the ownership structure of enterprises are considered in the context of the avoidance of unfavourable deviations of the reported income from the consensus of analysts’ income predictions. First of all, the study emphasized the role of a specific earnings management instrument, that is, shaping discretionary revenue. According to the author, it is worth researching the relationships between the elements of the company’s ownership structure (such as high ownership concentration and managerial ownership) and selected earnings management instruments. Such studies complement the investigations using mainly synthetic earnings management indicators [9,11,18,19,21,22,23,25,30].

The paper showed that enterprises substantially enhanced discretionary revenue if their revenue and operating income prior to intentional shaping were marginally below analyst consensus. The elements of the ownership structure had limited influence on shaping discretionary revenue to meet the income benchmarks, but it could not be inferred that they were irrelevant.

2. Prior Research and Hypothesis Development

2.1. Meeting Income Benchmarks as an Encouragement for Earnings Management and the Function of Revenue Identification in Intentional Shaping of Earnings

Stock prices tend to decline if the reported earnings are less than the expected earnings and this creates an incentive for upward earnings management [34,35]. This is especially true considering that meeting the forecasted earnings is rewarded by the capital market [36,37]. Avoiding negative surprises in terms of forecasted earnings has been settled in empirical studies [1,2,3,4,5]. Burgstahler, D. et al. indicated infrequent reporting of earnings below the benchmark, and frequent reporting of earnings above the benchmark [6,38,39,40]. It was also proven in the literature that some of the factors more susceptible to the goal of avoiding unfavourable deviations from the predicted profits include a more institutional ownership structure, exposure to a greater risk of claims by stakeholders and greater dependency on the achievement of expected earnings [41].

One of the tools of earnings management is the identification of revenues, which is largely due to the value relevance of revenues [42,43,44]. According to Marquardt et al. [45] companies that issue shares have an upward trend in revenues. Graham J.R. et al. [46] showed that earnings are driven up mainly by discretionary revenues. Revenue identification is also used to smooth out earnings [47,48]. It was also proven in the literature that revenue management (in the form of price discounts) is used to avoid reporting losses [28]. Research also showed that increasing discretionary revenues is used to avoid negative deviations from the forecasted earnings [7,49]. On the basis of companies registered in WSE in Poland, Piosik [49] found that companies expanded discretionary revenue when enterprises marginally lost consensus predictions for operating profit and net earnings, however investigation was conducted only on data from the period prior to COVID-19 pandemic. It would be recommended to investigate how these regularities develop over the period that also covers the COVID-19 pandemic.

I, therefore, formulate the following hypothesis:

Hypothesis 1 (H1).

If revenue, operating income and net earnings prior to intentional shaping are lightly below the consensus of the analysts’ predictions for the fourth quarter, enterprises expand discretionary revenue to avoid adverse deviations from the consensus.

2.2. The Function of Ownership Structure in Earnings Management

2.2.1. Ownership Concentration and Earnings Management

In the literature, there are many pieces of research on the function of ownership concentration in intentional shaping of earnings (both accrual and real). On the other hand, there has been much less research into the importance of ownership concentration in earnings management by identifying revenues. In investigation of the implication of the extent of ownership concentration in earnings management, one of the two hypotheses relating to the impact of concentration on the ability to monitor a company’s activity was usually taken into account: the hypotheses of efficient monitoring and expropriation [8,50,51]. The literature showed positive relationships between accrual earnings management and the extent of ownership concentration [8,9,10,11]. However, there was also evidence of a negative correlation between the extent of earnings management and the magnitude of ownership concentration [12,30]. Additionally, some authors showed more complicated, curvilinear relationships between the magnitude of the ownership concentration and the extent of earnings management, especially accounting earnings management [13,14] and real earnings management [33].

However, the results of some research negated the occurrence of a possible correlation between earnings management and ownership concentration [17,52,53].

Since most studies showed that a high degree of ownership concentration tended to coincide with shaping profits upwards, I adopted the assumptions of the expropriation hypothesis in my study. Moreover, in this study, ownership concentration was applied to corporations, not to individuals. Furthermore, Achleitner et al. [54] discovered that family firms are less absorbed in earnings management, contrasted with non-family firms. Therefore, I formulated a further hypothesis:

Hypothesis 2 (H2).

I hypothesize that in circumstances of avoiding unfavourable deviations from the predicted income (revenue, operating profit, net earnings), a high degree of concentration favours the growth of discretionary revenues by reporting entities.

2.2.2. A Function of Managerial Ownership in Earnings Management

One of two perspectives was usually considered in research on the importance of managerial ownership in monitoring the activity of a company, namely the entrenchment hypothesis [55] and the alignment of interest hypothesis [15,16]. In compliance with the alignment of interest hypothesis, the presence of managerial ownership in capital was a limiting factor for managerial opportunism and thus the extent of earnings management [15,16,17,18,19,20,30,33]. In turn, as reported by the entrenchment hypothesis, managerial ownership was able to strengthen managerial expedience [55]. This might also result in a more active involvement of the management board in earnings management, as demonstrated in [21,22,23,24,25]. According to the results of some studies, the relationships between the sizes of managerial shares and the amounts of earnings management were non-monotonous and rather curvilinear ([12,56,57,58,59,60]). However, there were studies that show no correlation between managerial ownership and earnings management [61].

In my study, the presence of managerial ownership in equity is expected to reduce discretionary revenue when the forecasted income has been missed by a small amount. This hypothesis is also justified by other research results from Poland, according to which the sizes of earnings management performed by transactions before mergers and acquisitions werereduced when equity included managerial ownership [62]. Consequently, the following hypothesis is proposed:

Hypothesis 3 (H3).

The presence of managerial ownership in equity is expected to reduce discretionary revenue, when income prior to intentional shaping fails to achieve the forecasted income marginally.

2.2.3. The Role of Institutional Investors in Intentional Shaping of Earnings

Institutional investors are also able to influence company’s controlling activities, containing intentional shaping of earnings. As reported by the efficient monitoring hypothesis, institutional investors can mitigate the sizes of earnings management, as documented in the literature [26,27,28,29,30,31]. However, according to the expropriation hypothesis, there are many constraints of the mitigating function of institutional investors in terms of earnings management. Almazan et al. [63] stated that it depends on the cost of monitoring. Claessens and Fan [64] and Porter [65] showed that institutional investors do not take a significant part in the monitoring activities of a company. Malik [66] showed that the use of real earnings management to prevent from losses or negative deviations from profit forecasts was lower in the presence of sophisticated, larger institutional investors. This study is specific as it only covers companies for which income predictions (for revenue, operating result and net earnings) are developed. In Poland, such companies are characterised by a large proportion of institutional investors in ownership. The median share of institutional investors in ownership in this group of enterprises is over 30%. In fact, in some companies, the proportion of institutional investors in ownership exceeds even 50%. For this reason, I can investigate if the higher proportion of institutional investors in equity mitigates/increases the size of discretionary revenue while avoiding negative deviations from forecast income. Given the significant share of institutional investors in the ownership of a certain group of enterprises, the expropriation hypothesis can be considered in the same way as in the case of a higher degree of concentration of ownership. Consequently, I formulated the next hypothesis:

Hypothesis 4 (H4).

When income (revenue, operating income, net earnings) prior to intentional shaping is marginally below the income forecast, the presence of a significant proportion of institutional investors in equity induces an additional increase in discretionary revenue.

3. Data and Methodology

3.1. Recognition of Discretionary Revenue

In this research, I recognized discretionary revenue using the technique of Stubben [67]. According to this technique, the normal amounts of changes in trade receivables were first identified. In accordance with the econometric model applied, changes in trade receivables in annual periods were described as a function of the sum of changes in realized sales revenues for the first three quarters, changes in revenues in the fourth quarter (all these changes annually) and the intercept. In the second step, the discretionary revenue was identified as the difference between the actual changes in the trade receivables and the normal amounts of changes in the receivables, determined in the first step (residuals). In both steps, all variables were scaled with the sum of assets. I identified discretionary revenues separately for industries and years.

Other approaches to recognizing discretionary revenue were also described in the literature [7,68]. However, as these approaches used net cash flow or cash inflows one year in advance, I opted out of their use, as the research period would be reduced by the year 2020 (COVID-19 pandemic period).

3.2. Research Design

In this study, I used Caylor’s modified approach [7] to describe the relationship between discretionary revenue, the indicator of income (before intentional shaping) achieving the estimate of income prediction to a small extent, the indicator of marginal failure of income (before intentional shaping) to meet the income forecast, and the control variables. Income was accounted for through three variables: revenue R; operating income, OPINC; and net earnings attributable to controlling entity, NIC.

The model becomes:

DR = α₀ + α₁ PISIMISS_t + α₂ PISIBEAT_t + α₃ PISIMISS_t × CONC_t + α₄ PISIMISS_t × INST_t + α₅ PISIMISS × MNG_t + α₆LnA_t-1 + α₇logM/B_t-1 + α₈GEAR_t-1 + α₉PAN-F_t-1 + α₁₀SGRTH_t + ε_t

DR = discretionary revenue (identified using the approach described in Section 3.1).

PISIMISS = (prior to intentional shaping miss) the just miss indicator variable is equal to 1 if an enterprise announces a small negative surprise for income prior to intentional shaping in the year t: (a) of no more than 3% of the revenue gap, measured as the percentage of negative surprise based on revenue before intentional shaping and divided by consensus on revenue, variable PISIMISS_R; (b) of no more than 2% operating income surprise (based operating income before intentional shaping) below market consensus of the end of the prior year’s total assets, variable PISIMISS_OPINC; (c) of no more than 2% net earnings surprise (based on net earnings prior to intentional shaping) below market consensus of the end of the prior year’s total assets; otherwise 0, variable PISIMISS_NIC. Therefore, I use three variants of the variable PISIMISS: PISIMISS_R, PISIMISS_OPINC, and PISIMISS_NIC. I anticipate a positive assess for this variable in compliance with the earnings management hypothesis.

PISIBEAT = (prior to intentional shaping beat) is defined as an indicator variable equal to 1 if an enterprise announces a non-negative earnings surprise (based on prior to shaping income) in the year t: (a) of less than 3% of revenue market consensus; (b) of less than 2% operating income positive surprise (based on operating income before intentional shaping) of the end of the prior year’s total assets; and (c) of less than 2% net earnings surprise (based on prior to shaping earnings) of the end of the prior year’s total assets. Therefore, I use three variants of the variable PISIBEAT: PISIBEAT_R, PISIBEAT_OPINC, and PISIBEAT_NIC. I anticipate a negative assess for the variable in compliance with the earnings management hypothesis.

CONC = an indicator variable up to 1 if corporations own more than 40% of total equity as at end of the period in the sample and 0 otherwise; according to an expropriation hypothesis I expect a positive estimate of the interaction of PISIMISS and CONC variables.

INST = an indicator variable up to 1 if institutional investors hold more than 35% of the total common equity and 0 otherwise; according to an expropriation hypothesis I expect a positive estimate of the interaction of PISIMISS and INST variables.

MNG = an indicator variable up to 1 if the members of management board hold more than 3% of the company’s shares and 0 otherwise; in agreement with an alignment of interest hypothesis I anticipate a negative assess of the interaction of PISIMISS and MNG variables.

I also considered the independent variables listed below:

LnA = the proxy of the size of company; the natural logarithm of the opening balance of the enterprise’s total assets.

M/B = a price to book ratio.

GEAR = gearing ratio, that is total debt divided by total assets as at beginning of the year.

PAN-F = delayed PAN-F score model, a gauge of the comprehensive financial situation of the enterprise (https://serwis.notoria.pl/news/espi-ebi (accessed on 25 September 2021)).

SGRTH = sales growth; an index of the annual dynamics of sales revenues.

In accordance with Caylor’s research approach [7], not only negative income surprises were introduced into the research project, but also positive ones. This is dictated by the intention to show not only the discontinuation of discretionary revenues, but moreover, a negative estimate of the parameter is expected, if the income prior to intentional shaping is slightly greater than the income prediction. This would indicate a reduction in discretionary revenues under these conditions.

The thresholds adopted for the values of 0 or 1 for the CONC, INST, and MNG variables resulted from the characteristics of the distribution of individual variables in the sample. So, for CONC variable, the limit was adopted at the percentile level k = 0.6. For the INST variable, the limit at the percentile level k = 0.65 was accepted. These assumptions guarantee a relatively high degree of ownership concentration. For the MNG variable, the limit was set at the percentile level k = 0.7.

PAN-F model is often used in Polish setting for bankruptcy prediction. The formula of the model is as follows: F = −2.478 + 9.478 (operating income/assets) + 3.613 (equity/assets) + 3.246 ((net earnings + depreciation)/liabilities)) + 0.455 current ratio + 0.802 (revenue/assets).

4. Results

4.1. Sample Choice and Descriptive Statistics

4.1.1. Sample Choice

The research population comprised all companies registered in the WSE, for which consensus forecasts are prepared for income predictions made by analysts, mainly from brokerage houses, between 2016 and 2020 (239 companies). The companies were analysed across the sectors in which they operate, as discretionary revenues were estimated across industries and years. Due to this modelling approach, all sectors with fewer than eight companies were deleted from the sample. All enterprises for which there were absent data on estimated consensus forecasts of companies’ revenue, operating income, and net earnings were also removed from the sample. After the next stage of selecting the sample, companies from five sectors were included in the sample, but companies from the construction and development sector were eliminated (17 companies). This sector was removed since the models showed no significance of predicting the value of the ΔAR variable (change in accounts receivable, and insignificant coefficients and determination). Ultimately, 230 firm-year observations were included in the sample, creating a data panel of 46 firms over a 5-year period across four sectors (

Table 1. Sample properties in line with companies’ industries.

No	Industry	Companies	Frequency	% Share
1.	Electromechanical industry (EL)	8	40	17.39
2	Massive and power industry (E&M)	14	70	30.43
3.	Computer Technology, Games and Telecommunications (IG&T)	15	75	32.61
4.	Contract services (CS)	9	45	19.57
	Total	46	230	100.00

). Observations restored from financial reports were acquired from the NOTORIA database (https://serwis.notoria.pl/news/espi-ebi (accessed on 25 September 2021)). Details on the elements of the companies’ ownership structure were obtained from the database https://equityrt.com/ (accessed on 25 April 2021). Sources regarding income prediction consensus were obtained from the resources of PAP (Polish Press Agency) (http://biznes.pap.pl/ (accessed on 15 February 2021)).

Three outstanding observations from Table 1 were rejected for modelling purposes due to extreme values for variable SGRTH. Consequently, 227 firm-year observations were used for modelling.

4.1.2. Descriptive Statistics

Table 2. Descriptive statistics (N = 227).

	Mean	Median	25% Quartile	75% Quartile	Std. Dev.
DR	0.000	−0.001	−0.013	0.017	0.028
PISIMISS_R	0.084	0.000	0.000	0.000	0.278
PISIMISS_OPINC	0.159	0.000	0.000	0.000	0.366
PISIMISS_NIC	0.176	0.000	0.000	0.000	0.382
PISIBEAT_R	0.119	0.000	0.000	0.000	0.324
PISIBEAT_OPINC	0.229	0.000	0.000	0.000	0.421
PISIBEAT_NIC	0.256	0.000	0.000	1.000	0.437
CONC	0.410	0.000	0.000	1.000	0.493
INST	0.361	0.000	0.000	1.000	0.481
MNG	0.295	0.000	0.000	1.000	0.457
LnA	14.226	13.953	12.954	15.263	1.883
M/BV	2.536	1.349	0.719	2.253	4.218
GEAR	0.446	0.470	0.314	0.558	0.179
PAN-F	4.544	2.465	1.585	3.769	8.443
SGRTH	0.061	0.050	−0.034	0.164	0.211

shows descriptive data for the response and predictor variables applied in model 4. Descriptive statistics were obtained using TIBCO Statistica, version 13.3.

The mean rate of prior to intentional shaping just missing consensus signifies that it was relatively slight for revenue (8.4%), and medium for operating income (15.9%) and for net earnings (17.6%).

4.2. Panel Data Analysis

Panel regression investigation was carried out in R 4.0.3 using the “plm” package from the R library [69]. The examination of the cross-sectional period features of the panel was conducted through the “lmtest” package [70]. The dependent variable DR and the independent (control) variable M/BV were normalized by the ordered quantile transformation method [71], thanks to the “bestNormalize” package from the R library [72]. Before normalization, the test results showed a greater degree of non-normality, using the Kolmogorov-Smirnov test for a normal distribution (hereinafter referred to as KS) KS = 0.47; p < 0.001 (DR), KS = 0.59; p < 0.001 (M/BV) than after KS = 0.01; p = 1 (DR), KS = 0.01; p < 0.001 (MBV). Despite adjusting for the normality of the dependent variable distribution, heteroscedasticity was still observed in all tested panel regression models. Adjustment of standard error and significance estimates due to the inspected heteroscedasticity was conducted using the “ARELLANO” technique which rectifies the covariance matrix of the variables [73,74]. The panel in the study has medium N and short T, therefore, the autocorrelation, time delays and stationarity were not tested.

Model diagnostics:

For the purpose of choosing a suitable panel model, distinctive tests were accomplished to indicate the goodness of fit for particular panel regression estimates. Additionally, an examination of heteroscedasticity of the irregular component variance was carried out. The results of the diagnostic tests are shown in

Table 3. Explanatory survey of the examined models.

Model	Lagrange Multiplier Test	Hausman Test	F Test for Individual Effects	Breusch-Pagan Test (Heteroscedasticity)
R	df = 1, χ = 3.74; p = 0.053	df = 12, χ = 9.08; p = 0.696	df1 = 44, df2 = 169, F = 0.72; p = 0.899	BP = 27.84, df = 13; p < 0.01
OPINC	df = 1, χ = 0.78; p = 0.377	df = 7, χ = 11.20; p = 0.130	df1 = 44, df2 = 169, F = 0.52; p = 0.993	BP = 33.68, df = 13; p < 0.01
NIC	df = 1, χ = 2.07; p = 0.151	df = 7, χ = 12.03; p = 0.099	df1 = 44, df2 = 169, F = 1.40; p = 0.067	BP = 44.35, df = 13; p < 0.001
	df = degree of freedom

.

The evaluation of the outcomes in Table 3 indicated that for individual _R, _OPINC and _NIC models, the OLS regression model (not including the variation along the considered enterprises and time) proved to be better fitted to the data than both the within and random effects models. In the circumstances where the Hausman test is significant, the fixed effects estimates are more appropriate. However, there are also insignificant F test for individual effects; this test points out that the means of dependent variables in panel groups are similar. In this situation, OLS is unbiased and the data is more suited to this model. The Breuch-Pagan test was carried out to test the heteroscedasticity of variances in particular models. The investigation with this test indicated the significant incident of the fact of dissimilarity of variance of the random component with respect to the expected estimates of the response variable in all three OLS models. By virtue of this heterogeneity, the standard errors in all models were adjusted through the “ARELLANO” procedure.

Regression results

Model _R:

Table 4. Influence of the modelled _R variables on the level of DR measurement.

	B Coeff	s.e.	t	p
(Intercept)	0.01	0.73	0.01	0.991
LnA	0.03	0.05	0.61	0.546
M_BV	0.11	0.08	1.37	0.171
GEAR	−0.64	0.41	−1.57	0.119
PAN-F	−0.04	0.01	−5.70	0.000 ***
SGRTH	0.06	0.52	0.11	0.915
PISIMISS_R	0.43	0.19	2.24	0.026 *
PISIBEAT_R	0.17	0.15	1.14	0.256
CONC	0.03	0.20	0.15	0.877
INST	−0.01	0.19	−0.05	0.962
MNG	0.07	0.14	0.54	0.593
PISIMISS_R:CONC	−0.31	0.29	−1.06	0.292
PISIMISS_R:INST	−0.15	0.31	−0.50	0.620
PISIMISS_R:MNG	−0.45	0.23	−1.97	0.050 *

F-statistic: 1.21839 on 13 and 213 DF, p-value: 0.26754; Adj. R-Squared: 0.012406, N = 227; Signif. codes: 0 ‘***’ 0.001 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 (‘.’—the result was on the statistical trend); Note: B = non-standardized regression coefficient, s.e. = standard error of coefficient B; t = t statistic; p = statistical significance (s.e. are estimated by the robust covariance matrix estimation).

introduces the OLS regression estimates for Formula (2). Table 4 indicates the associations between the extent of discretionary revenue and the slight breakdown to attain the consensus on revenue prior to intentional shaping (_R) and the role of ownership structure in situation of failure to accomplish the consensus on analysts’ prediction for revenue as well as also control variables.

The analysis of the results of model (_R) showed no significance when predicting the value of the variable DR: F (13, 213) = 1.22; p = 0.268, adj. R² = 0.01. However, the analysis of the model coefficients after correcting with the “ARELLANO” method showed a statistical significance of the PAN-F variable (B = −0.04, t = 5.70; p < 0.001), PISIMISS_R (B = −0.44, t = 2.24; p < 0.05) and the PISIMISS_R * MNG interaction (B = −0.45, t = 1.97; p < 0.05).

The modelling results showed that enterprises expanded discretionary revenues if revenue prior to intentional shaping marginally declined below the prediction consensus for revenue. However, it was shown that the inherence of managerial ownership in equity significantly mitigated discretionary revenue growth if revenue prior to intentional shaping fell marginally under the revenue benchmark. The increased proportion of institutional investors in equity and the ownership concentration demonstrated an insignificant influence on increasing discretionary revenue, when revenue (prior to intentional shaping) dropped marginally below the revenue benchmark. The results showed that the higher degree of ownership concentration did not influence the avoidance of unfavourable deviations of the reported income (revenue, operating income, net earnings) from the income prediction.

Model _OPINC

The analysis of the results of the model (_OPINC) showed significant predictions of the variable DR: F (10, 171) = 4.66 p < 0.001, adj. R² = 0.17. The analysis of the model coefficients after correcting with the “ARELLANO” method showed the influence of the GEAR variable (B = −0.68, t = 1.77; p = 0.079—influence on the border of the statistical tendency), as well as a statistically significant influence of the PAN-F variable (B = −0.04, t = 4.91; p < 0.001), PISIMISS OPINC (B = −0.83, t = 2.81; p < 0.01) and PISIBEAT OPINC (B = −0.47, t = 3.11; p < 0.01). The analysis did not reveal any other significant correlations. The outcomes of the investigation were presented in

Table 5. Influence of the modelled _OPINC variables on the level of DR measurement.

	B Coeff	s.e.	T	p
(Intercept)	0.54	0.73	0.73	0.464
LnA	−0.02	0.04	−0.56	0.579
M_BV	0.10	0.08	1.17	0.244
GEAR	−0.68	0.39	−1.77	0.079
PAN-F	−0.04	0.01	−4.91	0.000 ***
SGRTH	0.26	0.46	0.55	0.581
PISIMISS_OPINC	0.83	0.30	2.81	0.005 **
PISIBEAT_OPINC	0.47	0.15	3.11	0.002 **
CONC	−0.06	0.21	−0.27	0.789
INST	−0.02	0.19	−0.09	0.932
MNG	0.05	0.16	0.32	0.747
PISIMISS_OPINC:CONC	0.35	0.30	1.15	0.251
PISIMISS_OPINC:INST	0.34	0.34	1.00	0.317
PISIMISS_OPINC:MNG	0.14	0.34	0.41	0.679

F-statistic: 4.66399 on 13 and 213 DF, p-value: 0.00000045693; Adj. R-Squared: 0.17407, N = 227; Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘.’ 0.1 ‘ ’ 1 (‘.’—the result was on the statistical trend); Note: B = non-standardized regression coefficient, s.e. = standard error of coefficient B; t = t statistic; p = statistical significance (s.e. are estimated by the robust covariance matrix estimation).

.

From the data in Table 5, I concluded that reporting entities shaped discretionary revenue upward if their operating income prior to intentional shaping (OPINC) was marginally below the operating profit prediction consensus. Moreover, if operating income prior to intentional shaping was at, or slightly above, the prediction consensus for operating profit, reporting entities increased discretionary revenue as well. This research result is inconsistent with the expectations listed in Section 2, and demonstrated that even when the operating income before intentional shaping equalled or slightly exceeded the amount of the prediction, discretionary revenues were increased. This might indicate the tendency of reporting entities not only to avoid unfavourable deviations from the operating income forecast, but also the motivation to exceed the reported operating result above the prediction. It might also indicate the specific characteristics of the predicted operating income as a benchmark. On the other hand, it should be taken into account that in the other models (R_, OPINC) such a regularity did not occur, and therefore it is possible that the result might be spurious. The analysis did not state that the factors related to the ownership structure had any impact on this relationship (i.e., they did not significantly increase or decrease the discretionary revenue when operating income prior to intentional shaping was marginally below the prediction for operating income).

Model _NIC

The analysis of the results of the model (_NIC) showed significant predictions of the variable DR: F (10, 171) = 3.10; p < 0.001, adj. R² = 0.11. The analysis of the model coefficients after correcting with the “ARELLANO” method showed the statistical significance of the PAN-F variable (B = −0.04, t = 5.51; p < 0.001), as well as the result on the bound of the statistical trend of the influence of the GEAR variable (B = −0.81, t = 1.89; p = 0.060), interaction of PISIMISS_NIC and CONC: B = 0.54, t = 1.82; p = 0.070 and interaction of PISIMISS_NIC and INS: B = 0.51, t = 1.76; p = 0.080. The outcomes are presented in

Table 6. Influence of the modelled _NIC variables on the level of DR measurement.

	B Coeff	s.e.	T	p
(Intercept)	0.43	0.70	0.62	0.538
LnA	0.00	0.04	0.03	0.973
M_BV	0.10	0.09	1.20	0.232
GEAR	−0.81	0.43	−1.89	0.060
PAN-F	−0.04	0.01	−5.51	0.000 ***
SGRTH	0.08	0.48	0.17	0.862
PISIMISS_NIC	0.37	0.32	1.18	0.239
PISIBEAT_NIC	0.13	0.17	0.77	0.440
CONC	−0.16	0.21	−0.77	0.439
IST	−0.10	0.20	−0.51	0.610
MNG	−0.01	0.18	−0.08	0.940
PISIMISS_NIC:CONC	0.54	0.30	1.82	0.070
PISIMISS_NIC:INST	0.51	0.29	1.76	0.080
PISIMISS_NIC:MNG	0.06	0.33	0.18	0.854

F-statistic: 3.10094 on 13 and 213 DF, p-value: 0.00030409; Adj. R-Squared: 0.10782, N = 227; Signif. codes: 0 ‘***’ 0.001 ‘.’ 0.1 ‘ ’ 1 (‘.’—the result was on the statistical trend); Note: B = non-standardized regression coefficient, s.e. = standard error of coefficient B; t = t statistic; p = statistical significance (s.e. are estimated by the robust covariance matrix estimation).

.

The data in Table 6 showed that the only significant factor correlated with the DR variable was the PAN-F control variable. This means that the company’s growing financial position was accompanied by a reduction in discretionary revenues. In general, reporting entities did not increase discretionary revenues, if their net earnings prior to intentional shaping were marginally below the prediction for net earnings (the parameter estimate for PISIMISS_NIC is not statistically significant). However, it should be stated that the existence of higher ownership concentration and higher proportion of institutional investors in equity were on the bound of the statistical trend to expand discretionary revenue when net earnings prior to intentional shaping dropped marginally below prediction consensus for net earnings (p < 0.10). That might be evidence of a weak confirmation of the expropriation hypothesis and certainly not of efficient monitoring hypothesis.

Table 7. Outcomes of hypotheses verification.

Hypothesis	The Formulation of the Hypothesis	Results of Hypothesis Confirmation
H1	If income prior to intentional shaping (revenue, operating income and net earnings) is marginally below the consensus of analysts’ predictions for income for the fourth quarter, reporting entities increase discretionary revenue in order to avoid adverse deviations from the consensus.	Acknowledged for revenue and operating income.Not confirmed for net earnings.
H2	I hypothesize that in circumstances of avoiding unfavourable deviations from the forecasted income (revenue, operating profit, net earnings), ownership concentration favours the growth of discretionary revenues by reporting entities.	Not confirmed for revenue and operating income.The result was on the statistical trend for net earnings.
H3	The presence of managerial ownership in equity is expected to reduce discretionary revenue in circumstances of failure to meet the forecasted income (revenue, operating income and net earnings) to a small extent.	Acknowledged for revenue.Not confirmed for operating income and net earnings.
H4	When income prior to intentional shaping (revenue, operating income, net earnings) is slightly below the income forecast, the presence of a significant proportion of institutional investors in equity induces an additional increase in discretionary revenue.	Not affirmed for revenue and operating income.The result was on the statistical trend for net earnings.

summarized the details on the confirmation of hypotheses results.

5. Discussion

The study largely confirms the earnings management hypothesis [1,2,3,4,5,6] but it expands the range of income measures, supplementing them with revenue and operating income. This means enterprises expand discretionary revenue, if income prior to intentional shaping is marginally below analyst consensus on the prediction. This regularity applies to predictions for revenue and operating profit. I do not provide evidence that it relates to net earnings. In the studies conducted so far, the basic benchmark of the income was net earnings, or net earnings before the effects of discontinued operations [1,2,3,4,5,6,38,39]. However, it ought to be indicated that the synthetic indicators of earnings management (accrual-based and real earnings management) are not analysed in this study, but only one instrument is considered (i.e., discretionary revenues). In this study, I identified the role of discretionary revenue, and confirmed that managing net earnings does not turn out to be a significant incentive, while the forecasts of revenues and operating profit prove to be significant benchmarks. The results are different from Caylor, where net earnings proved to be a significant benchmark in the use of discretionary revenue [7]. By extending the study period with additional data from the COVID-19 pandemic, the obtained results turned out to be different than in the Piosik’s study [49]. In this study, the revenue benchmark and operating income prediction turned out to be more important in shaping discretionary revenues. The outcomes can therefore be clarified by the actuality that revenues and operating results were more sensitive to the verified instrument (discretionary revenue) than in the case of net earnings. This issue has so far been poorly addressed in the literature on the subject. Therefore, the following conclusions can be drawn from the investigation: predicted sales revenues and operating profit are important indicators of corporate performance and constitute an important benchmark in shaping discretionary revenues.

The study also shows that the elements of ownership structure (ownership concentration, managerial ownership and larger shares of institutional investors) have a limited impact on discretionary revenue formation, when income prior to intentional shaping (revenue, operating income and net earnings) is below analyst consensus.

I show that the presence of managerial ownership alleviates the growth in discretionary revenues, if revenue prior to intentional shaping is marginally below the analysts’ forecast for revenue. The results are in accordance with the alignment of the interests hypothesis [15,16,17,18,19,20,30,33], albeit in terms of using discretionary revenue in order to achieve predicted revenue. However, I do not find any impact of the presence of managerial ownership on the efforts to avoid unfavourable deviations from the predictions of the operating income and net earnings. This may be in line with the scepticism hypothesis concerning the function of managerial ownership [61] and is incompatible with the alignment of the interests hypothesis [15,16,17,18,19,20,30,33]. The research results do not confirm, however, the entrenchment hypothesis [21,22,23,24,25,55] in terms of discretionary revenue as an earnings management instrument, if we take into account operating profit and net earnings benchmarks. The study does not analyse the curvilinear relationships in this respect, but there were no such grounds either, taking into account the results of previous investigations on earnings management in Poland [33,49]. Therefore, we conclude that the function of managerial ownership in earnings management by means of discretionary revenue is relevant, but only in terms of achieving the revenue benchmark. However, it is not significant for operating income and net earnings benchmarks.

The impact of greater ownership concentration and a high proportion of institutional investors in equity is limited in circumstances of avoiding unfavourable deviations from the forecasted income. I showed that their presence is irrelevant in avoiding unfavourable deviations from forecasts on revenues and operating income. In this case, the results are inconsistent with both the efficient monitoring hypothesis [12,30] and the expropriation hypotheses [8,9,10,11]. Therefore, they correspond more closely to the hypothesis claiming a correlation between earnings management and ownership concentration [17,52,53] and claiming the role of institutional investors [64,65]. However, it is important to note that the study analysed the factor of the presence of a significant share of institutional investors in ownership and not the presence of any proportion.

On the other hand, it is not conclusive to assess the importance of ownership concentration and a higher institutional investor participation in the utilization of discretionary income for the sake of preventing from adverse deviations from the forecast net earnings. The interaction parameters in this case are positive and they tend towards statistical significance. This means that the presence of a higher ownership concentration and a high proportion of institutional investors in equity are on the bound of correlation with an increase in discretionary revenue to avoid adverse bias from net earnings predictions. This regularity may be in line with the expropriation hypothesis [8,9,10,11]. However, the conclusions from this aspect of the study are limited by obtaining relatively high errors of estimate. Summing up, a high degree of ownership concentration and a high proportion of institutional investors in ownership are not correlated with growth in discretionary revenue in order to achieve predicted revenue and operating income (operating level), but they are on the bound of statistical significance to be associated with an increase in discretionary revenue to attain net earnings benchmarks. This issue probably requires further research.

6. Conclusions

In this paper, I supported that companies expanded discretionary revenue if their revenue and operating income prior to intentional shaping were marginally below analysts’ consensus on these figures for the fourth quarter. I did not confirm this regularity in the case of net earnings forecasts. This means that discretionary revenue, as an earnings management instrument, are important in achieving benchmarks on operating level (revenue and operating income).

I also examined whether the elements of the company’s ownership structure impacted the volume of discretionary revenue in circumstances of avoiding unfavourable deviations from the forecasted income. I showed that the presence of managerial ownership alleviated the amount of discretionary revenues if revenue prior to intentional shaping was marginally below the revenue prediction. However, I did not confirm this pattern in the case of operating income and net earnings prior to intentional shaping. It can therefore be concluded that the presence of managerial ownership was relevant only in the case of striving to achieve predicted revenue, and is not significant in an attempt to achieve predicted operating income and net earnings.

Ownership concentration and a higher proportion of institutional investors in equity did not lead to an increase in discretionary revenue if revenue and operating income prior to intentional shaping were marginally below the analyst consensus. Therefore, I demonstrated the hypothesis of disbelief, regarding the function of increased ownership concentration in the discretionary shaping of revenue to achieve the operating income and revenue benchmarks. However, I found that growth in discretionary revenue to avoid adverse deviations from predicted net earnings amidst increased ownership concentration and increased institutional investor in ownership was on the bound of statistical significance. These results are closer to the expropriation hypothesis in terms of the function of increased ownership concentration in earnings management processes.

There are a few limitations of the investigation: only WSE listed companies were taken, and there were relatively medium number of observations. The duration of this investigation was also relatively short. However, the role of ownership structure elements in achieving income benchmarks by shaping discretionary revenue can be further analysed internationally in research in the future.

This study brings added value to both science and practice. Three areas of contribution to knowledge could be mentioned. The first involves the description of which income benchmarks are used for revenue recognition, and in particular for discretionary revenue. Secondly, our study identified which elements of the ownership structure are relevant to revenue recognition, if income before intentional shaping fails to achieve the consensus of analysts’ predictions, considering different types of income (revenue, operating income and net earnings). The paper also refers to issues of sustainable development. Financial pellucidity, containing clearness of transactions, is one of the mainstays of sustainability [75,76]. The research sphere analysing the relationships between the ownership structure, monitoring activities through the accounting information and sustainability is an important part of contemporary finance and accounting [77,78,79,80]. Earnings management influences monitoring activities of the enterprises, therefore it has a direct relationship with sustainability and economic performance [81,82]. This study examines the motivators that influence the magnitudes of discretionary revenue, which is relevant to the security of economic transactions. The importance of the study, in practice, is to detect discretionary revenues to avoid adverse deviations from forecasted income and to evaluate the importance of ownership structure under these circumstances. This is important for analysts, investors, and auditors.