Whether investing more than commitments indicates firms' better performance : based on airline industry in the United States

(1)

Master Thesis

Whether investing more than commitments indicates firms' better

performance: Based on airline industry in the United States

Supervisor: Dr. Tomislav Ladika

Mingting Lu

MSc Business Economics, Finance track

Amsterdam Business School

University of Amsterdam

(2)

Abstract

Firms which face business uncertainties can choose to either invest today or wait until the future when the uncertainties are presumably resolved. Therefore, for a firm in one fiscal year, there will be a difference between the actual investment and the committed investment which is decided before that fiscal year. What my thesis care about is comparing the actual investment and the committed investment of a firm during a year, and examining whether the difference between them is related to the firm's performance in that year. In other words, when a company decides to invest more than its committed investment which is signed before, can this decision indicate firm's better performance in that year? To be concrete, the performance of firm is measured by firm's annual sales. Focusing on representative public companies in the airline industry of the United States of America, this paper tries to point out that there exists a significant relationship between the investment difference (i.e. committed investment subtracts actual investment) of a firm and its performance when the actual investment is made. The concluded relationship means that if a firm actually invests more than what it planned before, that is because the firm's business are growing. Besides, comparing committed investment with actual investment is a potential indicator of firm growth opportunities, and investing more than planned investments can be treated as an indicator of firm's better performance and growing business.

(3)

-1. Introduction

To continue the operations and grow in the market, many firms have to make investments every year. But how much to be invested is not always a decision made year-by-year. Actually, sometimes firms will plan their future investments several years in advance. Since some of the investment decisions are capable of being delayed in reality, firms will not plan all its investments in advance. On the contrary, those firms will only sign part of its all investment opportunities as committed investments. Committed investments are the contractual obligations in the future. More specifically, firms often sign some purchase contracts in advance to guarantee their operations in the future. When those contracts take effect, they will make the firm invests the amount that are signed before. The amount of the purchase contracts is the firm's committed investment. For instance, in 2000, an airline may sign a contract which decide that it will purchase 10 Boeing-747 by 1800 million dollars in 2001 and purchase 8 Airbus 330 by 1600 million dollars in 2003. Then this airline's committed investment in 2001 is 1800 million dollars, and its committed investment in 2003 is 1600 million dollars.

When firm's actual investments exceed the amount it had initially planned, it may indicate that the firm's growth opportunities have been improved, and it performs better. Since facing uncertainties of business in investment strategies, most firms can choose to either seize the investment opportunities right now or wait until the future when the uncertainties could be presumably resolved. That is why there exists a difference between the actual investment and the committed investment of a firm during one fiscal year. Furthermore, firms will increase their investments only if they find profitable information which they can benefit from or uncertainties of their business are resolved. Intuitively,

(5)

investing more than commitments (i.e. the actual investment is more than committed investment) could be treated as an indicator of firm's better performance and growth opportunities. Based on this intuition, this thesis pays attention to comparing the actual investments and the committed investments of the selected representative firms, and examining whether the difference between the two kinds of investments indeed has a significant relationship with the firm's growth opportunities and performances, which are measured by the annual value of firm's sales.

This paper focuses on American public companies in airline industry for the sample, because committed investment is more meaningful in this industry and companies in airline industry provide the most relevant information in their 10-k statements. Moreover, another important reason to use airlines in my sample is that they make big and discrete purchases (i.e. aircrafts) frequently. In fact, firms have to report how many investments they have committed to do in next few years each year. Besides, the information of firm’s actual investments could be collected from its financial reports, so all of the information about firm's investments is available and it could be used to do research on firm’s investment strategies.

To achieve the aim of the research, this paper decides to use two different ways to measure how the committed investments are different from the actual investments. In the first method, I subtract the actual investment from the committed investment to calculate the mathematical difference between them. Then the relationship between the calculated difference (i.e. committed investment subtracts actual investment) and firm’s performance can be estimated. Besides, in the second method, this paper constructs a ratio which is calculated by the committed investment divided by the actual

(6)

investment to measure the difference between the actual investment and the committed investment, and examine whether this ratio could be used as a predictor of firm's performance. The reason to define this ratio is, just calculating the difference between the committed investment and the actual investment directly can't eliminate the effect of firm’s size on its performance. More specifically, to calculate differences or the ratios in the above two methods, this paper will find out the data on sample firms' committed investments, which are decided in advance for future operations, during the research time period and collect the respective actual investments. After the construction of all the variables, the main purpose of this thesis is figuring out the relationship between the difference of the two kinds of investment and the firm's performance. Considering the control variables, which will be explained detailed in Section 4, this paper tries to answer that relationship by regression analyses. If the relationship is negative, which means when the firm increases its actual investment relative to its committed investment in this year (i.e. the difference or the ratio decreases), then it tends to have a better performance in this year, thus the results could be concluded that investing more than the committed amount could be treated as an indicator of firm's better performance. On the other hand, if the relationship mentioned before is positive or ambiguous, then it means that we can't treat the difference between actual investment and committed investment as an good indicator of firm's growth opportunities. Maybe firms with declining business tend to invest more, or there is no relationship between firm's investment decisions and growth opportunities.

The hypothesis of this research is that increasing actual investment relative to committed investment (i.e. a smaller difference or a lower ratio) is an indicator of firm’s better performance and growth opportunities. There are two main reasons lead to this hypothesis. Firstly, if a firm's committed

(7)

investment becomes more smaller than its actual investment in the future, it means that this firm can delay some of its investment decisions and choose to invest in the future. In other words, this firm don't have to plan all its commitments in advance so that it has managerial flexibilities. Furthermore, the larger the difference between committed investment and actual investment is, the more the managerial flexibilities this firm has, because large difference means the firm can delay more of its commitments and make decisions later. Therefore, since this firm can benefit from its managerial flexibilities, as a result, it will have more sales and make a higher profit, which could improve its future performance. Secondly, firms will only increase their investment if they find profitable information from which they can benefit or some previous uncertainties of their business are resolved. Hence, investing more than commitments (i.e. the actual investment is more than committed investment) could be treated as an indicator of firm's better performance and growing business.

Much research has examined what factors will influence firm’s investment decisions, but room for improvement still exists, since little is known about how firm’s investment decisions are related to firm’s performance. Realizing the limitation of current studies, this paper tries to open up a new research field in firm investment strategies and provide with a feasible method to estimate firm’s performance. Furthermore, estimating the relationship between firm’s investment decisions and the firm’s performance, this paper will also figure out evidences that could be applied for making investment strategies. Lastly, the results of this paper could also be new applications of real options valuation theory, because some firms value flexibility (i.e. pointed out by the real options literature), but my thesis controls for determinants of real options.

(8)

This paper is organized as the following way: The background of this research and the related studies are briefly introduced in Section 2. The methodology and hypotheses of this research are explained in Section 3. The data and its descriptive statistics of this paper are introduced in Section 4, and the results of the regression analyses are shown in Section 5. Finally, Section 6 makes a conclusion of this paper.

2. Literature Review

Since the topic of this thesis is a new research field, there are very limited numbers of literature related to this thesis question. The related prior studies could be divided into two groups, one group of the papers focus on the research of real options analysis, which motivate this thesis and inspire the idea of regression model in this paper; the other group contains the papers that are mainly studies about the effect of investment plans and the uncertainties in business on firm’s investment strategies.

From the perspective of real options valuation, on a theoretical level, Abel (1983) and Abel, Dixit, Eberly, and Pindyck (1996) showed how opportunities for future expansion or contraction can be valued as options and how companies can benefit from those options. Trigeorgis (2002) made an overview of the real options by providing a comprehensive example, and it concluded a summary of empirical implications of real options to figure out what we have known about real options and investment under uncertainty. Besides, this paper also predicted that firms in the industry with lower uncertainty are more likely to plan investment in advance, firms have investment opportunities with

(9)

shorter horizons tend to plan investment in advance, and high interest rate will also lead to more investment plans of the firms. Since all those three factors affect firm's investment decisions, the research of Trigeorgis provided with three potential control variables for my research, and all these factors will be considered in the methodology part of this thesis. Further, this working paper also came up with what fields the future research can focus on about the real options, those opinions helped spark the inspiration of my paper. Dixit and Pindyck (1995) examined the shortcomings of the conventional approaches to decision making about investment and presented a better framework for thinking about capital investment decisions. This study argued that, if a company decides to go ahead with an investment, that means it gives up the option to wait for new information and invest in the future. The value of the option, which is an opportunity cost, should be considered into the cost of the investment. Thus, if using NPV method to judge an investment opportunity, the NPV rule should be modified: the present value of expected future cash flows must exceed the cost of the project by an amount equals to the opportunity cost of the real options. This research provided with a very comprehensive explanation about the idea of real options analysis. Trigeorgis (1993) came up with a more detailed overview of the existing real options literature and applications, as well as presented useful principles for quantifying the value of various real options. Above papers inspire me to do this research, because if a firm is able to decide invest today or later, that means investment opportunities are capable of being delayed and this firm has real options. This research can examine whether firms can benefit from their investment decisions (e.g. real options) in the airline industry.

Previous studies have also examined the implication of real options using mathematical models and ample data. Bulan (2005) investigated real options behavior in capital budgeting decisions using a

(10)

firm-level panel data set of U.S. companies in the manufacturing sector. Finally, the research found that increased industry uncertainty displays a pronounced negative effect on firm investment. Bulan, Mayer, and Somerville (2006) examined the extent to which uncertainty delays investment and the effect of competition on this relationship by a sample of 1214 condominium developments in Vancouver, Canada built from 1979-1998. The paper found that increases in both idiosyncratic and systematic risk lead developers to delay new real estate investment. The research on the implication of real options helps my paper provides with the hypothesis. Since firms can benefit from their real options and a larger difference between actual investment and committed investment represents a higher real option, it makes sense that the difference between actual investment and committed investment will have a positive effect on firm's future performance.

Furthermore, there are some related studies focused on the effects of investment plans and the uncertainty in investment. Since they pointed out that facing the uncertainty, firm’s actual investment could be different from their prior committed investment, the value of my thesis could be reflected. To be concrete, the result of Smyth and Driver (1968) was that, given the level of committed investment, the actual level of investment is determined by capacity constraints rather than by changes in output. Lamont (2000) said that investment plans, from the U.S. government survey of firms, are highly informative measures of expected investment and explain more than three quarters of the variation in real annual aggregate investment growth. Moreover, plans have substantial forecasting power for excess stock returns. Guiso and Parigi (1999) investigated the effects of uncertainty on the investment decisions of a sample of Italian manufacturing firms, and found uncertainty weakened the response of investment. Furthermore, this paper pointed out the effect of

(11)

uncertainty is stronger for firms that cannot easily reverse investment decisions and for those with substantial market power.

Since the research question of this paper is figuring out the effect of the difference of committed investment and actual investment on firm's future performance, the key step is finding a proper way to measure firm's performance. Among the different indicators to evaluate firm's performance, net income and sales are two basic and intuitive ones, and this paper also tried to use the annual change in net income and sales as index to judge firm's performance. Besides, as an indicator of how profitable a firm is relative to its total assets, return on assets (ROA) can help this thesis know the efficiency of firms and then evaluate their performance. Defined by Needles, owers, and Crosson (2008), ROA is calculated by dividing a firm's annual net income by its total assets. Furthermore, introduced by Tobin (1969), Tobin's Q is the ratio between the market value and the replacement value of the same physical assets, and this ratio can also be used as a potential method to estimate firm's performance. However, refer to Tobin's Q, there exist different calculations of firm's market value and replacement value. For example, Lewellen and Badrinath (1997) examined the methods commonly employed to estimate Tobin's Q ratios and found them to be flawed in design. They proposed an alternative procedure which is simpler and more accurate. Lang and Litzenberger (1989), on the other hand, use a different way to calculate firm's market value and replacement value. The specific method to measure firm's performance in this paper will be clarified deeper in Section 3 and Section 4.

(12)

3. Methodology and hypothesis

3.1 Hypothesis of the research

The hypothesis of this research is that firms which end up investing more than committed may be growing rapidly, and have better performances. In other words, the mathematical difference between committed investment decided before and actual investment (i.e. committed investment subtracts actual investment) of a firm have a negative relationship with its performance in that year. And, increasing actual investment is an indicator of firm's better performance. There are two intuitive reasons help arrive at this hypothesis. The first reason is that, the existence of a difference between firm's committed investment and actual investment means that this firm can delay some of its investment decisions and choose whether to invest in the future. In other words, this firm has managerial flexibilities between the year when the commitment is decided and the year when the commitment happens. Furthermore, more actual investment relative to committed investment means the firm can delay more of its investment opportunities and make more decisions in the future. Therefore, since firms can benefit from their managerial flexibilities, it is possible that they could have made a higher profit and improved the performance when their actual investments become larger than their committed investments. The second reason is that, according to the theory of real options valuation, if investment opportunities are capable of being delayed, we can think that the firm has options to decide invest now or later. Since firm can decide whether to exercise the options, and it will only exercise the options when it can benefit from the options. Thus, the options are valuable and they can increase the firm's value. In other words, they could help the firm performs better in the future.

(13)

Contradict to the hypothesis, there are also two other possible empirical results. The first one is that there is no relationship between firm's investment activities (the difference between committed investment and actual investment) and its performance. The second possible result is that, when a firm's actual expenditure increases relative to its committed investment, which is decided before, it may indicate the firm's worse performance.

3.2 Methodology of the research

3.2.1 Basic structure

Some firms, like financial companies and companies which provide services, don't have regular committed investments. Therefore, to analyze whether higher actual investment than committed investment indicates better firm's performance, this paper has to find out enough firms in some specific industries that tend to invest "big" machines or facilities regularly and report how much they are committed to invest in the future. In the United States, there indeed exist some industries meet the requirements, such as railroads industry, sea-based shipping industry, airline industry and so on. Because firms in these industries need to invest or purchase enough equipments or related facilities frequently, and many of the purchases are often signed advanced. For this research, the samples are the selected firms in American airline industry, because this industry is more mature and developed. Moreover, there are over sixty firms in this industry and they can constitute a competitive market. Furthermore, airlines usually make big and discrete purchases (i.e. airplanes). Last but not least,

(14)

airlines usually report their future commitments clearly in their annual 10-k filings, the needed data can be collected accurate and effective.

This paper collects 25 representative airline firms in the United States. These firms have operated for a relative long period and report their committed investments more clearly than other airlines. Furthermore, these 25 firms locate in different states, so they can represent different macro-economic development levels. For the selected firms, this research tracks their related financial data starts in 2000, so that I can find out the firms' performance and their investment activities in a long time period. Moreover, since some of these firms have not reported part of their financial data in 2013 now and they will report 2013 annual data on June 2014, the time period chosen for all the sample firms is from 2000 to 2012. Therefore, for each of the firms, this thesis has its data on 13 years, and then this thesis can use a regression with panel data to study the research topic. To test the above hypothesis, this thesis has a purpose of examining whether the difference between committed investment and actual investment of a firm is related to its performance. Therefore, there are two main variables should be considered, the variable which measures firm's performance and the variable which measures the difference between committed investment and actual investment. Since this paper assumes that the firm's performance would be related to the difference between committed expenditure and actual expenditure. The former variable is the dependent variable, and the latter variable is the independent variable as well as the explanatory variable.

Furthermore, though some airlines only report annually how much they are committed to invest in the future one year, most airlines report annually how much they are committed to purchase in the

(15)

future five years. For example, in 2005, American Airline Incorporation reported how much it is committed to invest in 2006, 2007, 2008, 2009, and 2010. For analyses of these firms' committed investments, except the effects of committed investments reported one year in advance, this paper also study the effects of commitments reported two years, three years, four years, and five years in advance.

3.2.2 Dependent variable

There are some potential variables and indicators to measure the performance of a firm. For instance, net incomes and sales can reflect the firm's ability to make a profit. ROA ratio (Return on Assets) measures the scale of the firm's earnings relative to its total assets. Tobin's Q is a commonly used ratio which shows how the market evaluate the firm. This thesis selects one of the most accurate indicators to be the dependent variable to measure firm's performance.

Among the several potential choices, since ROA ratio e uals to net income divided by total assets eedles, owers, and Crosson, 2008), it includes the effect of firm's annual net income and also considers the effect of firm's size, so ROA ratio is an indicator better than net income. Furthermore, Tobin's Q is the ratio between the market value and the replacement value of the same physical asset, and it can also be used as a potential method to estimate firm's performance. Similar to P/E ratio, Tobin's Q is a ratio that can measure how a firm is valued by the market relative to its book value. Besides, stock returns could also be used as a measure of firm's performance. Though the stock price and market value could be influenced by some factors such as political policies, inside trading, irrational behaviors and so on, those don't mean that stock returns and Tobin's Q are bad measures of

(16)

firm's performance. On the contrary, market-based measures of firm performance have the advantage that they incorporate not only what happens in one fiscal year, but also what investors expect to happen in the future. Also, ROA is closely related to sales, so there is nearly no difference between choosing sales or ROA as a measure. Due to the limitation of research time, among sales, stock returns, Tobin's Q and ROA ratio, this paper decides to use sales as the measure to evaluate firm's performance. I prefer to use sales because I believe, compared to Tobin's Q and ROA ratio, sales is a more directly and clearly way to measure the improvement of a firm's business, especially for airline companies. More sales means the firm has more customers and more market shares, which means it has growth opportunities. Thus, this paper chooses sales data as the variable to measure firm's performance.

3.2.3 Independent variable: Explanatory variable

Since the research question of this paper is whether the difference between committed investment and actual investment of a firm can indicate its growth opportunities (or performance), the explanatory variable in the regression analyses should be the difference between committed investment decided in advance and actual investment of a firm in the same year. To be concrete, there are multiple ways to measure the difference between committed investment and actual investment, and this paper has picked two of them which make the most sense:

(1) The first method is most straightforward. This research directly calculate the mathematical difference between firm's committed investment and actual investment to generate the needed explanatory variable, and the difference equals to committed investment subtracts actual investment. However, an obvious imperfection of this method is that there exists omitted variable bias. Firstly,

(17)

the size of the firm affect the difference between this firm's committed investment and actual investment, for large scale firms tend to invest more than small firms. Besides, in a mature market, especially in airline industry, large enterprises usually have more businesses than small companies, hence the firm's size also have an impact on firm's performance. Since firm's size affects its investment activities and performance simultaneously, this paper has to eliminate the effect of firm's size when making a regression analysis. Therefore, the regression model accounts for the effect of firm's size by control variables, which can represent the firm's size. More details of the selection of control variables are discussed later in section 3.2.4.

(2) The alternative method can solve the obvious flaw in the former method. Because firm's size will influence the scale of firm's investment, to account for the effect of selected firms' size, this paper defines a proxy variable to measure the difference of firm's committed investment and actual investment. The proxy variable equals to the committed investment divided by the actual investment. Then, the difference between committed investment and actual investment will be revealed as a percentage. So that the effect of firm's size on investment activities is excluded, and the relationship between the difference of the two kinds of investment and the firm's future performance will be figured out. However, this method still has its own flaw, because it can't distinguish two situations when committed investments equal to zero in both of the situations. Take for example, suppose that in both of year 2005 and 2007, United Airlines reported that it has no commitment in the next year. But, it invested 100 million dollars in 2006 and invested 200 million dollars in 2008. If dividing the committed investment by the actual investment to measure the difference between this firm's committed investment and actual investment, the explanatory variables in these two years (i.e. 2006 and 2008) both equal to zero. But, since United Airlines invested 100 million dollars more than its

(18)

commitment in 2006 and invested 200 million dollars more than its commitment in 2008, its performances and investment activities in 2006 and 2008 must be different, but this method can't distinguish these differences.

All in all, this paper uses the above two kinds of methods to measure the difference between committed investment and actual investment. It means this research does two regression analyses, and the regression model uses these two possible explanatory variables respectively. The purpose is that making the final regression results more convincing.

3.2.4 Independent variable: Control variables

As explained above, to eliminate the potential omitted variable bias in the regressions, this paper has to consider the control variables. More specially, variables which simultaneously affect the ratio between the committed investment and the actual investment (or mathematical difference between the committed investment and the actual investment) and the firm's performance (i.e. sales) should be considered.

Since this paper collected the panel data on the sample firms during 13 years, I will use an entity and time fixed effects regression model to examine how the difference between the actual investment and the committed investment of a firm in different years is related to firm's performance. According to Trigeorgis (2002), firm's managerial flexibility or real option value may be higher (other things being equal) for industries with higher uncertainty, for investment opportunities with longer horizons (i.e. those can be delayed longer), and when market interest rates are higher. This idea can be analogized

(19)

in my research. Because this paper only focuses on airline industry, all the sample firms have the same level of industry uncertainty. Besides, the effect of investment horizons is included in the entity fixed effect in the model for investment horizons usually don't change over time. What's more, the effect of interest rates is included in the time fixed effect in the model since interest rates don't change over firms, all the firms face the same interest rate in one fiscal year. Therefore, to conclude proper control variables, I have to find out the factors that not only are related to investment activities and firm's sales, but also change over time and different firms.

As discussed before, the control variables that related to the firm's size should be considered. The reason is that firm's size has an influence on firm's investment activities, and large scale companies tend to invest more than small companies, then firm's size will affect the difference between firm's committed investment and actual investment. Furthermore, since large firms usually have more sales and net incomes than small firms, firm's size also have an effect on firm's performance. In a word, factor which can measure firm's size accurately is chosen as control variable in the regression model, and the factor chosen in this paper is firm's total assets. Since different firms have different total assets and the total assets of one firm will also change over time. The total assets variable changes over time and firms, and it fits the requirements of the control variables.

Except for firm's total assets, this paper still should consider other possible control variables. Since the dependent variable of this paper is firm's performance, it's important to control for recent firm performance in the regressions, because this could lead to more investment and higher sales. There are two variables in this paper used to control for recent performance, cash flows in the previous year

(20)

and stock returns in the previous year. Firstly, there obviously exists a relationship between firm's investment activities and firm's cash flows, moreover, cash flows (e.g. in investment activities, in financing activities and in operating activities) will obviously have an effect on investing activities. Secondly, similar to cash flows, stock returns are also related to firm's performance and investments. Finally, cash flows and stock returns of a firm change over time and different firms have different cash flows as well as stock returns. Hence, cash flows and stock returns in the previous year are added to the control variables.

Last but not least, as the real options literature discusses the importance of uncertainty, this paper tries to proxy for this using market-book ratio, which can be used to capture whether a firm's business is growing rapidly or stable. Therefore, market-to-book ratio is also considered as a control variable in this research.

3.2.5 Regression analyses with panel data

Since two different methods to measure the difference between firm's committed investment and actual investment are used in this paper, this paper does regression analyses of these two explanatory variables respectively.

(1) When choose the difference between firm's committed investment and actual investment as the explanatory variable, the explanatory variable equals to committed investment subtracts actual investment. In the regression model, the performance of the firm will be measured by the level of firm's sales, which is the dependent variable. Besides, except the firm fixed effects and the time fixed

(21)

effects, the independent variables also include the four control variables: firm's total assets, firm's cash flows in the previous year, firm's stock returns in the previous year, and firm's market-to-book ratio, which all change over time and firms. The expression of the panel data regression model is showed as below:

Where is the level of firm's annual sales in year "t"; is the result for firm "i" in year "t", which equals to the committed investment minus the actual investment; is the firm fixed effect for firm "i", which can help me control the variables that change over entities, such as the horizons of firm's commitments; is the time fixed effect in year "t", which can help me control the variables that change over time, like market interest rates and annual shocks to the entire industry; and is the coefficient this paper interested in, for it represents the relationship between the difference of the two kinds of investment and firm's performance.

What's more, as I have said, though some airlines only report annually how much they are committed to invest in the future one year, most airlines also report annually how much they are committed to purchase in the future five years. Since is the result that equals to the committed investment subtracts the actual investment, then for firm "i" in year "t", there are five possible expressions for in the regression equation:

① ②

(22)

③ ④ ⑤

In this way, this paper regresses the level of firm's sales on the five possible respectively. Then I not only can figure out the relationship between the exceed investments relative to commitments reported last year and today's performance, but also can find out the relationship between exceed investments relative to committed investments reported 2 years, 3 years, 4 years and 5 years earlier and today's performance.

(2) Choosing the explanatory variable which equals to the committed investment divided by the actual investment, this paper still can achieve its purpose. And there will be only one change in the regression model. In the regression equation, different from method (1), is the result for firm "i" in year "t", which equals to the committed investment divided by the actual investment. Besides, for firm "i", in year "t", the five possible expressions for in this regression equation are:

① ② ③

④ ⑤

(23)

As explained before, in this case, this paper also regress the level of firm's sales on the five possible

respectively. And method (2) can also achieve the purpose of method (1).

This paper does regression analyses of these two explanatory variables respectively, moreover, for each of these two explanatory variables, this paper runs regressions for eleven times. To be concrete, in the first time, an OLS regression is run, and this paper regresses firm's sales on . In this way, the data on firm's sales and investments is treated as sectional data. In the second time, this thesis uses the collected panel data and regresses firm's sales on , in this time, firm fixed effects are added. In the third time, this paper continues to add time fixed effects in the regression. Next, control variables are considered in the regressions. In the fourth regression, this paper adds the firm's total assets to the third regression. In the fifth and sixth regression, the control variables which can control firm's performance (i.e. cash flows and stock returns) are considered one by one, and in the seventh regression, the factor of firm business' stability (i.e. market-to-book ratio) is included in the regression. Furthermore, in the eighth, ninth, tenth and eleventh regressions, the in the former regressions will be changed to , , , and respectively, so that this paper can move forward to study the effects of commitments which are signed two years, three years, four years or five years in advance. Finally, this paper divides all the data into two parts: time period from 2000 to 2008, and time period from 2008 to 2012. Then, focusing on the data before 2008 and comparing it with the data in the whole time period, I can figure out how financial crisis affect on the research question. The financial crisis in 2008 changed a lot of things in corporate finance and investment strategies, and it made firms had to face many challenges when borrowing money during that time, especially in 2008 and 2009. Hence, it is meaningful to study the crisis's effect. When

(24)

paying attention to the data before 2008, this paper repeats the method of regression used for the whole data between 2000 and 2012.

3.2.6 Hypothesis test

Finally, after running the regressions, the significance of the coefficient of the explanatory variable (i.e. ) should be tested, so that this paper can test the hypothesis and examine whether the difference between the actual investment and the committed investment can significantly indicate the firms' growth opportunities and performance.

No matter measured by the mathematical difference between firm's committed investment and actual investment (i.e. committed investment subtracts actual investment), or measured by the committed investment divided by the actual investment, the explanatory variable will decreases when actual investment increases compared to committed investment. Therefore, if is significant negative, it means when firms increase their actual capital expenditures relative to their original committed investments (i.e. invest more than commitments), the explanatory variables will decrease and these firms will have a better performance. On the other hand, if is significant positive, the relationship means that firms usually have a worse performance when they decide to increase their actual investments relative to their committed investments.

In conclusion, on one hand, if is significant negative, then the difference between committed investment and actual investment has a positive relationship with firm's performance. In other words, when firms increase their actual investments relative to their committed investments, it indicates that

(25)

those firms are experiencing growth opportunities and performing better (i.e. more sales). On the other hand, if is significant positive, then the conclusion would be: the difference between the committed investment and the actual investment has a negative relationship with firm's performance. Or to say, when firms are decreasing their actual investments relative to their committed investments, it seems that their performances (i.e. annual sales) are becoming better. Furthermore, if the results of the regression analyses can't reject the conclusion that significantly equals to zero, then this paper has to admit that there is no relationship between firms' performances and the differences between their committed investments and actual investments. Besides, the difference between committed investment and actual investment can't used to indicate whether firm's business is growing.

As discussed in section 3.1, the hypothesis of this research is that, the difference between committed investment and actual investment (i.e. committed investment subtracts actual investment) has a significant negative relationship with firm's performance. To be concrete, if a firm decides increase its actual expenditures relative to its original committed investment, then this can indicate that this firm tends to have a better performance (i.e. annual sales) and its business is growing. Therefore, the null hypothesis and alternative hypothesis of the regressions should be:

(26)

4. Data and descriptive statistics

4.1 The method of data collection

The data is collected based on the requirements explained in the research methodology. Since there are dependent variable, explanatory variables and control variables in the regression model, which have been explained in section 3, the main data of this paper is consists of the collected data on investment commitments of the twenty-five selected firms in airline industry, the data on actual investments of those firms, and the data on annual sales of those firms. Besides, the data also includes total assets, previous cash flows, previous stock returns and market-to-book ratios of the twenty-five firms. All the data is collected during the time period from 2000 to 2012. To be concrete, firms’ committed investments could be collected from their annual 10-k reports, and the data on firms' actual investments, annual sales, total assets, cash flows, stock returns and market-to-book ratios are collected from North American fundamentals annual version of Compustat.

The reasons to choose years from 2000 to 2012 as the time window are as follows: Obviously, the data before 2000 isn't representative because it is too far from now, and some of the firms have not reported part of their financial data in 2013 now (i.e. they will report 2013 annual data on June 2004). Thus, the time period from 2000 to 2012 is chosen. Furthermore, in this case, enough information can be collected for bull markets as well as bear markets, because the effect of the financial crisis is considered. Even though the financial crisis in 2008 changed a lot of things in corporate finance, especially in investment strategies. It would still be good to have a look at post-crisis market and examine how the market was influenced.

(27)

Due to the difficulty of collecting data from 10-k filings, this paper will only focus on representative companies in American airline industry, because firms in airline industry provide very detailed information about how much they have committed to spend on airplanes and other equipments over the next few years. This is meaningful for this research because I can calculate the ratios (i.e. the committed investments divided by the actual investments) and the differences between committed investments and actual investment not only for one year, but also for two, three, four and five years into the future. The sample of this thesis will be chosen based on all public companies in the United States. First of all, I keep the companies whose sic2 codes equal to 45, which means they are belong to airline industry, and drop the companies in other industries. Next, I pick up quite well-known U.S. airlines which provide most of Americans' air services. I then added a few more regional airlines that are less famous. Finally, there are twenty-five representative firms left, which conform to the selection criterion of this paper. The twenty-five airline companies are eligible and they compose the sample of this paper. Besides, the sample includes all airlines that a normal person would ever travel on, and a few specialized firms. However, because of bankruptcies, mergers and acquisitions, not all these airlines have full data (i.e. annual actual investments, annual committed investments, sales, total assets, cash flows, stock returns and market-to-book ratios) from 2000 to 2012. As long as didn't report the needed data in one or some years, the firm would be treated as don't have full information during 2000 to 2012. Table 1 lists the names of the selected firms. As shown this table, among the 25 firms, only 14 have full data during the time period from 2000 to 2012.

(28)

4.2 Data collection on dependent variable and summary statistics

As explained in section 3, the dependent variable chosen by this paper is firm's annual sales. The data collection of this part doesn't take a long time, because the needed data can be collected from the firms' financial reports from 2000 to 2012, and all the firms clearly report their annual sales in their income statements. Still, because some firms don't have full information during the chosen period, not all the firms have available data on sales for the whole time period. Table 3B shows the descriptive statistics of the 25 firms' annual sales from 2000 to 2012. As shown in this table, due to the difference between firms' size, the sales differ widely across the 25 sample firms. From the perspective of mean value, Fedex Corporation is the largest company whose average sales during the period is about 31215 million dollars, and Great Lakes Aviation, LTD. is the most small airline whose average sales is only about 106 million dollars. Therefore, it is obvious that these firms' sizes and levels of sales are really different, so the samples are representative. Besides, this table also shows the 25th percentile, median, and 75th percentile of different firms' annual sales. Focusing on the standard deviations of different firms' sales, which are reported in the last column, the Table 3B shows that most of the firms have unstable annual sales during the time period, because nearly all the standard deviations are considerable compared to the corresponding mean sales. The data on sales in this paper is in million U.S. dollars. Except the data on sales, another important job is collecting the data on independent variables.

(29)

4.3 Data collection on independent variables and summary statistics

First of all, for the independent variables, the main data this paper interested in is firms' actual investments (i.e. capital expenditures) and committed investments in time period from 2000 to 2012. As explained before, the data on the twenty-five firms' actual investments can be collected from their financial reports through Compustat. Table 2A summarizes the data on capital expenditures of the 25 selected firms in American airline industry from 2000 to 2012, all the figures are in million dollars. Due to the reason of foundation, bankruptcy, merger and acquisition, there are 11 firms don't have full information during that time period.

Meanwhile, a summary of descriptive statistics for the airlines' capital expenditures is done. Table 2B in the Appendix shows the descriptive statistics of the 25 firms' annual capital expenditures from 2000 to 2012. The figures in Table 2B are in million U.S. dollars. As shown in that table, due to the difference between firms' size, the level of annual capital expenditures differs widely across the 25 firms. The average capital expenditures, reported in the fourth column of Table 2B, varies between 1.8 million dollars and 2536 million dollars for the various firms. From the perspective of mean values of the capital expenditures from 2000 to 2012, Fedex Corporation is the largest company and Great Lakes Aviation, LTD. is the most small airline. Therefore, it is obvious that these firms' sizes and levels of capital expenditures are really different, and that is why the samples are representative. Besides, focusing on the standard deviations of different firms' expenditures, which are reported in the last column in Table 2B, it seems that most of the firms have unstable investment during the 13 years, because nearly all the standard deviations are considerable compared to the corresponding

(30)

capital expenditures. Hence, this paper have to find out the methods discussed before to eliminate the effect of firm's size on the results of this research.

One step further, the most important part of the data is the available information of the twenty-five firms' committed investments, which is found out from the airlines' 10-k filings reported annually. All the firms have to report how much investment they have committed to do in next few years every year, and that information is always reported in their annual 10-k filings. Nonetheless, the collection of this part of data takes a long time because different firms use different ways to report their committed investments, even though the differences are slight. Generally, some of the firms tend to report their commitments in the future one year, and some of the firms tend to report their commitments in one year, two years, three years, four years and five years, which means the latter kind of firms provide more available information. Moreover, there are some firms publish their commitments in a different way. They announce their committed investment in the future one year, but they combine their commitments in two years with commitments in three years, and combine their commitments in four years with commitments in five years. In this case, this paper assumes that the firms' commitments in two years are same to their reported commitments in three years, and their commitments in four years are same to the commitments in five years.

Referring to the 10-k filings of the selected twenty-five companies, this paper collects the dollar value of purchase obligations of each firm from 2000 to 2012, and all the commitments are recorded in million dollars. For airline industry, the most important investments are purchases of aircrafts. Because the lives of aircrafts are limited, all the airlines will report how many airplanes and engines

(31)

for the aircrafts they have to purchase in the future. Hence, the purchase obligations are usually purchase commitments of aircrafts and other purchase obligations which are related to equipments such as engines and ground facilities. Since most of the flights are served with meals and drinks, airlines also purchase goods for their on board services. Furthermore, the purchase obligations of some firms also include but not limited to the insurance, technology, marketing, maintenance, technology, outsourced human resource services, and other third party services as well as products. American firms usually publish the above committed investments in "Contractual Obligation" part of their 10-k reports. However, there are still some firms don't report their commitments clearly. For instance, they announce how many aircrafts they want to buy in the future and the time horizon for these commitments, but they don't make it clear that how many the commitments are in each year of the whole horizon. Since these cases only happen a little, this paper assume that the commitments are distributed evenly during the time period. Because only the obligations related to investments are considered in this paper, when calculating the commitments of purchases, I ignore firms' debt obligations and other unrelated obligations. Moreover, lease payments are not included in firms' investments. Thus, even though some airlines signed lease contracts to get new aircrafts, these contractual obligations will not treated as committed investment. Based on the collection criterion and assumptions, the committed investments of the firms are calculated and summarized from the 10-k reports.

After finished collection of the data on actual investments (i.e. capital expenditures) and committed investments of all the airlines in Table 1 from year 2000 to 2012, the first job this paper has to do is checking how large the committed investments are. In other words, I want to examine whether, at the

(32)

time the commitment was decided, did the committed obligations seem big or small. Then, the importance of committed investments to companies can be proved. This paper answers this question by dividing firm's committed investments decided in each year by firm's total assets one year before that year for all the data. For example, an airline reported in 2005 that its commitments in one year, in two years, in three years, in four years and in five years equal to 500 million dollars, 400 million dollars, 300 million dollars, 200 million dollars, and 100 million dollars respectively. And its total assets in 2004 is 5000 million dollars. To measure how large those commitments are, this paper will divide the five commitments decided in 2005 by the firm's total assets in 2004. Table 3A summarizes the results of this method, the first column represents the commitments in different horizons (i.e. commitment in one year, in two years, in three years, in four years and in five years) divided by past total assets, thus the figures in this table are calculated by commitments divided by past total assets for all the firms. The second column represents the number of observations, and it shows that not all the firms report full data on future five years and as the horizon of commitments increases, the number of available firms decreases. In the sixth column, the statistics show that the commitment in one year occupies the past total assets most, and the average percentage is 12.9%. Besides, the commitment in five years occupies the past total assets least, and the average percentage is 6.8%. Obviously, if a firm's committed investments are approximately equal to 10% of its past total assets, these investment can be a big deal. Therefore, the percentages in sixth column mean that the commitments of the firms from 2000 to 2012 are quite large. In a word, as shown in Table 3A, the committed investments decided different years in advance occupy around 10 percents of firm's total assets. And because airlines' total assets are usually very considerable, the decisions of committed

(33)

investments are indeed very important to firms. Hence, the research on committed investments in this paper is meaningful.

Using the collected data, I have to generate the explanatory variables used in this paper. As explained in section 3, the first kind of explanatory variable is the difference between committed investment and actual investment and the second kind of explanatory variable is the ratio between committed investment and actual investment. Therefore, using the collected data on firms' investments and the methodology explained before, this paper complete the generation of explanatory variables in the regression model. For each of these two explanatory variables, there are five measures of the variable. The difference between these five measures is how many years (i.e. one year, two years, three years, four years or five years) the committed investments are signed in advance.

Last but not least, to make the analysis convincing and eliminate omitted control bias, this paper also have to collect data on control variables. The control variables in the regression model are composed of four factors, which are firms' total assets, firms' previous cash flows, firms' previous stock returns, and firms' market-to-book ratios. All the above mentioned data can be collected from Compustat. To be concrete, total assets are published in income statements, and cash flows are published in cash flow statements (i.e. the sum of cash flows in investing activities, financing activities and operating activities). Furthermore, stock returns and market-to-book ratios need to be calculated based on the available data in Compustat. After picking up the annual year-end stock price PRCC_F, this paper measures the stock returns as the percentage change in PRCC_F from one year to the next. And, the market-to-book ratios are equal to dividing firm's market value by its book value. Same to the time

(34)

periods chosen for other variables, the time horizon for all the data on control variables is also from 2000 to 2012. Besides, except the four control variable came up with by this paper, there are still two factors which may affect firms' investment activities and firm' sales performance at the same time in the regression model (As explained in Trigeorgis (2002)). According to that research, the first factor is market interest rates for all years from 2000 to 2012, and the second one is the horizon of firm's investments. However, since having collected the data on twenty-five firms for thirteen continuous years, this paper sums up a panel data (i.e. unbalanced panel data) which is composed of twenty-five entities in thirteen continuous years, and uses this panel data to do regression analyses. Thus, due to the advantage of panel data analyses, market interest rates are considered in the time fixed effects and horizons of the investments are considered in the entity fixed effects. In this case, I don't need their specific data.

From Table 3C to Table 3F, those four tables summarize descriptive statistics for the four control variables in this paper. The figures in Table 3C and Table 3D are in million U.S. dollars, and those in Table 3E and 3F are percentile ratios. More specifically, first of all, Table 3C shows the descriptive statistics of the twenty-five firms' total assets from 2000 to 2012. As shown in that table, the level of firms' size differs widely across the 25 firms, this result is same to the results concluded from Table 2B and Table 3B. Then, Table 3D shows the descriptive statistics of the twenty-five firms' cash flows. This thesis combines the cash flows from investing activities, cash flows from financing activities, and cash flows from operating activities in Compustat and gets the data on firms' total cash flows. Besides, to measure firm's previous performance, the cash flows in each year of this research are cash flows happened in the previous year. As shown in this table, during the time period between

(35)

2000 and 2012, larger firms tend to have larger cash flows and smaller firms tend to have smaller cash flows. Moreover, the cash flows of all the firms are unstable, and some of them change a lot over the time. Except the cash flows, this paper also use the stock returns in the previous year to measure firm's recent performance. Table 3E displays the descriptive statistics of the twenty-five firms' previous stock returns from 2000 to 2012. Not all the airlines report full available data on their stock prices during the research time, when this paper can't get the available data on a firm's stock returns in some years, this table gives "NA" for that firm in those years. For the available firms, this paper picks up the annual year-end stock price PRCC_F variable in Compustat and measures the stock returns of firms as the percentage change in PRCC_F. It have to be emphasize that, to measure firm's previous performance, the statistics of stock returns in each year in this research are actual stock returns in the previous year. As shown in this table, the stock returns change a lot over time as well as firms. Finally, to capture whether a firm's business is growing rapidly or stable, this paper uses the firms' market-to-book ratios, and the descriptive statistics of them are shown in Table 3F. As shown in this table, when firm's market-to-book ratio is smaller than 1, it means this firm is bearish; and when firm's ratio is larger than 1, it means this firm is bullish, and this firm's business is growing rapidly.

5. Results of the regression analyses

5.1 Results of time period from 2000 to 2012

(36)

To begin with, the analyses subtract actual investments from committed investments to measure how large the difference between committed investments and actual investments are. Table 4A shows the results of the twenty-five firms from all years from 2000 to 2012, and the data set has been described in Section 4. Before the regressions, the dependent variable, the explanatory variable and control variables are winsorized by 1 percent. First of all, the first six regressions in Table 4A pay attention to the committed investments decided one year in advance, and some of them include additional potential determinants of firms' performances along with firm and time fixed effects. Moreover, the base specification, which is reported in column (6), includes all the control variables came up with in this paper. Columns (7) and (8) of Table 4A report the results when the commitments decided one year in advance in the base specification are changed into commitments decided two year or three years in advance. Furthermore, this research also analyzes the situations when commitments are decided four years or five years in advance, but the results are not included in this table.

As shown in Table 4A, with firm fixed effects and time fixed effects are added in the regression one by one, the coefficient on explanatory variable decreases, and the procedure is displayed in column (1) through (3). However, after firm and time effects were added in the second and third regression, the t-value -2.57 under the coefficient -1.1702 in column (3) means that the coefficient on the difference between committed investment decided one year in advance and actual investment is still significant at 5% significant level. Then, the regressions start to consider the effects of control variables. From column (4) to column (6), firm's total assets, firm's cash flows, firm's stock returns, and firm's market-to-book ratios in different years are added one by one as control variables. With control variables, the coefficient on the difference between committed investment decided one year

(37)

in advance and actual investment changes from -1.1702 in the third regression to -1.1828 in the sixth regression, and it is still significant at 5% significant level, because the t-value of the coefficient, as shown in column (6), is -2.90. Meanwhile, all the coefficients on total assets and stock returns from regression (4) to regression (6) are significant at 1% level. That means these control variables have significant effects on firm's performance and they are also related to firm's investment activities. To be concrete, according to the results of base specification, firm's total assets have significant positive relationship with firm's performance, and firm's previous stock returns have significant negative relationship with firm's performance. In regression (6), the coefficient on total assets is 0.5989 and the coefficient on stock returns is -355.4, the t-values under the coefficients prove their significant effects. Besides, firm's cash flows have a significant negative relationship (i.e. significant at 5% level) with firm's performance, the coefficient on this control variable is -0.6371 in column (6). Obviously, firms with more total assets usually have a bigger size, which can help them have more sales and increase their performances. Next, previous stock returns and cash flows measure the airline's performance before, and the results show that they have negative relationship with current sales. Lastly, market-to-book ratios, which are added to capture the stability of firm's growth, don't have close connection with firm's sales.

In conclusion, the Table 4A shows four interesting main results. Firstly of all, including the additional control variables increases the estimated effect of the coefficient on the difference between committed investment and actual investment (committed investment minus actual investment) from -1.1702 in column (3) to -1.1828 in column (6), which means if the difference is decreased by one million dollars, it is because that the airline's sales may increase about 1.1828

(38)

million dollars, and the relationship is significant. Since decrease of the difference means the firm decides more actual investments, hence investing more than committed can be closely related to firm's growing business. One method to evaluate the magnitude of this coefficient is to suppose that a airline increasing its actual investments by 10 million dollars relative to its commitments which are decided one year before, then its sales in this year may have increased about million dollars. This estimated effect is large, because according to Table 2B, the average value of most of the firms' capital expenditures is more than 30 million dollars, and the average value of more than half of those firms' capital expenditures is hundreds of million dollars. Which means most increase of actual investments may be due to firm's business is growing. Besides, the t-value of the coefficient on explanatory variable is -2.90, therefore the hypothesis that the difference between firm's committed investment and actual investment has no relationship with firm's performance can be rejected at the 1% significance level. Moreover, the coefficient is significant negative, so that it can be concluded that the mathematical difference between committed investment decided one year in advance and actual investment (i.e. committed investment subtracts actual investment) of a firm has a significant negative relationship with its performance in that year. Investing more than its commitments, which is decided one year before, is because of the firm's performance becomes better and the firm has growth opportunities. .

Secondly, this paper also extends the regressions to consider the effect of committed investments which are decided from two years to five year in advance. To achieve that goal, in regressions (7), (8), (9) and (10), the committed investments decided one year in advance are changed to committed investments decided two years, three years, four years and five years in advance respectively. The

(39)

coefficients on explanatory variables in regressions (7), (8), (9), and (10) are -0.6769, -1.2200, -0.5829 and -0.1023, besides, the t-values of these coefficients are -1.65, -2.93, -1.36 and -0.81. Because the results in regression (9) and (10) may be influenced by the limitation of the time period in this research, only the results of regressions (7) and (8) are shown in Table 4A. The results shows that all the differences between firm’s committed investments, which are made from one year to five years in advance, and its actual investments, have significant negative relationship with its sales performance when the commitments happen. And it means increasing firm’s actual investment relative to its commitments decided more than one year ago can also indicates that this firm has a better performance and more growth opportunities now. Besides, one step further, the results also show that the connection between firm's current performance and the difference between actual investments and committed investments decided more earlier is weaker. That makes sense, because firms will not decide most of its investments as committed several years in advance, then investing more than commitments decided several years ago maybe because the firm changes its strategies or some uncertainties are resolved during these several years, but not because it has grow opportunities now.

Thirdly, the control variables also have a explanatory power on firm’s performance. In regressions (6) through (10), in which committed investments are measured by commitments decided from one year to five years in advance, the coefficients on the three control variables (i.e. total assets, stock returns and cash flows) are significant at different significant level. And the effects of those control variables can't be neglected.

Whether investing more than commitments indicates firms' better performance : based on airline industry in the United States

Master Thesis