Benchmarks and targets under the SGP: evaluating safe deficit targets using NiGEM.
The Euro Area Stability and Growth Pact (SGP) has come under strain, with Germany, France and Portugal all breaching the conditions of the Pact. All have continuing deficits in excess of 3 per cent of GDP, and there appears to be little political will to deal with these breaches. The expansionary fiscal policies of the two largest countries in the Euro Area have called into doubt the sustainability of the Pact, and some redesign appears inevitable. It is difficult to conceive of European Monetary Union without a Pact of some form, and the potential causes of variation in budget deficits have to be taken into account when designing any new pact. In this note we discuss the variability of budget deficits and the potential safe limits if breaches are to be avoided.
We discuss minimum benchmarks for fiscal policy and budget targets within the SGP, and look at the causes for recent breaches in the fiscal guidelines set out in the SGP as well as reporting stochastic simulations on the Institute Global Model NiGEM. These extend Barrell and Dury (2001), Dury and Pina (2003), Barrell and Pina, (2003) and Barrell, Hurst and Pina (2003). If shocks are moderate then setting budget deficit targets of 0-1 per cent of GDP up to five years ahead would give a good probability of not exceeding borrowing of 3 per cent of GDP. However, if budget targets are set some distance from this range, as they have been, then breaches become more likely. The small open economies, Netherlands, Belgium, Portugal, Ireland and Finland, would have more variable deficits than the larger countries if they faced the same shocks. The UK, France, Germany and Italy all face similar levels of the variability of deficits with common shocks.
Budget deficits change either because the Government changes its stance, or the environment alters. There are large random elements to tax receipts and to spending, and deficits can worsen for reasons that are difficult to model. In addition, deficits will worsen as output moves away from its anticipated level. If demand shocks open up an output gap their impact will be partly offset by automatic fiscal stabilisers, but this involves an automatic worsening of the fiscal position. Estimates of the impact of demand shocks on the budget deficit can help us evaluate the cyclical component of a budget change. Supply shocks have been relatively common in Europe in the past ten years, and we find that offsetting the automatic stabilisers in the face of supply shocks stabilises output as well as the budget deficit.
Cyclical, discretionary and random components in the Budget
Tax revenues tend to fall along with the cycle, whilst expenditure rises, and a 1 per cent deviation in GDP from its trend might cause a 0.5 per cent of GDP change in the budget deficit. 1 However, it is difficult to assess what the cyclical position of the economy might be, and there are a large number of ways of extracting the information, as Massmann, Mitchell and Weale (2003) discuss. (2) We use the Approximate Band Pass Filter (ABP) on past data and on our forecast to estimate the output gap for the European economies, and hence to estimate the cyclical component in observed budget changes. (3) Chart 1 plots the non-cyclical element in budget changes in the European economies between 1999 and 2002, and compares it to the NIESR forecast of the deficit for 2003. (4) Much of the deterioration in the Euro Areas fiscal position that we have seen in the last three years cannot be attributed to the cycle, and is likely to be the consequence of discretionary action, especially in France and Germany.
We should not assume that any non-cyclical movement in the budget represents discretionary policy. Random variations in tax receipts or in spending may push the budget outturn well away from its predicted level without there being a cyclical explanation or a discretionary policy driving the change. Economic models cannot encompass all of the factors driving the economy, and models explaining tax receipts are bound to be incomplete descriptions. For instance a shift in the pattern of consumption away from taxed to untaxed goods may not be picked up in our equations for indirect tax receipts unless we have very detailed models.
NiGEM is an estimated world model, which uses a 'New-Keynesian' framework in that agents are presumed to be forward-looking but nominal rigidities slow the process of adjustment to external events. (5) Economies are linked through the effects of trade and competitiveness and are fully simultaneous. There are also links between countries in their financial markets as we model the structure and composition of wealth, emphasising the role and origin of foreign assets and liabilities. We have forward-looking wages, forward-looking consumption, forward-looking exchange rates and long-term interest rates are the forward convolution of short-term interest rates. (6)
Each country has a description of its domestic economy that can be broken up into sectors: the government, the labour market, consumption behaviour and financial markets. The supply side of the economy drives output in the long run through its impact on factor demands and on prices. It is important to close the model with monetary and fiscal rules. In our experiments we set interest rates using a combined policy of nominal aggregate targeting and inflation rate targeting, or two pillar strategy as advocated by the European Central Bank. Our fiscal policy rules have to be set in the context of our description of the public sector. We have models of direct and indirect taxes, and of government spending. We consider the financing of the government deficit (BUD), and we allow either money (M) or bond finance (DEBT). The debt stock affects interest payments and forms part of private sector wealth.
Current fiscal revenues can be disaggregated into personal taxes (TAX, which includes both personal income tax and social security contributions) which depend on personal incomes, corporate taxes (CTAX) which depend on longer-term profitability and other taxes (mainly indirect; MTAX) which depend on consumer expenditure. Transfers to individuals (TRAN) depend upon prices and on unemployment, and hence these vary with the economic cycle. Government consumption and investment (GC and GI), which are assumed to be on plan except for random fluctuations, are not influenced by the cycle (although plans may be adjusted in downturns). As GC and GI are in constant prices, we convert them to nominal terms using the private consumption deflator CED and the GDP deflator P, respectively. Government interest payments (GIP) are modelled as the payments on a perpetual inventory with the change in the debt stock each period attracting the long-term interest rate in the issue period until it is replaced. (7) The budget balance thus reads:
(1) BUD = TAX + MTAX + CTAX - TRAN - GIP - GC*CED - GI*P
We normally assume budget deficits are kept within bounds in the longer term, and taxes rise to do this. We can describe the simple fiscal rule as:
(2) [Tax.sub.t] = [Tax.sub.t-1] + [phi] [GBRT - GBR]
Where Tax is the direct tax rate, GBR is the government surplus target and actual surplus. The feedback parameter [phi] is designed to remove an excess deficit in less than five years. If fiscal solvency is 'off', it is turned back on again after our trial period.
Understanding shocks and uncertainty
We define a shock as the part of the level or change in an economic variable, such as investment, that we cannot explain using other economic variables. Take the equation for employment, (8) for instance, where the number of people employed (EE) depends on the level of output (Y), the rate of technical progress (TP) and the real wage (W/P). We may write this (where Log denotes the natural logarithm) as:
(3) Log(EE) = a + b*Log(Y) + c* Log (W/P) + d*TP + residual
The residual, or shock, is the part of the evolution of employment that cannot be explained by economic factors. As such it is exogenous to the economic process we are considering. In the stochastic simulations we analyse below we have to draw residuals from the past in order to apply them to economic relationships in the future. (9)
We can decompose shocks into demand shocks, supply shocks and fiscal shocks in table 1. Demand shocks are those that affect the level of spending in the economy, and we include consumption, investment, equity prices and exports and have no long-run effects on the level of output in the economy, but their effects on the price level will depend upon the monetary rule in place. Supply shocks are those affecting the labour market, such as employment, self-employment, hours and real wage (compensation equation) shocks, and may be temporary or permanent, when output is permanently affected. Tax shocks are those that directly affect the budget deficit independently of economic events elsewhere.
We may take the weighted average of these shocks by country to see who has the largest shocks and which countries faced more supply shocks, demand and fiscal shocks. Table 2 gives these indicators using France as a base. We weight together demand shocks using shares in GDP multiplied by the standard deviation of the individual shock over the period 1991Q1-1999Q4. Hence if exports are twice as important in GDP as investment, but the shocks have half the standard deviation, then export shocks take the same weight as investment shocks. We weight tax shocks by the importance of the item in the overall budget adjusted for the standard deviation of their shocks. Supply shocks are the standard deviation weighted average of shocks to employment (labour demand) and the wage equation (labour supply).
Demand shocks were similar in scale in the UK, France, Belgium and the Netherlands in the 1990s, whilst they were relatively high in Greece, Ireland Portugal and Finland. Supply shocks were at their highest in Finland, as we might expect given the scale of the impact of the collapse of the USSR on that country. Supply shocks were relatively large in Belgium, Germany, Portugal and Italy and relatively moderate in the UK, the Netherlands and France. Unexplained movements in tax receipts or in transfers were particularly large in the UK, followed by Greece and Finland.
Setting safe budget targets
The Euro Area has a Pact on fiscal policy amongst its members. (10) The Treaty of Amsterdam requires that the members of the Euro Area do not have deficits in excess of 3 per cent of GDP. They face penalties if their deficits are excessive for a sustained period unless they have been facing low output growth. The Broad Economic Policy Guidelines procedure that administers the Pact requires that deficits are set broadly in balance or surplus in the medium term to ensure that the chance of exceeding a 3 per cent deficit is low.
We can evaluate budget deficit targets using stochastic simulations on the Institute model NiGEM. (11) Stochastic simulations involve taking the model and repeatedly shocking all the equations in it. (12) We build up one trial with 20 model runs, and each experiment involves 150 trials. Each trial starts with a model run that has all the shocks from one quarter in the 1990s applied to it. The model is run with forward expectations and policy rules in place. A new future is created and in the second period of the trial a randomly selected quarter from the 1990s is applied. This is repeated until a 20 quarter trial is complete. The techniques are explored in Barrell, Dury and Hurst (2003) and Barrell, Byrne and Dury (2003), and the most recent version of our approach, which is used here, is discussed in Barrell, Hurst and Kirsanova (2003).
Different economies will be subject to different shocks at different times, and safe budget targets will depend on the shocks and on the structure of the economy. Some shocks may come from outside through trade and the exchange rate; others may come from changes in behaviour that we fail to capture in our analyses of the economy. There are many potential sources of uncertainty in any projection for government budgets. For instance, consumers may decide to spend less than anticipated, or they may switch expenditure to goods with poor tax receipts. Both would have an impact on projections for revenues. The shocks will vary significantly between countries, and this will give very different outcomes, as we will see below. However, we can also apply the shocks from one country to others to see how structures differ in response to similar shocks.
Applying French shocks
We have undertaken a set of stochastics where all countries in the EU are shocked with the residuals from the French equations as these had the least variable shocks overall in the period we are considering. We can therefore see whether the variability across countries of the budget deficit targets we discussed in Barrell and Dury (2000) and (2001) came from the structure of the economies or the structure of shocks they faced.
Table 3 reports deficit targets that would ensure that there was only a one in twenty chance that over a five-year period a 3 per cent of GDP deficit would not be exceeded. (13) Budget deficit targets in the range 0.2-1.4 would be sensible even though we have no fiscal feedback in place. (14) If it faced French shocks then Germany would need almost the same target as France at around 1.4, and the UK, Italy, Spain, Austria and Greece would need slightly tighter targets at around 1.0 per cent of GDP. The Irish, the Finns and Dutch would need targets around 0.6. The Belgians and the Portuguese would need to be closer to balance. (15) A similar distinction between targets that could be set for small and large countries is brought out in 'Public Finances in EMU 2002, Part II; where the Commission suggests that the larger countries could safely aim for cyclically adjusted deficits of 1-1.5 per cent of GDP, whilst others could aim for 0-1 per cent, except for Finland where a surplus might be required.
The absence of fiscal feedbacks gives a worst case scenario for deficits. Politicians might not react to emerging deficits in the first year that they emerge, but good housekeeping suggests that they do have to react eventually. This is in part because an emerging deficit is as likely to be associated with shocks to tax revenues as it is to changes in the economic cycle. If we put the fiscal feedback in place in order for undesired deficits to be removed within five years, we get significantly less variable budgets, as can be seen from table 4.
If we were to use the small shock 'solvency on' scenario as the basis for setting deficit targets, they would vary from 1.0 per cent of GDP for Belgium to 2.0 per cent for France, with most countries having targets around 1.5 per cent of GDP. Having fiscal solvency rules in place suggests a willingness to observe the Pact. The UK would need a target of 1.8 per cent to stay within the SGP floor 95 per cent of the time, if it followed the same good housekeeping rules. These figures, including those with the smallest and largest target in the group, are similar to those reported in Barrell and Dury (2000) using shocks from a shorter period.
In table 5 we form the ratio of the outturns in tables 3 and 4, and differences greater than 3 per cent can be seen as significant. (16) In general, budget deficits are 50 per cent more volatile if we do not have the solvency rule in place in the first five years of our trials when we face French shocks. For all countries, we find that income is more stable in response to these shocks if we have our good housekeeping rule in place. Inflation is also less volatile in all countries in the face of these shocks if we use a solvency rule. We can conclude that, depending upon the nature of the shocks faced, fiscal pacts will reduce the volatility of output and inflation. The stabilising properties of the automatic parts of the budget are limited in the face of some shocks, and in the face of supply shocks allowing automatic stabilisers to work may be counterproductive.
Looking at own country shocks
Shocks have differed between countries, and they have generally been larger than those experienced by the French. In setting budget targets it is necessary to take account of these differences and make judgements about the scale of shocks in the future. Table 6 repeats the first experiment above with the solvency feedback not operating but with own country shocks applied. The target deficits are much more variable, and in order to have only a one in twenty chance of the deficit exceeding 3 per cent of GDP, only France, Spain Netherlands and Ireland would be able to run deficits. All other countries would have to have a surplus. (17)
Table 7 gives the ratios of the RMSD in solvency off simulations with own shocks as compared to French shocks. When own shocks are applied the world becomes more volatile, with interest rates and budget deficits increasing in volatility. French output volatility is essentially unchanged, but the budget becomes slightly more volatile. German output is markedly more volatile, as is the budget deficit. The UK has an increase in output volatility if it has its own shocks, but not as large as is the case for Germany. The increase in volatility of tax revenues means that the volatility of the deficit in the UK increases markedly.
Budget targets in EMU cannot be set on the assumption that large shocks will not exist, but it is probably unwise to take the more extreme values we see in table 6. If targets were to be set on the basis of the worst increase above, Greece, then all countries would have to run surpluses. In general, deficit volatility increases by no more than 50 per cent, and hence we could set targets that were proportionately tighter than those in table 3 by this amount. However, even these may be seen as too tight, since for most countries there is little chance of exceeding the limit even without a feedback rule in place. Table 8 looks at our results with a fiscal feedback rule in place, and table 9 compares them to table 6. Basing a forecast and a set of policy rules on the assumption that there would be no reaction to emerging deficits is unwise, and having a fiscal feedback rule in place reflects the reactions of reasonable governments to deficits.
It is clear that if we trust governments to react to severe shocks, then we could set deficit targets around 1 per cent of GDP in all cases, except those of Greece and Belgium even if they faced their past shocks. It would perhaps be better to set a common target for all and expect those who faced the most severe shocks to react more than our simple rule requires. Hence we would argue that a 1 per cent deficit target is wise and sustainable, and high debt countries such as Belgium will have to do more work than others to stay within the bounds. In general the volatility of the budget deficit is much higher when there is no solvency rule in place, as we would expect. Output volatility is generally a little lower when there is no fiscal feedback, but this is not always the case, as we saw when we applied French shocks to other countries. In our experiments with own shocks, Ireland, Greece, Belgium, Spain, Italy, and Germany have lower output volatility with the feedback in place than in its absence. Only Finland had larger supply shocks than these countries in the 1990s, at least as indicated on table 2, whilst the remaining six countries had lower ones. Hence we might conclude that, although leaving the automatic stabilisers to work when we have demand shocks is wise, offsetting them when we have supply shocks is also wise. The impact of the solvency rule also reflects the fact that many shocks to taxes are also shocks to disposable income, and randomly increased revenue will frequently be associated with lower incomes and lower spending and activity.
Time profiles for deficit targets
The results in tables 3-9 give a summary statistic on budget deficit volatility over a five-year period. The time profile and the probability distribution of outcomes are of interest for policymakers, as the potential distribution of outcomes in the fist two or three years is of more importance when discussing immediate changes in fiscal policy. As we would expect, the safe limits get further from the target as we move into the future, especially if we have no feedback rule in place. In table 10 we set the deficit target to avoid a 3 per cent of GDP deficit in nineteen years out of twenty in each of the five years of our trials using the small shocks with no feedback rule in place (i.e. table 3). In table 11 we repeat the targets but with the rather disparate shocks each economy faced in the 1990s and with solvency feedbacks in place (i.e. table 8). We report results for the UK in EMU for comparability with other countries. The assumption makes little difference to the safe targets, but in general the UK could have a marginally looser target outside.
If small shocks with no solvency were to be the base case for setting deficit targets, then a range of 1.5-2.3 per cent of GDP for these targets would be possible for the next twelve months, and a range of 0.8-1.8 per cent of GDP would be feasible in the following year. Even in the third year of the policy programme most countries could aim to have a deficit in excess of 0.8 per cent of GDP, with only Portugal, the Netherlands and Ireland facing problems if they left tax rates unchanged. We get similar bounds for the first three years if we have solvency in place and tax rates adjust but apply own shocks. Only Greece needs to plan to run a surplus from the second year, and it could plan to have a stronger feedback on the deficit if it were unlucky and faced large deficits.
It is common to present policy in relation to fan charts for outcomes that give different probability distributions at each point in time. Charts 2-4 give the distribution of budget outcomes for the UK around the baseline path under the assumption of own shocks with solvency, both inside and outside EMU and own shocks without solvency outside EMU. It is clear from a comparison of charts 2 and 3 that membership of EMU has little impact on the variability of the deficit in the UK. Removing the solvency condition significantly affects the distribution of outcomes, as we might expect if there is no intention to maintain good housekeeping.
Fiscal pacts put bounds on borrowing, and the bounds have to be set carefully in the light of evidence. We have used our model, NiGEM, to evaluate safe targets for European deficits. We consider issues on the scale of shocks, their nature and the response of governments to emerging deficits. We conclude that cyclically adjusted target deficits of 1 per cent of GDP would ensure that governments seldom had to borrow more than 3 per cent of GDP, especially if they stood ready to raise taxes when the deficit deteriorated for reasons separate from cyclical developments. However both discretionary policy and automatic stabilisers can be counterproductive in the face of supply shocks.
If the automatic stabilisers are weak, then there is a case for using active fiscal policy to stabilise output. Discretionary policy in response to a downturn is clearly useful, but the potential for discretion raises the perceived volatility of output and interest rates, and inhibits investment. Systematic fiscal policy, as discussed in HM Treasury (2003) and Barrell and Weale (pp. 5-8 of this Review) could be used to stabilise output in the UK without increasing perceptions of potential volatility. Automatic responses could also easily be built into the redesigned SGP, and we expect that this will indeed happen.
Table 1. A categorisation of shocks Variable Type Description Consumption Demand Forward looking, based on optimisation, income and wealth Compensation Supply NAIRU bargaining equation, productivity, unemployment Corporate tax Tax Tax rate on base Self employment Supply Labour demand Employees in employment Supply Labour demand derived from production function Equity prices Demand Forward looking based on discounted profits Investment Demand Accelerator also depending on production function Indirect TAX Tax Tax rate on base Direct TAX Tax Tax rate on base Transfers Tax Expenditure depending on unemployment Exports of goods Demand External Demand curve Exports of services Demand External Demand curve Table 2. The relative importance of shocks Indexed shocks from 1991Q1 - 1999Q4 Demand shock Tax shocks Supply shocks Belgium 1.19 0.64 2.59 Finland 1.75 1.83 3.09 France 1.00 1.00 1.00 Germany 1.26 1.38 2.17 Greece 2.15 2.45 1.31 Ireland 2.29 1.35 1.86 Italy 1.28 1.20 2.37 Netherlands 1.19 1.06 1.46 Austria 1.64 0.64 1.56 Portugal 1.99 1.55 2.46 Spain 1.26 1.13 1.91 UK 0.93 2.45 1.39 Table 3. RMSD for small-scale shocks, solvency off FR shocks UK Germany France Italy Spain Neths solvency off GBR 1.18 1.00 0.95 1.18 1.16 1.38 Output 1.47 1.75 1.50 1.28 1.67 1.50 Inflation 0.62 0.37 0.45 0.35 0.28 0.52 target 95% 1.06 1.35 1.44 1.07 1.09 0.73 target 99% 0.25 0.67 0.79 0.26 0.30 -0.21 FR shocks Belgium Portugal Austria Greece Ireland Finland solvency off GBR 1.72 1.55 1.12 1.27 1.44 1.51 Output 1.88 1.38 2.61 1.76 3.13 1.41 Inflation 0.33 0.32 0.27 0.33 0.38 0.31 target 95% 0.18 0.45 1.16 0.92 0.64 0.52 target 99% -0.98 -0.61 0.40 0.05 -0.34 -0.51 Notes: No Solvency Feedback, UK in EMU, Two Pillar Strategy in Place. GBR is the ratio of the government deficit to GDP. Root Mean Squared Deviations are used in rows 4 and 5 which give one sided probability limits of a 5 per cent and 1 per cent chance of the deficit target being breached. Shocks applied to all countries at the same time. Table 4. RMSD for small-scale shocks, solvency on FR Shocks UK Germany France Italy Spain Neths solvency on GBR 0.76 0.67 0.58 0.77 0.85 0.95 Output 1.55 1.84 1.60 1.42 1.86 1.54 Inflation 0.64 0.39 0.46 0.37 0.30 0.53 target 95% 1.76 1.91 2.05 1.74 1.61 1.44 target 99% 1.23 1.45 1.65 1.21 1.02 0.79 FR Shocks Belgium Portugal Austria Greece Ireland Finland solvency on GBR 1.23 0.98 0.72 0.82 0.85 0.90 Output 1.94 1.50 2.67 1.85 3.21 1.45 Inflation 0.35 0.33 0.28 0.34 0.40 0.32 target 95% 0.98 1.39 1.82 1.66 1.60 1.52 target 99% 0.12 0.72 1.33 1.09 1.01 0.90 Notes: No Solvency Feedback, UK in EMU, Two Pillar Strategy in Place. GBR is the ratio of the government deficit to GDP. Root Mean Squared Deviations are used in rows 4 and 5 which give one sided probability limits of a 5 per cent and 1 per cent chance of the deficit target being breached. Shocks applied to all countries at the same time. Table 5. Ratio of outturns with small-scale shocks Ratio UK Germany France Italy Spain Neths FR shocks off: on GBR 1.56 1.51 1.65 1.54 1.37 1.45 Output 0.95 0.95 0.93 0.90 0.89 0.98 Inflation 0.98 0.95 0.97 0.95 0.95 0.98 Ratio Belgium Portugal Austria Greece Ireland Finland FR shocks off: on GBR 1.39 1.59 1.56 1.55 1.69 1.68 Output 0.97 0.92 0.98 0.95 0.98 0.97 Inflation 0.95 1.00 0.96 0.99 0.94 0.96 Note: Ratio of RMSD. Table 6. Setting deficit targets facing different shocks, solvency off Own shocks UK Germany France Italy Spain Neths solvency off GBR 2.73 2.18 1.00 1.84 1.46 1.47 Output 2.39 3.05 1.52 1.75 2.57 1.86 Inflation 0.62 0.62 0.45 0.94 0.66 0.83 target 95% -1.48 -0.58 1.35 -0.02 0.61 0.59 target 99% -3.37 -2.08 0.66 -1.29 -0.40 -0.42 Own shocks Belgium Portugal Austria Greece Ireland Finland solvency off GBR 2.92 2.16 1.83 3.21 1.46 1.86 Output 2.34 2.86 4.41 4.33 4.23 2.34 Inflation 0.43 1.28 0.32 2.67 0.69 0.50 target 95% -1.78 -0.55 0.00 -2.27 0.61 -0.05 target 99% -3.79 -2.04 -1.26 -4.49 -0.40 -1.34 Notes: No Solvency Feedback, UK in EMU, Two Pillar Strategy in Place. GBR is the ratio of the government deficit to GDP. Root Mean Squared Deviations are used in rows 4 and 5 which give one sided probability limits of a 5 per cent and 1 per cent chance of the deficit target being breached. Shocks applied to all countries at the same time. Table 7. French and own shocks compared (ratio of RMSDs) Ratio UK Germany France Italy Spain solvency off own: Fr GBR 2.31 2.17 1.05 1.56 1.25 Output 1.62 1.75 1.02 1.37 1.54 Inflation 1.00 1.68 0.99 2.73 2.33 Ratio Neths Belgium Portugal Austria solvency off own: Fr GBR 1.06 1.70 1.39 1.63 Output 1.24 1.24 2.07 1.69 Inflation 1.60 1.28 3.95 1.20 Ratio Greece Ireland Finland solvency off own: Fr GBR 2.53 1.01 1.23 Output 2.46 1.35 1.66 Inflation 8.01 1.82 1.63 Note: Ratio of RMSD. Table 8. Budget outcomes with own shocks and solvency in place Own shocks UK Germany France Italy Spain solvency on GBR 1.39 1.36 0.65 1.19 1.11 Output 2.34 3.12 1.65 1.79 3.05 Inflation 0.65 0.61 0.45 0.97 0.66 target 95% 0.71 0.77 1.94 1.05 1.18 target 99% -0.24 -0.15 1.50 0.24 0.43 Own shocks Neths Belgium Portugal Austria solvency on GBR 0.94 2.47 1.46 1.11 Output 1.85 2.37 2.74 4.35 Inflation 0.83 0.43 1.32 0.32 target 95% 1.46 -1.05 0.60 1.18 target 99% 0.82 -2.74 -0.39 0.43 Own shocks Greece Ireland Finland solvency on GBR 1.94 0.90 1.12 Output 4.55 4.41 2.33 Inflation 2.69 0.68 0.52 target 95% -0.19 1.53 1.17 target 99% -1.51 0.92 0.41 Notes: No Solvency Feedback UK in EMU, Two Pillar Strategy in Place. GBR is the ratio of the government deficit to GDP. Root Mean Squared Deviations are used in rows 4 and 5 which give one sided probability limits of a 5 per cent and 1 per cent chance of the deficit target being breached. Shocks applied to all countries at the same time. Table 9. Ratio of solvency on and off withown shocks (ratio of RMSDs) Ratio UK Germany France Italy Spain Neths own shocks off: on GBR 1.96 1.61 1.56 1.55 1.32 1.56 Output 1.02 0.98 0.93 0.98 0.84 1.01 Inflation 0.97 1.01 0.98 0.97 0.99 1.00 Ratio Belgium Portugal Austria Greece Ireland Finland own shocks off: on GBR 1.18 1.48 1.65 1.65 1.63 1.67 Output 0.99 1.04 1.01 0.95 0.96 1.00 Inflation 0.98 0.98 1.00 0.99 1.01 0.97 Note: Ratio of RMSD. Table 10. Deficit targets set to avoid the 3 per cent floor in nineteen years out of twenty when facing small shocks with no fiscal feedback First Year Second year Third year French shocks UK -2.01 -1.50 -0.95 Germany -2.30 -1.44 -1.43 France -2.24 -1.76 -1.48 Italy -2.04 -1.47 -1.13 Spain -2.13 -1.45 -0.96 Netherlands -1.53 -1.06 -0.80 Belgium -2.00 -1.14 -0.78 Portugal -1.67 -0.80 -0.51 Austria -2.19 -1.36 -1.17 Greece -1.92 -1.15 -0.88 Ireland -2.00 -1.12 -0.34 Finland -1.71 -1.06 -0.61 Fourth year Fifth year French shocks UK -0.83 -0.77 Germany -1.20 -1.30 France -1.30 -1.25 Italy -0.89 -0.86 Spain -1.10 -0.79 Netherlands -0.81 -0.75 Belgium -1.09 -0.76 Portugal -0.21 0.13 Austria -1.11 -1.00 Greece -0.49 -0.43 Ireland 0.31 0.81 Finland 0.09 0.01 Notes: No Solvency Feedback. UK in EMU, Two Pillar Strategy in Place. GBR is the ratio of the government deficit to GDP. Root Mean Squared Deviations are used in rows 4 and 5 which give one sided probability limits of a 5 per cent and 1 per cent chance of the deficit target being breached. Table 11. Deficit targets set to avoid the 3 per cent floor in nineteen years out of twenty when facing own shocks with a fiscal feedback First Year Second year Third year UK -1.67 -0.97 -0.79 Germany -1.72 -1.21 -1.29 France -2.32 -2.21 -2.19 Italy -2.08 -1.05 -0.90 Spain -1.96 -1.71 -1.59 Netherlands -2.08 -2.03 -1.99 Belgium -0.49 -0.42 -0.84 Portugal -1.53 -0.99 -0.89 Austria -1.76 -1.51 -1.24 Greece -0.53 0.32 0.79 Ireland -2.35 -1.80 -1.50 Finland -1.68 -1.02 -0.98 Fourth year Fifth year UK -0.58 -0.86 Germany -1.00 -1.07 France -2.16 -2.04 Italy -1.20 -1.06 Spain -1.35 -1.02 Netherlands -2.11 -2.06 Belgium -0.48 -0.59 Portugal -0.68 -0.70 Austria -1.31 -1.06 Greece 0.83 0.86 Ireland -1.55 -1.31 Finland -0.64 -0.82 Notes: Solvency Feedback in place, UK in EMU, Two Pillar Strategy in Place. GBR is the ratio of the government deficit to GDP. Root Mean Squared Deviations are used in rows 4 and 5 which give one sided probability limits of a 5 per cent and 1 per cent chance of the deficit target being breached.
(1) See HM Treasury 'Fiscal Stabilisation and EMU' p. 49 for a discussion.
(2) Even if we can measure the cycle, it is important to know what has driven a particular cycle before one can assess its implications for the budget position. Cycles driven by changing exports have very different budgetary implications from those driven by changes in consumption. A slowdown in activity driven by falling export demand is likely to have noticeably less impact on revenues than one driven by weak consumer spending, as the latter is more tax rich than the former.
(3) See the UK and World Chapters in this Review.
(4) We have removed the impact of 3G licence revenues on reported deficits.
(5) The theoretical structure and the relevant simulation properties of NiGEM are described in NIESR (2002) and Barrell, Dury, Hurst and Pain (2001).
(6) We use the Extended Path Method to obtain model consistent expectations.
(7) The perpetual inventory attempts to take account of countries like Italy and Belgium where there are large proportions of short-term public debt. Our simple model cannot take account of the complexities of debt finance, and there are residuals on these equations, and these are used in stochastics.
(8) Equations have been estimated with error correction related dynamics. We leave these out only for the sake of exposition.
(9) Barrell, Dury and Hurst (2003) discuss forward looking equations. Where we assume rational expectations, the next period's outcome is taken as the expected value in the current period.
(10) The Pact is discussed in detail in Brunila, Buti and Franco (2001) and recent developments are analysed in Barrell and Weale on pp. 5-8 of this Review.
(11) We have also compared the results with those with a different monetary rule and also with shocks from 1993Q1-1997Q4 rather than 1991Q1-199Q4 to ensure comparability with Barrell and Dury (2000). Neither change makes a substantial difference to our results. Hughes Hallett and McAdam (2003) undertake a related exercise on the public version of the IMF model, Multimod. This model is calibrated, and therefore there are no estimation-based residuals to draw, and hence their results are not directly comparable to ours, as they depend on their assumptions on the unobserved error process.
(12) This includes shocking the arbitrage related exchange rate equation. In some experiments this has a noticeable impact on the results. Including exchange rate shocks has little impact on the volatility of the budget deficit. In general, budget volatilities are 3-5 per cent higher with shocked exchange rates than without.
(13) The RMSD in the table can be seen as the standard deviation, and 1.64 standard deviations give the 95 per cent one sided confidence interval and leaves 5 per cent in the upper tail. Our target deficit can be calculated as (3-1.64*row 1, table x) The 1 per cent numbers are equivalent to those in Barrell and Dury (2000).
(14) We also include targets in table 3 for a 1 per cent failure rate target, which might be considered too tight.
(15) If we wished there to be only a 1 per cent chance that the targets would be exceeded, we could set targets of 0.3-0.8 per cent of GDP for the larger countries, but Ireland, Belgium, Finland, Portugal and the Netherlands would need to run surpluses.
(16) See Barrell, Hurst and Kirsanova (2003) for a discussion of test of significance.
(17) The UK results are discussed in more detail in Barrell and Hurst (2003).
Barrell, R., Byrne J. and Dury K. (2003), 'The implications of diversity in consumption behaviour for the choice of monetary policy rules in Europe', Economic Modelling.
Barrell, R. and Dury, K. (2000), 'An evaluation of monetary targeting regimes', National Institute Economic Review, 174, October.
--(2001). 'The Stability and Growth Pact, will it ever be breached? An analysis using stochastic simulations', in Brunila, A., Buti, M. and Franco, D. (eds), The Stability and Growth Pact.' The Architecture of Fiscal Policy in EMU.
Barrell, R., Dury, K. and Hurst, I. (2003), 'International monetary policy co-ordination: an evaluation using a large econometric model', Economic Modelling.
Barrell, R., Dury, K., Hurst, I. and Pain, N. (2001), 'Modelling the world economy: the NIESR model NIGEM', presented at an ENEPRI workshop, Paris, July.
Barrell, R., Hurst, I. and Kirsanova, T. (2003), 'Choosing the regime in an uncertain world, the UK and monetary union', National Institute Discussion Paper No. 209.
Barrell, R., Hurst, I. and Pina, A. (2003), 'Fiscal targets, automatic stabilisers and their effects on output', Fiscal Policy, Bank of Italy Conference Volume.
Barrell, R. and Pina, A. (2003), 'How important are automatic stabilisers in Europe?', Economic Modelling.
Brunila, A., Buti, M. and Franco, D. (eds) (2001), The Stability and Growth Pact: The Architecture of Fiscal Policy in EMU.
Brunila, A., Buti, M. and in't Veld, J.W. (2002), 'Fiscal policy in Europe: how effective are the automatic stabilisers?', European Commission Economic Paper No. 177.
Dury, K. and Pina, A.M. (2003), 'Fiscal policy in EMU: simulating the operation of the Stability Pact', Journal of Policy Modelling.
HM Treasury (2003), 'Fiscal stabilisation and EMU', Discussion paper, available on website.
Hughes Hallett, A.J. and McAdam, P. (2003), 'Deficit targeting strategies', Journal of Common Market Studies, 41, 3.
Massmann, M., Mitchell, J. and Weale, M. (2003), 'Business cycles and turning points: a survey of statistical techniques', National Institute Economic Review, 183, January.
NIESR (2002), NiGEM Model Manuals available at www.niesr.ac.uk.
Ian Hurst, Senior Research Fellow and Research Fellow, National Institute of Economic and Social Research. e-mail: email@example.com, firstname.lastname@example.org.
The authors would like to thank Stephen Hall, Rebecca Riley, Martin Weale, Olga Pomerantz and Jan in't Veld and participants in a Treasury Seminar for their comments. This paper is the result of work commissioned by the UK Treasury as background to the Treasury's continuing work on UK and EU fiscal policy frameworks.
|Printer friendly Cite/link Email Feedback|
|Author:||Barrell, Ray; Hurst, Ian|
|Publication:||National Institute Economic Review|
|Date:||Jul 1, 2003|
|Previous Article:||Changing the inflation target.|
|Next Article:||Annualised hours contracts: the way forward in labour market flexibility?|