Lies, damn lies, and making use of statistics

Lies, damn lies, and making use of statistics

Using statistics often requires a compromise of some sort, because a statistic is usually a simplification of a more complex issue. This means stats are often misused, or misunderstood.

We were recently criticised (along with the Institute for Fiscal Studies (IFS) – always a good organisation to be grouped with) by Ian Mulheirn on our use of the local authority level Valuation Office Authority (VOA) data on two-bed private sector housing rents. We were told that the VOA data does not show how rents in the aggregate have changed over time, and that the correct source for rent inflation is the Index of Private Housing Rental Prices (IPHRP).

…Quite correct. The IPHRP is the best measure of rent inflation we have, despite still being an experimental statistic itself.[1]

So why have we and the IFS sought out other sources of data on the private rental market? Unlike Ian, who is keen to promote his technical description of the housing market, we want to better understand why so many of our clients are looking for help.[2]

Mix adjustment

The key difference between the IPHRP and VOA is that the former uses a mix adjustment process. Homes available in the rental market are changing all the time. If a small two-bed home is taken out of the rental market, and a large two-bed replaces it, prices on average will go up – but that’s not the same as rent inflation being high. Mix adjustment tries to remove this issue by using a smaller sample of the full VOA data, focusing on the exact same homes over time.[3]

We chose the VOA dataset to look at a much more local level (something the government seems happy to do when setting Housing Benefit levels) because renters can’t mix adjust themselves. Regardless of the statistical process used to produce a good index, like the IPHRP, that is no consolation to a renter looking for a home. Households have to choose a rental home from what is available in the area that they need to live when they go looking. In the last 10 years, as house prices have soared, more and more families have been forced into the private rental sector so choice is limited.

You can’t always get what you want

Ian says using the VOA data exaggerates the changes in rent rises, but the difference between VOA and IPHRP numbers could suggest that unaffordable stock is coming into the rental market. Whether that’s because it’s in ever more expensive areas, or because the properties are of a higher quality, it still makes the options available less affordable for those looking for a home. It’s like needing to buy a car for work but only having the option of a Rolls Royce… It may be a nice car, but it’s still going to break the bank – but sir, this year’s model is only 2% more expensive than last year!’.

Fundamentally, the IPHRP is an index of rents, but we want to know what renters are having to pay – subtly different, but that’s statistics for you.


The truth of the matter is, much like the VOA data, the IPHRP has its limitations:

  • it only gives an inflation figures for a controlled stock of rental homes at regional level only
  • it does not provide the average price of a rental property in any of the regions
  • it doesn’t give numbers for different sizes of homes (the VOA is split by number of bedrooms, which we used)
  • both are based on data that is voluntarily[4] provided by landlords – so if a rogue landlord doesn’t want to disclose big rental increases, they don’t have to
  • it cannot describe what is happening in a more local housing markets – the types of markets that a renter will have to use as their search area so as to stay close to work, schools, family and friends.

The IPHRP can only give you a percentage by region. What use is knowing rents in the East Midland are only going up by 2.9% – the region stretches from Aynho near Banbry, to Brocklesby, south of Grimsby (they’re 134 miles apart).[5]  That’s not a commute, is it?

So, in short, we will continue to make use of data that may better help us to understand the failings of the housing market – because accepting a single statistical source seems like a gross simplification.


[1] The experimental statistic status means that although it is produced to rigorous methodological standards, it has not yet been confirmed as meeting the standards required of a national statistic.

[2] Details of the numbers of calls, face-to-face visits and searches on our website can be found here in our annual Impact Report.

[3] Yes, the IPHRP uses the VOA data. However, mix adjustment is hard to do as there is so much change. The IPHRP substitutes homes from the sample and uses around 120,000 of the near half million records collected.

[4] Private rents are collected voluntarily and currently government are not able to mandate landlords to disclose the rent they charge should they chose not to (

[5] According to Google Maps, using the walking function to plot direction.