An empirical comparative analysis of stock-‐valuation assessments based
on operating performance between firms in core and peripheral Eurozone
countries throughout the 2008 financial crisis.
Abstract
Since the beginning of the financial crisis in 2008, Eurozone countries and firms have experienced considerable struggle and pressure to demonstrate financial solidity and consistent growth. However, these effects have not been the same for all, as greater risk aversion prompted considerably higher financing costs and weaker expectations for periphery Eurozone members. As Europe starts demonstrating slow signs of recovery, questions emerge on whether there is evidence of significant divergences in how firms’ operating performance are being assessed by the markets comparatively between firms in core and periphery Eurozone countries. This research paper investigates the
relationship between firms’ operating performance indicators and firm-‐specific returns, to provide an accurate and reliable scope on how both have related from 2008 until the conclusion of 2013, and on whether the market has been applying ‘double-‐standards’ in evaluating operating performance in favour of firms in one of the regions.
Bachelor Thesis Economics and Finance Luís Nunes e Costa Pontes Calhau Student number: 6144489 University of Amsterdam
Faculty of Economics and Business Supervisor: Marijn Kool
Table of Contents
1. Introduction ... 4
2. Literature Review ... 6
2.1 Pricing Determination ... 6
2.2 Operating Performance Indicators ... 8
3. Research method ... 10
3.1 Data Collection ... 11
3.1.1 Countries and Firms ... 11
3.1.2 Operating Performance Indicators ... 14
3.1.3 Time Period Investigated ... 15
3.1.4 Sources for Data Collection ... 15
3.2 Calculations ... 15
3.2.1 Firm-‐specific Stock Price Returns ... 15
3.2.2 The Value for Comparison ... 16
3.3 Comparative Analytical Tests ... 17
3.3.1 Tests for Significant Differences Between Populations ... 17
3.3.2 Correlations and Regressions ... 18
3.4 Multiple Linear Regression Model ... 18
4. Results Discussion ... 20
4.1 Tests for Significant Differences Between Populations ... 20
4.2 Correlations and Regressions ... 21
4.2.1 Change in Operating Revenue ... 22
4.2.2 Change in Operating Profit ... 23
4.2.3 Change in Operating Profit Margin ... 24
4.2.4 Change in Operating Profit/Total Assets ... 25
4.2.5 Change in Operating Profit/Total Debt ... 26
4.2.6 Change in Net Cash Flow from Operations ... 27
4.3 Multiple Linear Regression Model ... 28
5. Conclusion ... 31
6. Evaluation of Study ... 33
6.1 Literary Framework ... 33
6.2 Data Collection ... 34
7. Bibliography ... 35
Appendices ... 38
Appendix A – Informational Supplements ... 39
1. Introduction
From 2008 to 2014, European countries have faced increasing difficulties in setting their economies on track to achieve financial stability and sustainable growth. The 2008 financial crisis has proven to be a great challenge to several economic sectors, eventually becoming a prompter for important structural reforms attempting to address micro and macroeconomic issues that turned unsustainable for many European
members.
With a decline in domestic consumption and significant tightening in credit conditions, among other factors, European firms have and continue encountering difficulty in improving performance levels and growing, or even recovering. For a wide range of reasons, however, the effects perceived in Europe varied among members. As evidenced by the widened sovereign bond yield differentials1 (Barrios, Iversen,
Lewandowska, & Setzer, 2009), peripheral Eurozone countries (e.g. Greece, Portugal or Spain) have faced considerably greater fall in investor confidence, consequently
resulting in self-‐fulfilling negative-‐outlook expectations that incited further pressures on their respective economies and companies. Consistently, sharp falls in investor
confidence were observed in the Eurozone throughout this period, as shown by data on the EMU (European Monetary Union) “Sentix” indicator2.
As demonstrated by the ECB’s (European Central Bank) global risk aversion indicator (Appendix A – Figure 4), the current period of significant financial distress has prompted investors to keep away from riskier investments. Consequently, investors adopted strategies to better safeguard resources, intended to mitigate risks of asset price falls and financial losses. Money flows to core Eurozone countries3 (e.g. France,
Germany or The Netherlands) instilled therefore no surprise, as investors attempted to prevent losses in the riskier weaker European economies (De Santis, 2012). As such, while Eurozone firms’ stock price movements continued to account for the whole Eurozone region’s systematic risks, disparities between them would not only denote
1
Please see figures 1 and 2 in Appendix A 2
See figure 3 in Appendix A
3
Figure 5 in Appendix A demonstrates comparatively higher Target2 balances in core Eurozone countries throughout the crisis, against periphery countries. Target2 balances indicate capital flows (Westermann, 2014)
risk-‐perceptions regarding each European member’s situation, but also firm-‐specific aspects.
In many cases, firm-‐specific operating performance indicators demonstrated fragilities, generally affected by macroeconomic dynamics and other systematic risk elements throughout the financial crisis arguably commenced in 20084. Yet, it is unclear
how decisive firm-‐specific operating performance has been in investing assessments comparatively between core and periphery Eurozone nations. This investigation aims to understand this by examining the following research question: how have firm operating performance indicators determined firm-‐specific stock price returns comparatively between core and peripheral Eurozone countries throughout the 2008 financial crisis? What differences can be observed and what do they mean? Most importantly, are they significant?
Ultimately, if observed a significant divergence in favour of firms in one of the regions, inferences could be made of a biased assessment by the markets on firm performance in favour of one of them, therefore denoting a possible misestimate of actual and potential recovery rates at an idiosyncratic level. Firms’ financing conditions and investor confidence tend to have strong ties with stock and risk assessments. Thus, perceptions of misevaluations could influence firm value, risk and perceived-‐risk, and credit ratings, while affecting recovery levels in the Eurozone at both a micro and macroeconomic level.
The following section in this study offers a thorough reflection on past papers, particularly focused in this analysis’ relevant factors -‐ namely stock pricing
determination and company operations. It provides a deeper understanding on the concerning subject, and presents grounding work supporting the construction of an accurate and reliable method and analysis. The methodology will subsequently provide a clear investigation design with careful reasoning and considerations on data
collection, analysis and presentation of data.
4
In the 2009 review “Economic Crisis in Europe: Causes, Consequences and Responses”, presented by the European Commission Directorate-‐General for Economic and Financial Affairs, authors establish summer of 2007 as the beginning of the financial crisis that would follow, making reference to the first spike in the 3-‐month interbank spreads against T-‐bills or overnight indexed swaps (OISs) (Appendix A – Figure 6), as “BNP Paribas froze redemptions for three investment funds, citing its inability to value structured products” (Buti & Székely, 2009).
2. Literature Review
Stock pricing analysis is generally considered particularly complex, where statistical tests are conducted and interpreted to attempt understand and verify determinants, mispricing indications and potential errors. The possible factors to consider are numerous, and while some may prove irrelevant for explanatory argumentation in a specific time-‐period, they may prove significant in another.
Pricing evaluation can be based on two different methods. Analysis of ‘fundamentals’ accounts for systematic (macroeconomic), systemic and industry-‐ related, and firm-‐specific variables, to measure a stock’s intrinsic value. It comprises an assessment of firms’ financial positions and prospectus, by inspecting financial reports, estimating future growth and considering movements in macro and microeconomic indicators. Alternatively, a ‘technical’ analysis employs quantitative techniques, such as supporting and resistance price levels, golden ratios, and Fibonacci sequencing, to learn historical stock pricing movements and trends to help better predict future prices.
This paper attempts to evaluate the fundamental contribution of operating performance to firms’ returns in the Eurozone since 2008. While the study focuses on a recent time period, there is extensive literature on which to build a reliable
methodological process for this investigation, which also assists in interpreting tests’ results obtained and drawing conclusive argumentation.
2.1 Pricing Determination
Stock prices reflect market players’ expectations on firms’ future returns and associated risks. Work by Harry Markowitz and William Sharpe on efficient portfolios lead to the conception of the “single index model” in 1963. Resultant from an analysis on between-‐asset relationships, the model, also referred to as the “one-‐factor model”, determined returns by:
𝒓𝒊 = 𝜶𝒊+ 𝜷𝒊𝒓𝒎+ 𝜺𝒊
where stock “i” returns, denoted by 𝒓𝒊, are defined by return due to firm-‐specific factors
(𝜶𝒊), sensitivity to market index returns (𝜷𝒊) multiplied by market index return denoted by 𝒓𝒎, and residual with mean zero and finite variance (𝜺𝒊). The formula was based on
the assumption of a common factor for all securities considered, a benchmark, from which each security reacted.
With an additional assumption on the shared opportunity for all investors to borrow or lend at the same interest-‐rate level (i.e. risk-‐free rate), Sharpe (1964) and Lintner (1965) would build on the return function and define the equation by:
𝒓𝒊 = 𝜶𝒊+ 𝒓𝒇+ 𝜷𝒊 𝒓𝒎− 𝒓𝒇 + 𝜺𝒊
where stock “i” returns, denoted by 𝒓𝒊, are defined by return due to firm-‐specific factors (𝜶𝒊), risk-‐free rate (𝒓𝒊), sensitivity to market returns (𝜷𝒊) multiplied by market risk
premium, with market return denoted by 𝒓𝒎, and residual returns, or unsystematic risk
(𝜺𝒊).
Later on, Fama and French (1993) would propose the inclusion of two new variables in the returns function: small-‐minus-‐big (SMB) and high-‐minus-‐low (HML). By carrying out regression tests, they found evidence that firm size by market cap, with small market capitalization firms providing higher returns due to higher risks, and book-‐to-‐market ratios, with higher returns associated to the higher risk from companies with higher book-‐to-‐market ratios (value stocks), better explained returns,
consequently leading to the formulation of the “three-‐factor model”. Given the inconsistencies in the single-‐index model’s returns’ predictive capacity perceived by Fama and French, both factors SMB and HML would provide significant improvements in characterizing systematic risks associated with securities and portfolios.
Carhart (1997) would make an additional contribution to the model, with one last significant variable: momentum (MOM). Upon the observation of premiums that would most likely associate with firms whose stock price upward movements persisted, as opposed to those experiencing declines, Carhart (1997) proposed the four-‐factor model that included market betas and Fama and French’s factors:
𝒓𝒊= 𝜶𝒊+ 𝒓𝒇+ 𝜷𝟏𝒊 𝒓𝒎− 𝒓𝒇 + 𝜷𝟐𝒊𝑺𝑴𝑩 + 𝜷𝟑𝒊𝑯𝑴𝑳 + 𝜷𝟒𝒊𝑴𝑶𝑴 + 𝜺𝒊
The momentum factor is based on the behavioural idea that market players will invest and maintain position in stocks that have historically shown continuous value appreciation. Contrastingly, market players will be progressively discouraged to hold stocks with constant historical underperformance and declines, thus further inciting downward pressures on asset prices. As a result, the momentum factor is calculated as the average between the best performing stocks minus the average between the worst performing (winners-‐minus-‐losers).
2.2 Operating Performance Indicators
Similarly to Fama and French (1993, 1996), past works have utilized regression techniques to not only test already conceptualized models and theories regarding security returns determination, but also investigate interactions with other variables. Using the same methods, Mehrani and Mehrani (2003) and Saghafi and Salimi (2005) examined the Tehran Stock Exchange for relationships between returns and
fundamental accounting variables. Both works provided evidence of variable
significance on firms’ stock returns, namely changes in operating profit, profit margins and pre-‐tax profit (Mehrani & Mehrani, 2003), changes in profitability and total assets (Saghafi & Salimi, 2005).
Further studies conducted on the Tehran Stock Exchange show indications of the importance operational ratios have on returns. Ghasempour, A., Ghasempour, M., and Bahonar (2013) built on the aforementioned research (Mehrani & Mehrani, 2003) by investigating a longer time period, as well as additional operating and profitability ratios, to conclude a significance in firms’ returns. Ratios included: changes in return on assets, changes in debt to asset ratios, and changes in cash flow ratios.
This paper focuses on extending the researched operating performance independent variables, to evaluate their relevance on European firms’ specific returns throughout the Eurozone crisis since 2008. While the past works referenced study a stock exchange of different dimensions and compositions (Tehran Stock Exchange), authors make a selection of indicators to represent company operating performance, in line with general accounting techniques. Given an assessment of operations is also attempted throughout this investigation, similar indicators are utilized, thus accounting
for performance in sales, operating profits and margins, and adding indicators that adjust them for company size and leverage.
In addition, the study also assesses firm-‐specific returns against the capacity of firms to generate cash. The ability to pay-‐out dividends to investors and its
sustainability, for instance, is dependable on cash generation, as they are generally paid in cash. Additionally, companies with declining levels of cash generation from
operations may indicate inefficiencies in cash collection and a weakened financial strength, which may discourage investors from buying/holding firms’ stocks, and lead to stock price falls. Cash flows gained considerable attention throughout the 2008 financial crisis, as they provided investors an extra view on how resilient each firm’s finances were, and how well they would be able to maintain operations and meet potential obligations, given the credit tightening observed.
3. Research method
The aim of this research is to primarily determine if significant variations exist between core and peripheral Eurozone countries, concerning how firm operating performance indicators weigh in on firm-‐specific stock price returns. To address this, the study envelops a statistical comparative analysis between both Eurozone regions, allowing verifying for a significant divergence based on values calculated using the formula:
𝝎𝒊=𝜶𝒊+ 𝜺𝒊 𝚫𝝁𝒊
where value 𝝎𝒊 is defined as return due to firm-‐specific factors in a pre-‐determined time
period, as denoted by 𝜶𝒊, plus residual returns, or unsystematic risk (𝜺𝒊), per change in
each operating performance indicator 𝝁𝒊 for the same pre-‐determined time period,
denoted by 𝚫𝝁𝒊.
The investigation proceeds with an attempt to show how significant each
operating performance indicator was in determining returns due to firm-‐specific factors throughout the time period studied, with a series of individual comparative assessments between both regions. Finally, a regression analysis is carried out based on a model enveloping all significant indicators considered, providing a generalized view on how all operating factors help explain firm-‐specific returns achieved for each region. The model is determined by:
𝜶𝒊+ 𝜺𝒊 = 𝜷𝒊𝟎+ 𝜷𝒊𝟏 𝚫𝝁𝒊𝟏+ 𝜷𝒊𝟐 𝚫𝝁𝒊𝟐+ ⋯ + 𝜷𝒊𝒑 𝚫𝝁𝒊𝒑+ 𝜺
The complete empirical design, comprising details and reasoning on data collection, computational steps taken, including the above, and analytical techniques used, is presented below.
3.1 Data Collection
3.1.1 Countries and Firms
The countries analysed throughout this investigation are Eurozone members. With a common currency and centralized monetary policy-‐making, besides the shared judicial and executive governmental body (i.e. European Union), there is greater confluence of the systematic factors constituted in the stock pricing of the Eurozone firms considered. Adding to this risk equivalence is the Eurogroup, formed by member countries’ finance ministers to “ensure a close coordination of economic policies” (European Union, 2014). These shared factors are key in allowing for a reliable comparative study of the companies selected at a firm-‐specific level, as several
macroeconomic and regulatory variables (e.g. central bank policy rates, exchange rates) are the same for Eurozone members.
The comparison intended for study is between core and peripheral Eurozone member countries. For this investigation specifically, the distinction between the terms core and peripheral attempts to account for economic strength since 2008, based on country growth, debt levels and inferred risk. Core Eurozone members refer to euro currency countries that have presented comparatively higher Gross Domestic Product (GDP) momentum, higher Target2 balances and lower risk implied by bond credit spreads since 2008. Conversely, periphery Eurozone members refer to euro currency countries that showed weaker economic performance since 2008, with comparatively low GDP momentum, lower Target2 balances and higher risk implied by bond credit spreads.
To ensure further variable equalization concerning potential unwanted macroeconomic effects, an additional sovereign categorization was added to country selection. Disparities in investment assessments and decisions arise when considering developed, emerging and undeveloped marketplaces. Market-‐making technology and efficiency, number and size of market players, regulatory framework, and market dynamics are some of the factors that can have a significant effect, and are therefore important to address when evaluating pricing assessments. To address this, all countries selected pertain to the FTSE Europe Developed country classification, as determined by FTSE’s Quality of Markets Assessment Matrix Criteria (FTSE, 2014).
In line with these selection parameters, the core Eurozone members examined are: France, Germany and The Netherlands. These Eurozone members are classified by FTSE as Europe Developed countries, and have shown the lowest credit spreads5,
comparatively better Target2 balances and high GDP momentum, as demonstrated by figures 5 and 7 in Appendix A; results that indicate greater resilience and better performance throughout the financial crisis.
The periphery Eurozone members examined are: Italy, Portugal and Spain. Similarly, they are Eurozone members also classified by FTSE as Europe Developed countries. As figures 1, 5 and 7 from Appendix A show, they have experienced greater credit spreads and therefore a comparatively high decline in investor confidence with very low levels of Target2 balances, leading to strong falls in GDP momentum. Also, these countries have been subject of structural reform programs and/or financial assistance defined by the European Commission (EC), the ECB and the International Monetary Fund (IMF).
Greece and Ireland also fill the criteria used to select the periphery-‐representing countries. Nonetheless, there are particular aspects that show accounting for both could be inadequate for this study. Figures show the two countries experienced the highest credit spreads throughout the crisis6, yet not the lowest Target2 balances7. Additionally,
as figure 78 shows, the crisis saw both countries experience very unusual GDP
momentum relatively to the rest of the Eurozone. These observations are indicative of the difficulty including both members could be to interpret computations and results based on stock price movements and returns. As will also be explained further in this section, the countries chosen for this study allow for a maximum number of indexed firms to be selected while maintaining the same industry representation across all countries. This reduces industry momentum effects on the data, thus permitting greater reliability of results. Selecting Greece and Ireland would not allow for this without reducing the number of firms to be examined.
The firms selected for this investigation are country index components. Exclusively selecting benchmark-‐indexed firms for each country reduces potential
5
Please refer to figure 2 in Appendix A 6
See figure 1 in Appendix A
7
See figure 5 in Appendix A 8
Presented in Appendix A
impact in results originating from different levels of importance each firm has in their respective markets, and ensures accessibility to extensive data collection on stock prices and financial statements. Moreover, due to significant variations in how operating performance is assessed, as well as in their regulatory frameworks, banks and other financial institutions are not considered in this research.
As aforementioned (p.12), to allow for a more accurate and reliable comparative analysis each firm is representative of a predetermined industry, consequently
addressing potential systemic and industry-‐specific effects. The only industries that are represented by at least one firm in each of the countries selected are: Materials,
Consumer Staples, Industrials, Energy and Communications. In agreement with all the parameters explained, the firms researched are as follows:
Eurozone Core
-‐ France (CAC40 Index): Vallourec SA (Materials), Carrefour SA (Consumer Staples), Vinci SA (Industrials), Total SA (Energy), Orange (Communications)
-‐ Germany (DAX Index): Lanxess AG (Materials), Henkel VZ (Consumer Staples), Siemens Aktiengesellschaft (Industrials), E.ON SE (Energy), Deutsche Telekom AG (Communications)
-‐ The Netherlands (AEX Index): Akzo Nobel (Materials), Koninklijke Ahold NV (Consumer Staples), Koninklijke Boskalis NV (Industrials), Royal Dutch Shell Plc (Energy), Koninklijke KPN NV (Communications)
Eurozone Periphery
-‐ Italy (FTSE MIB Index): Tenaris SA (Materials), Tod’s SpA (Consumer
Discretionary9), Atlantia SpA (Industrials), Eni SpA (Energy), Telecom Italia SpA
(Communications)
9
Italy’s index FTSE MIB does not have a component representative of Consumer Staples. For this case it is replaced by consumer-‐related product industry Consumer Discretionary
-‐ Portugal (PSI20 Index): Semapa R (Materials), Jerónimo Martins (Consumer Staples), Mota-‐Engil (Industrials), Galp Energia (Energy), PT Telecom SGPS N (Communications)
-‐ Spain (IBEX35 Index): Viscofan SA (Materials), Ebro Foods SA (Consumer Staples), ACS (Industrials), Repsol SA (Energy), Telefónica SA (Communications)
3.1.2 Operating Performance Indicators
The operating performance variables selected for analysis attempt to address two key aspects concerning the commanding research question of this investigation: include a range of indicators that provide a comprehensive and meaningful scope on operating profitability and efficiency, and that allows for a reliable, accurate and sound comparable assessment between the firms in question, given time constraints and information accessibility limitations specific to this investigation. Moreover, important considerations are made to account for variances in exogenous factors regarding macroeconomic dynamics and policies, as they are beyond each firm’s control.
To minimize potential impact of governmental action and other external aspects on the comparison between firms in terms of operating performance, thus permitting an analysis representative of firms’ actual operational capabilities and success, the
profitability indicators selected exclude interest-‐payments on loans and tax payments. Also, additional profitability indicators adjusting for company size (total assets) and debt load (total debt) are included in the study for a fairer and more accurate comparison. Given this investigation involves a study on relationships between operating performance and firm-‐specific returns (i.e. changes in stock prices due to firm-‐specific aspects), the indicators selected are at the same time percentage changes for a pre-‐determined time period. Accordingly, they are the following:
-‐ Change in Total Operating Revenue -‐ Change in Operating Profit
-‐ Change in Operating Profit Margin
-‐ Change in Operating Profit on Total Assets -‐ Change in Operating Profit on Total Debt -‐ Change in Net Cash Flow from Operations
3.1.3 Time Period Investigated
The topic at hand studies the time period since the beginning of the sub-‐prime lending and sovereign debt crisis, arguably10 and generally accepted as having
commenced in 2008 with the collapse of Lehmann Brothers and the contagion that spread to European sovereign states and institutions. Consequently, under examination is the time period from 2008 through to 2014 (i.e. 2008 – end of 2013).
3.1.4 Sources for Data Collection
The computational design utilized encompasses index and stock price returns, as well as additional related calculations, derived from index values and each firm’s stock prices collected from Yahoo Finance. This allowed for a quick and reliable daily and monthly stock price data collection, valuable given time and accessibility constraints. Stock prices from Yahoo Finance are adjusted for stock-‐splits and dividend payments, thus preventing their impact on returns calculated. Moreover, operating performance values were obtained from annual consolidated financial statements presented in the annual reports published by each of the firms examined.
3.2 Calculations
3.2.1 Firm-‐specific Stock Price Returns
Figures calculated to represent stock price returns due to firm-‐specific factors, thus attempting exclusion of systematic influences, are based on the Single Index Model, as developed by William F. Sharpe (1964), and Lintner’s capital asset pricing model (1965). Given time and information constraints, SMB (Small market capitalization – minus – Big market capitalization), HML (High price to book ratio – minus – low price to book ratio) factors, as proposed by Fama and French (1993), and the MOM (momentum) factor (Carhart, 1997) are not accounted for. Consequently:
𝒓𝒊 = 𝜶𝒊+ 𝒓𝒇+ 𝜷𝒊 𝒓𝒎− 𝒓𝒇 + 𝜺𝒊
10
Please refer to footnote number 4 (p.5) for argumentations on the date the financial crisis is considered to have begun
where stock “i” returns, denoted by 𝒓𝒊, are defined by return due to firm-‐specific factors (𝜶𝒊), risk-‐free rate (𝒓𝒊), sensitivity to market index returns (𝜷𝒊) multiplied by market
risk premium, with market index return – benchmark – denoted by 𝒓𝒎, and residual returns, or unsystematic risk (𝜺𝒊). Accordingly, the values representative of returns due
to each firm’s specific factors are denoted by:
𝜶𝒊+ 𝜺𝒊= 𝒓𝒊− 𝒓𝒇+ 𝜷𝒊 𝒓𝒎− 𝒓𝒇
To represent 𝒓𝒊, yearly stock price returns were calculated for each firm from
the closing price of each country’s last trading day of 2008 through to the closing price of each country’s last trading day of 2013. Similarly, yearly returns were calculated for each country market index for the same time period for their corresponding firms, thus adjusting for the country-‐specific shared factors mentioned by Sharpe in the “Single Index Model”. Risk-‐free interest rates are determined by the European Central Bank’s main refinancing rate decisions. Given their variances within a year period, values were collected from the European Central Bank and an average was computed for each year.
For each year, firms’ stock return sensitivities to their respective market index returns (𝜷𝒊) were calculated with the formula:
𝜷𝒊 =
𝑪𝒐𝒗 𝒓𝒊, 𝒓𝒎 𝑽𝒂𝒓 𝒓𝒎
A 𝜷𝒊 value was calculated for each firm for each year (2009, 2010, 2011, 2012 and 2013). For maximum accuracy and reliability, each market index and firm variances are based on daily price returns for each year, computed from daily stock prices from the closing price of each country’s last trading day of 2008 through to the closing price of each country’s last trading day of 2013.
3.2.2 The Value for Comparison
Data collected on each firm’s operating performance indicators from consolidated financial statements from 2008 until the conclusion of 2013, provide values for change in operating performance for 2009, 2010, 2011, 2012 and 2013. To
determine whether significant variances exist between both regions core and periphery, values were calculated based on the formula aforementioned:
𝝎𝒊=
𝜶𝒊+ 𝜺𝒊 𝚫𝝁𝒊
where value 𝝎𝒊 is defined as return due to firm-‐specific factors in a pre-‐determined time
period, as denoted by 𝜶𝒊, plus residual returns, or unsystematic risk (𝜺𝒊), per change in
each operating performance indicator 𝝁𝒊 for the same pre-‐determined time period, as denoted by 𝚫𝝁𝒊.
Having both firm-‐specific returns and changes in operating performance reflecting the same pre-‐determined time period is based on a crucial yet reasonable assumption. As all companies examined are market index components, and given the quick market pricing reactions in the developed countries’ exchanges, yearly operating performances are assumed to be priced-‐in already by the last trading day of each year. Given the companies’ statuses as index components, several analysts and the companies themselves provide signals on operating expectations throughout the year, thus
converging stock prices and returns to a value that accounts for its operating results. Any differences would be insignificant, given observations account for yearly returns.
Figures computed for 𝝎𝒊 are solely intended to show whether significant divergences can be inferred for each different operating performance variable. Firm-‐ specific returns per change in an operating performance indicator for the same
determined time period are not to be judged, given this part of the study examines each operating performance variable independently, and firm-‐specific returns are based on more than one factor. As an example, negative returns due to firm-‐specific factors may occur at the same time one particular operating performance indicator increases, showing the difficulty in interpreting 𝝎𝒊 values.
3.3 Comparative Analytical Tests
3.3.1 Tests for Significant Differences Between Populations
Primarily, given the central focus of this investigation, 𝝎𝒊 values are examined between both populations, core and periphery Eurozone, and tested for any significant
divergences for each operating performance indicator. Given the figures do not allow for a judgement on their value alone, as previously explained, a two-‐tailed t-‐test for
population comparison analysis is used. The test for each indicator will provide conclusive evidence on whether significant differences exist regarding how each operating performance indicator weighted in on returns due to firm-‐specific factors for core and periphery Eurozone countries throughout the 2008 financial crisis. All tests are conducted with a significance level of 5%.
3.3.2 Correlations and Regressions
While examining for statistically significant differences between both regions through group comparative t-‐tests, further tests are carried out to attempt understand how changes in each 𝝁𝒊 (operating performance indicator) determined returns due to firm-‐specific factors for core and periphery Eurozone (𝜶𝒊+ 𝜺𝒊). In this manner, statistical results are provided to attest to the relevance of each variable for the pre-‐ determined period between 2008 and the end of 2013, as well as how they have moved against returns due to idiosyncratic factors through the same time period – clarifications that cannot be obtained from the comparative assessment of 𝝎𝒊 values.
In line with these aspects, Pearson-‐correlation tests, as well as regression analyses, are conducted for each independent variable (operating performance indicator) against returns due to firm-‐specific factors, individually for each Eurozone region under study (i.e. core and periphery).
3.4 Multiple Linear Regression Model
Lastly, in line with the methods applied for stock returns determination by independent variables in the single, three-‐factor and four-‐factor model, as well as interactions between returns and profitability ratios (Ghasempour, Ghasempour, & Bahonar, 2013), a model comprising all operating performance factors considered is statistically tested with a multiple linear regression, as to verify how well has overall operating performance explained firm-‐specific returns in the core and periphery Eurozone regions. Results should provide additional signals on the difference in the
weight operations has had in returns from 2008 to the end of 2013 for both regions, as well as the role each operating indicator has had in the presence of the other
independent variables. A thorough evaluation will ensue, also accounting for the observations from the individual assessments made beforehand. In this case, the following full model is utilized:
𝜶𝒊+ 𝜺𝒊 = 𝜷𝒊𝟎+ 𝜷𝒊𝟏 𝚫𝝁𝒊𝟏+ 𝜷𝒊𝟐 𝚫𝝁𝒊𝟐+ ⋯ + 𝜷𝒊𝒑 𝚫𝝁𝒊𝒑+ 𝜺
where returns due to firm-‐specific factors in the studied time period (𝜶𝒊) plus residual returns, or unsystematic risk (𝜺𝒊), is determined by operating performance indicators, as
denoted by 𝝁𝒊𝒑.
Similarly to tests for population differences, all correlation tests and regressions are conducted with a significance level of 5%.
4. Results Discussion
4.1 Tests for Significant Differences Between Populations
Change in Operating Revenue Change in Operating Profit Change in Operating Margin Change in Operating Profit/Total Assets Change in Operating Profit/Total Debt Change in Net Cash Flows from Operations Core Sample Size 75 75 75 75 75 75 Periphery Sample Size 75 75 75 75 75 75 Core Mean -‐3,80197 0,23024 0,84665 0,99453 -‐1,00503 -‐0,81168 Periphery Mean 2,12742 -‐5,95786 -‐1,33328 -‐1,0054 0,2594 -‐1,78748 Core Variance 369,54113 6,35408 51,17613 104,01548 84,60127 18,43975 Periphery Variance 540,05482 4 986,1054 222,56223 183,16756 29,19932 84,09046 Two-‐Tailed T-‐Test p-‐ level (5%) 0,09074 0,44938 0,25569 0,30843 0,30634 0,4053
The tests for difference conducted concerning both core and periphery Eurozone regions demonstrate that at a significance level of 5% there are no significant
divergences on 𝝎𝒊 values when 𝚫𝝁𝒊 is determined by any of the operating performance
variables considered. Consequently, results obtained with this particular method suggest that for the time period from 2008 until the completion of 2013, inferences cannot be made about meaningful variances existing in how returns due to firm-‐specific factors were valued for changes in operating performance for France, Germany and Netherlands (core Eurozone region), relative to the Eurozone peripheral Italy, Portugal and Spain. Consequently, this implies no significant bias existed.
Findings indicate no significant divergence between the values for firms in core and periphery countries regarding change in operating revenue, change in operating profit and change in operating margin. P-‐values observed, 0,09074, 0,44938 and 0,25569 respectively, show high probabilities of the occurrences given an equalization of population parameters, thus supporting the variation insignificance. Notably, data shows a considerably higher variance within the periphery group for the three
independent variables – in particular for change in operating profit. These observations are nonetheless inconclusive, given the difficulty in interpreting the 𝝎𝒊 values alone.
Moving onto changes in operating profitability ratios, more specifically adjusting operating profits to an indicator of company size (total assets), it is possible to still see that with this particular computational approach differences between the core and periphery countries remain statistically non-‐significant. The high p-‐values above the 5% significance level demonstrate that when the formula for 𝝎𝒊 values is based on how
much operating profit is earned per asset unit, there is no indication of significant variations on how they weigh-‐in on firm-‐specific returns between the two regions. Similarly, there seems to be no significant difference when accounting for changes in operating profit per unit of debt. In spite of the increased concern over corporate debt levels and the impact from increased sovereign credit spreads, factors that led to outflows from riskier periphery Eurozone members to firms in the core throughout the crisis, results still show no significant difference between the two regions.
The last operating indicator evaluated (i.e. change in net cash flow from
operations) follows the trend from the other independent variables: data collected and values computed offer no statistically significant variances between the core and
periphery Eurozone firms under examination. All together, results point out that there is no evidence of partiality on the effect operating performance had in firm-‐specific
returns toward any of the two regions examined. Alternatively, these observations may also be indicative of method inadequateness. As presented by figures 1, 2, 5 and 7 in ‘Appendix A’, credits spreads, Target2 balances and GDP momentum since 2008 varied significantly between the two regions, thus generating doubts on whether the 𝝎𝒊 method utilized is suitable to detect a bias on stock returns. However, 𝝎𝒊 values calculated were based on returns computed to represent firm-‐specific factors only, removing macroeconomic effects, possibly making these doubts unjustified.
4.2 Correlations and Regressions
In spite of the results demonstrated by the comparative analyses performed in the previous section, correlation and regression analyses can provide a perspective on how returns due to firm-‐specific factors have related with each operating performance indicator from 2008 until the end of 2013. While the 𝝎𝒊 method suggests no significant
differences between the two regions, the following section offers a scope for each region on: how each indicator has moved against firm-‐specific returns, their explanatory power for the time period determined, and whether on their own, each indicator can be
regarded as a significant variable for firm-‐specific returns in each region. Accordingly, the following observations further expand on what could not be derived and interpreted from the 𝝎𝒊 method.
On account of the possible inadequateness regarding the method utilized to investigate for significant variances, discussions on correlational and regression differences are not strictly dependent on the insignificance of divergences observed, thus permitting more accurate and reliable evaluations, and conclusive remarks. Moreover, the following tests also allow for examination between operating
performance indicators for the Eurozone altogether, on their relationship with firm-‐ specific returns throughout the 2008 financial crisis.
4.2.1 Change in Operating Revenue
Core Periphery
Total Number of Observations 75 75
Pearson Correlation R 0,04510 0,01415
Pearson Correlation p-‐level 0,70080 0,90407
Correlation Significance (5%) No No
R Square 0,00203 0,00020
Adjusted R Square -‐0,01164 -‐0,01350
Standard Error S 0,23699 0,28802
Model Regression p-‐level 0,70080 0,90407
Model Significance (5%) No No
Intercept Coefficient 0,03484 0,14338
Variable Coefficient 0,08153 0,01523
Variable p-‐level 0,70080 0,90407
Variable Significance (5%) No No
Evidence collected between 2008 and the conclusion of 2013 suggest a greater, yet very weak relationship between firm-‐specific returns and change in operating revenue in core members’ firms with respect to the periphery, with core “R” (0,0451) > periphery “R” (0,01415). Additionally, a higher “R Square” and lower “Standard Error S” indicates operating revenue performance has greater explanatory power in determining firm-‐specific returns in the core region. Tests also present higher variable coefficients for the core region relative to the periphery, suggesting that core Eurozone firms experienced a higher change in firm-‐specific returns for every growth unit of operating revenue. According to the comparative analysis in the first section of results however, this difference is not significant.
Notwithstanding these observations, very low “R Square” figures in both regions, as well as p-‐values above the significance level set of 5% reveal that operating revenue performance alone does not prove to have been a strong determinant in firm-‐specific returns for either region throughout the 2008 financial crisis. Low Pearson correlation values for both core and periphery further support this notion.
4.2.2 Change in Operating Profit
Core Periphery
Total Number of Observations 75 75
Pearson Correlation R 0,22433 0,15900
Pearson Correlation p-‐level 0,05301 0,17303
Correlation Significance (5%) No No
R Square 0,05032 0,02528
Adjusted R Square 0,03731 0,01193
Standard Error S 0,23118 0,28439
Model Regression p-‐level 0,05301 0,17303
Model Significance (5%) No No
Intercept Coefficient 0,04458 0,14582
Variable Coefficient 0,04470 0,05503
