Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ FINANCIAL PROFILING OF PROFITABLE INTERNET FIRMS

Yolanda Fuertes-Callén* Department of Accounting and Finance University of Saragossa, Spain [email protected] José Antonio Laínez-Gadea Department of Accounting and Finance University of Saragossa, Spain [email protected]

Biographical notes: Yolanda Fuertes is an Assistant Lecturer at the Department of Accounting and Finance of the University of Saragossa. She has been a research visitor at the Universities of Sheffield (United Kingdom) and Stern at New York (USA). Her research interests include: E-business; multivariate mathematical models; Intellectual Capital, and Information Technologies in Accounting and Finance. Dr. Fuertes has published articles in Journals such as Decision Support Systems, The International Journal of Digital Accounting Research or Online Information Review. José Antonio Laínez is a Professor in Accounting and Finance at the University of Saragossa in Spain. His main area of researh is related to international accounting, harmonisation, international accounting standards, and capital market studies. He has published a large number of books and articles in the financial reporting area in Journals such as The European Accounting Review, Advances in International Accounting and The International Journal of Accounting. He has given numerous conferences to spanish, european and Latin American universities. Laínez’s international reputation in the area of international accounting is reflected both in the continuing provision of major EU funding over the last 5 years and in invitations from the United Nations Organization to reprensent to Spain in the “Intergovermental Working Group of Experts on Accounting and Reporting” and from the European Comission to represent to Spain in the “Accounting Advisory Forum”. *

Corresponding author. Gran Vía 2, 50005 Zaragoza, Spain. Telephone: +34 976 761000 Ext. 4652. Fax number:

+34 976 761769

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________

ABSTRACT

This paper seeks to identify the characteristics that differentiate Internet firms achieving positive operating profits from persistent loss makers. The study uses a sample of firms operating different online business models during the period 1998-2002. Our findings indicate that profitable firms achieve better solvency and liquidity due to the contribution of fresh capital by investors and increasing self-financing, and have lower indebtedness compared to loss makers. The continuing inability of a large number of Internet businesses to achieve profitability leads both investors and venture capital firms, the main sources of finance open to the firms during the period 1998-2000, to concentrate on other sectors and tighten the conditions that they are willing to fund ventures. These actions oblige firms to borrow while continuing losses and rising debt affect their solvency. The tremendous changes that took place in the Internet sector during the period—from initial boom to large-scale bust—further affected our ability to generalize results. As a result, we were confined to the time periods referred to in the financial statements of the sample firms. In this light, it will be of considerable interest if our findings are confirmed in the future as pertinent data becomes available. Our study can help both the academic community and business professionals better understand the turbulence in electronic markets over the past five years. This paper provides an opening for an interesting line of research aimed at clarifying the specifics of firms operating in the Internet sector.

Key Words: Accounting, Internet firms, e-commerce, financial analysis, profitability, logit analysis.

1

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Financial Profiling of Profitable Internet Firms 1. Introduction Profitability is the major challenge facing Internet businesses. While we should not doubt the importance already attained by the Internet and its new technologies, earnings and the outlook for profits are fundamental factors to investors backing online companies. The bursting of the Internet bubble has forced investors to shift their focus from the prospects for growth of issuing firms to the question of if and when profitability can be achieved. The new mantra for investors has become “path-to-profitability” (Jain and Kini, 2004). Issuing firms that are unable to credibly demonstrate a clear path to profitability are swiftly punished with steeply lower valuations. They also face obstacles in securing external financing. The study of business profitability and the identification of the economic and financial variables that characterize the most profitable e-business firms are some of the most interesting avenues of research open to financial economists and business administrators. E-business is, however, still in its infancy and faces numerous problems associated with the use of financial data, such as the lack of historical data, precipitous changes in expenses and revenues, the absence of profits, the constant creation and destruction of firms, and numerous mergers and takeovers. As a result, few studies of the profitability of Internet firms have appeared to date. To gain a greater understanding of these factors, this paper analyzes a sample of Internet firms during the period 1998-2002 and identifies the operational and financial variables that best distinguish the most profitable online businesses from those that are unprofitable. Using logit analysis, we have applied a static analysis of the financial variables that characterize the most profitable firms. We have also measured changes in dynamic variables in the relative position of each company for each of several financial ratios. Our findings are of interest to dot-coms because they provide a model that will help managers emulate successful businesses through benchmarking policies. Our findings are also of interest to investors because they will help them make better decisions when evaluating whether to back a dot-com or some other venture.

2

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ We have organized our paper into five sections: the first presents an overview of the topic; the second reviews earlier studies of business profitability; the third describes the analytical methodology we have applied in detail; and the fourth presents our findings from the application of this methodology to our sample of firms. The fifth section concludes with a summary of our findings and a discussion of future work. 2. Overview The second half of the 1990s witnessed the appearance of new technologies— especially the Internet—that generated enormous expectations with regard to the creation of new businesses and the transformation of the traditional economy. As a result, such firms enjoyed an unprecedented explosion in value as investors paid enormous sums for loss-making companies that nevertheless held out the promise of abundant future returns. Unfortunately, many companies failed to achieve as expected. For example, the online fashion store Boo.com attracted support from major investors worldwide. When it filed for bankruptcy, many New Economy firms—which until then had enjoyed lofty share prices— began to fall. In the second half of 2000, IPOs to fund the second round of start-ups were put on hold, new projects became ever scarcer, and high profile failures began to appear all over the world. Such events do not mean that the Internet has no future. They do mean, however, that e-businesses are undergoing an initial and necessary restructuring—as has happened in the past with all emerging industries. Because it is particularly important for emerging industries to achieve high market share, profitability sometimes becomes a secondary consideration. The case of Internet firms is no exception. In order to attract customers, many Internet companies paid scant attention to the costs incurred; some even went so far as to sell their products at a loss. The results, as we have seen, can be disastrous. Firms in a new industry need to set a level for costs they can sustain if they are to position themselves in the market—just as firms in established industries do. Unless the investment translates into revenue streams in the short term, they will go out of business. The Internet firms that avoided the worst errors—and those that already had a clear idea of how to successfully generate a profit from their customers—have gradually made progress and created a new generation of profitable online survivors. Examples include

3

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Yahoo!, Google, Expedia and Amazon. The current feeling in the U.S. and other markets in the English-speaking world is that the Internet models that survived the bursting of the technology bubble are both powerful and profitable. Today e-commerce is attracting more and more customers, online advertising has recovered, and the share prices of many Internet firms are rising much faster than market average. 3. Studies of profitability The study of factors associated with high business profitability is a field that has attracted considerable attention from researchers. During the late 1970s and early 1980s, various researchers (Steer and Cable, 1978; Armour and Teece, 1978; Teece, 1981) turned their attention to the links between the organizational structures adopted by firms and their profitability. They found that organizations with a multi-divisional structure were more profitable than holdings. However, more recent research (Ezzamel and Watson, 1993) into the relationship between profitability and organizational structure has shown that there is no particular design that is clearly associated with higher returns in today’s business context. Other researchers have opted to use regression techniques based on some profitability ratio as the dependent variable and one or more other factors as independent variables. Their aim has been to measure the relationship between profitability and the size of firms in order to establish the existence of economies of scale. Gort (1963) and Harris (1976) produced seminal work in this area. All of these studies concluded that firm size alone has limited power to explain levels of profitability because the correlation to the functions is relatively low. Furthermore, the direction of the regression coefficients is unstable. Subsequent studies (Gillingham, 1980; Chaganti and Chaganti, 1983; Woo, 1983) have taken a less ambitious tack by seeking to establish the variables that distinguish profitable firms from those which are not. The goal was achieved by making the profitability variable discrete in order to apply a multivariate classification model. Other studies (Toy et al. 1974; Kester, 1986; Barton and Gordon, 1988; Titman and Wessels, 1988; Rajan and Zingales, 1995; Michaelas et al. 1999) have examined the nature and determinants of corporate financial structure and found that gearing is negatively related to

4

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ the level of profitability. This result agrees with the pecking order theory (Myers, 1984; Myers and Majluf, 1984), which states that companies with high levels of profits tend to finance investments out of retained earnings rather than by raising debt finance. The search for the factors that affect the profitability of online firms began to attract the interest of researchers after the Internet bubble burst in 2000. To date, however, only a few papers dealing in depth with this issue have been published. Bughin (2003), for example, considers whether and how companies can be profitable on-line. The study identifies distinct patterns of competitive behavior within e-tailing, including maximum reach in the segment and high-value niche strategies. A high-value niche strategy occurs, for example, when an e-tailer keeps a laser focus on high margin products and/or on customer segments. It also occurs when an e-tailer attempts to convert traffic into customers at the lowest possible cost. Bughin’s paper proves that customer conversion and retention, but not traffic per se, are critical to achieving profitability in transactional ecommerce. Jain, Jayaraman and Kini (2004) focus on the factors that influence the probability and timing of post-issue profitability in Internet IPO firms. They find that the probability of increased profits and the decline in time-to-profitability are related to the age of the firm, human capital deployed, pre-market investor demand, and management changes (i.e., ownership, senior officers, directors) at the time of the IPO. The study also shows that the probability of decreased profitability and increased time-to-profitability are related to the proceeds of the IPO, the risk of the firm, venture capital participation, proportion of outsiders on the board, relevant industry experience held by the CEO, and the reputation of the CFO. Bughin and Hagel (2000) examine virtual communities with the aim of investigating the economic returns of marketing outlays aimed at communities. They show that such communities are no exception among B2C models in that they exhibit decreasing returns-to-scale in marketing. In their opinion, the strong membership franchise of the community should be used as a clear means to increasing marketing effectiveness, and hopefully one of the many paths used to strive quickly towards profitability. In this context, we believe it is important to identify the characteristics that best distinguish profitable firms in the Internet sector from those that are not. In order to

5

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ accomplish this, we have conducted an empirical study based on the financial statements of a sample of online firms for the period 1998-2002. 4. Method Our discussion of methodology focuses on three key issues: 1) the process by which the sample was obtained; 2) the variables applied; and 3) our classification technique (i.e., logit analysis). It also includes a specific description of how we applied logit analysis. 4.1. Description of the sample For the purposes of the empirical study, we selected a sample of Internet stocks listed on the NASDAQ technology market during the period 1998-2002. Due to the youth of the firms operating in this sector, the majority of which had not even been formed prior to 1998, we could not extend the period of time for our study. This is clearly an obstacle to the analysis of such firms because it is impossible to perform financial analyses on them. We selected the firms in the sample according to two criteria. First, they had to belong to the Internet sector. Second, they had to offer enough information that we could carry out our analysis. We discovered that classifying a firm as belonging to an Internet industry is not an easy task since the sector includes a complex mixture of businesses. Therefore, we used a number of different criteria, such as revenue sources (e.g., commissions, advertising income, and/or electronic sales) and the nature or scope of the e-business, for example, whether the relationship was between firms (B2B) or between the firm and the consumer (B2C). We selected our firms by using the Wall Street Research Net© WSRN.com in InternetStockList (http://www.internetstocklist.com), which groups online firms into different categories by main activity. Demers and Lev, 2001; Hand, 2001; Lazer et al. 2001; and Rajgopal et al. 2001 also used this classification in their studies. The inclusion of a firm in one of these categories does not mean that it obtains no revenues from other activities, but that its principal revenue source is that of the reference group. The InternetStockList places firms into thirteen categories: Table 1

6

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ For the purposes of analysis, we have assumed that all of these business models can be combined in four umbrella categories comprising network infrastructure providers (network infrastructure), information providers (portals and content), and firms selling products and services online (online shops and services). With respect to the latter category, we believe that it is necessary to analyse the two business models separately given the distinct nature of the goods traded. To meet the second criteria for our study, enough information had to be available for each firm that we could carry out our analysis. Our data came from the COMPUSTAT database, which compiles the balance sheets and income statements disclosed by firms annually. Because the sector has such a short history and has not yet stabilized, we further reduced the size of the sample by requiring that the financial information cover the entire five-year period of the study. The final sample comprises 156 firms operating the four business models mentioned above. (We identify the firms in Appendix 1.) 4.2 Variables We based our study on the information contained in the balance sheets and income statements published by the firms comprising the sample for each year of the period. We then constructed a battery of financial ratios using the factors conventionally considered relevant to assessing a company’s operational and financial situation: liquidity, profitability, indebtedness, and intensity of activity. Like any other technique, the analysis of ratios is subject to certain limitations that must be taken into account in assessing a firm’s true situation. This is even more valid in the case of Internet firms because of the frequent appearance of extreme observations and the difficulty of interpretation in cases where the denominator takes a negative value. Table 2 presents the 18 ratios on which we based our analysis. It also illustrates how we defined the ratios and the manner in which we constructed them. Table 2 One of the limitations in using ratios is evident in two listings in our table: V12 ROE (after tax) and V16 ROE (pre-tax). Because both terms are negative in some firms, particularly in the first year, the ratio is meaningless. Therefore, we have eliminated such observations from the analysis.

7

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ 4.3. Design of the study and methodology employed For the purposes of our study, we sought to identify the characteristics of Internet firms that have succeeded in achieving unusual levels of profitability for their respective sectors of activity. We then compared these companies to those that have experienced lower relative profitability. The initial studies performed for this purpose make profitability discrete in order to apply a multivariate classification model (Walter, 1959; Haslem and Longbrake, 1971; Gillingham, 1980; Chaganti and Chaganti, 1983; Woo, 1983). We applied the logit analysis technique within this study as follows: 1) Static analysis. The dependent variable considered is whether the firm belongs to the high or to the low profitability group in any given year. The formally independent variables consist of the ratios representing the company’s economic and financial situation in the same year. This analysis, which is performed for each year from 1998-2002, provides an approximation of the characteristics of profitable firms in each year in relation to the norm for the sector. 2) Dynamic analysis: The dependent variable is whether the firm is included in the high or low profitability group in a given year Xt. The formally independent variables consist of relative changes. That is to say, they illustrate the differences arising in each ratio in the period between Xt and Xt-i (where i = 1,2,3 or 4). This analysis throws light on the parameters undergoing the greatest change in firms belonging to the high profitability group as compared to the firms in the low profitability group. We applied the logit analysis technique to obtain the probability that an observation will fall within a given set depending on the behavior of the independent variables. The logit analysis poses no constraints on the normal distribution of the independent variables or on the equality of variance-covariance matrices, as would be the case if we employed other classification techniques such as discriminant analysis (Hair, 1999). Nevertheless, other problems arise from the use of financial ratios as independent variables, as described by researchers such as Joy and Tollefson (1975). Given the range and number of ratios that could be calculated, the model contains numerous possible variables. When applying the methodology, therefore, we performed a univariable ANOVA analysis to establish the power of each variable considered on an individual basis

8

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ and to distinguish between the different groupings of firms. In order to avoid the problem of multicolinearity, we used an iterative method to select variables stepwise rather than introducing all of the ratios at once. The following section discusses our results. 5. Characteristic parameters of profitable Internet firms In this section, we examine the characteristics common among firms that achieve higher profitability than is the norm in their business sector. We then compare these with the characteristics of less profitable firms. As explained above, we have applied the logit technique for this purpose by applying a static and a dynamic analysis. In order to classify each firm in the sample on the basis of the profitability obtained, we defined a dichotomous indicator that takes a value of 0 where profitability is low in relation to the norm and 1 where it is relatively high. We used a measure of operating profitability, rather than any indicator of financial profitability, in order to analyze the return on investment and eliminate the impact of the firm’s financial structure on its profitability. We then selected the high and low profitability groups by discarding the second and third quartiles for the dependent variable, defined as relative profitability. This procedure is similar to that employed by Walter (1959) and Woo (1983), who defined the high profitability group as that formed by the most profitable 25% of firms and the low profitability group as the least profitable 25%. Despite the drawback inherent in this method of discarding half of the sample, it does have the advantage of clearly identifying the characteristic patterns of high and low profitability that appear between the “best” and “worst” firms. The formally independent variables are the 18 ratios defining the key aspects of the firms’ operating and financial positions. We eliminated direct profitability components, such as margin and asset turnover ratios, because we assumed that their relationship with operating profitability is already proven and that their inclusion in the model is therefore unnecessary. Also, the inclusion of these ratios could distort the classificatory power of other indicators that have a less obvious relationship with profitability a priori.

9

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ 5.1 Static analysis The objective of the static analysis is to distinguish those characteristics that best differentiate Internet firms achieving profitability above the norm from those that are relatively unprofitable. We performed this analysis for each of the years in the period 19982002. To this end, we took the inclusion of each firm within the high or low profitability group (groups 1 and 0) in year X as the dependent variable, and the operating and financial ratios for the same year as the independent variables. On the basis of the prior univariable analysis, we obtained the variables exhibiting clear differences between high and low profitability firms over the period. The greatest differences observed between the two groups in 1998, 2001 and 2002 are mainly concentrated in the variables V7 (total assets/total liabilities), V8 (cash and ST investments/current liabilities) and V18 (current liabilities/total liabilities). These differences indicate that the most profitable firms are more liquid and more solvent. We did not observe significant differences between the two groups in 2000. However, it should be noted that this was the year the Internet bubble burst, which caused a general downturn in the sector that affected all firms to a greater or lesser degree. After this analysis, we then applied the binary logistic regression methodology. As recommended by Hosmer and Lemshow (1989), we included the variables in the model using the forward conditional stepwise method, which inputs variables into the regression model one-by-one. Table 3 shows the models obtained for each of the years and presents the variables and their respective coefficients. Table 3 Confirming the results of the univariable analysis, one of the most influential variables is V7 (total assets/total debt) because it indicates the firm’s solvency. This variable forms part of the model in the years 1998 and 2001. The direction of the coefficients indicates that the more solvent a firm in relation to the sector norm, the more likely it is that it will fall within the high profitability group. Similarly, the inverse of this ratio reflects the level of a firm’s total indebtedness. The results indicate that the firms included in the high profitability group are less indebted on average than those in the low profitability firms.

10

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Variable V6 (sales/working capital), which also appears in the model in 1998 and 1999, did not provide any significant differences between the groups in the prior ANOVA analysis. This variable belongs to the group of activity ratios like V1 (sales/employees) that form part of the model in 1999 together with V6. It describes the relationship between the level of a company’s operations and the assets needed to maintain the business. In this case, V6 reflects the level of working capital required to sustain a given level of sales. The direction of the coefficients for this variable in 1998 and 1999 reveals a higher level of working capital turnover in profitable firms than in the group comprising low profitability undertakings. Lower turnover ratios, indicating longer shelf life for inventory and slower collection of receivables, could be indicators of a weak demand for a firm’s products. They could also indicate sales to customers whose ability to pay is uncertain. This may, in turn, signal one or more of the following: 1) the firm’s income is overstated; 2) future production cutbacks may be required; or 3) potential liquidity problems may exist. In order to reach a judgement on the adequacy of working capital, however, we would need to take its components into account by considering the amount and evolution of available cash and bank balances, accounts receivable, inventories, and current liabilities. Individual analysis of the amount and evolution of these items could reveal the existence of problems with inventories, the collection of receivables, or payments to suppliers. It is also necessary to consider the disparate nature of the items comprising working capital and, therefore, the total figure. For example, transfers from one caption to another may not affect total working capital but could nevertheless condition the firm’s liquidity. In such cases, the amount of working capital in itself contributes little or nothing to our understanding of the business since it could give rise to erroneous interpretations. Similarly, any opinion that might be voiced with regard to the adequacy of working capital needs to take into account non-accounting information such as the general background, the sector analyzed, and the economic outlook. In 1998 V1 (sales/employee) differences appear between the high and low profitability groups, with the former obtaining better productivity from staff. If we continue to analyze this relationship, we find that the low profitability firms increased their productivity in 2001. This rise in V1 is the result of significant staff cuts carried out by 11

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ these firms during the period. For example, Amazon laid off 1,200 employees, Easylink laid off 814, and Internet Pictures Corp laid off 457. As shown in Table 3, none of the variables reaches the threshold for significance to be included in the model in 2000. This result is consistent with the prior mean differences analysis, which showed that no differences emerged between the groups because of the crisis reigning in the sector in that year. Finally, only variable V8 (current assets/current liabilities), which indicates liquidity, is included in the model in 2002. The ANOVA analysis reveals the presence of significant differences between the high and low profitability groups for this variable in 2002 and, indeed, in 1998 and 1999 as well. The differences indicate that the firms forming part of the high profitability group maintained a stronger current ratio than did the unprofitable firms. The Hosmer-Lemeshow test indicates the existence of statistically significant differences between observed and predicted classifications. Table 4, which uses non significant chi-squared values, shows a good fit in the model for each of the four years considered. Table 4 We have employed classification matrices as a final tool to measure the fit of the model. These matrices reflect the ratio of correct classifications at each step of the calculation process. Table 5 shows that the overall percentages for correct prediction in the model are high, with an average rate of 80% in classificatory accuracy. Table 5 An exception occurs in 2001, when the classification percentage drops to 63.6%. Also note that the systems we have constructed classify firms in the low profitability group better than they do profitable firms. This indicates that the businesses comprising the latter category share fewer common features in terms of the formally independent variables. To put this another way, the low profitability firms are more similar to each other, while the profitable firms exhibit more differentiated characteristics. 5.2 Dynamic analysis Our independent variables consist of the changes in ratios between each of the years analyzed (Xt) and prior years (Xt-i, where i= 1, 2, 3 or 4), whereas our dependent variable consists of the high or low profitability category in the year analyzed (Xt). The results of 12

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ the analysis of mean differences indicate that there are significant differences between the high and low profitability groups throughout the period for variables V7 (total assets/total liabilities) and V18 (current liabilities/total liabilities and equity). Differences also occur in the variables that indicate liquidity (V8 and V9), mainly in the periods referenced to 1999 (1999-2000, 1999-2001, 1999-2002). The models shown in Table 6 were obtained by applying the proposed logit analysis to each of the periods Xt -Xt-i. Table 6 The only variable forming part of the model is V7 (total assets/total debt) in all cases except the period 1998-1999. V7 indicates that the solvency of profitable firms increased in comparison to borrowings in the periods Xt-1-Xt. This is due to the fact that investors in these firms made significant capital contributions, mainly in cash. The result was falling indebtedness and rising equity. At the same time, tangible and intangible fixed assets increased, as well as financial investments in other firms in the sector. When considering intervals of more than one year (Xt-2 and Xt-3), all of the models point to V18 (current liabilities/total liabilities and equity) as the variable that best differentiates between the two groups of firms. The negative beta coefficient for variable V18 indicates that the most profitable firms reduced short-term debt over the period, while the liability of the least profitable firms rose period after period. Meanwhile, variable V17 (long term debt/total liabilities and equity), which forms part of the model in the period 1999-2002, confirms that the most profitable firms also lowered their level of long-term indebtedness. Determining the appropriate level of a company’s debt is a major responsibility of corporate financial officers; it is also one of the most difficult problems in finance. The name of the game here is risk, which basically consists of two components: business risk and financial risk. Business risk stems from the nature of the business in which a firm engages, while financial risk stems from the way in which a business finances its assets. In order to keep overall risk within acceptable levels, the basic rule is that a company operating in an environment with a high level of business risk (e.g., highly competitive; low entry barriers favoring the entry of competitors; or potentially high product substitution) should not add a high level of financial risk. Rather, it should maintain a low debt-equity

13

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ ratio. As a result, many companies, such as those in the automotive industry and the high technology sector, currently operate conservative financial policies. Another of the variables forming part of the models is V1 (sales/employees), which shows that the firms in the low profitability group in 2001 and 2002 obtained higher employee productivity than did those in the high profitability group. These figures confirm the results of the static analysis. Variable V1 rose for unprofitable firms because of numerous lay-offs in those years. In fact, 81.2% of businesses in the group cut their headcount, including 24/7 Real Media (719 employees), Amazon (1,200), Covad Communications (900), Easylink (814), Interliant (683), Niku (709) and Sportsline (317). In contrast, 66% of the firms in the high profitability group increased their headcounts. For example, Cisco Systems grew from 34,000 employees in 2000 to 38,000 in 2001; Microsoft grew from 39,100 to 47,600; eFunds grew from 1,927 to 2,560; PEC Solutions grew from 588 to 1,248; and Ebay grew from 1,927 to 2,560. In the last and longest of the periods analyzed (1998-2002), the model consists of three variables: V1 (sales/employees), V7 (assets-total (net)/liabilities-total) and V3 (liabilities-total/shareholders’ equity). Only variable V7 is significant, however, indicating once again that the most profitable firms succeeded in increasing solvency over the five years analyzed (1998-2002), while solvency in the least profitable firms deteriorated. Table 7 reflects the percentage of correct classifications in each step of the calculation process for all periods except 1998-2000 and 1998-2001, for which no valid model was obtained. Note that we calculated the classification table to measure the fit of the models. Table 7 In view of the percentages for correct classification obtained (on average between 71% and 84%), we may infer that the functions discussed above are reasonably valid. This reflects the power of the variables to classify firms in the high or low profitability group. Furthermore, we can observe that the models are better at classifying the firms in the high profitability group than in the low profitability group. This suggests that high profitability companies are more similar to each other in terms of changes in formally independent variables. Finally, overall classification efficiency rises in the last years of the study (2001 and 2002). For example, in the period 2000-2002, it reaches 91.3% compared to a score of 67.4% over the five years taken as a whole (1998-2002). 14

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ 6. Conclusions The bursting of the Internet bubble forced investors to shift their focus from the growth prospects of issuing firms to the question of if and when profitability can be achieved. As a result, Internet companies that failed to generate positive cash flows within the expected time frame found that they were unable to tap capital markets and were caught in a vicious cycle of events that frequently led to de-listing and even bankruptcy. The purpose of the present study is to identify the economic and financial characteristics that best distinguish the Internet firms that have successfully achieved positive levels of operating profitability from those that continue to incur significant losses. In order to accomplish this, we performed a static analysis that characterized the most profitable firms based on financial variables for the years in question. We then introduced dynamic variables to measure changes in the relative positions of the companies for each of the ratios. The results of the study indicate that firms with positive profitability during the period achieved greater solvency and higher liquidity than did unprofitable firms. The analysis also reveals the absence of significant differences in the financial characteristics of profitable and unprofitable firms in 2000. This is due to the technology crisis that affected all companies to a greater or lesser degree that year. The dynamic analysis confirms the result of the static analysis and reveals that the most profitable firms have progressively improved solvency levels. The development of public Internet companies throughout the period in question indicates that investors rewarded firms that showed profitability (or profit potential) by participating in equity offerings and boosting stock prices. As a direct result of such investor intervention, these companies improved their cash position, which led to greater solvency and higher liquidity. Analysis of the various sub-periods shows that the most profitable firms reduced indebtedness, mainly current liabilities, and increased self-financing. This can be demonstrated by analyzing the characteristics of the sector in which the firms operate (e.g., highly competitive environment, low entry barriers, potentially high product substitution) since the high level of risk inherent in the sector recommends a conservative financial policy that holds overall risk down to acceptable levels. Meanwhile, the fact that a large number of Internet firms have continued to fail to reach profitability has led both investors and venture capital firms to concentrate on other 15

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ industries and tighten the conditions under which they are willing to fund ventures. Since investors and venture capital firms represent the main sources of finance open to the sector during the period 1998-2000, firms have been forced to raise debt financing. At the same time, continuing operating losses and the rising cost of borrowing have caused deterioration in the solvency of such firms. We believe that our study can help both the academic community and business professionals better understand the turbulence in electronic markets over the past five years. In addition, we believe that our findings provide an opening for an interesting line of research aimed at clarifying the specifics of firms operating in the Internet sector. Our findings illustrate the key features defining the economic and financial characteristics of businesses that have succeeded in achieving profitability. Such information could prove invaluable to investors seeking to make better decisions on whether to invest in dot-coms or other ventures. Our research also has limitations, however. Because the Internet sector has emerged only recently, it was impossible to consider a longer period for analysis. This hampered our ability to apply techniques, such as panel data analysis, that might have provided further insight and information. The tremendous changes that took place in the Internet sector during the period—from initial boom to large-scale bust—further affected our ability to generalize results. As a result, we were confined to the time periods referred to in the financial statements of the sample firms. In this light, it will be of considerable interest if our findings are confirmed in the future as pertinent data becomes available.

16

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ References Armour H, Teece D. Organisational structure and economic performance: a test of the multidivisional hypothesis. Bell Journal of Economics 1978; 9:106-122. Barton S, Gordon P. Corporate strategy and capital structure. Strategic Management Journal 1988;9:623-632. Bughin J. Finding the Path(s) towards profitable E-Commerce. Working Paper 2003; 7: Deutsche Bank Research. Bughin J, Hagel J. The Operational Performance of Virtual Communities - Towards a Successful Business Model? Electronic Markets 2000; 10 (4 ): 237-243. Chaganti R, Chaganti R. A profile of profitable and not-so-profitable small businesses. Journal of Small Business Management 1983; 21 (3): 43-51. Demers E, Lev B. A rude awakening: Internet shakeout in 2000. Review of Accounting Studies 2001; 6 (3): 331-359. Ezzamel M, Watson R. Organisational form, ownership structure and corporate performance: a contextual empirical analysis of UK companies. British Journal of Management 1993; 4: 161-176. Gillingham D. A comparison between the attribute profiles of profitable and unprofitable companies in the United Kingdom and Canada. Management International Review 1980; 20 (4): 64-73. Gort M. Analysis of stability and change in market shares. Journal of Political Economy 1963; 71 (1): 51-63. Hair J, Anderson R, Tatham R. and Black, W. Multivariate Data Analysis. Prentice Hall International, 1999. Hand JR. The Role of Book Income, Web Traffic, and Supply and Demand in the Pricing of U.S. Internet Stocks. European Finance Review 2001; 5 (3): 295-317. Harris M. Entry and barriers to entry. Industrial Organization Review 1976; 3: 165-175. Haslem J, Longbrake W. A discriminant analysis of commercial bank profitability. Quartely Rewiew of Economics & Business 1971; 11 (3): 39-46. Hosmer D, Lemeshow S. Applied Logistic Regression. New York: Wiley & Sons, 1989.

17

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Jain B, Jayaraman N, Kini O. Dot.Com to Dot.Gone: Determinants of Probability of Profitability and Time to Profitability of Internet IPO Firms. Paper presented to the FMA European Conference, Zurich, Switzerland, (June 2004). Joy O, Tollefson J. On the financial applications of discriminant analysis. Journal of Financial and Quantitative Analysis 1975; 10: 723-739. Kester C. Capital and ownership structure: A comparison of United States and Japanese manufacturing corporations. Financial Management 1986; 15(Spring): 5-16. Lazer R, Lev B. and Livnat, J. Internet traffic measures and portfolio returns. Financial Analysts Journal 2001; (May-June): 30-40. Michaelas N, Chittenden F, Poutziouris P. Financial policy and capital structure choice in U.K. SMEs: Empirical evidence from company panel data. Small Business Economics 1999; 12: 113-130. Myers S. The capital structure puzzle. Journal of Finance 1984; 34 (3): 575-592. Myers S, Majluf N. Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics 1984; 13: 187-221. Rajan R, Zingales L. What do we know about capital structure? Some evidence from international data. Journal of Finance 1995; 50 (5): 1421-1460. Rajgopal S, Kotha S, Venkatachalam M. Why is Web Traffic Value-Relevant for Internet Firms? Working Paper, (2001), University of Washington. Steer P, Cable P. Internal organization and profit: an empirical analysis of large UK companies. Journal of Industrial Economics 1978; 27 (1): 13-30. Teece D.J. Internal organization and economic performance: an empirical analysis of the profitability of principal firms. Journal of Industrial Economics 1981; 30 (2): 173-199. Titman S, Wessels R. The determinants of capital structure choice. Journal of Finance 1988; 43 (1):1-19. Toy N, Stonehill A, Remmers L, Wright R, Beekhuisen T. A comparative international study of growth, profitability and risk as determinants of corporate debt ratios in the manufacturing sector. Journal of Financial and Quantitative Analysis 1974; 9 (November): 875-886. Walter J. A discriminant function for earnings-price ratios of large industrial corporations. Review of Economics & Statistics 1959; 41 (February): 44-52.

18

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Woo C. Y. Evaluation of the strategies and performance of low ROI market share leaders. Strategic Management Journal 1983; 4(2): 123-135.

19

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Appendix 1 List of firms forming the sample Company 1-800-Flowers.Com 24/7 Real Media Inc Agile Software Corp Allscripts Healthcare Soltns Amazon.Com Inc American Greetings Ariba Inc Art Technology Group Inc Asiainfo Hldgs Inc Ask Jeeves Inc Aware Inc Bea Systems Inc Blue Coat Systems Inc Braun Consulting Inc Broadvision Inc Checkfree Corp Chordiant Software Inc Cisco Systems Inc Click2learn Inc Cnet Networks Inc Commerce One Inc Concord Communications Inc Coolsavings Inc Corillian Corp Covad Communications Group Critical Path Inc Cybersource Corp Cysive Inc Dell Inc Deltathree Inc Diamondcluster Intl Inc Digex Inc Digimarc Corp Digital Impact Inc Digital Insight Corp Digital River Inc Digitalthink Inc Doubleclick Inc Drugstore.Com Inc Dsl.Net Inc E.Piphany Inc Easylink Services Cp Ebay Inc

Ticker Company FLWS Liberate Technologies TFSM Lightspan Inc AGIL Loudeye Corp MDRX Macromedia Inc AMZN Marimba Inc AM Marketwatch.Com Inc ARBA Martha Stewart Living Omnimd ARTG Matrixone Inc ASIA Mcdata Corp ASKJ Mercury Interactive Corp AWRE Micromuse Inc BEAS Microsoft Corp BCSI Modem Media Inc BRNC Multex.Com Inc BVSN Navisite Inc CKFR Neoforma Inc CHRD Neon Systems Inc CSCO Net Perceptions Inc CLKS Net2phone Inc CNET Netiq Corp CMRC Netopia Inc CCRD Netratings Inc 3CSAV Netscout Systems Inc CORI Netsolve Inc 3COVD Nexprise Inc CPTH Niku Corp CYBS Novell Inc CYSV Nuance Communications Inc DELL Onvia.Com Inc DDDC Onyx Software Corp DTPI Oracle Corp 3DIGX Packeteer Inc DMRC Palmone Inc DIGI Paradyne Networks Inc DGIN Pec Solutions Inc DRIV Persistence Software Inc DTHK Petsmart Inc DCLK Portal Software Inc DSCM Priceline.Com Inc DSLN Primus Knowledge Solutions EPNY Qualcomm Inc EASY Quest Software Inc EBAY Red Hat Inc

Ticker LBRT LSPN LOUD MACR MRBA MKTW MSO MONE MCDTA MERQ MUSE MSFT MMPT MLTX NAVI NEOF NEON NETP NTOP NTIQ NTPA NTRT NTCT NTSL NXPS NIKU NOVL NUAN ONVI ONXS ORCL PKTR PLMO PDYN PECS PRSW PETM PRSF PCLN PKSI QCOM QSFT RHAT

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Efunds Corp EFDS Redback Networks Inc RBAK

21

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Company

Ticker

Company

Ticker

Embarcadero Technologies Inc Emerge Interactive Inc Entrust Inc Expedia Inc Extended Systems Inc F5 Networks Inc Firepond Inc Forrester Research Inc Foundry Networks Inc Freemarkets Inc Gartner Inc Gemstar-Tv Guide Intl Inc Goamerica Inc Gric Communications Inc Hollywood Media Corp Ibasis Inc Infonet Services Corp Inforte Corp Infospace Inc Ingram Micro Inc Inktomi Corp Inrange Technologies Corp Interland Inc Interliant Inc Internap Network Svcs Corp Internet Cap Group Inc Internet Pictures Corp Internet Security Sys Inc Interwoven Inc Intraware Inc Intuit Inc Itxc Corp Ivillage Inc Keynote Systems Inc

EMBT EMRG ENTU EXPE XTND FFIV FIRE FORR FDRY FMKT IT.B GMST GOAM GRIC HOLL 3IBAS IN INFT INSP IM INKT INRG INLD INIT INAP ICGE IPIX ISSX IWOV ITRA INTU ITXC IVIL KEYN

Register.Com Inc Rsa Security Inc Sabre Hldgs Corp Sagent Technology Inc Scm Microsystems Inc Selectica Inc Siebel Systems Inc Sportsline.Com Inc Stamps.Com Inc Switchboard Inc Sybase Inc Tanning Technology Corp Terayon Commun Systems Inc Tibco Software Inc Tumbleweed Communications Turnstone Systems Inc Tut Systems Inc Ulticom Inc Universal Access Global Hldg Verisign Inc Verity Inc Versata Inc Verticalnet Inc Vialink Co Vicinity Corp Vignette Corp Visual Networks Inc Vitria Technology Webex Communications Inc Webmd Corp Webmethods Inc Websense Inc Worldgate Communications Inc Yahoo Inc

RCOM RSAS TSG 3SGNT SCMM SLTC SEBL SPLN STMP SWBD SY TANN TERN TIBX TMWD TSTN TUTS ULCM UAXS VRSN VRTY VATA VERT 3VLNK VCNT VIGN VNWK VITR WEBX HLTH WEBM WBSN WGAT YHOO

22

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Tables Table 1. Internet business models per Wall Street Research Network 1. Advertising 2. Consultants/Designers 3. Content/Community 4. E-Commerce enablers 5. E-tailing 6. Financial Services 7. Internet ServiceProvider(ISP) /Access

8. Internet Services 9. Performance software 10. Search /Portals 11. Security 12. Speed / bandwidth 13. Wireless

Table 2. Definition of the ratios applied in the study V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V11 V12 V13 V14 V15 V16 V17 V18

Definition Sales / employees Net income / employees Liabilities-Total / Shareholders' Equity Current Liabilities / Liabilities-Total (Non current liabilities + equity) / Net fixed assets Sales / Working capital Assets-Total (net) / Liabilities total Current Assets / Current Liabilities Cash & ST Investments / Current Liabilities EBIT / Total Assets Net income / Sales Net income / shareholders’ equity EBIT / Sales Sales / Total Assets Gross Profit / Revenues Pre-tax Income (EBT) / shareholders’ equity LT Debt-Total / Total liabilities & equity

Name Sales per employee Return per employee Liabilities-to-Equity Ratio Current Liabilities to Total Liabilities Long term finance to Net fixed assets Working capital turnover Solvency ratio Current ratio Cash ratio ROA Profit Margin ROE (after tax) Margin before interest and tax Total assets turnover Gross Margin ROE (pre-tax) Long term debt to total liabilities & equity Current term debt to total liabilities & Current liabilities-Total / Total liabilities & equity equity

23

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Table 3. Results of the logit model for the period 1998-2002 1998 V6 Step 2 V7 V1 1999 V6 Step 2

Sales / Working capital Total assets / Total liabilities Constant Sales / employees Sales / Working capital Constant

---------------------------------2001 V7 Total assets / Total liabilities Step 1 Constant 2002 V8 Current assets / Current liabilities Step 1 Constant 2000

Beta 0.11 0.67 -3.46 0.02 0.02 -2.50

E.T. 0.05 0.19 0.82 0.00 0.01 0.65

Wald Sig. Exp(B) 4.27 0.04 1.11 12.02 0.00 1.96 17.60 0.00 0.03 15.62 0.00 1.02 4.74 0.03 1.02 14.71 0.00 0.08

--

--

--

--

--

0.45 -1.43 1.43 -3.28

0.20 0.69 0.44 1.00

4.87 4.31 10.45 10.70

0.03 0.04 0.00 0.00

1.56 0.24 4.16 0.04

Table 4. Hosmer – Lemeshow tests for the period 1998-2002

1998 1999 2001 2002

Step 1 Step 2 Step 1 Step 2 Step 1 Step 1

Chi-square 13.75 12.98 12.98 11.11 9.19 7.98

df 8 8 8 8 8 8

Sig. 0.09 0.11 0.11 0.20 0.33 0.44

Table 5. Logistic regression classification table Predicted Observed

1998 Step 2

Profitability

Low High

Overall percentage 1999

Step 2

Profitability

Low High

Overall percentage 2001

Step 1

Profitability

Low High

Overall percentage 2002

Step 1

Profitability Overall percentage

Low High

Profitability Percentage correct Low High 30 9 76.9 11 28 71.8 74.4 33 6 84.6 8 31 79.5 82.1 15 7 68.2 9 13 59.1 63.6 20 3 87.0 5 16 76.2 81.8 24

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________

Table 6. Results of the logistic regression in the dynamic analysis Variable Current liabilities / Total liabilities V18 1998-1999 and equity Constant

Beta

S.E. Wald Sig. Exp(B)

2.37

0.94 6.31 0.01

10.68

0.41

0.27 2.31 0.13

1.50

1999-2000 V7

0.21 0.20

0.08 6.22 0.01 0.27 0.53 0.47

1.23 1.22

Xt-i - Xt

1998-2000

Total assets / Total liabilities Constant

----- ------------------------

2000-2001 V7

--------------------------

Total assets / Total liabilities Constant V1 Sales / employees Current liabilities / Total liabilities 1999-2001 V18 and equity Constant 1998-2001 ----- ------------------------

0.96 1.34 -.009

0.32 9.16 0.00 0.51 6.94 0.01 .003 7.46 .006

2.61 3.83 .991

-2.44

1.19 4.18 .041

.087

2001-2002 V7

1.53 0.51 .002

0.52 8.52 0.00 0.38 1.87 0.17 .003 .626 .429

4.63 1.67 1.00

-23.23

7.24 10.27 .001

.000

1.83

.658 7.73 .005

6.23

-7.14

3.49 4.17 .041

.001

-8.21

2.91 7.95 .005

.000

.935 -.004 .381

.469 3.97 .046 .002 3.70 .054 .155 6.08 .014

2.55 .996 1.46

.002

.002 1.01 .313

1.00

.693

.420 2.72 .099

1.99

Total assets / Total liabilities Constant V1 Sales / employees Current liabilities / Total liabilities 2000-2002 V18 and equity Constant LT Debt / Total liabilities and V17 equity 1999-2002 Current liabilities / Total liabilities V18 and equity Constant V1 Sales / employees V7 Assets-Total (net) / Liabilities total 1998-2002 V3 Liabilities-Total / Shareholders' Equity Constant

.916

.384 5.70 .017 2.50 --------------------------

25

Journal of Internet Business

Issue 3 - 2006

_________________________________________________________________________ Table 7. Logistic regression classification table for the dynamic analysis Xt-i -Xt

Observed Step 1 Profitability

1998-1999 Overall percentage Step 1 Profitability 1999-2000 Overall percentage Step 1 Profitability 2000-2001 Overall percentage Step 2 Profitability 1999-2001 Overall percentage Step 2 Profitability 2001-2002 Overall percentage Step 2 Profitability 2000-2002 Overall percentage Step 2 Profitability 1999-2002 Overall percentage Step 3 Profitability 1998-2002 Overall percentage

Low High Low High Low High Low High Low High Low High Low High Low High

Predicted Profitability Percentage correct Low High 24 15 61.5 7 32 82.1 71.8 20 12 62.5 6 26 81.3 71.9 16 5 76.2 2 20 90.9 83.7 30 9 76.9 13 26 66.7 71.8 16 6 72.7 4 19 82.6 77.8 20 3 87.0 1 22 95.7 91.3 18 5 78.3 2 20 90.9 84.4 14 9 60.9 6 17 73.9 67.4

26