The Scottish Government

« Previous | Contents | Next »

Listen

Scottish Economic Statistics 2004

A5 Analysis of firm closure and survival

Paul Teasdale and Jessica Minty, Scottish Executive

Summary

This article presents measures of enterprise closure using data from VAT registrations, Companies House registrations, the Inter Departmental Business Register (IDBR), the Labour Force Survey and company insolvency statistics. It looks at the number of closures and the probability of closure for different types of business. It includes comparisons between Scotland and the UK as a whole, and over time. All of the sources show a remarkable stability in survival rates over time.

Introduction

There are several data sources that monitor business closures in Scotland. Each measures something slightly different. This article describes the sources, and considers three types of measure:

• Closure or death rate - that is firms closing in a given year;
• The survival rate - the proportion of firms surviving a given number of years;
• The hazard rate - the probability of a firm, having survived to a certain age, closing in the forthcoming period.

Before we go any further we should note that a "closure" does not necessarily mean a business "failure". A study for Barclays Bank suggests that only a fifth of closures are associated with financial difficulties ⁹. It simply means that a business has ceased trading as a separate entity. That covers several possible outcomes:

• The owner may have moved on to a waged job that pays better than the business. The Labour Force Survey shows that the majority of those who leave self-employment go rapidly into a job;
• For many medium sized and large firms, closure is associated with acquisition by another firm. This could be a sign of success. For many owners of growing firms this is an objective. (A recent survey for the FSB found that 7 per cent of respondents wished to sell the business in the next two years) ¹⁰;
• Some firms may cease to be registered for VAT because their turnover falls below the threshold ¹¹.

Measuring closures

Each of the data sources provides several types of measure:

• The number of closures or de-registrations;
• Closures as a proportion of the total population of businesses;
• And, in the case of VAT de-registrations, closures relative to the adult population.

VAT de-registrations

The most commonly used source of business numbers - births and deaths - is VAT registrations and de-registrations. This information comes from administrative records, and is published annually by the Department of Trade and Industry. Enterprises with a turnover above a certain threshold (currently 58,000) are required to register, although many below the threshold also register. At the end of 2002 there were 124,200 businesses registered in Scotland. Because it comes from administrative records the VAT data can provide very detailed information by location, by sector and over time.

Chart A5.1: VAT deregistrations in Scotland 1994-2002

The number of de-registrations in Scotland has remained fairly stable for a decade at just under 12,000 each year. Every year about 10 per cent of the stock of VAT registered businesses close. There is very little variation between most regions of the UK, although the rate tends to be slightly higher in London and the South East. The "death rate" varies between sectors, with a lower turnover for the sectors with relatively high entry costs (e.g. manufacturing, agriculture).

However, looking at de-registrations per 10,000 adults there has been a fall in Scotland from 29 to 27 over the period 1994 to 2002. In the UK, the rate was 37 in 1994 and in 2002 but fell to as low as 32 in 1997 and 1998. The higher rate in the UK reflects the greater rate of business births in England.

Inter-Departmental Business Register

The Inter-Departmental Business Register (IDBR) builds on VAT registration data and includes those additional enterprises that are registered for PAYE but not VAT (i.e. they have employees but turnover below the threshold). The IDBR recorded 147,500 such firms in Scotland in 2003. We can use the IDBR to show how the death rate varies by company size (Table A5.1) or age (Chart A5.2).

The analysis below includes only enterprises (not public sector) based in Scotland. This avoids the potential distortions caused by enterprises closing only their Scottish operations.

Table A5.1: Enterprise closures between 2000 and 2003 by size of enterprise

Chart A5.2: The proportions of enterprises closing in 2000-2003 by year of registration

Companies House Register

A register of companies is maintained by Companies House. There were around 99,000 companies on the register in Scotland at the end of March 2004. About 34 per cent of businesses on the IDBR are registered as companies (20 per cent of all enterprises) - although not all companies are on the IDBR as the IDBR does not include companies that are not trading. The most recent data for 2003/04 show that 8,400 Scottish companies left the register during the year. Chart A5.3 shows the trend over the past 15 years, alongside the similar trend in England and Wales ¹². The number of de-registrations has risen in the past decade. This follows an increase in the total number of businesses registering as companies. This rise in the proportion of businesses (especially new businesses) registering as companies was intensified by changes to the legislation on corporation tax after 2001.

Chart A5.3: Companies House deregistrations 1980-2003

Insolvencies

The DTI releases data on insolvencies every quarter. The total number of company insolvencies comprises compulsory liquidations and creditors' voluntary liquidations. Over the past decade, liquidations have averaged about 600 per year in Scotland and around 14,000 in England and Wales. In recent years the number has risen, in part, because of the greater number of companies registered. Chart A5.4 shows that, as a proportion of the stock of all companies registered, the number becoming insolvent has been very stable at about 0.8 per cent each year. The exception to this trend was the spike at the start of 2003. The proportion of businesses becoming insolvent has been consistently lower in Scotland than in England and Wales (where the legislative regime is slightly different). This is probably due more to a lower business birth rate in Scotland than to higher survival rates.

Chart A5.4: Quarterly insolvencies as a proportion of all registered companies in Scotland and England & Wales 1994-2004

Table A5.2 shows that, in 2003, the number of insolvencies was equal to about 7.5 per cent of the number of company closures ¹³. This has remained fairly constant over the last ten years. In England & Wales, insolvencies are equal to just over 8 per cent of company closures, a fall from a peak of 17 per cent in 1995.

Table A5.2: Comparison of company deregistrations and insolvencies in Scotland 1994-2003

Along with the company insolvency statistics the DTI releases figures for bankruptcies of individuals. Over the last ten years there has been an upward trend in the number of bankruptcies in Scotland and in England & Wales. In 2003 there were 3,328 individual sequestrations in Scotland - equal to about 8 per 10,000 adults.

Labour Force Survey

Useful supporting data can be drawn from the LFS, which provides valuable information on the people who are self-employed, as well as the businesses. It is the main source of information on the one person businesses below the VAT threshold. The LFS reports that in Spring 2003 there were 20,000 people who had been self-employed a year earlier but were no longer (9 per cent of the total self-employed in 2002). ¹⁴ Not all of these changes in status will be associated with a business closure: for example, some people may have left a continuing business; or the legal status could have changed so self-employed people become employees of their own company.

Survival Rates

We turn now to survival - that is the proportion of businesses in a cohort still in existence after a given length of time. The data most commonly quoted refer to one year or three years. The reason for looking at these points is that, as we shall see, survival rates are much higher after that time.

VAT data

The most commonly used indicators of survival make use of VAT data. Of firms that registered in Scotland in 1999, 65.3 per cent were still trading three years on and of those firms registering in 2001, 92.5 survived one year. Chart A5.5 shows the average survival rates in Scotland for one, three, five and eight years after registering for VAT. The median life of firms is five years; in other words, half will close within five years of registration.

Chart A5.5: Business survival rates in Scotland at one, three, five and eight years

Survival rates in Scotland do not differ significantly from those in rest of the UK. Table A5.3 shows one and three year survival rates of recent cohorts for Scotland and for the UK.

Table A5.3: Percentage of businesses surviving one and three years after registering for VAT in Scotland and the UK

Survival rates vary according to sector - with higher rates in those sectors with high entry barriers as start up costs are greater and new firms are generally larger. A comparison of survival rates across Scotland shows that survival rates tend to be higher in rural areas. However, rates vary only slightly between UK regions, and there is not a stable ranking of regions. Table A5.3 shows survival rates have increased since the early 90's.

It might be thought that a higher birth rate (per head of population) would lead to a lower survival rate: as it could mean more marginal businesses being formed by people with less entrepreneurial ability. However, there does not appear to be much statistical support for this. In the cross sectional data, survival rates are highest in the parts of Scotland with the highest birth rate; and in the time-series the survival rate has been fairly stable despite fluctuations in the birth rate. It would appear that conditions favouring births also support survival. It is because of this stability that we can be confident that movements in the number of closures will follow very closely, but with a lag, the number of births. It also follows that an increase in the birth rate should lead to an increase in the stock of businesses.

Companies House

The VAT data are useful when looking at survival rates of firms registered in the last few years and can provide details by sector or location. However, the information is limited in the length of the period it covers. VAT was introduced in 1973 and there is a discontinuity in the data in 1993. Along with the data on companies registering and deregistering, Companies House produces data on the survival rates of businesses dating back to those that registered in 1862. This provides us with a time series over 140 years. Plotting this data in a chart shows the remarkable consistency in survival rates over 140 years despite changes in the economic conditions. Only two periods appear to stand out: lower survival among companies registered in the early 1980s and higher survival rates for firms registering in the 1950s. Overall we see that the greatest risk of closure occurs in the first four years of trading. As with the VAT data, the median life of a business is less than five years. Only 54 per cent of businesses that registered in 2000, and only one third of businesses that registered in 1995 were still trading in 2004.

Chart A5.6: Proportion of companies still trading in 2004 by date of company registration

Labour Force Survey

The LFS provides information on the length of time in business for the self-employed. The pattern is similar to the VAT and company data, but for the self-employed (mostly) below the VAT threshold the survival rate is lower, with a median life of around two years.

The Hazard Function

Another way to present the survival data is as a hazard function, as shown in Chart A5.7. From this we can see the probability of a firm closing in the coming period given that it has survived, say, four years. The story here is similar to that shown in chart A5.2, but where that compared death rates of cohorts of different ages in 2000-03, each line on chart A5.7 follows a single cohort over a number of years.

Chart A5.7: Exit hazard function, probability of closure in next 3 months for three cohorts

This helps to identify the most vulnerable points in the life of the enterprise. The chart shows that the probability of closure peaks around 12 to 24 months after registration. A number of reasonable explanations can be given for this. This peak explains why the number of deaths follows the number of start ups with a two year lag. The hazard function for the UK as a whole is very similar, following similar patterns, levels and shifts between cohorts. This suggests that the differences between Scotland and England lie not in survival rates but in the birth rates.

Chart A5.8: Comparisons of exit hazard functions of closure in next 3 months in Scotland and UK, for 1999-2001 cohort

Conclusions

This article has drawn together from several sources data pertaining to business closures. All of these sources show a considerable degree of stability in survival rates over time, with some fluctuation, which may be related to the economic cycle. Comparing business survival rates between Scotland and the UK average, we found very few differences. This indicates that the differences in size and growth of the business population can be explained by differences in birth rates rather than variations in survival.

« Previous | Contents | Next »