The impact of interest and the housing market on the UK economy....... The Chancellor has asked Professor David Miles to examine the UK market for longer-term fixed rate mortgages. This paper by Adrian Cooper, which is part of a study commissioned by the Miles Review, presents the results of a series of simulations using the of the UK economy to investigate the contribution of the housing market to macroeconomic volatility and the implications of changing the structure of mortgage finance from the current variable rate system linked to short-term interest to a fixed rate system linked to long. The main findings are that the housing market has been a contributor to past volatility in the UK economy, and that moving to a fixed rate structure would reduce the impact of a change in interest on key macroeconomic indicators.... I. Introduction This paper presents the results of a series of simulations conducted using the of the UK economy to inform the Miles Review of the UK mortgage market. The paper is organised as follows: In Section II we illustrate the importance of house price volatility for the macroeconomy by presenting two counterfactual scenarios one holding house prices constant relative to income since 9 and another in which house prices are assumed to have risen at a constant nominal rate since 9. In Section III we consider whether this volatility would have been lower if households demand for housing and mortgage financing had been related to long-term interest rather than to the predominantly variable rate mortgage products actually available over the last years. In Section IV we consider the impact of a change in short-term interest on the UK economy and the extent to which the scale of that impact reflects the direct effect of mortgage interest on house prices, mortgage borrowing and consumer spending. In Section V we compare the impact of a change in short-term interest on the UK economy under four forms of mortgage finance: to short-term interest. to long-term interest rate. Fixed long-term rate mortgages with no prepayment. Fixed long-term rate mortgages with free prepayment. II. The Importance of House Price Volatility for the Macroeconomy The UK housing market is widely considered to be an important source and/or channel of transmission of volatility at the macroeconomic level. As Chart shows, there have been very marked cycles in house prices most notably the boom of the late 9s, the recession of the early 99s and the boom of the last few years. Typically, these have been associated with marked cycles in overall economic growth, as shown in Chart. How important has the past volatility in house prices been in causing volatility at the macroeconomic level? We have used the OEF Model to inform the answer to this question by calculating how the economy might have performed if house prices had followed a less cyclical profile than actually observed since the early 9s. We have constructed counterfactual ie historic scenarios under two alternative assumptions for house prices: (i) House prices remained constant relative to household disposable income from 9 onwards (ie house price inflation was equal to household income growth). (ii) House prices rose each year from 9 at a given nominal rate taken to be.% pa, which is the average annual increase between 9 and. See, for example, EMU and the monetary transmission mechanism and Housing, consumption and EMU, HM Treasury EMU Studies, June Interest and the housing market January
Chart : UK house price inflation % year % year Chart : House price scenarios 9= 9= Constant growth rate (.% a year) Constant relative to household income Nominal Real 9 9 9 99 99 99 99 99 Source: ONS 9 9 9 9 99 99 99 99 99 999 % year Chart : UK house prices and GDP Real GDP (RHS) Nominal house prices (LHS) 9 9 9 99 99 99 99 99 Source: ONS % year Chart compares the profile for house prices in the counterfactual scenarios with that actually recorded. Clearly, house prices are much less volatile in the two counterfactuals, although volatility in disposable income means that there are still significant movements in house prices in case (i). The difference between the behaviour of macroeconomic variables such as GDP as actually recorded which we take as our base case and that predicted in these counterfactual scenarios then provides a measure of the contribution of house price volatility to macroeconomic volatility. Two points should be noted in interpreting these simulations: First, we have made no allowance for any impacts that changes in house prices may have on consumer confidence over-and-above that already implied in the equations. In practice, however, such affects could be an important part of the transmission of volatility in house prices to overall macroeconomic volatility. Similarly, we have not allowed for any direct impacts on investment in dwellings or housing turnover. Second, our counterfactual simulations allow for a reaction on the part of the UK monetary authorities to changes in the macroeconomic environment. This is based on a version of the Taylor rule that relates interest to both the output gap and the difference between inflation and its target rate. So, in periods when the counterfactual scenarios imply more modest house price inflation than in the base case, and therefore more subdued overall demand, this is likely to be partly offset by interest being lower than in the base case, and vice-versa. It should be stressed, however, that the parameters of this Taylor rule have not been changed to reflect the alternative behaviour of house prices in the counterfactual scenarios. In practice, the monetary authorities could well have adopted a very different policy rule in these circumstances, given the implied change in the monetary transmission mechanism. Chart compares the profile of GDP growth in each of these counterfactual scenarios with its actual behaviour. Chart does the same for RPIX inflation. Table compares the volatility of growth in GDP, consumer spending and house prices across the scenarios. The key points to note are: The cycle in both GDP growth and RPIX inflation are significantly less marked in the two counterfactual scenarios than in the base case. Nevertheless, considerable volatility remains, and marginally more so in the case where house prices are held constant relative to incomes. The implication of these counterfactual simulations is that the behaviour of house prices accounts for part but by no means all of the volatility of the UK economy over the last years. January Interest and the housing market
Table : Impact of house prices on volatility deviations of year-on-year growth, 9QQ Base case Constant house prices relative to income Linear trend in house prices GDP.9.. Consumer..9. spending House prices... % year % year Chart : GDP growth in scenarios Constant house price growth House prices constant relative to household income 9 9 9 9 99 99 99 99 99 999 III. The Impact of Short-Term vs Long- Term Interest Rates on Volatility In this section, we consider whether the volatility seen in house prices and in the overall macroeconomy would have been lower since the early 9s if mortgage interest had been linked to long-term rather than short-term interest. Chart compares the behaviour of short (-month interbank), long (-year) and average mortgage since 9. It shows the close relationship historically between changes in average mortgage and short-term interest. But it should be noted that the differential between mortgage and short is not constant in particular, the margin tends to narrow when short are very high (eg 99-9, 999) and widen when short fall sharply (eg 99-9, -). Chart also clearly shows that the cycles in long-term interest in the UK have been much more muted than those in either short or average mortgage, although all three have been on a similar downward trend over the last years or so. % % Short-term Chart : UK interest Mortgage Chart : RPIX inflation in scenarios % year % year Long-term 9 9 9 9 99 99 99 99 99 999 Constant house price growth House prices constant relative to household income 9 9 9 9 99 99 99 99 99 999 We have used the to construct a counterfactual scenario in which it is assumed that average mortgage historically were linked to long rather than short. In preparing this scenario, we assume that mortgages historically would have been -year fixed rate arrangements with no prepayment allowed, with the interest rate on new mortgages taken out in any period given by that quarter s long rate plus a margin of basis points. The average mortgage rate paid by households would then effectively have been equal to a -year moving average of long-term interest plus a basis points margin. Interest and the housing market January
Before presenting the results, it is important to stress that such a counterfactual exercise does need to be interpreted with caution. It is possible that the response of the housing market to changes in interest would have been very different in the past had mortgages been linked to long rather than short, whereas here we assume that the OEF Model equations would have remained valid. Chart compares the profile of house prices since 9 in our counterfactual scenario with its actual behaviour. Strikingly, it suggests that linking mortgage to long rather than short would have exacerbated the house price boom in the late 9s, since long were generally at least as low as short over this period and at times (eg 9-9) they were considerably lower. The recession in house prices in 99-9 would have been delayed had mortgages been linked to long, since the peak in long was much lower than that in short. However, house prices would ultimately have fallen even further than actually recorded, since the average rate paid by households under a system of -year fixed rate mortgages would have fallen only gradually in the early 99s. More recently, house prices would have still risen very strongly under a mortgage system linked to long. 9= 9= Chart : House price scenarios Long in house price/mortgage equations Long also in consumption equation 9 9 9 9 99 99 99 99 99 999 These differences in the behaviour of house prices are reflected in the performance of the overall macroeconomy. In calculating these effects, we have produced two versions of our counterfactual scenario: (a) The is modified so that the equations for house prices, mortgage borrowing and household interest payments are linked to long rather than short. (b) As (a) but with the consumer spending equation also linked to long rather than short. (Such a change might be justified if households perceptions of the cost of borrowing/foregoing saving is more closely related to the mortgage interest rate they pay than other measures of interest, although this is clearly arguable.) The same caveats with regard to consumer confidence and monetary policy apply here as in the scenarios presented in Section II. Chart shows the behaviour of GDP in these scenarios, while Chart 9 does the same for RPIX. Table compares the volatility of growth in GDP, consumer spending and house prices across the scenarios. The key points to note are: The suggests that the cycles in GDP growth and RPIX inflation would have been even more marked had mortgage been linked to long rather than short. There would have been slightly less volatility if consumer spending as well as house prices had been linked to long. This result partly reflects the additional volatility of house prices under these scenarios. But it also appears to reflect the impact of the change in the relationship between interest and the behaviour of the macroeconomy that is not compensated for by a change in the way in which the monetary authorities are assumed to set interest. While, as explained in Section II, the OEF Model includes an interest rate reaction function, this has not been reparameterised to take into account the different structure of the economy implied by linking mortgage to long rather than short. % year % year Long also in consumption equation Chart : GDP growth in scenarios Long in house price/mortgage equations 9 9 9 9 99 99 99 99 99 999 January Interest and the housing market
% year % year Chart 9: RPIX inflation in scenarios Long in house price/mortgage equations 9 9 9 9 99 99 99 99 99 999 Table : Impact of using long-term rather than shortterm linked mortgage on volatility deviations of year-on-year growth, 9QQ Base case House price/mortgage borrowing equations linked to long Long also in consumption equation Consumption equation also linked to long GDP.9.9. Consumer... spending House prices... Chart compares the change in GDP relative to its baseline projection in response to the rise in interest under three assumptions: (a) The standard (b) A modified version of the in which the direct impact of the change in interest on housing prices is switched off. (c) As (b) but with the direct impact of the change in interest on consumer spending also switched off. Chart shows the profile of house prices in the three scenarios, while Chart shows the profile of consumer spending. The standard suggests that a % point increase in short rate will reduce GDP by up to.% relative to base, with the peak impact coming after two years. Switching off the direct effect of interest on housing prices and mortgage borrowing reduces this peak GDP impact to.% and delays it by two quarters, while switching off the effect on consumer spending as well reduces it further to.% and delays the peak by another quarter. So, about a third of the impact of higher interest on GDP ope via the direct effect on house prices, with a further sixth operating via the direct effect on consumer spending. But this still leaves a significant impact of interest on the economy, reflecting, for example, impacts on business and dwellings investment, and on the exchange rate. IV. The Sensitivity of the UK Economy to Interest Rate Changes In this section, we consider the impact of changes in short-term interest on the UK economy and assess the extent to which these impacts reflect the direct effect of mortgage interest on house prices, mortgage borrowing and consumer spending. We do this by considering an increase of basis points in short lasting for two years. Thereafter, interest are determined by the interest rate reaction function in the. This policy change could be thought of as a reaction to an unexpected inflation shock in the context of a credible monetary framework. Longterm interest are assumed to be an average of expected short, with -year rising by basis points immediately when short rise and then gradually falling back to their base level over two years. No direct effect on consumption equation as well as house prices/ mortgage borrowing Chart : Impact of short on GDP No direct effect on house price/mortgage equations Interest and the housing market January
Chart : Impact of short on consumption Chart : Impact of short on house prices No direct effect on consumption equation as well as house prices/ mortgage borrowing - No direct effect on consumption equation as well as house prices/ mortgage borrowing - No direct effect on house price/mortgage equations - No direct effect on house price/mortgage equations - V. The Impact of Interest Rate Changes Under Alternative Forms of Mortgage Finance In this section, we consider the impact of a change in short-term interest on the UK economy under four forms of mortgage finance: (a) to short-term interest ie as in Section IV. (b) to long-term interest ie as in Section III. (c) Fixed long-term rate mortgages with no prepayment In this case, we assume that households take out mortgages on which the interest are fixed for years and which do not allow early prepayment. This means that housing demand and consumption would then be influenced by a weighted average of long-term interest over the last years. (d) Fixed long-term rate mortgages with free prepayment In this case, we assume that households take out mortgages on which the interest are fixed for years but that they are able to prepay these and take out a new loan at no cost. So, when long-term mortgage fall below their previous lowest rate since the mortgage was taken out, households will refinance. As in Section IV, we consider the effect of an increase of basis points in interest for two years. And we also consider the impact both where the change to the form of mortgage finance applies only to the equations for house prices and where the change also applies to the equation for consumer spending. Chart : Impact of short on GDP Chart a: Impact of short on GDP *Assuming house prices linked to long *Assuming consumption as well as house prices linked to long January Interest and the housing market
Chart : Impact of short on consumption Chart a: Impact of short on consumption *Assuming house prices linked to long *Assuming consumption as well as house prices linked to long Chart : Impact of short on house prices Chart a: Impact of short on house prices - - Fixed long-term mortgages with no prepayment* *Assuming house prices and mortgage borrowing linked to long - - - - *Assuming consumption as well as house prices linked to long - - Chart compares the impact on GDP of an increase in short under the first three forms of mortgage finance. (We consider the results with fixed long-term interest with free prepayment separately below.) Chart shows the impacts on house prices, while Chart shows the impact on consumer spending. These results clearly show: (a) The impact of changes in interest on the economy are much larger under the current variable rate mortgage system linked to short-term interest than under a variable rate system linked to long-term interest. This is because mortgage are assumed to rise by the full basis points that short rise, while long-term interest increase by only basis points. (b) The impact of higher short is even more muted in a system of fixed long-term interest with no prepayment. This is because only those households either taking on a mortgage for the first time or having to remortgage a loan that has come to the end of its term are directly affected by the increase in long. Under a mature system of -year fixed rate mortgages, this would be expected to be only around % of households in the first year following a change in interest. So, the impact of a basis point increase in long on the average mortgage rate paid under such a structure would be only around basis points in the first year, although it might increase in subsequent years depending on how sustained is the increase in long. Not surprisingly, the impact of increased interest is greater under all of these mortgage systems if there is a direct effect on consumer spending behaviour as well as on house prices and mortgage borrowing. The impact of a change in interest under a fixed long-term interest system with free prepayment is more complicated to analyse. For one thing, it is likely to be asymmetric ie the scale of response to an increase in interest is likely to be different to that in response to a fall in interest. This is because households are free to choose whether to prepay their loans and take out a new mortgage and will only do so if the terms available make this attractive. This may well be so for many households after a fall in interest, in Interest and the housing market January
which case the effect on the overall economy could be significant. But it is a rather different matter following an increase in interest, since households can then simply choose to hold on to their existing mortgages, which would leave their mortgage payments unaffected. However, there is a further complication in analysing the impact of interest rate changes when there is free prepayment, because the response of households then depends on the path that interest would otherwise have taken - ie it is base dependent. To see this, consider the following simple example. Table : Illustrative scenarios for mortgage and refinancing Mortgage (%) Case A After rate Base rise in period Case B After rate Base rise in period Period.... Period.... Do households refinance in period? No No Yes No Here we set out two alternative base forecasts for mortgage. In period, mortgage are.% in both, which we assume to be the lowest level of mortgage for at least a decade. On this basis, with free prepayment, all mortgage borrowers would remortgage in period at.%. The base forecasts for period are, however, different. In case A, long-term interest are assumed to remain unchanged, leaving mortgage at.%. In case B, long are assumed to fall by basis points in period, reducing mortgage to.%. As a result, in the base version of scenario A, no-one would remortgage in period they would have nothing to gain having already taken out a mortgage at.%. In contrast, in the base version of scenario B, all households would remortgage again, since they could then take advantage of the new lower interest rate available to them. Now consider the impact on scenarios A and B if we analyse the effect of an interest rate increase in period that pushes up long by basis points compared with their respective base forecasts. In case A, this would mean that mortgage in period are now.%. In case B, mortgage in period are now.%. This change in interest would have no effect on households refinancing decisions under scenario A compared with the base case. Households were already choosing in the base case not to remortgage in period, and would clearly still not do so after the interest rate increase. In contrast, the effect of the increase on mortgage under scenario B is substantial. With now.% in period, there is no longer an incentive for any household to remortgage, whereas in the base case all households were remortgaging. So, the impact of the increase in interest relative to the base forecast is much greater under scenario B than A. This implies that the impact of a change in monetary policy on the economy would have to be analysed much more carefully under a mortgage system that allowed free prepayment than under the existing structure of mortgage finance since its effects would depend on the precise past and expected profile of interest. Simple ready reckoners of the effects of interest rate changes would be even less appropriate than they are under the current mortgage system. However, it should be acknowledged that the assumption that all households would respond to any incentive to remortgage under such a system is unlikely to hold in practice, given the other non-monetary costs of remortgages (eg searching out the best deal, completing paperwork, etc) and the natural inertia of many people. This would tend to dampen the effect of changes in interest in such a system and reduce the degree to which the impacts are base dependent. Moreover, these asymmetries and base dependencies will only be quantitatively significant in circumstances where long-term interest change by large amounts. They do not cause significant differences in the simulations we present here although there are some because we are considering a shock in which longterm interest change by at most only basis points. In analysing the effects of a change in interest under such a mortgage system with the OEF Model, we have considered two alternative profiles for long-term interest as shown in Table. In the first of these, there would be no incentive to remortgage in whether or not interest increased relative to base. But in the second, households would be expected to refinance in January Interest and the housing market
Q if interest were at base levels, but would not do so if were basis points higher. Chart summarises the impact on GDP for both an increase in short of basis points against these two baseline forecasts, and a basis point reduction in short against the first of the baselines. The effects of an increase in short in this case are typically slightly larger than under the assumption of fixed long-rate mortgages with no prepayment, although the extent of these differences will depend on the exact baseline profile of mortgage. But they are again substantially less than under the current system of mortgages. Table : Assumptions for long-term interest used in simulations OEF central forecast, used as baseline for simulations of both bp rise in short and bp fall Alternative forecast used as baseline for alternative simulation of impact of bp rise in short Q.. Q.. Q.. Q.. Q..9 Q.. Q.. Q.. Q..9 Q onwards.. Effects under fixed long-term mortgages with free prepayment* Chart : Impact of short on GDP Effects under fixed long-term mortgages with free prepayment* Chart a: Impact of short on GDP *Assuming consumption as well as house prices linked to long Conclusions Impact of a rate cut Impact of a rate increase under two alternative baseline forecasts for long *Assuming house prices and mortgage borrowing linked to long Impact of a rate cut Impact of a rate increase under two alternative baseline forecasts for long This paper has set out the results of a series of simulations using the to investigate the contribution of the housing market to macroeconomic volatility and the implications of changing the structure of mortgage finance from the current variable rate system linked to shortterm interest to a fixed rate system linked to long. The main findings are that the housing market has been a contributor to past volatility in the UK economy, and that moving to a fixed rate structure would reduce the impact of a change in interest on key macroeconomic indicators. It should be stressed, however, that does not mean that the UK economy would necessarily be more stable under a fixed rate mortgage system. That would depend on how monetary policy was then determined, and whether or not the UK was in EMU. Interest and the housing market January