Amsterdam Business School
Master Thesis
Earnings management and Dutch hospitals in the DTC regulation period
Name: C.G.A.H. Miedema
Student number: 2008005
Date: 22 June 2014
Supervisor: B.J. van Praag
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
2 Abstract
In the last decade the Dutch health care sector is deregulated from budget funding to performance funding model, where in 2015 the hospital funding will be completely based upon performance. This thesis addressed earnings management by Dutch hospitals in the diagnostic treatment combination (in short: DTC) regulation period. In this regulation period, from 2005 till 2012 hospitals are funded per diagnostic treatment combination. A diagnostic treatment combination is a series of operations or proceedings for a disease. There is a fixed prices segment, annually determined by government for DTC’s and negotiable segment for DTC’s.
In the DTC regulation period there are several incentives changes initiated by the reforms in the Dutch hospital sector. Before there was incentive to merge, as larger hospitals received more funding. Also, there is an incentive to produce as hospitals are funded based upon actual output, instead of agreed upon in the previous period. Another change in incentives is the incentive to produce efficiently as for more operations per DTC no additional funding is granted. Also, there is a countervailing power between hospitals in the negotiable price segment.
The Dutch Healthcare Authorization benchmarks each hospital annually and explains variances in an annual report. These reports are very detailed, per specialism and diagnostic treatment combination. Both the amount of DTC’s as well as the price per DTC are carefully analyzed. Also health insurers analyze the amount of DTC’s invoiced and prices carefully. If a hospital falls out of the benchmark, health insures perform an audit on the patient records. Also, hospitals with relatively high earnings or profit, in relation to the benchmark, are audited by health insurers. If health insures find a declaration unlawful, the hospital has to pay the health insurer back. Therefore management has an incentive to use accrual based or actual earnings management to prevent this increasing supervision, according to Eldenburg et al (2011) and Leone and Van Horn (2005).
If health insurers or the Dutch health authority sees a significant increase in financial performance, this will effect next year’s funding and prices. According to Yetman (2011) and Jegers (2010) that not for profit organizations use earnings management to maintain or gain government subsidies and donations. Therefore the first hypothesis is that hospitals manage earnings in a range just above zero.
3
The data is hand collected from annual reports by author (approximately 30%) and other employees of a Dutch audit firm. The data contains of 813 firm year observations from 2003 to 2012, for which 2003 and 2004 are only used for scaling purposes.
Burgstahler and Dichev (1997) assume that cases are normally distributed if there is no earnings management. Therefore we first analyzed the distribution of operating income. Earnings management is measured using the Jones (1991) model for discretionary and non-discretionary accruals. The empirical model of Leone and Van Horn (2005) is used to test the zero-profit hypothesis. This model measures the correlation between discretionary accruals and earnings before discretionary accruals. We also performed an analysis of variances and Barletts (1937) test for equal variances.
We find empirical evidence of hospitals managing their earnings in a range just above zero in the Leone and Van Horn (2005) model for regular hospitals. We also find empirical evidence for a significant difference in the distribution of operating income before and after discretionary accruals in Barlett’s test (1937).
However, there are limiting conditions. As current years revenue is highly significantly correlated to other operating expenses and personnel costs, as also shown by Bouwens et al (2006) in the pre DTC regulation period. Therefore further research into this strong correlations is recommended.
4 Table of contents
Abstract ... 2
Table of contents... 4
1. Introduction ... 6
2. The Dutch hospital sector ... 9
2.1 Introduction to the Dutch hospital sector ... 9
2.2 Functional budgeting system from 1988 to 2005 ... 11
2.3 Deregulating hospital funding from 2005 to 2011 ... 12
2.3.1 Implementation of DTC replacing functional budgets ... 12
2.3.2 Guaranteed system for cost of capital ... 14
2.4 DTC’s on their way to transparency from 2012 ... 14
2.4.1 Transaction model for performance funding ... 15
2.4.2 Abolishment of guaranteed system for capital costs... 15
2.4.3 Control model for medical specialists fees ... 15
3. Earnings management ... 16
3.1 Definitions of earnings management ... 16
3.2 Incentives for earnings management ... 17
3.2.1 Meeting expectations of the capital market... 18
3.2.2 Contractual motives ... 19
3.4 Incentives for earnings management in hospitals ... 20
3.5 Measurement of earning management ... 21
3.5.1 The Healy model ... 21
3.5.1 The DeAngelo model ... 22
3.5.2 The Jones model ... 22
3.5.3 The modified Jones model ... 22
5
4.1 Normal distribution of operating income ... 23
4.2 Zero-profit hypothesis ... 23
5. Research methodology ... 25
5.1 Sample – Hand collected data ... 25
5.2 The distribution of operating income ... 26
5.3 Empirical model for the measurement of discretionary accruals ... 26
5.3 Empirical model zero-profit hypothesis part one ... 28
5.4 Empirical model zero-profit hypothesis part two ... 29
6. Research results ... 30
6.1 Distribution of operating income... 30
6.2 Analysis of outliers ... 30
6.3 Discretionary and non-discretionary accruals ... 32
6.4 Testing the zero-profit hypothesis part one ... 35
6.5 Testing the zero-profit hypothesis part two ... 41
7. Conclusion ... 42
6 1. Introduction
In the last decade the Dutch hospital sector is deregulated from budget to performance funding. In 2015 the hospital funding will be completely based upon performance. This thesis addressed earnings management by Dutch hospitals in the diagnostic treatment combination (in short: DTC) regulation period. In this regulation period hospitals are funded per diagnostic treatment combination. A diagnostic treatment combination is a series of operations or proceedings for a disease.
While most literature about the effect of deregulation focuses on hospital quality, this thesis focuses on the effect on earnings management. In this paper the term deregulation is used for the reforms related to the changes in the funding structure of Dutch hospitals, which allow a greater role of market mechanisms. According to Kent (1989) incentives are an important part of understanding decisions made in non-profit organizations based upon changes in institutional environment. Prior research has also researched the effect of the deregulation in the Dutch hospital sector on various variables. Blank and Eggink (2011b) find a positive relationship between the productivity of hospitals and the changes in institutional context. For another variable, competition, Schut et al (2005) find that although the significant changes in institutional context the competition in the health care sector is still limited. Also according to Blank and Dumaij (2011a) changes in institutional context let to more cost efficiency in Dutch hospitals.
This paper offers insight into the effects of policy reforms on the expected manager behavior and thereby their financial performance. Especially given that the objective of the reforms was to provide good health care to the entire population at lowest possible costs. In 1988 eleven percent of the gross national product was used for health care, in 2012 this is almost fifteen have percent of the GNP according to the central bureau of statistics (2011b). So, this is a major cost category for the government and population. This is a continuous process to insure funding and performance are aligned properly. This paper provides an indication of the effectiveness of this market reforms measured in earnings management of health care providers. Therefore this study is relevant for current and further policy makers. Also, Hein (2013) in the financial newspaper of the Netherlands reviews the financial situation of Dutch hospitals as critical. As the question hereby is, if de deregulation of the Dutch hospital market led to a critical or different financial situation. Before, hospitals were mainly funded by government funds or governmental guaranteed loans. In the DTC
7
regulation period loans aren’t guaranteed by government and direct governmental funding decreases significantly. Therefore hospitals depend more on free market finance providers, like banks. This paper provides a better understanding of the financial performance of hospitals during this deregulation period. Therefore this paper is relevant for decision making in providing funding to Dutch hospitals. This is because this knowledge makes it easier to review a hospitals business plan.
Scott (2003) defines earnings management as the accounting policy choice to achieve some specific objective. This definition incorporates earnings management to maximize income (Burgstahler and Dichev 1997), to minimize income (Leone and Van Horn 2005), or to smooth income overtime (Bartov 1993). Bouwens et al (2006) show that Dutch hospitals manage earnings by avoiding loss. Managers also include future performance in manipulating income and therefore use income smoothing techniques. Dutch hospitals manipulate their income upwards to receive favorable financing conditions. However this research is conducted in the pre DTC period and can therefore not be generalized over the after DTC period. Bouwens et al (2006) show abnormal distribution of scaled earnings in a range just above zero. This research is relevant because earnings management after DTC at Dutch hospitals is not yet researched.
In the DTC-period managements incentives change, as there is no incentive to merge any more, as prices are no longer related to hospital size. There is an incentive to produce, as DTC products are directly related to actual production, as posed to the prospectively determined parameters. There also is an incentive to produce efficiently as for more operations no additional funding is granted any more. As there also is a negotiable price segment there is another incentive to produce. No more funding incentive to merge, as prices are no longer related to hospital size. Therefore this is a deregulation measurement, where there is more financial freedom. The first three incentive changes could lead to significantly better financial performance. However due to the negotiable part of prices managers have incentives to minimize operating income to overcome problems in next year’s negotiations with health providers. If health insurers or the Dutch health authority sees an increase in financial performance, this will effect next year’s funding and prices. As also the supervision on hospitals and their performance is increased, another incentive to manage earnings in a range just above zero.
8
To test the zero-profit hypothesis, this paper follows the Jones (1991) model for measuring the discretionary portion of accruals. This is one of the most commonly used methodology in the accounting literature. The modified Jones model, by Dechow et al (1995) is not used because hospitals debtors from regulation period through regulation period fluctuate rather easily. An example given is the change in debtor’s and other receivables from 2011 to 2012. Due to the change in funding, prices have to be negotiated with the health insurer. This leads to a high accrual of to be invoiced revenue and a lower trade debtor’s accrual. The modified Jones model assumes that all changes in credit sales are due to earnings management. However due to changes in regulation, as explained above, many of these changes are due to changes in regulation. Therefore we don’t assume that all changes in credit sales are due to earnings management.
9 2. The Dutch hospital sector
2.1 Introduction to the Dutch hospital sector
While most literature about the effect of deregulation focuses on hospital quality, this thesis focuses on the effect on earnings management. In the last decade the Dutch hospital sector is deregulated from budget to a performance funding, where in 2015 the hospital funding will be completely based upon performance. In this paper the term deregulation is used for the reforms related to the changes in the funding structure of Dutch hospitals, which allow a greater role of market mechanisms. According to Kent (1989) financial performance is an important part of understanding decisions made in non-profit organizations based upon changes in institutional environment. The background of current and prior regulation policies of Dutch hospitals is outlined. This is done by means of literature review of research into the Dutch hospital sector. But first the current general funding and legal structure of hospitals is explained.
In the Netherlands care is guaranteed for all Dutch citizens. However, the hospital care is not freely accessible. The care is accessible if a general practitioner gives an indication for the need of secondary care or in case of an emergency. Else, provided care has to be paid by civilians directly. All Dutch citizens are obligated to close a health insurance, to ensure care for everyone. Therefore hospitals are funded through various channels. Hospitals, are directly or indirectly funded through Dutch government, health insurers and civilians. The Dutch hospitals can be divided into three categories, based upon their financing and planning. The three categories are:
1. Categorical hospitals, which provide one or a few specialisms (26 hospitals in 2012) 2. General hospitals, which provide a large range of specialisms (54 hospitals in 2012) 3. Academic hospitals, which provide a large range of specialisms, conduct academic
research and complex patient care (8 hospitals in 2012)
The general hospitals and categorical hospitals are similar in funding and size. However academic hospitals are different as they also receive funding for the education of medical specialists and research.This is because academic hospitals are expected to serve a public interest by activities in the field of scientific education and research. This additional funding is paid by employers,
10
government and civilians through the Dutch health insurance fund. In size there is also a difference between regular and academic hospitals. For example the Erasmus MC has total assets of approximately 1.5 milliard euros and has 10,097 FTE’s (according to the annual report of 2012).The largest regular hospital in FTE’s in 2012, hospital Isala, reports only 3,949 FTE’s. Also the largest regular hospital in total assets is the hospital Antonius with 582.5 million. This is also shown in total revenue, where Erasmus reports approximately 1.2 milliard euros and the largest hospital in revenue, Antonius, only reports 454.2 million euros.
As many hospitals are established, directly or indirectly in a foundation. The board of directors is appointed in the articles of association. Only three hospitals have another legal structure. Other possible legal structures are the association and the private company. Only one hospital is legally organized in an association, this is the Diaconessehuis. The Haven hospital and the Slotervaart hospital are the only hospitals who are legally conducted in a private company. However regardless the type of legal structure operating for profit is prohibited. All realized profit must be used for a pre-defined social purpose. The board of directors is ultimately responsible for all hospital activities, including the medical activities. As there is information asymmetry between medical personnel and the board of directors daunting task. After all they do not perform medical treatments and the quality of care depends mainly on the medical specialists. Therefore it is legally required to appoint an external supervisory board.
Medical specialists can be employed through an employment agreement or can be freely established by means of an admission agreement. The majority of medical specialists are freely established and arranged in partnerships between specialists. Each medical specialist is required to join the medical staff. The medical staff is the organ that provides the communication between medical specialists and the Board of Directors. The hospital internal codes of conduct and the medical staff provides which medical-specific regulations and agreements. Medical specialists are also responsible for the registration of operations. Operations can be, among other, nursing days, x-rays, cytostatic drugs or consults. The current regulation, pooling of operations in a care product leads to funding. Therefore specialists have an incentive to record and perform operations. Prior and current policies and their incentives are explained in the next paragraphs.
11
2.2 Functional budgeting system from 1988 to 2005
In this period (1988-2005) Dutch hospitals are financed through a pre-fixed functional budget. This budget is paid from public resources by the Dutch government. This budget is determined by 4 components (Blank and Dumaij 2011a):
1. Availability component (supply area)
2. Capacity component (recognized beds and weighted specialists) 3. Location component (recognized capacity)
4. Production component (production agreements for hospital admissions, nursing days etc.)
The components are measured in production parameters. A production parameter, for example is the amount of recognized beds. The funding for this parameter is the amount of recognized beds multiplied by the fixed price (determined by national tariff agency).Even if eighty percent of the beds were empty, the funding amount of this parameter did not change. Therefore there is a gap between the measurement of production and actual production of a hospital. Another issue was that these parameters were prospectively determined and depended on the ‘agreed upon expected output’ rather than the actual output (Schut et al 2005). Therefore there was another gap between actual parameters and funded parameters on hospital care.
Also, larger hospitals received more funding for certain parameters. Larger hospitals were expected to treat more severe patients, and therefore have more costs. This led to an undesirable side effect, an incentive to merge, according to Blank and Dumaij (2011a).
For the built of a hospital the institution needed a permit by the College Built of health care institutions. The compensation for the plant, property and equipment is determined by the permit application and does not depend on actual future production. Also the government acted as guarantor for the mortgage loans. Therefore there is a lack of countervailing power according to Windhorst (2007).
12 In simplified form hospital funding in this period:
1. Agreed upon production parameter X fixed prices (higher prices for larger hospitals) = Functional budget = Revenue
2. Amount of capital compensation on permit = Capital income
Kutzin (2001) provides a conceptual framework of health care provider payment methods and treatment incentives. According to him this kind of funding through prospective budgets leads to two treatment incentives. The first incentive is to underprovide treatment and second to shift patients to other providers. This is because there are no financial consequences of underprovide or to shift patients.
2.3 Deregulating hospital funding from 2005 to 2011
In 2005, the Dutch government started deregulating the hospital financing structure, to introduce managed competition. For two reasons they wanted to gradually reform the market structure in the hospital sector. The first reason is they wanted to have a good price – quality ratio. Main social objectives hereby were waiting times, technical medical profession, efficiency, choice in health care and innovation. The second reason was that health care must be available and affordable for everyone (Central Plan Bureau 2005). According to Hassaart (2011) the main objective for this deregulation measure was that funding should depend on performance and funding should enforce efficiency.
Therefore, the objectives to deregulate were the flaws in the functional budgeting system as mentioned in paragraph 2.2;
1. Incentive to merge (due to higher funding of large hospitals)
2. Lack of countervailing power (due to government guarantees on loans)
3. Lack of an incentive to operate efficiently (funding based upon agreed production parameters, rather than actual production)
2.3.1 Implementation of DTC replacing functional budgets
In 2005 the Dutch government implemented funding per diagnostic treatment combination. A diagnostic treatment combination (DTC) is a series of operations. One operation is for example
13
to put an arm in a plaster. The diagnostic treatment combination is the broken arm, consisting of variance operations, such as plaster put, X-ray, appointment with specialist and removing plaster.
The DTC’s are directly related to the production per patient, in contrast to the parameters in the functional budgeting period. As more operations in one DTC, did not lead to extra funding, there is an incentive to minimize the operations in one DTC, and therefore to produce more efficient.
The funding of hospitals is divided into two segments:
1. A-segment – Regulated prices for DTC’s, these prices are determined by the Dutch health care authority (NZa)
2. B-segment – Free negotiable prices between health insurer and hospital
The fixed prices do not depend on the size of a hospital, there is no funding incentive to merge anymore.
The negotiable prices in segment B give a countervailing power incentive. However, according to Hassaart (2011) the negotiable B-segment, measured in actual revenue was only 5-13 percent from 2005-2008 and 19.9 percent in 2009. Measured in the amount of DTC’s, the negotiable B-segment was 10-28 percent from 2005-2008 and 34 percent for 2009-2011.
In simplified form hospital funding in this period:
1. Amount of DTC’s multiplied by fixed price = Revenue A-segment
2. Amount of DTC’s multiplied by negotiated price = Revenue B-segment
3. Amount of capital compensation on permit = Capital income
According to the conceptual framework of Kutzin (2001) the DBC structure is classified as a case-based funding scheme according to fee schedule. This is because the price is prospectively determined by the Dutch health care authority (Nza) and payment is done retrospectively per item of service provided (DTC). The treatment incentives given by Kutzin (2001) are to increase volume of less severe patients in each DTC and decrease services (operations) per case.
14 2.3.2 Guaranteed system for cost of capital
In 2009 the Dutch government abolishes the permit regulation for the built of hospitals. There are no new permits granted to hospitals. Hospitals need to integrate their capital investment in property plant and equipment in their DTC prices. This deregulation measure gives an incentive to use capital efficiently, as only production leads to funding, negotiated or fixed. The health insurer, a for-profit organization, will not pay for insufficient use of capital. As before, if the permit was granted, the capital costs were compensated. But to ensure a good transition, there is a transaction period till 2016, see paragraph 2.4.2.
2.4 DTC’s on their way to transparency from 2012
In 2012 the Dutch government introduced DTC’s on the way to transparently. Hospitals transferred from the system of funding through regulated prices to full performance funding. The desired final model of complete performance funding is introduced in the following steps:
1. The abolition of hospital budgets, from regulated price performance of the A-segment to the A-A-segment falling within performance funding;
2. Expansion of the free segment(from 34 percent in 2011 to 70 percent in 2012,measured in DTC’s);
3. Introduction of a fixed segment with availability contributions for specific care functions;
4. Full reimbursement of the capital costs in rates
With the introduction of the new funding system DTCs On their way to Transparency (DOT) in 2012, approximately 30,000 diagnosis treatment combinations (DTC) are replaced by approximately 4,400 DTC care products.
In the final performance funding model, health care providers are paid per output performance, based on applicable integral rates. Besides the free segment full free prices remains a regulated segment with maximum rates for care that is not suitable for free pricing. In addition there will be a fixed segment for care that would be insufficiently secured with performance funding. However, in negotiations with health insurers, a maximum amount of funding is agreed. Provided care above this maximum is not funded though health insurers. Therefore the financial risk of providing care above this maximum is for the hospital itself.
15
The final model is applicable from 2015, till then the transition regulation is applicable. During the transition phase, the final model already implemented, with the exception of the integral rates involving the fee component for specialists is recorded in the overall rate and differentiated premium for capital costs, complemented by flanking policies. As this thesis only researches Dutch hospital performance till 2013, the final model is not part of this thesis but shortly simplified explained to understand current transition regulation. The flanking policy comprises three transition models:
1. Transaction model for performance funding (NZa, 2012c) 2. Guarantee system for capital costs (NZa, 2012b)
3. Control model for medical specialist fees (NZa, 2012a) 2.4.1 Transaction model for performance funding
The transaction model functions as a safety net for the transaction from functional budgets (A-segment) to performance funding. Hospitals receive funding for ninety percent (2012) of the difference between their revenue based upon performance funding and upon functional budget. The compensation for the difference is gradually reduced till zero in 2015.
2.4.2 Abolishment of guaranteed system for capital costs
The regulation for capital costs was separate from other hospital funding till 2009.To prevent difficulties in integrating cost of capital into hospital prices the Dutch Healthcare Authorization (NZa) compensates hospitals for certain capital costs. This compensation is gradually reduced till zero in 2017. This means that capital cost have to be included in DOT-prices.
2.4.3 Control model for medical specialists fees
Medical specialists are employed based upon a fee agreement or are employed through payroll. The costs for medical specialists employed through fee agreement are a significant expense for a hospital. Therefore the Dutch HealthCare Authorization (NZa) made the agreement with fee- specialists that their increase in salary till 2014 stays within the macroeconomic framework of two and a half percent.
16 3. Earnings management
3.1 Definitions of earnings management
Many studies, as well as in the not-for profit as the profit sector have researched earnings management. This is because in many sectors managers have incentives to manipulate their earnings. Based upon literature there are various definitions of earnings management. However, there are common factors of these definitions. In this paragraph the most frequently used definitions of earnings management are explained.
Earnings management is only useful when there is information asymmetry. Information asymmetry leads to a demand of information and information value. As for example, the annual report is means of communication between stakeholder and management. If stakeholder would have the same information as the manager, the annual report has no information value for the stakeholder. Also, because the stakeholder has all the information he sees through efforts of management to manipulate earnings. Therefore management would not have an incentive to manipulate earnings, or to public an annual report as there is no demand for this kind information, as all information is publically known.
Scott (2003) defines earnings management as the accounting policy choice to achieve some specific objective. This definition incorporates earnings management to maximize income (Burgstahler and Dichev 1997), to minimize income (Leone and Van Horn 2005), or to smooth income overtime (Bartov 1993). However this definition does not specify which objectives are specifically targeted. This definition is also limited, as it explains earnings management as the use of accounting policy use, but managers can also use real actions to manipulate earnings. This brings us to the two kinds of earnings management, real activity earnings management and accrual earnings management.
Accrual based earnings management is the manipulation of earnings through accruals within generally accepted accounting principles, according to Leone and Van Horn (2005). Accruals contain assumptions and estimations and therefore estimations errors. As these estimations or assumptions are deliberately manipulated to maximize income, to minimize income, or to smooth income overtime, there is accrual based earnings management. If there is no intent to manipulate earnings estimation errors are part of the inherent risk of accounting through accruals. This is the
17
trade-off between reliability of accruals and value relevance of cash flows. Dechow et al (1998; 1994) show that earnings is a better measure of performance than underlying cash flows.
Real activity earnings management is the manipulation of earnings by real transactions according to Eldenburg et al (2011). Mangers influence transaction dates to maximize income, to minimize income, or to smooth income overtime. These are physical and real transactions, however they only occur by management intention to manipulate earnings. Leone and Van Horn (2005) define real earnings management as the intent of managers to increase or lower spending’s to manipulate earnings.
Healy and Wahlen (1999) define earnings management as: “Earnings management occurs when managers use judgment in financial reporting and structuring transactions to either mislead some stakeholders about underlying economic performance of the company or influence contractual outcomes that depend on reported accounting numbers.” The definition earnings management from Healy and Wahlen (1999), is expressively and exclusively based on misleading stakeholders. But earnings management also has information value. The blocked communication with stakeholders can be unblocked and therefore management can communicate private information to stakeholders. However all definitions of earnings management imply intent by management and therefore also imply that they have incentives to do so. In the next paragraph the incentives for earnings management in the profit and not for profit sector are outlined. As there are as well similarities as differences between the profit and not for profit sector both our outlined shortly. In the following paragraphs.
3.2 Incentives for earnings management
Managers have different incentives to (ab) use the freedom associated with accounting principles so as to achieve certain desirable earnings. According to Healy and Wahlen (1999) there are 3 main incentives to manipulate earnings:
1. Meet expectations of the capital market (capital market expectations and valuation) 2. Contract motives (contracts written in terms of accounting numbers)
3. Monopoly control and other government intervention (antitrust or other government regulation)
18
Eldenburg et al (2011) ad hereby that profit organizations also have incentives to outperform expectations to create more shareholder value and they have incentives to report more earnings than prior year. These incentives can be decided into two kind of motives, compensation motives or motives to emulate a benchmark.
3.2.1 Meeting expectations of the capital market
As there is information asymmetry many companies are reviewed by stakeholder based upon their external reporting. As for example investors and analysts determine stock prices by discounting future cash flows. The company’s financial reports, including financial statements, quarterly and interim notifications, can play an important role in valuing the shares. As therefore this information has value in financial decision making and is therefore decision useful.
As the end of an accounting period nears, managers can observe the firm’s underlying earnings as well as analyst forecasts. During this time they can estimate any shortfall from the forecast and use income-increasing abnormal working capital accruals to eliminate this gap. Other accrual choices (deferred taxes, accrued expenses, provisions) are based upon assumptions and estimations. Managers have incentives to do so due to signaling theory and costs of capital.
These external reports are based upon accounting, therefore managers have opportunities to influence short-term results by accrual based earnings management, according to Healy and Wahlen (1999). In this study they refer to Burgstahler and Eames (1998) who find empirical evidence that managers are take action to apply earnings management to match the predictions of the market or to go by them. Roychowdhury (2006) also finds similar empirical evidence. However, notice that this stream of research focusses on the opportunistic use of accounting, by window-dressing and misleading users of financial statements.
Kasznik (2002) show that the market penalizes firms that previously meet analyst expectations but subsequently fail to. Positive abnormal accruals are negatively correlated to future earnings. This is because this discretionary accrual leads to a reverse in future earnings. Therefore there is a cost of capital in meeting expectations through positive abnormal accruals in future accounting periods. However, if subsequent earnings meet expectations easily, these reversal does not affect future market values. The aggressive use of positive abnormal accruals to meet analyst expectations is also more likely to raise the suspicions of auditors and the board of directors.
19 3.2.2 Contractual motives
A company can be seen as a group of contracts. The company has contracts with employees, lenders, creditors, suppliers and other stakeholders. The aim of the company is to minimize the contract costs, negotiation costs, monitoring costs and costs of renegotiations. Yetman (2011) and Jegers (2010) find that not for profit organizations use earnings management to maintain or gain government subsidies and donations.
Just as profit businesses, non-profit companies do not want to show a negative external financial reporting. The motivation and reasons are therefore comparable, according to Leone and Van Horn (2005). Not for profit organizations only use earnings management when there are small losses, so that the reporting to stakeholders is positive. The authors conclude that hospital managers influence the profit, so they do not have to report a loss. Reporting a loss can lead to higher transaction costs with creditors and or the government. Bugstahler (1997) also finds in the for profit sector that managers are loss averse. This can be due to managers opportunistically avoid reporting earnings decreases and losses to decrease transaction costs with stakeholders.
There is also another contractual motive, if a loss is reported job security of management can decrease. Not for profit organizations with an oversight body have another motive to apply earnings management. If reporting a loss leads to increased supervision it can be very costly. Therefore management has an incentive to use accrual based or actual earnings management to prevent this increasing supervision, according to Eldenburg et al (2011) and Leone and Van Horn (2005).
Leone and Van Horn (2005) conclude that managers manage earnings as close as possible to zero. They also find that managers in the not for profit sector try to keep variations in income as low as possible. Managers even try to keep the variations in the gain as low as possible. The result, according to the authors, is that if there is made too much profit, a hospital loses the tax-free gain or that there is more pressure on price agreements with third parties. This is in line with the income smoothing according to Bartov (1993). Hereby managers try to manipulate earnings fluctuations to a pre-determent amount that is considered to be normal for the company. According to Bartov (1993) there are many incentives for this kind of income smoothing. For example, to overcome higher targets for following years or to ensure that the company can report a yearly growth.
20
Krishnan et al (2011) show that the stakeholders, make the choice of a donation or a fee conditional on the financial ratios. If these numbers are relatively high in relation to other organizations in the same branch, they are more likely to be positive towards donations. This can increase the compensation for managers. Leone and Van Horn (2005) indicate that donors make a choice based on the amount of charity and the profitability of the hospital. Krishnan et al (2002) indicate that donor’s choice depending on two ratios, namely the program service ratio and the fundraising ratio.
The program service ratio is composed of the service cost divided by the total cost. The fundraising ratio consists of the fundraising expenses divided by the total collected donations. Some parts of the ratios (program service costs, total costs, fundraising costs and received donations) are based upon accounting and can therefore be manipulated by managers.
3.4 Incentives for earnings management in hospitals
There are many models for not-for-profit incentives at hospitals for earnings management. According to Wilcox (2002) these models can be divided into two streams of research. The first stream argues that there is no difference in profit-maximizing between for-profit and not for-profit hospitals. However, there is according to Danzon (1982), a difference between the appropriations of result. In for-profit hospitals result is appropriated by the payment of dividend. In not-for-profit hospitals the result is appropriated to different shareholders, like specialists, hospital staff, administrators and outsourced agents. The second stream argues that for-profit hospitals operate more efficient as they have lower costs and higher revenues (see overview paper James et al 1987). Wilcox (2002) however finds that for-profit status is unrelated to earnings management in United State hospitals. But there is a significant relationship between the financial relationship of a system hospital and an independent hospital according to Wilcox 2002. This is because the cost advantages hospitals obtain due to size, better known as economies of scale.
Bouwens et al (2006) show that Dutch hospitals manage earnings by avoiding loss. Managers also include future performance in manipulating income and therefore use income smoothing techniques. Dutch hospitals manipulate their income upwards in the year to receive favorable conditions. However this research is conducted in the pre DTC period and can therefore
21
not be generalized over the after DTC period. Bouwens et al (2006) show abnormal distribution of scaled earnings in a range just above zero.
3.5 Measurement of earning management
In the previous sections the phenomenon of earnings management is outlined. The definitions and motivations of managers to apply earnings management are outlined. In this section the methods to empirically research earnings management are explained. Earnings management is no easy phenomenon to research. However prior researchers have developed methods to detect earnings management.
There is a difference between discretionary accruals and non-discretionary accruals. Discretionary accruals are accruals that are not the result of normal operating activities, but are influenced by managers. Non-discretionary accruals are the result of normal business operations. The management does not manipulate these accruals. Each in the following paragraphs explained model is then based on a calculation that is splitting the non-discretionary part of the discretionary accruals. These models are the Healy model, the DeAngelo model, the Jones model, the modified Jones model and the industry model.
Each of these models in based upon the nature of accruals. Accruals are used to shift cash flows over time and create a better measure of firm performance. For example, receivables lead to subsequent collection in t+1. However, if this accrual is wrong it will lead to a reversal of this accrual in t+1. These models try to calculate this abnormal portion of accruals, which is due to earnings management.
3.5.1 The Healy model
Healy (1985) assumes that each period of the non-discretionary accruals are constant and only the discretionary accruals vary by period. The discretionary accruals can then be calculated as the difference between total accruals and non-discretionary accruals. Healy (1985) defines accruals as the difference between reported earnings and cash flow on an operational basis. He uses the total accruals as a proxy for the discretionary accruals because the non-discretionary accruals are not observable. In his model there are three groups, one group where he assumes that there is positive earnings management and two groups where he assumes negative earnings management. The total
22
average of accruals over time where no earnings management is expected is than the normal part of non-discretionary accruals. Based upon this portion the discretionary accruals can be calculated.
3.5.1 The DeAngelo model
In the DeAngelo (1986) model non-discretionary accruals are seen as constant and the discretionary accruals are calculated from the total accruals. DeAngelo (1986) compares the total accruals for the year with the total accruals from the previous year. The DeAngelo model is comparable, with the Healy model. The big difference is that the DeAngelo model takes the prior periods into account. In this model the first difference of total accruals are calculated. The first difference of the total accruals is meant the difference between the total accruals of the year when you want to measure earnings management and total accruals from the previous year. DeAngelo (1986) assumes that the first differences had an expected value of 0 under the null hypothesis of no earnings management.
3.5.2 The Jones model
Jones (1991) based its measurement model for earnings management on the measurement of the model DeAngelo (1986). The measurement model of Jones has some additions. The accruals are dependent on the economic conditions of the company. Where the revenue affects the level of non-discretionary accruals this factor will have to be included in the measurement model. According to Jones, his non-discretionary accruals, unlike the models in Healy and DeAngelo are the residuals of the model. The variability of the non-discretionary accruals is explained by changes in revenues and the change in the value of fixed assets in comparison with the previous year. The difference between total accruals and non-discretionary accruals are discretionary accruals.
3.5.3 The modified Jones model
Dechow et al (1995) modifies the Jones model by adding the change in receivables in the same period. This original model assumes that managers do not apply earnings management on revenue. However the modified Jones model implies that all changes in credit sales are a result of earnings management. This is based upon the assumption that it is easier to earnings management on credit sales than cash receivables.
23 4. Hypothesis development
4.1 Normal distribution of operating income
Burgstahler and Dichev (1997) assume that cases are normally distributed if there is no earnings management. Therefore, if all income is normally distributed and there are no outliers, there is no earnings management. Therefore, before researching further, we first checked if operating income of hospitals is normally distributed. Operating income is hereby defined as the earnings before interest and taxes. Taxes however form a relatively small role in the profit and loss account. This is because due their social function they are not obligated to pay corporate taxes. As hospitals defer in size operating income is lagged by the total assets.
4.2 Zero-profit hypothesis
Before 2005 hospitals where funded through functional budgeting. Brouwens (2006) has shown that Dutch hospitals manage earnings in a range just above zero in the pre-DTC-period. From 2005 till 2012 hospitals are funded through DTC’s, where for the B-segment prices are negotiable. This deregulation means a shift in incentives, as explained in chapter 2 and summarized below.
1. No more funding incentive to merge, as prices are no longer related to hospital size 2. Incentive to produce, as the DTC’s are directly related to actual production, as a
posed to prospectively determined production parameters
3. Incentive to produce efficiently, as for more operations no additional funding is granted, as a posed to prospectively determined production parameters
4. Incentive to produce efficiently, as there is countervailing power between hospitals in the negotiable price segment
Therefore this is a deregulation measurement, where there is more financial freedom. The first three incentive changes should lead to significantly better financial performance. However due to the negotiable part of prices managers have incentives to minimize operating income to overcome problems in next year’s negotiations with health providers. If health insurers or the Dutch health authority sees a significant increase in financial performance, this will effect next year’s funding and prices.
24
For the annual reports of Dutch hospitals there are in addition to the regular Dutch generally accepted accounting principles, there are specific accounting principles, namely the RJ 655. This accounting principle specifies the exact way of presenting funding.
The Dutch Healthcare Authorization benchmarks each hospital annually and explains variances in an annual report. These reports are very detailed, per specialism and diagnostic treatment combination. Both the amount of DTC’s as well as the price per DTC are carefully analyzed. Also health insurers analyze the amount of DTC’s invoiced and prices carefully. If a hospital falls out of the benchmark, by for example having a relatively high number of broken arms, health insures perform an audit on the patient records. Also, hospitals with relatively high earnings or profit, in relation to the benchmark, are audited by health insurers. If health insures find a declaration unlawful, the hospital has to pay the health insurer back. Therefore there is a risk in outperforming other hospitals in profit or earnings. The supervision on hospitals and their performance is therefore increased, another incentive to manage earnings in a range just above zero.
Therefore the first hypotheses is that:
25 5. Research methodology
5.1 Sample – Hand collected data
The audited annual reports of Dutch hospitals are published on a website of the Dutch department of health care, welfare and sport. These audited annual reports are audited, mainly (98 percent) by big four audit firm. For performing the analyses of hospitals that manipulate their earnings to report profit just a range above zero, data from audited annual reports for all hospitals in the Netherlands for the period 2003 till 2012 are derived. The DTC regulation is implemented in 2005, however data of 2003 and 2004 is collected for lagged control variables in several regressions. As Brouwers (2006) has already researched earnings management in the pre-DTC regulation periods this research focusses on earnings management in the DTC-regulation period. The period of 8 years is chosen because this minimizes the risk on measuring spot trends and other incidental effects, according to Kent (1989). This eight years can be divided into three periods, as already described in chapter 2. Having such an approach allows the research to contribute on existing literature discussing the effects regulation on earnings management in not-for profit hospitals.
For the observations the full balance sheet, income statement and disclosures are obtained from the financial statements. The data is hand collected by author and other employees of a Dutch audit firm. As therefore human errors are threat for the accuracy of used data a numerous of safeguards are taken. The mathematical correctness of the used balance sheet and profit and loss accounted is verified per observation. Also every observation is verified by at least one second reviewer based upon the source data (annual report). After importing the data into SPSS the author reviewed all variables based upon minimum, maximum and mean. Only observations with sufficient data for calculating all the proxies for the analysis are kept. In table 3.1 below the number of observations per hospital are displayed.
Table 5.0.1 – Number of observations per year
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Total
N 76 76 76 76 84 84 85 85 85 85 813
Due to merges in the period 2003 till 2008, data of 9 hospitals are eliminated from the sample till 2008. This leaves the data collection with a total of 813 observations.
26
5.2 The distribution of operating income
Burgstahler and Dichev (1997) assume that cases are normally distributed if there is no earnings management. Therefore, to test our other hypothesis we have to rule out that the operating income of hospitals is normally distributed. In this study therefore to control if there are any abnormal distributions of operating income a histogram is used. This method is also used by Leone and Van Horn (2005) and Bouwens et al (2006).
5.3 Empirical model for the measurement of discretionary accruals
In the theoretical framework the measurement methods of earnings management are explained including their strengths and weaknesses. To test the zero-profit hypothesis, this paper follows the Jones (1991) model for measuring the discretionary portion of accruals. This is one of the most commonly used methodology in the accounting literature. The modified Jones model, by Dechow et al (1995) is not used because hospitals debtors from regulation period through regulation period fluctuate rather easily. An example given is the change in debtor’s and other receivables from 2011 to 2012. As due to the change in funding, al prices have to be negotiated with the health insurer. This leads to a high accrual of to be invoiced revenue and a lower trade debtor’s accrual. The modified Jones model assumes that all changes in credit sales are due to earnings management. However due to changes in regulation, as explained above, many of these changes are due to changes in regulation. Therefore we don’t assume that all changes in credit sales are due to earnings management. Bouwens et al (2006) have used the modified Jones model and the Jones model in research to measure the extent of earnings management and had similar results for both methods.
To capture the attempts of hospitals earnings management the paper follows the Jones (1991) model for measuring the discretionary portion of accruals;
∆ TAit = 1
+ ∆ REVit + PPEit + εit(1)
27 Where:
∆ TAit = difference in working capital accruals for hospital in year t to t-1
Ait-1 = total assets for hospital in year t-1
∆ REVit = difference in revenue from t to t-1
PPEit = plant property and equipment for hospital in year t
The accruals are estimated as a function of the change in plant, property and equipment and the change in activities (change in revenue). The change is defined as the difference between t and t-1. The working capital accruals are calculated as the change in current assets minus the change in current liabilities minus the change in cash added with the change in short term liabilities with credit providers and minus depreciation expenses. The change in current assets is defined as the change in inventory, inventories, work in progress, specific hospital receivables, debtors and other accruals. The current liabilities are a function of the change of provisions, work in progress (credit), specific hospital payables, creditors and other liabilities. The change in short term liabilities with credit providers are calculated as the change in the short term part of the long term loans and current accounts. The depreciation expense is the depreciation and amortization expense in t. The discretionary accruals are defined by the error term of this regression. Discretionary accruals are calculated as the unstandardized betas from the model multiplied by the variables and the differences between the actual accruals scaled by the total assets in period t-1.
Academic hospitals differ, as explained in chapter 2, from regular hospitals. An academic hospital has a basic care that matches that of the general hospitals, namely regular patient care and training function for medical specialists. In addition to these activities, the academic hospitals have a so-called top reference function, a workplace function (research and education for the medical faculty) and a development function (development of new medical technologies and treatments). These functions lead to a difference in size of the hospitals, activities and funding. Therefore the accruals for academic hospitals correlated differently with revenue than for the regular hospitals due to differences in activities. Therefore we applied the Jones (1991) model separately for academic and regular hospitals.
28
5.3 Empirical model zero-profit hypothesis part one
We use two models for the empirical analysis of the zero profit hypothesis, the Leone and Van Horn (2005) regression model and the analysis of variance. The empirical model of Leone and Van Horn (2005) is used for testing the zero-profit hypothesis, is explained in this paragraph.
DAit = λ0 + λ1EBIDAit + λ2INCOMEit-1 + λ2DAit-1 + ε (2)
Where:
DAit = discretionary accruals in period t scaled by total assets in t-1 (REG
EBIDAit-1 = operating income minus discretionary accruals in period t scaled
by total assets in t-1
INCOMEit-1 = operating income in period t-1 scaled by total assets in t-2
DAit-1 = discretionary accruals in period t-1 scaled by total assets in t-2
Discretionary accruals (DA) are calculated as the unstandardized betas from the model multiplied by the variables and the differences between the actual accruals scaled by the total assets in period t-1. DA represents the discretionary, in other words unexpected, portion of accruals which represent, according to the Jones (1991) model empirical evidence for earnings management. The independent variable EBIDAit-1 represents the earnings before manipulation. This independent variable is calculated as operating income minis discretionary accruals scaled by total assets in t-1. The sample is divided into two categories, hospitals with positive earnings (gain) before manipulation and hospitals with negative earnings (loss) before manipulation. If our results are consistent with our hypothesis we expect a negative coefficient at λ1. In the cases of losses before manipulation by discretionary accruals we expect upwards earnings management and therefore positive abnormal accruals. For the cases of gains before manipulation by discretionary accruals downwards earnings management is expected and therefore negative abnormal accruals, according to Leone and Van Horn (2005).
The model also contains two control variables. The control variable INCOMEit-1 is added to the regression as Kothari et al (2005) have shown that discretionary accruals are positively
29
correlated to prior year earnings. The lagged downwards manipulation of earnings will lead to upward manipulation in future years. Therefore a negative coefficient is expected for λ2. The control variable DA it-1 is added to the model by Leone and Van Horn (2005) to correct for the possible autocorrelation between prior year discretionary accruals and current earnings. This is because over time the sum of the total accruals have to be zero due to the reversal effect of manipulated accruals. As accruals only shift income overtime, we also expect that the mean of the discretionary accruals is close to zero.
5.4 Empirical model zero-profit hypothesis part two
As due to the mechanical relationship between DAit and EBIDAit the regression of Leone
and Van Horn (2005) could contain a measurement error. This kind of measurement error is also shown by prior research by DeFond and Park (1997). The incentives for earnings management by hospitals are disclosed in paragraph 4.2.
An analysis of variances analysis squares the differences between the mean of the categories to the categories mean and the categories in relation to the overall mean. The squared differences between the overall mean of and the category means is not assumed to be significantly different for scaled earnings before and after discretionary accruals the sample. This is because discretionary accruals only shift earnings overtime and therefore the means of income before and after earnings management are not assumed to be significantly different. As both groups have the same number of observations it is not so important if populations have the same standard deviation for performing the analysis of variances.
The Barlett’s test of equal variances (1937) is used to test equal variances against the alternative for scaled earnings before and after earnings management. If hospitals manage earnings the sample variances before and after discretionary accruals are expected to be significantly different. If hospitals use discretionary accruals to shift earnings over time, this will affect the distribution of income before and after discretionary accruals and therefore the standard deviation.
30 6. Research results
6.1 Distribution of operating income
Bugstahler (1997) assume that operating incomes are normally distributed if there is no earnings management. Discontinuities in this distribution can be an indication for earnings management. In figure 6.1 below a histogram of operating income scaled by total assets to correct for size effects is displayed. As this is only an indicator, all 813 hospitals from the sample are included, and therefore also contain 2003 and 2004, before the DTC implementation. The discontinuity around zero is obvious, 560 of the 812 cases lie between the zero and 0.2. The mean of the scaled operating income is 0.01 with a standard deviation of 0.3. The number of observations is 813, from 2003-2012. Therefore the number of observations for operating profit just above zero is higher than the number of observations of an operating profit just below zero. This is an indication for earnings management whereby hospital managers try to manipulate their operating income just above zero (binocular evidence).
Figure 6.1 – Histogram of scaled operating income
6.2 Analysis of outliers
We have reviewed the measured discretionary accruals of all hospitals. We have observed some exceptional circumstances in accruals from t to t-1. The discretionary accruals of some
31
hospitals are therefore not measured properly. By analyzing these outliers in detail we observed special circumstances disclosed in the annual reports. These circumstances made our used instrument not suitable for measuring discretionary accruals. In the following paragraphs the rational for excluding these observations is explained.
The Dutch hospital Leveste explains in their annual report of 2012 that the hospital was not able to negotiate prices timely with health insurers. As the hospital therefore has no pricelists they were not able to invoice revenue till late December 2012. This leads to a significant increase in other receivables in relation to 2011. In 2011 the other receivables are 8 million euro’s and in 2012 the other receivables are 64 million, an increase of 700%. As not invoicing revenue was not a management strategy, as it leads to no incoming cash flow for almost an entire year. Our method of measuring discretionary accruals in this observation leads to the measurement of the late billing instead of the intended measurement of the management manipulation. This billing issue was a national problem. But the majority of the hospitals had already solved a large portion of this issue by the end of the year.
The hospital Isala reports in 2004 a received financing surplus of periods before 2003 of 80 million euro’s. This amount must be repaid in the following years. As this mutation form t to t-1 captures an accrual which is not related to current earnings our measurement tool of discretionary accruals cannot be used and is therefore excluded from the sample. This is the same for the hospital Meander in the year 2008 and hospital Medical Centrum Alkmaar where they received a financing surplus of 39 respectively 61 million euro’s and is also excluded.
The Dutch permit regulation for the built of a new hospital can lead to significant mutations in the accruals form one year to another. The observation of hospital Röpcke Zweers in 2004 and 2006 is an outlier due to presenting the current account position for the built of the new hospital under the other receivables and conservation investments (permit amount). The Kennemer Gasthuis in 2005 and 2007 has presented the permit similar. In the subsequent period after the built of the new hospital this current account is reversed, which leads to a measurement error in measuring discretionary accruals. After all the discretionary accrual measured is not due to managerial manipulation of earnings but due to investments in fixed assets and therefore all four cases are excluded.
32
The last excluded outlier is hospital Spaarne because of new to current account with new related party for exploitation of the parking garage in 2004. Due to this new related party there is a significant difference between accruals in t to t-1. This difference in accruals is due to restructuring the business activities and does not capture managerial attempts of earnings management, this observation is excluded from the sample.
6.3 Discretionary and non-discretionary accruals
Table 6.1 contains the descriptive statistics of the unscaled variables used in the regression for measuring discretionary accruals. This descriptive statistics contains the unscaled variables in million euros. In this table the diversity of the dataset is visible. As for the delta variables, ⌂ TAit and ⌂ REVit, data for from prior years is used to calculate this variable, the total observations is decreased with 90 hospitals. Also, the difference between the number of observations all hospitals and the sum of regular hospitals and academic hospitals is nine. The remaining difference in number of observations is caused by excluding the in paragraph 4.2 mentioned hospitals. The mean of total assets and revenue for academic hospitals is 3.3 times the mean of regular hospitals. This is consistent with our research methodology where we divided the sample into these two categories based upon their characteristics. Hereby the skewness in the sample, driven by academic hospitals is resolved. However, there is still skewness driven by relatively small number of very large regular hospitals.
A relatively large part of the total assets of hospitals are the plant, property and equipment. This is approximately 90 percent for regular hospitals and 99.8 percent for academic hospitals. These percentages are based upon the ratio of the mean plant of property and equipment and total assets. This relatively large part can be explained by the relatively high costs of medical equipment, especially for academic hospitals who for research purposes use newly developed equipment and special equipment. Another explanation is the functional budgeting system, where historically hospitals where funded by parameters based upon size and thereby the magnitude of the plant, property and equipment. This was one of the undesired side effect of this funding structure according to Blank and Dumaij (2011a). Thereby, until 2012 all hospitals were compensated for their deprecation expenses as long as they have a permit for the PPE. Therefore there is no incentive to use capital efficiently.
33
Table 6.1 - Descriptive statistics discretionary accruals
N Minimum Maximum Mean SD
All hospitals
Total assets 813 31,113 1590,600 214,699 194,600
Total revenue 813 26,617 1236,600 199,184 191,771
⌂ Total working capital accrualsit 723 -164,200 106,650 -11,495 28,572
⌂ Revenueit 723 -97,527 173,112 11,138 18,958
Plant, Property and equipment (PPEit) 813 14,920 1088,500 138,045 126,617
Regular hospitals
Total assets 724 31,113 582,500 161,411 91,808
Total revenue 724 14,920 403,400 104,068 63,428
⌂ Total working capital accrualsit 642 -156,500 106,650 -11,237 22,409
⌂ Revenueit 642 -97,527 105,928 7,482 11,382
Plant, Property and equipment (PPEit) 724 26,617 454,200 144,673 79,977
Academic hospitals
Total assets 80 300,600 1590,600 694,946 226,982
Total revenue 80 198,400 1088,500 444,830 147,252
⌂ Total working capital accrualsit 72 -164,200 99,410 -18,3671 60,481
⌂ Revenueit 72 -44,595 173,112 43,868 35,710
Plant, Property and equipment (PPEit) 80 328,100 1236,600 693,793 208,425
As mentioned before, to measure the attempts of hospitals earnings management the paper follows the Jones (1991) model for measuring the discretionary portion of accruals. In table 6.2 the output of these two regressions is displayed. As this table shows that for regular hospitals 30.2 percent of all variances is explained by the model. For academic hospitals this is only 4.9 percent. This difference is probably due to the difference between academic hospitals themselves, as for example the academic hospital Maastricht is for example half of the size of the VU. In addition, the number of observations is, because the Netherlands only has nine academic hospitals, much lower. Therefore the regression for regular hospitals is more reliable as the model appears to remove a larger portion of the non-discretionary component of accruals. The model for academic
34
hospitals is not significant and therefore we cannot distinguish variances due to earnings management or regular activities properly.
Table 6.2 –Jones model for the measurement of discretionary accruals
The residuals of the model are used to measure the discretionary portion of accruals. The discretionary accruals are calculated as the difference between the actual mutation in working capital accruals (lagged by total assets) and the expected mutation in working capital accruals (lagged by total assets) from t to t-1. The expected mutation in working capital accruals is calculated by multiplying the betas of table 6.2 with the independent variables. If there is no earnings management, the error term should be zero. Discretionary accruals are calculated as the unstandardized betas from the model multiplied by the variables and the differences between the actual accruals scaled by the total assets in period to -1. The further the error term deviates from zero, the more is manipulated by management. However, if there is equal portion of downwards and upwards earnings management the mean is also expected to be zero. This is because positive discretionary accruals are related to positive manipulation (higher income) and negative discretionary accruals are related to negative manipulation (lower income) according to Leone and Van Horn (2005). In table 6.3 the descriptive of the unstandardized residuals of both models is displayed. Hereby the mean is close to zero, what implicates that there is an equal part of upwards as downwards earnings management. But this can also be due to the reversal effect of accruals, as
R R² Adjus. R² SE (Est.)
Regular hospitals 0,549 0,302 0,299 0,115
Academic hospitals 0,298 0,089 0,049 0,097
B SE T Sig
Regular hospitals
1scaled by TAit-1 (constant) -1,335 0,905 -1,475 0,141
REVit scaled by Tait-1 -0,133 0,067 -1,981 0,048
PPEit scaled by Tait-1 -0,082 0,014 -6,053 0,000
Academic hospitals
1scaled by TAit-1 (constant) -15,781 20,192 -0,782 0,437
REVit scaled by Tait-1 0,165 0,179 0,920 0,361
