skip to content

Keynes Fund


Summary of Project Results

The performance of small and medium sized enterprises (SMEs) in the 19th century has been little examined beyond case studies because of the lack of available large scale data. However, data on the workforce size for 1851-81 for most of the population of British firms and has now become available, deposited by the grant holder in the UKDS open access database: the British Census of Entrepreneurs (BBCE) (Bennett et al., 2020a).

These derive from digital records of the censuses for the nineteenth century in I-CeM (Schürer and Higgs, 2014). From I-CeM information on SMEs has been extracted, parsed and coded in BBCE, supported by a major ESRC-funded project, and other financial supports BBCE identifies all proprietors recorded in the census and their workforce using a question used in the British censuses for 1851-81 that requested information on each proprietor’s business: “In TRADES, MANUFACTURES, or other Business, the Employer must, in all cases, be distinguished. – Example: ‘Carpenter – Master, employing 6 men and 2 boys;’ inserting always the number of persons of the trade in their employ, if any” (1861 Census: Householder Schedule). Though not a well-designed question by modern standards, the responses to this question abstracted in BBCE give valuable alphanumeric data on workforce size and sector. There are 200,000 to 300,000 individual employers who gave this information for the period. In addition BBCE also identified 1.1-1.6 million one-person business proprietors (trading on ‘own account’) that had no employees. An on-line Atlas of the BBCE data at (Bennett et al., 2020b), as well as a User Guide (Bennett, et al., 2019), are now fully available.

The Keynes project has linked these proprietors and their firm-size data to profits data for each individual proprietor’s firm. Profits data for individual SMEs are derived from contemporary tax records for income tax on profits from ‘trades and professions’, which covers all non-farm businesses. Individual tax records were supposed to have been destroyed, but some have survived, mostly now held at local Record Offices (ROs) derived either from copies of local assessors’ files or from personal records, with others available from scattered records held at The National Archives (TNA). Extensive search of all catalogues led to about 4,400 profit records for three years 1861, 1871 and 1881, with a further 350 derived from TNA. Of these tax records the project linked 2,985 with the proprietor’s census record. This is the basis of the project analysis. These linked proprietors were both employers and own account operations (i.e. businesses with, and without other employees). Although a very small proportion of the total proprietors of the period identified in BBCE, they prove to be sufficiently representative by sector and profits/tax characteristics to allow robust analysis.

The original research plan envisaged developing econometric estimates from the individual level data for profit and workforce size, and their interaction, with controls to allow estimation of the effects of sector, portfolios, proprietor age, gender, marital status, household structure, and location. The estimates were to be interpreted in terms of sector markets, and locational opportunities, intermediated by demographic and personal factors, and aspects of regulatory depth in each sector. The analysis was aimed mainly at 1851-81, but it was hoped that additional estimates for 1891 might be possible to assess larger, though more problematic, comparator data on profits for 1891. In addition it was hoped to develop estimates for the impact of rail treatment effects over 1851-81. It was hoped that sufficient candidate locations would be available to create a matched sample of non-treated locations to develop causal estimates for a panel of individual SMEs in different places to estimate treatment effects on profits. Railways addressed potential market failures by connecting previously disconnected markets that experienced incomplete trades due to local monopolies selling less than optimum to raise prices, and/or monopsony that bought fewer inputs.

In the event a number of modifications to the project plan were necessary:

  • the data were Innovative and challenging, not previously available or used by anyone else in this way so that a lot of methodical development was required to align and process to a format capable of useful analysis.

  • The core research for the Keynes project was to extract profits data from the tax records. At the time of the application most local RO data had already been collected, and TNA information was added to this. A method was developed to align data on profits for some records that only recorded tax duty payable which had to be adjusted.

  • The project had to focus on 1861-1881. The possible extension to 1891, and perhaps to later years was postponed; similarly the use of 1851 data proved problematic because of changed tax (and hence profits) definitions. A decision was made to focus the project resources to allow a full analysis of 1861-81 and use the information on 1851 and 1891 and later in future research.

  • The final outcome was a set of profits data for 4,750 business proprietors over 1861-81. As with all record-linkage, precise matching of individuals between their tax and census records was challenging. Although taxes and censuses both used residential addresses, some names were not well recorded, some people were very mobile, and the actual tax years available were not always precise matches. However, profits data was linked for 2,985 census individuals. The main difficulties of linkage related to records in the TNA sources and some local ROs giving insufficient residential precision or imperfect/insufficient name recording to allow accurate record linkage.

  • The reduced years of coverage limited the extent to which rail treatment effects could be investigated. It was not possible to create matched samples of treated and non-treated locations to assess potential market failures. However, rail was included in the econometric estimates and showed important differences between locations. In addition it proved possible to use measures of waterway access to show how different transport modes influenced different types of location.

  • Three post-doctoral researchers were engaged using the Keynes Fund support. Their contributions differed somewhat for the original project specification because of their availability, as agreed with the Fund Managers. The final contributions were: Dr Harry Smith - initial data linkage, data template and coding, and methodology development; Dr Carry van Lieshout - completion of data linkage, data coding, and database construction; Dr Piero Montebruno - database alignment, merging with BBCE, addition of other data (transport etc.) and initial econometric estimates.

The outcome at a first stage of processing has been development of a draft research paper using econometric cross-sectional estimates of profit and profitability (per worker) for 1861-81 controlling for sector, portfolios, partnership participation, proprietor age, gender, marital status, household structure, and location by urban/rural location, population density, and access to railways and waterways.

The analysis demonstrates that profits varied profoundly by sector and location and chiefly reflected contrasts of ease of sector market entry, and differences in externalities between towns and rural areas. Larger profits generally needed a larger urban base to support business needs, especially larger workforces. However, profitability was usually more influenced by sector market conditions than location, at both aggregate and detailed sector levels.

Some of the largest differences in sector market opportunities reflected the constraints of market entry and barriers to access to the professions. Some of the highest profitability levels were physicians and solicitors, with lower profits and profitability occurring in newer professions such as engineers and accountants. Other high profitability was for personal services and agricultural processing, with retail and refreshments also significantly above average. There was low profitability in manufacturing, and very low in construction and some maker-dealing trades. The results contrast with many dominant narratives of the nineteenth century which emphasise the high profits of manufactures and London financial markets. Profitability in sectors such as professions, agricultural processing, food sales and refreshments have rarely been the subject of previous analysis but were crucial parts of the economies of large cities and medium-sized towns. They also provided widespread access for large numbers of people to higher profit opportunities across the country. In comparison, making good in many of the sectors in manufacturing that offered high returns on capital was often more challenging and risky. Although they employed huge numbers of workers, in 1881 there were only about 500 large-scale manufacturing employers with over 1,000 employees, and even firms that employed over 20 workers only numbered 22,000 in all manufacturing. In contrast, for physicians and solicitors alone, there were 23,500 employers and sole proprietor in England and Wales, spread across the country; with agricultural processing, food sales and refreshments offering much more numerous and more geographically accessible opportunities than manufacturing. Hence, for most entrepreneurs the incentives to develop businesses were widespread in a wide range of sectors. Along with manufactures and financial markets this was a high period for British economic opportunity.

The analysis also has important implications for the sources of incentives. Variations of profitability between localities were generally more limited than between sectors, which is indicative of increasing market integration. However, important spatial effects occurred in the expected direction: rural areas generally offered significantly lowest scope for profit and profitability, with urban areas and especially London significantly higher. At a finer-grained level, however, rural areas differed internally, with closeness to a rail station and the presence of a station or waterway generally offering some higher profit opportunities. In the most geographically isolated areas where wider competition was limited, traditional maker-dealers maintained significantly higher profitability compared to other sectors. Differences in firm-level organisation (measured by portfolios and partnership) and demographic factors were generally less important than sector or location.

However, much remains unexplained and is the subject of future research developments. The evidence is constrained by the samples and size of data available, but a substantial residual suggests an important agency and individual entrepreneurial characteristics that differ in a way that it has not been possible to examine in the historical data. Modern studies of entrepreneurship often focus on the role of personality which is not measureable in historic data. This may explain the residual unexplained variance. However, the analysis so far shows we should not overstate the role of agency. The dominant sources of differentiation of profitability were firstly sectors, secondly locations, and thirdly business organisation and resources.

Further research can be developed with these data, not least because they are not anonymised and hence offer potential for further record linkage at local level, and perhaps to combine with other data on larger enterprises.


Impact and outputs

The first stage of draft analysis has been completed by Bob Bennett in June 2020, 4 months after the end of the Keynes support. This draft analysis, co-authored with the team involved in the Keynes funding, will be sent for publication and put onto the Keynes website. The source data material will be developed over the next year for deposit, either alongside BBCE at UKDS, or at the Cambridge Apollo depository.

Dissemination of the work is beginning with the Keynes Fund workshop in September 2020. The work will be presented at future UK, US and European conferences and workshops.

An important aspect of future dissemination will be linking of all Keynes project outputs to the related BBCE database deposit materials at UKDS, the separate user interface for BBCE, BBCE Atlas of Entrepreneurship, and the ESRC and other funding support websites:


Any possible future plans

The analysis so far has concentrated on assessing the main economic features of the 1861-81 data. This will be taken further in several directions:

The main localities where a proportion of the records are concentrated can be developed in greater depth using further record linkage at local level, and combining with other data on larger enterprises trading in these localities that reported profits elsewhere, using other archival sources. This should broaden the view possible on the local/non-local sources of profit and its leadership. More data from other sources can also be developed to assess extent of markets; e.g. local, national or export orientation.

The original project plan of extending to 1891 profits data can be developed, though widespread workforce data was no longer collected by the census so that the type of analysis possible will differ: focusing on choice between statuses – as employer, own account or worker. Some early stages of development have been made with the BBCE data assessing status switching over time; this can be taken further by integrating with the Keynes-supported profits data.

The use of 1851 profits data can directly extend the analysis developed for 1861-81. The census collected workforce data for this period, which is in BBCE. The challenge is to align the tax definitions (and hence the measurement of profits) between 1851 and 1861-81. This is tractable, within bounds, but requires a substantial adjustment exercise.

A major benefit of aligning the 1851 data is that it will allow development of a modest analysis of rail treatment effects, as originally planned, for all 1851-81. It will allow a few matched sample areas of treated and non-treated locations to be used to assess rail impacts and effects of potential market failures.



Bennett, R., Smith, H., Van Lieshout, C., Montebruno, P., & Newton, G. (2019) The British Business Census of Entrepreneurs 1851-1911 (BBCE): User Guide, Cambridge, University depository,

Bennett, R., Smith, H., van Lieshout, C., Montebruno, P., & Newton, G. (2020a) British Business Census of Entrepreneurs, 1851-1911 [data collection], Colchester, Essex, UK Data Archive [distributor] SN: 8600,

Bennett, R. J., van Lieshout, C., Smith, H., Montebruno, P. & Lucas-Smith, M. (2020b) BBCE: Atlas of Entrepreneurship.

Schürer, K. and Higgs, E. (2014) Integrated Census Microdata (I-CeM), 1851-1911 [data collection], Colchester, Essex, UK Data Archive [distributor] SN: 7481,



Prof. Robert J Bennett


Professor Robert J Bennett is Director of Research, Emeritus Professor at the Department of Geography, and Emeritus Fellow of St. Catharine's College, University of Cambridge. His research interests are Analytical Economic Geography, Business Management and Public Policy.


Share this Project