In September 2008, the UK government announced a surprise stimulus policy in response to a dramatic fall in the housing market: a property transaction tax on houses sold in a certain price range was temporarily eliminated. This column reports research showing that this stimulus boosted transaction volumes by 20% and increased consumer spending by an amount equal to the forgone tax revenue. Cutting transaction taxes during economic downturns can thus be an effective way to stimulate both the housing market and the broader economy.
Throughout 2008, the UK housing market was in freefall. As Figure 1 shows, between the fourth quarter of 2007 and the first quarter of 2009, the number of residential property transactions plummeted by 60%. Over the same period, house prices fell by almost 20%.
Amid this market turmoil, the government stepped in to shore up the housing market. As part of a surprise stimulus policy announced in September 2008, the government temporarily eliminated the property transaction tax — the stamp duty land tax — on properties sold between £125,000 and £175,000, estimating that this would allow 60% of all properties sold to pay no tax.
But is cutting a small transaction tax (in this case 1% of the value of a property) an effective way of stimulating the housing market or the wider economy? Prior research on stimulus programmes that encourage people to purchase durables suggests that the effects are short-lived and are undone very quickly after the stimulus is removed (Mian and Sufi, 2012). This would imply that stimulus policies have to be sustained throughout the economic downturn to have a sizable impact, which would be expensive for a government to maintain.
Furthermore, it could be argued, a temporary stimulus to house transactions is unlikely to encourage construction of new houses, so the policy will mainly cause additional trading of existing houses. Shuffling assets between people isn’t an obvious way to increase GDP in a recession.
Our research provides new evidence that challenges these views (Best and Kleven 2017). We show that the UK housing market stimulus programme increased monthly transaction volumes by 20% while it was in place, and only 30% of this boost was undone by a lull in activity following the withdrawal of the programme.
We also find that the increase in transactions had large effects on consumer spending. This is due to the fact that trading houses implies moving house for residential property owners, and moving tends to trigger considerable spending.
Our estimates suggest that each pound of forgone revenue during the temporary tax cut generated at least one pound of additional consumer spending. By contrast, income tax rebates — another common form of fiscal stimulus — generate much less spending (Shapiro and Slemrod, 2003a,b; Johnson et al, 2006; Agarwal et al, 2007; Parker et al, 2013).
Tax stimulus propels housing market activity
On 1 September 2008, the government announced a temporary tax exemption that lasted until 31 December 2009: the ‘stamp duty holiday’. Houses sold in the price range of £125,000 to £175,000 became exempt from tax, whereas such transactions would normally have been subject to a tax of 1% on the property value.
Was this stimulus measure successful at increasing the number of houses sold during the holiday period? And if so, was this just a short-lived effect? How many of the extra house sales would have happened at a later date anyway, and from how far in the future were house sales brought forward?
To investigate these questions, we study the evolution of housing market activity in the targeted price range £125,000–£175,000, and compare it to activity in a neighbouring, untargeted price range.
When doing this type of analysis, we carefully adjust for the fact that homebuyers may move between the targeted and untargeted price ranges in response to the stimulus programme, thus distorting the comparison of transaction activity in the two groups. The results that we discuss below are therefore not affected by such price responses.
The visual evidence in Figure 2 shows that the stimulus programme was effective both at slowing the fall in the housing market at the end of 2008 and at speeding its recovery in 2009. The programme increased activity in the targeted price range by an average of 20% in the months that it was in place, according to our estimation.
Moreover, the reversal of these effects after the withdrawal of the programme appears to be relatively modest. We estimate that the stimulus programme lowered activity levels in the targeted range by 8% per month for about a year.
Figure 2: Effects of the Stimulus Program
Notes: The red series show the evolution of the number of monthly sales in the price range targeted by the stimulus. The blue series show the evolution of our comparison group. To facilitate comparison, we normalize both series so that they each average to zero in the months leading up to the tax stimulus in September 2008. The solid vertical lines in the figure mark the beginning and end of the stimulus programme.
The stimulus programme was effective both at slowing the fall in the housing market at the end of 2008 and at speeding its recovery in 2009
As Figure 2 shows, the two series move in parallel up until the tax stimulus is announced and then they start to diverge. Transaction levels are higher in the targeted group in every month that the policy is in place. The end date of the stimulus was pre-announced and so we see a spike in activity in the final month of the stimulus as people rush to complete transactions before the programme is removed.
Immediately after the removal of the policy, transaction levels in the targeted price range drop sharply and stay marginally below transaction levels in the comparison range for about a year. This slump in activity suggests that some of the additional transactions during the stimulus programme were the result of re-timing – that is, they were brought forward from the period after the tax holiday and are therefore missing in the post-stimulus period. After about a year, however, the series return to moving in parallel as they did before the stimulus.
To assess how much of the stimulus was undone in the months following the end of the holiday, Figure 3 shows the cumulative impact of the stimulus. The green line climbs rapidly during the stimulus programme and then falls gradually during 2010 due to stimulus reversal.
But the line levels off in 2011 and 2012, at which point only about 30% of the stimulus effect was undone. This implies that the large stimulus effect did not primarily reflect short-term timing effects. Instead, the effect came from transactions that would not have taken place otherwise or that were brought forward from far into the future.
Figure 3: Cumulative Impact of the Stimulus
Notes: The blue and red series show increased activity in the targeted and comparison price ranges, respectively, while the green series shows the cumulative difference between the two groups. All series are normalized to average to zero in the months before the stimulus.
This evidence stands in contrast to findings from another popular type of stimulus programme during the Great Recession: temporary subsidies to car transactions such as the ‘cash-for-clunkers’ programme in the United States. The additional car purchases induced by cash-for-clunkers reflected re-timing over the very short-term (Mian and Sufi, 2012). As a result, the stimulus effect was fully undone within 10 months of the programme ending.
Why are the effects from the two types of programmes so different?
First, housing markets operate differently from car markets. In particular, homebuyers are often constrained by the need to make large down-payments out of their liquid savings. As our study shows, the imposition of transaction taxes reinforces such down-payment constraints, amplifying the effect of taxes on prices and transaction volumes.
This amplification is particularly strong when there are many liquidity-constrained and highly-leveraged households, as around the Great Recession. As a result, tax cuts to housing during downturns can have particularly large stimulative effects.
Second, the US cash-for-clunkers programme was very short (one month) and largely unanticipated, giving households very little time to respond. It seems natural that such a short programme primarily induced households who were already planning to buy a new car to do so sooner.
The contrast between our findings and the cash-for-clunkers findings suggests that, for stimulus to be successful, governments must commit to the policy for long enough to allow households to change their spending plans.
Around the time of the UK stamp duty holiday, the US government was pursuing a similar policy: the First-Time Homebuyer Credit (FTHC) between 2008 and 2010. The US policy also provided a tax subsidy to house transactions, but one that was targeted to first-time buyers only.
Berger et al. (2017) find that the US housing stimulus had strikingly similar effects to those of the UK stimulus. FTHC increased transaction volumes among first-time buyers by between 16-25%, and there was little reversal of this effect within two years following the policy.
Moving houses triggers lots of spending
Why would we care that policy-makers are able to boost trading of (mostly) existing houses? A reshuffling of housing assets does not obviously affect spending or GDP, which is the main goal of stabilization policies.
But when households move, they spend significant amounts on items such as repairs and improvements, removals, furniture, appliances, and commissions. This implies that stimulating house purchases can have large effects on household spending.
Stimulating house purchases can have large effects on household spending.
To estimate such moving-related spending, we make use of a different data source that surveys households in detail on their spending patterns and on when they bought their homes. We estimate the effect of moving on spending based on spikes in moving-related spending categories around the time of moving (an event study approach), controlling for potential confounders that may impact both moving and spending. We check that the observed spikes in moving-related spending are not merely crowding out other forms of spending (say, restaurant visits or clothes).
We conservatively estimate that moving triggers additional spending of about 5.1% of the house value in the year of the move, and another 0.7% of the house value in the subsequent year. Together with our estimates of the increase in housing market activity during the stimulus (20%) and the size of the tax cut (1 percentage point), this implies that the spending increase per pound of tax cut equals 1.02 (in the year of the move) or 1.15 (in the two years after the move).
To compare, previous micro studies of the spending effect of tax stimulus, such as income tax rebates, have found much smaller effects, in the range of 0.3 to 0.7 (Shapiro and Slemrod 2003a,b; Johnson et al. 2006; Agarwal et al. 2007; Parker et al. 2013).
The moving-related spending channel we propose has recently been confirmed for the United States. Benmelech et al. (2017) estimate that moving houses in the United States induces substantial spending on home durables and improvements. Their effects are similar in magnitudes to what we find for the UK.
They also show that the types of house buyers most similar to those who benefited from the UK stamp duty holiday — younger, lower-income, first-time buyers — are the most responsive. These findings confirm that stimulus policies targeting the transfer of an existing housing asset may generate large increases in consumer spending and hence in GDP.
What have we learnt?
Governments have a range of fiscal tools at their disposal to stimulate the economy during economic downturns. Tax stimulus is often more feasible than spending stimulus for political and practical reasons, as we saw in governments’ responses to the Great Recession. Our findings offer a number of lessons for the design of tax stimulus policies.
First, even small tax subsidies to house transactions can have large impacts on housing market activity during downturns. While we have studied the temporary elimination of a transaction tax in the UK, our findings are applicable to a wider set of instruments such as the US homebuyer tax credit for first-time buyers or other transaction subsidies. We have argued that the large effects arise from how transaction taxes/subsidies interact with down-payment constraints and leverage.
Second, stimulating trading of existing houses can have real effects, because of the complementarities between moving houses and spending. We find large effects of moving on household spending in moving-related categories that do not crowd out spending in other categories. As a result, the spending effect of the UK transaction tax stimulus per pound of forgone tax revenue was around 1. This is higher than the spending effect of conventional tax stimulus such as income tax rebates.
Third and finally, the stimulus impact of such tax subsidies are not reversed after the withdrawal of the policy. This is because the additional house transactions do not reflect short-term timing responses alone, but also transactions that would not have taken place otherwise or that have been brought forward from far into the future.
To conclude, our research shows that tax subsidies to house transactions can be a very powerful form of fiscal stimulus that policy-makers should seriously consider in future downturns.
This column summarises ‘Housing Market Responses to Transaction Taxes: Evidence From Notches and Stimulus in the U.K,’ by Michael Best (Columbia) and Henrik Kleven (LSE), forthcoming in Review of Economic Studies.