1 Roderick Kloeze 1814850 July 12th 2012
The costs of red tape – determinants
and cross-country differences
Master Thesis
University of Groningen
2
Content
1. Introduction ... 5
1.1 Research aim and question ... 7
2. Literature review ... 9
2.1 Definitions and measurements of red tape ... 9
2.1.1 Is red tape always bad? ... 11
2.1.2 Red tape and formalization ... 12
2.1.3 Public-private differences in red tape ... 13
2.2 The costs of regulation... 14
2.2.1 Economic impact of regulation ... 14
2.2.2 The costs of administrative delay ... 16
2.3 Perceptual or objective measures? ... 17
2.4 The origins of red tape ... 18
2.5 Summary ... 20
3. Model and hypotheses ... 21
3.1 The regulatory stockpile ... 23
3.2 Quality of design ... 23
3.3 Predictability of application ... 24
4. Data and methods ... 25
4.1 Data ... 25
4.2 Costs of red tape ... 27
4.3 Independent variables ... 28
4.3.1 Regulatory stockpile... 28
3 4.3.3 Predictability of application ... 29 4.4 Control variables ... 30 4.5 Statistical model ... 32 4.6 Method of estimation ... 32 4.6.1 Homoskedasticity ... 32 4.6.2 Endogeneity ... 33 4.6.3 Multicollinearity ... 34 4.6.4 Normality ... 35 5. Results ... 36 5.1 Descriptive statistics ... 36 5.2 Regression results ... 38 5.3 Robustness tests ... 42
5.3.1 Disaggregated cost indicators ... 42
5.3.2 Alternative model specification ... 44
5.3.3 Country level regressions ... 45
5.3.4 Alternative estimation method ... 47
5.4 Discussion of findings ... 48
6. Conclusions, limitations and future research ... 49
6.1 Conclusions ... 49
6.2 Added value ... 52
6.3 Limitations and future research ... 54
7. Appendix ... 58
4
Abstract
This research investigates cross-country differences in the costs of red tape and identifies several determinants of these costs. The focus is on red tape originated from the government that small businesses have to comply with. The determinants identified here are the regulatory stockpile, the quality of design and the predictability of application. The results show that a larger regulatory stockpile increases the costs of red tape, whereas higher quality of design and predictability of application lower the costs of red tape. The added value of this study is the data set used and the multi-dimensional view of red tape.
5
1. Introduction
Low economic growth and high levels of government debt and deficits have increased the interest of governments in structural reforms to boost competitiveness and reduce
unemployment (see for example, The Economist, February 18th 2012). One aspect that often appears in these discussions is “cutting red tape”; the drive to reduce red tape actually
precedes the current economic problems and has received wide attention, especially in Europe (Wegrich, 2009). In the US the passing of the Sarbanes-Oxley Act, following the Enron scandal, also sparked interest in the (economic) costs and benefits of regulation (e.g. Engel, Hayes and Wang, 2007; and Zhang, 2007). Academic research into the effects of regulation and red tape also precede the current interest, however the impact of this academic research on
government policy has been rather limited (Pandey and Moynihan, 2006). The current process of reducing the costs of regulation is mainly driven by the development of a standardized, easy to understand measure, the so-called Standard Cost Model (SCM) (see SCM Network, 2011); which originated from outside the academic world (Wegrich, 2009).
Red tape has become somewhat of a catchphrase and many countries and organizations are said to suffer from the malicious effects of red tape. However, these discussions have often neglected the use of a proper definition of red tape (Helm, 2006). There is a large research tradition that has focused on the definition of red tape and how to differentiate red tape from general forms of regulation, formalization and bureaucratization (Bozeman 1993; Pandey and Scott, 2002; DeHart-Davis and Pandey, 2005). The next section will discuss the most important definitions and measurement techniques in detail. This section will discuss some studies and concepts that are important with regard to deriving the research question for this study. Bozeman (1993) identifies four different types of red tape, based on whether red tape originates from within or outside the organization, and on whether its effects are within or outside the organization. The first type of red tape is “ordinary red tape”, red tape that
6 which is imposed from outside the organization, but has effects within. This type of red tape is often associated with (excessive) government regulation. The academic research tradition that investigates this type of red tape was started by Djankov et al. (2002). The other 2 types of red tape, “interorganizational red tape” (internal, internal) and “pass-through red tape” (external, external) generally receive less attention. This study will focus on “external control red tape” however, the focus and type of data will differ from the Djankov et al. (2002) style studies. Ordinary red tape research has focused mainly on the existence and effects of internal red tape in public and private organizations (Bozeman, 1993; Pandey and Kingsley, 2000; Feeney and Bozeman; 2009; Brewer and Walker, 2010). This strand of research has identified many
variables that influence (perceptions of) red tape as well as variables that moderate the effects of red tape (Pandey and Kingsley, 2000; Pandey and Moynihan, 2006; DeHart-Davis and
7 The institution-based view has been expanded to also cover strategic management (Peng, et al., 2009). In this institution-based view, the strategy, conduct and performance of a firm is in large part dictated by the external institutional environment prevalent in a certain country or region. Red tape and regulation are part of this institutional environment (Scott, 2001) and could entail large costs and thus be detrimental to firm performance. Cross-country differences in red tape could therefore be used to explain cross-country differences in firm performance.
The main difference between the two research traditions described above is with regard to the measurement of red tape and specifically between objective and subjective/perceptual
measures. The studies in the tradition of Djankov et al. (2002) attempt to objectively measure red tape. The advantage is that such measures are verifiable and can be (relatively easily) reproduced across countries. The disadvantage is that such measures may not fully cover all costs of red tape and does not differentiate between useful rules and regulations and red tape (Arruñada, 2007). There are other objective measures such as the Standard Cost Model (SCM Network, 2011), that have the same advantages and disadvantages. Subjective/perceptual measures are based on questionnaires, and some standardized measures have evolved (Pandey and Scott, 2002). One of the advantages of these kinds of measures is that they can account for the fact that what constitutes red tape often differs from one person to the next (Feeney and Bozeman, 2009). The advantages and disadvantages of perceptual and objective measures will be discussed in more detail in section 2. This study will use a perceptual measure derived from a questionnaire distributed by the OECD (2001).
1.1 Research aim and question
The aim of this study is to investigate if red tape differs across countries. And, if this is so, what factors accounts for the differences in red tape. The main research question is therefore: Does
the level of red tape differ between countries and why? The reasons why red tape are expected
8 countries: red tape in the World Bank studies is treated as a one-dimensional concept, whereas the studies of internal red tape have shown that it is a more complex phenomenon. So both from an empirical as well as a theoretical point of view this is an underexplored area of research.
There are some studies that propose how red tape comes into existence through the political process (Rosenfeld, 1984; Bozeman, 1993 and Helm, 2006), which is discussed in more detail in the next section. The conclusion of this process is that red tape may come into existence due to both overregulation and badly designed regulation. Based on this process, red tape can be expected to differ across countries because of differences in: the regulatory stockpile, the quality of these rules and regulations and the predictability of enforcement of rules and regulations. This point will be elaborated upon below, for now it suffices to state the sub-questions of this paper:
1) What is the relation between the regulatory stockpile that a business faces and has to comply with and the costs of red tape?
2) What is the relation between the quality of the design of the rules and regulations that a business has to comply with and the costs of red tape?
3) What is the relation between the predictability of the enforcement of the rules and regulations that a business has to comply with and the costs of red tape?
This paper continues as follows: the next section discusses the relevant literature with regard to the costs of regulation, with a focus on red tape. This section discusses definitions and common measures of red tape. Other important research topics that arise from the literature are also discussed. The section ends with a discussion on how red tape comes into existence, which will form the basis of the discussion why red tape can be expected to differ across countries. Section 3 discusses why red tape differs across countries and derives the hypotheses that will be investigated in this study. Section 4 discusses the data that will be used, and the section that follows discusses the results of statistical analysis. Finally, section 6 concludes and offers
9
2. Literature review
The current research tradition with regard to red tape received a strong impetus by Bozeman (1993), who clearly defined key concepts and aspects of red tape and set out implications for future research. Previous important work discussed by Bozeman (1993) includes Kaufman (1977), Buchanan (1975), Rosenfeld (1984) and Baldwin (1990). These studies already indentify several questions that have been dominant in later research: What is red tape and how should it be measured? Is red tape always bad? How is red tape different from formalization? Is red tape higher in public or in private organizations?
2.1 Definitions and measurements of red tape
Rosenveld (1984) provides one of the earliest definitions of red tape and many later definitions are similar to his. The definition that he proposes is: “guidelines, procedures, forms and
government intervention that are perceived as excessive, unwieldy, or pointless in relationship to decision-making or implementation of decisions” (Rosenveld, 1984: 603). This early definition
sets out two important characteristics of red tape that are often repeated: red tape as excessive rules and regulations and red tape as a perception or impression.
One of the goals of Bozeman (1993) was to properly define the concept of red tape and provide additional detail, first he defines several aspects of red tape, such as: rule density, functional object and rule efficacy. The definition of red tape that he settles on and that has been often quoted in later research is: “rules, regulations, and procedures that remain in force and entail a
compliance burden for the organization but have no efficacy for the rules’ functional object”
(Bozeman, 1993: 283). Where functional object refers to: “the organizationally sanctioned
purpose for which a rule is created, the problem it seeks to solve, the opportunity it exploits”.
And rule efficacy is defined as: “the extent to which a given rule addresses effectively the
functional object for which it was designed”. An important drawback of this definition is that it
is rather broad in scope (Pandey and Moynihan, 2006).
Pandey and Kingsley (2000: 782) provide a slightly narrower definition: “impressions on the part
10
to the organization”. The advantage of this definition is that it considers red tape as a
characteristic of the organization rather than of every individual rule. As a result this definition is more workable in empirical research. In empirical research a definition close to the two presented above is often used: “burdensome administrative rules and procedures that have
negative impacts on the organizations’ effectiveness” (Pandey and Kingsley, 2000; Feeney and
Bozeman, 2009; Feeney and DeHart-Davis, 2009).
Other studies go beyond this measure of red tape by including questionnaire items that deal with more specific aspects of red tape (Pandey and Kingsley, 2000; DeHart-Davis and Pandey, 2005; Pandey and Moynihan, 2006; Coursey and Pandey, 2007; Brewer and Walker, 2010). These studies are based on surveys from the National Administrative Studies Project (NASP). Next to the overall measure of red tape these studies often include one or two measures of “personnel red tape”, that are concerned with the ease of firing underperforming employees, the ease of promoting well-performing ones and the connection between pay and
performance. Other questionnaire items deal with red tape in procurement or information systems. These more specific items do not necessarily deal with rules that are excessive and serve no purpose, so it could be argued that they do not deal purely with red tape.
Finally, another interesting measure of red tape is proposed by Pandey and Bretschneider (1997). They measure red tape as the part of administrative delays that cannot be attributed to organizational characteristics. Administrative delays are regressed on several organizational characteristics and the residual is taken to be red tape. This measure has the advantage of being objective, furthermore as red tape is considered to be a residual, this measurement captures the delays that serve no function and are therefore pointless. This very closely captures the definition of red tape. Despite these advantages the measure is not often used, due to several disadvantages. First of all, for the residuals to truly capture red tape no
11
2.1.1 Is red tape always bad?
The definitions of red tape cited above suggest that red tape is a necessarily negative
phenomenon, however not all rules and regulations have detrimental effects. Additionally, the line dividing red tape from useful rules and regulations is blurred and often depends on a certain point of view. Several terms have been introduced to deal with these issues: white or green tape and stakeholder red tape, each is discussed in turn.
Before defining red tape Bozeman (1993) first considers the rules and regulation in general, he acknowledges that many rules and regulations have important functions and benefits in
addition to their costs. He notes that in everyday discussions the distinction between rules that have benefits and red tape, which does not, is not always clear. As a result, rules that are frustrating but do have benefits are sometimes mistakenly considered to be red tape. To avoid the contradictory notion of beneficial red tape, he labels such rules “white tape”. So red tape is defined as those rules that serve no purpose at all, whereas white tape refers to those rules and regulations that are frustrating but do serve an ultimate purpose.
In a similar vein DeHart-Davis (2009) considers effective rules that improve organizational performance. The label that she uses for such rules is “green tape” and should be viewed as the exact opposite of red tape. So, green tape refers to rules that serve a functional object and she argues, therefore, that the simple notion of cutting rules will not necessarily reduce red tape and may even harm organizational performance. DeHart-Davis (2009) identifies five
characteristics that are associated with effective rules. These are: written requirements, valid means-ends relationships, optimal control, consistent application and purposes that are understood by stakeholders. The requirement of valid means-ends relationships is especially important, green tape are rules that achieve an end, whereas red tape are rules that serve no end, or functional object (Bozeman, 1993). The importance of the view of stakeholders in determining whether a rule is good or bad has also been recognized by other authors.
12 from the other. Different stakeholders may have different views of what constitutes red tape, as they differ in opinion whether or not a rule serves a functional object and achieves a valid end. Kaufman (1977: 4) captures this tellingly: “[o]ne person’s red tape may be another’s
treasured safeguard”. To accommodate this insight Bozeman (1993: 284) also proposes a
definition of stakeholder red tape similar to the definition of general red tape: “organizational
rules, regulations, and procedures that remain in force and entail a compliance burden, but serve no object valued by a given stakeholder group”.
If this definition is followed a rule can still serve a functional object, but it will be perceived as red tape by a certain stakeholder group that does not value this object. As a result what one group considers red tape can be considered by the other to be white or green. Studies that use questionnaire data aimed at one particular group necessarily measure stakeholder red tape, because the respondents will consider a rule pointless or excessive based on the value they attach to the object served. However, few studies explicitly account for this effect. Feeney and Bozeman (2009), do explicitly study stakeholder red tape by comparing two groups that deal with the same red tape. They study both the staff and external consultants of the Georgia Department of Transportation and find that one group (the staff) perceives more red tape in the organization than the other. This result is however limited to organizational red tape, in terms of contracting both parties perceive the same levels of red tape.
2.1.2 Red tape and formalization
13 tape is the result of attempts at formalization, either as a result of good rules gone bad or as new rules born bad. This point is elaborated upon in a later section of this literature review. The concept of formalization is similar to the aforementioned concepts of white and green tape, however it is more concerned with an organizational characteristic than with whether or not a particular rule is beneficial to the organization. Pugh et al. (1968: 75) define formalization as: “the extent to which rules, procedures, instructions and communications are written”. Written rules and procedures are one of the requirements of green tape described by DeHart-Davis (2009), indicating that the right level of formalization can lead to rules and procedures that are beneficial to the organization. Bozeman (1993) adds to the definition above that not only the extent of written rules and procedures are important for formalization, but also their number. Formalization and red tape are thus both concerned with rules and procedures, however there is one crucial difference, formalization is a natural and inherent part of any well-functioning organization, whereas red tape is not. Formalization allows organizations to
operate efficiently and predictably, and as Wintrobe (1982) argues there is an optimal level that balances rules and procedures with employee discretion. The optimal level of red tape, on the other hand, is none at all, red tape always detracts from organizational performance.
2.1.3 Public-private differences in red tape
14 that public organizations are characterized by higher levels of red tape, but find that work alienation is as strong a predictor of perceived red tape as the sector an employee is active in. However, the overall conclusion of these studies that public organizations are characterized by higher levels of red tape than private organizations stands. Bozeman (1993) argues that this increased red tape in public organizations is mainly in the ordinary red tape category (that is, red tape that is originated and has an impact, within the organization).
Indeed the focus of most of the studies cited above is on ordinary red tape, whereas this study focuses on external control red tape. However, many of the findings above can be applied to external control red tape and the sources of both types of red tape may be similar. This is especially the case for internal red tape in public organizations and the external, government originated red tape faced by businesses. Because, red tape in these cases originates from the same source, the political process.
2.2 The costs of regulation
The measures of red tape discussed above generally ask for the perception of respondents about the degree to which regulation is perceived to be pointless or excessive. Other measures try to be more objective with regard to the costs of administrative procedures, however these measures are generally unable to determine whether rules and regulations are excessive or pointless. They are therefore not concerned with red tape as has been defined above, however these measures are more popular among policy makers. Two common measures are discussed here: the standard cost model (SCM) and the Worldbank “Doing Business” indicators.
2.2.1 Economic impact of regulation
The “Doing Business” indicators investigate the level and effects of regulation on a cross-country basis and from an economic perspective. Studies in this tradition often focus on a broad cross-section of countries and are concerned with the entry of new businesses,
15 seem to assume that all forms of regulation have a negative impact on these economic
variables, which is one of the main criticism of these indicators (Arruñada, 2007).
In a broader perspective these studies are based on the institutional view to economics that argues that economic growth and wealth ultimately depend on the institutional framework of a country (Helpman, 2004). North (1990) defines these institutions as the “rules of the game” and formal regulation make up a large part of the institutional framework. Scott (2001) considers regulation as one of the three pillars of institutional theory (together with norms and culture). This research tradition argues that the right rules and regulations improve economic
performance, whereas ill-designed regulation can hold back economic growth.
The study by Djankov et al. (2002) that started the research tradition based on the “Doing
Business” indicators focuses on the regulation of entry of new businesses. The costs of
regulation are linked to indicators of corruption and autocracy. The underlying assumption in their study is that regulation of entry is only to the benefit of incumbents and those that reap the revenues. So, it is expected that corruption and autocracy are positively correlated to regulation of entry, and they find strong evidence of this relation. The main purpose of Djankov et al. (2002) is to estimate the direct costs of the regulation of entry and compare these across countries. Their sample consists of 85 countries from different continents and income
categories and is thus very extensive. The costs of entry are reported as a percentage of per capita GDP to allow comparison across countries. The differences across countries are substantial: the number of procedures needed to start a business range from 2 to 21, the minimum number of days needed to complete them from 2 to 80 and the costs from around 1.5% of per capita GDP to over 450%. Even if not all of these procedures can be considered pointless, these results are a clear indications that red tape differs strongly across countries. It is unlikely that such large differences in regulation can be explained by differences in the “optimal” level of regulation alone, part of the observed differences must be due to differences in red tape.
16 substantial and are not often considered in estimates of proposed regulation (Harrington, Morgenstern and Nelson, 2000). Using various datasets and measures, regulation has been shown to reduce new business creation in fast growing sectors (Ciccone and Papaioannou, 2007), reduce investment (Alesina et al., 2005) and affecting going private decisions (Engel et al., 2007). These studies are closer to red tape in that they determine whether regulation has positive or negative overall economic effects. Those rules and regulations with negative effects could be considered red tape, whereas those with positive effects are legitimate regulation. However, it is possible that not all the costs and benefits of regulation are captured, so these results need to be interpreted carefully.
The study of the effects of the Sarbanes-Oxley Act (SOX) by Zhang (2007) are interesting in this regard. SOX was passed, following the Enron scandal, to protect shareholders better against corporate fraud, so the new rules and regulations should be beneficial to shareholders.
However, using an event study method, Zhang (2007) finds that the passing of these new rules and regulations caused share prices in the US to fall. Evidently, shareholders believed that the costs of the law were greater than the benefits, while the law was designed to protect them. In this case, there is, therefore, a strong argument to label the new rules as red tape. Such an extensive event study test to determine whether a new rule should be labeled red tape or not, is however not plausible in many cases.
2.2.2 The costs of administrative delay
Another method of estimating the costs of government generated red tape is the Standard Cost Method (SCM). The goal of the SCM is to quantify the total costs of administrative procedures and these estimates are often matched with a target for the reduction of costs (Wegrich, 2009). This method of assessing the costs of regulation has been adopted throughout the European Union over the past decade. The costs of administrative burdens can be quite substantial, in The Netherlands the costs of administrative delays has been put at 3.6% of GDP and in
17 compliance, so it ignores indirect and structural costs of regulation as well as potential benefits. So in this regard the SCM does not truly measure red tape and could both under- or
overestimate the costs of regulation to society.
According to Wegrich (2009) the SCM has spread across countries at a much faster pace than earlier efforts to cut red tape. The SCM method has the advantage that it delivers an
internationally comparable single measure of the overall costs of administrative burdens, obtained through a fairly standardized process. The process of estimating the SCM measure of administrative burdens is as follows (Keyworth, 2006): first, the relevant administrative
procedures are identified, then the number of firms that these procedures apply to.
Subsequently an assessment of the costs of compliance for the average firm is obtained, either from the concerned firms or by direct observation of a firm completing the requirements. Finally, the number of firms that have to comply with a certain administrative procedure is multiplied by the cost estimate to obtain the total administrative burden of a given rule.
2.3 Perceptual or objective measures?
The discussion above identifies several measures of red tape and/or regulatory burdens, and all these measures have their own advantages and disadvantages. Therefore, there is no best overall measure of red tape for all situations. However, the questionnaire based measure comes closest to actually measuring red tape, as rules that are pointless and excessive. The other measures tend to confuse red tape with either formalization or general potentially beneficial regulation. Furthermore, the questionnaire measures tend to ask for perceptions or impressions rather than try to put an objectively figure on red tape. Overall in the theoretical literature the questionnaire measure is adopted more often (Coursey and Pandey, 2007), whereas the other measures are more popular in policy circles (Keyworth, 2006).
18 tape when it is put on the policy agenda by an effort to reduce it. Furthermore, perceptions compare the current level of red tape to some implicit benchmark and the exact benchmark is not always clear. These caveats should also include the possibility that managers might report exaggerated levels of red tape to serve their own purposes. Baldwin (1990: 9) on the other hand offers several arguments in favor of perceptual data. First of all, perceptual data allow managers to respond not just on the number of rules and procedures they face, but also to what degree these are oppressive or frustrating. Furthermore, he argues “that behavior is typically a response to perceptions, not objective information”. So, perceptual measures should give a better reflection of the constraints faced by managers, as well as the degree to which these constraints serve no purpose. Additionally, Feeney and Bozeman (2009: 710) observe that “[t]here is an emerging consensus that perceptions of red tape matter and that these
perceptions affect behavior in complicated ways”.
2.4 The origins of red tape
19 The reasons listed by Bozeman (1993) for “rules born bad” are: inadequate comprehension, self-aggrandizement, negative sum compromise, over-control and negative sum process. All these reasons point to instances where the rule making process is faulty, or corrupted by special interests. The natural result of a faulty is process is faulty rules that serve no
organizational or social function. The reasons for “good rules gone bad” are: rule drift, rule entropy, change in implementation, change in the functional object, change in the rule’s
efficacy, rule strain, accretion and misapplication. These reasons all concern instances where an intrinsically good rule is applied in ways not originally intended or loses its meaning over time, due to inertia of the rule and a changing environment.
20 groups that are important during elections. So the political process provides a setting where there is excessive demand for and supply of regulation, increasing external red tape for businesses and internal red tape for government organizations.
2.5 Summary
The above literature review has defined red tape and discussed several often used measures of both red tape and administrative burdens. The different measures have been contrasted and the overall conclusions seems to be that perceptual measure most closely measure red tape, whereas the other measures focus more on administrative burdens. Finally, the reasons why red tape exists at all, given the generally accepted pointlessness of red tape, have been discussed. The sources of costs of red tape have been touched upon, but they will be
21
3. Model and hypotheses
As mentioned the main aim of this research is to determine what causes the differences in regulatory costs between countries. The focus is on three aspects of regulation: the number of rules, the quality of rules and the predictability of rules. These will be discussed in detail below. The main research question, as defined above, is therefore: Does the level of red tape differ
between countries and why?
The introduction and literature review have already provided part of the (theoretical) answer to this question. It has been shown that countries differ in their level of regulation (Djankov et al., 2002), and it is theoretically possible that these differences reflect differences in underlying circumstances and that all countries have instituted optimal levels of regulation. However, this is unlikely, at least part of the differences in regulation should realistically be attributed to differences in red tape (because total regulation is the sum of useful regulation and red tape). These differences in regulation are attributed to differences in institutional frameworks. So, it makes sense to attribute differences in red tape across countries to differences in institutions as well. Specifically, as the section on the origins of red tape explains, the political process and the quality of government can cause red tape. And therefore they can also cause differences in red tape across countries, as the political process and quality of government differ across countries with institutional differences.
22 the regulatory stockpile that a business faces and has to comply with is a natural first
determinant of the costs of red tape.
However, the regulatory stockpile that need to be complied with is not the only determinant of red tape. An equal number of rules and regulations that needs to be complied with can lead to different costs of red tape depending on the design of these rules and their application. The main criticism of Arruñada (2007) with regard to the Doing Business studies, is that these studies ignore such considerations. For example, the costs of red tape are lower if two rules are consistent then if they are inconsistent with each other. The section on the origins of red tape shows that there is a large chance that the political process may produce regulation that is of low quality. As Rosenfeld (1984) argues, the result of compromise may be regulation that is vaguely worded or ambiguous. It is then left to the bureaucracy to refine and implement such regulation, the result of which will be unnecessarily many exceptions and caveats. As a result regulation will be excessive with regard to the functional object. The process of pleasing constituencies and special interests, as described by Helm (2006), can lead to rules that are inconsistent with each other. Because, as rules are designed per constituency or interest group, there is a large chance that oversight is lost with regard to whether or not rules are consistent or overlap.
The quality of government can also lead to problems with the predictability of the enforcement of rules and regulations. The most extreme cases concern serious infractions of good
government, such as corruption, which lead to inconsistent application of rules between different businesses or to unexpected payments. However, even if such cases are put aside there may still be considerable uncertainty with regard to how a rule may be applied to a specific business. Vague and ambiguous regulation that comes out of the political process (Rosenfeld, 1984), may cause different agencies or different civil servants within one agency to interpret the same rule in different ways. And as a result the application of rules may differ across firms and time. Furthermore, businesses may need to make payments they did not expect, as a result of vague or misunderstood regulation.
23
3.1 The regulatory stockpile
The studies in the tradition of Djankov et al. (2002), show that the costs of these administrative procedures can be substantial. The average number of procedures to start a business over the entire sample of Djankov et al. (2002) is 10.48, taking at least 47.40 business days. The costs of these procedures are 47.08% of per capita GDP. And these costs can be prohibitively high and severely slow down the rate of new business entry (Ciccone and Papaioannou, 2007). As discussed above, rules and regulations are not the same as red tape, however the regulatory stockpile is an indicator of red tape, as the compliance with a rule is a necessary condition for it to be considered red tape. Additionally, Feeney and Bozeman (2009), in a study of internal red tape, find that those respondents who feel that the focal organization has to many rules also perceive higher levels of organizational and contracting red tape. This result can be explained by two different types of logic. Firstly, if the number of rules that need to be complied with is higher, the number (and perhaps the share) that is excessive or obsolete is larger. Secondly, if more frustrating, but useful, rules need to be complied with the chance that an individual manger will view them all to be pointless or excessive is larger. So these rules will be perceived as red tape even if strictly speaking they are not, however as has been observed before actions often depend on perceptions not on the objective reality (Baldwin, 1990). This leads to the following hypothesis:
Hypothesis 1: A larger stockpile of regulations that need to be complied with will increase the (perceived) costs of rules and regulation.
3.2 Quality of design
The second independent variable concerns the quality of design of rules and procedures. This variable encompasses several indicators of quality, such as the ease of understanding rules and procedures, whether or not it is clear which agency to contact and whether rules are designed to achieve their object as effectively as possible. One of the main criticism of both the “Doing
Business” and SCM measures of the costs of regulation is that these treat all regulation as bad
24 fundamental component of regulatory reform and regulatory management in a large number of countries”. Furthermore, as observed above, DeHart-Davis (2009) finds that well-designed rules are less likely to be perceived as “red tape” and more likely as “green tape”. So, if rules and regulations are well-designed and applied, the perceived negative effect is lower. Additionally, the discussion concerning “rules born bad” (Bozeman, 1993) clearly show that the design of regulation is a key determinant of red tape. These are all reasons why low quality regulation increases the perceived costs of red tape. More objectively, better quality of service should reduce the costs of regulation, given the number of rules/procedures. First of all, if rules are easy to understand and it is clear who to contact the time spend to comply with regulation is reduced. Also, rules that are easy to grasp will lower the need for legal advice and investment into specialized software. So lower quality regulation increases the costs of “transacting” with the government and since compliance with regulation is a “transaction” that cannot be (legally) avoided, it raises the costs of red tape. This leads to the following hypothesis:
Hypothesis 2: Higher perceived quality of the design of rules and regulation, lowers the perceived costs of rules and regulations.
3.3 Predictability of application
25 and enforcement are more predictable. Indeed, the theory of transaction cost economics argues that a transaction that is characterized by high uncertainty should either be internalized or there should be unilateral adaption (Williamson, 1991). However, a “transaction” with a regulatory body cannot, by definition, be internalized, so the increased “transaction” costs due to adaptation have to be borne by the organization. This leads to the following hypothesis:
Hypothesis 3: The more predictably rules and regulations are enforced, the lower the perceived costs of these rules and regulations.
4. Data and methods
4.1 Data
The data that will be used in this study are from a dataset collected for the OECD (2001)
Business views on red tape study, a survey of nearly 8000 small and medium sized firms in 11
OECD countries. The main objective of the OECD (2001) study is to explore differences in total regulatory costs and quality across different countries, and identify weaknesses and strategies for improvement, as such the data is well suited to test the above hypotheses. The
disadvantage of this dataset is that it is concerned with the costs of regulation, and therefore not red tape per se. The above literature review has clearly shown that regulation and red tape are different concepts. However, this dataset will be used because it has several advantages that outweigh this drawback. This dataset uses perceptual, questionnaire based measures which is common practice in red tape research. The use of this dataset allows this practice to extend beyond the traditional internal red tape studies, to those that are concerned with external control red tape. This dataset is (to my knowledge) the only dataset that is concerned with regulation/red tape that applies a questionnaire method on a cross-country basis.
26 The data are concerned with the costs and effects of regulation as well as several indicators of quality and enforcement. The data was gathered for three areas of regulation were examined: employment regulation, environment regulation and tax regulation. The three different regulatory dimensions are defined by the survey report as:
Employment regulations: These include hiring and firing employees, complying with health and safety standards, workers rights’, consulting with worker councils or unions, statistical reporting of employment-related data, administering employment-related or payroll taxes, social security and pensions, or other mandatory employee benefits (such as, maternity leave and sick leave). Environment regulations: These include licenses, permits, planning and environmental impact assessments; complying with emission/discharge and hazardous substance requirements, process or product quality standards, pollution control and product regulations; environmental reporting and testing, record-keeping and day-to-day administrative requirements related to the environment, such as environmental levies and taxes; eco-labelling of products or processes. Tax regulations: These include business profits tax/corporate income tax, other taxes on capital and assets (e.g. dividend tax, property tax), sales taxes (e.g. VAT, general sales taxes), and tax deduction requests (such as PAYE income taxes) (OECD 2001: 43-44).
These areas of regulation were examined because they are consistently considered to be major areas of regulation in both domestic and international policy. Furthermore, these areas had been identified in previous studies as areas of regulation that impose easy to measure direct compliance burdens. The focus of the survey was on small- and medium-sized enterprises (SMEs), firms that employ a maximum of 500 employees. The choice of SMEs as the unit of analysis was made for two reasons. Firstly, small firms are more exposed to regulatory burdens and their performance and behaviour is more sensitive to regulation. And, secondly, a pilot survey showed large firms have more difficulty in responding to the survey as no single person or department is responsible for compliance with all regulation.
27 then distributed by mail in 11 countries: Australia, Austria, Belgium, Finland, Iceland, Mexico, Norway, New Zealand, Portugal, Spain, and, Sweden. Each firm in the sample received one questionnaire dealing with either labor, environmental or tax regulation. So no single
respondent provided information on all areas of regulation. Response rates are often a concern for postal questionnaires (Bryman and Bell, 2007). In this case the average overall response rate was 40%, ranging from 78% in Australia to only 18% in Mexico and Portugal. However, for each country at least 300 surveys were returned, so the sample size is relatively large per country. Response rates on individual questionnaire items varied and some items have many missing values. The OECD (2001) report observes that despite the low response rate on certain items the results of the survey can be extended to all firms with similar characteristics in the participating countries. However, the report does observe that quantitative measures from Mexico suffer from quality problems. The remainder of this section discusses the questionnaire items that will be used to construct the variables needed to test the hypotheses.
4.2 Costs of red tape
As discussed, red tape is costly by definition (Bozeman, 1993) and the costs of red tape come in many forms. First of all, there are administrative compliance costs in terms of the labor hours spend on complying with regulation, as measured by the SCM. Direct costs of compliance can also include investment in specialized software and the need to hire outside (legal) council. Measuring the costs of red tape in this way ignores some factors, which are generally difficult to measure. The most important of these are the indirect effects on economic behavior, for
example investment in software needed to comply with regulation may displace investment in productive resources. These indirect costs are potentially a large part of the true costs of red tape, however because such costs are difficult to measure they are not often included in research (Harrington, Morgenstern and Nelson, 2000). Another indirect effect ignored here are effects on the morale and motivation of those that have to comply with pointless or excessive regulation. DeHart-Davis and Pandey (2005) show that internal red tape can lead to feelings of work alienation among employees.
28
How many hours are spent on an average month by staff and management in your business complying with regulations
Estimate your annual computer or software expenses mostly used to comply regulations
How much money does your business spend during an average month on hiring outside services to comply with regulations
The first and third items concern monthly costs, whereas the second item concerns annual costs. So, the first and third items will be multiplied by twelve to obtain annual estimates. Furthermore the first item will be multiplied by an estimate of hourly labor costs obtained from the OECD (2001: 100) report that accompanies the survey. The estimated labor costs are on a country basis, ideally business level data would be used, however these are not available. These different costs items will then be summed to obtain an overall cost measure. In addition, the individual items will be used separately as dependent variable as well, because the
independent variables may have different effects on each of the individual cost components. Specifically, a difference may be expected between the first measure and the other two, as the first measure concerns internal costs, whereas the other two deal with external expenditures. As the cost figures are not normally distributed the natural logarithm of the costs variable will be used.
4.3 Independent variables
4.3.1 Regulatory stockpile
29 complex or length than others. Again the data will come from the OECD (2001) survey, and the items that will be used are:
During the past year, how many separate decisions or permissions did your business request from a government to comply with regulations (open question)
Despite the number of regulations, it is still feasible to comply fully (4-point scale)
The first item is a continuous variable and the responses range between 0 and 300, whereas the second variable is a discreet questionnaire item that has been coded 1 to 4. The first item will be multiplied by the inverse of the second item. This means that the weight of the number of rules and procedures will be lower, the more feasible it is to comply with them.
4.3.2 Quality of design
As regards the quality of design of the rules and regulations a business has to comply with, the survey provides several questionnaire items that are useful:
Regulations are easy to understand (4-point scale)
Regulations achieve their objectives as simply as possible (4-point scale)
Regulations are consistent with one another (4-point scale)
These items are appropriate for several reasons, first of all, easy to understand regulations, that achieve their objectives as simply as possible are the very opposite of the definition of red tape as proposed by Bozeman (1993). The Australian Office of Regulation and Review (ORR)
considers minimum regulation necessary to achieve objectives as one of the characteristics of high quality regulation (Radaelli, 2004). Furthermore, the UK Better Regulation Task Force (BRTF) lists consistency as one of the five general principles that good regulation should meet (Helm, 2006). These measures are all questionnaire items that range from 1 to 4, they will be summed and the final variable that will be used ranges from 3 to 12. The Cronbach’s alpha for this measure is 0.71, just above the common cut-off point of 0.7.
4.3.3 Predictability of application
30
Officials give definite answers (4-point scale)
It is clear who is responsible for decisions (4-point scale)
The process for appeals and complaints is clear (4-point scale)
Decisions are consistent and predictable over time and among similar business (4-point scale)
Additional or unexpected payments are not required (4-point scale)
You get the same view no matter who you contact (4-point scale)
Some of the arguments in favor of regulation that is consistent and consistently applied have been laid out in the previous section on quality. Furthermore, regulations that are consistent and predictable over time are in less danger of experience rule drift or change in
implementation or one of the other causes of “good rules turned bad” observed by Bozeman (1993). Furthermore, unexpected additional payments, or differing views from the same agency can be expected to increase the costs of compliance beyond the minimum necessary, cases which are ignored in the SCM and Djankov et al. (2002) measures. The fact that these measures ignore the possibility of additional payment due to inadequate application is another of their drawbacks (Arruñada, 2007). Definite answers, clarity as to who is responsible and a
transparent process of appeals are all aspects of accountability, and greater accountability should lead to more predictable enforcement. Accountability is considered to be a
characteristic of good regulation according to both the ORR and BRTF (Radaelli, 2004 and Helm, 2006). The individual items range from 1 to 4 and these will be summed as well to obtain a range for the final variable from 5 to 20. The Cronbach’s alpha for this measure is 0.72, also above the usual rule of thumb.
4.4 Control variables
This study will control for several firm characteristics that might influence red tape, as well as the area of regulation and the country where the firm is located.
Firm size: The size of the firm is an important control variable (Rainey et al., 1995), the
31 is smaller for a large firm than a small firm (Engel et al., 2007). The square of firm size will also be included to account for the fact that small firms often do not have to comply with all regulations. Size will be measured in terms of number of employees at the time of the OECD (2001) survey.
Firm age: The age of the firm will be included as an indication of the experience it has in
dealing with bureaucratic procedures. Experienced managers that have dealt with many different agencies and procedures are less likely to perceive the negative effects of red tape and may even know methods to reduce the objective burden (Pandey and
Kingsley, 2000). The measure from the OECD (2001) survey ranges from 1 to 3, indicating firms that are less than 2 years old, 2 till 5 years old and more than 5 years old.
Foreign ownership: Foreign ownership might mean that the firm experiences more red
tape because it has to comply with special host country regulation. Also, the firm might have to deal with red tape from the country where the owners are located.
Economic sector: Rules and regulations differ across sectors as well as across countries,
so the costs of red tape are likely to differ as well across sectors. Controlling for sector will therefore be important, the instrument used will come from the OECD (2001) survey. The survey classifies firms into 16 main sectors, a division that is not very detailed, however it is the only data available and will therefore be used. A list of sectors can be found in the appendix, table A8.
Regulatory area: The regulatory burden is likely to differ across the three regulatory
areas (employment, environment and tax) surveyed in the OECD (2001) survey. In addition regulation may be perceived as more appropriate in certain areas than others. Therefore, to control for these effects, dummy variables for the area of regulation will be included.
Country: Country dummies will be included in the analysis to control for country specific
32
4.5 Statistical model
The statistical model that will be used here estimates the costs of red tape as a function of the regulatory stockpile, the quality of design and the predictability of application. Different specifications of the costs variables will be used and different sets of control variables will be included in the final analysis. In general, the statistical model will look as follows:
(1) Where:
CoRT = The costs of red tape RS = The regulatory stockpile Qod = The quality of design
Poa = The predictability of application
Zi = The set of control variables (including sector and regulatory area dummies) Ci = A set of country dummies
ε = The error term
4.6 Method of estimation
The dependent variable for this study is a continuous measure of the costs of complying with regulation. Of the independent variables the regulatory stockpile is continues and the other two are sums of ordinal variables. Summing these variables places them closer to a continuous scale, however not perfectly so. The discrete nature of some of the variables as well as other flaws in the data might cause OLS estimators to be inconsistent. Therefore this section will investigate whether or not the data violate the crucial OLS requirements of: homoskedasticity, endogeneity, multi-collinearity and normality. These assumption are necessary to ensure that OLS is the best, linear, unbiased estimator (BLUE).
4.6.1 Homoskedasticity
The OLS estimator assumes that the regression errors have the same variance for all
33 incorrect hypotheses tests. The OLS estimator may no longer be the best, however it will still be linear and unbiased, so that if heteroskedasticity is controlled for the estimates do not change. If heteroskedasticity is present a pattern can be observed if the least squares residuals are plotted against the independent variables. These plots can be found in appendix 1. The plots for quality and predictability do not show any clearly discernible pattern. The plot for the
regulatory stockpile does show a pattern that seems to suggest reducing variance. Plots of the residuals against the control variables also indicate heteroskedasticity (for the number of employees and age of the firm). A Breush-Pagan can determine officially if the regression suffers from heteroskedasticity. This test confirms that the regression specification suffers from heteroskedasticity (chi2 = 47.54 p = 0.000). This indicates that either robust standard errors or generalized least squares should be used to correctly estimate the model.
4.6.2 Endogeneity
34 does not seem to suggest that this is a problem. The costs of red tape does not enter into the considerations in that discussion which focuses on the over-supply and demand for regulation. In conclusion, the issue cannot be settled entirely in favor of no endogeneity. Here it will be assumed that the regulatory stockpile is not endogenous, however this is certainly an issue that needs to be explored further in future research.
Theoretically, another problem could occur with regard to the quality and predictability variables. In addition to responding to the independent variables the reported perceptions of red tape will vary based on characteristics of the business (which will be controlled for) as well as individual circumstances. These individual circumstances can have an effect on perceptions of red tape (Pandey and Kingsley, 2000). The responses to the questionnaire items that deal with quality and predictability could thus be correlated with individual circumstances that are unobserved. Such common variance due to the fact that information is only received from one respondent is a characteristic problem of questionnaire data. It is difficult to control for this effect, as no data is available in the OECD (2001) dataset, and should therefore be kept in mind in interpreting the results and could form an avenue for future research. However, the values for Cronbach’s alpha show that each set of items measures an underlying construct. In addition, principle-component factor analysis of all eight items results in two retained factors. In other words these items measure two different underlying constructs and there seems to be no endogeneity problem with regard to individual circumstances.
4.6.3 Multicollinearity
Multicollinearity refers to a situation where the independent variables are perfectly correlated so that the x’s are exact linear functions of each other. If this is the case, one of the
35 estimated coefficients are inflated as a result of collinearity. The VIF for the independent
variables are all low at less than 1.5 indicating that multicollinearity is not very important for these variables. The VIF is substantially higher for the control variable of number of employees as well as for most of the dummy variables and especially for the sector dummies. This is to be expected, a dummy is one if all others are zero and vice versa. However, this is not necessarily a problem that needs solving. As the results section will show, the estimated equation has the expected signs and the independent variables are significant. The same is true for many of the control variables, with the exception of the sector dummies. If this is the case “there is no reason to try and identify or mitigate collinearity” (Hill et al., 2008: 155). Furthermore, the VIF values indicate that collinearity is likely to only be a problem for the (sector) dummies which are not of primary concern here.
4.6.4 Normality
36 number of observations is in certain cases much more limited. This has to be kept in mind if results of individual countries are analyzed.
5. Results
5.1 Descriptive statistics
Table 1 shows the mean, standard deviation maximum and minimum as well as the number of observations of the costs variable, for the entire sample and per country. The countries are ranked by mean costs of regulation. The average costs of regulation are around $390,000, the averages per country range from $8,900 to around $2.2 million. Spanish and Portuguese firms on average face the highest costs of compliance, whereas those from Australia and New Zealand face the lowest burdens. These results are in keeping with popular perceptions of an overregulated Southern Europe and lightly regulated Anglo-Saxon countries.
Table 1: Descriptive statistics cost variable, entire sample and per country (in 1000’s) Dollars
37 These results hold also if the costs are divided by the number of employees of the firm to
control for firm size. Spanish, Portuguese and Icelandic firms face high costs of regulation and those from Australia and New Zealand the least. The Scandinavian countries and Austria are at the intermediate levels, consistent with a common perception that these are countries that regulate strongly, but do not overburden their firms as much as Southern Europeans. The most surprising figure is the moderate regulatory burden observed in Mexico, however the OECD (2001) report observes that quantitative data from Mexico can be unreliable, so this figure should not necessarily be taken at face value.
Table 2 shows the descriptive statistic of the independent and control variables, for the entire sample. Also included is the measure for the number of rules and procedures that a firm has to comply with. Tables of the per country descriptive statistics for the independent variables can be found in the appendix. The average number of rules and procedures that has to be complied with is 5.02, after adjusting for the degree to which these are feasible to comply with the average drops to 2.66, so in many cases firms report that it is quite feasible to comply with regulations. The minimum number of rules that need to be complied with is 0, and the
maximum is in both cases 300, so the firm that needs to comply with the most regulations also concludes that it is not feasible to comply, which is to be expected. The averages for quality and predictability are 6.37 and 11.10 respectively, so more or less at the middle of the range for each variable. The sample covers the entire possible range for both variables. The average firm has 67 employees and is between 2 and 5 years old. Finally, 13% of the firms are fully or
partially foreign owned.
Table 2: Descriptive statistics for the independent and control variables, entire sample
Variable Mean Standard deviation Minimum Maximum
Stockpile 2.66 8.20 0 300
Rules and procedures 5.02 12.03 0 300
Quality 6.37 1.88 3 12
Predictability 11.10 2.70 5 20
Employees 67.18 104.63 1 910
Age 2.87 0.39 1 3
38 Table A2 in the appendix displays the descriptive statistics for the regulatory stockpile variable per country, ranked from the high to low. On average Swedish firms face the highest stockpile, whereas Mexican firms face the lowest stockpile. The ranking of countries along the stockpile variable differs from the ranking along the costs variable, an indication that the costs of red tape are determined by more than just the regulatory stockpile. The means displayed in this table do not represent the number of rules and procedures that need to be complied with, as they are weighted by the perceived feasibility of compliance. This might explain the relatively low means for countries that are sometimes seen as overregulated. As Radaelli (2004) argues, perceptions are compared to an implicit benchmark, the benchmark for feasibility of
compliance might be higher in highly regulated countries than in lowly regulated ones. So that the mean of the regulatory stockpile variable for countries that impose many rules and
regulations converges somewhat towards those that impose fewer.
Tables A3 and A4 in the appendix display the descriptive statistics for the quality and
predictability variables, per country. Mexico and Spain score high on both dimensions, whereas Sweden and Australia score quite low on both dimensions. In addition, Belgium scores low on the quality dimension and Portugal on predictability. Interestingly, countries that impose few rules and regulations tend to rank low on the quality and predictability dimensions, this could again be caused by low implicit benchmarks (Radaelli, 2004). Another reason could be that countries focus their limited legislative resources on either instituting a certain number of rules and procedures, or on designing ones that cover many areas and are therefore of low quality or unpredictably applied. A table of correlations between the dependent, independent and
control variables can be found in table A5, in the appendix.
5.2 Regression results
39 on the costs of red tape. Firms face more regulatory costs as they increase in size, however as the quadratic term shows, the effect is diminished as firms grow even larger (the coefficient of the quadratic term is however very small). On average, for each extra employee regulatory costs increase by 0.8%. Foreign firms also face higher regulatory costs, likely due to
inexperience with the host country regulatory system or due to special provisions in certain laws for foreign firms. These extra costs are substantial, on average firms owned by foreigners face 30% higher regulatory costs than domestically owned firms. Both these control variables are significant in most specifications. The age variable is insignificant, whereas most of the country dummies are strongly significant, and the R2 is quite high at 0.47.
The introduction of the independent variables in model 2 raises the R2 to 0.49 an increase of 0.02, so the increase in explanatory power is relatively modest. So firm characteristics and country characteristics other than the regulatory framework explain a much larger fraction of the variation in red tape than does the regulatory framework. The model confirms all
hypotheses and each will be discussed in turn. Hypothesis 1, that an increase in the regulatory stockpile faced by firms increases the costs of compliance, is confirmed. The coefficient is statistically strongly significant (p<0.001) and the sign is positive, as expected. Economically speaking, the magnitude of the coefficient of the regulatory stockpile is also significant.
Depending on the ease of compliance, on average each extra rule increases costs of compliance by between 0.53% and 2.1%.1 The average firm faces costs of compliance of around $380,000 so that an extra rule increases costs by approximately $2,000 if ease of compliance is at the highest level (a score of 4) and approximately $8,000 if it is at the lowest level (a score of 1). Figure A6 in the appendix shows how the costs of compliance change for a firm facing $350,000 in compliance costs, if the number of rules and procedures that needs to be complied with increases, for each level of feasibility of compliance.
1 If the number of rules and procedures that a firm has to be complied with is not adjusted for the feasibility of
40 Table 3: Regression results of the impact of regulatory structure on the costs of red tape, all countries
Model 1 Model 2 Model 3 Model 4 Model 5
All areas Employment Environment Tax
Stockpile 0.021*** 0.043*** 0.016*** 0.029** (0.005) (0.01) (0.003) (0.01) Quality -0.124*** -0.119*** -0.132** -0.118*** (-0.019) (-0.034) (-0.049) (-0.024) Predictability -0.031* 0.004 -0.024 -0.057*** (-0.013) (0.023) (-0.03) (-0.017) Employees 0.008*** 0.008*** 0.008*** 0.009*** 0.006*** (0.001) (0.001) (0.001) (0.002) (0.001) Employees2 -0.000*** -0.000*** -0.000*** -0.000** -0.000*** (0) (0) (0) (0) (0) Age 0.133 0.109 0.043 0.025 0.213 (0.084) (0.083) (0.125) (0.189) (0.131) Foreign 0.309** 0.284** 0.141 0.444* 0.269† (0.104) (0.102) (0.174) (0.222) (0.153) Constant 11.906*** 13.025*** 8.757*** 15.162*** 11.868*** (0.467) (0.473) (0.625) (1.279) (0.589) Observations(N) 2990 2990 1032 754 1204 R-squared 0.47 0.49 0.46 0.39 0.62 Adj. R-squared 0.46 0.49 0.44 0.36 0.61 F 92.48 93.23 30.72 27.25 64.78
Δ R-squared (Comp. to model 1) 0.02 -0.01 -0.08 0.15
41 Hypothesis 2, that higher quality of design reduces the costs of compliance, is also confirmed. The coefficient is strongly significant (p<0.001) and the sign in negative, as expected. The coefficient is economically speaking also large, at 12.4%. This variable ranges from 3 to 12, so that everything else equal, the costs of compliance for a firm that faces the lowest quality regulation (a score of 3) is 112% higher than that of a firm that faces the best quality regulation (a score of 12). For the average firm this means a difference of almost $425,000.
Hypothesis 3, that greater predictability of application lowers the costs of compliance, is also confirmed. The coefficient is not as strongly significant as the coefficients of the other two variables, but still sufficiently so (p<0.05), the sign is negative, again as expected. Economically the coefficient is also slightly smaller than that of the quality variable, but at 3.1% it is still quite high. Predictability ranges from 5 to 20, so that everything else equal the difference in cost of compliance from the lowest to the highest quality is 46.5%. For the average firm this means a difference of close to $177,000.
Columns 3 until 5 in table 3 provide a first robustness check, the data is disaggregated in the three different regulatory areas studied by the OECD. The results for hypotheses 1 and 2 hold for all regulatory areas and the magnitude of the coefficient is similar for each area. Only with regard to the regulatory stockpile variable in the employment sample is the coefficient quite different (twice as large). These results thus support the conclusion that both the regulatory stockpile as well as the quality of regulation are important in determining the costs of red tape. The coefficient for predictability is only significant in the sample of tax regulation, and much more strongly so than in the entire sample (p<0.001). Apparently unpredictability of application is only a concern in matters of taxation. Intuitively this makes sense, taxation is a sensitive area with regard to compliance with regulation, so that firms will do much to make sure that they are on the right side of the law. Unpredictability of application is therefore likely a greater source of concern with regard to taxation, where accidental violations have larger
42
5.3 Robustness tests
This section performs several more robustness tests to ensure that the results arrived at above are correct and not spurious depending on certain data or specifications. The first set of tests replicates the model above, for each of the different cost indicators separately. The following analysis tests an alternative regression equation. The next set of tests has the same
specification as the regressions above, however they are performed for each country individually.
5.3.1 Disaggregated cost indicators
Columns 1 until 4 in table 4 show the regression results for different cost indicators. The first indicator is the internal costs of red tape, the estimated costs of the time spend by employees on complying with government regulation and procedures. Hypotheses 1-3 are again
confirmed, as all coefficients are significant (p<0.001, p<0.01 and p<0.05 respectively) and the sign of the coefficient is in the right direction. In magnitude the coefficients for the regulatory stockpile and predictability of application are similar to model 2 in table 3. The coefficient for the quality of design, however, is much smaller. This indicates that for internal costs the
problem of bad design is less of an issue than for external costs. External costs, in column 2, are defined as the sum of ICT and outside council costs. They reflect money paid to outside parties, rather than time spend internally, to comply with rules and regulation. Again, hypotheses 1-3 are confirmed, all coefficients are highly significant (p<0.001, p<0.001 and p<0.05 respectively) and the sign of the coefficients are in the right direction as well. The coefficients for the
43 Table 4: Regression results of the impact of regulatory structure on the costs of red tape for different cost specifications
Model 1 Model 2 Model 3 Model 4 Model 5
Internal costs External costs Software costs Outside council External costs as costs independent variable
Stockpile 0.021*** 0.015*** 0.011† 0.013*** 0.016** (0.006) (0.004) (0.006) (0.003) (0.005) Quality -0.043** -0.083*** -0.091*** -0.069*** -0.016 (-0.014) (-0.018) (-0.028) (-0.017) (-0.015) Predictability -0.025* -0.025* -0.022 -0.030** -0.016 (-0.01) (-0.012) (-0.017) (-0.012) (-0.01) External costs 0.201*** (0.018) Employees 0.006*** 0.007*** 0.006*** 0.007*** 0.005*** (0.001) (0.001) (0.001) (0.001) (0.001) Employees2 -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** (0) (0) (0) (0) (0) Age 0.065 0.074 0.025 0.048 0.07 (0.059) (0.073) (0.086) (0.067) (0.063) Foreign 0.200** 0.181* -0.022 0.143 0.162* (0.074) (0.092) (-0.138) (0.094) (0.075) Constant 8.190*** 13.445*** 8.355*** 12.642*** 5.463*** (0.31) (0.362) (0.426) (0.405) (0.417) Observations(N) 2948 2621 924 2535 2579 R-squared 0.23 0.61 0.65 0.62 0.26 Adj. R-squared 0.22 0.61 0.63 0.62 0.25 F 23.43 130.56 56.22 151.36 24.66 Δ R-squared (Comp. to model 1, table 3) -0.24 0.14 0.18 0.15 -0.21
