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.

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.

III. The Implementation of the Golden Rule Over the Cycle1

A. Introduction

1. The golden rule is a cornerstone of the U.K.’s fiscal framework. Introduced in 1998, it states that over the economic cycle, the government should borrow only to invest and not to fund current spending. (A more technical definition of the golden rule is that over the cycle, the cumulative balance on the current budget as a percentage of GDP should be nonnegative; progress against meeting the rule is measured by the average balance on the current budget as a percentage of GDP since the start of the cycle.) As explained in Balls and O’Donnell (2002), the rationale for the above specification of the golden rule is to remove any bias against capital spending and to ensure fairness between generations.

2. Since its introduction, the U.K.’s golden rule has prompted numerous discussions in academic and policy circles, as well as the media. In the theoretical literature, for instance, Kell (2001) has evaluated the U.K.’s golden rule against a benchmark of an “ideal” fiscal rule, concluding that the rule measures strongly in many respects. On the policy front, the performance of the golden rule has been discussed extensively and regularly by the HM Treasury in its Pre-Budget (PBR) and Budget reports, as well as by the Institute of Fiscal Studies (IFS) in its annual “Green Budget.” In the media, over 2000 news reports have mentioned the golden rule during the past year.

uA03fig01

Average cumulative surplus on the current budget in 2005/06

(in percent of GDP)

Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A003

3. The recent wave of interest has been partly due to the fact that as the cycle has come close to an end, the safety margin for meeting the rule has shrunk and the risk of breaching it has risen. According to the HM Treasury’s estimates and projections, the current economic cycle started in 1999/2000 and will finish in 2005/06. Since the start of the cycle, however, the projected average balance on the current budget in 2005/06, which could be interpreted as the safety margin for meeting the golden rule, has declined from over 1 percent of GDP in the 2000 PBR to about 0.1 percent of GDP in the 2004 PBR (see chart). On one hand, the depletion of the safety margin with the unfolding of the economic cycle is not surprising—in fact, one could argue that the safety margin has served its purpose and should be zero at the end of the cycle. On the other hand, however, the low level of the safety margin has increased the risk of breaching the golden rule if budgetary projections for the remaining part of the cycle do not materialize; either due to an adverse output shock or to any other source giving rise to forecast errors.

4. Stepping back from the current conjuncture, this paper poses the following questions.

  • Question 1. What is the relationship between the size of the safety margin and the risk of breaching the golden rule? The objective of the paper is to quantify the (ex-ante) safety margin m needed to ensure that the probability of violating the rule is less than p percent over the cycle.

  • Question 2. How does the size of the safety margin depend on the characteristics of the likely shocks that hit the economy? The approach of the paper is to consider the impact of output, asset price, and discretionary policy shocks on the size of the safety margin (for a given probability of meeting the golden rule).

  • Question 3. What are the policy implications of subscribing to the above probabilistic approach to the golden rule? In addressing this question, the paper builds on the answers to Questions 1 and 2, as well as on existing proposals to update the U.K.’s fiscal framework (see Emmerson, Frayne, and Love (2004)).

5. The rest of the paper is organized as follows. Section B provides additional background on the introduction, formulation, and implementation of the golden rule. Section C describes the methodology used to tackle Questions 1 and 2, which is to simulate a reduced-form stochastic model of the current budget balance. Section D presents and discusses the simulation results, i.e., the size of the safety margin needed to meet the golden rule under various assumptions about the type and magnitude of shocks that affect the fiscal balance. Section E concludes.

B. Background and Motivation

6. Lessons from past policy experience made an important contribution to the design of the U.K. fiscal framework. As discussed in HM Treasury (1997), the sharp deterioration in the fiscal balance between the late 1980s and the early 1990s—which was partly attributed to large errors in estimating the level of output relative to its trend—taught policy-makers two key lessons. First, it is critical to take a prudent approach by building in a margin for uncertainty in the fiscal projections. Second, it is important to be open and transparent by establishing fiscal rules that remain stable over the economic cycles and allow objective ex post evaluation. The U.K.’s fiscal framework, and the golden rule, in particular, incorporate these two principles in order to help avoid past mistakes. Balls and O’Donnell (2002) elaborate that “of course, it is not possible to remove all source of uncertainty. But by taking a prudent approach, including using cautious assumptions and publishing cyclically adjusted estimates of the key fiscal indicators, the risk of mistakes can be minimized.”

7. What are the margins of caution in the current implementation of the golden rule? Several elements play a role in incorporating a margin for uncertainty in the fiscal projections, and, therefore, the golden rule. In principle, the combination of these margins of caution should accommodate likely adverse shocks within the fiscal rule.

  • Cautious assumptions. The forecasts for revenue and expenditures use 11 key assumptions—on privatization receipts, trend growth, claimant unemployment, interest rates, equity prices, VAT receipts, consistency of price indices, composition of GDP growth, funding, oil prices, and underlying market share of smuggled tobacco—that are audited on a three-year rolling basis by the National Audit Office (NAO) to be reasonable and cautious. The auditing of the trend growth assumption was prompted by errors in the estimates of the level of trend output in the late 1980s. The VAT assumption was included mainly because receipts in this tax category consistently underperformed relative to forecasts in the early 1990s.

  • Annually Managed Expenditure (AME) margin. The expenditure projections include a buffer should outturns for AME programs turn out to be worse than expected.

  • Safety margin. Fiscal policy is also set so that the current budget shows an ex-ante surplus, i.e., a safety margin for meeting the golden rule.

  • Alternative scenario. Finally, the PBR and Budget documents present an alternative projection of the cyclically-adjusted current balance and the average surplus on the current budget in which trend output is assumed to be 1 percent lower than in the main projections. Under this cautious scenario, the average surplus on the current budget was shown to be negative in the 2004 Pre-Budget Report.

8. In spite of these margins of caution, the ex-post realizations of the current balance reveal that the risk of breaching the golden rule at the end of the cycle can still be non-trivial. As already mentioned, the key indicator of progress against the golden rule, i.e., the average current budget balance since the start of the cycle, is projected by the HM Treasury to be only 0.1 percent of GDP in 2005/06 (see Table III.1). Keeping in mind that the historical one-year-ahead absolute error in forecasting fiscal balances has been, on average, about 1 percent of GDP, one could conclude that the risk of breaking the golden rule during the last year of the cycle is well above zero. It is difficult to assess to what extent the present situation came about because the safety margin was not large enough or the underlying assumptions were not cautious enough to accommodate unexpected shocks and/or discretionary changes in fiscal policy. Nonetheless, it is noteworthy that the steady depletion of the safety margin for meeting the golden rule has been due to successive downward revisions to current balances (see Table III.1). For example, the 2003 PBR showed a substantial worsening in the cyclically-adjusted balances in both 2003/04 and 2004/05, reflecting lower-than-expected receipts from income and corporation taxes and additional spending on social benefits and defense.

Table III.1

Current Balance Revisions Since the Start of the Cycle

article image
Source: HM Treasury

9. The recent experience suggests that output shocks may not always be the main source of uncertainty over the cycle. It also illustrates that the key generator of uncertainty in one cycle may be different from that in another. While trend growth surprises may play an important role in one cycle, uncertainty about asset price changes—leading to revenue surprises in certain tax categories, such as corporation tax and income tax receipts—may be critical in another. Even within revenue categories, VAT receipts may underperform in one cycle, but corporation tax receipts may turn out weaker than expected in another. In addition, discretionary policy shocks can also play a role in eroding the safety margin for meeting the golden rule. Therefore, the experience with the golden rule so far motivates the questions addressed in the remaining sections.

C. Methodology

10. The performance of the golden rule is simulated under various realizations of output and asset-price fluctuations, as well as discretionary policy. The simulation methodology builds on three steps.2 The first step is to specify a simple reduced-form model of the current budget balance. The second step is to calibrate the parameters of the model in order to simulate it under stochastic shocks that broadly match the historical U.K. data on output and asset price fluctuations, as well as hypothetical discretionary policy shocks that affect the fiscal balances. The last step is to compute the cumulative current balance that corresponds to each generated output cycle (given the realizations of the asset price cycles and the assumptions about discretionary policy) and to examine the resulting distribution of cumulative current account balance for all simulated cycles.

Model specification

11. The current budget balance is modeled as follows.

  • For each cycle i, total current revenues, rit, are defined to be the sum of the following five components: (i) starting revenues, rio; (ii) underlying (or structural) revenues, rits(μt), which vary with trend growth, μt; (iii) cyclical revenues, ritc(ot), which depend on the output gap,o t; (iv) asset-price related revenues rith(ηt) and rite(δt), determined by whether the housing and equity markets are in a boom or a bust (ηt, δt = 1 or -1)3; and (v) discretionary change in revenues, ritd, which is assumed to be exogenous.

(1)rit=rio+rits(μt)+ritc(ot)+rith(ηt)+rite(δt)+ritd
  • For each cycle i, total current expenditures, eit, are equal to be the sum of the following four elements: (i) starting expenditures, eio: (ii) underlying (or structural) expenditures, eits(μt), which vary with trend growth, μt; (iii) cyclical expenditures, eitc(ot), which depend on the output gap,ot; and (iv) discretionary change in expenditures, eitd, which is assumed to be exogenous.

(2)eit=eio+eits(μt)+eitc(ot)+eitd
  • The corresponding current balance, bit, is the difference between current revenues, rit, and current expenditures, eit. Using Equations (1) and (2), the total current balance can be decomposed into five parts, affected by the starting balance, trend growth, the output gap, asset (house and equity) prices, and discretionary policy (see Equation 3’).

(3)bit=rit-eit=(rit-eio)+(rits(μt)-eits(μt))++(rith(ηt)+rite(δt))+(ritd-eitd)(ritc(ot)-eitc(ot))+
(3)bit=bits(μt)+bitc(ot)+bith(ηt)+bite(δ)t+bitd
  • For each cycle i, Equation (3’) is normalized so that bitc(0)=0 and bits(μ¯)=0, where μ¯ is the long-term average trend growth of the economy. All variables are expressed as percentages of GDP. The (semi) elasticity of the current balance with respect to output, ε, is taken to be 0.7, which is consistent with recent estimates.4 The revenue effects from asset prices are assumed to be: i) zero in normal times, i.e., bith(0)=bite(0)=0, and ii) symmetric and constant in booms and busts, i.e., bith(1)=-bith(-1)=θ and bite(1)=-bite(-1)=ω. (Note that the parameters θ and ω need to be chosen.) Using these assumptions, Equation (3’) takes the simplified form below that is used in the simulations.

(3)bit=bio+ε(μit-μ¯)+εoit+bita(ηt,δt)+bitd
  • The baseline assumption about the starting value of the current balance is that bio zero for all i. By resetting the starting value, one could estimate the safety margin for a representative cycle. However, this assumption can be relaxed in order to examine the knock-off effects of discretionary policy shocks in one cycle on the following cycles by assuming that bio is equal to the sum of all discretionary shocks from previous cycles (bio=Σj=0i-1bjod).

Parameter calibration and cycle generation

12. The next step is to choose stochastic processes and calibrate their parameters to generate series for output and asset price cycles that match the characteristics of historical data reasonably well. The ultimate objective is to feed Equation (3”) with simulated output shocks (μt, ot) and asset price shocks (ηt and δt) generated by the calibrated stochastic processes.

Output cycles

13. The starting point is to establish the characteristics of the U.K. business cycle that need to be matched by the simulated data. Unlike the NBER in the U.S., there is no official arbiter of the timing of the U.K. cycle. However, previous studies have adopted various techniques to analyze the U.K. business cycle (Artis (2002), Krolzig and Toro (2001), Ravn (1997), Birchenhall, Osborn, and Sensier (2000)). Starting with a classical definition of the cycle, this paper uses a modified version of the Bry-Boschan (BB) algorithm (as in Harding and Pagan (1999a)) to find peaks and troughs in the U.K. real output series from 1955Q1 to 2004Q2. The resulting turning points match quite closely with those reported in Artis (2002). The average duration and amplitude of recessions and expansions are calculated to provide a reference point for the simulated data.

14. As illustrated in Harding and Pagan (1999b), a simple model of output growth captures the characteristics of the U.K. cycle quite well. In particular, quarterly output growth is assumed to be a random walk with drift, i.e., Δyt = μ + ξt, where μparameter (= 0.062) and ξt is the error term that is normal and independently distributed with a standard deviation σ(=0.009). The above stochastic process and parameter values are used to generate 50000 observations of quarterly real output. Note that by construction, the first and second moments of the actual and simulated series coincide. Thus the simulated data are analyzed for peaks and troughs and compared to the actual data on the basis of cycle duration and amplitude. The results show that the simulated cycles resemble the actual ones (see table).

Characteristics of Actual and Simulated Output Cycles

article image

15. The simulated data are then used to derive cycles based on crossing points between trend and actual output, in the spirit of the HM Treasury definition. The precise methodology used by the Treasury to calculate trend growth and the output gap cannot be applied, since no measures of economic slack are available to determine the on-trend points in the cycle. As an alternative, the simulated series are detrended using a Hodrick-Prescott filter. The data frequency is changed from quarterly to yearly. The output cycles used in the simulations are obtained by identifying the crossing points between trend and actual output.5 In each cycle, actual output crosses trend output three times. The average length of resulting cycles (n = 1750) is 7 years, with the shortest and longest cycle lasting 2 and 18 years, respectively (see chart).

uA03fig02

Length of Output Cycle

(in years)

Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A003

Asset price cycles

16. House price and equity price cycles are generated in a similar way to output cycles. In particular, quarterly data on real house and stock prices are used to estimate the parameters of the stochastic processes. (Admittedly, the specification of real house prices as a random walk with drift does not fit the data well, as a lot of serial correlation is left in the errors.) The characteristics of actual and simulated asset price cycles are shown above.

Characteristics of Actual and Simulated House and Equity Price Cycles

article image

17. Boom and bust episodes are identified using the simulated asset price series. In particular, the housing (stock) market is defined to be in a boom (bust) in a given year if the house (equity) prices are significantly above their trend values in that year, i.e., in the 90th (10th) percentile of the distribution of deviations from the trend line.6 During the boom and bust periods, revenues are assumed to differ from their normal times due to the direct effect of asset prices on certain tax categories, such as stamp duty, capital gains tax, corporation and income taxes, etc.7 The magnitude of the house price and equity price effects,θ and ω is assumed to be in the range of 0.8 to 1.6 percentage points of GDP in each boom year, which is broadly consistent with existing estimates in the literature (see Eschenbach and SchU.K.necht (2002, 2004)).

Discretionary policy

18. Discretionary shocks are assumed to be exogenous and stochastic. The exogeneity assumption is made for the sake of simplicity and tractability. (A possible extension of the model is to make discretionary policy endogenous, although this would involve the difficult task of calibrating a fiscal response function.) The distribution of discretionary policy shocks is taken to be uniform and symmetric at zero. In particular, bitd ˜U[-d, d], where d is the upper bound of the distribution. In the simulation, d takes on several values (d = 0.01, 0.02). Note that the symmetry assumption is consistent with the “integral” definition of the U.K.’s golden rule, which allows fiscal easing during one part of the cycle as long as it is compensated by fiscal tightening during another.

Cumulative balance computation

19. Using the simulated series for output and asset price shocks, the cumulative current balance is computed for each simulated economic cycle. For instance, the cumulative balance for a cycle i, which lasts Ti years, is calculated as the sum of the current balances throughout the cycle. These current balances are derived using Equation (3”), given the realizations of output and asset price shocks (μt, oi, ηt and δt) for this particular cycle.

(4)ci=Σt=1Tibit

Since each simulated cycle gives rise to a cumulative balance ci, the total number of realizations of the golden rule is as large as the number of simulated output cycles.

20. The distribution of cumulative balance outturns can be used to examine in what proportion of the cases the golden rule is broken and by how much. Given this information, one can compute how large the safety margin, s, should be so that only a given proportion p of the overall distribution corresponds to negative values of the cumulative balance. This (ex-ante) safety margin s would correspond to a (1-p) probability of meeting the golden rule. In addition, one could examine how the size of the margin varies with the nature and the magnitude of the underlying shocks stemming from output and asset-price fluctuations and discretionary policy.8

D. Results and Discussion

21. The impact of output uncertainty on the simulated distribution of cumulative budget balances is significant even in the absence of other shocks. In 25 percent of the cases, the cumulative current deficit exceeds 2.9 percent of GDP (see chart). To ensure that the golden rule is met with a 75 percent probability, the corresponding annual safety margin needs to be about 0.45 percent of GDP. If this buffer is maintained in all simulated cycles, only 25 percent of the overall distribution of cumulative balances will be negative (see chart).

uA03fig03

Distribution of Cumulative Balances: Before Safety Margin

(disc. shock=0, equity shock=0, house price shock=0)

Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A003

uA03fig04

Distribution of Cumulative Balances: After Safety Margin

(disc. shock=0, equity shock=0, house price shock=0)

Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A003

22. The effect of asset price uncertainty on the shape and dispersion of the cumulative distribution relative to the previous case is quite striking. If equity and house price shocks are added to the simulations, the distribution of cumulative balances becomes more dispersed and the associated margin of safety rises compared to the case of output shocks only (see chart). This happens because even though an asset price boom (or bust) is a low probability event, the size of the shock (when it occurs) is large. Now the annual safety margin associated with a 75 percent chance of meeting the golden rule is estimated to be about 0.6 percent of GDP (see chart). However, the marginal impact of asset prices on the safety margin depends heavily on how large the revenue effects of equity and house prices are assumed to be. The result presented above is based on the premise that revenues are higher by about 1.2 percent of GDP in each boom year and lower by the same amount in each bust year. If the size of the revenue effect is assumed to be 1.6 percent per year, the distribution of cumulative balances becomes even more dispersed and the safety margin increases to 0.7 percent of GDP. But if the revenue effect is taken to be 0.8 percent per year, the corresponding safety margin falls to 0.5 percent of GDP.

uA03fig05

Distribution of Cumulative Balances: Before Safety Margin

disc. shock=0, equity shock=0.012, house price shock=0.012)

Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A003

uA03fig06

Distribution of Cumulative Balances: After Safety Margin

disc. shock=0, equity shock=0.012, house price shock=0.012)

Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A003

23. Discretionary shocks also have an impact on the distribution of cumulative balances. Adding discretionary shocks to the preceding simulations could increase the safety margin further. For example, assuming that the starting current balance is zero for each cycle and the width of the discretionary shock distribution d is 0.02 (see para. 19), the distribution of cumulative balances acquires more weight in its tails compared to the previous case of output and asset price shocks. In particular, the safety margin corresponding to a 75 percent chance of meeting the golden rule rises further to 0.7 percent of GDP.9 However, the marginal effect of discretionary shocks on the safety margin declines as the narrowing width of the underlying distribution, d, decreases the size of the discretionary shocks.

uA03fig07

Distribution of Cumulative Balances: Before Safety Margin

disc. shock=0.02, equity shock=0.012, house price shock=0.012)

Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A003

uA03fig08

Distribution of Cumulative Balances: After Safety Margin

disc. shock=0.02, equity shock=0.012, house price shock=0.012)

Citation: IMF Staff Country Reports 2005, 081; 10.5089/9781451814286.002.A003

24. The relationship between the size of the safety margin and the probability of meeting the golden rule is summarized below for various sources of uncertainty (see Table III.2). The overall margin of safety rises sharply as the desired probability of adhering to the rule increases. For instance, it almost doubles if the probability goes up from 75 to 90 percent and doubles again if the probability rises further to 99 percent. It is also worth noting that for a given probability of meeting the golden rule and parameter specification, the overall safety margin is larger than the safety margin due to any particular source of uncertainty. Nonetheless, it is smaller than the sum of the safety margins from each source of uncertainty, as various types of shocks partly offset each other. The results also illustrate how the overall margin evolves depending on the assumed magnitude of the asset price effect (see Specifications 1-3). For example, the desired buffer rises from about 1¾ to 1¾ percent of GDP as the assumption about the revenue effect of an asset price boom changes from 0.8 to 1.6 percentage points of GDP.

Table III.2.

Sources of Uncertainty and the Annual Safety Margin

article image

25. The above analysis is subject to a number of limitations. First, the model specification is very simple and assumes that all types of shocks are uncorrelated. In practice, however, asset price shocks are likely to be positively correlated with output, raising the size of the safety margin. In contrast, if policy is allowed to respond to output and asset price shocks, the discretionary shocks are likely to be negatively correlated with output and asset price shocks, decreasing the desired magnitude of the fiscal buffer. Second, the true parameters of the model may be different from the calibrated parameters used in the simulations. For instance, the volatility of output may have declined permanently, reducing the required magnitude for a current surplus buffer. In spite of these limitations, however, the results are useful in highlighting the implications of implementing an asymmetric fiscal rule and illustrating the relationship between the safety margin and the uncertainty stemming from various sources.

26. Going back to the three questions posed at the beginning of the paper, one could draw on the empirical findings to answer them as follows.

  • What is the relationship between the size of the safety margin and the risk of breaching the golden rule? The findings suggest that there is a strong inverse, nonlinear relationship between the size of the safety margin and the risk of breaking the rule. The baseline results show that the (ex-ante) safety margin required to meet the golden rule with a 75 percent chance during a representative cycle is around 0.6 percent of GDP (see Table III.2). However, increasing the probability to 99 percent requires a safety margin of 2.3 percent of GDP. Making the plausible assumption that maintaining a very high safety margin involves a substantial cost to society, one could infer that policy-markers are likely to choose a safety margin that is associated with some positive probability of breaching the golden rule.

  • How does the size of the safety margin depend on the characteristics of the likely shocks that hit the economy? The simulation results indicate that output shocks are not the sole determinant of the safety margin. Providing against asset price shocks also requires a substantial safety margin. Its size depends on the underlying assumptions about the magnitude of the direct effect of asset prices on fiscal revenues. At the beginning of each cycle, uncertainty about discretionary policy shocks can raise the level of the safety margin as well.

  • What are the policy implications of subscribing to the above probabilistic approach to the golden rule? The estimated relationship between the size of the safety margin and the risk of breaching the rule suggests that providing for all uncertainty can be very costly. Therefore, it is important that policy makers be explicit about the size of the safety margin that is incorporated in the fiscal projections and the associated probability of breaking the golden rule. This could be achieved by using central (rather than cautious) assumptions and stating explicitly how large the overall safety margin is. In addition, the ex-ante and ex-post probability of breaching the golden rule could be communicated by using a fan chart of the current balance, as well as the associated cumulative balance. (These proposals are also put forward in Emmerson, Frayne, and Love (2004).)

E. Conclusions

27. The discussion in this paper leads to the following conclusions. First, given that the golden rule is asymmetric (current balance or better over the cycle), the government would need to target a small current surplus ex ante if it wants the probability of meeting the rule to be higher than 50 percent ex post. Second, the higher the desired probability of meeting the rule, the higher the safety margin. For example, the results suggests that the average current surplus needed to meet the golden rule with a 75 percent chance during a representative cycle is about ½ percent of GDP. Increasing this probability to 99 percent requires a significantly higher current surplus (2¼ percent of GDP). In other words, attempting to drive the risk of breaching the rule to close to zero could be very costly from macroeconomic and intergenerational perspectives. Thus, it is important to be explicit about both the desired probability of meeting the rule and the associated target for the average current surplus. Third, the analysis illustrates that the safety margin depends not only on output shocks but also on asset price and discretionary policy shocks. In particular, the overall safety margin is larger than the safety margin due to any particular source of uncertainty, although it is smaller than the sum of the safety margins due to specific sources, as different types of shocks partly offset each other.

APPENDIX III.1 Data Description and Sources

This appendix describes the data used in the simulations.

Output: Quarterly real GDP series from 1960Q1 to 2004Q2. Source: ONS.

House prices: Quarterly real house prices. Mix-adjusted house price index from 1968Q2 to 2004Q2, deflated by GDP deflator. Source: ODPM.

Equity prices: Quarterly real equity prices. Price index (FTSE All-Shares) from 1962Q2 to 2004Q2, deflated by GDP deflator. Sources: Bloomberg, ONS.

References

  • Artis, M. (2002), “Dating the Business Cycle in Britain,” CGBCR Discussion Paper Series, available at http://www.ses.man.ac.U.K./cgbcr/discussi.htm

    • Search Google Scholar
    • Export Citation
  • Balls, E., and G. O’Donnell (2002), Reforming Britain’s Economic and Financial Policy: Towards Greater Economic Stability, London.

  • Birchenhall, C., Osborn, D. and M. Sensier (2000), “Predicting U.K. Business Cycle Regimes,Centre for Growth and Business Cycle Research, available at http://www.ses.man.ac.U.K./cgbcr/

    • Search Google Scholar
    • Export Citation
  • Emmerson, C., C. Frayne, and S. Love (2004), “Updating the U.K.’s Code for Fiscal Stability,IFS Working Paper No. 04/29.

  • Eschenbach, F. and L. SchU.K.necht (2002), “Asset Prices and Fiscal Balances,” ECB Working Paper No. 141.

  • Eschenbach, F. and L. SchU.K.necht (2004), “Budgetary Risks from Real Estate and Stock Markets,Economic Policy, July 2004, pp. 313346.

    • Search Google Scholar
    • Export Citation
  • Harding, D. and A. Pagan (1999), “Knowing the Cycle,Melbourne Institute Working Paper No. 12/99.

  • Harding, D. and A. Pagan (1999), “Dissecting the Cycle,” Melbourne Institute Working Paper No. 13/99.

  • HM Treasury (1997), “Fiscal Policy: Lessons from the Last Economic Cycle,available at http://www.hm-treasury.gov.U.K./mediastore/otherfiles/lessons.pdf.

    • Search Google Scholar
    • Export Citation
  • HM Treasury (2003), “End of Year Fiscal Report,available at http://www.hmtreasury.gov.U.K./media/324/70/end_of_year_352[1].pdf

  • Kell, M. (2001), “Assessment of Fiscal Rules in the United Kingdom,IMF Working Paper No. 01/91.

  • Krolzig, H. and J. Toro (2001), “Classical and Modern Business Cycle Measurement: The European Case,” Institute of Economics and Statistics Oxford Discussion Paper, available at http://ideas.repec.org/p/cea/doctra/e2002_20.html

    • Search Google Scholar
    • Export Citation
  • Ravn , M. (1997), “Permanent and Transitory Shocks, and the U.K. Business Cycle,Journal of Applied Econometrics, Vol. 12, pp. 2748.

    • Search Google Scholar
    • Export Citation
1

Prepared by Petya Koeva.

2

All simulations were conducted using a collection of custom-built Java programs.

3

Non-linearity is assumed to capture the buoyancy of certain taxes (e.g., capital gains tax, corporation tax, income tax on bonuses, etc.) during booms. This specification ignores any indirect impact of asset prices on fiscal revenue (via output) and considers the direct effect only.

4

The choice of elasticity is based on the empirical analysis conducted by the HM Treasury (2003), which suggests that if GDP growth were one percentage point lower than assumed in a given year, the surplus on the capital budget would be lower by 0.5 percent of GDP in the first year and by a further 0.2 percent of GDP in the following year. While these coefficients are obtained by regressing spending and revenue ratios to GDP on estimates of contemporaneous and lagged output gaps, this paper assumes that the same elasticity is applicable to surprises in trend growth in the short run (over the cycle). In practice, discretionary policy is likely to respond to such surprises. However, this possibility is not explored below, and discretionary policy is taken to be exogenous.

5

This is different from the classical definition of cycles discussed above, which is based on turning points and uses the level of output only.

6

About 17 percent of the simulated output cycles have no such periods; 40 percent have a stock market boom (or bust); and 40 percent have a housing market boom (or bust).

7

For example, the equity market boom of the late 1990s was associated with a substantial rise in company profits and individual bonuses, which boosted corporation and income tax receipts.

8

The above setup assumes that the safety margin is set at the beginning of the cycle and is not adjusted as the cycle unfolds.

9

If the assumption of zero starting current balance for each cycle is relaxed to allow for discretionary shocks to carry over (see para. 13, bullet 5), the probability that the golden rule will be broken is high even if the above safety margin is implemented. This occurs because if the starting balance for a given cycle is negative (reflecting historical discretionary shocks), the entire distribution of the safety margin for that cycle is worsened.

United Kingdom: Selected Issues
Author: International Monetary Fund
  • View in gallery

    Average cumulative surplus on the current budget in 2005/06

    (in percent of GDP)

  • View in gallery

    Length of Output Cycle

    (in years)

  • View in gallery

    Distribution of Cumulative Balances: Before Safety Margin

    (disc. shock=0, equity shock=0, house price shock=0)

  • View in gallery

    Distribution of Cumulative Balances: After Safety Margin

    (disc. shock=0, equity shock=0, house price shock=0)

  • View in gallery

    Distribution of Cumulative Balances: Before Safety Margin

    disc. shock=0, equity shock=0.012, house price shock=0.012)

  • View in gallery

    Distribution of Cumulative Balances: After Safety Margin

    disc. shock=0, equity shock=0.012, house price shock=0.012)

  • View in gallery

    Distribution of Cumulative Balances: Before Safety Margin

    disc. shock=0.02, equity shock=0.012, house price shock=0.012)

  • View in gallery

    Distribution of Cumulative Balances: After Safety Margin

    disc. shock=0.02, equity shock=0.012, house price shock=0.012)