# Active Voices: An Empirical Study of the Effects of Small Firm Lobbying on New Firm Survival

Viktorie Sevcenko[1], Department of Philosophy, Politics and Economics, University of Warwick

## Abstract

This study examines the presence of direct negative effects of the activities undertaken by the Federation of Small Businesses (FSB) on the survival rates of small startups, testing the hypothesis that the established small firms may be 'crowding out' startups by proposing policies aimed at helping themselves and thereby increasing competition for newly formed enterprises. The paper uses a combination of data sources, including publicly available data from the Office for National Statistics, as well as data collected by the author from secondary and tertiary sources. Employing the Ordinary Least Squares method and the non-parametric Cox Proportional Hazards Model, the study regresses two versions of an original index of FSB's lobbying activities on individual firm chances of survival and small firm group dynamics. The study finds partial support for the potentially detrimental effects of small firm lobbying on startup survival rates and post-entry performance.

Keywords: Interest groups, lobbying, FSB, firm, survival.

## Disclaimer

This work contains statistical data from ONS which is Crown copyright and reproduced with the permission of the controller of HMSO and Queen's Printer for Scotland. The use of the ONS statistical data in this work does not imply the endorsement of the ONS in relation to the interpretation or analysis of the statistical data. This work uses research datasets which may not exactly reproduce National Statistics aggregates.

## Introduction

The small firm sector constitutes the bulk of all enterprises in the UK, employs more than half of the workforce and contributes disproportionately to the creation of new products and services. The importance of these firms is rarely overlooked in government, but given their multitude and variety, representing the entire sector's interests to policymakers is no simple task. A typical small business interest group will therefore poll its members on their concerns and aim to alleviate them. However, as Olson (1971) showed in his The Logic of Collective Action, large interest groups could only exist if their members are offered exclusive benefits in return for their contributions, and the lobbying is carried out as a 'by-product', financed by the contributions. This in turn means that the firms joining such organisations may be doing so more for the benefits received rather than being involved in the shaping of policy, but to which extent are the policies of these groups nevertheless influenced by the opinions of their members? And why is this question important?

The case of small business lobby groups is peculiar because these groups attempt to represent an entire sector of the economy, where frequently the only common factor is firm size. Moreover, despite their frequently large memberships, most small firms do not belong to any interest group or trade association (Storey, 1994; Grant, 1993). Nevertheless, the policies proposed and lobbied for by, for example, the Federation of Small Businesses, will, if accepted by the government, affect all small firms. The extent to which these policies are desirable by at least the majority of small enterprises is therefore an important concern.

What, then, is the effect of membership opinion on the policy-making process in a small business lobby group? The case examined in this paper is that of the Federation of Small Businesses (FSB), the most representative small firm interest group in the UK today. With membership of over 210,000 businesses, this group's mission is to 'be and remain the largest and most effective organization promoting and protecting the interests of the self employed and small business owners within the UK' (FSB, 2011: para. 5). In addition, the mission statement is accompanied by six principal objectives (FSB, 2011):

1. To recruit continuously at a high level;

2. To retain as many members as possible;

3. To promote the interests of members;

5. To publicize the benefits of self-employed and business ownership;

6. To identify the continuously changing needs of members.

Therefore, based on these objectives, the goal of the FSB is to support all small firms, regardless of their age, since the FSB also aims to publicize the benefits of self-employment and business ownership, presumably to attract new market entrants. The group does not pose any restrictions on new members, except that they should employ fewer than 250 people, and it is constantly recruiting. Furthermore, the FSB pledges to promote the interests and protect the businesses of its all its members. However, to what extent can a group protect the interests of established business and yet encourage new market entrants and members, who are likely to pose competition for the existing firms? This paper will therefore examine the possibility that the FSB focuses more on the established companies at the expense of new entrants.

## FSB Membership Structure and Literature Review

New firms are often defined as those in existence for less than two years since establishment, or at least no older than five years (Storey, 1994). Only 8% of the FSB's members have been in business for one to two years (FSB, 2010), whereas 73% are at least six years old, with over a quarter of the member businesses being over 21 years old. The reason lies partly in the scarcity of capital and its higher opportunity cost for new firms, and partly in the overall small proportion of new registrations compared with the entire private sector (ONS, 2010) – new firms represent a larger proportion of the FSB than of the UK small firm sector in general. However, the problem also lies in the need for the FSB to retain its members and their contributions. If a new firm can only be seen as a start-up, and hence has special needs for its first two years of existence, then its status will change as it ages, but the FSB's retention goals are usually as high as 80% each year (FSB, 2009). This means that in order to have a strong representation of new business views, the FSB must recruit startups at an increasingly faster rate, but given that its growth has been relatively stable and slowing down in the recent years (FSB, 2009), this has not been possible.

As a result, the views represented by the members of the FSB and hence the FSB itself are increasingly those of the older and established firms. This may not be a serious problem, as long as the policies proposed by the FSB do not have significantly different effects on the two groups. However, the first two years after establishment are also the most difficult in terms of survival, and a time when companies need substantial support. In comparison to their more experienced counterparts, new firms experience much steeper learning curves and relatively higher costs of compliance with regulation (Audretsch, 1995; Mata and Portugal, 1994; FSB, 2010). They may therefore require other policies and services than those desired by the older firms, even if they are of the same size. Moreover, new entrants pose a threat to existing small firms (Audretsch, 1995; Mata and Portugal, 1994), meaning that existing FSB members may not welcome policies aimed to significantly increase startup chances of success.

The idea that there exists an optimal size and type of membership for an interest group to operate effectively, and that the provision of club goods helps establish that size, dates back to Buchanan (1965) and his economic theory of clubs. The central idea here, however, differs from that proposed by Buchanan and subsequent theorists of club theory. Even if generalized, the FSB's policy goods are not excludable; they usually affect all small firms in the UK. At the same time, its exclusive goods – the services it provides for its members – are non-rival and not subject to diminishing returns with a rising number of members. In fact, services such as networking events only increase in value with additional members. However, the existing FSB members may nevertheless be 'crowding out' new establishments. I therefore aim to test the following hypothesis: provided that the government is neutral as to which policy to implement, the policy proposals of the FSB are diminishing the chances of successful entry by new establishments.

## Data

Two groups of data were collected: data on FSB's policies and data on small and medium-sized enterprises (SMEs).

##### FSB's Lobbying Activities

Perhaps the most difficult aspect of this research has been acquiring data on firm and industry lobbying activities ('lobbying' in this context is used very broadly, relating to a number of activities a firm or a representative body could potentially undertake in relation to government; this includes provision of information, direct financial support etc.). In contrast to the United States, there are no official records of lobbying in the UK andno official registration of lobbyists, and lobbying bodies prefer to keep this information private. Therefore, following the methods used by Bernhagen and Trani (2010) and Harris (1999), data was collected from all published materials on the Federation for Small Businesses, including newspapers, FSB's publications on their website and all forms of electronic press. The exact collection process is described in Appendix 1, but a short summary follows.

Printed materials by third parties were searched for using Factiva (an online newspaper publication search engine), implementing an adopted version of the search methods used by Bernhagen and Trani (2010) and focusing on the years 1999 to 2008 to match the available firm-level data. The matches were filtered through to isolate specific instances reporting FSB's involvement in a given issue. Duplicates were eliminated, and the issues were allowed repetition only if the FSB added a singnificant new suggestion in each.

Additionally, publications from the FSB's website and their press releases were used in order to obtain further evidence of their efforts and to examine many of the proposals in greater detail. Finally, other sources of online and print news reports or academic studies providing additional evidence were used to obtain as broad a picture of the organisation's activities as possible. In total, 391 suggestions, campaigns and initiatives undertaken by the FSBwere identified in the decade between 1999 and 2008, and classified according to the topic they referred to and the year they were raised in.

The total of 20 topics were not identified prior to searching for data, but were instead used to classify the 391 items found. The main idea of each item was usually clearly stated in the newsreport and/or an FSB publication, but grouping these into 20 distinct categories was less straightforward. The initial method was to use the classification the FSB uses for its policies, which lists 13 topics, but these were later expanded to take separate account of issues mentioned by the FSB and/or existing literature as especially important for new firms. The EU category was eliminated (detailed explanation of how each topic was identified and constructed can be found in Appendix 1).The following 13 topics are used by the FSB:

No. Topic No. Topic
2. Crime 9. Regulation
3. Employment 10. Rural Affairs
4. Environment 11. Skills
5. European Union 12. Tax
6. Health & Safety 13. Trade & Industry
7. Local Government

Table 1: Topics addressed by the FSB

The final 20 topics are listed in Table 2, with the third column containing the number of times each topic was addressed by the FSB over the years.

Topic No.
1 VAT regulations 8
2 Other tax-related regulations 23
3 Employment 48
4 Self-employment & incorporation 2
5 Environment 12
6 Rent, lease & business rates 17
7 Property purchasing & regional planning 4
8 R&D and IP support 6
9 Crime 25
11 Infrastructure, transport & fuel 51
12 Skills & staff training 26
13 Procurement 13
15 Telecom, post & other communication services 16
16 Rural affairs 5
17 Fair access & competition policy (Local Government) 22
18 Regulation & bureaucracy 45
19 Health & safety 10
20 Insolvency & bankruptcy 6

In addition, the proposals were classified according to the magnitude of the likely short-term effects they would have on small firms if adopted. To determine the magnitude, each proposal was analyzed and its main goal decided using the following scale:

Magnitude Main message in the proposal
1 Raise awareness of the issue; Review an existing policy; Encourage and Promote a given policy
3 Amend a policy; Increase investment/decrease taxes in line with inflation; Provide active business support
4 Introduce an entirely new policy/Abolish an existing policy; Direct investment/tax red uctions above inflation rates
5 Introduce an entirely new, large-scale project aimed specifically at small firms

Finally, the magnitudes were added to obtain two yearly figures: (1) a total sum of magnitude to approximate the level of activity by the FSB in that year, obtained by adding up the magnitudes across all topics for each year; and (2) an average 'magnitude' per proposal per year, obtained by dividing (1) by the number of proposals in that year. These values were then used to create an index of FSB activity.

This data was then used to construct an index of FSB' s activities. The index was chosen as a method because it allows data to be compiled into a potentially meaningful yearly figure, and since the goal is to assess the impact of FSB's overall effort on the small firms, a single yearly figure is preferable. In addition, only FSB's proposals were taken into account, leaving out the opinion of government leaders and other small firm interest groups because the goal of the was to analyse the effect of FSB's lobbying generally, including those policies that were proposed and not adopted.

The indices were constructed using a cumulative value of each of the two magnitude measures obtained by adding each year's value to the two values in the previous years:

$\gamma_t = \chi_t + \chi_{t-1} + \chi_{t-2}\qquad(1)$

Where $\gamma_t$ stands for the effect of FSB's activities in year t, and $\chi_t$, $\chi_{t-1}$ and $\chi_{t-2}$ stand for the total value of activities undertaken in years t, t-1 and t-2. This method was used to account for the frequently long time it takes for a new proposal to be adopted. The indices were then constructed by taking 2002 as the base year (the year of the first observed exit) and obtaining the relative values of activity in each of the subsequent years. Table 4 depicts the index values that were obtained. These indices were then regressed along with control variables on a number of datasets using OLS and the Cox Proportional Hazards methods.

Year Total Magnitude Index Average Magnitude Index
2001 88.5 96.50
2002 100 100.0
2003 50.0 107.2
2004 54.4 89.50
2005 65.7 83.00
2006 58.5 85.00
2007 48.4 105.5
2008 52.8 101.7

Table 4: Yearly index values

##### Tailored Index

The rationale behind creating a tailored index was the following: each firm exhibits some characteristics, which in theory should predict the firm's sensitivity to a given topic lobbied for by the FSB. For instance: firms located in areas with greater incidence of crime should benefit to a greater extent from policies to reduce crime than should firms in relatively safe areas. Alternatively, firms spending a greater proportion of their turnover on business rates should benefit more from policies to reduce these than firms paying a smaller proportion. In this fashion, six of the 20 topics addressed by the FSB were selected to approximate the potential importance of each proposal for each firm:

1. Crime;

3. Number of employees;

4. Proportion of turnover spent on telecom services;

5. Proportion of turnover spent on transportation;

6. Average expenditure on RandD within the industry.

The rest of the topics were either unobservable in this fashion or considered applicable to all firms in very similar ways. Therefore in addition to a tailored value, each firm was assigned the overall index for the year, and the two values were summed for the total tailored index. Each of the firms was then given a categorical rating of its 'sensitivity' depending on the proportion of its turnover spent on the given product or service, or given its location or industry type. Table 5 presents the data:

Topic Calculation Method Used Categorical Outcomes
Crime Determined by Location 0 - 4; each interval representing a standard deviation from the mean (level 3)
Business Rates Business Rates Exp / Turnover 0 - 4; with 0.5 for missing values
Number of Employees No. of employees at startup 0 - 4; with 0.5 for missing values
Telecom Services Telecom Services Exp / Turnover 0 - 4; with 0.5 for missing values
Transportation Transport Exp / Turnover 0 - 4; with 0.5 for missing values
R&D Expenditure Average R&D Exp in Industry 0 - 4; with 0.5 for missing values

Table 5: Categorisation of topic importance

The Tailored Index was composed in the following way: for each year, the value of the index was adapted to each firm by multiplying the values of Total Activity for the topic, by the categorical value the firm possessed for that topic, and obtaining the sum of these values and the original value for the year of each type of index. For example, a firm with the following values: untailored index value in 2002 = 100, Crime = 2, Business Rates = 1, Number of Employees = 3, Telecom Services = 2 and RandD Expenditure =0.5, would have a total tailored index for the year 2002 equal:

$Index(total)_{2002} = (2 \times 16) + (1 \times 35) + (3 \times 15) + (2 \times 21) + (0.5 \times 5) + 100 = 256.5\qquad(2)$

Where the values 16, 35, 15, 21 and 4 reflect the level of activity by the FSB on that given topic in that year, and 100 reflects the overall cumulative level of activity for this year and two previous years. This new index was then considered 'tailored' to each firm, attempting to evaluate the extent to which the FSB's activities were affecting it in particular.

##### Firm-Level Data

The data on small firms was obtained from three sources: the Office for National Statistics website, the Virtual Microdata Laboratory run by the Office for National Statistics, and the FAME database. Aggregate data on small and medium sized firms, their growth and their multitude was acquired from the Office for National Statistics website. The dataset contains all UK firms across all registers, and records their size, birth and death, location, industry type. For the purpose of this analysis, data was obtained on groups of companies arranged by the number of employees they employ. The firms' life spans were regressed on the indices.

Firm-level data used to test the Tailored index was obtained from the Virtual Microdata Lab. Two datasets were combined for this purpose: the Business Structure Database (BSD) and the Annual Respondents Database (ARD2). Both of them were supplied by the ONS and were accessible in full length only through a few select data centres, to access which, training was required. Both datasets are extremely detailed and contain more firm-specific data on small firms than any alternative source.

The BSD dataset contains enterprises registered for VAT and/or PAYE, covering nearly 99% of the private enterprise output in the UK. It provides clear data on firm birth (new firm registrations) and death (firm de-registrations), without double counting (the same business units are not counted twice, even if they fall into a period of inactivity or are acquired by another firm). The database further contains information on each firm's location, itsclassification code (SIC 2003), the number of local and reporting units[2] within the enterprise, and the number of employees. The data includes all companies that have ever entered the database, even if they are no longer alive. It therefore serves as the ideal base dataset for this paper.

The ARD2 is a panel dataset designed for a number of purposes, containing information on over 900 different variables on most firms in the UK. It is collected yearly from a subset of the entire firm population, with thresholds for different firm sizes[3]. However, it does provide approximate values, usually using the last available year, for the other firms who were not sent a questionnaire in a given year, to maintain the database complete. For these reasons, this dataset is the ideal way to track important variables for specific firm populations.

However, as with any data source, there are some limitations. The datasets do not merge easily, and given the different collection methods of each, the variables are not always comparable. The ARD2 was not collected for the purpose of longitudinal research, and firms entering the database and exiting it do not always correspond to new firms and failing firms. Instead, a wave of new firms might be an expansion of the coverage of the ARD2 and not a reflection of an increase in the number of firms.

Moreover, neither of the two datasets are able to classify exit, which is problematic given that (as discussed earlier) much of firm de-registration is the result of firms merging together, firms being acquired by others or the entrepreneurs finding another, more lucrative position. Nevertheless, not only are many studies affected by this same problem (Mata and Portugal, 1994; Audretsch, 1995), but one may also argue that if a firm exits and is not merged with or bought by another for a sum far beyond the market value of its assets, the firm was not profitable enough to offset the entrepreneur's opportunity costs and, therefore, failed in at least the expectations the entrepreneur had for it. Further limitations of the datasets are the fact that extremely small firms are only seldom sent the annual questionnaires, which means that data on these may be dated.

From the two datasets, the BSD and the ARD2, a sample of 17,195 manufacturing firms was obtained[4], with incorporation dates of 1997 and 2002 – 2008 inclusive, and observed failures between 2002 and 2008 inclusive. These years were chosen to test the hypothesis that older firms – those incorporated in 1997 – were affected by the index differently than new firms.

The dataset used here was a combination of the BSD and ARD2 for the years 2001 to 2009. Table 6 presents the births and deaths in each cohort:

Year Births Deaths
1997 6664 3438
2002 1547 627
2003 1690 728
2004 1529 798
2005 1404 850
2006 1481 1052
2007 1435 1241
2008 1264 1222

Table 6: Births and deaths by year (Source: ONS)

Of the 17,195 firms in total, by the end of 2009, 10,191 of them failed. This represents a 59.7% failure rate, of which almost 20% were firms incorporated in 1997[5]. Each cohort was observed across the following variables: number of employees, wage rates, turnover rates, crime rates in the area[6], business rates, expenditure on telecom services, expenditure on transportation and the average expenditure on RandD in the each firm's industry. These variables were chosen in accordance with the topics for which the FSB campaigns (listed in Table 1). Not all topics were regressed as control variables due to them being either unobservable in this manner or concerning all firms in very similar ways. Therefore, only those topics, which could benefit or harm specific companies disproportionally were included.

## Models and Methods

To estimate the changes in the employment composition and proportion of small firms the OLS estimator was used. However, simple OLS and other linear methods of estimation are insufficient when it comes to analysing the length of time for which firms remain in a market (Mata and Portugal, 1994). These models do not take into account censored events, and given the nature of survival analysis, some firms (observations) will inevitably be censored due to the event (failure) not occurring for all within the given time frame (Mata and Portugal, 1994). To account for this, the paper will use survival analysis, utilising the non-parametric Cox Proportional Hazards Model, a common model for this type of analysis. The model has the following function:

$h(t) = \lim_{\Delta{t}\rightarrow0}\left[\frac{(t\leq{T}\leq{t}+\Delta{t}\mid{T}\geq{t})}{\Delta{t}}\right]= \frac{f(t)}{S(t)}\qquad(3)$

Where T is the duration of the firm's existence, f(t) is the probability density function and S(t) is the survival function. The estimated Cox Proportional Hazards Model (Cox, 1972) will then be:

$\ln{h(t)} = \ln{h_0(t)} + X\beta\qquad(4)$

Where h(t) is the baseline hazard function, X is a vector of explanatory variables and $\beta$ is a vector of parameters. The baseline hazard is estimated by setting X equal to 0. Accordingly, the effect of a unit change in a covariate is to produce a constant proportional change in the hazard rate (Mata and Portugal, 1994, p. 231). Put another way, the effect of each covariate is to act multiplicatively on h0(t). The non-parametric nature of the model allows one not to have to define a specific distribution of its baseline hazard, thereby not risking misspecification (Cox, 1972; Mata and Portugal, 1994).

However, the assumption made to allow this is that of 'a constant effect of a covariate on survival over the entire follow-up period' (Zhang and Huang, 2004: 267). This means that each covariate will have the same effect on the dependent variable throughout the entire length of analysis. This, of course, may not be true for all firms, but appropriate in this analysis because the examined variables are not likely to change in their impact throughout time or are assumed to have similar changes in their effect on all firms.

## Results and Discussion

##### Cox Model Regressions

These calculations were performed for all firms in the sample, and the resulting index was regressed on all entry cohorts between 2002 and 2008 simultaneously, using the Cox Proportional Hazards model of survival. Table 7 therefore presents the results for the survival hazard facing newly incorporated firms only[7]:

Model Specification I II III Total Activity Index 0.0610*** 0.0216*** Magnitude Index 0.0041*** 0.0709*** No. Observations 17014 17014 17014 Prob > chi2 0.0000 0.0045 0.0000

Table 7: Cox Model regression results (Source: ONS)
[*** Significant at the 1% level]

The three regressions show a very significant trend in the predicted direction. Instead of decreasing the likelihood of premature exit, lobbying by the FSB seems to increase the likelihood, therefore supporting the original hypothesis that policies promoted by the FSB favour existing and established small firms rather than new firms specifically. In fact, there seems to be evidence of 'crowding out' of new firms, whereby the FSB's policies cause new firms to exit the market prematurely. According to the first two regressions, a single unit increase in the value of the index leads either to a 6.96 percent increase or a 0.408 percent increase in the exit hazard for a newly incorporated firm.

Moreover, as the third model suggests, the Total Activity Index in the first model captures some of the effect of the Magnitude Index in the second model, which means that the more the FSB lobbies for specific causes, the greater the negative impact on new firm survival. These results may therefore suggest that the greater the effort by the FSB on each topic, the greater their achievements on the subject and the worse the outcome for each new firm.

However, besides potential flaws in the design of the index, the indices may also be inversely correlated with firm performance: the higher the small firm mortality rates and the more difficult small firms find it to make sufficient profits, the more likely will the FSB be to lobby to improve the situation. There may, therefore be a problem of reverse causality, although the cumulative nature of the index does help offset this hazard somewhat.

##### OLS Group Regression Results

As the tailored index is positively related to failure for individual firms within their first year of operation, it may be of interest to examine the impact of the index on existing small firms over time. To check for these effects, the composition of small firm cohorts is examined. The data obtained from the ONS website on all registered small and medium-sized firms is separated into groups by the number of their employees, each group employing zero, between zero and 10, between 10 and 50, and so on. Including economy-wide control variables, the following questions were examined: 1) changes in the proportion of the total number of firms in economy, that fall within each size band and 2) changes in the proportion of the total workforce that each group employs.

To maintain consistency, the groups were only analyzed between the years 2002 and 2008. The economy-wide control variables used were the unemployment rates, Gross Value added rates and the CPI index, following the previous study findings summarized in Storey (1994: p.68). The predicted effects of the indices are as follows: the indices should cause an downward trend in the number of very small firms (employing 0 and less than 10 employees) because these groups contain the vast majority of startups, and an upward trend in the proportion of the workforce employed by the existing firms, reflecting their growth. For this analysis, the Ordinary Least Squares method was used and the results obtained[8] are presented in Table 8.

Prop. Enter
Independent Variables Dependent Variables
0 10 50 250
CPI Index -1.0866 1.3808* 0.3346 0.0269
Gross Value Added 1.0764 -1.3587* -0.3166 -0.0187
Unemployment Rate -0.7763 1.2657 0.4132 -0.0354
Constant 81.6835* 11.5567 -0.5770 -0.1978
Adjusted R-squared 0.9672 0.9943 0.9211 0.8764
F(4, 1) 37.82 219.11 15.60 9.87
Prob>F 0.1213 0.0506 0.1874 0.2339
No. Observations 6 6 6 6
Prop. Employed
Independent Variables Dependent Variables
x=0 0<x<10 10≤x<50 50≤x<250
CPI Index -0.0126* 0.0953** 0.0728* -0.1264
Gross Value Added -0.8344 -1.7915* -1.3457 2.6898
Unemployment Rate 0.8154 1.6589* 1.2528 -1.8183
Constant -0.4539 -1.8168 -1.2150 -3.3540
Constant 23.0625 38.2933 29.1854 -52.5609
Adjusted R-squared 0.9712 0.9690 0.9290 0.6674
F(4, 1) 43.11 40.07 17.36 3.51
Prob>F 0.1137 0.1179 0.1779 0.3783
No. Observations 6 6 6 6

Table 8: OLS regression results
[** Significant at the 5% level and * Significant at the 10% level]

As shown, the regressions did not yield significant results. There is only a slight negative trend among the smallest firms, but those employing fewer than 10 employees show a positive relation to the index. Total Activity Index seems not to affect the proportion employed, except for the slightly negative effect on the smallest firms, reflecting the downward in employment or outright failure among these firms. For the number of firms as proportion of total enterprise population in the private sector, the larger firms reaped greater benefits, suggesting that either many firms have grown from having less than 10 employees to employing more, or new, large firms entered.

So what does this mean? The evidence presented here does not show a significant effect of lobbying activities on cohort dynamics, and must be approached with care, for the index is based on secondary and tertiary sources and the aggregate data may conceal serious differences among industry and geographical groups. Moreover, the OLS method does not necessarily show a causal relationship, and again, there may be inverse correlation between firm performance and FSB activities, although the index does attempt to account for the last possibility by using cumulative values including previous years. However, the evidence so far points at the possibility that the FSB's activities are poor predictors of the success of any given firm. The effects of the legislation for which the FSB lobbies may not necessarily be strong enough to push an able firm out of business nor keep a poorly performing one in business. Moreover, being a large organisation serving an increasingly heterogeneous group of firms, the FSB is not always able to accommodate all wants equally efficiently.

## Conclusion

The hypothesis tested in this study stated that small business interest groups may pursue policies that are detrimental to the success of new firms. The analysis suggested that this outcome may be inevitable due to the nature of the problem: new firms are only new for a short time, but the structure of the FSB, and most other interest groups, is such that member retention is key to the organization's ability to collect contributions and finance its lobbying activity as a 'by-product'. One would therefore expect to find a negative effect of the FSB's activities overall on the success of new establishments, provided that the government is neutral as to which policy to implement. The empirical results have yielded some evidence in support for this hypothesis, somewhat confirming these expectations.

However, does this mean that the FSB is hostile to new firms? The answer is probably 'no', but the theory of small firm lobbying may need rethinking nevertheless. Small and established enterprises do not necessarily posses the same traits as small startups. Their small size may, among others, be the result of the owners' choice, but their age reflects extensive market experience and company learning, regardless of whether or not the learning was active or passive. Moreover, new entrants are a source of competition for older firms, and the purpose of business lobbies is to protect the interests of its members. The policy wishes of these two groups may therefore not only differ, but even clash enough to justify the creation of a separate group representing startups only. However, unlike with existing members, who are registered and able to contribute, nascent entrepreneurs and startups are not even easily identifiable, let alone possessing sufficient funds for contribution. And in this case, the possibility of lobbying as a by-product may not be realisable.

Further research would need to examine in greater depth the differences in policy preferences of startups and older firms, relating these back to the activities of the interest groups. In light of this knowledge, the interest groups' activities could be split into groups according to their likely positive or negative impact on new versus established firms. This would yield some preliminary insight into the possible effects of the proposed policies on small firms. Following this, the process by which policy proposals by the small business interest groups are received and reviewed by the respective authorities would need to be addressed, since the effect of lobbying on the small firms' wellbeing is not direct, but transmitted by a range of intermediaries. Furthermore, data on lobbying activities could be collected from a broader range of sources, including interviews and surveys of the interest group members and employees, as well as non-member small business owners. Finally, the reasons why companies join and exit small business interest groups needs to be investigated in greater detail.

## Acknowledgements

I would like to thank Prof. Wyn Grant and Prof. Dennis Leech for their support and very helpful suggestions, and Dr Catherine Hanley without whom this project would not have taken place. Finally, I would also like to thank the Reinvention Centre for enabling me to undertake this study and supporting me financially.

## List of Tables

Table 1: Topics addressed by the FSB

Table 4: Yearly index values

Table 5: Categorisation of topic importance

Table 6: Births and deaths by year

Table 7: Cox Model regression results

Table 8: OLS regression results

## Appendix 1

##### Search terms employed to collect data for the index
All of these words At least one of these words This exact phrase
federation of small business * concerned policy legislation regulation suggest condemned proposed agreed disagreed urged criticized FSB
federation of small business * concerned policy propose proposal FSB
ederation of small business * opposed policy legislation regulation suggest condemned proposed agreed disagreed urged FSB
federation of small business * proposed policy legislation regulation suggest condemned proposed agreed disagreed urged FSB
federation of small business policy legislation regulation suggest condemned proposed agreed disagreed urged FSB
federation of small business proposal legislation MP tackle policy address intend plan solve promise FSB
federation of small business proposal legislation MP tackle policy address intend plan solve promise small business survival performance output growth FSB
federation of small business policy legislation regulation FSB
federation of small business regulation proposal address MP Government regulatory law FSB
federation of small business government announce introduce policy shift FSB
##### Index topic construction

The European Union was eliminated as a category, and all policy proposals in this group were distributed among the 20 groups used, since this distinction should not in theory affect small and new firms differently. Tax was split into two categories: VAT tax and other tax-related regulations, since crossing the VAT payment threshold has been identified as an important step for new businesses by the FSB (FSB, 2010). In addition, seven new groups were created. Self-employment and incorporation regulation was added because it was identified as one of the key and most difficult steps for a start-up in the first year or existence (Storey, 1994). Rent, lease and business rates regulation was added because these costs represent an especially high proportion of turnover for start-ups (Storey and Greene, 2010). Property purchasing and regional planning were added as a counterpart of the Rent, lease and business rates regulation topic, since if the company is not renting premises, it must own them, or do neither. In the latter case, this decision may be affected by policies on each of the two options.

The RandD credits and intellectual property rights topic was added to isolate the FSB's efforts to promote and ease the appropriation of innovation by small firms and startups, often seen as especially important for the creation of new ventures (Dodgson, Gann and Salter, 2008). Infrastructure, transport and fuel group was added because small firms are frequently dependent for their existence on transport and these costs form a large proportion of their turnover (FSB, 2005; FSB, 2000). Access to finance and cash-flow/late payment problems was added because financing is a crucial topic for firms of all sizes, and especially the small and new firms (Storey, 1994). Finally, Insolvency and bankruptcy regulation was added as a control because it has been shown to influence the way in which firm exit is measured and its consequences (Storey, 1994; Storey and Greene, 2010).

##### Classification of Magnitude

The following terms were used as keywords in order to classify a proposal in terms of its likely effect on a firm (if the proposal contained one of these words or a majority of the words within each section, it was classified accordingly):

###### Level 1

The FSB would suggest that government should:

"Review", "Define", "Work together", "Communicate", "Increase Communication", "Cooperate", "Investigate", "Do more" (without specific suggestions), "Consider", "Discuss", "Explore", "Increase Attention to", "Reconsider", "Increase Awareness of", "Monitor", "Encourage"

Or the FSB expressed:

"Concern", "Dismay", "Shock" (all without specific policy proposals)

###### Level 3

The FSB would suggest that government should:

"Modify", "Add", "Enhance", "Increase" (with specific instructions, usually concerning funding or support), "Strengthen", "Introduce" (usually in addition to existing legislation)

###### Level 4

The FSB would suggest that government should:

"Compensate", "Develop", "Introduce" (new legislation or project)

Or the FSB would suggest that firms should not:

"Be charged", "Pay" (for something in particular), "Have to Register" (for something particular or in a certain place/with a certain authority), "Have to Comply" (with something in particular)

###### Level 5

The FSB would suggest that government should:

"Elect a New Minister", "Create a New Committee", "Open or support a New Bank"

## Notes

[1] Victoria graduated from the University of Warwick in July 2010 with a degree in Philosophy, Politics and Economics and is currently reading for an MSc at Imperial College London. She is hoping to continue her studies next year and pursue a PhD in Management.

[2] Local units are the smallest possible business units, and data collected on them from the reporting units, larger groups of local units (if there are any). The data used in this study is on the local-unit level.

[3] Different percentages of the total population are surveyed each year, with decisions made depending on firm size and the industry it belongs to. Further detail on coverage may be obtained upon request.

[4] Only manufacturing firms were used in this sample due to the tailored index being too specific to be comparable across all industries.

[5] All failure rates are those observed in the year subsequent to their incorporation or in 2002, for the 1997 cohort.

[6] Values for crime rates were obtained from the publicly available data on the ONS website.

[7] Firm size, industry classification, location, average industry entry and exit rates, and proportion of turnover spent on each of the six key variables were used as control variables.

[8] Due to the low significance of the effects of the magnitude index for any of the groups, only total activity index is presented.

## References

Audretsch, D. (1995), 'Innovation, Growth and Survival', International Journal of Industrial Organization, 13, 441-57

Bernhagen, P. and B. Trani (2010), Who Gets What in British Politics? Analyzing Media Reports on Lobbying around Government Policy, 2001-2007, Aberdeen: University of Aberdeen, available at http://www.abdn.ac.uk/~pol209/Bernhagen_Trani_MPSA09.pdf, accessed 12 September 2010

Buchanan, J. M. (1965), 'An Economic Theory of Clubs't, Economica, 32 (125), 1-14

Cox, D. (1972), 'Regression Models and Life-Tables', Journal of the Royal Statistical Society, 34 (2), 187-220

Dodgson, M., D. Gann and A. Salter (2008), The Management of Technological Innovation, New York: Oxford University Press.

Federation of Small Businesses (2000), UK Manifesto 2000, available at http://www.fsb.org.uk/, accessed 11 September 2010

Federation of Small Businesses (2005), Better Transport, available at http://www.fsb.org.uk/, accessed 10 September 2010

Federation of Small Businesses (2009), Annual Report 2009, available at http://www.fsb.org.uk/, accessed 10 March 2011

Federation of Small Businesses (2010), 'Voice of Small Business' Annual Survey: Report of Key Findings, available at http://www.fsb.org.uk/, accessed 20 September 2010

Grant, W. (1993), Business and Politics in Britain, 2nd ed., London, Macmillan

Grant, W. (2000), Pressure groups and British Politics, Basingstoke, Macmillan

Harris, C. P. (1999), 'Lobbying and Public Affairs in the UK: The Relationship to Political Marketing', unpublished PhD thesis, Manchester Metropolitan University

Jordan, G. and D. Halpin (2003), 'Cultivating Small Business Influence in the UK: The Federation of Small Businesses' journey from outsider to insider', Journal of Public Affairs, 3 (4), 313-25

Jordan, G. and W. A. Maloney (1997), 'Acoounting for Sub-governments: Explaining the Persistence of Policy Communities', Administration and Society, 29 (5): 557-83

King, R. and N. Nugent (eds.) (1979), Responsible Rebels, London: Hodder and Stoughton

Mata, J. and P. Portugal (1994), 'Life Duration of New Firms', The Journal of Industrial Economics, 42 (3), 227-45

Olson, M. (1971), The Logic of Collective Action, London: Harvard University Press

ONS (2010), VAT Stats Tables 1994 to 2007, available at http://stats.bis.gov.uk/ed/vat/index.htm, accessed 10 November 2010

Storey, D. J. (1994), Understanding the Small Business Sector, London: Thomson Business Press

Storey, D. J. and F. J. Greene (2010), Small Business and Entrepreneurship, London: Pearson

Verbeek, M. (2009), A Guide to Modern Econometrics, 3rd ed., West Sussex: John Wiley and Sons Ltd.

Zhang, H. and C. Huang (2006), 'Nonparametric Survival Analysis on Time-Dependent Covariate Effects in Case-Cohort Sampling Design', Statistica Sinica, 16, 267-85

To cite this paper please use the following details: Sevcenko, V. (2011), 'Active Voices: An Empirical Study of the Effects of Small Firm Lobbying on New Firm Survival', Reinvention: a Journal of Undergraduate Research, Volume 4, Issue 1, http://www.warwick.ac.uk/go/reinventionjournal/archive/volume4issue1/sevcenko. Date accessed [insert date]. If you cite this article or use it in any teaching or other related activities please let us know by e-mailing us at Reinventionjournal at warwick dot ac dot uk.