Small Business Bonus Scheme: evaluation

This report presents the results of an evaluation of the Small Business Bonus Scheme (SBBS), and provides recommendations in relation to the SBBS and non-domestic rates relief more broadly.


1. The Fraser of Allander Institute is an independent research institute that is part of the Department of Economics, Strathclyde Business School, University of Strathclyde.

2. See, p18.

3. See

4. Scottish Ministers have set out their approach to taxation here:

5. Whilst non-domestic rates are often referred to as 'business rates', charities, public and third sector organisations are also liable.

6. Source: Scottish Local Government Finance Statistics 2019-20,

7. Further information can be found at

8. For a full list of currently available reliefs please see

9. Source:

10. Report of the Barclay Review of Non-Domestic Rates, Scottish Government (2017):

11. We have inflated the numbers in Figure 1.1 to 2020 prices as we make an explicit comparison over time and calculate total spending over the period. Previous figures cited up to this point are not inflated in this way, and are in current (i.e. "in year") prices. This sum perhaps overstates the extent of the government spending on the SBBS over the period since some properties that benefit would be eligible for alternative reliefs in its absence. We find that this is likely to be true of very few properties however.

12. The What Works Centre for Local Economic Growth is run in partnership with LSE and the Centre for Cities:

13. Much of their evidence scoring is based on the Maryland Scientific Scale, detail on which can be found in their evidence scoring guide:

14. We were given access to versions of the PTOA database and Billing Snapshot that also included some information from the VR.

15. With some exceptions: Oct. 2009; Jan. and Oct. 2010; July 2011; July and Oct. 2012; and the April 2017 data was dated 6 April, not 1 April.

16. To do so we selected the appropriate month of the VR/PTOA database: October in 2011, 2012, 2013, 2015 and 2016 and May and June respectively in 2018 and 2019; since the Billing Snapshot data for 2018 and 2019 do not align with a pair of quarterly VR and PTOA datasets, data from the May 2018 and June 2019 VR and PTOA were provided specifically for this purpose.

17. Although note that not all of these properties will be eligible for relief, in particular if they constitute part of a multi-site business whose cumulative RV exceeds £35,000.

18. We use the median to report the average as the data contain large outliers that would skew the mean value.

19. Throughout, when we refer to RV as it relates to SBBS, we are referring to RV net of any residential apportionment.

20. For example, if a property with an RV below £18,000 is actually part of a larger multi-site business with a cumulative RV above £35,000, but is not attributed to it in the data, then we would classify this property (identified in the data as a single-site business) as eligible when it is in fact not.

21. For example, see

22. See the background notes of the Scottish Government NDR relief statistics publication for detailed information The Scottish Government also publishes annual rates relief guidance for Local Authorities that includes a full description of mandatory and discretionary reliefs here:

23. As of 1 April 2020 this is actually a legal requirement for SBBS eligibility:

24. Discretionary sports relief for charitable properties that do not qualify for mandatory sports relief is 100% centrally-funded and Hardship relief, which is 95% funded, but only for awards as a result of the Glasgow Sauchiehall Fire.

25. There have been changes to the property/rental market over the past decade that could have influenced this, namely the rise of self-catering properties. If domestic properties are used heavily for short-term rentals, they become subject to non-domestic rates. See

26. We used the VR and PTOA dataset information from January 2020 in order to use the most up-to-date address information when the survey was launched.

27. Our questions on business activity (for example turnover and employment) were in fact backward looking and so should have been completed based on pre-pandemic performance. There is one exception to this – Question 7 asks "How would you describe the current position of the business?". There were still questions that could be affected by timing of the survey, however.

28. Because RVs do not differ greatly on average across local authorities, our stratified random sampling strategy means that the distribution of properties in the sample of businesses to which the survey was sent is broadly identical to that in the sampling frame – the population of properties from which we drew the sample. Appendix Figure A.2 shows the distribution of properties across local authorities in both the sampling frame and the administered sample for comparison. There are slightly larger differences in the distribution of properties across local authorities in the entire population of businesses, primarily because our sampling frame excluded small shootings which are predominantly in rural local authorities (see Figure 4.5).

29. FTE employment represents the number of full-time staff the total hours worked in a business are equivalent to. For example, two part-time employees working half full-time hours represent one FTE employee.

30. SBBS eligibility is based only on a business's properties within Scotland. We asked this question to understand the extent to which businesses with properties in Scotland also have interests elsewhere, which might imply something about their size.

31. For multi-sites, there were only 9 responses from eligible non-claimers and 4 from those classified as ineligible.

32. Had we predicted in advance these severe discrepancies we would have included this as a question in the survey.

33. There is one exception to this – Q7 asks respondents about the recent growth of their business.

34. We entered many of these manually and most responses that reported to be contracting actually noted this was because of Covid-19 and the lockdown.

35. The comparison across RV more or less amounts to a comparison of claimers either side of the threshold: 23 of the 33 in the £14,000-15,000 group receive 100%, and 21 of the 28 in the £15,000-£16,000 receive 25%. Only a handful are eligible non-claimers/missing.

36. For example, the gross NDR bill for a property with an RV of £14,500 with a poundage of 49.8p is £7,221, corresponding to £601.75 per month. If the business only receives 25% relief compared to 100% relief, its costs will increase by £451.31 per month. While this is above the £300 considered in this question (which was used in the survey as it is representative of typical NDR savings businesses enjoy as a result of SBBS), answers will be suggestive of the effect.

37. For a business with an RV of £15,500, this represents a cost saving of £474.69 per month.

38. Possible responses were: I pay the rates bill; I use a rating agent to pay the rates bill; my landlord pays the rates bill; I don't know who pays the rates bill; or my head office pays the rates bill.

39. The fact that some respondents recorded that their "business" was a charity again shows the shortcomings of identifying suitable properties to survey from the VR and Billing Snapshot data.

40. Again, we note that it is possible for responses to have been impacted across the key dates during the beginning of the Covid-19 pandemic. When disaggregating them by whether or not they were recorded before or after each of the key dates, however, there is very little difference in the proportion reporting each degree of impact.

41. Question 29 is discussed alongside the response to question 24 above.

42. The same is true for multi-site businesses, but we focus here on single-site businesses to provide as clean a comparison as possible given the uncertainty in constructing business entities from the data. Moreover, note that the comparison of multi-site businesses that receive 100% relief with those that receive partial relief is more nuanced because, for example, a two-property business with a cumulative RV of £20,000 could be comprised of two properties with an RV of £10,000 which therefore receives 25% relief on both, or one property with an RV of £19,000 and one with an RV of £1,000 that therefore only receives 25% relief on the low-RV property. Thus, drawing any wider inferences from an econometric analysis of multi-site businesses would be problematic.

43. The design of the SBBS policy means that it is in fact not possible to make a comparison of businesses who received 100% with an appropriate counterfactual group that received no relief. This is because the policy is universal, meaning the only way to carry out such a comparison would be by utilising either those businesses that did not take up the SBBS relief they were eligible for, or those that have an RV of above £18,000. Both groups are likely to systematically differ from the pool of 100% recipients in ways that make this comparison unlikely to be informative.

44. All of the following expressions also condition on observable characteristics of businesses, for example industry. We contain these in the vector for each business but keep the conditioning implicit to simplify the expressions.

45. Suppose, for instance, that a business is attributed an RV of £15,250. Whether or not the business engages in an appeals process is likely to be influenced by the need for the business to save on NDR, so it is not unreasonable to assume that businesses with greater need for NDR relief are more likely to appeal.

46. This is not to say that these enterprises have been perfectly identified, but they are the group about which we have most certainty.

47. We have simplified slightly here. Businesses could not be explicitly excluded based on whether they were PAYE and/or VAT registered. The Scottish Government assessed the likelihood that this would be the case for properties based on their RV and property description. Whilst not perfect, given the detailed nature of property descriptions we assume this is sufficient information to determine this likelihood with a reasonable degree of accuracy. Further, matches for properties which were included in the matching exercise but that weren't deemed sufficiently accurate were discarded. This again affects the SBBS effect we can hope to identify. However, we cannot define this group along any meaningful dimension and thus cannot address it directly. Nevertheless, this should be borne in mind when understanding the effect we aim to estimate. It is also unlikely that the condition of being PAYE and/or VAT registered is not met for all businesses close to the threshold, but these conditions must also be borne in mind.

48. In practice it would be possible to implement a version of the estimator we use that adjusts for non-perfect "compliance" with the SBBS, e.g. the situation in which a business eligible for 100% relief does not claim. Given the uncertainty around the accuracy of cumulative RVs we have discussed, we opt to focus on eligible claimers as this is the sample most likely to have accurately linked data.

49. We cannot rule out that inconsistencies in matching – i.e. properties being matched incorrectly to enterprises – contribute to this large variation. For a similar reason it is also worth noting from Table 7.6 that the turnover and employment among enterprises with an RV between 0 and £1,000 appears implausibly high when compared with all other groups.

50. To be clear we are not dividing the sample up in an attempt to 'find' results. This would be akin to 'p-hacking'. Rather, ex-ante we believe that – for the reasons we outline throughout this section regarding similarity – the effect of the policy might differ depending on business type. We use these descriptive tools, as well as more formal tools in the next section, to test this.

51. The match rates in the full exercise and in the sample are not strictly comparable for two main reasons: Firstly, the full exercise used a wider range of address and postcode information than the sample and secondly in the sample we were able to interrogate the matches manually to ensure that they were no spurious ones being thrown up; something we weren't able to do systematically for the full match.

52. The valuation roll provides a flag that indicates whether a property is occupied by a proprietor, tenant or occupier. This flag is used to derive the business name and the proprietor, tenant and occupier address. Property address refers to a property's physical address. Ratepayer name and address are the names and addresses local authorities use for billing purposes.



Back to top