Does increased capital grant actually deliver more social rented housing? Evidence from a quasi-experiment in England
Dr Michael Marshall (University of Sheffield)
Dr Meng Le Zhang (University of Cardiff)
Background
Long-term decline in social rented housing
Increased affordability pressure, especially among renters
Estimates suggest 90,000 new social rented homes needed annually (Bramley, 2018)
2
Background
Government in 2014
“This review [of how to increase social housebuilding] will assess if councils are making sufficient use of their existing powers and flexibilities to deliver new social housing. For instance, councils could use their property portfolio more effectively to finance housebuilding by selling expensive vacant properties and using the receipts to build new affordable homes.”
Government in 2025
3
Why look at grant? Isn’t it obvious?
Political scepticism
Funder uncertainty on:
Lack of academic research on causality
4
What does previous research tell us?
Efficacy of subsidy depends upon policy design and the institutional context (Yates and Whitehead, 1998)
Institutional context:
Comparative research suggests higher grant-per-home delivered proportionately more social rented housing in Scotland (Gibb, 2021)
5
2018 policy change
Social rented grant available in areas of ‘high affordability pressure’
Gap between weekly social and private rents >= £50
Social rent grant could be used outside these areas, but couldn’t get extra grant
June 2018 - Homes England’s budget increased by £1.67bn
Target = 12,500 social rent homes
6
Source: Shared Ownership and Affordable Homes Programme 2016 to 2021 Addendum to the Prospectus, June 2018
Policy in action
Region | Low-cost homeownership | Affordable rent | Social rent | All programme average |
East Midlands | 32,050 | 38,252 | 46,050 | 36,293 |
East of England | 32,736 | 37,323 | 64,390 | 39,736 |
North East | 36,263 | 35,866 | 52,850 | 36,231 |
North West | 32,008 | 38,117 | 55,409 | 36,620 |
South East | 80,304 | 38,238 | 56,643 | 39,189 |
South West | 33,443 | 36,174 | 64,151 | 38,103 |
West Midlands | 30,721 | 38,808 | 50,267 | 37,399 |
Yorkshire and The Humber | 33,919 | 38,485 | 58,701 | 37,519 |
England | 33,018 | 37,818 | 57,580 | 37,497 |
Methods
Quasi-experimental approach
Fuzzy regression discontinuity design (RDD)
Local average treatment effect (LATE)
Four outcomes using 2019/20 data:
Robustness checks
8
Social rented starts
Increased probability of capital grant take up by 44.6 percentage points
0.71 starts per 1,000 dwellings (p < 0.05)
58.7 homes per authority
3,405 homes in 2019/20
9
HA starts
0.68 starts per 1,000 dwellings (p < 0.05)
56.2 homes per authority
95.7% of the total increase
10
LA starts
No evidence of effect
May have lagged effect in following years
Observational data suggests delivery is greater in London
11
Affordable housing starts
No evidence of effect
Not catalysing delivery of more affordable housing overall
12
Robustness checks
13
Robustness checks
Controlling for private starts, median earnings, household growth, private sales, existing social housing supply, and population over 65
Different specification of the functional form
Triangular kernel
Placebo tests – random cutoffs and median earnings
Varying the bandwidth
14
Conclusions
Increased capital grant for social rented housing does have a causal effect on supply
The 2018 policy change could plausibly have met its target, but events got in the way (e.g. COVID-19, Trussonomics)
Effect was much greater on HAs than LAs – institutional context matters!
No evidence of catalytic effect on total number of affordable homes. Massive injection of cash is needed to get more homes overall and higher proportion of social rented homes
Limitations of research
15