Scottish Government

4 Modelling future RSL housebuilding

Key points

• Assuming a reduction in the grant rate to 50% and factoring in a number of other reasonable assumptions, it is estimated that RSLs have the capacity to build 3,895 homes, annually, over the next 20 years. This compares with the Scottish Government's current 'core programme' of about 3,600 homes.
• Rents rising at rates above RPI+1% p.a. would increase capacity, while unit management costs continuing to increase at above-inflation rates would erode it. However, the scale of these effects would be somewhat smaller than for local authorities.
• Halving the assumed RSL grant rate (to 25%) would reduce RSL housebuilding capacity by only 27%. Such a grant rate would be similar to that currently in force for local authorities and, at this level, would imply a similar housebuilding capacity for each of the two sectors, yielding a combined national output capacity of around 5,300 homes p.a., and costing just under £200M - around two-thirds of the Scottish Government's current total budget for local authority housebuilding grant and the RSL core programme.
• At a common grant rate of 35%, the two sectors would have a combined national output capacity of around 6,300 homes p.a., with full utilisation of this capacity costing around £310M p.a. in Government grant - slightly higher than the 2010/11 AHIP 'core programme' budget.

Outline methodology and key assumptions

4.1. This chapter presents outputs from a financial modelling exercise to project the housebuilding capacity of housing associations or Registered Social Landlords - RSLs - in the period to 2035. In this initial section we describe, in broad terms, the method used in this process. While, in the very widest sense, this task is in principle the same as that of assessing the financial capacity of the council sector, as discussed in Chapter 3, the sectors differ in important respects and therefore the nature of the task as well as its practical implementation both differ significantly.

4.2. While councils remain within the public sector and essentially have to meet relatively simple public accounting rules and constraints (basically balancing their HRAs from year to year), RSLs are quasi-private corporate bodies which must provide business plans which satisfy not just the Scottish Housing Regulator ( SHR) but also the banks and other bodies which finance their borrowings, short and long-term. This necessity to have a relationship with private lenders (or investors in bonds) applies particularly to RSLs aspiring to undertake new development and investment.

4.3. At the core of this relationship with lenders/investors is an assessment of the prospective financial performance of the organisation, an assessment which must have regard to liquidity, solvency, risk, and the adequacy of returns and reserves necessary to deem the organisation's business plan as sustainable and prudent. This assessment generally refers to certain key financial ratios, and these provide the main basis for this financial modelling exercise. However, it is also worth emphasising that, in their relationships with lenders, the regulator and other stakeholders, RSLs will also be judged on more qualitative matters, particularly relating to the quality of management.

4.4. The essence of the modelling approach adopted for the RSL sector therefore has been one of assuming that associations attempt to expand by investing more in new building, but only insofar as this financial/business planning accountability relationship with lenders and other stakeholders is likely to permit it. We undertake a 'micro-simulation' whereby, for each RSL in each year, we assess its performance on a basket of ratios, and depending on its standing we allow it to expand progressively, continue development at a previous level, or not develop at all. This is done for 75 RSLs, comprising all of the larger associations and a sample of smaller and more specialist associations.

4.5. For reasons explained in Chapter 1, the RSL projections were undertaken separately for two groups of associations: the 30 which authorised access to recent financial projection data, and another 45 for which, in the absence of such authorisation, we have based our work on SHR published accounts and performance data. Because these provide a different range and time coverage of base data, the technical details of the models are different for these two groups, particularly in respect of the establishment of the base position. However, for both groups, forward projections are made from year 2011 onwards using common assumptions rather than the assumptions with associations themselves might have made in their forward projections ⁴².

4.6. It is very important to emphasise that this modelling is concerned with financial capacity to undertake different levels of development. In other words, it is seeking out the upper limits to levels of development, were this to be prioritised by the sector, which would be compatible with robust and prudent financial and business planning and the normal internal and external constraints on this. It is not a forecast of what RSLs would actually do, given the particular set of parameters. Some RSLs might actually build at around the levels our model suggest, while others might not, for a variety of reasons mainly to do with balancing other priorities. This key point needs to be borne in mind in relation to the projects set out both in this chapter and in Chapter 3, and particularly when considering the possible implications for the grant regime in the future. Thus, there is no underlying assumption that all RSLs will wish to develop continuously and to fully utilise their financial capacity for this purpose.

Financial stress indicators

4.7. Five financial ratios are, as noted, used to rate and score performance/ robustness year-by-year. In each case two thresholds are applied, one for Red vs. Amber, and one for Amber vs. Green. The numbers currently used were chosen by the researchers (but also informed by discussions with stakeholders). However, these parameters of the model can be changed ⁴³.

a) Interest cover: This is surplus before interest and depreciation, over gross interest payable. This is a measure of whether you have enough coming in after meeting expected bills to pay the interest which you know you will have to pay. Lenders will be concerned about this. Depreciation isn't a real bill you have to pay, so it can be discounted. Thresholds are 100% and 125%.

b) Net Surplus over Turnover: This is surplus after interest, depreciation, tax if any and including sales. This is a basic measure of how far your head is above water relative to the size of the body. This does allow for making provision for depreciation. Thresholds are 1% and 3%.

c) Return on Capital: Net surplus (as previous) over capital reserves ⁴⁴. A sort of classic rate of profit. However, a bit misleading because assets are often valued conservatively at historic cost less HAG less debt, not at current value (even on tenanted basis). Thresholds are 1% and 3%.

d) Gearing: Net Debt over (Total Assets minus Net Debt). This is a measure of how highly loaded up you are with debt, relative to your own capital. High gearing can make for high rates of profit on a small capital base, but is much more risky. This ratio indicates how much fluctuations in performance are 'geared up' or amplified by the effect of debt. Unlike the other indicators, these are regarded as maxima rather than minima. Thresholds are 80% and 60% ⁴⁵.

e) Liquidity: Current assets over current liabilities ('current' meaning realisable within one year). This measures whether will you have enough cash coming in, when your current debtors pay up, to meet the bills of your current creditors plus paying off any debts which have to be repaid within the year (including long-term debt reaching term). Thresholds are 1.0 and 1.2.

4.8. Overall, the modelling indicates that it is typically the surplus-based measures which kick in and act as a ceiling in most cases. In the second group of RSLs (the '45') there are many cases of low and negative surpluses in the early years, which gradually improve later. The liquidity measure also looks rather low for many in this group, but this is partly a consequence of the way we have estimated some of the components. Adjustments are made through the projection period to correct this where possible. A 'liquidity shortfall' is calculated, based on the difference between current liabilities and current assets, and cash and investments are left in place and allowed to build up rather than being used to defray borrowing if this shortfall is present. This is intended to mimic the likely approach in business planning which would include taking action in the short term to correct deficiencies in liquidity.

Feedback mechanisms

4.9. The housing association financial capacity model is not a simple extrapolation, because it includes feedback mechanisms. By this we mean processes which change values for individual RSLs from a simple extrapolation from previous values, depending on some outcome from the previous year. The financial stress ratio indicators play a key role here, because, the more these fall short of threshold values, the less new development, if any, the RSL is able to carry out.

4.10. This is the most important feedback in the model. Performance scores are calculated from three bandings (Red=0, Amber=1, Green=2) for each of the five component measures identified above. RSLs scoring nearly all Green (9-10) are enabled to develop at a progressively increasing level. Those scoring an intermediate 5-8 can develop at this past annual rate, while those scoring below 5 cannot develop in that year.

4.11. Clearly, the development programme has many consequences in terms of augmenting the stock and hence affecting rental income and running costs. It also leads to increased borrowing and debt, and to changes in the assets shown in the balance sheet. The RSL model differs from the local authority model in explicitly modelling all of these.

4.12. Another significant feedback added to the model is a mechanism to use excess cash (where surplus exceeds 20% of annual turnover) - or financial assets - to defray the amount of borrowing needed to support new investment. The rationale here is that, since borrowing rates are typically higher than saving rates, it is generally better to reduce borrowing rather than increase saving. However, if the liquidity ratio is low, cash is first diverted into correcting this. As noted above, this approach is intended to mimic, in some degree, the kinds of treasury management and decisions which RSLs would make when faced with situations of either growing cash or liquidity shortfalls.

4.13. A further aggregate mechanism is used to try to control HAG expenditure to stay roughly within a predetermined resource envelope (separately for the two groups of 30 and 45 RSLs). The ratio between each year's HAG expenditure and the HAG envelope is calculated. A rolling, lagged moving average of this ratio is then applied to adjust new development numbers down (for overspending) or up (for underspending). This does not work perfectly in every time period, and when using the model it is necessary to apply judgement, having regard to cumulative over- or under-spending at some future target date (generally 2030). No similar mechanism has been incorporated within the local authority modelling, which simply generates the total grant spend.

Assumptions

4.14. So far as possible we have tried to follow a common approach and common set of assumptions for the RSL and local authority projections. The approach typically involves a central assumption and high and low variations about this. This applies to both the economic context assumptions and those relating to policy constraints and inputs. Tables 4.1 and 4.2 show the assumptions used in the main RSL modelling in respect of these two broad categories of factor. So, for example, our central assumption for future inflation ( RPI) is 2.5% p.a., and we test variations between 1.5% and 3.5%. We assume constant real unit management costs but unit maintenance and new build costs rising by 1% p.a. ⁴⁶ This assumes fairly tight control on management costs and the benefits of economies of scale, while recognising that building related costs tend to display an upward trend in real terms. The central assumption for the borrowing rate of 5.5% reflects past experience and actual RSL business plans ⁴⁷ but may be rather optimistic about the future prospects for funding costs, which were discussed in Chapter 2.

Table 4.1 - National economic assumptions used in RSL capacity projections

Table 4.2 - Policy assumptions used in RSL capacity projections

* HAG envelope excludes top sliced programmes

4.15. The most important policy assumptions are the grant ( HAG) rate and the associated public spending available to support it ( HAG envelope), the extent of any real terms increases in rents, and the level of new build procurement costs.

4.16. The central assumption for HAG envelope - £270M - is just below the 2010/11 'core programme' AHIP budget (£290M net of top sliced items) ⁵¹. The central HAG rate is chosen to roughly exhaust this (real terms) budget over a 20-year horizon. It will be noted that this HAG rate (50%) is somewhat below the recent/current rate, which exceeds 60%. The low grant assumption is chosen for consistency with the current figures for local authority sector new build. In practice, across different scenarios grant rates may be varied within this range in order to roughly balance out the demand for and supply of grant money over a run of years, given the assumed envelope (see also para 4.22).

4.17. Our central assumption is that rents would continue to reflect recent practice in rising at a rate just above RPI inflation (' RPI+1%', which would typically be below the increase in earnings). Variant assumptions involve an initial upward shift in rents, by either 15% or 30% in real terms over five years. After that they would then revert to an RPI+1% pattern. Clearly these variant assumptions would assume a will on the part of Scottish Government and the RSL sector to countenance rent rises to support new provision, and a judgement that this would not seriously infringe affordability or Housing Benefit cost criteria. Relevant here is the affordability analysis set out in Table 1.6 ( Chapter 1) which shows that, by comparison with the rents charged by their English counterparts, Scottish RSL tenancies have remained relatively affordable. Assumed rent increases in the 2009-13 business plan forecasts for the 30 RSLs supplying these projections averaged 3.32% p.a., which is close to our assumed 3.5% (=2.5+1.0).

4.18. SHQS achievement costs are assumed to be included within the business plan forecasts for the 30 RSLs providing these. In practice, the costs are in some cases found within capitalised repairs etc, while in other cases they are contained within the planned maintenance budget. For the 45 RSLs modelled from a base on accounts data, we use SHR published information on proportion of stock below the standard and apply an assumed unit cost for remediation to this number of units, phased over the period to 2015.

4.19. Our central assumption on procurement costs is that a new build unit would cost on average £120k at today's prices. This is at the lower end of recently observed unit costs. From our knowledge of housebuilding costs and viability models and from advice received from Scottish Government Housing Investment Division, we understand that £100k would be a basic average construction cost and £8-10k might be fees/oncosts, with the balance being land costs. Thus our high assumption would include significant land acquisition costs whereas our low assumption would imply a norm of no or negligible land costs, based on either free public sector land or effective use of the s.75 mechanism to the same end. Our central assumption could be interpreted as a mixture of some sites with positive land prices paid and other sites at zero cost.

4.20. For a number of assumptions, provision is made for these to change differently over an initial transition period before settling down to a steady long-term trend. The transition time period can also be varied, although this is typically set at five years. Key variables subject to transitional change are rents (where a rent shift is assumed in variant scenarios), HAG rate, and the (real) unit cost of new stock. It should be noted that the latter assumes convergence to a common value which may be net of (part of) land etc costs by virtue of s.75 or use of free/cheap public land.

Sector-specific modelling methodology and assumptions

4.21. It is appropriate to comment at this point on the rationale for certain differences in assumptions or approach between the housing association model and the local authority model. The local authority model takes each authority's own projections up to 2014 and then imposes future trend assumptions thereafter. By contrast, the RSL model takes 2011 as the first projection year. However, RSLs have widely different starting points in relation to a number of parameters. Therefore, we set up the model to enable convergence towards common values over a transition period, typically set at five years. Also incorporated within our approach is the view that policy might reasonably proceed in two steps, with a transition phase to new values, then a later phase where general long-term trends apply. The model is set up in this flexible way, and we have made certain judgements about these initial transitions (but they can be varied).

4.22. The other rationale for differences in the baseline assumptions for the RSL model and that for the local authority model is the different starting points. Housing associations operate within an established provision system which has been working with certain typical levels of grant over an extended period. Local authorities are a new provision system which has only recently been re-established using initially a quite different grant mechanism and grant level. Our baseline assumptions for RSLs were intended to represent starting from roughly where we are now but adjusting the grant rate downwards slightly, for two reasons. Firstly, the Scottish Government has indicated from Firm Foundations onwards a wish to see lower grant costs. Secondly, we are modelling capacity limits and in this context we set grant rates to be consistent with a given grant envelope (50% being consistent with the starting value for the grant envelope). In order to impose an effectively uniform grant regime on both sectors it is necessary to make a sharper move from the existing position. This common low grant regime is modelled as one of the variants on the baseline RSL model, rather than the baseline itself.

4.23. There are some more detailed differences in the way particular assumptions have been reflected. One example is the allowance for geographical variations in the level of need for additional affordable housing (see Section 3.1). Because many RSLs work across local authority boundaries, need measures derived at local authority level cannot be applied as absolute numerical caps. Therefore, the approach adopted to constraining output in low or zero need areas is slightly different in practice, although similar in general motivation and intent.

4.24. Although some parameters are subject to convergence across the sector in the early years of the projection ⁵², we do not attempt at this stage to model convergence in rent levels, nor in M&M unit cost levels. In this respect the RSL approach differs slightly from that reported in Chapter 3 for local authorities (where some rent convergence is one of the variant options). It would be possible to model convergence in either of these key variables, but this would add complexity and possibly controversy about the exercise, without necessarily adding very much to the extra capacity for housebuilding numbers.

Headline capacity scenario for RSLs

4.25. The overall results summaries for the two groups of RSLs are shown in Table 4.3. The upper part of the table shows the analysis of the 45 RSLs based on accounts data, including grossing up for the smaller RSLs included on a sample basis. It should be noted that this includes smaller RSLs sampled at one in five. The lower part shows the results for the 30 RSLs with projection data supplied. The two groups were building 2,600 and 2,850 units p.a. respectively in the base period 2008-10.

4.26. The 30 group are more active in development and have slightly more capacity to build in the future, averaging 2,060 p.a. compared with 1,840 p.a.. The 30 group are on average more geared for development with higher current and prospective debt levels, so in that sense their capacity to expand further may be more limited. A further indication of this is that their average financial performance flag scores fall to a somewhat lower level after 2016. The 45 group, however, are a more mixed cohort. Relatively few qualify for expanded output because at least one or two of their critical financial ratios falls short. In addition, more of these RSLs are located in geographical areas which we rate as having no or low need for additional affordable housing ⁵³. For these reasons, some apparent latent financial capacity within this group goes untapped. An indicator of this is the rising balance of cash and financial investments for this group.

4.27. Taking them all together, there is capacity to build about 3,895 p.a. on baseline assumptions. Capacity is rather lower (about 3,250) in the initial period 2011-15, and somewhat higher (about 4,130) after 2025, but the general impression is one of a fairly stable picture over time. For both groups debt would increase over time, rather faster for the 45 but at a higher level for the 30 ⁵⁴.

4.28. At first sight, these figures may appear somewhat alarming in that they seem to suggest a maximum capacity for RSL housebuilding well below the levels recorded in recent years - see Table 1.2. However, the overallRSL figures for Scotland are not, in fact, the appropriate comparator. This is because the £270M grant funding envelope for our modelling relates only to the AHIP 'core programme' and not to RSL construction activity funded from 'topsliced' funds (e.g. Glasgow reprovisioning programme, LCHO). Hence, the appropriate benchmark for the 3,895 capacity estimate is, in fact, 3,600 (assuming a £270M budget and grant funding per unit of £75K) ⁵⁵. Nevertheless, given that the modelled estimate is a 'maximum capacity' projection rather than an output forecast, it does suggest that it may be difficult to maintain future output at levels recently seen. Possible reasons for this apparently lower capacity are discussed below in the context of sensitivity tests, but clearly one factor is the moderate level and rate of grant assumed in this baseline run.

Table 4.3 - Main outputs of RSL capacity projections on baseline assumptions
(a) Summary results: 45 RSLs⁵⁶

(b) Summary results: 30 RSLs

4.29. It is useful to break the national output total down in different ways, to see in which geographical areas and for which types of RSL the capacity numbers are higher or lower, both in absolute terms and relative to recent output levels. Table 4.4 provides an analysis by region, using whole-local authority approximations to city-region strategic planning areas and other commonsense regional grouping ⁵⁹. It also shows the numbers by size and type of RSL and by a banding based on the need for additional affordable housing at local authority level ⁶⁰.

Table 4.4 - RSL new build output capacity projections on baseline assumptions by region, association class and need level of locality*

*Calibrated in average units p.a.

4.30. The region with the largest capacity is Glasgow and Clyde Valley which accounts for a third of the total, followed by South East Scotland which accounts for rather over one-fifth, and Ayrshire-South West which accounts for 15%. Capacity looks small for Aberdeen/shire, Central and Highland & Island regions. The only region with an increased capacity relative to recent build rates is Ayrshire-South West

4.31. Unsurprisingly, large general-purpose associations with over 1,000 stock account for the majority of capacity (62%). The stock transfer sector accounts for nearly a quarter of the total, but this is by no means dominated by the local authority whole stock transfer group ⁶¹. Smaller 'general needs' RSLs account for 12% of national capacity. Organisations primarily focussed on older people or other special need groups are shown as having relatively small capacity, but this partly reflects the fact that the sample did not fully represent these groups ⁶².

4.32. There is some degree of mismatch between the geography of capacity and the geography of need. Over a fifth (23%) of the capacity in this period is in localities shown as having no additional need for affordable housing, with another 13% in localities with limited need. The areas with the higher levels of need (above the national average) only have 35% of the capacity. If a strict view were taken of need, then it could be argued that the effective national capacity is less than shown here. This is addressed further in the sensitivity analyses and also when we compare with local authority capacity.

Alternative scenarios for housing associations

4.33. The projection model has been set up to enable projected new build numbers and other financial outputs to be easily obtained for different sets of assumptions. The ranges of main assumptions built in were discussed in section 4.1 above. Table 4.5 below provides a summary comparison of the sensitivity of new build numbers (averaged over the period 2011-35) to different assumptions. Mainly this looks at the impact of changing one assumption at a time ⁶³. However, near the bottom of the table some combined scenarios are compared. The table shows the outputs from the two groups of RSLs separately and the sum of the two; it also shows the absolute and percentage difference from the baseline.

Economic variables

4.34. Inflation does not make much difference, although higher inflation would increase capacity slightly, and vice versa. The trend in real management costs is much more significant - if real costs per unit continued to rise at 2% p.a., output capacity would fall by 19%. However, a 2% p.a. reduction in management costs would have a much smaller positive effect (8%). This probably reflects the fact that it is in the nature of this kind of model that certain relationships between indicators are non-linear. Many RSLs are projecting low increases in management costs, perhaps expecting to reap more scale economies ⁶⁴.

Table 4.5 - Sensitivity analysis summary of RSL new build output capacity projections - average annual no. of units for variations in individual assumptions , 2011-2035*

*Note that this time period is slightly different to that analysed in Table 4.4 - hence, the slight difference in the 'central estimates' between the two tables.

4.35. The borrowing rate is also quite significant, as confirmed by some of stakeholder interviews, where there is considerable nervousness about possibly higher rates in future. A rise of 1.5% points would reduce new build capacity by 575 units p.a. (15%). The benefit of a 1.5% point lower rate would be less, around 520 units (13%). The risk of an adverse outcome here currently seems greater than the chances of a favourable outcome.

Grant rates

4.36. We would expect the HAG resource envelope and grant rates to make a substantial difference and this is confirmed from the comparison between high and low HAG rates (it is necessary to adjust rates consistently with the HAG envelope). So raising the HAG rate from 50% to 60% would raise output capacity by less than one quarter (23%) but it would also increase the grant envelope (i.e. public subsidy cost) by a half (from £270m to £410m). This is because more units have to be financed and the average cost per unit is higher. The opposite strategy, of reducing grant rate to 25%, would reduce output by 27%, but would see total envelope (spend) drop by more than half to £116m. It is interesting that the system appears capable of sustaining a significant level of output (2,850) even on such a low grant rate. We return to this point below when discussing the direct linkage between local authority and RSL scenarios (see in particular Tables 4.7 and 4.8).

Rents

4.37. Future rent variations are, as expected, quite significant. A difference of 10% points in the initial rent shift (i.e. seeing rents rise by 15% vs. 5%, or RPI+3%, over the first 5 years) would alter output capacity by about 585 units p.a. (15%). A difference of 25% (i.e. 30% vs. 5%, or RPI+6%) would enable an extra 1,535 units p.a. (40% more) to be built. These scenarios would require progressively lower grant rates (45% and 35%) to work in terms of a fixed grant envelope.

4.38. A difference of 1% point p.a. in the longer-term rent trend (i.e. assuming annual uprating by RPI+2%) also makes a difference, particularly later on. On average the impact of 1% point p.a. on the long-term rent rise is 13% more output for an increase, or 23% less output for a reduction.

4.39. The other significant parameter which is partly related to policy is the unit cost of new provision. Higher unit costs (by £20k) would reduce output by 530 units p.a. (15%), while lower unit costs would increase output by slightly more (16%). We return to this issue in the subsequent chapter when discussing Section 75 planning agreements. There is also some uncertainty and debate about the likely future trend in real procurement costs. We assume 1% rise p.a. in real terms in the baseline, for various reasons including the generally low technical progress in construction and the element of land price/house price effect in the figure. An alternative assumption of zero real increase would also raise output capacity by about 475 units p.a. (12%).

Local need

4.40. A policy parameter which comes from outwith the financial capacity model per se is the level of need for additional affordable housing in the localities where particular RSLs mainly work. The baseline assumption is that RSLs based in areas with zero or low need may continue to develop, but only at their previous level ⁶⁵, and not at a greatly enhanced level. A more stringent assumption would be to say that RSLs working in zero need authorities would undertake no new development. This would have a negative impact on output averaging 575 (15%) over the whole period (Table 4.5). This figure is lower than the 835 shown in Table 4.4 for 'zero need' areas, the main reason being that the 'zero need filter' scenario redistributes some of the grant to other areas through the model's feedback mechanisms.

4.41. In considering policy options here, it is necessary to consider what part of new RSL (or local authority) investment in new build is not related to the shortage of affordable housing but to some (presumably high priority) regeneration or replacement agenda. It may be that some of this is covered by the current 'top slice' element of the AHIP budget.

Combined scenarios

4.42. We have constructed several combined scenarios intended to capture possible combinations of circumstances (see Table 4.6). In particular, we present two bundles of assumptions which are either favourable or unfavourable, from the viewpoint of achieving social sector output. The favourable combination includes a larger (higher) HAG envelope, a high rent shift, a higher growth profile and low unit costs for new build. This combination would enable output to rise to 7,995 units, an increase of 105%. This would not be achieved immediately - there would be a buildup period of about a decade.

4.43. The corresponding unfavourable combination is not an exact mirror image, because we feel that some factors are more likely to be adverse than favourable, in the foreseeable future. This scenario entails a rising real cost of management, a higher borrowing rate, a smaller (lower) HAG grant rate (33%) and envelope (£134m), no initial upward rent shift, and higher new build unit costs. This scenario would reduce annual output by 53% to only 1,835 units.

4.44. The individual assumptions involved here are not far-fetched, when considered individually. A combination like the unfavourable one is not inconceivable, at least in the short to medium term. Therefore, we have to conclude that the range of potential outcomes is indeed quite wide. Furthermore, it is clear that in order for the RSL sector alone to achieve higher output than in the recent past, it would be necessary for there to be a combination of conditions which are somewhat more favourable than in our baseline assessment. Conversely, it is not at all implausible to expect the level of achievable output to be markedly below that achieved recently.

4.45. One further variant assumption included in Table 4.6 concerns possible requirements for further expenditure on improvements to the existing stock post achievement of the SHQS in 2016. Whereas the baseline assumes that this element of expenditure steps down by half after this date, the variant entails continuing this expenditure at the same level indefinitely. This might cover work on improved energy efficiency for example. The impact is shown in the last row of Table 4.6, and amounts to a reduction of about 510 units p.a. (13%). This might be an underestimate of the impact, given the way it has been treated in the models for the two groups of RSLs, particularly for the group of 30 RSLs ⁶⁶. It would also be an underestimate if the extra expenditure for energy measures were greater in scale.

Table 4.6 - Sensitivity analysis summary of RSL new build output capacity projections - average annual no. of units for variations in packages of assumptions, 2014-2034

Low grant scenarios

4.46. There is perhaps, at the present juncture, a particular interest in combined scenarios focussed around lower grant levels. Two obvious reasons for this are (a) the prospects of reduced availability of public spending resources, and (b) the emergence of the local authority new build sector operating, currently, on significantly lower grant rates. We present within Table 4.6 two additional scenarios incorporating low grant (25%), one of which combines this with certain other favourable assumptions which to some extent compensate for the low grant level ('Low Grant & Cost'), and the other of which (responding to suggestions from stakeholders commenting on the draft report) includes low grant, higher borrowing rate, a medium upward shift in rents, and static real procurement costs.

4.47. The more favourable of these two scenarios combines a low grant rate (25%) and constrained HAG envelope (£134m) with a low level of unit costs for new build (maximising the use of s.75 and free public land). Under this scenario RSL output is rather below the baseline level, with a reduction of 13% to 3,376 units p.a. This demonstrates that the notion of operating a lean system in terms of grant need not be completely hopeless in terms of the prospects for output, if there is an ability to keep costs down. The further variant shown in Table 4.6 has a similar net effect, seeing output down by 468 units to 3,426 (a 12% reduction). Higher borrowing costs and lower grant are not fully offset by higher rents and less cost growth. Other results in Table 4.5 show that combinations of lower costs, including management costs, and somewhat higher rents could increase output with lower average grant rates, but these would require the maintenance of a grant budget similar to the present level.

4.48. The scenario which is intended to be directly comparable with the central local authority sector scenario has the same low grant rate (25%) with a low grant envelope, but this time no reduction in unit costs for new build. This is shown as 'Low HAG Rate' in Table 4.5. This would generate RSL output of 2,850 units p.a. over the whole period, a reduction of 27% on our baseline, and not much over half of the level of RSL new build in the last six years.

Comparison with local authority projections

4.49. At this point it is useful to draw on the results of the local authority projections as set out in Chapter 3 and make a like-for-like comparison with the last of the scenarios described above for the RSL sector. To reiterate, these are both low grant scenarios, with average unit grant in the £30-35k range in today's values. Table 4.7 shows the comparison over twenty-year periods broken down by region and local need banding. It also shows the local authority projection numbers excluding outputs in excess of the local need estimate (which can be directly observed at local authority level). A rough comparison of zero need-filtered RSL output can be made by subtracting the output shown in column 2 against the zero need band (i.e. 756 units), from the total output (2,661), which equates to around 1,900 units.

4.50. Broadly speaking, the two sectors could produce similar levels of output (2,660 and 2,485) from a similar grant regime. The combined total would be 5,145, which can be compared with the 2009/10 Scottish Government 'mainstream social rented' grant funding approvals total - 4,861 ⁶⁷. However, if we filtered out the low/zero need localities, the modelled output would be less, at about 3,900 units p.a.. It appears from this table that the grant cost would still be higher for the RSL part of this output. This may be because of the initial transition period, as the model shows the RSL sector spending over the grant envelope in the early period ⁶⁸.

Table 4.7 - Comparison of RSL and local authority output on common low grant assumptions - average annual units

4.51. It is interesting to compare the modelled outputs of the two sectors in the different regions and in the different types of locality in terms of need. The RSL sector could deliver more in Ayrshire- SW, Glasgow and Clyde Valley, Highlands and Islands and Tayside areas (leaving aside issues about need), whereas the local authority sector appears able to deliver more in Aberdeen/shire, Central, and South East Scotland regions. The local authority sector seems to be able to deliver more in the higher need districts whilst the RSL sector can deliver more in the lowest and moderate need bands.

4.52. How would the two sectors respond to variations in the grant rate? Table 4.8 gives a picture of this by comparing headline output from the two models at ten different grant rates, using otherwise common baseline assumptions ⁶⁹. Although the levels of output are (as noted above) similar, especially at the 25% grant level, the output increases with grant at different rates in the two sectors. At 5-25% grant the RSL sector capacity is greater - this may be because of slack in the sector in terms of cash balances, higher rents, etc. However, as grant rates increase the local authority sector appears able to expand output more, suggesting that this has more latent financial capacity or ability to lever in borrowing ⁷⁰. However, any such expansion would entail a large (and probably unrealistic) increase in the total level of grant.

4.53. Analysis of such a wide range of grant rates is theoretically interesting but raises further technical and substantive issues. For example, above the level of about 65%, new social rented provision can arguably be delivered without cross-subsidy from an existing social rented stock. Furthermore, as grants rise above this level, they may be used to cross-subsidise other activities, or indeed to push up land or procurement costs. While it would be interesting to debate these issues, it is unlikely that (in the case of the high grant rates and total grant costs) these will be realised in the foreseeable future.

Table 4.8 - Annual output and grant costs (2010 prices) at different grant rates for both RSLs and local authorities

4.54. The figures in Table 4.8 can perhaps be better appreciated from graphical presentations, as in Figures 4.1-4.3. Figure 4.1 shows how output rises slowly (less than proportionately) with grant rate, but that it rises more steeply for the local authority sector in the middle range of grant rates. Although some RSLs are able to be somewhat more responsive at higher grant rates, others are constrained by their mixed financial performance scores, while some are also limited by the assumptions we have made about need. Figure 4.2 shows the corresponding pattern of grant costs associated with these grant rates and outputs. These tend to curve upwards because higher grant rate has a dual effect, on both the development capacity numbers and on the per unit grant cost. Figure 4.3 shows the overall relationship for the two sectors combined between grant outlay and output.

4.55. These analyses also provide some indication of where grant rates might be pitched, and what levels of output might be expected, given different levels of overall public spending resources available. For example, if the resource envelope were just slightly above its current level (at £309m) then one might look at grant rates at the c.35% level with a maximum expected output of just over 6,000. A halving of resources to £160m might be compatible with a maximum capacity of about 5,000 units with grant rates around 20-25%. It should be emphasized, however, that this is based on a strict capacity approach, and may not be a realistic picture of how an optimal resource allocation mechanism would work in practice. It may be necessary and appropriate to set grant rates rather higher than the minimum, so as to generate sufficient bids from providers to be confident of fully utilising the budget envelope and also to enable the Scottish Government to apply some basic sense and quality controls on the programme.

Figure 4.1 - Output capacity by grant rate and sector

Figure 4.2 - Grant cost by grant rate and sector

Figure 4.3 - Combined output and combined grant input