Abstract
The Selected Issues paper examines the dynamics of the inflation process in the United Kingdom, particularly the influence of external shocks. The paper provides a brief summary of existing empirical work. It presents the expectations-augmented Phillips curve model used in the analysis. The basic conclusion is that low inflation in recent years can be explained reasonably well by a combination of increased competitive pressures. The paper also presents some estimates of the degree of possible overvaluation in housing prices and outlines the links between housing prices and the macroeconomy.
II. How Should Policymakers Respond to a Decline in House Prices?1
A. Introduction
1. For a sustained period, house prices in the United Kingdom have been appreciating sharply, and given the high correlation between house price and real activity, a possible retrenchment could have important macroeconomic implications. In this paper, MULTIMOD, one of the IMF’s multi-country macroeconomic models is used to examine the implications of a possible retrenchment in real house prices in the United Kingdom. The focus of the analysis is on how policy might best respond to a sharp fall in house prices, in particular the relative roles of monetary and fiscal policy. A key dimension of this policy-mix question is how the fiscal response might complicate meeting the fiscal rules.
2. Several factors account for the high correlation between house prices and the real economy. First, strong preferences for home ownership result in a high income elasticity of housing demand. Further, the predominance of variable-rate mortgages implies that housing demand is also highly interest rate sensitive. While these factors lead to large cyclical fluctuation in housing demand, the price elasticity of housing supply is quite low resulting in demand fluctuations being quickly transmitted into prices. Additionally, financial accelerator effects operating through the impact of house prices on household balance sheets contribute further to overall variability in consumption demand. The confluence of these factors means that there is a high degree of simultaneity in the cyclical correlation of house prices and the macro economy that policymakers need to be cognizant of as they strive to achieve their objectives.
3. The analysis focuses on four key questions. First, given the sustained period of rapid house price appreciation, what do various estimates of the sustainable level for house prices suggest about the possible degree of current overvaluation? Second, given the historically high correlation between real house prices and real activity, how large an impact on the real economy could a decline in house prices have? Third, how can policy be conducted to mitigate the macroeconomic disruption arising from a possible decline in house prices? Fourth, if house price declines do materialize, what policy mix would best safeguard the credibility of the monetary and fiscal policy frameworks?
4. Although there is uncertainty about the extent of possible overvaluation in the housing market, how it may be resolved, and the resulting broad macroeconomic implications, risk analysis is central to good macro policy management. The analysis presented in this paper should be interpreted in exactly that context, analysis of a possibly significant risk to continued strong and stable macroeconomic growth in the United Kingdom. As such, the analysis is tilted toward examining the worst possible outcomes to gauge the economy’s resilience and consider policy options that can mitigate the potential negative consequences.
5. The remainder of the paper is structured as follows. Section B presents some estimates of the degree of possible overvaluation in house prices and outlines the links between house prices and the macro economy. In Section C, a brief overview of MULTIMOD is presented focusing primarily on the transmission channels of the shocks considered and the characterizations of monetary and fiscal policy. The simulation results are presented in Section D. Section E concludes.
B. House Prices and The Macro Economy
Current house prices and sustainable levels
6. A range of indicators suggest that current house prices in the United Kingdom may be above the sustainable level. The ratio of house prices to rents, the ratio of house prices to average household disposable income, real house prices relative to linear time trends and an econometric real house price equation estimated in IMF (2003), all suggest that house prices are currently above what can be thought of as a sustainable or equilibrium value.
7. The current ratios of house prices to rent and to average household disposable income are at or close to their historical peaks. Within an asset-pricing context, the ratio of house prices to rents can be thought of as a price-earnings ratio, with rents reflecting prospective income.2 Looking at the evolution of this ratio graphed in the right panel of Figure II.1 indicates that the current level is close to the historical peaked reached in early 1989. Experience suggests that this ratio will likely move back toward its long-run average of 19. The evolution of the ratio of house prices to average household disposable income in the left panel of Figure II.2 presents a similar story. In spite of strong growth in disposable income, the recent rapid pace of house price appreciation has resulted in a current level of the ratio that is about 25 percent above its historical average.
The Ratios of House Prices to Rents and Average Household Disposable Income
Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A002
House Prices and Estimated Equilibrium Levels
Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A002
8. Simple linear time trends and an econometric model of real house prices suggest a range of overvaluation that depends on the time period used for estimation. Estimating a simple linear time trend up to 1999Q4 suggest overvaluation of roughly 65 percent in 2004Q2 (top panel in Figure II.2). However, extending the estimation period to 2004Q2 suggests more modest overvaluation of about 35 percent. This difference is not surprising as extending the sample period allows more of the recent run-up in house prices to be interpreted as trend. The econometric error-correction model, presented in IMF (2003) and updated in Appendix 1, models real house prices as a function of real disposable income and the real inter-bank interest rate. Interpreting the cointegrating relationship as providing an estimate of equilibrium house prices suggests that the extent of overvaluation ranges from 34 percent, if the equation is estimated up to 1999Q4, to 60 percent, if the equation is estimated up to 2004Q2 (bottom panel of Figure II.2). This difference arises largely because the coefficient on disposable income in the cointegrating vector, which is quite high in the United Kingdom relative to other industrial countries, declines from close to 2 to 1.5 when the equation is estimated over the longer sample period. Although no intuitive reason for the decline in the estimated coefficient is readily apparent, the change highlights the fact that all of these estimates should be interpreted cautiously, with the focus being on the broad picture coming from the range of techniques rather than any single estimate alone.
9. Plausible arguments can be made that the sustainable level for house prices may be higher than these indicators suggests. With the establishment of credible monetary and fiscal policy frameworks, both nominal and real interest rates in the United Kingdom are likely permanently lower than many would have expected in the 1980s or early 1990s. Permanently lower nominal and real interest rates could have raised the sustainable ratios of house prices to rents and disposable income more than the historical average would suggest.3 The impact of such structural change may also not be fully captured in the linear time-series estimates of equilibrium or trend real house prices either. In addition, demographic factors have also been cited as adding to housing demand thereby increasing sustainable price levels. Consequently, one should be careful about drawing precise conclusions about the extent of overvaluation suggested by these indicators.
10. Although reasonable people could conclude that the range of possible overvaluation in the housing market is quite wide, cautious policymakers should be thinking about the implications of a possible retrenchment, including a large one. Considering the various estimates of overvaluation along with the arguments as to why these estimation techniques may not be fully capturing structural effects from significant improvements in monetary and fiscal frameworks, the overvaluation in the housing market could lie within the 0 to 60 percent range. With so many indicators pointing toward some overvaluation along with the historical tendency of some markets to over shoot, the possibility of some correction in real house prices cannot be completely discounted. Policymakers need to remain cognizant of the risk of a correction and its macroeconomic implications.
House prices and real activity
11. There is a high degree of correlation between real house prices and real economic activity in the United Kingdom. As indicated by the deviation of house prices from a linear time trend and the output gap graphed in Figure II.3, house price cycles have tended to follow the business cycle quite closely in the United Kingdom. Periods of excess demand tend to be associated with house prices rising above trend. The relationship between house prices and consumption, as studied in detail in HM Treasury (2003b), underlies the high correlation.4 Factors such as strong preferences for the consumption of housing services, inelastic supply, variable-rate mortgages and a strong financial accelerator channel linked to house prices are cited as the key reasons for the high correlation between real activity and house prices. What is somewhat surprising with the current cycle is that house prices have not yet started to move back toward trend even though real activity is now approaching trend from below.
The Output Gap and the Deviation of House Prices from Trend
Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A002
1/ Source: Staff estimates.12. In an international context, the income elasticity of house prices in the United Kingdom tends to be quite high. The estimated U.K. income elasticity of house prices is well above unity as indicated by the house price equation estimates presented in Appendix 1 and those in mentioned in Chapter IV of this paper. The pooled income elasticity of house prices for a range of industrial countries presented in Otrok and Terrones (2005) is considerably lower than this at roughly ½. Strong preferences for housing coupled with relatively low price elasticity of supply (see Barker (2004)) are the factors that result in this high income elasticity of house prices.
13. Housing and consumption demand in the United Kingdom are sensitive to movements in short-tem interest rate because of the predominance of variable-rate mortgages. The structure of mortgage finance influences housing demand because of its direct effect on affordability. When short-term interest rates are low because monetary policy is attempting to support economic recovery, housing affordability is high and aggregate demand and housing demand rise together. Further, because of the large stock of mortgage debt financed largely with variable-rate mortgages, fluctuations in short-term interest rates have an enormous impact on homeowners’ debt-service costs and, consequently, the amount of disposable income available to finance consumption expenditures. When interest rates are low, debt-service costs fall and a larger portion of household disposable income is available to finance consumption expenditure.5
14. Rising house prices themselves also stimulate consumption because innovations in the U.K. financial markets over the 1980s have made it easier for households to access the equity in their homes to finance current consumption. This financial accelerator channel and its implications for monetary policy in the United Kingdom are modeled in Aoki, Proudman and Vlieghe (2002). When house prices rise, those not moving can use the increase in equity to secure loans that carry much lower costs than conventional unsecured consumer loans. Those moving house can also use built up equity to increase funds available for current consumption financed at low interest rates. Effectively this reduces the relative price of current consumption, encouraging households to consume. Further, if a large portion of home owners face binding liquidity constraints, the impact of access to credit through mortgage equity withdrawal could be quite large.
15. In IMF (2002), a reduced-form error-correction model of U.K. consumption is presented that captures the impact of these channels. This model contains a long-run cointegrating relationship between consumption, income, housing wealth, financial wealth and real interest rates. The short-run dynamics are influenced by unemployment, housing wealth, financial wealth and loans to households secured on dwellings. The model is as follows:
where c is log consumption, un is the unemployment rate, sl is log secured liabilities, hw is log housing wealth, fw is log financial wealth, r is the real short-term interest rate, and Δ is the first difference operator.6 The estimated coefficients illustrate that the channels through which housing has been cited as influencing consumption are important. In the long run, households treat housing as wealth that can use to support current consumption. In fact the estimates suggest that housing wealth is even more important in this regard than is financial wealth (coefficient of 0.1 on housing wealth versus 0.04 on financial wealth). In terms of the short-run dynamics of consumption, changes in housing wealth are important, as are loans secured against housing wealth. It is worth noting that the consumption model estimated above is very similar in structure and response coefficients to other empirical models of U.K. consumption.7 This model is embedded within the MULTIMOD simulation analysis conducted below and is used to increase the coherence between the model’s response and the response found in more detailed analysis of U.K. consumption behavior.
16. Evidence presented in Bank of England (2004) suggests that the correlation between the growth in house prices and the growth in consumption expenditure may have weakened in the last few years. Although this may indicate that the negative impact on consumption from declining house prices may be much weaker than longer historical averages suggest, in risk analysis, like that presented in this paper, it may be prudent to consider the possible worst case scenarios. Further, the growth rate relationship becoming weaker recently does not rule out the possibility that the high correlation throughout a large portion of current phase of house-price appreciation has lead to a levels imbalance that will unwind if house price fall precipitously.
C. MULTIMOD: An Overview
17. This section present a very simple overview of MULTIMOD, one of the IMF’s multi-country models of the world economy. A slightly more detailed description of the model is contained in Appendix 2 and the interested reader is directed to Laxton and others (1998) for a more complete description of the model’s structure, estimation and properties.
18. The U.K. block of MULTIMOD models the behavior of five types of economic agents: households, firms, nonresidents, fiscal authorities, and monetary authorities.
Households consume goods, supply labor and accumulate financial assets in the form of government bonds, housing and claims on the capital used by firms. Households are split between those whose consumption in each period is equal to a fraction of their combined financial and human wealth, and those that can consume only their disposable income. This latter group of households face liquidity constraints that prevent them from borrowing against their human wealth.
Firms combine labor and capital under Cobb Douglas production technology with the objective of maximizing the net present value of their expected future stream of profits.
International trade is motivated by the assumption that nonresidents view a country’s composite good as being an imperfect substitute for their own home-produced composite good. This assumption leads to the modeling of trade volumes as functions of activity and relative prices. In addition to trading, nonresidents can also hold domestic financial assets or alternatively supply foreign financial assets to domestic residents depending on whether the country is a net debtor or net creditor.
Fiscal authorities purchase goods and services and provide transfers that they finance through taxation or debt issue. The fiscal authorities have targets for the ratios of expenditures, transfers and debt to GDP. Cyclical variation in economic activity leads to deviations from target ratios and the tax rate on labor income is gradually adjust to restore government debt to its target relative to GDP.
The monetary authorities stabilize inflation by adjusting the short-term nominal interest rate according to a Taylor-type monetary policy reaction function.
19. MULTIMOD contains a complete description of relative prices. Consumer prices can be functions of up to four key prices: the world price of oil, the world price of non-oil primary commodities, non-oil GDP deflators and exchange rates. MULTIMOD, like most macroeconomic policy models, relies on a reduced-form Phillips curve to characterize the behavior of core inflation. Core inflation is a function of lagged inflation, expected future inflation and goods market disequilibria. The behavior of the nominal exchange rate is governed by uncovered interest parity.
20. Agents in MULTIMOD are required to form expectations of the future evolution of many variables. For example, households must form expectations about future labor income and firms must form expectations about future profit streams. In MULTIMOD, it is assumed that expectations of all future variables are perfectly rational (model-consistent) except expectations of non-oil GDP inflation and possibly the exchange rate. Here the model relies on a mixture of backward-looking and model-consistent expectations to generate the empirically observed persistence.
21. Although it is difficult to precisely capture the variation in quarterly data with a model of annual frequency, the estimation technique employed to determine MULTIMOD’s parameter values has resulted in a model that broadly captures the properties in the U.K. data. Although comparing impulse responses from backward-looking VARs to impulse responses from forward-looking models with endogenous policy reaction functions is fraught with conceptual and methodological difficulties, the impulse response results presented in Table II.1 provide a useful check. The table contains the first year responses of consumption and investment in the U.K. block of MULTIMOD to a 100 basis point increase in the short-term nominal interest rate and the average VAR responses for the first four quarters reported in IMF (2004). These results indicate that in the first year when the experiments are most comparable, MULTIMOD closely replicates the VAR results. Beyond the first year, however, consumption and investment in MULTIMOD recover very quickly because the endogenous policy reaction function takes over and quickly reduces interest rates to re-anchor inflation at the target rate. In the VAR results, the negative implications of the increase in interest rates take roughly four years to dissipate.8
First-Year Impulse Responses to a 100 Basis Point Increase in Interest Rates
First-Year Impulse Responses to a 100 Basis Point Increase in Interest Rates
| Consumption | Investment | |
| IMF (2004) VAR | -0.28 | -0.85 |
| MULTIMOD | -0.30 | -1.01 |
First-Year Impulse Responses to a 100 Basis Point Increase in Interest Rates
| Consumption | Investment | |
| IMF (2004) VAR | -0.28 | -0.85 |
| MULTIMOD | -0.30 | -1.01 |
D. Multimod Simulation Analysis
22. In this section MULTIMOD is used to examine the implications of declines in the real value of the U.K. housing stock relative to the baseline. The assumption in the baseline is that real house prices exhibit a mild gradual decline with a portion of the required re-equilibration occurring through continued income growth. The shocks that are considered are viewed to be temporary with prices falling sharply in the first year and then gradually converging back to the baseline by the sixth year of the simulation. Initially real house price declines of 10, 20 and 30 percent are considered. Based around the most severe shock considered, in which house prices fall by 30 percent, several other scenarios are presented that examine the possible implications of other factors that might either precipitate the retrenchment or complicate the macroeconomic adjustment.
The Base-Case Model
23. The base-case version of MULTIMOD that is used for this analysis incorporates a relatively benign characterization of the inflation process. The Phillips curve is linear in excess demand, and the pass-through of exchange rate movements into consumption prices as well as the resistance of workers to declines in their real wages arising from import price shocks are half of the effect suggested by estimation over the 1980 to 2001 period.9 MULTIMOD nests a Phillips curve that can be either linear or nonlinear in excess demand.10 Although there is considerable evidence from the 1970s and 1980s suggesting that inflation was nonlinear in excess demand, the improved inflation control achieved since the early 1990s in many industrial countries suggests that this nonlinear relationship may no longer be the appropriate characterization of the inflation process. In addition, the improved credibility of central banks and their clear commitment to inflation control suggest there are fewer strategic arguments for embodying such nonlinear relationships into policy models.11 As noted in Hunt and Isard (2003), there is mounting empirical evidence that, since about 1990, the pass-through of exchange rate movements into consumer prices is considerably less than that which was evident during the 1980s. Because MULTIMOD is annual, estimating on short sample periods is not feasible and other evidence has been used to improve its data coherence. MULTIMOD’s Philips curve also embodies a channel through which workers can resist the reductions in their real consumption wage that arise from import price shocks.
Although historically the data suggest that such a channel has been important in the transmission of these types of shocks, more recent evidence again suggests that the advent of credible inflation targeting regimes has weakened this channel. Consequently, the estimated parameter that captures this effect has also been reduced by half, consistent with the change made to the exchange rate pass-through effect.
24. Monetary policy is characterized by a Taylor-type reaction function and it is assumed that automatic stabilizers are allowed to operate, but tax rates and spending remain unchanged while aggregate demand is below aggregate supply. The response of monetary policy to the shock will have a significant impact on its effect and the appropriate characterization of the policy reaction function is a potentially contentious issue. However, it can be cogently argued that a Taylor-type monetary policy rule is useful for illustrative purposes because of its general acceptance as a realistic characterization of how policymakers respond and as a reaction function that is relatively robust to the types of uncertainty monetary policymakers face given plausible characterizations of social preferences over inflation and output variability.12 The Taylor-type rule used as the benchmark includes a coefficient of 0.5 on the contemporaneous output gap and coefficient of 0.5 on the one-year ahead deviation of inflation from target. The fiscal policy response is countercyclical only through automatic stabilizers and for the first four years of the shock, the model’s endogenous fiscal reaction function is turned off. This is necessary to prevent the endogenous policy rule from responding to the increase in government debt above its target level by increasing tax rates while the economy is experiencing a period of excess supply. Under this shock, the baseline characterization of fiscal policy would otherwise be procyclical.
25. The impact of the decline in house prices on consumption is augmented using the reduced-form consumption error-correction model presented in IMF (2002). Although MULTIMOD incorporates the housing stock as part of the wealth out of which forward-looking households consume, the magnitude of housing wealth is dwarfed by the magnitude of human wealth (the net presented value of the future stream of labor income). Because policymakers respond to the shock by reducing interest rates used to discount labor income, increases in human wealth partially offset the impact of declining house values and the overall response of consumption is less that what reduced-form evidence would suggest. Consequently, to improve the model’s ability to capture the impact on consumption, the error-correction consumption model from IMF (2002) is used as a guide. This consumption equation is embedded into the MULTIMOD simulations so that the general equilibrium effects coming through income, interest rates and unemployment as well as the house price declines are captured. Judgment is added to the consumption of forward-looking households until the impact of the house price shock on total consumption broadly matches that coming from the error-correction model. It is also worth noting that positive judgment is added to investment because given the share of residential investment in total fixed investment in the United Kingdom, the initial model response of total investment to house price declines was too large. On balance, the net impact of the added judgment on GDP roughly cancels, but the impact on the individual components of GDP is now more consistent with how the shock would be expected to play out.
The Base-Case Simulations
26. Declines in real house prices have an immediate negative impact on real activity. The base-case response to shocks of 10, 20, and 30 percent are presented in Figure II.4. Initially, the house price declines increase the user cost of capital in housing through declines in the current and expected real price of housing and, consequently, residential investment falls. The declines in housing wealth and the judgment added to match the reduce-form error-correction consumption model reduces the consumption of forward-looking households. In turn, the consumption of liquidity-constrained households declines as their disposable income falls. Weaker real activity has knock-on effects for future profit streams and business fixed investment falls as well.
Impact of Potential Declines in Real House Prices Solid line 30 percent, Dashed Line 20 Percent, Dotted Line 10 Percent Percent or Percentage Point Deviation from Baseline
Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A002
Source: MULTIMOD.27. The monetary authority responds quickly to the shock and nominal interest rates decline sharply in the first year. In the case of the 30 percent decline in house prices, the nominal interest rate declines by about 240 basis points. When the house price declines are smaller, so are the responses of short-term interest rates. With real interest rates falling, the real exchange rate depreciates. This easing in monetary conditions helps to support real activity and output begins to recover back toward baseline in the second year. After five years, the maximum cumulative loss in GDP is roughly 6 percent. With investment falling in the shock, potential output also declines and the maximum cumulative output gap after five years is lower, at roughly 4 percent. In the first year, CPI inflation only declines slightly below baseline as positive inflationary effects from the depreciation in the exchange rate partially offset the disinflationary impact from excess supply. With tax rates held fixed, expenditure on automatic stabilizers increasing, and revenue falling because of weaker activity, the fiscal position deteriorates. Under the 30 percent decline in house prices, the ratio of government debt to GDP ratio rises by just over 3 percentage points by the fourth year.
Expectations of slower productivity growth
28. If recent house price appreciation has been fueled by inflated expectations of productivity growth, it is possible that a realignment of expectations could precipitate the retrenchment in house prices. Alternatively, changing expectations of productivity growth could complicate the macroeconomic adjustment to a sharp fall in house prices. To examine this, a shock to productivity is added to the house price decline of 30 percent. The productivity shock considered reduces future total factor productivity growth temporarily by 0.25 percentage points for five years, after which time productivity growth returns to baseline. The results for the key macro variables from this experiment are graphed in Figure II.5 along with the results from the house price shock alone.
A 30 Percent Decline in Real House Prices and Slower Productivity Growth (solid), Base-Case Simulation (dashed) Percent or Percentage Point Deviation From Baseline
Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A002
Source: MULTIMOD.29. Not surprisingly, with productivity declining at the same time as house prices fall, the impact on GDP is larger. Output declines by almost 2½ percent in the first year. While the cumulative output loss is larger because of the supply component, the cumulative output gap after five years is very similar to the base-case. With forward-looking households and firms responding quickly once they realize productivity prospects over the medium term are lower than expected, monetary policy eases by more to support aggregate demand. Consequently, the net impact on inflation and the output gap is similar to the base case. The fiscal position, however, deteriorates even more with the debt-to-GDP ratio rising by over 4 percentage points by the fourth year. In part this reflects the permanently lower level of GDP. In summary, provided the monetary authority recognizes the supply component and eases appropriately, a realignment of expected productivity growth of the magnitude considered does not greatly complicate the macro adjustment to a decline in house prices. However, the fiscal position deteriorates further.
Alternative fiscal responses
30. If a sharp decline in house prices were to materialize, there may be some temptation to bolster demand through temporary fiscal initiatives. This, however, would only worsen the deterioration in the cumulative deficit and debt positions. To illustrate the macroeconomic implications of alternative fiscal responses, the 30 percent decline in house prices is simulated under two alternative fiscal responses relative to the base case. Under the first alternative, to help support activity, fiscal policy is eased through a cut in labor income taxes by an amount that roughly doubles the increase in the debt-to-GDP ratio relative to the base case.13 For the second alternative, the model’s endogenous fiscal policy rule, which seeks to stabilize the debt-to-GDP ratio by adjusting taxes, is switched on. Under both of these alternatives, monetary policy is allowed to respond endogenously according to the specified Taylor-type interest rate reaction function. The results for the key macro variables under these alternative fiscal responses are graphed in Figure II.6 along with the results under the base-case fiscal assumptions.
Alternative Fiscal Responses Base Case (solid), Easier Fiscal Stance (dotted), Tighter Fiscal Stance (dashed) Percent or Percentage Point Deviation From Baseline
Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A002
Source: MULTIMOD.31. With monetary policy responding endogenously, alternative fiscal responses do not have a material affect on the macroeconomic outcome, but debt and deficit profiles could be significantly different. With the model’s endogenous fiscal rule operating (dashed line), a tighter fiscal stance than in the base case reduces the cumulative increase in the government’s deficit-to-GDP ratio by 2 percentage points and cuts in half the increase in the debt-to-GDP ratio. The net impact on GDP, however, is negligible as tighter fiscal policy is almost completely offset by easier monetary policy. With the fiscal tax rule operating on labor income taxes, consumption is weaker. However, lower interest rates and a weaker real exchange rate lead to higher investment and net exports. CPI inflation is only marginally lower in the medium term. Under the second alternative, in which tax rates are lowered such that the accumulation in debt roughly doubles (dotted line), real activity is only marginally stronger as higher interest rates largely offset the impact of lower taxes. The path for CPI inflation is also only slightly higher under the easier fiscal stance.
32. These simulation results would appear to argue strongly for allowing monetary policy to respond to the shock and keeping fiscal policy focused on satisfying the golden rule and the sustainable investment rule. In doing so fiscal policy would avoid the risk of undermining the credibility of the fiscal framework. For example, if equilibrium real interest rates were to increase temporarily, but persistently, as a result of the missed fiscal objectives, then the negative impact of the shock could be larger and much longer lived. There are potentially significant benefits to relying on monetary policy alone to offset the impact of a sharp decline in house prices.
A less benign inflation process
33. Although it would appear prudent to rely on a monetary policy response alone should a sharp retrenchment in house prices materialize, it is also important to consider the potential risks to the monetary policy framework. The base-case version of MULTIMOD used for the simulations embodies a somewhat benign structure for the inflation process. It contains mild pass-through of exchange rate movements into CPI inflation, low resistance on the part of workers to declines in their real consumption wage and a linear response of inflation to goods market disequilibria. An interesting question to ask is what might happen if the monetary authority based their initial interest rate response to the shock on these assumptions, but it turned out that the inflation process was in fact nonlinear, the pass-through was stronger for depreciations than for appreciations (asymmetric), and workers were more resistant to declines in their real consumption wages. This would appear to be stacking the deck as much as possible against the monetary authority.
34. To consider the implications of the monetary authority underestimating the difficulty of its inflation control task, the following simulation experiment is run. First, the monetary policy response for the first year and a half is set exogenously to the response achieved under the benign inflation structure when house prices fall by 30 percent and fiscal policy is tighter than in the base-case simulation. However, judgment is added to the inflation response in the first two years so that it is identical to that which would be achieved under full pass-through of exchange rate movements into consumer prices, more resistance to declines in the real consumption wage, the nonlinear Phillips curve and the aggressive monetary policy easing.14 The results of this experiment are graphed in Figure II.7 along with the results under the benign inflation assumption that generated the initial policy response.
Less Benign Inflation Process Than Assumed by the Initial Monetary Policy Response (solid), Benign Inflation Process (dotted) Percent or Percentage Point Deviation From Baseline
Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A002
Source: MULTIMOD.35. Even if the inflation process turns out to be much less benign than assumed in the initial easing in monetary policy, there appears to be little risk that the credibility of the inflation targeting framework would be threatened. In the first year, inflation now rises above baseline by just over 1 percentage point and accelerates further to just over 1.2 percentage points above baseline in the second year. Even holding the policy rate fixed for the first half of the second year, the increase in the last half would see the policy rate rising back close to baseline. The policy rate would then remain close to baseline thereafter.
It is useful to keep in mind that with higher inflation, real interest rates are still well below baseline helping to offset the negative effect of the shock on aggregate demand. These results suggest that if monetary policy alone responded aggressively to the shock and in doing so was underestimating the difficulty of its inflation control task, and remained ignorant of the mistake for a year and half, inflation in the second year would accelerate to roughly three percent, given the baseline. Such an acceleration in inflation would not require a letter of explanation from the Governor to the Chancellor. Consequently, this simulation suggests that there is very little risk to the monetary policy framework of an aggressive monetary response to a sharp decline in house price.
E. Conclusions
36. Analysis suggests that the range of potential overvaluation in house prices is wide. Some traditional techniques for estimating the sustainable level of house prices suggest that overvaluation could be in the range of 25 to 60 percent. At the same time, cogent arguments can be made that improvements in monetary and fiscal policy frameworks have increased sustainable prices beyond what these linear estimation techniques can capture, suggesting there is little, if any, overvaluation at all. From a policymaker’s perspective, there does appear to be some risk that house prices are overvalued and, consequently, the possibility of some correction cannot be ruled. Policymakers need to be cognizant of the risk of such a correction and the possible implications.
37. Simulation analysis of the impact of declining house prices illustrates that the macroeconomic consequences could be significant. Real house price declines of between 10 and 30 percent could lead to cumulative losses in GDP relative to baseline ranging between 2 and 6 percent after five years, provided policy responds actively to offset the impact. If the house price decline is accompanied by a realignment of expectations of future productivity growth, aside from the real supply effects of such a realignment, such a development would not appear to complicate significantly the macro adjustment process. This is conditional, of course, on the monetary authority correctly perceiving the shock and responding accordingly. Overall these results may be indicative of the largest impact that declining house prices could have because consumption in these simulations responds as it has historically to variations in house prices. Although there are some indicators suggesting that this relationship may be weakening, when conducting stress test like the one considered in this paper, it is prudent to focus on the worst possible outcomes, particularly when evidence to the contrary is so preliminary.
38. Even if house prices declined by a large amount and households responded strongly, decisive policy responses can significantly mitigate the impact, with GDP growth falling below baseline in only the first year of the shock. If real house prices were to fall by 30 percent, either a sharp reduction in interest rates coupled with automatic stabilizers or a combination of lower interest rates, lower labor income taxes, and automatic stabilizers result in GDP growth declining by only 2 percentage points in the first year. Supportive policy would subsequently result in above-baseline growth for the next four years as the level of economic activity is returned to baseline.
39. Although both monetary and fiscal policy can effectively mitigate the impact of declining house prices, it would appear sensible to allow monetary policy alone to respond. If fiscal policymakers reduced income tax rates to help support household income and the monetary policy response factored this in, the broad macroeconomic picture would remain largely unchanged, but the fiscal position would deteriorate even more than it does when only fiscal automatic stabilizers operate. If, on the other hand, fiscal policy focused more on safeguarding fiscal objectives than assumed in the base-case simulation, monetary policy could largely offset the broad macroeconomic impact. The resulting improvement in the fiscal position would reduce the risk of violating the golden rule or the sustainable investment rule, and thereby help maintain the credibility of the fiscal framework. With the effect of a sharp fall in house prices largely falling on household consumption, monetary policy has a powerful offset because of the high proportion of variable-rate mortgages, and, consequently, it is highly effective under this shock. At the same time, the risk to the monetary framework of relying exclusively on monetary policy to respond to the shock appears to be quite low.
APPENDIX II.1 An Error-Correction Model of House Prices
1. This section presented the updated estimates of the error-correction model of U.K. house prices reported in IMF (2003). The empirical methodology used in this analysis was to estimate a reduced-form equation of log real house prices (p) as a function of log real disposable income per household (y), and real inter-bank rates (r).1 The estimation employed an error-correction model where real house prices adjust to their long-run equilibrium while responding to short-run movements in house prices in previous quarters, interest rates and real income per household.
2. Data on house prices and number of households were obtained from the Office of the Deputy Prime Minister, real disposable income from the National Statistics Office and three month inter-bank rates from the Bank of England.2 We estimated the model over three sample periods, 1972Q4 to 1999Q4, 1972Q4 to 2001Q3 and 1972Q4 to 2004Q2.
3. All series are I(1), i.e., contain a unit root, and real house prices, real income and interest rates co-integrate (Tables II.A.1, II.A.2 and II.A.3). The model shows a long-run income elasticity of house prices larger than that from previous literature and high degree of persistence in real house prices as far as two quarters back (Table II.A.4). The model estimated over the 1972Q4 to 1999Q4 period is presented below.3 Estimation results over the other sample periods are available upon request.
Unit Root Tests 1972:4–1999:4
Unit Root Tests 1972:4–1999:4
| Variable | Test Speficication | Lag | ADF t-statistic |
|---|---|---|---|
| Real House Prices | Levels | 5 | -1.10 |
| 1st Differences | 4 | -3.25 ** | |
| Real Disposable Income per Household 1 | Levels | 1 | 3.08 |
| 1st Differences | 0 | -13.01 ** | |
| Real Disposable Income per Household 2 | Levels | 1 | 3.36 |
| 1st Differences | 0 | -6.70 ** | |
| Real 3m Interest Rate | Levels | 1 | -1.93 |
| 1st Differences | 0 | -7.39 ** | |
| Null Hypothesis: Series has a unit root Test critical values: | 1% level | -2.59 | |
| 5% level | -1.94 | ||
| 10% level | -1.62 |
Unit Root Tests 1972:4–1999:4
| Variable | Test Speficication | Lag | ADF t-statistic |
|---|---|---|---|
| Real House Prices | Levels | 5 | -1.10 |
| 1st Differences | 4 | -3.25 ** | |
| Real Disposable Income per Household 1 | Levels | 1 | 3.08 |
| 1st Differences | 0 | -13.01 ** | |
| Real Disposable Income per Household 2 | Levels | 1 | 3.36 |
| 1st Differences | 0 | -6.70 ** | |
| Real 3m Interest Rate | Levels | 1 | -1.93 |
| 1st Differences | 0 | -7.39 ** | |
| Null Hypothesis: Series has a unit root Test critical values: | 1% level | -2.59 | |
| 5% level | -1.94 | ||
| 10% level | -1.62 |
Lag Order Selection Criteria
indicates lag order selected by the criterion
Lag Order Selection Criteria
| The lag length of the VAR system used to perform cointegration analysis was selected using the criteria presented below. |
||||||
| Lag | LogL | LR | FPE | AIC | SC | HQ |
| 0 | -117.8 NA | 0.0 | 2.3 | 2.4 | 2.3 | |
| 1 | 359.6 | 918.3 | 0.0 | -6.6 | -6.3 | -6.5 |
| 2 | 404.6 | 84.0 | 0.0 | -7.3 | -6.8* | -7.1* |
| 3 | 415.5 | 19.7* 1.30e-07* | -7.3* | -6.6 | -7.0 | |
| 4 | 420.3 | 8.41 | 0.0 | -7.3 | -6.3 | -6.9 |
indicates lag order selected by the criterion
Lag Order Selection Criteria
| The lag length of the VAR system used to perform cointegration analysis was selected using the criteria presented below. |
||||||
| Lag | LogL | LR | FPE | AIC | SC | HQ |
| 0 | -117.8 NA | 0.0 | 2.3 | 2.4 | 2.3 | |
| 1 | 359.6 | 918.3 | 0.0 | -6.6 | -6.3 | -6.5 |
| 2 | 404.6 | 84.0 | 0.0 | -7.3 | -6.8* | -7.1* |
| 3 | 415.5 | 19.7* 1.30e-07* | -7.3* | -6.6 | -7.0 | |
| 4 | 420.3 | 8.41 | 0.0 | -7.3 | -6.3 | -6.9 |
indicates lag order selected by the criterion
Lag Order Selection Criteria
indicates lag order selected by the criterion
Lag Order Selection Criteria
| The lag length of the VAR system used to perform cointegration analysis was selected using the criteria presented below. |
||||||
| Lag | LogL | LR | FPE | AIC | SC | HQ |
| 0 | -128.9 | NA | 0.0 | 2.5 | 2.6 | 2.5 |
| 1 | 357.2 | 935.3 | 0.0 | -6.6 | -6.3 | -6.5 |
| 2 | 402.9 | 85.2 | 0.0 | -7.3 | -6.7* | -7.1* |
| 3 | 413.1 | 18.5* 1.36e-07* | -7.3* | -6.5 | -7.0 | |
| 4 | 417.7 | 8.1 | 0.0 | -7.2 | -6.2 | -6.8 |
indicates lag order selected by the criterion
Lag Order Selection Criteria
| The lag length of the VAR system used to perform cointegration analysis was selected using the criteria presented below. |
||||||
| Lag | LogL | LR | FPE | AIC | SC | HQ |
| 0 | -128.9 | NA | 0.0 | 2.5 | 2.6 | 2.5 |
| 1 | 357.2 | 935.3 | 0.0 | -6.6 | -6.3 | -6.5 |
| 2 | 402.9 | 85.2 | 0.0 | -7.3 | -6.7* | -7.1* |
| 3 | 413.1 | 18.5* 1.36e-07* | -7.3* | -6.5 | -7.0 | |
| 4 | 417.7 | 8.1 | 0.0 | -7.2 | -6.2 | -6.8 |
indicates lag order selected by the criterion
Cointegration Test
Cointegration Test
| Hypothesized Number of Cointegrating Vectors |
Eigenvalue | Trace Statistic |
5 Percent Critical Value |
1 Percent Critical Value |
|---|---|---|---|---|
| A. Model with Real Disposable Income 1 | ||||
| None * | 0.21 | 41.41 | 35.19 | 0.01 |
| At most 1 | 0.08 | 16.69 | 20.26 | 0.14 |
| At most 2 | 0.07 | 7.97 | 9.16 | 0.08 |
| B. Model with Real Disposable Income 2 | ||||
| None ** | 0.22 | 42.11 | 34.91 | 41.07 |
| At most 1 | 0.08 | 15.69 | 19.96 | 24.60 |
| At most 2 | 0.07 | 7.22 | 9.24 | 12.97 |
Cointegration Test
| Hypothesized Number of Cointegrating Vectors |
Eigenvalue | Trace Statistic |
5 Percent Critical Value |
1 Percent Critical Value |
|---|---|---|---|---|
| A. Model with Real Disposable Income 1 | ||||
| None * | 0.21 | 41.41 | 35.19 | 0.01 |
| At most 1 | 0.08 | 16.69 | 20.26 | 0.14 |
| At most 2 | 0.07 | 7.97 | 9.16 | 0.08 |
| B. Model with Real Disposable Income 2 | ||||
| None ** | 0.22 | 42.11 | 34.91 | 41.07 |
| At most 1 | 0.08 | 15.69 | 19.96 | 24.60 |
| At most 2 | 0.07 | 7.22 | 9.24 | 12.97 |
APPENDIX II.2 An Overview of MULTIMOD
1. This appendix present a very simple overview of MULTIMOD, the IMF’s multi-country model of the world economy. A slightly more detailed description of the Model is contained in the appendix and the interested reader is directed to Laxton and others (1998) for a more complete description of the model’s structure, estimation and properties.
2. MULTIMOD has a two-tired structure. The first tier is a static representation that describes the long-run equilibrium of the economy where countries can be characterized as either net debtors or net creditors. The steady-state model is derived in a manner that makes it exactly consistent with the behavioral structure that determines the dynamic adjustment towards this full stock-flow equilibrium. This steady-state representation can be used to conduct comparative static analysis of the impact of permanent shocks to the economy. It also provides the terminal conditions exploited by the solution algorithm for solving the complete model. MULTIMOD’s second tier is a dynamic representation that describes the transition path that the economy takes to the long-run equilibrium.
3. For the simulations presented in this paper we use a slightly modified version of MULTIMOD Mark IIIB. It contains individual blocks for four industrial countries: the United Kingdom, the United Sates, Japan, and Canada. There are also two aggregate industrial country blocks. The first consists of the Euro area and the second consists of all remaining industrial countries. Each industrial country/block has an identical structure, but the estimated parameter values may vary. Developing countries are aggregated into two blocks. The main developing country block is made up of net debtor countries. The remaining developing country block consists of net creditor countries that are primarily those that export large quantities of oil. Both the developing country blocks are very simple with the key distinguishing feature being that the net debtor countries face a borrowing constraint. Analysis can be done with either individual industrial country/blocks or the complete model of the world economy.
4. Each industrial country block models the behavior of five types of economic agents: households, firms, nonresidents, fiscal authorities and monetary authorities. Each industrial country produces a single composite tradable good. Nonresidents perceive this composite tradable good to be an imperfect substitute for their own home-produced composite tradable good. The main developing country model as well as the international trade accounts distinguish among three types of tradable goods: the composite good, oil, and non-oil primary commodities. The main developing country model also includes a non-tradable manufactured good.
Households
5. In MULTIMOD, households consume the traded goods, supply labor and accumulate financial assets in the form of government bonds and claims on the capital used by firms. In the industrial country blocks, household behavior is based on an extended version of the Blanchard (1985) finite-planning-horizon model. Because current generations are disconnected from future generations, the model embodies non-Ricardian features where changes in government savings can affect national savings, interest rates and asset accumulation.
6. The basic Blanchard framework for household behavior has been extended along several dimensions. First, households are split between those whose consumption in each period is equal to a fraction of their combined financial and human wealth and those that can consume only their disposable income each period. This latter group of households face liquidity constraints that prevent them from borrowing against their human wealth (the present value of their expected life-time labor income). Further, households’ labor income profiles are age dependant. These extensions allow changes in taxes to have more short-term impact on economic activity and mean that population dynamics have important implications for consumption and saving. Households’ supply of labor is assumed to be perfectly inelastic with respect to the real wage.
Firms
7. Firms in MULTIMOD combine labor and capital under Cobb Douglas production technology with the objective of maximizing the net present value of their expected future stream of profits. Firms are assumed to be perfectly competitive. Capital accumulation is based on the q theory of Tobin (1969) with the addition of costly adjustment. Adjustment costs are quadratic around the steady-state level of investment. Differences between the market price of capital and its replacement cost lead firms to change their desired level of capital. Costly adjustment means that firms adjust investment flows gradually to achieve their new desired level for the capital stock.
Nonresidents and international trade
8. Unlike the explicit optimization theory determining the behavior of households and firms, international trade is motivated by the assumption that nonresidents view a country’s composite good as being an imperfect substitute for their own home-produced composite good. This assumption leads to the modeling of trade volumes as functions of activity and relative prices. Activity variables are constructed from input/output tables allowing for different import propensities in consumption, investment, government expenditure and exports. Domestic activity is the scale variable driving imports and nonresident activity is the scale variable driving exports. In addition to trading, nonresidents can also hold domestic financial assets or alternatively supply foreign financial assets to domestic residents depending on whether the country is a net debtor or net creditor. It is assumed that the financial assets held or supplied by nonresidents are government bonds. Global consistency ensures that worldwide trade flows balance and global net foreign asset positions sum to zero.
Fiscal authorities
9. The fiscal authorities in MULTIMOD purchase goods and services and provide transfers that they finance through taxation or debt issue. The fiscal authorities have targets for the ratios of expenditures, transfers and debt to GDP. Cyclical variation in economic activity leads to deviations from target ratios. To restore government debt to its target relative to GDP, the fiscal authorities gradually adjust the tax rate on labor income. Because households supply labor inelastically, this labor income tax is effectively a lump sum tax. Transfer and expenditure target ratios are automatically restored as economic activity stabilizes. The Mark IIIB version also incorporates endogenous fiscal transfers that respond to the degree of slack in economy and thus act as automatic stabilizers.
Monetary authorities
10. In MULTIMOD, the role of the monetary authority is to provide the nominal anchor. Which is achieved by targeting the rate of inflation. The monetary authorities adjust the short-term nominal interest rate according to a Taylor-type monetary policy reaction function. The short-term nominal rate is adjusted relative to a neutral nominal rate in response to the gap between inflation and its target rate and the gap between current output and potential output. The response coefficient on the inflation gap is set at 0.5 and the response coefficient on the output gap has been set to 0.5. The model user can choose whether the inflation and output gap terms in a Taylor-type policy rule are lagged, contemporaneous or one-period-ahead forecasts. This version of the model permits solutions with different countries/blocks choosing different long-run target rates of inflation.
Prices
11. MULTIMOD contains a complete description of relative prices. Industrial country prices can be functions of up to four key prices: the world price of oil, the world price of non-oil primary commodities, non-oil GDP deflators and exchange rates. The world price of oil is exogenous and the world price of non-oil primary commodities adjusts instantaneously to clear the non-oil commodities market. The behavior of the non-oil GDP deflator (referred to as core inflation) is described by a reduce-form Phillips curve and uncovered interest parity determines exchange rate behavior. How these prices are combined to generate the full of set relative prices depends on the individual country’s/block’s trading relationship with the rest of the world.
12. MULTIMOD, like most macroeconomic policy models, relies on a reduced-form Phillips curve to characterize the behavior of core inflation in the industrial countries. Core inflation (the rate of inflation in the non-oil GDP deflator) is a function of lagged inflation, expected future inflation and a goods market disequilibria. The natural-rate hypothesis is imposed in the estimation of parameter values. The model nests both a linear and nonlinear inflation process which the model user can switch between. The nonlinearity is such that inflation is more responsive to excess demand in the labor market than it is to excess supply. Although, the specification does not include explicit wage rates, the dynamics of inflation and inflation expectations are characterized in a manner that implicitly recognizes important features of wage-setting behavior (in particular, contracting lags and wage-push elements). Further, a real-wage catch-up term has been included in the Phillips curve to capture the effect of households resisting the erosions in the purchasing power of their real wage that can arise from changes in import prices.
13. The behavior of the nominal exchange rate is governed by uncovered interest parity. The exchange rate will deviate from the expected future exchange rate in proportion to the gap between the domestic short-term interest and the foreign short-term interest. All exchange rates are expressed in terms of the United States currency.
Expectations
14. The agents in MULTIMOD are required to form expectations of the future evolution of many variables. For example, households must form expectations about future labor income and firms must form expectations about future profit streams. In MULTIMOD, it is assumed that expectations of all future variables are perfectly rational (model-consistent) except expectations of non-oil GDP inflation and possibly the exchange rate. Here the model relies on a mixture of backward-looking and model-consistent expectations to generate empirically observed persistence.
References
Aoki, K., J. Proudman, and G. Vlieghe , 2002, “House Prices, Consumption, and Monetary Policy: a Financial Accelerator Approach,” Bank of England Working Paper No.169.
Bank of England, 2000, Economic Models at the Bank of England: September 2000 Update, London: Bank of England.
Bank of England, 2004, “Inflation Report,” Bank of England (November).
Barker, K., 2004, “Delivering Stability: Securing Our Future Housing Needs,” HM Treasury.
Barker, K., 2005, “The Housing Market and the Wider Economy,” Speech at the Institute for Economic Affairs State of the Economy Conference.
Blanchard, Olivier J., 1985, “Debts, Deficits and Finite Horizons,” Journal of Political Economy, Vol. 93 (April), pp. 223–47.
Clarida, R., J. Gali and M. Gertler, 1997, “Monetary Policy Rules in Practice: Some International Evidence,” NBER Working Paper No. 6254.
HM Treasury, 2003a, “U.K. Membership of the Single Currency: An Assessment of the Five Economic Tests,” HM Treasury, June 2003.
HM Treasury, 2003b, “Housing, Consumption and EMU,” EMU Studies, HM Treasury, June 2003.
Hunt, B., and P. Isard, 2003, “Some Implications for Monetary Policy of Uncertain Exchange Rate Pass-Through,” Scottish Journal of Political Economy, Vol. 50, No. 5, pp 567–584.
IMF, 2002, “Why has U.K. household consumption been so strong,” in United Kingdom: Selected Issues, IMF Country Report No. 02/52.
IMF, 2003, “An analysis of house prices in the United Kingdom,” in United Kingdom: Selected Issues, IMF Country Report No. 03/47.
IMF, 2004, “The interest sensitivity of U.K. demand,” in United Kingdom: Selected Issues, IMF Country Report No. 04/55.
Laxton, D., P. Isard, H. Faruqee, E. Prasad, and B. Turtelboom, 1998, “MULTIMOD Mark III: The Core Dynamic and Steady-State Models,” IMF Occasional Paper No. 164.
Laxton, D., D. Rose, and R. Tetlow, 1994, “Monetary Policy, Uncertainty and the Presumption of Linearity,’ Technical Report No. 63, Ottawa: Bank of Canada.
Levin, A., V. Wieland and J. Williams, 1999, Robustness of Simple Monetary Policy Rules Under Model Uncertainty,” in Monetary Policy Rules ed. By John Taylor, Chicago: University of Chicago Press and NBER, pp 263–316.
Miles, D., 2004, “The U.K. Mortgage Market: Taking a Longer-Term View,” HM Treasury.
Nickell, Stephen, 2004, “Household Debt, House Prices and Consumption Growth,” Speech at Bloomberg in London.
Otrok, C., and M. Terrones, 2005, “House Prices, Financial Returns and Macroeconomic Fluctuations: International Evidence,” forthcoming IMF Working Paper.
Rudebusch, G. and L. Svensson, 1999, “Policy Rules for Inflation Targeting,” in Monetary Policy Rules ed. By John Taylor, Chicago: University of Chicago Press and NBER, pp 263–316.
Tobin, James, 1969, “A General Equilibrium Approach to Monetary Theory,” Journal of Money Credit and Banking, Vol. 1 (February), pp. 15–29.
Prepared by Ben Hunt with help from Marialuz Moreno Badia in updating the estimated U.K. house price equation initially presented in IMF (2003). The author also acknowledges the assistance of Jared Bebee in preparing the charts.
Using a framework similar to the dividend-discount model used for equity valuation implies the following:
where P is the house price, E is the rent and R is the discount rate.
Assuming that rents grow at a constant rate, g, rf is the risk-free rate, and δ is the housing risk premium then the housing price-earnings ratio can be derived from the above as:
See Nickell (2004) and Barker (2005).
HM Treasury (2003b) is one of the studies accompanying the assessment of the five economic tests analyzing the decision on EMU membership in HM Treasury (2003a).
Miles (2004) contains an in-depth study of the U.K. mortgage market and outlines the policy options for encouraging households to move more toward fixed-rate mortgages.
For estimation details please see IMF Country Report No. 02/52.
See HM Treasury (2003b) and Bank of England (2000).
Attempting to match the VAR experiment more closely by exogenizing interest rates for an extended period of time in MULTIMOD leads to instability. This result is not surprising and in fact should be expected in a rational expectations model where policy actions are required for stability.
For more details on this version of MULTIMOD see Hunt and Isard (2003).
The nonlinearity is such that one percentage point of excess demand generates more inflationary pressure than the same amount of excess supply generates in deflationary pressure.
See for example Laxton and Tetlow (1995).
See for example Clarida, Gali and Gertler (1997), Rudebusch and Svensson (1998) and Levin, Wieland and Williams (1999).
The fiscal easing was implemented via a cut in labor income taxes rather than increased expenditure because this shock is falling on household wealth and income, and reducing income taxes is the most direct way to offset this.
This experiment stacks the deck more against the monetary authority than simply setting interest rates exogenously and using the version of the model with the nonlinear Phillips curve and the full pass-through. The way the experiment is conducted here, the pass-through effect is asymmetric and the monetary authority does not benefit as much from deflationary pressures coming from the appreciation in the exchange rate that occurs in the third year and beyond.
Our model is a reduced form equation that can be derived from two structural equations (for supply and demand). Since the model does not contain a quantity indicator, as it would in an structural demand equation, it is not possible to recover the parameters of the housing demand schedule.
Real house prices are calculated by deflating the ODPM house price index by the Retail Price Index (RPI). Nominal interest rates using a 8-quarter moving average of RPI.
T-statistics in parenthesis. All variables except interest rates are in logs, Δ denotes first differences, p represents real house prices, r real short-term interest rates, and y represents real income per household.
