Getting a Grip Understanding and Controlling Suppliers Prices in Three Steps

(1)

Getting a Grip

Understanding and Controlling Suppliers Prices in Three Steps

Michiel Buysing Damste S1271083

(2)

Getting a Grip

Understanding and Controlling Suppliers Prices in Three Steps

-Public Version -

Michiel Buysing Damste

S1271083

University Supervisors: Drs. D.A. Vegter Dr. B. Goldengorin

Large company Supervisor: ……….

Groningen, 2007 University of Groningen

Faculty:

Management and Organisation Specialisation:

(3)

Preface

In front of you is the result of the final challenge the University of Groningen gave me, writing a final thesis based on an internship. About six months ago the kick-off of the project was given by Large company by providing me a tough problem which was hard to define. “Can you help us with this? …….”, “Well, I think so……” I replied, not sure yet how to do it.

The goal for Large company in this project was to get a better understanding of the prices paid for Research. Why is the price for this research three times as expensive in London than in Moscow? Why is the price for Research as it is and what factors drive the price? These questions are examples of problems Large company faced regarding the prices for Research.

Alongside this problem, Large company provided me a very pleasant environment to do this project which helped me a lot to do this project successfully. I have felt that people are always willing to help each other and always provided me with a lot of support. Therefore I would like to thank all my colleagues in Large company. I would like to thank my supervisor in Large company especially for guiding me and thinking with me in the process of conducting this research.

A project like this is also guided by a coordinator from the university. I would like to thank Dennis Vegter for guiding me through this last phase of my studies. Although the pretty significant distance between Rotterdam and Groningen I feel we had a good cooperation in which I was given helpful feedback. Also I would like to thank my co-assessor Boris Goldengorin. He managed to overlook the entire problem very quickly and provided me with useful feedback.

Finally, I would like to thank my family for their unconditional support during my entire studies. Also my friends deserve a “thank you” here for a great time which motivated me and made me enjoy my studies a lot!

Michiel Buysing Damsté Groningen, May 2007

(4)

Executive summary

This research aims to improve the understanding of suppliers’ prices for Research within Large company. In Large company NPI Supply Management Europe establishes contracts with suppliers of Research. The marketers and the CMI department order researches according to the prices that are in these contracts.

There is a lack of information and knowledge on the prices for Research in Large company. More specifically, it is not clear what activities influence the price of Research and how the prices and the influence of the activities change when the characteristics of the research change. By the characteristics of the research is meant for example the length of the interviews. The second part of this problem regards the influence of the country where the research takes place on the price for Research.

Information and knowledge on the prices is necessary in order to improve the bargaining power of Large company to its suppliers. The more knowledge on the prices for Research the more powerful Large company will be in the negotiations. Therefore the output of this research should be a tool that improves the understanding of the prices on Research and will be helpful to improve the bargaining power.

In order to come to a desirable output this research is split into two phases. The first phase is the diagnoses phase followed by the design phase. In the diagnoses phase the focus is on the allocation of the causes of the problem. The design phase is about transforming the problem and its causes found in the diagnoses phase into an applicable solution.

The diagnoses phase starts with identifying the major costs for Research. Relative weight in the total price and linear correlation analysis proved that the price for fieldwork is a substantial part of the total price and has a direct influence on the total price. Therefore the price for fieldwork is chosen as the focus area for this research.

The diagnoses phase is split up in two parts: a quantitative analysis and a qualitative analysis. The quantitative analysis tries to find relationships between variables and the price and looks for consistency over countries. To look for the relationships mentioned above the data from current contracts with suppliers are used. In these contracts the price for fieldwork is related to data collection method, interview length and incidence rate. In order to find relations, linear correlation analysis is used. With the use of graphs the behaviour of the prices based on the different variables, as interview length, are compared. Although direct relationships were found between the price and the variables, the behaviour of the price showed very different patterns. The behaviour found is that contradicting that no conclusions can be based on the behaviour of the price based on these variables.

In order to check for consistency in suppliers’ prices over countries all prices are made equal by the ESOMAR index. The ESOMAR index is an index to compare prices over countries specialised on Research. The prices over countries are compared for consistency with the repeated measures ANOVA method. For all types of research very little relationships were found over countries. This is an indication that the prices in the contracts are not equally competitive for all countries.

The qualitative analysis aims to identify the activities that bring costs with them and influence the price significantly. These activities are analysed in the way they influence the price for fieldwork based on the principle of product costing. The data used for the qualitative analysis is collected in expert interviews within Large company as well as with two suppliers. In the analysis is found that labour costs, incentives and the cost for location (in case of central location research) are the major costs. In the qualitative analysis is found that the influence of the cooperation rate is higher than expected. Another finding in this Chapter is that the

(5)

data collection method has a large influence on the price for fieldwork. Not taking exceptions into account can be said that the more advanced the data collection method the cheaper the price for fieldwork. According to this finding is found that the preferred supplier for Type A does not master CAWI research.

The conclusion from the diagnoses phase is that the lack of knowledge on the price that was assumed in the problem description is proved by the strange behaviour of the price found over the different rate cards. This in combination with the little consistency in the price over countries shows that there is also a lack of control of the suppliers’ prices. The qualitative analysis has given a better understanding which can be a good starting point for the design phase.

Based on the findings in the first part of this Chapter the second part of the research’ target is to transform the current situation into a desired future situation. The design aims to improve the understanding and control of the suppliers’ prices. The design phase can be split into two phases. The first step is necessary to improve the understanding of the price for fieldwork and is also necessary to design a valid control tool. The second part is the design of the actual control tool.

In the diagnoses phase is found that the incentives and cooperation rate have a bigger influence on the price than was thought before. In the current situation incentives are still included in one price for all variable costs while the behaviour of the price is contradicting with the behaviour of the price for interview length and incidence rate. Therefore in the design the price for incentives is taken out of the variable costs and put into a separate framework. This provides the opportunity to increase the knowledge because it will provide knowledge on a lower level which will improve the knowledge on the level desired.

The control tool consists of three steps. The standards set in these three steps are guided by the regulating loop. The regulating loop is a good tool to control prices because it compares the actual situation with the standards how it should be which can lead to taking actions.

The firststep of the control tool checks the behaviour of prices with the standards following

from the activity analysis from the qualitative analysis. The second step is to prevent suppliers from moving costs from fieldwork to client servicing just to meet the standards set in the first step. Therefore the relative weight of both prices are compared to previous situations and is the price for client servicing checked with the ESOMAR index for commercial tariffs. The third step is to prevent the suppliers from applying a give and take strategy. By this is meant that the supplier gives beneficial rates in one country but gain more in another. To check for consistent rates the repeated measures ANOVA test is used after prices are made equal by the general ESOMAR index.

The new design will be beneficiary for Large company in several ways. It will improve the understanding of the prices for Research and because of the separation of the incentives from the other costs. It will help the understanding for other types of research as well, since it will provide knowledge on the cooperation rate in the different countries. Besides the previous it will make the results obtained in the negotiations equally competitive all over the rate card and will improve the possibilities to price researches outside of the rate card. Maybe most important of all it reduces the ability of the suppliers to ‘fool around’ with prices. The three step method also helps to make the sourcing process more efficient in terms of dependability, quality and speed. The general conclusion is that the knowledge on prices and the control of prices for Research is improved and that this research is a good first step to further improve the knowledge on prices for Research for which recommendations are given in this research.

(6)

α

= .05) WHEN ALL

µ

’S ARE EQUAL 57 -APPENDIX C: CORRELATION INCIDENCE RATE AND INTERVIEW LENGTH ON PRICE

FIELDWORK 59

-APPENDIX D: ESOMAR RELATIVE PRICE INDEX FOR COUNTRIES INCLUDED IN THIS

RESEARCH 60

-APPENDIX E: FISHERS LEAST SIGNIFICANT DISTANCE FOR TYPE A AND TYPE B 61 -APPENDIX F: STANDARD RESEARCHES FOR TYPE B AND TYPE B COMPARING DATA

COLLECTION METHODS 63

-APPENDIX G: BEHAVIOUR OF THE PRICE FOR FIELDWORK ON INCIDENCE RATE 65 APPENDIX H: BEHAVIOUR OF THE PRICE FOR FIELDWORK ON INTERVIEW LENGTH: 66

(8)

(9)

Chapter 1 Organisation and processes

In this Chapter an overview of the organisation and its products will be given. The role of NPI Supply Management Europe within the total organisation will be described focussing on the Marketing Services team, the environment this research is performed in. In the second part of this Chapter the processes within the Marketing Services team will be described.

NPI Supply Management Europe

In Large company Europe there is a specific group of supply managers who deals with the procurement of non production items (NPI), which is everything bought with 3rd parties, excluding packaging and ingredients. NPI Supply Management is organised on a regional level although for some items globally.

NPI engages as a business partner with its internal customers and makes use of all available knowledge and resource from external partnerships (e.g. outsourcing) to accelerate delivery of the mission. NPI seeks selectively strategic partnerships with suppliers, delivering mutual benefits, including supplier innovation.

In NPI Supply Management Europe there are several teams dealing with different areas: - ESP; Employee Site and Professional Services

The ESP team work primarily to negotiate, contract manage and track the performance of selected supplier agreements on fleet, HR services (temporary labour, learning/courses, Rewards), travel & expenses & site services (reprographics, furniture, facilities management, couriers, office supplies & trademarks).

- E&T; Engineering and Technical services The Engineering and Technical team focuses on equipment, technical services and utilities. - MM; Marketing Materials The Marketing Materials team focuses on promotional items, point of sale materials, commercial print management and Brand activation agencies

- MS; Marketing Services The Marketing Services team focuses on media, creative, PR and communication agencies and Research. This research takes place in the area of Research.

These teams base their sourcing activities on the following 7 step Strategic Sourcing methodology which is the guideline for everything that is purchased:

(10)

Figure 1.2 7-step strategic sourcing methodology

The methodology is based on an AT Kearney report where strategic sourcing is described as a disciplined, systematic process for reducing the total costs of externally purchased materials goods and services while maintaining/improving levels of quality, service and technology.

Research

The purchasing of Research consists of different types of research. Within Large company two types of Research are distinguished: After launch research and Before launch research. After launch research researches are all researches for products that are already on the market and can be purchased by consumers. Research in this field is done on a continuous base. Examples of After launch research researches are research to determine the market share opposed to the competitors, determine the group of consumers that are buying the products, etc.

Besides After launch research there is Brand Development. This type of research is executed on an ad hoc basis and can be as well quantitative as qualitative. Before launch research focuses on ideas for products, packaging, or advertisements that are not yet on the market.

Confidential

Figure 1.3 Overview Research

This research focuses on the Before launch research, there are five types of research Large company orders for the purpose of Brand Development.

1. Profile cluster and Supply Market

2. Sourcing strategy for cluster

3. Supplier Portfolio Generation

4. Selection of implementation path

5. Negotiation and Selection of Supplier(s)

7. Continuous Benchmarking of Supply market

6. Operational Integration with Suppliers

O

V

E

R

V

I

E

W

(11)

Type A (TYPE A): In this research the product or a prototype of a product is tested with potential customers. Large company has selected several suppliers (max. 3 per region) as preferred suppliers, where the internal customer the department (CMI) should order its research. This type of research is rather standardised with relatively little differentiation. Type c (TYPE C): This research takes place in a early stage of the development of a new product. During this research is tested whether the potential customer would be willing to buy a certain product. Company X from company Y has been the supplier for this research to Large company for a period in time. Therefore they have great knowledge about Large company and its products and delivers high quality research (perceived by CMI). There are other suppliers in the market for TYPE C but the experience and knowledge of company X combined with their high quality database (supported by company Y) makes them the preferred supplier for CMI.

Type B (TYPE B): This research tests commercials, finished and unfinished, with consumers to see if it has the right communication effect. Starting from the beginning of 2007 this type of testing is going to be applied for all types of advertisement. The market for Type B is similar as the market for TYPE C there are more suppliers which can perform the same sort of research although CMI has one preferred supplier which is company Z. The difference in quality between company Z and its competitors is that big that company Z has been identified as the only supplier of Type B research for Large company.

Type D: These studies are very different from the previous ones. In this research all the habits of the consumer are covered. Questions like: How does the customer like to pour his dressing on a salad?, are covered. This type of research is done once every two / three years. Opposed to the other types of research, the person that orders does not know what he or she is looking for up front.

Type E: In this research the packaging of the products are tested with the consumers. What does appeal to the consumer and brings out the right message? There are many potential suppliers that are able to perform this type of research. Within Large company there is no clear coordination of the procurement of this research. The outsourcing partner in the United States has taken responsibility to source this for North America as well as Europe. The reality in Europe at the moment is that CMI picks its own suppliers at prices that are not negotiated before hand.

Figure 1.4 Innovation funnel

Feasibility Capability Launch and roll out Ideas TYP E C TYP E C TYP E A CTY PE A TYP E A TYP E C CTY PE Or TYP L A U N C H

(12)

In the figure above can be found where in the innovation process the different researches are executed.

Conclusion

This Chapter set the environment for this research, the position of the NPI Supply Management team Europe has been described and the positions of the Marketing services team within NPI. The emphasis for this thesis will be on Research. This Chapter can be seen as basis for further analysis and general understanding.

(13)

Chapter 2 Research Design

The background for this research will be described in this Chapter. Out of the problem description a research plan will be extracted and the relevance of the problem will be shown. This will lead to the goal and research question for this research, followed by the research structure.

Perceived problem

As described in the previous Chapter the marketing services team is responsible for the negotiation of the contracts with the suppliers for Research.

In 2006 the marketing services team made, in cooperation with one of the preferred suppliers for Type A, a rate card for the purchased research based on the variables CMI uses to order the research. A preferred supplier is a supplier with whom Large company prefers to do business for a certain type of research. This is supported by a contract. CMI is encouraged/obliged to do business with these suppliers. With other suppliers for this type of research is agreed to use the same structure based on variables demanded by how the customer, CMI, orders the research with the supplier. The rate card covers a large percentage of all the researches but is not exhaustive and sometimes there are arguments with suppliers about prices of researches that are not in the rate card.

For the other markets of Before launch research, suppliers have not been as cooperative as the suppliers for Type A. There are rate cards but about the prices negotiated it is unknown whether they are competitive. The markets for other research within Before launch research are different from the market for Type A. In the market for Type A there are several suppliers per region to do research, which has improved the bargaining power of Large company. For other types of research there are only few suppliers and due to quality issues, demand from CMI and switching costs, the current suppliers have become a sort of monopolists. Therefore Large company does not have much bargaining power. As a result Large company does not have enough information and thus knowledge than would be desirable. Besides the issues above Large company has found that very different prices are charged over different countries and feels that prices negotiated in the different countries are not all as competitive as they would like them to be.

The rate card for Type A contains the market prices for the 28 countries included. To determine the prices for researches in countries that are not included in the rate card, the ESOMAR index is used to determine the price relative to a country with similar characteristics that is included in the rate card. According to the suppliers this method is not delivering accurate prices for countries that are not in an economically stable environment. There are two main problems identified:

• There is insufficient information about the negotiated prices. More specifically, it is not clear what activities influence the price of Research and how the prices and the influence of the activities change when the characteristics of the research change. By the characteristics of the research is meant for example the length of the interviews. The second part of this problem regards the influence of the country where the research takes place on the price for Research.

• To translate the prices in unstable economic regions for countries included in the rate

card to prices for countries not included in the rate card there is no clear accurate method.

In this research the focus will be on the first problem identified. This problem is experienced for all types of research. Although there is more knowledge on the prices for Type B there are still some questions to be answered here as well. The second problem will not be

(14)

discussed in this research. At the moment the second problem is more perceived by the suppliers than with Large company itself, although Large company is searching for a better method to translate prices.

Validation of research

Bargaining power is one of the five competitive forces Porter (1980) identifies. Porter identifies a number of characteristics that influence the bargaining power of a buyer, which can be found in figure 2.2.

Figure 2.2 Bargaining power

Information on the supply market leavers is identified as a characteristic that provides the buyer with power. Information about the suppliers’ levers can be of benefit to Large company for two reasons. The first regards the negotiations. When there is a better understanding of how the price is determined and how the price behaves plus what activities drive the price, Large company can get a better knowledge on the prices for this research, which will help with negotiations because prices can be negotiated faster and in more detail.

On the other hand there is the internal aspect. When there is a better understanding of the suppliers’ prices. CMI can adapt its ordering behaviour accordingly to achieve savings. Concluded can be that there is a lack of information on the suppliers’ leavers. This information is hard to obtain because services are hard to valuate. Besides in most markets there is little competition for the suppliers this reduces the opportunities to easily obtain information. A service is hard to quantify into useful information but is according to Porter (1980) an important driver for bargaining power.

Problem statement

This research aims to change the current situation of lack of knowledge on prices in Research. The widely used and accepted Diagnoses, Design, and Implementation (DDI) model by De Leeuw (2000) will be used to conduct this research. This model is widely accepted and used as a guideline for doing research that will lead to a desired future situation based on a diagnosis of the current situation. This research will focus on the diagnoses and design phase. The diagnoses phase is a translation of the problem situation into a management problem. The focus is explicitly on the allocation of the causes of the problem. The design phase is about transforming the allocated problem from the diagnose phase into a concrete solution.

(15)

Research objective:

Improve the knowledge on the prices of the purchased services for Before launch research in order to negotiate competitive prices.

According to (Agger,1913) Competitive price is the result of free competition and equals the cost of production. Besides actual outlays cost of production includes "normal profits," which may be considered to be of such an amount as to afford no "extra inducement to enter the business or leave it." This definition is applicable for how Large company approaches its suppliers today.

Research question:

What drives the prices for the different types of Before launch research and how can the information from the separate researches be used to get a better knowledge on the prices for Before launch research for the future?

To answer this question, constraints are determined to set the boundaries for this research.

Constraints

-This research will only focus on competitive prices for European countries since this is the region NPI SM Europe is directly responsible for.

-Because of the irregularity of purchasing and different characteristics, Type D research is excluded from this research.

-Because the responsibility for the procurement of Type E research is not for NPI SM Europe also this research is excluded from this research.

-The markets the suppliers operate in are seen as a fixed factor in this research. There are no indications that the situation in the markets will change within a short period of time. The first constraint means this research will focus on the European countries which are present in at least two out of the three rate cards and/or are of high importance for Large company. The countries in the rate cards represent the countries where research is ordered most often. Below you can find the countries that fit the constraints and therefore will be included in this research. It will be interesting to see if, in similar cases, processes and the related prices differ between these countries.

1.Czech Republic 2.Denmark 3.France 4.Germany 5.Hungary 6.Ireland 7.Italy 8.Netherlands 9.Poland 10.Russia 11.Spain 12.Sweden 13.United Kingdom

(16)

The processes will be influenced by the variables set by CMI. But what is exactly meant by these variables? These variables define the constraints for how the research should be done. Examples of variables are number of products, incidence rate of the target group, and number of open questions. These variables will be further described in the next Chapter. Depending on how CMI determines the variables, activities the supplier executes and as a result the price for a research is influenced.

Subquestions:

The sub questions for this research are just like the main structure for this research designed according to the DDI model of De Leeuw (2000):

Diagnoses:

What are the main components of the price for before launch research?

What influence do the variables determined by CMI (price drivers) have on the price drivers? To what extend are differences in prices between types of research caused by difference in data collection method?

How is the price influenced by the country the research is conducted in?

What activities are the price drivers influenced by? To what extend is there a difference between the countries?

Design:

How can the current tools be adapted/extended to get a better understanding of the suppliers’ prices?

The sub-questions are focused to get a better knowledge on the several parts the total price is constructed from. Since the suppliers in Research face little direct competition it is hard to conduct a valid research about their prices opposed to their peers. By fragmenting the total price in smaller pieces and analyse these and compare them to the prices for similar activities in other types of research, Large company will get a better understanding. The knowledge on the small parts of the prices will help to speed up negotiations and will be useful to negotiate better prices.

Output

The result of this research should be a tool that improves the understanding for Large company on the prices of Research. The data that is the outcome of this research has to answer to several criteria set by Baarda en De Goede (2000).

-Reliability: undependable on chance or coincidence.

-Validity: extend to which you actually measure what you should.

-Internal validity: Quality of the conclusion, are the right relations assigned?

Below is continued with the research structure, which will lead to the desired output.

Research structure

Below the structure of the research plan can be found. It shows the steps the research will be conducted in. The guideline for the structure is Diagnoses, Design, and implementation model.

(17)

Figure 2.3 Research structure

Conclusion

In this Chapter the guidelines for the research have been set. Two main problems have been identified and validated which lead to the objective and a central research question. This research question will be explored and answered in the remainder of this research guided by the sub questions and research structure above taking all constraints into account.

Diagnoses Research Setup Ch 1: Introduction Ch 2: Research design Design Ch 7: Conclusion Ch 8: Design

Ch 9: Conclusions and recommendations Ch 4: Identification of main price drivers

Ch 3: Methods

Ch 5: Quantitative analysis of the influence of variables and country on price driver

Ch 6: Qualitative analysis of activities price drivers

(18)

Chapter 3 Methods

This Chapter is about the methods and techniques that will be used to conduct the research. A statistical data analysis will be used to have a look at the rate cards of the different types of research and the relation between the rate cards. For the analysis of the processes that relate to the rate card an activity analysis will be executed.

Approach

The method that De Leeuw (1996) calls the ‘De Ballentent’ (Ball Tent) is used to describe the methods and techniques used in this research. This method describes the research approach as congruence decisions over all the balls in the Ball Tent, figure 3.1.

Figure 3.1 ‘Ballentent’, De Leeuw

The choices made in this research and their relations are described according to this figure. The balls Research Question and Used Concepts are already described extensively in Chapter two and will not be described in this Chapter.

Data resources

Six types of data resources are distinguished by De Leeuw (1996): Documents, media, reality, model of reality, databases and the experience of the researcher. The data used in this research are empirical. The data used in this research are partly extracted from contracts with suppliers for Research. These contracts contain rate cards that contain the prices negotiated by NPI Supply Management Europe. Since these are real life prices which are seriously negotiated by Large company’s strategic sourcing strategy, therefore not influenced by coincidence, the data are considered reliable.

The qualitative data used in this research comes from expert interviews and observations of the researcher. Therefore the data collected can be subject to interpersonal appraiser reliability (Baarda en de Goede, 2001). To optimise the interpersonal appraiser reliability two people will be present when the data is collected. Data collection for both observations and interviews will be structured (Collecting data where the method of questioning and/or observing is determined before hand, Baarda en de Goede, 2001).

(19)

Measurement and observation methods

The data from the contracts are not direct applicable information for research. Data can become information when data is processed (van Beek, e.a., 2000). The rate cards used for this research do not all have the same structure and the data in the rate cards need to be aligned to make a valid comparison. The data from the separate rate cards are processed in order to get valuable information for the research, this will happen in the next Chapter. The price of a research is influenced by the variables that the customer, Large company (CMI), sets that will determine the characteristics of the research asked for. According to De Leeuw (1996) the definition of a variable can be split in two: the conceptual definition and the operational definition. The conceptual definition tells what is measured and what the goal of the measurement is. The operational definition is about how the measuring is actually done. The definitions in table 3.1 are used in the quantitative analysis on the variables used by CMI in Chapter 5 and 6.

Name of variable Conceptual definition Operational definition

Respondents Average number of people that

participate in a research.

# of respondents that have to participate in the research.

Incidence rate Part of the total population that should participate in the research.

% target group of the total population.

Boost People needed extra than

naturally would be in a target group with n respondents in order to get n number of respondents in the sample.

% of extra respondents out of the total number of respondents in the sample size.

Cell Group of respondents testing one

product.

# of groups testing one product. Interview length Length an interview should have. Required interview length in minutes.

Products Number of products included in a

research.

# of products included in the research.

Location Where the field work is done,

central location or in-home.

Is the research done at a central location or in-home.

Monadic/sequential The sequence of testing the

products.

Are the products in the research tested at the same moment (monadic) or after each other (sequential).

Legs Number of separate moments of

collecting data.

Number of extra moments than regular for data collection.

Questions The number of questions that are

in a research.

# of questions in the research.

Open questions Number of open questions that

are in a research.

# of open questions in the research. Data collection method How is the data collected in the

fieldwork?

Internet/ telephone/pen and paper etc.

Table 3.1 Definition of variable used by CMI

Analytical methods

In this research several quantitative analysis will be used in order to support the qualitative research. For the quantitative analysis, statistics will be used. In order to speed the process software programs like MS Excel and SPSS for windows will be used to support the calculations.

To look for reasoning behind pricing mechanisms, correlations and difference tests are used. Correlation analysis is used to show relationships between variables. The variable we think

(20)

of as the causative agent is the independent variable, and the variable that is used to measure the effect, the outcome variable, is called the dependent variable (Corty, 2007). Ordinarily, the independent variable is thought to influence the dependent variable, rather than the other way around (Freedman e.a., 1998).

When both variables are measured at the interval or ratio level, then the correlation of first choice is the Pearson product moment correlation coefficient, commonly called the Pearson r, or r_P(Carter and Williamson, 1996):

y x xy p

S

r

=

₍₁₎

Where: S_xy = The co-variance of the sample

S_x = Standard error for variable x

S_y = Standard error for variable y

The value of r is between -1 and +1 where -1 is a perfect negative correlation and +1 is a perfect positive correlation. Where r = 0 there is no relation between variable x and y. There are some assumptions that have to be taken into account when using Pearson product moment correlation coefficient (Corty, 2007).

-Both x and y are interval- or ratio- level variables. -The relationship between x and y is linear.

These two assumptions can not be violated at any time, there are three other assumptions that can be violated under certain conditions.

-The sample is a random sample from the population.

-Both x and y are normally distributed (can be violated if N is large enough). -The variables display homoscedasticity (can be violated if N is large enough).

If both variables are ordinal, or one is ordinal and the other interval or ratio, then the correlation of first choice is the Spearman rank order correlation coefficient, commonly called Spearman’s r or

r

_S:

n

d

r

s

=

−

∑

3 2

6

1

₍₂₎ Where: i i Y X

rank

d

2

=

−

n = the number of paired scores r = the correlation coefficient

For r_s the outcomes and of the formula are the same as for r_p, -1 is perfect negative correlation; +1 is perfect positive correlation; 0 is no correlation. Though there is a difference in assumptions:

-Ordinal or interval data is required to rank the data.

-Although r_s does not assume a linear relationship between the two variables, it does assume that the direction of the relationship is consistent (either rising or falling).

(21)

In order to find differences over types of research and countries a difference test will be executed. Assuming that the prices for Type B are competitive, two-sample difference tests can be done to find differences over the types of research. Because the selection process for elements in the samples is influenced by each other the relation between the different types of research is calculated by a Dependent-sample t test. This test is designed to determine the difference between two dependent samples. Further research into multi-sample difference test made clear that there is a better way to compare countries: the

Repeated-measures ANOVA (analysis of variance) test. This is because the

dependent-samples t test does not partition the total variability into its components, and as a result it ends up overestimating the percentage of variability in the dependent variable that is predicted by the grouping variable (Corty, 2007).

To track down the differences in pricing over countries the Repeated-measures ANOVA test is used. This test searches for significant difference in a range of samples. Before you are able to use it you have to commit to a couple of assumptions:

-The samples are dependent

-The level of measurement of the dependent variable is on interval- or ratio- level. -Random samples (can be violated, but has to be considered)

-The variables are normally distributed (can be violated if N is large enough).

-Sphericity (if this assumption is violated, the F ratio for the repeated-measures ANOVA can be corrected).

The sphericity assumption holds that the variances of the difference scores for all possible pair combinations are equal.

residual group

MS MS

F = (3)

Where: F = the affect of the grouping variable

MS_group = mean squares group MS_residual = mean squares residual

group group group df SS MS = and residual residual residual

df

SS

MS

=

(4)

Where: SS_... = sum of squares

df_... = degrees of freedom

1

− =k

df_group and df_residual =(n−1)(k−1) (5)

Where: k = number of groups

n = sample size per group

In the repeated-measures ANOVA test it is of importance that variability caused by variability in either sample or group is not taken into account. Residual variability is the equivalent to the variability that is left after the other sources of variability have been removed (Corty, 2007).

To see if the value found for F is significantly different the following decision rule is in place:F_observed ≥F_cv. F_cv(critical value) can be found in the table in Appendix B. The value of

(22)

cv

F is determined by the degrees of freedom of the numerator and the denominator. The degrees of freedom of the numerator is k -1, where k is the number of groups. The degrees of freedom of the denominator is calculated by (n – 1) (k – 1) where n is the sample size per group.

When there is a significant difference found with the formula above it tells us that at least two out of the means being compared are statistically different. Now we should determine an overall effect size, and find out which means have a significant difference.

To determine the variance in the dependent variable predicted by the variable, for a repeated-measures ANOVA the following formula is used:

residual total residual group MS SS MS k SS + − − = ( 1) 2

ω

(6)

Where:

ω

2= omega squared; the percentage of variance in the dependent

variable that is predicted by the independent variable

SS_group = Sum of squares for treatment

k = number of groups being compared

MS_residual = mean square residual SS_total = sum of squares total

2

ω

provides an estimate, for the population, of the percentage of variance accounted for by

the independent variable. To find an estimate of the degree of relationship between the grouping variable and the dependent variable is found by taking the square root of

2

ω

(Corty, 2007).

Now we are going to find out which pair(s) of means are statistically different. In order to find this difference the Fisher LSD (least significant difference) test is used. It finds the minimum amount by which two means must differ to conclude that the difference is significant at the

α

level selected.

n MS t

LSD

=

_cv

2

residual (7)

Where: LSD= Fischer LSD (least significant difference)

cv

t = the critical value of t, two tailed, with df = df_residual and

α

set at the desired level (can be found in appendix X)

MS_residual = mean square residual

n= sample size per group in the repeated measures ANOVA

After having calculated the LSD it is easy to calculate the confidence interval between two groups with the following equation:

LSD

M

CI

=

₁

−

₂

±

%

95

(8)

Where:

M

₁ = mean of one group

(23)

Activity Cost Pool

For the qualitative analysis the activities executed, that together make the process behind the price, are analysed. To compare the activities and their behaviour the principles of Product costing will be used in this research. Product costing is the process of measuring and assigning to products and services the costs of the activities performed to design and produce them (Atkinson, Banker, Kaplan and Young, 2001). Besides assigning the cost, the behaviour of the costs over different values of the variables is performed research on. Therefore the principle of the activity cost pool is used. Activity cost pool (Garrison, Noreen, Seal, 2003) is a bucket in which costs are accumulated that relate to a single activity in the Activity Based Costing system. Every variable can be seen as a cost pool. The goal of this analysis is to see how a change in the variable, changes the activities in the cost pool. Later in this research the behaviour of the activities is compared to the price behaviour.

Conclusion

In this Chapter the theoretical framework is given that will be used to solve the research question given in Chapter two. Besides the data resources, measurement- and observation- methods that will be used, there is special attention for the statistical analysis that will be used to indicate irregular behaviour in rate cards. For the qualitative analysis an activity analysis will be executed.

(24)

Chapter 4 Main Costs

In this Chapter the focus area for the rest of the research will be identified. With several analyses the importance of components of the price for the total price of Research will be identified.

Rate card breakdown

The quantitative data used in this research is subtracted from the contracts Large company has with its suppliers. The lay-out and therefore the information the rate card in the contracts contain, are not aligned. These will first be aligned before any calculations will be made. The prices shown in the rate cards are shown point by point below.

Type A:

• Fixed fieldwork costs: The set up costs for fieldwork independent on the number of

interviews (e.g. supervision).

• Costs per interview: The costs per interview based on the total interview length, the

incidence rate, multiplied with the total number of respondents.

• Cost of coding: The costs for the creation of the open end question code frame and

the costs for the coding itself.

• Additional leg: The costs of adding an additional product to the interview

• Indirect costs: The level of complexity of a study determines these costs. The

complexity is related to the number of products and the length of interview per product. All the costs that do not relate to any of the previous price drivers.

Type B:

• Fieldwork costs: All the costs related to fieldwork.

• Data processing: The costs for processing the data collected in the fieldwork

• Client servicing: All the cost that do not relate to fieldwork or data processing

• Picture Sorting: The costs for this special add- on service

Type C:

• Everything is in one price

Below you can find a schematic overview of the information available in the contracts:

Type A Type B Type C

Fixed Fieldwork Costs Costs per Interview

Indirect Costs Client Servicing

Coding Data Processing

Fieldwork Costs

1 Price Table 4.1 rate card break-down

In order to make calculations and comparisons it is necessary to combine the costs per interview and fixed fieldwork costs in Type B. This way the relative influence of fieldwork costs on the total price can be compared with the fieldwork costs in Type B.

In the rate card for Type B there can be found that for the preferred supplier in Type B, data processing is part of the of the indirect costs. In the cost breakdown of the preferred supplier for Type B, coding is part of data processing process. This means that indirect costs and coding for Type B and client servicing and data processing costs for Type B can be combined to make a valid comparison on total costs and the influence of fieldwork costs on it.

(25)

Type A Type B Type C Fieldwork Costs

1 Price Fieldowrk Costs

Other Costs Other Costs

Table 4.2 Rate cards compared

Costs

In order to say something about the influence of fieldwork costs on the total price, the relative influence on the total price will be calculated as well as the correlation between the fieldwork costs and the total price. The correlation is calculated in order to see if an increase in fieldwork costs will lead to an increase in the total price. Alongside these calculations, correlations will be checked over the different countries. Correlation over countries will extend the possibilities for calculations in the remainder of this research. If there is a significant correlation, the prices for one country can be predicted by the price for another country.

First the relative influence of the price for fieldwork on the total price will be calculated. The rate card for Type A consists out of 4 types of research: central location monadic; in home monadic; central location sequential monadic; in home sequential monadic. Where central location/in home determines the location where the fieldwork is conducted and where monadic/sequential monadic determines the number of stages in time that the fieldwork is conducted in. To be exclusive all types of Type A research are in the calculations. In order to get most valuable data for Large company most ordered researches are taken for the sample.

The sample size for Type B will be the entire population.

In the rate card for Type C there is only one price mentioned and no difference is made between fieldwork costs and other costs like data processing or client servicing. Therefore no information is available for this type of calculation.

As can be found in table 4.3 below there is a big difference in the influence the fieldwork costs have on the total price between Type A and Type B. Though for both types of research can be said that fieldwork costs are a big part of the total costs.

FWC % of Total costs

TYPE A TYPE B TYPE C

Table 4.3 The average influence of the price for fieldwork on the total price per type of research.

The big influence of fieldwork costs on the total price is an incentive to focus on the fieldwork costs. To see if there is a direct influence of fieldwork costs on the total price the correlation will be calculated.

In order to calculate the correlation over countries the Pearson product moment correlation coefficient will be used when the samples commit to the assumptions for this tool. The first assumption that can not be violated is that both the x and the y variable are interval- or ratio- level variables. The fieldwork costs and the total costs are both on ratio level. The second assumption is that the relationship between x and y is linear. A linear relationship is one that can be described by a straight line (Corty, 2007). In this sample the normality assumption is

(26)

violated but the Pearson product moment correlation coefficient is robust to this violation because N is large enough.

Confidential

Figure 4.1 Correlation between the price for fieldwork and the total price for Type B

As can be found from the scatter plot in figure 4.1 there is a linear relationship between X and X in TYPE A. The Pearson product moment correlation coefficient over all countries is

p

r =

** Correlation is significant at the 0.01 level (2-tailed).

Table 4.4 Correlation Type B

The sample used is only part of the total population. To find the 95% confidence interval for the correlation between fieldwork costs and total costs the Fisher’s z transformation for r is used. The

z

_r values for the correlation: are X and X.

r r z z s CI r 1.96 % 95 = ± (1) 3 1 − = N s_r (2)

Where s_r= the standard error of the sampling distribution

With a confidence interval of 95% that the correlation between the fieldwork costs and the total costs for Type A is between X and X. With a correlation of X can be concluded that there is a relatively strong positive relationship between fieldwork costs and the total costs for Type A. This correlation is not strong enough to base prospects for total prices on. Within countries the correlations are stronger, see table 4.5. Because of the smaller sample size the Spearman rank order correlation coefficient is used to calculate the correlation within countries for the price for fieldwork and the total price. All the correlations found are significant on a 0.01 level, but for the Netherlands and Spain (see Appendix A).

Country Correlation France Germany Hungary Italy Netherlands Poland Russia

(27)

Spain UK

Table 4.5 Correlations for the different countries in Type B between fieldwork costs and total costs.

From the above can be concluded that in Type B there is correlation found over countries. There is a strong correlation found within countries between the price for fieldwork and the total price which indicated that the price for fieldwork has a strong influence on the total price.

When we look at linearity for Type B, figure 4.2, three groups can be distinguished from the scatter plot. First the group of outliers in the top left attract attention. This group represents the values for Ireland. The second group that is different from most values are the values for the countries Czech Republic, Russia, Poland and Hungary. These countries distinguish themselves by mostly by low total price, the influence of the fieldwork costs on the total price differs per country.

Confidential

Figure 4.2 Correlation between the price for fieldwork and the total price for Type B

That leaves us to the main group of values that appear to share a linear relationship. The Pearson product moment correlation coefficient for all the countries in Type B is.

FWC Price

Total Price

Table 4.6 Correlation Type B

It is interesting to look at the correlation of the main group of values in figure 4.3. When we look at this graph we find a group that has a Pearson product moment correlation coefficient of X. The group that is left are the countries: Denmark, France, Germany, Italy, The Netherlands, Spain, Sweden and the United Kingdom.

FWC Price

Total Price

Table 4.7 Correlation Type B without outliers

Figure 4.3 Correlation between the price for fieldwork and the total price for Type B without outliers

(28)

The correlation between the price for fieldwork and the total price for Type B research in the different countries can be found in table 4.8 (more detailed information can be found in appendix A). The correlations found are all significant at the 0.01 level except for the Czech Republic where the correlation is significant at the 0.05 level (2-tailed). Just like in Type A can be found that there is a strong relation between the price for fieldwork and the total price. Country Correlation Czech rep. Denmark France Germany Hungary Ireland Italy Netherlands Poland Russia Spain Sweden UK

Table 4.8: Correlations between the price for fieldwork and the total price for the different countries in Type B testing.

Conclusion

In this Chapter price for fieldwork is identified as focus area for calculations and comparisons over researches. The price for fieldwork is a substantial part of the total price for Type A as well as for Type B. Besides, the price for fieldwork has a direct influence on the total price, proven by the correlations. Because of the substantial part of the total price and the direct influence it has on the total price, fieldwork costs will be the focus for the remainder of this research. Savings made in this part of the total price can deliver overall savings.

In the calculations on correlation over countries there is found relatively strong correlations for both Type A and (subgroups of) Type B. These correlations are not strong enough though to use them for future calculations.

(29)

Chapter 5 Quantitative Analysis

Now is concluded that fieldwork costs make a substantial part of the total costs for Large company suppliers and that there is a direct relationship between the price for fieldwork and the total price we go more into dept to see what influences the price for fieldwork and how. In this Chapter the influence of variables and the behaviour of the variable over different countries and types of research are investigated.

In the three rate cards for Type A, Type B and Type C there are two variables defined that directly influence the cost for fieldwork: incidence rate and length of interview. The variables will be discussed over the different types of researches. But first the data collection methods used will be discussed since they determine the characteristics of the process to great extend and are therefore expected have a big influence on the fixed- as well as variable price for fieldwork.

Data collection methods

To do a valid research on the pricing for Research the samples that are compared have to be the same. What probably influences the price for fieldwork considerably is the data collection method: in what way the data is collected and how the collected data is handled. In the table below the different methods used in the market are explained.

PAPI Paper Aided Personal Interview

The interviewer collects the data and fills in the data on paper.

PASI Paper Aided Self Interview

The interviewee fills in the answers on paper

CAPI Computer Aided Personal Interview

The interviewer collects the data and fills in the collected data directly into the computer system.

CASI Computer Aided Self Interview

The interviewee fills in the answers directly into the computer system.

CATI Computer Aided Telephone Interview

The interviewer collects the data by telephone and fills in the data directly into the computer system.

CAWI Computer Aided Web Interview

The interviewee is asked questions over the internet and fills in the answers directly into the (personal) computer. Table 5.1 Description of data collection methods

Because in this research the focus is on fieldwork we are mainly interested in the last two letters of the abbreviations and since they are expected to bring the most costs with them. In table 5.2 can be found which data collection method is used combined with type of research in what country. There are a lot of similarities between the different types of research based on data collection method. To maximize the validation of this research only the prices of the exact same data collection methods will be compared.

(30)

Confidential

Table 5.2 Data collection methods assigned to research and country

In the quote for the rate card is asked to use the data collection method that is most commonly used for that country. In table 5.2 can be seen which methods the suppliers choose for that. In table 5.3 the prices for samples that are standard for the type of research are taken to compare whether the location and data collection method of a research make a difference.

When we look at the prices for standard fieldwork in terms of incidence rate and interview length we find that for Type A in the western countries (Germany, The Netherlands, United Kingdom and Russia), in-home research is more expensive then central location research. For Central and Eastern European countries and Spain it is the other way around. In Type B research we find that for all countries that make a distinction between central and in-home research, in home research is cheaper, see table 5.3.

There has to be taken into account that there is no data available for in-home interview for Central and Eastern European (CEE) countries. For Type c there is no distinction made between in-home and central only between CAWI and the others. CAWI is found a cheaper data collection method then the others.

For Type B in-home is consistently cheaper, while for Type A there are differences. In table 5.3 is found that the data collection method used for in-home location in Type B is CAWI which is consistently cheaper then CASI or CAPI. In Type A it is the other way around. CAWI is used in Germany and the UK opposed to respectively CAPI and PASI. It is interesting to take a deeper look into this issue in the qualitative analysis.

Confidential

Table 5.3 Data collection comparison

Correlation

Incidence rate

In order to calculate the influence of incidence rate on the price for fieldwork the correlation

will be calculated. To see if there is a direct relation to a lower incidence rate and a higher price for fieldwork the Spearman product moment correlation coefficient is used. In the rate

card an interval is used to price the different situations. The assumption that ordinal or interval data is required to rank the data is not violated. The table below is used in Type B.

Incidence Rate 10, up to 15% Over 15, up to 20% Over 20, up to 30% Over 30, up to 40% Over 40, up to 50% Over 50, up to 60% Over 60, up to 70% Over 70, up to 80% Over 80% 1 2 3 4 5 6 7 8 9

Table 5.4 Ordinal values incidence rate Type A

For all researches in all countries the Spearman product moment correlation coefficient is X, except for Germany where it is X. This is caused by the fact that for Germany there are the same prices that account for multiple incidence rates. All rates are on a 0.01 significance scale.

(31)

Incidence rate 30 up to 50% 50 up to 75% 75%+ 1 2 3

Table 5.5 Ordinal values incidence rate for Type B

Here as well are almost everywhere perfect negative correlations, X, found except for 25 and 30 minute lasting CAWI interviews in The Netherlands where a correlation of X is found with a x significance interval. In Russia a correlation of X with a X significance is found. The most interesting outcome is that in Poland there is no correlation between incidence rate and price at all.

For Type C there are values of incidence rate given for Company X 2 research. This is a special research within Type C. Here is a perfect negative correlation of -1 is found as well. Interview length

For interview length the correlation is calculated in the same way as for incidence rate. For Type A is found that except for researches in Germany, Italy (in-home) and for 4 values of incidence rates in in-home research in the United Kingdom there is a perfect correlation of 1 between interview length and price for fieldwork.

For Type B the values found are less valid because the sample size per group (defined by country, location and incidence rate) is limited by three. Therefore we have to be careful with taking conclusions on the information found with this data. For most countries there is a direct relation between an increase in interview length and price for fieldwork (appendix C). For the following samples that is not the case:

Czech Republic Hungary Ireland Netherlands (central location) Poland Russia

For the countries above accounts that the Spearman product moment correlation coefficient is X (X significance level). This means that the price for fieldwork is the same for 2 out of the 3 interview length options.

In Type C there is no information on the influence interview length has on the price since these researches are more standard.

From the calculations above can be concluded that incidence rate and interview length have a significant influence on the price for fieldwork with Large company preferred suppliers. Incidence rate and interview length are here identified as main price drivers that directly relate to the input Large company gives to the suppliers. For Type B the interval chosen for interview length might be too small since many countries filled in the same price for at least two out of three variables. It is interesting to take a deeper look into these variables to see how they behave within and over different countries in the qualitative analysis.

Getting a Grip Understanding and Controlling Suppliers Prices in Three Steps

Getting a Grip

Understanding and Controlling Suppliers Prices in Three Steps

Getting a Grip

Understanding and Controlling Suppliers Prices in Three Steps

Preface

Executive summary

Table of contents

α

µ

Chapter 1 Organisation and processes

O

V

E

R

V

I

E

W

Chapter 2 Research Design

Chapter 3 Methods

S

S

S

r

=

r

n

n

d

r

=

−

−

∑

6

1

rank

rank

d

=

−

df

SS

MS

=

ω

ω

ω

ω

α

=

2

α

LSD

M

M

CI

=

−

±

%

95

M

Chapter 4 Main Costs

z

Chapter 5 Quantitative Analysis