How Does Earnings Management Affect Innovation Strategies of Firms?

Transcription

1 How Does Earnings Management Affect Innovation Strategies of Firms? Abstract This paper examines how earnings quality affects innovation strategies and their economic consequences. Previous literatures document that accounting information quality improves investment efficiency, because it reduces information asymmetry between firms and investors. Earnings management mainly affects accounting information quality. Innovation comes up with high uncertainty of future revenue and growth. Therefore, accounting information quality will play an important role on reducing information asymmetry when managers consider patent investment strategies. However, different earnings management methods have different impacts on cash flow volatility. When firms have more real earnings management activities, the cash flow volatility of firms will higher than those with less real earnings management. In contrast, conducting accrual earnings management mainly affects the reporting earnings instead of cash flow volatility. Previous literatures only examine how accrual earnings management affects investment efficiency. In-house patenting activities need consistently financial support. In contrast, patent purchasing activities need one-time financial support. Therefore, firms with high volatile cash flow have difficulty to invest in-house patenting activities. This paper tries to examine how real earnings management and accrual earnings management affect patent investment strategies and how these two earnings management strategies and patent investment strategies simultaneously affect firms value. Our findings support our arguments and indicate that different earnings management strategies have different impact on the innovation strategies and economic consequences. Key Words: Accrual Earnings Management, Real Earnings Management, Innovation Strategies

2 1. Introduction This paper examines how earnings management affects innovation strategies and their economic consequences. Biddle and Hilary (2006) indicate that high accounting information quality increase investment efficiency, because high quality accounting information reduces investors adverse selection or moral hazard and thus reduces financing costs. Because of the uncertainty of patent investment, capital providers will concern more about accounting information quality. Cohen and Zarowin (2010) indicate that after congress passes Sarbanes-Oxley Act (SOX), firms start to use real earnings management to manipulate earnings instead of using accrual earnings management, since accrual earnings management is easily to be detected by auditors. However, if firms start to use real earnings management, it not only affects earnings, but also affects cash flow. Brown and Petersen (2011) document that in-house patenting activities need to be consistently invested, because of high adjustment costs of in-house patenting activities. If firms manipulate earnings via real earnings management activities, then their cash flow will not be smooth. Therefore, real earnings management will disrupt in-house patenting. Moreover, when firms consider investing in-house patenting activities or patent purchasing, they will consider what the competitive advantages come up with these two patent investment strategies. Since these two strategies bring different investment risks, we will get further evidence to examine how these two patenting strategies contribute firm value. Innovation strategy not just only include in-house patenting activities, but also paten purchasing activities. Previous literatures have not examined how accrual earnings management and real earnings management affect firms to make patent investment strategies and their economic consequences yet. However, different patent investment 1

3 strategies will differently affect sale, market share, and profitability of firms (O Regan and Kling, 2011). We need to get further evidence to verify how the different patent investment strategies affect firm value. Brown, Fazzari, and Petersen (2009) indicate that firms have difficulty to obtain funding from debt and equity when they consider investing patenting activities. Since firms with high patenting activities have uncertainty of growth potential and limited collateral value of patenting results, those firms have difficulty to finance their patenting activities from debt holders. Because of information asymmetry, shareholders adverse selection forces firms to discount their issuing price - so called lemons premium and flotation costs. The uncertainty of patenting activities limits firms to obtain available funds for supporting their patenting activities. If firms reduce this information asymmetry by improving their accounting information quality, they will reduce their financing costs and release the constraint for patenting activities. Earnings management affects patent investment strategies. Biddle and Hilary (2006) present that accounting information quality positively increase investment efficiency. They argue that accounting information quality eliminates the information asymmetry between firms and capital providers. Due to information asymmetry, capital providers adverse selection and moral hazard will need extra premium to compensate the uncertainty they face. By using different earning management measurements (accounting aggressiveness, loss avoidance, earnings smoothing, and timeliness), they find that firms with high accounting information quality have higher investment efficiency than those with low accounting quality. Since patenting activities include high uncertainty of future growth and return, accounting 2

4 information quality will be the function of reducing information asymmetry. Whereby, high quality accounting information will play an important tool for reducing financing providers adverse selection or moral hazard. If that is the case, firms have less financing constraint for patenting activities. We infer that firms with less earnings management have high accounting information quality and high patenting investment activities. Different earnings management has different effects on patent investment strategies, since they affect cash flow volatility differently. Cohen and Zarowin (2010) indicate that real earnings management will impact cash flow. When managers conduct real earnings management, they might over produce, cut advertising expenses, cut R&D expenses, or cut SG&A. However, when managers conduct these real activities, they reduce future growth and cash flow. Once cash flow increases in current year and reduce in following years, cash flow volatility will increase. Brown and Petersen (2011) indicate that in-house patenting needs to be consistently invested because of high adjustment cost of in-house patenting. When firms start to new inhouse patenting activities, they will spend high costs on hiring skilled technology workers and training costs. When firms turn down in-house patenting activities to response temporary financing fractions, the costs of transferring skilled technology workers and dispose patenting-related equipment will be high. Because in-house patenting activities need consistent cash flow to support them, firms with volatile cash flow will have difficulty to keep investing in-house patenting activities. Cohen and Zarowin argue (2010) that real earnings management increases current net income but decreases future earnings and cash flow, whereby we infer that firms with high degree of real earnings management will be unable to conduct in-house patenting 3

5 activities, because real earnings management increases cash flow volatility. In-house patenting activities help firms take competitive advantage and increase firm value. O Regan and Kling (2011) argue that in-house patenting activities increase firms sale, market share, and profitability. Narula (2001) indicate that comparing with non-in-house patenting activities, in-house patenting activities help firms to take background competences and marginal competences. For firms, the advantage of in-house patenting activities is to have control over the right to develop their own niche. However, patent purchasing help firms to transfer the risk of failed developing patents. At the same time, it also limits firms to develop their own technology. If firms want to develop more advanced technology, they have to rely on outsiders technology or facilities. When the self-development ability is controlled by outside technology providers, the systematic risk of firms is high. CAPM model indicates that increasing systematic risk brings high beta and thus increases required return. We infer that firms with high portion of in-house patenting investments have high firm value than the firms with high portion of patent purchasing. To examine our hypotheses, we extract financial data from COMPUSTAT. Our patent purchasing data of S&P 500 firms are subtracted from LexisNexis Academic database. Both of these data are over First we examine the relation between earnings management and patent investment (including in-house patenting and patent purchasing). Second, we decompose patent investment into in-house patenting and patent purchasing and examine how real earnings management induces firms to conduct patent purchasing at a given level of accrual earnings management. Third, we examine whether firms with high level of patent purchasing have lower 4

6 firm value. Finally, we examine whether firms with high level of real earnings management and high level of patent purchasing have lowest firm value among all groups. For our models, we control leverage, refocus effect, profitability, outstanding shares, growth potential, firm size, financial constraint, and year effect. For measuring accrual earnings management, we use absolute value of discretionary accruals and estimated by cross-sectional Jones model (Jones, 1991) to measure the degree of earnings management. For measuring real earnings management, we follow Cohen and Zarowin (2010) to measure degree of real earnings management by decreases in discretionary expenses including advertising, R&D, and SG&A expenses. In order to capture the total effects of real earnings management, we combine two individual measures to compute comprehensive metrics of real earnings management activities. For our first real earnings management measurement, RM_1, we follow Zang (2006) and multiply abnormal cash flows from operations and abnormal discretionary expenses by minus one, separately. Then we aggregate these two numbers to get RM_1. Our patent purchasing data of S&P 500 firms is subtracted from LexisNexis Academic database. Both of these data are over When we navigate patenting purchasing data from LexisNexis Academic database, we have key words Firm Name AND Patent AND Acquisition and research it within major world publications. Our patenting purchasing information is including the announcement data of patent purchasing, the amount of patent purchasing, the number of patent purchasing, and the target firm information. Since we have data of 5

7 the amount of patent purchasing, we measure the value of PATENT_PURCHASE as to the amount of patent purchasing in certain year. We measure the value of in-house patenting as the amount of research and development expenses (XRD) in certain year. We measure the value of PATENT_PURCHASE+XRD as sum of the value of PATENT_PURCHASE and the value of XRD at certain year. Final, we measure the value of PATENT_PURCHASE/XRD as the ratio of the amount of patent purchasing (PATENT_PURCHASE) to the amount of research and development expenses (XRD). The main findings support our hypothesis that earnings management decreases patent investment. These outcomes also consist with Biddle and Hilary s (2006) argument that accounting information quality can improve firm-level investment, since high accounting information quality reduces adverse selection and moral hazard problem. Our findings also support Cohen and Zarowin (2010) that real earnings management will impact cash flow, since it will directly affect expenditures or investment strategies of firms. The evidences also support (Brown and Petersen, 2011) that firms with high real earnings management will conduct more patent purchasing activities, because of high volatility of cash flow resulted from real earnings management and the need of cash holding buffers for in-house patenting. Finally, we find out that no matter which kind of earnings management and thus the innovative activity a firm conducts, earnings management increases information asymmetry between investors and managers and impairs firms performance anyway regardless of what patent strategy has been executed. The contributions of this paper are two. First, we provide further evidence to 6

8 indicate that how real earnings management and accrual earnings management affect patent investments. Previous literatures either examine the relationship between accrual earnings management and in-house patenting or examine the relation between real earnings management and in-house patenting. However, these two earnings management methods have different impacts on cash flow volatility and also these two patenting investment have different cash flow needs (consistently invest or onetime invest). Therefore, we provide further evidence to explain earnings management strategies and patent investment strategies. Second, we provide further evidence to explain how these two patent investment strategies affect firm value. The article in Wall Street Journal on October 24, 2011 indicates: Many companies are devoted to innovation but they tend to have little to show for their spending on research and development, according to a new Booz & Co. report. We try to verify whether inhouse patenting activities are value-increasing activities. The remainder of this paper processes as following order. Section 2 develops the hypotheses. Section 3 describes the research design. Section 4 presents the main results. Section 5 presents additional test. Section 6 has our conclusions. 7

9 2. Literature Review 2.1 Earnings Management and Patent Investment Investors adverse selection increase cost of capital when companies finance their investment projects. Myers and Majluf (1984) indicate that due to information asymmetry between managers and outside investors, managers will use equity financing when stock price is overvalued. Investors know about this situation, therefore, when companies announce to use equity financing, investors will negatively react on stock price to get so call lemon premium. This lemon premium might force managers to forgo positive-npv project to financing outside capital. Williamson (1975) argues that by creating internal capital market with lower information asymmetry, firms do not face underinvestment problem due to introducing outside capital, especially when the outside capital market is imperfect market where has high information asymmetry between firms and investors. Due to imperfect information transparency, outside capital market, business-government relation, legal system, that make diversified-firms can be easier for survive than focus firms. Information asymmetry increase financing costs. Hess and Bhagat (1986) find that information asymmetry costs the outside investors more with regard to the expenses associated with transactions and gathering information. Lim, Thong, and Ding (2008) also support this argument. They document that due to complexity business organization, high information asymmetry between insiders and outsiders of firms induce managers to manipulate earnings before firms conduct SEOs. However, firms with certain characteristics might earn positive reaction when they announce to 8

10 SEO. Jensen (1986) proposes that announcing the intention to conduct SEOs, firms with the characteristic of high-profit-potential receive a more positive reaction from the market, while the firms with the characteristic of low-profit-potential receive a negative one. These findings support one idea that market might not always negatively react to SEOs announcement. Earnings management decreases investment efficiency. Biddle and Hilary (2006) argue that accounting information quality can improve firm-level investment, since high accounting information quality reduces the information asymmetry between firms and investors, and thus investors adverse selection and moral hazard problem will be eliminated. They find that higher accruals earnings management reduces firmlevel investment. They also indicate that accounting information quality play a relatively important role in the countries where stock market dominates the sources of capital. This finding indicates that transparent information is the key factor for reducing lemon premium when firms need to finance their investment projects. Biddle, Hilary, and Verdi (2009) also indicate that higher financial reporting quality can make valuable investments more visible and reduce investors adverse selection in the issuance of securities. Earnings management may damage patent investment. Brown, Fazzari, and Petersen (2009) indicate that the uncertain and volatile return of patent investments limit firms to finance their patent investment from external capital market. One of considerations by external financing providers for patent investments is the uncertainty of future cash flow of patent investment. They indicate that even though equity financing is the main funding resource of patent investment, several 9

11 disadvantages of equity financing still restrict firms to invest patenting activities. When firms issue new shares, it requires pay a lemon premium to shareholders due to information asymmetry. Shareholders adverse selection problem downwards share price when firms announce to issue new shares. Due to uncertainty return of patent investment, information transparency plays an important role on eliminating lemon premium when firms finance their patent investment projects. Since Biddle and Hillary (2006) document that earnings quality reduces investment efficiency because of investors adverse selection, we infer that earnings quality dominates an important role on patent investments. Previous literatures document that earnings quality reduces investment efficiency. Patent investments are characterized as high uncertain and more volatile future return. Earnings management decreases patent investment projects, because it increases information asymmetry and investors have difficulty on detecting uncertainty and volatile return of patent investments. Based on our inference, we develop our first hypothesis: H1: Earnings management decreases patent investments of firms. 2.2 Earnings Management, In-House Patenting, and Patent Purchasing Last section infers that earnings management may decrease accounting information quality and thus reduces patent investment. Different earnings management methods have different impacts on cash flow volatility. At the same time, in-house patenting and patent purchasing need different financing needs. This section 10

12 further investigates how earnings management methods affect patent investment strategies. Roychowdhury (2006), and Cohen and Zarowin (2010) indicate that firms not just only do real earnings management, but also do accrual earnings management. Cohen and Zarowin (2010) indicate that managers prefer manipulating earnings by real earnings management more than accrual earnings management in the post Sarbanes-Oxley Act (SOX) period. They argue that Sarbanes-Oxley Act accrual earnings management would be easily detected by auditors, whereby, managers prefer manipulating earnings by real activities. Real earnings management and accrual earnings management have different impact on firm cash flow. Roychowdhury (2006) argues that the purpose of earnings management is to meet certain earnings targets. The difference of real earnings management and accrual earnings management is its impacts on cash flows. Roychowdhury indicates that real earnings management affects cash flows, but accrual earnings management has impact on cash flow. If managers over produce or cut discretionary expenditures, such as R&D expenses, advertising expenses, or SG&A expenses, these real activities will directly affect firm cash flow. Real earnings management damages firms profitability and value. Cohen and Zarowin (2010) indicate that managers can manipulate earnings with three main real activities. First, they can boost sales through increase price discounts or more lenient credit terms; Second, they can conduct overproducing to increase inventories at the end of year and decrease cost of goods sold; Third, they can cut discretionary expenses, including R&D expenses, advertising, and SG&A expenses. They also 11

13 argue that reducing discretionary expenses increases current earnings and cash flow, but might reduce future earnings and cash flow. According to Cohen and Zarowin s argument, the volatility of cash flow will be high after firms conduct real earnings management. Hsu (2009), and Brown, Fazzari, and Petersen (2009) indicate that patent investment is an important factor for the firms future grow. Hsu (2009) find that patent investment can increase future stock returns and premiums, since patenting activities are the main driving force for economic growth and fluctuations. Cohen and Zarowin (2010) point out that firm with high real earnings management have lower future ROA. Because real earnings management affects future cash flow, we infer that real earnings management will increase cash flow volatility. In-house patenting needs consistent cash flow to support it (Brown and Petersen, 2011). Once firms involve in-house patent investments, they will face higher amount of adjustment costs, such as hiring skilled technology workers and training costs. If firms change in-house patenting strategies to response temporary financing fractions, they will suffer severe losses on transfer skilled technology workers or patenting facilities. Brown and Petersen (2011) indicate that there are three characteristics of in-house patenting. First, limited collateral value and information asymmetry of patenting investments, firms will face financing fractions for financing in-house patenting activities. Second, limited collateral value and uncertainty of future cash flow force firms to finance their in-house patenting activities with volatile sources of finance like internal cash flow or stock issues. Third, firms will spend high adjustment costs of in-house patenting activities, if they change their in-house patenting strategies. Because of high adjustment costs of in-house patenting investments, firms have to finance their in-house patenting investments with consistent cash flow. Once 12

14 the volatility of cash flow is high, firms will have difficulty to keep investing in-house patenting activities. Brown and Petersen (2011) indicate that cash holding buffers inhouse patenting activities when firms face financing fractions, since cash holding can help firms to avoid high amount of adjustment costs of in-house patenting activities. Real earnings management (using opportunistic reduction of R&D) result in high cash flow volatility (Roychowdhury, 2006). There is also empirical evidence that firms manage earnings using activities with cash flow consequences. Most of the evidence centers on the opportunistic reduction of R&D expenses. Bens, Nagar and Wong (2002) find that managers of firms facing EPS dilution because of exercises of employee stock options (ESOs) repurchase stock. Managers partially finance these repurchases by reducing R&D. There is similar evidence for firms facing EPS dilution from outstanding ESOs in Bens, Nagar, Skinner and Wong (2002). The Dechow and Sloan (1991) find CEOs in their final years reduce spending on R&D to increase short-term earnings. Bushee (1998) examines firms trying to meet previous year s earnings and find that they reduce R&D more if they have lower institutional ownership. He interprets this as evidence that the R&D reductions by this set of firms are potentially value-destroying and are prevented by the presence of sophisticated investors. Previous literatures only focus on how financing resources affect in-house patenting activities. Contrast to in-house patenting investments, patent purchasing has different financing need. In-house patenting activity needs consistent cash flow to support it, but patent purchasing doesn t. When firms involve patent purchasing, they don t have to invest huge amount of capital on hiring high skilled technology workers 13

15 and patenting facilities. Contrasting to in-house patenting activities, patenting purchasing doesn t have to pay high amount of adjustment costs. Therefore, firms don t need consistent cash flow to buffer the costs of patent purchasing. Based on Brown and Petersen s arguments (2011), in-house patenting activities need consistent financing resources to support them. Biddle and Hillary (2006) indicate that earnings management increases information asymmetry between investors and firms and whereby increases underinvestment problems. Since patent investments have higher degree of uncertainty, information transparency will be an important factor for the capital providers. According to Biddle and Hillary s findings, we infer that firms with high degree of earnings management will face underinvestment problems in patent investment. However, real earnings management and accrual earnings management have different impact on cash flow volatility. These different impacts will affect patenting investment strategies. According to Cohen and Zarowin s findings (2010), we infer that since real earnings management will impact cash flow, cash flow volatility of firms will be higher than it in the firms with high accrual earnings management. In contrast, since accrual earnings management will not affect cash flow, cash flow volatility of firms will be lower than it in the firms with high real earnings management. According to our previous inference, we develop our following hypothesis: H2: At given level of accrual earnings management, firms with high real earnings management will prefer to patent purchasing. 2.3 Patent Investment and Firm Value 14

16 In-house patenting and patent purchasing have different impacts on firm value. O Regan and Kling (2011) indicate that firms with significant in-house patenting investment outperform other firms in terms of sale, market share, and profitability. Comparing with non-in-house patenting, in-house patenting has more background competences and marginal competences (Narula, 2001). In contrast, patent purchasing has to rely on providers patent developing skill. Once the technology development skills are controlled by outsiders, firms will have difficulty to develop their own skill forward. Therefore, patent purchasing brings high systematic risk to firms. Moreover, Paster and Veronesi (2009) argue that new technologies bring high uncertainty of future productivity and come up with observed stock price patterns. When firms develop new technologies, the uncertainty of the future productivity increases idiosyncratic risk, because small scale of production and a low probability of a large-scale adoption. This argument indicates that developing in-house patenting increases idiosyncratic risk and patent purchasing increases systematic risk. According to CAPM model, increasing systematic risk brings high beta and thus increases required return. We infer that paten purchasing decreases firm value. Therefore, based on our inference, we develop the following hypothesis: H3: Firm with high level of patent purchasing has lower firm value than those with high level of in-house patenting. 2.4 Patent Investment, Earnings Management, and Firm Value Real earnings management increases patent purchasing. Cohen and Zarowin (2010) indicate that real earnings management will impact cash flow, since it will 15

17 directly affect expenditures or investment strategies of firms. Since real earnings management will affect cash flow, it will affect cash flow volatility of firms as well. In other words, since accrual earnings management will not affect cash flow, cash flow volatility of firms will be lower than it in the firms with high real earnings management. Brown and Petersen (2011) indicate that in-house patenting activities need consistent financing resources to support them, because firms have to pay high adjustment costs for turning down in-house activities. Based on Cohen and Zarowin s finding (2010), we infer that firms with high real earnings management will suffer high volatile cash flow, since they cannot consistently support in-house patenting. However, if firms want to buy patent from outside, they just consider whether they can afford it without considering cash flow volatility. Therefore, firms will consider to conduct patent purchasing when firms with high real earnings management. Under this situation, high patent purchasing will increase systematic risk and decrease firm value. According to our inference, at a given level of accrual earnings management, high real earnings management will induce firms to conduct patent purchasing because of high cash flow volatility. Since patent purchasing increases systematic risk, firms with high systematic risk will increase required return and thus decrease firm value. We develop our hypothesis as follow: H4: At the given level of accrual earnings management, firms with high real earnings management and high level of patent purchasing have lowest firm value among all groups. 16

18 3. Data and Methodology To examine our hypotheses, we obtain financial data and in-house patenting data from COMPUSTAT. Our patent purchasing data of S&P 500 firms are subtracted from LexisNexis Academic database. Both of these data are over When we navigate patenting purchasing data from LexisNexis Academic database, we have key words Firm Name AND Patent AND Acquisition and research it within major world publications. Our patenting purchasing information is including the announcement data of patent purchasing, the amount of patent purchasing, the number of patent purchasing, and the target firm information. For examine our hypothesis, first we examine the relation between earnings management and patent investment (including in-house patenting and patent purchasing). Second, we further decompose patent investment into in-house patenting and patent purchasing. We examine how real earnings management induces firms to conduct patent purchasing at a given level of accrual earnings management. Third, we examine whether firms with high level of patent purchasing have lower firm value comparing with the firms with low level of patent purchasing. Finally, we examine whether firms with high level of real earnings management and high level of patent purchasing have lowest firm value among all groups. For our earnings management equation, we control leverage, refocus effect, profitability, outstanding shares, growth potential, firm size, financial constraint, and year. For measuring accrual earnings management, we use absolute value of discretionary accruals and estimated by cross-sectional Jones model (Jones, 1991) to measure the degree of earnings management. For measuring real earnings management, we follow Cohen and Zarowin (2010) to measure degree of real 17

19 earnings management by decreases in discretionary expenses including advertising, R&D, and SG&A expenses. In order to capture the total effects of real earnings management, we combine two individual measures to compute comprehensive metrics of real earnings management activities. For our first measure, RM_1, consistent with Zang (2006), we first multiply abnormal cash flows from operations and abnormal discretionary expenses by negative one and then aggregate them into one measure. 3.1 Accrual Earnings Management For measure accrual earnings management, we follow Jones (1991) and use cross-sectional Jones model and cross-sectional Modified Jones model to estimate discretionary accrual (DAC). First, we estimate discretionary accrual from crosssectional Jones model as follow: TA it Assets it 1 = Assets it 1 + β 1 [ REV it PPE it ] + β Assets 2 [ ] + ε it 1 Assets it (1) it 1 Where TA is total accruals estimated from following equation. TAt = Current Assetst Current Liabilitiest Casht + Current Maturities of Long-Term Debtt Depreciation and Amortization Expenset. Assets are total assets. REV are change in revenue. PPE are gross property plant and equipment. The residuals (ε) from the discretionary accrual models are used as estimates of discretionary accruals. 3.2 Real Earnings Management We follow Cohen and Zarowin (2010) to measure degree of real earnings management by decreases in discretionary expenses including advertising, R&D, and 18

20 SG&A expenses. We estimate the abnormal discretionary expenses are estimated as the deviations from the following equation: DISX it 1 SALES = K Assets 1 + K it 1 i,t 1 Assets 2 + ε i,t 1 Assets it (2) i,t 1 Where DISX are discretionary expenses during the year, and are calculated by the sum of advertising expenses, R&D expenses and SG&A. In order to capture the total effects of real earnings management, we compute another proxy, RM1, consistent with Zang (2006); we first multiply abnormal cash flows from operations and abnormal discretionary expenses by negative one and then aggregate them into one measure. We multiply by negative one, so that the higher these amounts the more likely that the firm is engaging in sales manipulations and cutting discretionary expenditures to manager reported earnings upwards. We acknowledge that these individual variables underlying RM1 may have different implications for earnings that may dilute any results using these aggregated measures. We thus report results corresponding to both the aggregated measures as well as the individual real earnings management proxies (REM_DISX). 3.3 Patent Investment, In-House Patenting, and Patent Purchasing To measure innovative investment conduct by firms, we obtain financial data and in-house patenting data from COMPUSTAT and collect the data of research and development expenses as in-house patenting (XRD). Our patent purchasing data of S&P 500 firms are subtracted from LexisNexis Academic database. Both of these 19

21 data are over When we navigate patenting purchasing data from LexisNexis Academic database, we have key words Firm Name AND Patent AND Acquisition and research it within major world publications. Our patenting purchasing information is including the announcement data of patent purchasing, the amount of patent purchasing, the number of patent purchasing, and the target firm information. We use following four methodology to capture firms innovative investment, PATENT_PURCHASE, XRD, PATENT_PURCHASE+XRD, PATENT_PURCHASE/XRD. Since we have data of the amount of patent purchasing, we measure the value of PATENT_PURCHASE as to the amount of patent purchasing in certain year. We measure the value of in-house patenting as the amount of research and development expenses (XRD) in certain year. We measure the value of PATENT_PURCHASE+XRD as sum of the value of PATENT_PURCHASE and the value of XRD at certain year. Final, we measure the value of PATENT_PURCHASE/XRD as the ratio of the value of PATENT_PURCHASE to the value of XRD. 3.4 Models To test our first hypothesis, we follow Brown, Fazzari, and Petersen (2009) and develop to following model: Patent j,t = β 0 + β 1 Patent j, t 1 + β 2 EM j,t + β 3 MarketBook j, t + β 4 Sgwth j, t + 20

22 β 5 CashFlow j, t + β 6 CashFlow j, t 1 + β 7 StkIssues j, t + β 8 StkIssues j, t 1 + β 9 DbtIssues j, t + β 10 DbtIssues j, t 1 + Year + ε j,t (3) Where, Patent means four kind of patent investments (measured by PATENT_PURCHASE, XRD, PATENT_PURCHASE+XRD, PATENT_PURCHASE/XRD.). EM includes accrual earnings managements (AEM) and real earnings managements (REM). AEM means accrual earnings management measured by Jones model. REM means real earnings management measured by decreases in discretionary expenses including advertising, R&D, and SG&A expenses (REM_DISX); RM1, consistent with Zang (2006), we first multiply abnormal cash flows from operations and abnormal discretionary expenses by negative one and then aggregate them into one measure. MarketBook is Market-to-Book ratio. Sgwth is Sales growth. CashFlow is equal to sum of cash and short-term investments divided by total assets. StkIssues is equal to net cash raised from stock issued divided by total assets. DbtIssues is equal to net new long-term debt issued divided by total assets. Year is to control year fixed effect. To test our second hypothesis, we separate our sample into four groups: (1) firms with high degree of accrual earnings management and high degree of real earnings management; (2) firms with high degree of accrual earnings management and low degree of real earnings management; (3) firms with low degree of accrual earnings management and high degree of real earnings management; (4) firms with low degree of accrual earnings management and low degree of real earnings management. By following Cohen and Zarowin s (2010) argument, we infer that firms with high degree of real earnings management have high cash flow volatility and thus disable 21

23 to invest in-house patenting. By following Biddle and Hilary s (2006) argument, we infer that firms with high degree of accrual earnings management have underinvestment problem and thus disable to conduct patent purchasing. For testing our second hypothesis, we develop the following model: Patent j,t = β 0 + β 1 Patent j, t 1 + β 2 AEM j,t + β 3 REM j,t + β 4 AEM REM j,t +β 5 MarketBook j, t + β 6 Sgwth j, t + β 7 CashFlow j, t + β 8 CashFlow j, t 1 + β 9 StkIssues j, t + β 10 StkIssues j, t 1 + β 11 DbtIssues j, t + β 12 DbtIssues j, t 1 + Year + +ε j,t (4) Where, Patent means four kind of patent investments (measured by PATENT_PURCHASE, XRD, PATENT_PURCHASE+XRD, PATENT_PURCHASE/XRD.). EM could mean two different earnings managements, accrual earnings managements (AEM) and real earnings managements (REM). AEM means accrual earnings management measured by Jones model. REM means real earnings management measured by decreases in discretionary expenses including advertising, R&D, and SG&A expenses (REM_DISX); RM1, consistent with Zang (2006), we first multiply abnormal cash flows from operations and abnormal discretionary expenses by negative one and then aggregate them into one measure. And the rest of variables are defined as same as previous section. The third hypothesis examines whether patent purchasing reduces firm value because of high systematic risk of patent purchasing. For testing third hypothesis, we develop the following model: 22

24 TobinQ j, t = β 0 + β 1 Patent j, t + β 2 Firm Size j, t + β 3 Leverage j, t + β 4 Liquidity j, t + β 5 Profitability j, t + Year + ε j,t (5) TobinQ is equal to the ratio of market value of equity plus book value of total liabilities to book value of total assets. Patent means four kind of patent investments (measured by PATENT_PURCHASE, XRD, PATENT_PURCHASE+XRD, PATENT_PURCHASE/XRD.). Firm Size is the value of natural logarithm of total assets. Leverage is the ratio of total liabilities to total assets. Liquidity is the ratio of cash plus short term investment to total assets. Profitability is the ratio of EBITDA to total assets. Year is to control year fixed effect. Industry is to control industry fixed effect. ε is error term. The forth hypothesis further examine whether firms with real earnings management and conducting patent purchasing have less firm value among other group. For testing forth hypothesis, we develop the following model: TobinQ j, t = β 0 + β 1 Patent j, t + β 2 Patent AEM j, t + β 3 Patent REM j, t + β 4 AEM j,t + β 5 REM j,t + β 6 Firm Size j, t + β 7 Leverage j, t + β 8 Liquidity j, t + β 9 Profitability j, t + Year + ε j,t (6) TobinQ is the ratio of the sum of market value of equity and book value of total liabilities to book value of total assets. Patent means four kind of patent investments (measured by PATENT_PURCHASE, XRD, PATENT_PURCHASE+XRD, 23

25 PATENT_PURCHASE/XRD.). EM could mean two different earnings managements, accrual earnings managements (AEM) and real earnings managements (REM). AEM means accrual earnings management measured by Jones model. REM means real earnings management measured by decreases in discretionary expenses including advertising, R&D, and SG&A expenses (REM_DISX); And RM_1, consistent with Zang (2006), we first multiply abnormal cash flows from operations and abnormal discretionary expenses by negative one and then aggregate them into one measure. And the rest of variables are defined as same as previous section. Firm Size is the value of natural logarithm of total assets. Leverage is the ratio of total liabilities to total assets. Liquidity is the ratio of cash plus short term investment to total assets. Profitability is the ratio of EBITDA to total assets. Year is to control year fixed effect. Industry is to control industry fixed effect. ε is error term. 24

26 4. Empirical Results 4.1 Descriptive Statistic Table 1 presents the descriptive statistic of all variables. The mean value of patent measured by PATENT_PURCHASE, XRD, PATENT_PURCHASE+XRD and PATENT_PURCHASE/XRD respectively is 0.000, 0.029, 0.029, and The mean value of absolute value of discretion accruals are 0.133, 0.134, and 0.192, estimated by jones model( JONES ), Modified Jones Model( MODIFIEDJONES ), Performance Matched Jones Model( KOTHARIJONES ), respectively. The mean value of real earnings managements are 0 and 0.007, estimated by approaches used in Cohen and Zarowin (2010). These are measured by decreases in discretionary expenses including advertising, R&D, and SG&A expenses (REM_DISX); And RM1, consistent with Zang (2006), to carry out this measure, we first multiply abnormal cash flows from operations and abnormal discretionary expenses by negative one and then aggregate them into one measure. <Table 1 Inserts Here> 4.2 Univariate Comparisons Table 2 panel (A) shows our t-test on main variables within Low Accrual Earnings Management and High Accrual Earnings Management. We find that firms with high accrual earnings management have less patent purchasing. The means of PATENT_PURCHASE among these two categories are and This finding supports previous literatures which indicate accrual earnings management results in information asymmetry between investors and owners, and hence decreasing firms 25

27 patent purchasing activities. The other finding about firms with low accrual earnings management against high accrual earnings management relates to in-house patenting is that firms with high accrual earnings management have higher in-house patenting. Based on Roychowdhury (2006) that real earnings management affect firms cash flow and Brown and Petersen s arguments (2011), in-house patenting activities need consistent financing resources to support them. Under same level of real earnings management, firms with higher accrual earnings management would involve itself into relative high level of in-house patenting. Panel B and Panel C of Table 2 show our t-test on main variables within firms with low real earnings management (measured by DISX and RM1) against firms with high real earnings management. In Panel B of Table 2, the means of XRD are 22,134 against 12,657. And in Panel C of Table 2, the means of XRD are 23,172 against 12,851. We find that firms with high real earnings management invest less on inhouse patenting. This finding consists with our hypothesis that firms with higher real earnings management would affect its cash flow and hence have difficulty maintain its in-house patenting. <Table 2 Inserts Here> 4.3 Multivariate Regression Analysis First, we examine whether earnings management decreases patent investment. Biddle and Hilary (2006) argue that accounting information quality can improve firmlevel investment, since high accounting information quality reduces adverse selection and moral hazard problem. They find that higher accruals earnings management reduces firm-level investment. Therefore, we predict that earnings management decreases patent investment projects, because it increases information asymmetry and 26

28 investors have difficulty on detecting uncertainty and volatile return of patent investments. Panel (A) and Panel (B) of Table 3 shows that the coefficients on real earnings management through DISX and RM1 are negative when patent investment are measured by PATENT_PURCHASE, XRD and PATENT_PURCHASE+XRD, respectively. These results indicate that doing real earnings management will increase the cash flow volatility and thus reduce innovation investment which needs the consistent cash flow supporting. Panel (C) of Table 3 shows that coefficients on accrual earnings management measured by JONES are positive when patent investment is measured by PATENT_PURCHASE, XRD and PATENT_PURCHASE+XRD, respectively. These results indicate that accrual earnings management doesn t impair the consistency of cash flow; therefore, firms have ability to keep investing innovation investment. So far, our findings don t support the first hypothesis, that is, information asymmetry may decrease patent investment of firms. However, our findings support cash flow volatility theory, that is, cash flow volatility is the main concern for the managers while they conduct innovation investment. We further examine whether at given level of accrual earnings management, firms with high real earnings management will prefer to patent purchasing. Cohen and Zarowin (2010) find that managers prefer manipulating earnings by real earnings management more than accrual earnings management for avoiding auditors detection. We find that firms like to do real earnings management and reduce in-house patenting despite these firms may conduct accrual earnings management in the same time. Next, we decompose patent investment into in-house patenting and patent purchasing and examine how real earnings management induces firms to conduct patent purchasing at a given level of accrual earnings management. 27

29 <Table 3 Inserts Here> Panel (A) and Panel (B) in Table 4 present whether firms with high real earnings management will prefer to patent purchasing after we consider the effect of accrual earnings management toward firms patent investment decision. We argue that firms with high real earnings management would conduct more patent purchasing activities, because of high volatility of cash flow resulted from real earnings management and the need of cash holding buffers for in-house patenting (Brown and Petersen, 2011). According to the results of Panel (A) and (B) in Table 4, the coefficients on real earnings management (DISX and RM1) are negative when the patent investments are measured by PATENT_PURCHASE and PATENT_PURCHASE+XRD. Our findings support our second hypothesis, that is, firms with high real earnings management will prefer investing patent purchasing activities instead of in-house patenting activities. These findings support our hypothesis that both information asymmetry and volatility of cash flow has negative influence on patent investments. And Panel (A) of Table 4 shows the coefficient on REAL_JONESDISX is negative against XRD and PATENT_PURCHASE+XRD. These results indicate that when firms conduct accrual earnings management and real earnings management simultaneously, the effect of accrual earnings management and real earnings management against firm s preference toward patent investment would offset each other. Next we examine whether firms with high level of patent purchasing have lower firm value. <Table 4 Inserts Here> O Regan and Kling (2011) indicate that firms with significant in-house patenting investment outperform other firms in terms of sale, market share, and profitability. Comparing with non-in-house patenting, in-house patenting has more background 28

30 competences and marginal competences (Narula, 2001). Table 5 shows a different view from previous literature, all kind of patent investment measured by different way (PATENT_PURCHASE, XRD, PATENT_PURCHASE+XRD and PATENT_PURCHASE/XRD) have positive coefficient on firms value (measured by TOBINSQ). Especially, we find that the coefficients on XRD and PATENT_PURCHASE+XRD are significantly positive. This result supports O Regan and Kling s (2011) argument that in-house patenting activity is the way for creating the value of firms. These findings support our third hypothesis that firm with high level of patent purchasing has lower firm value than those with high level of inhouse patenting. Next, we are going to examine whether a firm with high real earnings management and high level of patent purchase have lowest firm value. <Table 5 Inserts Here> Real earnings management increases patent purchasing. Cohen and Zarowin (2010) indicate that real earnings management will impact cash flow, since it will directly affect expenditures or investment strategies of firms. Since real earnings management will affect cash flow, it will affect cash flow volatility of firms as well. In other words, since accrual earnings management will not affect cash flow, cash flow volatility of firms will be lower than it in the firms with high real earnings management. Brown and Petersen (2011) indicate that in-house patenting activities need consistent financing resources to support them, because firms have to pay high adjustment costs for turning down in-house activities. Hence, firms with high real earnings management have high level of patent purchasing and have relative lower firm value. Table 6 shows the results. The coefficients on most interaction terms are negative, in term of XRDXDISX, XRDXJONES, (PATENT_PURCHASE+XRD)XDISX, and 29

31 (PATENT_PURCHASE+XRD)XJONES. The possible explanation of these results would be no matter which kind of earnings management a firm conducts, earnings management increases information asymmetry between investors and managers. Although conducting innovation activities represents growth potential of firms, it represents the uncertainty of future cash flow as well. In other words, firms with high innovation activities have high information asymmetry between investors and managers. Managing earnings just represents that firms with high innovation activities have high information asymmetry. And the firm value is impaired due to information asymmetry created by earnings management, despite the patent investment is positive to firm value. Our findings partially support the fourth hypothesis that firms with high real earnings management and high level of patent purchasing have lowest firm value among all groups. <Table 6 Inserts Here> 30