UNIVERSITY OF GRONINGEN
Antecedence of
Administrative Tape
Master thesis
S1655175 Wojciech Stefan Zieliński
2011-07-20
Supervised by: Gjalt de Jong
Compliance with various national rules and regulations has become part of standard operations for many private companies regardless of whether those rules and regulations have a positive or negative impact on private business. The core focus of this thesis is the discussion concerning red tape, which represents the negative effects of rules and regulations. The impact of company-specific characteristics, such as size and age, together with the government’s quality of service on the costs of compliance with national regulations that these companies suffer from is estimated with the use of Norwegian data obtained from the 2001 OECD report titled, “Businesses' Views on Red Tape:
Administrative and Regulatory Burdens on Small and Medium-sized Enterprises”. The results resolve the ambiguities concerning company size and prove the significant effects that the age of a company and governmental service quality have on perceptions of administrative tape.
1
1. IntroductionDuring the past centuries all modern societies have built vast legal systems that are rule- oriented and accompanied by various legal institutions (van Witteloostuijn and de Jong 2008). Well- functioning national legal systems became essential for a democratic society (North 2005).
Appropriate regulations are indispensable to achieve economic, social and environmental objectives (Acemoglu, Johnson, and Robinson 2000). Furthermore, rules help to protect property rights of employees, employers, investors and consumers (Giannetti and Koskinen 2010; Atanassov and Kim 2009; La Porta et al. 2000; Laporta et al. 1998). The status quo of the legal system is constantly challenged, which results in a dynamic nature as new rules are introduced or old ones are either changed or replaced, in order to fulfil new needs and meet new requirements (Maltzman and Shipan 2008). The dynamic nature of the legal system is welcomed as it helps bring about change towards new situations and improvement to the efficiency of the entire system. On the other hand, a dynamically changing system may lead to instabilities. Therefore, a legal system that offers both appropriate adjustments and stability is the foundation of social development and growth (van Witteloostuijn and de Jong 2011).
The previous section strived to emphasise the impact that the legal system bares on the economy and all stakeholders; whereas this section will try to investigate the impact of rules and regulations. Olsen (2006) and March and Olsen (1989) prove that rules have both positive and negative effects. Regulations that yield positive influences are viewed as administrative “green” tape (DeHart-Davis 2009). Positive effects of rules and proper institutions can be observed in the increase of economic development (Acemoglu, Johnson, and Robinson 2000; Hall and Jones 1999). Regulations strengthen property rights, which leads to innovation (Gould and Gruben 1996) and improves the functioning of the financial markets by ensuring higher levels of investment (Barth, Caprio Jr, and Levine 2001). Rules also help to organize life, by clarifying relationships and responsibilities between various stakeholders1 and providing means to protect themselves in case of any abuse (Adler and Borys 1996). Moreover, people’s safety is enhanced, because companies have to apply safety protocols, which results in both employees’ better work conditions and final consumer reassurance of product quality. Transparency is another positive aspect as end users have better understanding of products that they are buying. Even on the level of government and private business regulations may lead to efficiency as Blau and Meyer (1971) claim that requirements of filling precise forms are more efficient than permitting companies to provide the information according to their liking, as not all information might be delivered or it can be delivered in a very difficult format to process. However, mostly, administrative tape is known for its negative side in which case it is referred to as administrative “red”
tape. Rules that are too restrictive or regulations overblown in terms of size may hamper economic growth. For every issue there are procedures, which are not always the easiest to follow and may require a large amount of time to comply with. For ordinary people, dealing with governmental
1 such as employer-employee, producer-consumer or lender -borrower.
2
institutions normally means a day off as most governmental offices are open during working hours. As far as business is concerned obtaining permits is costly, may imply long waiting periods and require several trips (Mauro 1995). In case of any mistake one might end up filing dozens of papers that are required to clarify it. Sometimes one faces several institutions with conflicting regulations, which may result in, at best, a longer period for obtaining a decision or, in the worst case scenario, no decision at all. In addition, the bureaucratic buffer may prevent many viable businesses to operate, including reductions in entrepreneurial enterprises (Osborne and Gaebler 1992). As a result, compliance costs may sky- rocket reducing GDP (Djankov, McLiesh, and Ramalho 2006).It is estimated that red tape reduces GDP by 3%(OECD 2001). Therefore, for many countries around the world reduction of red tape has become a priority (Boeheim et al. 2006; OECD 2001; Helm 2006; World-Bank 2006). Such initiatives as “Better-regulation programs” are being launched in order to reduce the stock pile of regulations, improve existing regulations and design new regulations better in the first place (Radaelli 2005). One of the cornerstones of “better regulations programs” is a method called regulatory impact assessment, which aims at designing, implementing and monitoring better regulations by taking into account higher number of variables and the assessment of the impact that the newly designed regulation bear not only on the intended domain but also on the broader perspective (Kirkpatrick and Parker 2007). The Netherlands is often perceived as a good example that is actively trying to reduce administrative burden (OECD 2004, 2006). There are several reasons why the Netherlands is chosen, such as the explicit targets that aim at reducing red tape by 25 percent by 2012, the development of Standard Cost Model (SCM) that measures administrative burden in a more subjective way, the independent advisory board Actal and adequate institutions, which include the interdepartmental taskforce aiming at increasing cooperation between the departments (OECD 2006).
The Dutch approach has been copied by various countries. For example EU has recommended for its members to reduce 25 percent of their regulatory burden. Although many countries have tried to counteract red tape, it is imperative to evaluate the efforts. OECD (2010) reports that to a large extent most of the programs have failed in their attempt to reduce red tape. Surprisingly, the studies regarding the evidence-based evaluations of governmental programs aiming at improving the situation are scarce (van Witteloostuijn and de Jong 2008). Only van Witteloostuijn and de Jong (2008,()2010), who constructed the count database of Dutch national rules in higher education between 1960 and 2004, show that the number of rules in this domain increased by more than 800 percent. This rapid increase in the 1990s and 2000s suggests that the better-regulated programs have not stopped the rule generation process. In addition, the analysis and measurement of administrative tape appears to be challenging.
The field that is predominantly associated with administrative tape is called public administration research (Pandey and Scott 2002). These studies focus on three areas: the measurement and the definition of the concept (Coursey and Pandey 2007; Pandey and Scott 2002;
Bozeman 2000), the differences in perception of red tape between public and private managers
3
(Baldwin 1990; Rainey, Pandey, and Bozeman 1995; Pandey and Kingsley 2000) and the perceived impact of red tape on an individual manager (DeHart-Davis and Pandey 2005). Although public administration studies look in depth into causes and results of red tape, they limit themselves only to intraorganizational2 red tape (Bozeman 2000).Another group of studies that pays attention to administrative tape is business impact studies.
These studies try to assess the administrative costs and their impact on the government and private sector (Walker and Brewer 2009; Brewer and Walker 2010; OECD 2001). In comparison to public administration studies, business impact studies focus on external control3 red tape, but they miss in- depth analysis of issues regarding administrative tape. Moreover, business impact studies are criticized by Kitching (2006) for not addressing questions regarding the reasons behind different perceptions of administrative costs. Inspired by the lack of appropriate treatment of the issue by other researchers this thesis aims to contribute to the discussion by investigating the antecedents of administrative tape in the private sector with a focus on the external control setting. This thesis will try to answer the research question of why companies in the private sector have a different perception of administrative tape, which they experience while complying with national rules and regulations.
The business impact literature limits itself only to the discussion concerning the size of the company as the antecedent of administrative tape (Kitching 2006; OECD 2001). However, even in this case the results are mixed and inconclusive. Therefore, the size of the company was chosen as the first antecedent that will be tested in this thesis. Another important company-specific quality is the age of the company and the experience that accumulates with it. Therefore, the second factor which will be addressed is the age of the company. The final issue expounded will be the quality of the service provided by governmental institutions.
In the next section, a literature review will be presented, where the issues concerning administrative tape will be elaborated upon. The discussion will cover the measures of relevant concepts and definitions, followed by the hypothesis formulation and the theoretical logic explanation.
Afterwards, the research methodology applied in this paper will be offered; in this section, data collection issues, measures used and model specifications will be presented. Subsequently, empirical results including robustness analysis will be shown. Finally, a conclusion with the evaluation will be offered. This section will also include a reflection on opportunities for future studies.
2 Intraorganizational red tape is the type of red tape that has originated from rules and regulations developed in this organization and it only has negative impact within this organization (Bozeman 2000).
3 External control red tape is the type of red tape that originates from rules and developed by external organizations but has negative impact only on the organization in question (Bozeman 2000).
4
2. Literature review and Hypothesis2.1.1 Red Tape research
The concept of red tape originated in 17th-century England where all the official documents were tied with reddish tape, which started to be associated in the 19th-century with “inaction or delay caused by official sluggishness” (Chandler and Plano 1982). In the 1970s, Buchanan (1975) and Kaufman (1977) first addressed the issue of administrative tape in the field of public administration.
Buchanan (1975) strived to prove that a market-based solution is not as efficient as it was claimed to be, by comparing red tape in public and private organizations; whereas Kaufman (1977) made the first attempt to describe red tape phenomena. However, these first attempts have been criticized for their lack of appropriate definitions and concept ambiguities by Bozeman (2000). Kaufman’s work was continued by Rosenfeld (1984) who, inspired by Kaufman’s model, tried to test it empirically with the emphasis on red tape in the area of intergovernmental programs and federalism. Rosenfeld (1984) is also the first author who provided the first proper definition of red tape. He defines red tape as all the governmental interventions, rules, forms and procedures that are perceived as difficult to control, excessive or useless during decision- making and implementation processes. In the 1990s the interest in red tape started to grow, which led to many publications in the field (Pandey and Scott 2002). At the start of the decade, Baldwin (1990), though reconciled with Buchannan’s research, provides contradictory results. In addition, he defines red tape as the constraints that employees have to struggle with while performing day-to-day activities. Moreover, in his study, a distinction between formal and informal red tape as well as internal and external red tape is made. Bozeman and Reed (1992) disagreed with the prevailing view that red tape should be regarded as only the number of formal rules and regulations. It was proposed that red tape should be defined as the delays that are caused by rules and regulations, which should include the irritation from stagnation. One problem with this classification is that it is hard to separate the delays caused by red tape from delays caused by standard operations. Bretschneider and Bozeman (1995) and Pandey and Bretschneider (1997) tried to solve the issue using more advanced measures.
The main breakthrough was made by Bozeman (1993;()2000) who developed key concepts and definitions of red tape. According to him, red tape is a negative phenomenon, which should be regarded as pathology in every organization. The possibility that red tape might have any positive effects was fully rejected. He defined red tape as “rules, regulations and procedures that remain in force to entail a compliance burden but do not serve the legitimate purpose the rules were intended to serve”(Bozeman 2000). Furthermore, he claims that the reasons for red tape occurrence are either rules that were born badly or good rules that have gone badly. Several causes for the occurrence of this phenomena are provided: poor understanding of the function and the impact that rules have by rule-makers; over-control, which is represented by increase in a formalization as a response to uncertainty; compromise and democracy, which implies fulfilling the needs of various stakeholders
5
that are taking part in the rule-making process. Furthermore, compromise and democracy may lead to:rule drift and entropy, changes in the rule implementation and its function, and rule misapplication.
The most recent studies have changed their focus. DeHart- Davise and Pandey (2005) have started the discussion, which is based on a super-micro level, where questions regarding the impact that red tape has on managers are addressed. On the other hand, DeHart- Davise (2009) is breaking with Bozeman’s tradition of defining red tap as only negative and tries to reopen the discussion regarding the positive effects of green tape. On the other hand, Walker and Brewer (2008) discovers that there are significant differences in the perception of red tape along the hierarchical structure.
Therefore, they conclude that red tape has been understated so far because mostly top managers’
perceptions were investigated in other research. These findings lead them to the further investigation of why lower management perceives red tape as more burdensome (Brewer and Walker 2010).
2.1.2. Concept measurement
In this section a review of concept measurement in public administration research will be illustrated. The most widely used method to obtain the measures of administrative tape in the field was questionnaires. These questionnaires were using Likert-scale predefined type questions to measure the perception of red tape (Pandey and Scott 2002). Using perceptions of administrative tape is strongly embedded in this field of studies and argued to be the best measure of red tape (Bozeman, Reed, and Scott 1992; Rosenfeld 1984; Rainey, Pandey, and Bozeman 1995). Bozeman, Reed, and Scott (1992) argue that by using perception one can, in addition to normal costs, account for intangible and hard to measure costs such as irritation and other emotions. This method is claimed to be a good proxy when it is used appropriately, which can be done thanks to the developments in public administration literature (van Witteloostuijn and de Jong 2011). However, this method is not free of flaws.
Gouldner (1952) claims that there are two phenomena involved in red tape: the first one is a perception of rules and regulations by an individual; the second is objective practice or patterns.
Pandey and Kingsley (2000) declare that the same rules and regulations will be differently assessed by different observers based on their emotional state, which may result in a subjective outcome based solely on the emotional state of the person in question. It is this lack of objectivity and precise estimation that induced Bozeman, Reed, and Scott (1992) to look for other more objective measures.
Using Likert-type questions might not reflect the true perceptions of costs and might be biased due to stereotypes or one’s emotional state (Pandey and Scott 2002).
Another method that is used in public administration studies is measuring the delays that are caused by rules and regulations in terms of time that the critical tasks require (Bozeman, Reed, and Scott 1992; Bretschnieder and Bozeman 1995). The strength of this method is its higher level of objectivity and better reflection of the true cost (Bozeman, Reed, and Scott 1992). The weakness of this method is that organizations prioritise some tasks, which results in quicker finalisation of prioritised
6
tasks due to more efforts and attention spent on them. This may lead to an inappropriate depiction of true costs (Bozeman, Reed, and Scott 1992); besides, some people within an organization are more careful and prefer to double check, which takes more time.Using formalization is regarded as another method of calculating red tape(Pandey and Kingsley 2000). Despite Bozeman’s (1992) early warnings that using formalization as red tape is a misconception, other studies refined Bozeman’s definition in order to increase the applicability of the concept in empirical research. Pandey (1995) and Pandey and Kingsley (2000) suggest using formalization as a primitive form of red tape. Red tape was defined by them as “impressions on the part of managers that formalization in the form of burdensome rules and regulations is detrimental to the organization." Pandey and Scott (2002) acknowledge the perspective that red tape exist when managers view formalization as burdensome as beneficial, because it simplifies the operationalization of the concept due to the fact that there is no longer a need for a detailed case study of every rule. On the other hand, Bozeman and Scott (1996) and Bozeman (2000) argue that high or low levels of formalization are neither good or bad, which is a serious flaw in this method.
Bozemma Reed and Scot (1992) appeal for using more measures because, according to them, red tape was a multidimensional concept. In answer to their call, Rainey, Pandey, and Bozeman (1995) tried to compare all constructs developed at that time, such as: global measure that implies direct Likert-type question regarding red tape, rule enforcement measure, personnel rule constraints, personnel delays and number of decision makers. Their mixed findings led them to the conclusion that administrative tape as a construct requires revision, better clarification and the applications of a larger variety of measures. Pandey and Scott (2002) have come to a similar conclusion, which resulted in an extensive literature review and the reconciliation of constructs. It is strongly recommended to use appropriate measures and clear definitions of them. However, the problem with both studies is that, in general, they have stayed within the tradition of Likert-type scale surveys. Only a handful of studies have decided to use a different approach. Bozeman and DeHart- Davise (1999) used an extensive and detailed case study in which every rule is looked into and assessed, which allowed them to construct an appropriate red tape construct. However, the problem with this method is that it requires lots of time. In addition, since it is mostly a case study approach where only a limited number of rules can be investigated, all the flows of the case study analysis apply here (Pedhazur and Schmelkin 1991). Yet another interesting approach was used by Scott and Pandey (2000). They designed an experimental treatment, where all the test groups were supposed to assign red tape and value it in accordance with the service provided. However, despite of all the efforts, Coursay and Pandey (2007) criticize other studies for not addressing “the difficult question as to exactly how the red tape construct is formulated from these indicators”. The Confirmatory Factor analysis is proposed as a superior method of testing the models.
In summary, most of the debate in public administration has focused on the definitions, concept measurement, formalization and the differences between public and private management (Coursey
7
and Pandey 2007; Pandey and Scott 2002; Rainey, Pandey, and Bozeman 1995; Scott and Pandey 2000).2.2. Business impact studies
Business impact studies is the second group of research that aims at addressing topics related to administrative tape (Nijsen et al. 2009; Allers 1994). More importantly, these studies provide a foundation, by laying the theoretical and empirical background for this thesis. The discussion regarding red tape becomes relevant because it was proven that in OECD countries red tape costs are 3-4 per cent of GDP (OECD 2001). On the other hand, studies like Capelleras, Mole et al. (2008) provide evidence that the effects of red tape are minimal. In addition, Business impact studies are criticized by Carter, Mason et al. (2009) for their lack of inclusion of the beneficial effects that regulations may entail. The reason for such ambiguity might be that the causal link between performance of private firms and administrative tape is complex and not always negative (Helm 2006). In addition, Helm (2006) blames for the variations between the country findings the lack of universal approach, which would imply the usage of the same measures for concepts. Despite some criticism, the fact remains that regulations may bear large and burdensome effects. Regardless of the question of who is more right or wrong, business impact studies have also contributed to the discussion in many other ways.
The first contribution concerns the development of classifications and definitions of costs that arise due to administrative tape. There are several different classifications and many different types of costs used. Probably the best overview is provided by the OECD (1997) report see box 1.
Box 1: Main categories of regulatory burdens.
Source: Adapted from OECD (1997), “Regulatory quality and public sector reform,” in OECD (1997)
There are two sectors, public and private, that suffer from administrative costs. The public sector is subject to administrative costs due to several reasons. First, rules and regulations have to be developed, which includes hours that officials spend on the projects and numerous consultations that
8
every rule has to go through. After rules are developed they have to be enforced, because otherwise nobody would comply with them. At the end, every rule has to be maintained, which entails up-keep costs. The private sector incurs costs of complying with rules and regulations. For businesses, the main category of costs is administrative compliance costs, which consist of administrative charges, taxes, long term structural costs. Although compliance costs are the main focus of many business studies like OECD (2001), there are also other costs acknowledged such as capital cost4, opportunity costs, and psychological costs5.This section will give a brief overview of the field and will be an introduction to the second contribution of business impact studies. Kitching (2006) classifies business impact studies on three categories, such as compliance costs studies, business burden studies and business decision making and competitiveness studies. Compliance costs studies are the most influential group of studies. Only the impact of administrative compliance costs is investigated. These costs are measured ex post and without any potential gains that regulation may bring to the firm. Although it is admitted that such potential benefits exist, the difficulty in measurement prevents them from inclusion. It is also crucial to emphasise that compliance costs studies are only interested in costs arising from legal requirements, which means that administrative costs that are incurred in the company internally and irrespectively from those imposed by government are not taken into account. The method that compliance cost studies use is the Standard Costs Model. For an example of such studies see KPMG (2006) or OECD (2001). Business burden studies measure perceptions of business owners of the encumbrance caused by red tape. The emphasis is put on the concepts of costs in total which are not fully represented by compliance costs only. Therefore, they stay within the tradition of public administration studies and use Likert-type questions in the quantitative survey (Kitching 2006). For studies focusing in this area see authors like Carter, Mason et al. (2006) and Cosh and Hughes (2010). The main difference between the two types of studies is the method that they use for the calculation the costs of red tape. Business decision making and competitiveness studies in contrast to previous studies respond to Helms’s (2006) criticism of the absence of in-depth investigation of complex relationships between firm performance and the impact of regulations by using a quantitative and qualitative approach. However, the problem that these studies struggle with is the limited focus on only several regulations and their impacts on the companies, which hinders the generalizability of the results. For an example of such research see Arrowsmith, Gilman et al. (2003) or Walsh (2008). The second comprehension concerns the measurement of red tape in terms of costs.
An overview of the field shows that there are two dominant methods that are used. SCM, including many variations of it, is the first approach (Wegrich 2009). This method is rather an accounting measure, where the amount of time spent on complying with a single obligation is calculated. Subsequently, the number of cases, which include the number of affected companies and
4special investments are required such as buying specific ICT.
5 Frustration caused by regulatory requirements.
9
frequency, are summed. The total value obtained from this summation is then referred to as the quantification of administrative costs for this single rule. Afterwards, the quantifications are added up for a particular domain and are viewed as an overall burden for this particular area. For the calculation of these values two techniques are used. First, one uses the precise stopwatch approach, which measures exactly the time spent on compliance with the single rule. The second technique uses interviews where the subjects are asked to assess the amount of time required to comply with regulations (SCM-Network 2005). However, values obtained from the second technique lose the main advantages of subjectivity. In this case, the values are considered the perceptions of costs, due to the relativity of time. Yet this can be viewed as an advantage because the emotional costs are then partially depicted (OECD 2001). Sometimes the total amount of time is multiplied with the hourly wage of an interviewee6, which helps when the researcher wants to add other costs as well (OECD 2001). Although it is argued that this method is relatively accurate, systematic and the least subjective, it is emphasised that it is not designed to extract the precise costs of administrative tape (Nijsen et al.2009). Therefore, SCM is regarded as a proxy, which resembles the costs of red tape. In addition, Keyworth (2006) and Malyshev (2006) criticize this approach claiming that information requirements lead to measurement errors. Regardless of the criticism, this method is still viewed as a very good approximation which is the least subjective and most systematic.
Supporters of the second approach claim that SCM lacks all the other hard-to-quantify costs (Godwin and Lawson 2009; Allers 1994). In addition, they state that companies cannot disentangle the company-specific costs from those incurred by regulations, which further hampers the reliably of SCM (Chittenden et al. 2005). Supporters of this line of study advocate the predefined Likert-scale questionnaires, which represent the actual and total cost of red tape, including those hard-to-identify and disentangle components. This method is also not free from criticism. As it stays within the tradition of public administration studies, the same flaws apply here as well. In addition, OECD (2001) reports that this method may not represent true costs, but the strategic behaviour of respondents, who are always seeking the possibility to relieve the burden. Edwards, Ram, and Black (2004) support this claim by proving that this method shows rather stereotypical views regarding red tape that do not resemble the true costs. Kitching (2006) reports that most of the studies that used this approach had very poor and insignificant results. Finally, regulatory reform units complain that even if rules are simplified, resulting in actual significant cost reduction, the level of perception may not change (Edwards, Ram, and Black 2004).
2.3. Hypothesis
The reason why there is such a large debate regarding administrative tape definitions and measurement is that this concept is multidimensional and causal links are hard to establish (Coursey and Pandey 2007; Helm 2006). Consequently, a proper definition and specification is required on
6 In many cases the wage is an average hourly value of the people that comply with the regulations.
10
behalf of the researcher (Pandey and Scott 2002). Although many times it was stressed that administrative tape yields both positive and negative results, this thesis will only take into consideration negative aspects. Therefore, from now on administrative tape is associated with red tape only. The proper definition of red tape will include all the costs that the economy suffers from the existence of rules and regulations. However, the aim of this paper is to investigate private business only, which results in the narrowing of the costs of red tape to those incurred by private companies.Despite the sheer will of the author of this thesis to include all the costs of red tape, the hardships of measurement make it impossible. That is why the proxies that will be used will mostly be limited to costs of compliance. Therefore, red tape is defined as all the costs of compliance that a company has to incur while complying with rules and regulations.
The size of the company in many economic and business studies is researched and it is always viewed as an important indicator, because smaller companies seem to reflect different characteristics than bigger companies (Schaper et al. 2009). Although often size might not be accounted for as a separate variable, the results might be anyway corrected for the size of the company. Size is so relevant that even in comparing countries, GDP is normally corrected and compared on a per capita basis. Moreover, paradigms like resource-based view use size as a proxy to measure the resources of companies, implying that companies that have more labour, land or capital are more resourceful and can utilize these resources more efficiently (Penrose 1959). The size of a company was considered a crucial antecedent of administrative tape and gained most recognition from Business Impact studies (Kitching 2006; OECD 2001; Carter, Mason, and Tagg 2009).
The research of the OECD (2001), Crain (2005), Mors, Bucher et al. (2004) and McGregor- Lowndes and Ryan (2009) shows that the costs of compliance with regulations are linear and regressive, which means that smaller companies incur higher costs in comparison to larger companies.
There are two main theoretical explanations provided that try to address this relationship. It is claimed that dealing with governmental administration can be regarded as a fixed cost for every company. Since large companies have a greater resource base, it enables them to utilise these fixed costs more efficiently (Carter and Jones-Evans 2006; Kitching 2006). Secondly, large companies have enough power to influence and lobby governmental organizations, which allows them to further reduce administrative costs (Berney and Owens 1985; Carter, Mason, and Tagg 2009). Taking into consideration these findings and acknowledging the strength of theoretical foundation, this research will test the negative linear relationship between the size of a company and administrative tape.
H1: The size of a company is linearly and negatively associated with the perception of red tape.
Kitching (2006) reports that the relationship between the size of a company and the perception of red tape may not be linear. Moreover, in many business impact studies the linear relationship is assumed and treated only as such. It is consequently imperative to account for the possibility of having a nonlinear relationship. Therefore, following the findings of the previously
11
mentioned literature, the second hypothesis will be testing the negative relationship between size and perception of administrative tape, which is nonlinear.H2. The size of a company is nonlinear and negatively associated with the perception of red tape.
On the other hand, authors like Poutziouris, Chittenden et al.(2003) and Lancaster, Ward et al.
(2003) find a concave relationship between the size of a company and red tape. These authors fully acknowledge the power of the theory provided in the previous section. It is suggested that small business owners may not perceive red tape costs as so burdensome, because they may not be fully aware of all the requirements that they have to fulfil (Edwards, Ram, and Black 2004; Blackburn and Hart 2003). In addition, small companies may experience smaller costs due to the existence of many governmental programs aimed at increasing small entrepreneurship, which aim to simplify regulations for them. Moreover, many laws with which large companies have to comply do not require compliance from small businesses, for example, the existence of labour unions or corporate laws.
Therefore, these authors claim is that, initially, there is some threshold for which small companies should perceive lower values of costs, which in turn should rise with the increase in size of the company. However, at one point, the curve should inflect and the costs of administrative tape should decrease.
H3. The size of a company is nonlinear and concave (∩) in its impact on the perception of red tape.
An ecology perspective considers the age of a company as another important company attribute leading to significant changes in company characteristics (Carroll 1983; Freeman, Carroll, and Hannan 1983). Furthermore, Glisson and Martin (1980) stress the importance of jointly using the age and size of a company in one model. Therefore, this thesis will investigate age as one of the antecedents of red tape. The explanation of the hypothesised impact of age on the perception of administrative tape will be drawn from a knowledge-based view, organizational learning perspective and life cycle models.
A knowledge-based view stresses the importance of knowledge as the cause of competitive advantage (Eisenhardt and Santos 2002). A company that is well-endowed with knowledge is able to reduce all costs and is most efficient (Lubit 2001). Whenever a company repetitively encounters a problem, routines are created to tackle it, which further increases the efficiency of the company (Grant 1996). The development of knowledge and routines takes time (Levitt and March 1988). In summary, from a knowledge-based view, we can conclude that knowledge and routines decrease costs, but to develop it time is required. Life cycle models show that formalizations and routines are developed when a company is aging (Robbins and Barnwell 2002).
Taking into considerations the above theory and Edwards, Ram, and Black (2004) finding, that the need to comply with rules and regulations results in the creation of routines in an organization that allow more efficiency in complying with these rules, it can be argue that older companies that had
12
time to adjust to rules and regulations7 should possess enough knowledge and develop enough routines to deal with them most efficiently.H4. The age of a company is negatively associated with the perception of red tape.
Because rules are written down they should have the same impact on companies with similar characteristics. However, as it was pointed out, the results vary (Helm 2006), which means that regardless of companies’ characteristics, the costs of perceptions may differ. Edwards, Ram, and Black (2004) find that rules and regulations on their own are not as burdensome as it seems. What makes them burdensome is how they are enforced. This suggests that government service is crucial in determining the costs of administrative tape. Since service is volatile by nature (Kotler and Amstrong 2008), it may be the reason for the differences in the perception of costs. This relationship is not properly addressed in the literature, partially because the view that government quality and the service that it provides is considered one of the dimensions of red tape (Edwards, Ram, and Black 2004). However, authors like van Witteloostuijn and de Jong (2011) focus on the dimensions of rules and regulations only. Moreover, many governments consider rule reduction as the main burden relief.
Better communication or government service is then considered as an extra measure, which aims at reducing the burden caused by rules and regulations. The final hypothesis will address the impact that the volatility of service may yield on the perceptions of administrative costs.
H5. Bad quality of governmental service is positively associated with the perception of red tape.
3. Research Methods
In this section the methods that will be used to estimate the model and the data on which the hypothesis will be tested are discussed. First, issues regarding data collection will be addressed. Then the description of the variables and the procedures used to extract them will be deliberated upon.
Finally, the model, including the method of estimation, will be determined and followed by the analysis of the required assumptions that these methods yield.
3.1. Data
The data used in this thesis research is obtained from the OECD database. All the information regarding the procedures used while gathering the data are described in the OECD’s report
“Businesses’ Views on Red Tape”(2001). Using data developed by someone else limits the spectrum of the definitions and the methods that can be used for the measurement of concepts. Moreover, all the issues regarding the data gathering process that the OECD database suffers from and that are
7 The fact that rules and regulations are changing is taken into consideration. However, it is very rare that the whole system would change dramatically and most often the core of rules and regulations stays not changed. Therefore especially in case of OECD countries where systems are stable it is assumed that core rules and regulations do not change, or if they do they are often communicated in advance so every company can adjust.
13
described in their report are present in this thesis as well. Therefore, in the following section, all the relevant information from this report will be expounded upon.The process of developing this survey took several stages. In the first stage, from June 1995 to December 1997, a questionnaire was developed under the supervision of the OECD Working Party on Regulatory Management and Reform. The pilot study was conducted in four countries (UK, Sweden, Mexico, and Canada). Improvements to the final questionnaire were incorporated based on the results of this pilot study. In the second stage, a survey was launched between April 1998 and March 1999 in 118 OECD countries. It was a large-scale business mail survey with a total of 7859 surveyed companies; the data gathered is a cross-section tape. In the third stage, the OECD report was created.
This survey was designed to address several issues. The first goal was to measure and compare direct regulatory and administrative compliance costs across different countries, policy areas and businesses. The second goal was to assess business perceptions of regulations’ quality and the quality of regulatory administration. The third goal was to evaluate aspects of indirect costs arising from employment regulation. The final aim was to achieve all the measures by using a standardised methodology. Moreover, three areas of regulations were chosen in this survey: tax, employment and environment. These particular three areas where chosen because previous studies have proven that these three areas are the most burdensome for business. Companies received a questionnaire only from one area, which means that there are different companies present in different areas. There are four sections present in each questionnaire. The first section is composed with specific questions concerning each company, such as firm size or industry. The second section contains questions regarding business perceptions of regulatory quality, the new regulation development process and the level of compliance. The questions regarding the private sector perceptions of the quality of administration service are included in the third section. Questions in sections two and three are of a Likert-type. The final fourth section addresses the issues of perceptions of compliance costs, which consists of both SCM and Likert-type questions.
The data set limits in this study‘s scope to only small- and medium-sized business. The criterion used was the number of employees, which ranged from 1 to 5009. The reason for this restriction is that in smaller companies the person who is answering the questionnaire should have a better understanding of how much time is spent in this company on complying with regulations, because this person is most likely the one who is responsible for dealing with rules. Whereas in larger companies where many divisions are present it is hard to give a definite answer on how much time different employees have spent on complying with regulations. The companies were also categorized in three groups on the basis of the number of employees, 1-19, 20-49 and 50-499, with the assumptions that in the small companies the owner or general manager deals with the regulatory issues, in medium-sized
8 Australia, Austria, Belgium, Finland, Iceland, Mexico, New Zealand, Norway, Portugal, Spain, Sweden.
9 There is though some inconsistency as the descriptive statistic reviled that there are larger companies.
14
companies a single employee is more likely to have all the relevant information and in the large companies the particular departments are better informed.Using a large-scale mail survey has both disadvantages and advantages. To begin with the advantages, authors claim that the possibility of a comparison of different businesses among different countries is the main benefit. Another advantage is the large random sample, because then the low response rates that might occur are not perceived as problematic, especially when the characteristics of non-respondents can be assumed comparable with the respondents’. In addition, the mail survey allows respondents to decide without time pressure when to answer the questionnaire and gives them more time to think about the appropriate answer. Although having many countries involved allows for better generalizability of the findings and cross comparisons of the countries involved, at the same time it creates issues of its own. The differences in the size of the surveys that are launched in each country are based on the budget that each country decided to spend on the project and, in some cases, resulted in a low number of observations. In addition, cultural differences may cause different responses to the same questions because the interpretation of similar questions might differ.
Moreover, the response rates may vary due to national attitudes towards surveys. A good example of this argument is that, due to cultural differences, the survey in Mexico and Spain was conducted by phone. Business cycles are said to have a large impact on opinion surveys because companies that are badly performing during the survey might have more negative response than otherwise. Some disadvantages are subject-oriented. The bias results might be attributed to the measure used, for example, in the survey where SCM is used, which aims to calculate exact costs of compliance with regulations. However, the answers in this survey are provided by people, who do not keep exact records of their time spent on complying with regulations, which results in a subjective perception of costs and is therefore not intended as a precise measure. The last issue mentioned is the disentanglement of compliance costs between those that are caused by regulations and those that companies incur for their own sake.
Before a more detailed analysis is conducted, the boundaries of this thesis have to be specified.
The first choice is to limit this thesis to one country, which will enable the homogenization of the sample by reducing the cultural difference impact and other inherently country-specific issues that might be present in the data. The country that this thesis will focus on is Norway. There are several reasons why this particular country has been chosen. Norway is known for good business conditions and has a regulatory framework that is ranked high in terms of efficiency and quality (Schwab 2010).
However, at the same time, the rigidity of a regulatory framework is perceived as the biggest obstacle in doing business by Norwegian companies. The composition of good business conditions and an excellent regulatory quality presents Norway as an interesting case to study. Another initial restriction that will be imposed is in the area of regulations. Initially, the model will be tested on the tax area of regulation. Afterwards, the approach is extended in a robustness analysis by changing the area of regulations to employment.
15
3.1.1. Non-response ratesNon-response takes place when respondents do not reply to the questionnaire sent to them. If the non-response is random and the data obtained still represents the population then only the sample size is smaller but the results are valid. However, in case where the non-response is caused by some factor, then the sample does not represent the populations, which results in biased outcomes. Since the data is used from secondary sources, any analysis that could prove that the non-response is not an issue cannot be applied by the author of this thesis. Nevertheless, such an analysis was conducted in the OECD (2001) report. That is why this thesis will rely entirely on the procedure used by the authors of this report. Firstly, a comment will be given on the general treatment of non-response rates in the report, because the procedure used was designed to treat the data set as a whole. Then, a more detailed description of non-response rates in Norway will be provided.
The total sample consists of 7.859 independent replies out of 22.544 surveys sent, which gives a total response rate of 37 percent. The difference between this type of study and a study based in one country is that response rates differ significantly per country. This may hamper the cross-country comparability, which is relevant for the procedure used in non-response analysis. In order to ensure the comparability of different countries, OECD (2001) used the Gallup France statistical protocol, which states that there have to be at least 100 observations for a given area of regulations. The results for every country in the sample are above this threshold. Given this, and the fact that this was a random sample, it is stated that the companies possess the same characteristics across the countries.
The next step after ensuring that the data is comparable was to verify whether omitting the non- respondents in the analysis will harm the interpretability of the results. The procedure used to assess the likelihood of the underlined issue was the follow up postcard addressed to the non-respondents.
They were asked to provide reasons for not replying to the survey. In addition, they were requested for the assessment of the administrative costs. These postcards were sent in three countries: Mexico, Australia and Iceland, because it was already established that the company characteristics are similar in the entire sample, so the non-respondents are assumed to be the same. In addition, these countries have the lowest response rates in the data set. The motives for not replying were similar: length and complexity of questionnaire, lack of interest in the subject, inability to calculate the costs and lack of time. Those who were able to evaluate the administrative costs stated that they were moderate for Mexico and Iceland; in Australia, they replied moderate or low. This suggests that the data set does not suffer from a non-response rate issue. The only word of caution that they raised was that “in some of the strata in certain countries, the small number of responses increases the variance of quantitative estimates”. The procedure that they used to overcome this issue was to weight the quantitative results in order to imitate the structure of firms in each country in terms of sector and size. The overview of the response rate in each particular country is available in Appendix table 1.
For Norway, the percentage of responses is high: 50 percent. In the area of tax there are in total 344 observations which represent a 49 percent response rate. Any explicit worries regarding non-
16
response rates are not raised and there is no particular reason why the methods used by the OECD (2001) should be questioned. In addition, the sample sizes are large. Moreover, as far as Norway is concerned, the response rate is also large. Therefore it can be assumed that there is no non-response bias.3.1.2. Common method variance
All the data that is being used was obtained via the survey. Acquiring the data from one source, which used the same method, might result in a common method problem. When data suffers from common method variance problem then the variance which is attributed to the measurement method instead of the relevant constructs may result in systematic measurement errors and bias the estimates of the true relationship between theoretical constructs (Chang, Van Witteloostuijn, and Eden 2010).
Chang, Van Witteloostuijn, and Eden (2010) propose four solutions for this issue such as the use of a different source of information for dependent and independent variables, the mixture of questions and usage of different scales, specifying the relationships in a complex manner and using statistic methods to correct for the problem. The way to tackle this potential issue was the usage of three different dependent variables in the robustness analysis. This allowed for a comparison between the models.
Moreover, the authors of the OECD report claim that the results show larger variability between companies, even when they compare the companies from this same sector or size. It is suggested that the perception of costs, including those calculated with SCM method, differ. In order to prove that this data set does not suffer from common method variance Harman's single factor test was conducted.
This test has revealed no common factors, which demonstrates that in this data set there is no common method variance problem.
3.1.3. Missing values
In this section the problem of missing values is addressed. In this data set two issues will be encountered regarding not having full information, one is missing values and the other is “I have no opinion”. There are several ways that one can deal with this problem. None of them are perfect and all of them suffer from their own issues. In this thesis the method of listwise deletion will be used, which means that all the observations with missing values will be deleted. In the report itself the authors are in general satisfied regarding specific answers. In Norway there are only two areas which are pointed out as troublesome and implying the possibility of running into problems with missing values. The first case is the environment area of regulations, because some companies, especially in business, do not have to comply with any regulations regarding the environment. Due to the fact that the remaining two areas are sufficient to conduct the analysis, it is decided to not include this area in the robustness analysis. Another issue raised are the questions regarding the quality of service variable, specifically questions asking for the relationships regarding obtaining the permit or the decision in the tax area of regulations. In total, the sample size will be reduced from 344 to 187 in the full model. Although the
17
reduction seems large the sample size remains relatively large. It is fully recognized that missing values might impose a limitation to this study, but there is not much that can be done to improve it except for a thorough robustness analysis.3.2. Dependent variable
In this section the method chosen to measure red tape will be discussed. Coursey and Pandey (2007) recommend testing red tape using several measures. The data set allows one to calculate administrative tape in four different ways: SCM, SCM corrected for size, global measure of red tape and perception of costs of compliance. It is very appealing to use all the methods. However, in the initial part for this study only one method has been chosen. Afterwards in the robustness analysis other methods are investigated.
The method that has been chosen to estimate red tape variable is the SCM. As already mentioned in the literature review, this method calculates actual costs of compliance and is used as a proxy for the perception of costs. There are two components in the data set, which will be summed up in order to calculate the dependent variable. The first component consists of variable costs, which are the multiplication of number of hours monthly spent on complying with regulations and average hourly wage in a given country. In addition to variable costs respondents were also asked to evaluate how much they spend on external help such as consulting. Due to the fact that many companies may use outsourced help in addition to their variable costs these values will be treated as variable costs and added up. The second component will consist of fixed costs such as money spent on computers and programs that are used to comply with administrative tape.
3.3. Independent variables.
3.3.1. Size
The debate regarding the measurement of the size of a company is large (Calof 1993). There are many proxies used to measure size. The most common proxies are number of employees, turnover and market share. In the data set that is at the disposal there are two proxies that can be used for estimating the size of a company: number of employees in the firm and turnover of the company. In the business impact studies these two proxies were most widely used. In the initial model number of employees will be used as a proxy for firm size. The reason for choosing only one proxy for size is that both proxies correlate with one another, which leads to multicollinearity issue, which in turn results in only one of them being significant when both are present in the model. However, both of them are significant separately. Due to the fact that they correlate highly the number of employees is decided to be used as a proxy for the size of the company. In order to account for non-linear relationship the size variable will be squared and added up to the equation.
18
3.3.2. AgeAnother hypothesised relationship is the age of a company. The best way to measure age would be to have continuous data of the age of companies and check for exponential relationships. However, the data at the disposal has only three time categories: less than two years, between two and five years and over five years. Having data available only in three categories reduces the interpretability and the power of the results. Another problem that this data structure may cause is that one group might be overrepresented or underrepresented. The summary statistic shows that: 83 % companies are older than five years, 12 % are in the group between 2-5 years and 5 % are below 2 years. There is a large disproportion, which may further reduce the power of this variable.
3.3.3. Quality of service
The quality of service will be measured by using the perceptions of it by companies that are surveyed. In the data set there are two sets of 4-point likert-scale questions that measure the perception of quality of service (see appendix 2). The procedure to extract the variable is exploratory factor analysis. In addition the varimax rotation will be used in order to obtain easier to interpret outcomes. The results from factor rotated matrix suggest that there are two valid factors for which variance is above 1. The first factor is composed of three questions, because for those variables the factor loadings are above 0.5:
- Officials give definite answers.
- The response of the government authorities is given within a reasonable time.
- It is clear who is responsible for the decisions.
Index value of the sum of these questions will be created. These questions are predominantly dealing with the answers that the government organizations provide. Therefore, this variable will be called
“quality answer”. The second factor is composed of the two following questions:
- The process is flexible and not bureaucratic.
- Decisions are consistent and predictable over time and among similar business.
These questions deal predominantly with the process of obtaining the decision and the expected outcome of this process. Therefore this variable will be referred to as “quality of the process”.
3.3.4. Control variables
Theory suggests that foreigners have lower understanding of the national rules and regulations, therefore their perception of administrative tape should be more negative. In the data set there is a dummy variable which accounts for ownership. This dummy consist of a question of whether the company has fully national ownership or otherwise consist of a foreign one. Another control variable is industry. Accounting for industry differences is very important as the level of compliance cost might vary substantially among different industries due to the specific industry regulations. In the data set
19
there are fourteen different industries. Using such disaggregated approach may be inconclusive and scatter the data too much. Even in the OECD (2001) report the more aggregate measure is used. The table 1 provides an overview of all the industries and the way that they have been aggregated.There are several potential controversies regarding this classification. First of all it is made arbitrarily according to authors’ best understanding of the subject. Secondly, “no response” have been put as “other”, because it can be assumed that the respondent in such company that have not ticked any answer had problems with classifying their company to listed industries, implying that their company is operating in another industry. This is one of the rare cases where deleting missing value is not justified and one can logically argue to keep them. The last issue is that some industries that are present in the total sample are not present in Norway. In the initial analysis the aggregated measure of industry yielded better results and higher Cronbach’s alpha of the entire model.
Table 1: Overview of industries.
Industry Classification Presence in the
data
Fishing Other
Manufacturing Manufacturing and related
Construction Construction
Wholesale, retail trade Service
Repair of motor vehicles, cycles, etc. Service not there Hotels and Restaurants Service
Transport, storage and
communication Service
Financial intermediation Financial service
Real Estate Financial service
Renting of machinery Manufacturing and related not there Computers and related activities Manufacturing and related
Consulting and business activities Financial service Education and research Social service Health and Social work Social service
Other Services Service
Other (specify) Other
No Response Other
3.4. Model
The aim of this study is to understand antecedents of perception of administrative tape, caused by national rules and regulations in the private sector. To test the hypothesis presented in the
20
literature review the following model was developed with Y as the dependent variable, , , and as four independent variables and as the control variable:
The dependent variable Y is measuring perceptions of red tape. The independent variables are: size of the company ( ), size of the company that is squared ( ), age of the company ( ),quality service variables ( ). I will use two control variables industry( ), and ownership dummy ( ), which takes 0 value if foreign owned and 1 if nationally owned. The error term is .
3.5. Method of estimation
The quality of service as the independent variable is measured using an indexed likert-scale, which implies that this type of data is ordinal. Moreover, the age as an independent variable is categorical type of data. This means that the variables are composed from ordinal type and continuous type of data. Although index scores are used for quality service variables, which makes them closer to continuous type of data, the method that should be used is challenging. There is a debate among scholars regarding the appropriateness of the methods that should be applied in such a case. Lubke and Muthen (2004) claim that under very strict assumptions one can treat ordinal data in the same way as continuous data. In addition, Glass, Peckham, and Sanders (1972) claim that in F test ANOVA could provide appropriate p-values for an ordinal data. On the other hand, most methodological and statistical texts clearly state that for ordinal data the median or mode are the adequate measure of central tendency suggesting non-parametric statistics (Clegg 1998). Following the suggestion of Lubke and Muthen (2004), the ordinal data, like continuous data, will be treated and estimated by the model ordinary least squares (OLS). This means that very strong assumptions of the best linear unbiased estimates (BLUE) have to be tested (Griffiths and John 2008). If the data satisfies the BLUE requirements, OLS will be used as the method of estimation. Even if the data fulfills these conditions the interpretation of the coefficient magnitude for quality service and age variables would be impossible. Only the signs of the coefficient for these variables will be interpretable. The reminder of this section will be testing if the data fulfills BLUE requirements requirements such as heteroscedasticity, endogeneity, multicollinearity and normality.
3.5.1. Homoscedasticity
One of the very important assumptions of OLS is homoscedasticity (Newbold, Carlson, and Thorne 2007). It means that the variance of the error term is constant and alike for all error terms.
Var
If the variance of an error term is not equal then the assumption of homoscedasticity is violated and the data is said to be heteroscedastic. This causes a problem because the underlining property of OLS is to minimize the squares of the errors. If those errors have different variance then observations with higher errors variance are being over-weighed and the observations with lower error variance are
21
being underweighted. This means that observations that are more extreme case, which lie further away from the true line (high variance) meaning that they have the least information regarding the regression, are more emphasized. The implication in such a case is that if heteroscedasticity is present and other assumptions of OLS are met then the parameter estimates are still not biased, but standard errors are biased, as the OLS no longer deliver estimates with the lowest variance. Biased standard errors lead to wrong t-statistics and confidence intervals, which makes the interpretation of the hypothesis significance impossible. To test for heteroscedasitcy one should check the variance of the residuals. The Breush-Pagan test was chosen because it is one of the formal tests that are available to test for heteroscedasticity. Tests’ null hypothesis states that the variance of the error terms is constant. The outcome of the Breush-Pagan test is above 0.05, which means that the null hypothesis of Var is not rejected. This allows one to conclude that the data does not suffer from heteroscedastic.3.5.2. Endogeneity
When using OLS as the method of estimation the variables have to be strictly exogenous (Griffiths and John 2008). This means that the error terms cannot correlate with the independent variables. The cause of such correlation can be: measurement error, omitted variables, lagged dependent variable models with serial correlation, simultaneity and sample selection errors.
Endogeneity is a serious violation of the OLS; it may cause biased and misleading coefficients and significance tests. The only proper diagnostics available on the cross-section data is Hausmans’ tests.
However, it is not possible to perform this test since there are no instrumental variables at the disposal. Therefore, the only way to prove the lack of endogenity issue is a sound theoretical model and exploratory factor analysis. For most of the independent variables it is not expected to see any correlation with error terms. Moreover, they have been successfully used in the previous studies and none of the previous authors addressing this issue have pointed out any problems. The only variable that could be questioned is the quality service that is by some authors claimed to overlap with the concept of red tape. However, as argued before, it is my strong belief that this is not the case in this study. Especially since the questionnaire was designed in such a way to separate those two concepts.
Exploratory factor analysis has not revealed strong connections between all measures.
Although the theoretical and indirect method support lack of endogeneity in this model, the usage of appropriate methods would be recommended in future research to address this relationship.
3.5.3. Multicollinearity
Independent variables should not correlate highly with one another (Griffiths and John 2008). If this happens then the OLS suffers from multicollinearity issue. This phenomenon does not violate OLS assumptions and even if it is present OLS results are still BLUE. However, the presence of multicollinearity is not welcomed because standard errors are biased. This in turn may provide
