The National Fraud Authority estimates that housing tenancy fraud, including unlawful sub-letting, costs £900 million a year. This lengthens waiting lists for legitimate housing applicants. Councils recovered nearly 1,800 homes in 2011, with a total replacement value of nearly £264 million. But this is pitiful if anything like 1.6 million social homes prove to be sublet - see below.

Experian suggest that at least 160,000 social homes are unlawfully sublet in the UK, but as many as 160,000 social housing tenants in London alone may be subletting their properties. In Westminster a raid on one Paddington housing block revealed 75% of housing benefit claimants were not living in their registered properties and were illegally subletting them for thousands of pounds a week. Another raid on the luxury 600-flat Park West development on Edgware Road found 61% of claimants were subletting their properties. A study of social housing fraud in Westminster is reported here - it may be as much as £22m a year in Westminster alone.

Social housing fraud is the wickedest form of welfare fraud. There's more than the financial dimension: there isn't enough social housing available, so every detection gives a poor family in temporary accommodation a better life.

Freeing up sub-let properties would be the cheapest and quickest way to make more social housing available. There are 8 million council or housing association homes in England and 1.8 million households on the waiting list. If 20% of these homes right across the country did prove to be sublet, that would free up 1.6m homes.

So the scope for good is potentially massive.

Where do these figures come from? It's data matching - still much underused. A matching of 27,000 tenants found "indications of fraud" in 5,300 cases. That's almost 20%. Yet the government's official estimate is that less than 1% – 50,000 properties in the whole country – are fraudulent.

• In 2,120 cases (8% of the total) they found "red" indicators of fraud, where the registered tenant had a mortgage, bank account, active credit or utility bills at another residential address.
  
  
• In 3,180 cases (12%) they found "amber" indicators of fraud – active credit, bank accounts, Sky TV or utility bill records held by a person with a different surname at the tenancy address, but no such activity there by the registered tenant. (The "amber" figure excludes, as far as possible, legitimate residents with different surnames, such as live-in partners and family members from previous relationships. They excluded such people from its totals by data-matching them with the registered tenants.)
  
  
• There was also a third category of properties where the tenant could never be contacted by the council or where neighbours or maintenance men had reported suspicions. If these were included the figure of suspect tenancies could be nearer 25% in some areas.

Another report on more than 30 social housing providers showed potential fraud in 6% of tenancies. But this seems to cover different ground from the previous study, concentrating on people with multiple social tenancies.

Stoke-on-Trent alone has recovered 44 properties recently, with another 19 in process. They identify three main types of social housing fraud:

• Application fraud, where tenants have obtained properties by lying about their circumstances
  
  
• Tenancy succession fraud, where the legal tenant dies or moves out but friends or relatives keep the house
  
  
• Sub-letting, where a tenant rents out the council's property to rake in personal profits.

Not all of these would be caught by the two studies. So the overall number might be even higher than 20-25%.

Starting from data matching, Hillingdon has recovered nearly 40 London properties – equating to around £500,000 in realised revenue for the local authority. Hammersmith and Fulham has identified potential savings of £654,000 within the first eight months. It is expected to top £1 million in 2012-13.