these financial benefits, product-services can give rise to several competitive ... Personalized warranty. Help desk. Service supporting the customer. Training.
Exploring the linkage between servitization and financial performances: evidences from the HVAC industry. Filippo Visintin, Mario Rapaccini University of Florence Information Based Industrial Services Laboratory Department of Energy Engineering “Sergio Stecco” Via Cesare Lombroso 6/17 50134, Florence, Italy {filippo.visintin,mario.rapaccini}@unifi.it
Abstract. Based on previous research we identified 5 variables to measure the degree of servitization of manufacturing companies. Subsequently, we submitted a questionnaire to assess the value of these variables to 250 companies operating in the Italian HVAC industry, receiving 54 answers. Using the aforesaid 5 variables as clustering variables, we conducted a hierarchical cluster analysis. The cluster analysis produced three clusters (highly, medium, low servitized companies), between which the variables were significantly different in the mean. In order to test the effect of servitization on the firms’ financial indexes (ROS, EBITDA margin), we conducted a one-way ANOVA considering the financial indicator as dependent variable and the cluster to which companies belongs to as factors. The ANOVA revealed a significant effect of the level of servitization on both the financial indexes. Finally, we performed a contrast analysis thereby demonstrating that companies highly servitized perform better than those medium and low servitized. Keywords: servitization, HVAC, financial performance.
1 Introduction A generalized decrease of the returns on product’s sale, coupled with an increased focus on customer satisfaction, have been encouraging a rising number of companies to supplement their product offerings with services both before and after the sale. The rationale for this integration stems from the benefits that these services can generate [1], [7]. First and foremost, product-services can generate substantial revenues and profits [2]. These revenues can be hidden on a mark-up on product’s price, as well as explicit and coming from the sale of services, in isolation from the product. Moreover, they can provide a stable source of cash flows, being more resistant to the economic cycles that drive investments and equipment purchases [10]. In addition to these financial benefits, product-services can give rise to several competitive benefits.
Firstly, they can provide differentiation, making products more appealing. Such a benefit is particularly relevant in mature mass-markets, where low-cost competitors provoke fierce price competitions, especially if products are easy to copy and patents offer only limited protection against copying. Secondly, product-services can help achieve customer satisfaction [8] especially when products are complex and customers very service-demanding. Thirdly, by providing services it is possible to build strong, close, and positive relationships with customers. These relationships, in addition to increase customer (true) loyalty, allow the providers to collect reliability data as well as suggestions and complaints about the products and/or the services. These information help design products and/or services more tailored to the customers’ needs [5] and devise effective recovery actions whenever customers’ expectations are not met. Finally, being labor-intensive, services are less easy to imitate, and therefore represent a more sustainable source of competitive advantage. The shift from selling manufactured goods to offering packages of products and product-related services is usually termed as servitization [12]. Despite servitization it is, by no means, a new topic, it is undoubtedly the object of a renewed interest in literature ([1], [7], [8], [3], [4], In spite of all the discussion about the importance of servitization, however, there is remarkably little empirical evidence about the impact of servitization on firm’s financial performance [7]. This paper fills in, at least partially, this gap. In fact, on the one hand, we define a methodology to measure “how much” a manufacturing company is servitized; secondly, referring to companies operating in the Italian HVAC (Heating, Ventilation Air Conditioning) industry we identify a positive relationship between a higher level of servitization and better financial performance. We decided to focus on the HVAC industry for two reasons. Firstly, there are empirical evidences that in this industry servitization is an on-going process [13]. Secondly, the HVAC industry is characterized by huge installed base of long lasting products needing frequent maintenance. It determines a rather high demand for services and, as a result, a remarkable potential service business [13].
2 Methodology Building on previous research [2], [11], we prepared a questionnaire to inquire how companies operating in the Italian HVAC industry managed services. The questionnaire was submitted to the 250 companies belonging to the Italian Associations of the Manufacturers of HVAC equipments (Assotermica and UCC), thereby covering almost all the competitors operating in Italy. We received 54 answers. The sample obtained was, however, too heterogeneous (it included companies selling solar thermal system as well as heating tubular steel radiators) to compare the profitability of the companies within the sample. As a result, we narrowed the scope of our analysis and we selected only the companies producing condensing boilers, traditional boilers and water heaters. By doing so, we obtained a sample made of 26 companies. It is worth to mention that, however, all the most important competitors operating these segments are represented in the sample. We hypothesized to quantify the degree of servitization of a manufacturing company by means of five variables. These variables are described in the next paragraph.
2.1 Service portfolio variables With the first two variables we wanted to measure the width of company’s service portfolio. In order to do so we asked to indicate which of the services reported in the first column of Table 1were actually provided (directly or through third party service providers). Table 1. Service portfolio. Service offered Installation Fix and repair (under warranty) Substitution (under warranty) Provision of technical documentations Provision of the user’s manual Extended warranty Personalized warranty Help desk Training Technical and Maintenance consulting Maintenance Product upgrade Product check-up Spare parts provision Fix and repair (out of warranty)
Category Essential services
Service supporting the customer
Service supporting the product
Consistently with the classification provided by Mathieu [6] we subsequently classified these services in essential services, service supporting the customer (SSC) and service supporting the product (SSP) (second column of Table 1). Essential services are those services that companies are either obliged by law to provide or that are considered as market qualifiers by customers. Services supporting the customer (SSC) are those services offered, in addition to the essential services, to provide additional protection against risk (e.g. extended warranty services) and help customers to interact with the product. Services supporting the product (SSP), instead, are those services offered, in addition to the essential services, to ensure a correct product functioning over the time. Since all the companies in the sample provided essential services, these services were not considered in the analysis. The width of the service portfolio was thus measured by means of two numeral variables named SSC, SSP. These variables span from 0 to 5, and measure the number of SSCs and SSPs (respectively) that company actually provide to their customers. 2.2 Key Performance Indicators variables With the second two variables, we wanted to measure the extent to which the service performance were measured and controlled. In order to do so we asked to indicate which of the performance indicators reported in the first column of Table 2 were actually adopted.
Table 2. Indicators, subclasses and classes. Indicators % first time fix, % of failures that required more than one intervention because of a skill mismatch % of failures that required more than one intervention because of a spare parts shortage Mean Time to Repair Mean Time Between Failure Mean Downtime Number of failures fixed / Total number of failures % spare parts order evaded on time % of failures fixed remotely Preventive maint. Effectiveness (% failures avoided). Customer satisfaction index Spare parts delivery lead time, Number of spare parts stock-out Spare parts stock Spare parts rotation index Spare parts demanded /Spare parts in stock Workforce productivity indexes
Subclass
Class
First Time Fix indicators
Time/Response indicators
External KPI
Effectiveness indicators Custom. Satisf. Indicators Material-related indicators
Internal KPI
People-related indicators
These indicators were subsequently classified (see column 3 of table 2) in “external KPIs”, that is, indicators that measure the service performance according with a customer perspective, and “internal KPIs”, that is, indicators that measure the service performance according with an internal processes perspective. The first class of indicators was further subdivided in four sub-classes: “First Time Fix”, “Time/Response”, “Effectiveness” and “Customer Satisfaction” indicators (Blumberg, 1998, pp. 102). Hence, we defined a numeral discrete variable, named EKPI. Each time a company uses at least one indicator in a one of these sub-classes the EKPI variable increases by 1 (hence, EKPI ranges from 0 to 4). In the same fashion, we classified the second class of indicators in 2 subclasses: material related indicators and people-related indicators. Each time a company uses at least one indicator in one of these sub-classes the IKPI variable increases by 1 (hence, IKPI ranges from 0 to 2). 2.3 New product/service development variables Finally, we wanted to measure the extent to which the information coming from the field are used to design new products and services. Hence, we asked company how often such information are used. Three options were available: “never”, “only in presence of evident issues/opportunities”, “always”. Hence, we defined a numeric discrete variable named “NPSD”. NPSD assumes the value 0 if the information coming from the field are not used, 1 if they are used only sporadically and 2 if they are always used. By doing so we assumed that in the first case the information were
used the 0% of the time in the second case the 50% of time and in the third case the 100% of the time).
3 Analysis Using the previously mentioned 5 variables as clustering variables, we conducted an agglomerative hierarchical cluster analysis based on Euclidean distances and on the Ward’s algorithm. Before performing the analysis, we standardized the variables by calculating their z-scores (z=(x-μ)/σ ) in order to take into account the different scales of the variable (SSC, SSP variables span from 0 to 5, EKPI from 0 to 3, IKPI from 0 to 2, NPSD from 0 to 2). The agglomeration process was conducted by SPSS and produced the following agglomeration schedule in table 3 (the first 19 stages are omitted because of space limitations). Table 3. Agglomeration schedule. Stage
N° of clusters
20 21 22 23 24 25
6 5 4 3 2 1
Cluster Combined
Coeff.
Cluster 1 9 5 1 2 2 1
18,476 24,629 31,290 39,387 61,781 125,000
Cluster 2 25 10 12 5 9 2
Stage Cluster First Appears Cluster 1 Cluster 2 13 0 0 19 18 0 16 21 23 20 22 24
Next Stage 24 23 25 24 25 0
The coefficient reported in column 5 represents the distance between the clusters merged in each step. As can be noticed, as the number of clusters merged decrease from 3 to 2 such a distance increases a lot (from 39,39 to 61,78). It means that the clusters being merged are very different each other. Hence, the adequate number of clusters to consider is 3. As a result, we stopped the agglomeration process at the stage 23 thereby obtaining the following three clusters (Table 4). Table 4. Cluster analysis. Case 1 4 7 8 12 14 20 21 22 26 2
SSC 4 5 4 3 5 3 4 4 4 4 1
SSP 5 3 3 4 4 4 3 4 3 4 2
EKPI 4 4 4 3 4 3 3 4 3 3 2
IKPI 1 1 1 1 2 1 1 1 1 1 1
NPSD 2 2 2 2 2 2 2 2 2 2 1
Cluster 1 1 1 1 1 1 1 1 1 1 2
3 5 6 10 11 13 15 16 18 19 23 24 9 17 25
2 2 2 2 2 1 2 2 2 1 2 2 1 0 0
1 4 1 4 4 3 3 2 2 3 3 2 1 1 1
2 2 1 2 2 2 1 2 2 1 2 1 1 0 0
1 0 1 1 1 1 1 1 1 1 1 1 0 0 1
1 1 1 2 1 1 1 1 1 1 2 1 0 0 0
2 2 2 2 2 2 2 2 2 2 2 2 3 3 3
As can be noticed, the cluster analysis produced three clusters, between which the variables were significantly different in the mean. Looking at the clusters it is possible to observe that companies belonging to cluster 1 have a wider service portfolio including lot of SSCs and SSPs. In addition, their performance measurement systems include lot of external KPI, which denote a higher customer orientation. In addition, they systematically analyze customer’s feedbacks to improve their products and services. On the contrary, companies belonging to cluster 3 have a very small service portfolio including only essential service an few SSP and performance measurement systems including only very few external KPIs. In addition, customer’s feedbacks are neither analyzed, nor utilized as input to support the new services and products development. Finally, companies belonging to cluster 2 represent and intermediate situation. As a result, we considered the company belonging to the cluster 1 as “highly servitized”, the companies belonging to cluster 2 as “low servitized” and the company belonging to the cluster 3 as “non servitized”. Starting from their balance sheets and income statements, for each company in the sample we calculated the value of the EBITDA margin, ROA, ROI and ROS whose descriptives are reported in Table 5. Table 5. Descriptive statistics Indicators EBITDA margin ROA % ROS % ROI %
N 26 26 26 26
Mean 5,5702 5,1281 3,0100 8,1752
Std. Deviation 8,33121 5,58655 7,66188 12,42450
In order to test the hypothesis that servitization positively affect the firms’ profitability, for each financial indicators, we decided to perform a one-way ANOVA considering each financial indicator as dependent variable and the cluster to which companies belongs to as factors. In order to perform the ANOVA analysis, however, we needed to check for the homogeneity of variance and normality of the distribution of the indicators. Hence, we performed the Levene test and Kolgomorow-Smirnov test of normality (Table 6).
Table 6. Levene test of homogeneity of variances and test of normality of Kolmogorov-Smirnov Levene Statistic ,803 2,290 1,230 1,526
EBITDA margin ROA % ROS % ROI %
df1 2 2 2 2
df2
Sig.
23 23 23 23
,460 ,124 ,311 ,239
Kolmogorow Statistic ,182 ,131 ,206 ,129
df
Sig.
26 26 26 26
,027 ,200* ,006 ,200*
Unfortunately, both the ROA and the ROI indexes did not pass the test of normality (sig.>0,05) nor was possible to transform these variables to make them normally distributed. Hence, we decided to narrow our analysis to the ROS and to the EBIDTA margin thereby neglecting the effect of servitization on the capital turnover (ROI = (ROS x CT) = (operating margin/revenues) x (revenues/invested capital). Table 7. One way ANOVA Indicator EBITDA margin ROS %
Between Groups Within Groups Total Between Groups Within Groups Total
Sum of Squares 1499,527 235,702 1735,228 1346,593 121,019 1467,611
df 2 23 25 2 23 25
Mean Square 749,763 10,248
F 73,163
Sig. ,000
673,296 5,262
127,962
,000
The ANOVA (Table 7) reveals a significant effect of the level of servitization on both the ROS and the EBITDA margin. We therefore decide to test whether there are statistically significant differences between: (i) the ROS and the EBITDA margin of no servitized companies and servitized companies (first contrast: cluster 3 vs. clusters 1 and 2); (ii) the ROS and the EBITDA margin of low servitized companies and highly servitized companies (cluster 2 vs. cluster 1). The results of the contrasts test are reported in Table 8. Table 8. Contrast Tests Contrast EBITDA margin ROI %
1 2 1 2
Value of Contrast 44,0584 -6,8234 56,8311 -15,8169
Std. Error
t
Df
Sig.(2-tailed)
3,93408 1,34651 5,46659 1,87104
11,199 -5,067 10,396 -8,454
23 23 23 23
,000 ,000 ,000 ,000
As can be noticed the mean value of the ROS and EBITDA margin of the two experimental groups are significantly (p