Measuring sustained superior performance at the firm level

Documents de treball MEASURING SUSTAINED SUPERIOR PERFORMANCE AT THE FIRM LEVEL Emili Grifell - Tatjé Pilar Marquès - Gou Document de treball núm. 2002/8 Departament d'economia de l'empresa  E. Grifell - Tatjé, P. Marquès - Gou Coordinador documents de treball: Pere Ortín Ángel http://selene.uab.es/dep-economia-empresa/codi/documents.html e-mail: [email protected] Telèfon: 93 581451 Edita: Departament d'economia de l'empresa http://selene.uab.es/dep-economia-empresa/ Universitat Autònoma de Barcelona Facultat de Ciències Econòmiques i Empresarials Edifici B 08193-Bellaterra (Barcelona), Spain Tel. 93 5811209 Fax 93 5812555 Juliol, 2002 MEASURING SUSTAINED SUPERIOR PERFORMANCE AT THE FIRM LEVEL Emili Grifell - Tatjé Pilar Marquès - Gou Document de treball núm. 2002/8 La serie documents de treball d'economia de l'empresa presenta los avances y resultados de investigaciones en curso que han sido presentadas y discutidas en este departamento, no obstante las opiniones son responsabilidad de los autores. El documento no debe ser reproducido total ni parcialmente sin el consentimiento del autor o autores. Dirigir los comentarios y sugerencias directamente al autor, cuya dirección aparece en la página siguiente. A Working Paper in the documents de treball d'economia de l'empresa series is intended as a mean whereby a faculty researcher's thoughts and findings may be communicated to interested readers for their comments. Nevertheless, the ideas put forwards are responsibility of the author. Accordingly a Working Paper should not be quoted nor the data referred to without the written consent of the author. Please, direct your comments and suggestions to the author, which address shows up in the next page. 1 Measuring Sustained Superior Performance at the Firm Level Authors' names E. Grifell – Tatjé P. Marquès - Gou Departament d’Economia de l’Empresa Departament d'Organització i Gestió d'Empreses Facultat de Ciències Econòmiques Facultat de Ciències Econòmiques Universitat Autònoma de Barcelona Universitat de Girona Edifici B Campus Montilivi 08193 Bellaterra 17071 Girona Spain Spain [email protected] [email protected] 5 July 2002 Abstract This paper proposes a two-dimensional Strategic Performance Measure (SPM) to evaluate the achievement of sustained superior performance. This proposal builds primarily on the fact that, under the strategic management perspective, a firm's prevalent objective is the pursuit of sustained superior performance. Three basic conceptual dimensions stem from this objective: relativity, sign dependence, and dynamism. These are the foundations of the SPM, which carries out a separate evaluation of the attained superior performance and of its sustainability over time. In contrast to existing measures of performance, the SPM provides: (i) a dynamic approach by considering the progress or regress in performance over time, and (ii) a cardinal measurement of performance differences and its changes over time. The paper also proposes an axiomatic framework that a measure of strategic performance should comply with to be theoretically and managerially sound. Finally, an empirical illustration of the Spanish banking sector during 1987-1999 is herein provided by discussing some relevant cases. Key words Sustained superior performance, persistent performance, profit differences, sustainable competitive advantage. 2 1. Introduction Strategic management research focuses on explaining the heterogeneity of performance among firms. Performance evaluation uncovers such differences, addresses their measurement, and analysis. It provides a valuation of the outcome results attained by economic agents, in our case firms, which discriminates between success and failure in performance, and establishes the degree to which the former or the latter has been achieved. The importance of performance evaluation arises from the many consequences that successful or unsuccessful results can have in economic agents, not the least of which is the survival prospects of the organization, or the possibility to sustain a certain welfare situation. Furthermore, performance evaluation assesses the degree of success of the actions or strategies implemented, which will allow improvement in future decisions. Evaluating the performance of a firm implies the synthesis of the performance data available for a firm, which necessitates the use of measures or methods in charge of carrying out this synthesis. This aggregation aims to convey the relevant information about performance, therefore it has to consider how the judgements over performance results are made, thus reflecting the preferences of evaluators on results, in order to be theoretically and managerially sound. Therefore, information about the preferences of evaluators is required. Although preferences can vary across individuals, evaluators taking a strategic management point of view have reached a consensus on a firm's prevalent preference or objective: this is attaining superior economic performance to that of competitors (Rumelt, Schendel and Teece, 1994), assuming that it is the reward for having competitive advantages (e.g. Barney, 1997, Grant, 1998), which firms must achieve and sustain. The concept of competitive advantage dates back to Ansoff (1965) and was widely popularized by Porter (1979, 1980). A competitive advantage was to be the source of superior performance for firms, as it provided a strong competitive position (Ansoff, 1965: 110). Later, literature became more demanding requiring the sustainability of superior performance (Porter, 1985; Ghemawat, 1986; Barney, 1991; Conner, 1991; Amit and Shoemaker, 1993; Porter, 1996). This expansion of the concept includes a dynamic and strategic point of view, accepting that only some competitive advantages are difficult to imitate by competitors in the long run and, therefore, can lead to a sustained superior performance over time. Assuming this upgraded objective for firms, this paper proposes a 3 two-dimensional Strategic Performance Measure (SPM) to gauge sustained superior performance by carrying out a double evaluation: (i) the degree to which superior performance is obtained, as a result of achieving competitive advantages, and (ii) the sustainability of this superior performance. In this way, the SPM will report on the degree of achievement of Sustained Superior Performance (State 1). Otherwise, if this state has not been reached, the SPM will classify the performance of the firm into three other possible states: Eroding Superior Performance (State 2), Worsening Inferior Performance (State 3), and Reducing Inferior Performance (State 4). The definition of the four states is relevant as they all do take place. For example, the classic works of Mueller (1986, 1990) especially highlight the finding of firms whose performance would be classified in State 1, State 2, and State 4. Apart from presenting a new measure of strategic performance, this paper proposes a set of axiomatic properties which we believe should be held for other measures of strategic performance. These properties are based on three basic conceptual dimensions called: relativity, sign dependence and dynamism. These dimensions, in turn, stem from the objective of pursuing sustained superior performance. This paper proceeds as follows: the following section presents the background literature on the persistence of superior performance. The third section begins with the derivation of the three basic conceptual dimensions: relativity, sign dependence, and dynamism. It continues with the definition of the Strategic Performance Measure and its two components, and the analysis of the states of sustained superior performance and the three other possible states. It also sets out a graphic analysis of this definition, which enables the temporal analysis of superior performance and its sustainability. The theoretical evaluation of the two measures is presented in section four. An illustration of the proposed measures is given in section five, and a final section of conclusions ends the paper. 2. Background on Strategic Performance Evaluation The rationale for sustained superior performance is the possession of sustained competitive advantages. Therefore, the main task for managers is to find strategies that create, renew, and struggle to maintain competitive advantages, even in hypercompetitive contexts (D'Aveni, 1994). For this reason, literature has mainly focused on competitive advantage as the 4 dependent variable, i.e., in explaining the sources of sustained competitive advantage. As Wiggins and Ruefli (2002) point out, only a few studies have concentrated on the distribution of performance, or what they call the topography of performance. Studies of the distribution of performance are found in two literature sources: the first one is the literature on the persistence of profits, which has generally used time series methodologies, and is mainly found close to the domain of industrial organization; the second source of literature that we believe to be related to strategic performance is the one in ex post risk measurement. This branch of literature is relevant because it is concerned with what economic agents (e.g., firms, managers, investors) are averse to, which is necessarily related to some type of failure in performance, and failure is part of the strategic performance concern. Secondly, it must be noted that an important part of the literature on risk has an ex post focus, which means that it is actually carrying out an evaluation of realized outcomes. For example, the recent works of Miller and Reuer (1996), Miller and Leiblin (1996), Miller and Bromiley (1996), Ruefli, Collins and LaCugna (1999), and Reuer and Leiblin (2000) run mainly along the lines of ex post risk measurement. Another interesting feature of this body of research is that it is measure-oriented, which involves the development and evaluation of measures that synthesize ex ante or ex post performance. Time series approaches The main body of research on superior performance has been based on autoregressive time series methodologies. These methods were selected because the aim of this part of the literature was to study the dynamism of performance, namely the persistence or decay in performance. Mueller (1986, 1990) used this methodology with the purpose of examining the long-term persistence of superior ROA for large US industrial firms, mainly finding the convergence of performance towards the mean, although at a slower speed for the highestperforming firms, and some high-performing firms whose ROA increased over time. The same type of autoregressive methods were used in studies for other US and European firms or strategic business units, with similar findings (e.g. Geroski and Jacquemin, 1988; Jacobsen, 1988; Schohl, 1990; Droucopoulus and Lianos, 1993; Goddard and Wilson, 1996; Waring, 1996). More recently, Mueller and Raunig (1999) used the same autoregressive model developed in Mueller (1986) and Geroski (1990) to test whether the results from StructureConduct-Performance models estimated at the industry level are sensitive to the degree of heterogeneity of the firms in industries. Consistent with Mueller (1986) for the US over the 5 period 1950-1972, and Mueller (1990) for six other countries, his findings indicate that competitive forces require more than one year to eliminate short-term rents; that persistent differences in performance across firms exist within many industries; that it cannot be assumed that profits observed in an industry at a given point in time are near their long-term equilibrium which, in turn, is not the same for all industries; and that inter-, and withinindustry variations in profit rates are important in many cases. Therefore, the existence and persistence of profit differences is still an issue because the empirical findings encounter room for delays or violations of the expected decay in abnormal profits that the economic theory predicts. In the field of strategic management, Wiggins and Ruefli (2002) have recently made a contribution to the measurement of persistent superior performance. They use a new methodology introduced in Ruefli and Wiggins (2000) to stratify firms in groups of performance, and they investigate the stability over time of pertaining to the higher performance group by means of ordinal time series methods. Measure-oriented approach Many of the models and measures used to evaluate ex post risk have been borrowed from financial economics and statistical decision theory. The most traditional approach is the mean-variance model, which presents the mean as a measure of the central tendency of outcomes and the variance as a measure of its variability. The mean is widely accepted as a valuable description of a series of outcomes, but there is more discussion on the use of the variance, which is often presented as a measure of risk. Alternative approaches have abandoned variance to complement the mean with other measures aimed at considering what decision-makers perceive as risk, such as semi-variance, deviations below a target level or some derivative of a covariance, like the beta of the Capital Asset Pricing Model (Malkiel, 1989). However, strategy research has recently identified that reliance on these existing measures of risk may not be adequate for both the concept and the use in management research (e.g., Bettis and Thomas, 1990; Baird and Thomas, 1990; Miller and Leiblin, 1996; Ruefli, Collins and LaCugna, 1999). It can be considered that, the mean being a measure of the central tendency of a series of performance outcomes of a firm, a measure to complement it, namely ex post risk measure, should convey the information of outcomes which is relevant and not related to centrality. In 12 The property of dynamism is meant to reflect the sustainability of superior performance, i.e., the maintenance or the erosion of the advantage in performance. Therefore, the measure should increase if outcomes positively deviate from the reference levels, i.e., when it δ increases over time, and should decrease if outcomes negatively deviate from the reference levels, i.e., when it δ decreases. The measure of dynamic performance will be defined to incorporate this property. Let it z be the comparative performance change from period t-1 to period t as: 1− − = ititit z δ δ . If 0< it z, the firm has suffered an erosion in its outcomes relative to the reference levels in this transition from t-1 to t, failing to achieve any objective of sustainability. If 0= it z, the firm has maintained its position relative to reference levels, whatever it is. If 0> it z, the firm has improved its position, meeting its dynamic objective to improve performance over time. The aggregation of that behavior over time is carried out in the measure of Dynamic Performance ( i DP ) which for discrete historical outcomes is defined as follows:7 Definition 2: Dynamic Performance (DP) The measure of dynamic performance for a firm i in a period comprised between 1 and n is the mean value of it z from t=1 to n: () ∑ = =n t iti z n nDP 1 1 ,1 , being 1− −= ititit z δδ .8 If 0< i DP , the firm has, on average, suffered an erosion in its outcomes relative to the reference levels over this time-period, failing to achieve the objective of sustainability. If 0= i DP , the firm has maintained its position relative to reference levels in the period under analysis. If 0> i DP , the firm has, on average, improved its position, meeting its dynamic objective to increase performance. Apart from dynamism, i DP involves the other two dimensions because it ultimately aggregates on it δ , which complies with relativity and sign dependence, as can be seen in the expression of it z as a function of the original performance outcomes: [3] [4] 13 ( ) ( ) ( ) ( ) 1 1 1 11 − − − −− −−−=−−−=−= t ref t ref itit t ref it ref itititit xxxxxxxxz δδ .9 The definition of dynamic performance lets the final relative position of outcomes of a firm i, in δ , to be expressed as a function of the initial position, 0i δ , and the dynamic performance, as follows: iiin Dn P 0 + = δ δ . In turn, the final outcome value of the firm can be expressed as: ( ) i refref niin Dnxxxx P 00 +−+= . As can be seen in expression [7], i DP offers a simple interpretation and connection with the actual outcomes. This is due to its linearity. It is formulated as an average and can be interpreted as the average periodical rate of change of the relative position, which can be an increase in the distance to the reference level, when 0> i DP , a decrease when 0< i DP , or the maintenance of the relative position when 0 = i DP . The two component measures have been defined; however, the strategic evaluation of performance implies the integration of both evaluations to measure the degree of achievement of the objective of attaining a sustained superior performance. For this reason we formulate the Strategic Performance Measure as follows: Definition 3: Strategic Performance Measure (SPM) The Strategic Performance Measure (SPM) is a two-dimensional measure which evaluates the degree to which a firm has achieved sustained superior performance, assessing its relative position by means of the measure of Static Performance ( i SP ) and the temporal dynamics of the relative performance by means of the measure of Dynamic Performance ( i DP ). This two-dimensional approach allows the graphic representation and analysis of the performance of a firm, as is explored in the next section. Additionally, the performance over a long period of time can be divided into subperiods to introduce the analysis of the transitions in performance over time. [6] [5] [7] 14 3.3. The graphical display of the Strategic Performance Measure The signs of the two component measures of the SPM define four relevant performance states. In the most favorable one 0> i SP and 0≥ i DP , meaning that the firm has achieved, on average, a superior performance which has been sustained over time, or improved in the case that 0> i DP .10 In this case strategic performance has undoubtedly been successful. The extent of such achievement is measured statically by the value of i SP , and the average dynamic improvement rate is measured by the value of i DP . The strategic performance of a firm in that situation would be depicted in State 1 in Figure 1, and it is referred to as Sustained Superior Performance. The second best status would take place when 0> i SP but 0< i DP . In that case, the firm has, on average, achieved a superior performance measured by the value of i SP , but it has been eroding over time at the rate indicated by the value of i DP . This situation is depicted in State 2, and is named Eroding Superior Performance. -- Insert Figure 1 around here -- Should the firm be in a disadvantageous performance situation, its state will be in the lower quadrants, because 0≤ i SP . When the firm reduces the distance to targets, it reduces its inferior performance, that is, 0≥ i DP . It will be represented in State 4, in the right quadrant, and it is called Reducing Inferior Performance. The worst possible situation in strategic performance is found when a firm has an inferior performance, 0 < i SP , which is becoming worse over time, causing a negative dynamic performance, 0 < i DP . This has been named Worsening Inferior Performance, and it is represented in State 3 of Figure 1. The most desirable situation is being in State 1, but there are two other states which can be considered positively. In a short-term evaluation, firms would be required to have a superior performance, 0≥ i SP , thus being represented in the first row of Figure 1 (States 1 or 2). But in a long-term perspective, dynamic performance is what measures whether the progress is made or not with the aim of building superior performance, therefore increasing the importance of being preferably in the second column (States 1 or 4). However, there is one state which can not be considered good from any point of view: Worsening Inferior Performance is an undesirable state, in both a shortand a long-term perspective. 15 This two-dimensional approach to the measurement of strategic performance has several interesting features. In the first place, it defines objective criteria which classify the strategic performance of a firm into four possible states, considering not only superior performers but also inferior performers. Secondly, it distinguishes between the static position (superior or inferior performance) and the dynamic performance over time (improving or worsening), which could be assimilated to a shortand a long-run evaluation of performance, respectively. Thirdly, apart from the four states analysis, it offers a cardinal analysis, which is the synthesis of performance into a measure (two-dimensional) which allows the ordering of strategic performance according to the two dimensions. This could not be offered by the other dynamic approaches such as time series analysis and stratification techniques. Furthermore, the proposed Strategic Performance Measure will be tested to have some properties that make it conceptually and technically recommendable. 3.4 The transitions within the four states If the evaluation of strategic performance is carried out over a long time-period, it may well be worth considering making periodical measurements of the strategic performance achieved. Next we shall briefly exemplify how this analysis could be useful for strategic management. -- Insert Figure 2 around here -- According to literature, the performance of a firm can exhibit certain transitions along the four states of strategic performance. Figure 2 represents a hypothetical case. When a firm has achieved a differentiation advantage (Porter, 1985), it is expected to achieve a superior performance which will be maintained for a certain time-period, therefore being represented in SPM 1. If the advantage attained by the firm is only sustainable over a limited period, it is expected that the superior performance erodes over time, as imitators reproduce the advantages or innovators find minor advantages that allow them to catch up in performance. This occurrence would leave the performance evaluation of the firm in SPM 2. If the firm evaluated is not able to come to the market with more innovations in products, processes, or resources use, but some competitors do, the firm will lose the competitive advantage and therefore its superior performance. In this case, it will move away from SPM 2 because its static performance will be negative. If innovating competitors succeed in their advantages and increase performance along a time-period, the strategic performance of the firm under 16 analysis will move to SPM 3, where it does not have superior performance and its relative position even becomes worse over time. If the firm manages to react to that situation and place advantageous products into its markets, its performance can gain positions over time, that is, having a positive dynamic performance, which will bring the firm to SPM 4. Should this not happen, the evaluation of the firm would remain in State 3, which is the most dangerous state. To sum up, any trajectory that leads to State 3 (Worsening Inferior Performance) should be considered as extremely dangerous for the survival of the firm. On the contrary, trajectories leading to State 1 (Sustained Superior Performance) are the healthiest ones for a firm. The multiple observation of performance, in consecutive periods, provides a view of the stability of strategic performance, answering the questions of the changes of relative position in performance over time, and of the gain or loss in competitiveness, in performance terms. It allows the possibility to confirm several aspects of strategic performance: whether it is stable over time, whether its relative position is randomly moving from one state to another or not, whether relative gains and losses are produced over time; or whether a certain trajectory backed by theory stands or not. Also, it can be used to analyze the effects of the strategic interaction among firms, because the performance of relevant competitors can be analyzed in parallel, to evaluate the effects of their competition over time on strategic performance. In section 5 an application of the SPM to the Spanish banking sector is provided to illustrate the meaning of the measure and to study some relevant cases where different trajectories along the four states can be observed. 4. Theoretical Evaluation of a Measure of Strategic Performance In the previous section the two-dimensional measure of Strategic Performance has been defined, in terms of its two components: Static Performance and Dynamic Performance. The component measures were defined to comply with the three basic dimensions relevant to performance evaluation: relativity, sign dependence, and dynamism. In this section the three basic conceptual dimensions are transformed into an axiomatic framework that we defend should apply to any measure of strategic performance. The framework ensures that the measures make managerial sense and desirable mathematical properties. 17 The three basic dimensions, relativity, sign dependence, and dynamism, give rise to six mathematical properties. Two other conceptual properties are added to the evaluation system because they are supported by literature, and are compatible with and complementary to the previous six ones. The properties will be listed in turn and applied to the measures of Static Performance and Dynamic Performance as the components of the SPM. In the formulation of the properties, the expression good outcomes refers to those outcomes judged as successful because they achieve or exceed targets, i.e., the static outcomes which comply with 0> it δ or dynamic outcomes which comply with 0≥ it z. Bad outcomes will be the ones judged as failures in achieving targets, i.e., the static ones with 0 ≤ it δ or the dynamic ones with 0< it z. Property 1: Relativity in value. The contribution of an outcome to the value of performance assessment depends on the reference levels, as well as the value of the outcome itself. Property 2: Sign dependence. The marginal contribution of a good (bad) outcome is positive (negative), i.e., it increases (decreases) the value of the performance measure. Static evaluation: If δ ′ is a good (bad) static outcome, then SP(δ + δ') >(<) SP(δ). Dynamic evaluation: If z′is a good (bad) dynamic outcome, then DP(z + z') >(<) DP(z). Property 3: Monotonicity in value. The performance measure is continuous and increases (decreases) with the value of good (bad) outcomes. Static evaluation: If δ is a good (bad) static outcome, then 0/ > ∂ ∂ δ SP () 0/ <∂∂ δ SP . Dynamic evaluation: If zis a good (bad) dynamic outcome, then 0 / > ∂ ∂ z DP () 0 / < ∂∂ zDP . Property 4: Monotonicity in frequency. The performance measure is continuous and increases (decreases) with the frequency of good (bad) outcomes. Static evaluation: If p is the frequency of a good (bad) static outcome, then 0/ >∂∂ pS P () 0/ <∂∂ pS P . 18 Dynamic evaluation: If q is the frequency of a good (bad) dynamic outcome, then 0 / >∂ ∂ q DP () 0 / < ∂ ∂q D P. The following two properties only apply to the dynamic performance measure as they are derived from the dynamic dimension. Property 5: Dynamic sign dependence. If the outcomes of the firm, it x, increase (decrease) over time, the performance measure increases (decreases), ceteris paribus: being 1− −=∆ ititit xxx , then 0/>∆∂∂ ti xDP . Property 6: Dynamic relativity. If reference level outcomes, ref t x, increase (decrease) over time, the performance measure decreases (increases), ceteris paribus: being ref t ref t ref txxx 1− −=∆ , then 0/<∆∂∂ ref t xDP . There are two other conceptual properties which are not derived from the three basic dimensions but which are found in literature and have also been considered relevant. The first one is diminishing sensitivity, which considers that the marginal contribution of outcomes decreases with its magnitude. The rationale behind this property is that the psychological response to change is a concave function of the magnitude of this change. This property is proposed by the behavioral decision theory, and especially by the prospect theory (Kahneman and Tversky, 1979; Tversky and Kahneman, 1992). This property would require that the measure of performance should be concave for bad outcomes and convex for good outcomes. However, this property is not generally accepted. Widespread measures of ex post performance, such as the mean and the variance, do not comply with diminishing sensitivity, and even the variance presents increasing sensitivity. Because of the advantages of linearity and some criticisms on diminishing sensitivity, we propose non-increasing sensitivity. Property 7: Non-increasing sensitivity in the value of outcomes. Static evaluation: If δ is a good (bad) static outcome, then 0/ 22 ≤∂∂ δ SP () 0 / 22 ≥∂∂ δ S P . Dynamic evaluation: If zis a good (bad) dynamic outcome, then 0/22 ≤∂∂ zDP () 0 / 22 ≥∂∂ z D P. 19 Another conceptual property affects the reference or target levels, which are the values used to discriminate between good and bad outcomes. As already mentioned, valid reference levels must be relevant for the competitive arena of the firm or group under analysis. In strategic management literature, applications tend to use reference levels which are recalculated over time, often built with data from the industry. Aspiration levels from the behavioral theories of the firm (Cyert and March, 1963) also give support to changing reference levels over time. In the field of strategic management, the Strategic Reference Point theory (Fiegenbaum, Hart and Schendel, 1996) highlights the importance of time in defining reference levels. Therefore, the following conceptual, but not mathematical, property is proposed: Property 8: Reference levels change over time. The former eight properties describe the axiomatic properties for which we contend that any measure of performance should comply. The properties are intended to make the measure managerially and theoretically sound. Although some of the properties can be similarly found in literature, their joint consideration is not found, as far as we know. The set of properties applied to the SPM proposed in Definition 3 leads to the formulation of the following proposition: Proposition: The Strategic Performance Measure (SPM) complies with properties P1 to P8, as its constituent measures comply with their corresponding properties: 1. The measure of Static Performance, i SP , complies with all properties that apply to it (P1, P2, P3, P4, P7, P8). 2. The measure of Dynamic Performance, i DP , complies with all properties that apply to it (P1, P2, P3, P4, P5, P6, P7, P8). See Appendix 1 for proof of the Proposition. Compliance with the axiomatic properties proves that the SPM conveys the type of strategic performance evaluation defended in this paper, allowing SPM to be considered conceptually valid. The importance of this compliance depends, in turn, on the validity of the objective posed for firms: the search for sustainable superior performance. The prevalence of this objective maybe subject to discussion, but it is necessarily important either descriptively, normatively, or under both points of view. 20 5. Measuring Sustained Superior Performance in the Spanish Banking Sector To illustrate the measures proposed and their properties, this paper presents an application to the Spanish banking sector, which analyzes the performance of firms competing in the industry from 1987 to 1999. The Spanish banking sector is adequate for the application of the strategic performance evaluation because it is a sector with increasing competition, where strategic competition undoubtedly applies, especially for the biggest banks and savings banks, and for the innovative or differentiated medium banks. With this application, we shall show that static performance describes the degree of achievement of superior performance for a banking firm during a time-period, and dynamic performance captures the time trend of superior performance, which is its observed sustainability over time. The period under analysis is characterized by containing the final part of a long deregulation phase that had the objective to progressively comply with the Second Banking Directive of the European Union, which aimed at preparing the unified banking market in 1993.11 This liberalization implied an increase in the competitive possibilities for the banks, and a bigger one for the savings banks as they were previously affected by more restrictions. One of the main reactions to the new environment was the mergers wave, which significantly reduced the number of banking firms in the sector and, therefore, increased its dimension. Its peak period was in 1990, and it ended in 1995. After the consolidation wave, the leading role for strategic moves was mainly given to the slight post-merger rationalizations, stronger price competition, and branch expansion. Because of the structural and competitive changes, it is interesting to analyze the period under study in three parts: 1988-1991, 1992-1995, and 19961999.12 The first subperiod corresponds to the last steps in deregulation. The second would contain the first reactions to the new deregulated competitive framework, including the consolidation wave. The last period corresponds to a more stable post-regulatory phase. These subperiods contain 4 years each, therefore, the Static Performance will be based on these 4 years, and the Dynamic Performance on the corresponding 4 transitions in performance. According to the competitive situation, we would expect to find that Sustained Superior Performance (State 1) is an exceptional situation, as the increased competition erodes the previous privileged positions, situating the performance of the firm in Eroding Superior Performance (State 2) or even in the inferior performance states, in the case of more 21 hypercompetition. Theory would also predict that enlarged competitive possibilities would promote competitive moves, innovation, and more dynamism in the sector increasing the probabilities of transitions between performance states: from superior performance (States 1 and 2) to inferior performance (States 3 and 4), and conversely when some banks are able to build competitive advantages. The application uses the operating returns on assets as the relevant performance variable because it reflects the effect of strategies in the financial intermediation business, which is the traditional activity of banks. The SPM is computed using the operating returns´ sample mean as the reference level. The firm-level data is obtained from annual bank account records published by the Spanish Banking Association (AEB) and the Spanish Savings Banks Confederation (CECA). As a panel of results was preferable for the purpose of this application, we have reconstructed the returns of the merged firms in the years previous to the merger. The sample includes the banks which had a complete panel of data for the period and which did not present abnormal data in their annual accounts.13 This has led to a final sample of 50 savings banks and 37 banks. The results of computing the SPM for the sample show that Sustained Superior Performance (State 1) exists for some firms in every subperiod, but it is difficult to maintain. Actually, there is only one firm which keeps its SPM in State 1 for the three subperiods. This firm is labelled SB1 and it is a savings bank with a regional focus. The value of its sustained superior performance can be observed in table 1, which also presents the results of eight more banking firms that will be used as examples. The SPM of the firm is depicted in figure 3. It has achieved a superior performance because SP is positive in every subperiod, being around 0.5. This figure indicates that its performance has been approximately 0.5 points over the industry average. As regards to DP, it shows that its performance presented an increasing trend because DP is positive. Particularly, it increased at a rate of 0.142 in the first subperiod, but only at rate of 0.015 in the third one. In the same Figure there are two more examples of firms which keep their state during the three subperiods. B1 is a large bank with a national focus.14 Its SPM indicates that it has an Eroding Superior Performance (State 2). The temporal dynamics of SPM for B1 indicates that its superior performance (SP) has decreased over the three subperiods. The third bank represented in Figure 3 is B2, which is a small national bank specialized in commercial banking. Its SPM is Worsening Inferior Performance for the three subperiods. The inferior performance (SP) ranges from 0.591 in 28 Appendix 1 Proof of the proposition Property 1 Relativity in value. 1a. () 1 1 ∑ = −= n t ref titi xx n SP , therefore it is a function of i x and ref x. 1b. () ∑ = − − − =n t ititi n DP 2 1 1 1 δδ , thus, it is a function of it δ , which, in turn, is a function of i x and ref x. Property 2 Sign dependence. 2a. A good static outcome complies with 0'> δ and has a frequency of n 1. If it is added to the existing δ series, ceteris paribus, which has a certain static evaluation measured by () δ SP , then, () () δδδδδδ ′ += ′ +=+ ∑ =n SP nn SP n t it 111 ' 1 . As 0'> δ , then () ( ) δ δ δ SPSP >+ '. 2b. Similarly to 2a, for SP, if δ ′is a bad static outcome, then 0 < ′ δ , thus, () ( ) δ δ δ SPSP <+ '. 2c. For DP, if z′is a good dynamic outcome, 0'>z, and if it is added to the existing z series, ceteris paribus, 11 1 z n z n DP n t iti ′ += ∑ = , that is, ( ) ( ) )('.' zqzzDPzzDP + = + . As 0'>z, then () () zDPzzDP >+ '. 2d. Similarly to 2c, for DP, if z ′ is a bad static outcome, then 0< ′ z, therefore () () zDPzzDP <+ '. For the monotonicity properties we can express SP and DP in a more general expression, separating good from bad outcomes and considering the frequency of each of the outcomes, as follows: () )()( 00 δδδδδ δδ ppSP itit ∑∑ ≤> ⋅−⋅= , ( ) ( ) ∑ ∑ <≥ ⋅−⋅= 00 itit zz izqzzqzDP . Property 3 Monotonicity in value. 3a. For SP, if 0> δ , 0)(/ >=∂ ∂ δ δ pSP . 3b. For SP, if 0≤ δ , 0)(/ < −=∂ ∂ δ δ pSP . 3c. For DP, if 0≥z, 0)(/ >=∂ ∂ zqzDP . 3d. For DP, if 0<z, 0)(/ < −=∂ ∂ zqzDP . 29 Property 4 Monotonicity in frequency. 4a. For SP, if p is the frequency of a certain good static outcome, then 0/>=∂ ∂ δ pSP . 4b. For SP, if p is the frequency of a certain bad static outcome, then 0/ <−=∂∂ δ pSP . 4c. For DP, if q is the frequency of a certain good dynamic outcome, then 0/>=∂∂ zqDP . 4d. For DP, if q is the frequency of a certain bad dynamic outcome, then 0/ <−=∂∂ zqDP . The expression of the measure of Dynamic Performance can be rewritten in terms of the increments of outcomes for the firm under evaluation, 1− −=∆ t i t i t ixxx , and for the reference levels ref t ref t ref txxx 1− −=∆ , leaving the expression of i DP as follows: () ( ) ( ) ∑ ∑<∆−∆≥∆−∆ ⋅∆−∆−⋅∆−∆= 00 ref tti ref tti xx ref tit xx ref titi zqxxzqxxDP . Otherwise, it can be written as: ( ) ( ) ( ) ( ) ∑ ∑<∆−∆≥∆−∆ ⋅∆−∆−⋅∆−∆= 00 ref tti ref tti xx it ref t xx ref titi zqxxzqxxDP Property 5 Dynamic sign dependence. 5a. For a good dynamic outcome, that is, when 0≥∆−∆= ref titit xxz , then 0)(/ >=∆∂∂ zqxDP it . 5b. For a bad dynamic outcome, that is, when 0 <∆−∆= ref titit xxz , then 0)(/ >=∆∂∂ zqxDP it . Property 6 Dynamic relativity. 6a. For a good dynamic outcome, that is, when 0≥∆−∆= ref titit xxz , then 0)(/ <−=∆∂∂ zqxDP ref t. 6b. For a bad dynamic outcome, that is, when 0 <∆−∆= ref titit xxz , then 0)(/ <−=∆∂∂ zqxDP ref t. Property 7 Non-increasing sensitivity in the value of outcomes. 7a. For SP, 0/22 =∂∂ δ SP because SP is linear on δ . 7b. For DP, 0/22 =∂∂ zDP because DP is linear on z. Property 8 Reference levels change over time. 8a. For SP, as ( ) ref titit xx −= δ , reference levels are introduced by ref t xwhich are defined to vary over time. 8b. For DP, as () 1− −= ititit z δ δ , dynamic reference levels are variable over time as they are set as the previous period deviation. In turn, static reference levels are introduced by ref t xwhich are defined to vary over time. 30 Dynamic Performance 0 < i DP 0≥ i DP 0> i SP State 2 Eroding Superior Performance State 1 Sustained Superior Performance Static Performance 0≤ i SP State 3 Worsening Inferior Performance State 4 Reducing Inferior Performance Figure. 1 Four states in strategic performance evaluation Dynamic Performance 0 < i DP 0≥ i DP 0> i SP Eroding Superior Performance Sustained or Increasing Superior Performance Static Performance 0≤ i SP Worsening Inferior Performance Reducing Inferior Performance Figure. 2 Transition example of performance SPM 1 SPM 4 SPM 2 SPM 3 31 SPM SPM SPM 1988-1991 1992-1995 1996-1999 SP DP SP DP SP DP SB1 0.483% 0.142% 0.588% 0.024% 0.443% 0.015% B1 1.171% -0.092% 0.505% -0.129% 0.148% -0.031% B2 -0.591% -0.361% -1.401% -0.102% -1.628% -0.005% B3 2.061% 0.194% 2.013% -0.077% 1.570% -0.145% B4 2.099% 0.711% 1.011% -0.601% -0.252% -0.261% SB2 -0.670% -0.220% 0.150% 0.273% 0.351% 0.039% SB3 -0.646% -0.262% -0.251% 0.331% 0.489% 0.080% SB4 -0.702% -0.074% -0.478% 0.094% -0.496% -0.041% SB5 0.899% -0.153% -0.069% -0.231% -0.291% 0.079% Table. 1 Strategic Performance Measure results for the example firms 32 Strategic Performance Measure: Examples I SB1_1 SB1_2 SB1_3 B1_1 B1_2 B1_3 B2_1 B2_2 B2_3 -3,0% 0,0% 3,0% -0,5% 0,0% 0,5% Dynamic Performance Static Performance SB1_1 SB1_2 SB1_3 B1_1 B1_2 B1_3 B2_1 B2_2 B2_3 Figure 3. Strategic Performance Measure: Examples I 33 Strategic Performance Measure: Examples II B3_1 B3_2 B3_3 B4_1 B4_2 B4_3 SB2_1 SB2_2 SB2_3 -3,0% 0,0% 3,0% -1,0% 0,0% 1,0% Dynamic Performance Static Performance B3_1 B3_2 B3_3 B4_1 B4_2 B4_3 SB2_1 SB2_2 SB2_3 Figure 4. Strategic Performance Measure: Examples II 34 Strategic Performance Measure: Examples III SB3_1 SB3_2 SB3_3 SB4_1 SB4_2 SB4_3 SB5_1 SB5_2 SB5_3 -3,0% 0,0% 3,0% -0,5% 0,0% 0,5% Dynamic Performance Static Performance SB3_1 SB3_2 SB3_3 SB4_1 SB4_2 SB4_3 SB5_1 SB5_2 SB5_3 Figure 5. Strategic Performance Measure: Examples III 35 Footnotes 1 In ex post risk literature there have also been some time-series approaches (e.g., Bettis and Mahajan, 1990), but the majority of the works have been measure-oriented. 2 The definition of Besanko, Dranove and Shanley (1996) actually refers to competitive advantage, which they assimilate to superior performance as it is the expected cause-effect relationship. 3 See references in section 2. 4 The reference, or target level, is considered to be only dependent on time, in the sense that it is common for all firms belonging to the system under analysis, as the evaluation takes place in a context of strategic competition. 5 Should the outcomes be continuous, the expression for the static performance measure would be ∫ =dpSP iti δ , where p is the density function of outcomes. 6 The aggregation of the static performance, it δ , is carried out linearly to comply with sign dependence. Nevertheless, the aggregation could include the consideration of weights, which would leave its expression as: () ∑ = =n t ititi wnSP 1 ,1 δ . This would allow the definition of weights variable over time or over values of it δ , reflecting varying preferences over time or values, respectively. For the purpose of this article, the unweighted aggregation is preferred as it assures a convenient information content of the measures proposed, and we find no theoretical need to make a weighted aggregation. Further, the unweighted proposal does not cause the analysis any relevant loss of generality, because the inclusion of standard weights (positive) would not change the axiomatic and theoretical analysis provided. 7 Should the oucomes be continuous, ∫ =dqzDP iti , where q is the density function of outcomes' transitions. 8 The aggregation of the dynamic performance, it z, is carried out linearly to comply with sign dependence. Nevertheless, the aggregation could include the consideration of weights, which would leave its expression as: () ∑ = =n t ititi zwnDP 1 ,1 . See footnote 6 for further comments. 9 This expression also shows the equivalence between first considering relativity and then dynamism, or the reverse. 10 The signs of the inequalities have been defined to comply with the objective of sustained superior performance, superior performance being identified with 0> i SP , and sustained or increasing performance over time with 0≥ i DP . 11 The main policies included were the complete deregulation of interest rates (1987); the end of the branch expansion control (1985 for banks, 1989 for savings banks); the abolition of restrictions to the entrance of foreign banks (1988, 1993); and the removal of the liquidity requirements which implied compulsory investment in government bonds (1992). 12 Note that the initial period (1987) is not included in the evaluation as there is one initial period needed to compute the dynamic performance, z, of 1988. 13 The banks involved in mergers and acquisitions are not included in the sample because they do not have a complete panel, but they are considered in the reconstructed return series for the bank resulting from the consolidation. Abnormal observations were the ones that stated null assets or workers during the period under analysis, indicating that the bank was operating in special situations, such as whether it was temporarily inoperative, being restructured or operating in other activities but not in the traditional banking intermediation. 36 14 Large banks and savings banks are those belonging to the top ten in average assets during the period under analysis.