United Kingdom: Basic Data
|Demographic and other data:|
|Area||94,247 square miles (244,100 sq. km.)|
|Population (mid-2001)||59.5 million|
|Infant mortality (per 1,000 live births)||6.1|
|Doctors per 1,000 inhabitants||0.5|
|GDP per capita (2000)||SDR 18,360|
|Composition of GDP in 2000, at current prices||In billions of Pounds||Distribution in Percent|
|Total investment (including stockbuilding)||161.8||16.6|
|Total domestic demand||953.19||98.0|
|Exports of goods and services||254.3||26.2|
|Imports of goods and services||272.2||28.0|
|GDP at market prices (average estimate)||972.6||100|
|Selected economic data||1999||2000||2001|
|Output and unemployment:||(Annual percentage change)|
|Real GDP (at market prices, average estimate)||2.3||2.9||2.3 1/|
|Manufacturing production||0.8||0.3||0.6 1/|
|Average unemployment (in percent)||6.0||5.6||5.1 1/|
|Earnings and prices:|
|Average earnings in manufacturing||4.1||4.6||4.8 1/|
|Retail price index, excluding mortgage interest||2.2||2.1||2.4 1/|
|Money and interest rates:|
|M0 (end of period)||11.7||4.8||6.3 2/|
|M4 (end of period)||4.2||8.3||8.12/|
|3-month Interbank rate||5.4||6.1||4.0 3/|
|10-year government bond yield||5.6||4.9||4.9 3/|
|(In billions of pounds sterling)|
|Fiscal accounts (In percent of GDP):5/|
|General government balance||1.9||2.7||-1.3 4/|
|Public sector balance||1.8||4.3 6/||-1.1 4/|
|Public sector net debt||38.2||32.7||30.8 4/|
|Balance of payments:|
|Current account balance||-19.1||-17.0||-20.3 1/|
|(In percent of GDP)||-2.1||-1.8||-0.8 1/|
|Direct investment (net)||-76.0||-78.9||-77.5 6/|
|Portfolio investment (net)||104.5||89.1||96.8 6/|
|Gross reserves, official basis||22.2||28.3||26.4 2/|
I. The Macroeconomic Effects of U.K. Fiscal Policies: An Empirical Exploration1
A. Introduction and Overview
1. This paper studies from an empirical standpoint the impact of fiscal aggregates on the evolution of output and the real effective exchange rate (REER) in the United Kingdom during the period from the late 1970s to the present. It finds, first, that the size of the dynamic fiscal multipliers is small, and often statistically nonsignificant. Second, that the direction of the impact of taxes and government consumption, but not of social transfers, is, if anything, the reverse of that predicted by standard Keynesian models.2 For example, an increase in government consumption tends to depreciate the REER and slightly depress output. And third, that these unconventional effects of fiscal policy appear to result from the behavior of private domestic agents, who, on balance, tend to react to fiscal expansions by contracting consumption and investment (and vice versa). These results are consistent with a growing body of empirical studies, including those on expansionary fiscal contractions and on the Ricardian equivalence hypothesis. This paper, however, remains largely agnostic as to the factors underlying the non-Keynesian effects of budgetary policies, on which a consensus is yet to emerge.
2. The results, however, do not directly imply that fiscal policy has little or no effects on output and the exchange rate in all circumstances. The estimates of the fiscal multipliers reflect an extended period of time and it is possible, as argued below, that the effects of fiscal policy might be nonlinear, with different—and even opposite—effects depending on expectations of future policies and private sector confidence. Thus, for example, during 2001, private consumption did not slow down in response to the increase in government spending as the multipliers—estimated for the whole sample period—would predict. This could be, for instance, because the current budgeted impulse is not seen as threatening the state of the public finances or because interest rates have not risen—indeed have been cut significantly.
3. The study of the effects of fiscal policy on economic activity, including through the exchange rate, is a classic theme in macroeconomics and has been extensively discussed in the literature.3 Nevertheless, until recently, most of the work in this area had discussed the effects of fiscal policy from a theoretical angle and attempts to quantify empirically these effects had been typically based on simulations of large-scale macroeconomic models.4 As indicated in Blanchard and Perotti (1999), however, these simulations tend to postulate, more than explore empirically these effects. In particular, simulation methodologies are not well suited to capture the existence of confidence effects and nonlinearities in the transmission mechanism of fiscal policy, even if they were present, as the underlying macroeconometric models used for the simulations tend to assume them away.5
4. During the 1990s, research evaluating the existence of non-Keynesian effects of fiscal policy has gained significant impetus, generated in part by prominent episodes of fiscal consolidations that were associated with sustained expansions of output—such as, for example, in Denmark (1983-86), Ireland (1987-89), and Sweden (1983-89)—as well as from converse experiences—such as Sweden in the early 1990s and Japan in the 1990s.6 This research has also been propitiated by the increasing availability of national statistics relatively comparable across countries and the development of adequate econometrical instruments for dealing with dynamic effects—notably structural vector autoregression (VAR) techniques.
5. Without attempting here a review of this recent strand of the literature, it is worth mentioning some contributions that are relevant to this paper. Since Giavazzi and Pagano (1990), an increasing number of studies have identified instances of non-Keynesian effects of fiscal policy, particularly fiscal consolidations that had an expansionary effect.7 Although the literature is far from unanimous regarding the underlying causes of these unconventional effects, researchers have attempted to describe and typify sets of circumstances that appear associated to them.8
6. First, the literature stresses the role of forward-looking expectations of agents and confidence effects in the possible existence of nonlinearities in the transmission channels of fiscal policy. The hypothesis is that, while small policy shocks may have the conventional Keynesian effect, large and persistent policy changes that force an overall reassessment of agents’ expectations may give rise to expansionary fiscal contractions and contractionary fiscal expansions, as the case may be. Thus, Giavazzi, Jappelli, and Pagano (2000) find that, in OECD countries, non-Keynesian effects of fiscal policies tend to be associated with large and persistent fiscal impulses, although the level and dynamics of public debt do not always appear significant. In contrast, they find in a large sample of developing countries that non-Keynesian effects tend to be associated with rapidly changing public debt levels, while the size of the fiscal consolidation as such is less significant. They conjecture that the capacity to service public debt and the possibility of a default is less of an immediate concern in OECD countries, while it is a more relevant source of uncertainty in developing countries. In a similar vein, Baldacci, Cangiano, Mahfouz, and Schimmelpfenning (2001) finds in a wide sample of countries that favorable initial fiscal and external conditions tend to support fiscal policy having the conventional Keynesian effects. Alesina and Perotti (1997), Alesina and Ardagna (1998), and Perotti (1999), inter alia, also find evidence of nonlinear effects of fiscal policy. In virtually all cases, when the fiscal multipliers are estimated through methodologies that do not entail strong a priori assumptions on the transmission mechanism (i.e., through reduced-form specifications), the size of the multipliers turns out significantly smaller than suggested by simulations of structural macroeconometric models. Thus, with a methodology similar to this paper, Blanchard and Perotti (1999) find that the fiscal multipliers for the United States have the standard Keynesian signs but their size is smaller than suggested by earlier estimates.
7. The findings of this paper are consistent with this view. Although our sample period (1979Q1-2001Q3) was primarily determined by the availability of suitable data, the period is dominated by two large and sustained fiscal consolidations (1976-1988, and 1994-2000, see Figure 1) that started from high fiscal deficits and were accompanied by an equally sustained demand-driven recovery in growth.9 Estimates of the dynamic fiscal multipliers for the period 1963-78 (with a restricted set of variables that exclude the REER) have the conventional Keynesian signs, with the exception of the multiplier of government consumption, which remains close to zero and statistically nonsignificant.
Figure 1.UK: Output Growth and Fiscal Balance
8. A second conclusion in the literature is that the composition of budgetary policies matters. Drawing on the experience of twenty OECD countries, Alesina and Perotti (1995 and 1997) show that permanent improvements in the public finances are typically driven by spending cuts, whereas fiscal consolidations based on tax increases tend to be temporary. Political economy argues also in this direction: even if private demand were equally responsive ex ante to revenue and expenditure measures, the former would have a larger effect in shaping expectations of a successful fiscal adjustment—including because, being politically costlier, spending cuts would signal a government’s determination in consolidating the public finances.
9. Prompted by these arguments, the empirical analysis conducted here avoids the use of a single indicator of the fiscal policy stance (such as the fiscal deficit) and considers separately the effects of government consumption, social transfers, and tax revenue. We find that the dynamic multipliers associated to government consumption, social transfers, and taxes are indeed different in size and lag profile. The estimated multipliers on output of government consumption and taxes are very close to zero and statistically nonsignificant (but still with the reverse sign than predicted by Keynesian theory). By contrast, the estimated multiplier on output associated with spending on social transfers has the conventional sign: positive shocks to spending result in an expansion of output, mainly during the subsequent four quarters. The multipliers of all fiscal variables on the REER are statistically nonsignificant.10
10. The methodology used in this paper owes much to Blanchard and Perotti (1999). As in that paper, we use a structural vector autoregression (VAR) methodology. Once the VAR is estimated and the contemporaneous coefficients are identified, the impulse response functions of the (logarithms of) budgetary aggregates on (the logarithm of) output and the REER can be interpreted as the dynamic multipliers of the budgetary variables. To our knowledge, no such study has been carried out for the U.K. economy, although this or similar methodologies have been extensively used to analyze the effects of monetary policy in the United Kingdom and elsewhere and of fiscal policy in the United States and other countries. The next section describes the methodology in detail and the following section discusses the main findings and extensions of the basic model. Finally, we offer some conclusions.
11. The basic reduced-form specification of the VAR analyzed here is
where yt is the vector of the state variables, and xt are the exogenous variables, including a constant, a deterministic quadratic time trend, and dummy variables. The vector ut are the reduced-form non-orthogonal disturbances with covariance matrix ∑ and l is the maximum lag. The matrices β, Ht, and ∑ are 5x5 matrices of parameters to be estimated. The state variables of the basic model included in yt are the logarithms of government consumption, tax revenue, social transfers, GDP—all in real terms, seasonally adjusted—and the REER.11 Owing to data availability, the series for real tax revenue and social benefits are constructed on the basis of nominal data deflated by the GDP deflator.12 Social transfers are defined as the National Accounts category social benefits (including in kind) for the central government and local authorities.13 This includes, inter alia, social security benefits; state retirement pensions; sickness, incapacity, maternity, and other government-provided allowances; social assistance, including by means of refundable tax credits; and unemployment benefits. The REER is based on the CPI; although an estimate of the real exchange rate based on unit labor costs would have possibly been preferable, available estimates of the latter would have resulted in a significant loss of statistical degrees of freedom. The vector xt of exogenous variables includes dummy variables reflecting the different governments in office during the sample period.14 These dummy variables are introduced to take into account—albeit in an admittedly simplified form—different fiscal policy response functions associated with each government. The choice of the sample period was dictated by data availability and covers 1979Q1-2001Q3. The data have quarterly periodicity, and this relatively high data frequency plays an important role in the identification of the structural VAR, since it allows us to assume away contemporaneous influences between some variables. The maximum lag l was determined by the likelihood ratio test and set to four.
12. In order to compute the impulse response functions and variance decompositions, the structural form VAR needs to be identified from the estimated parameters from equation (1). To identify the VAR, we follow the ideas proposed by Blanchard and Perotti (1999) and use a priori information on the budgetary process and independent estimates of the elasticities of taxes and transfers with respect to output. The structural expression of the VAR has the form
where the residuals vt = Aut are restricted to be orthogonal. This restriction, however, is not enough to identify all the components of the 5x5 matrix A.
13. It is in choosing the additional restrictions that we use a priori information. Specifically, the matrix A is assumed to have the following structure:
where the x’s indicate unrestricted unknown coefficients. This postulated structure of the matrix A renders the structural VAR model exactly identified.
14. Bearing in mind that A represents the matrix of contemporaneous transmission of shocks across the state variables, the restrictions imposed by (3) can be motivated as follows. The first row indicates that government consumption is not influenced, within the quarter, by shocks to the other state variables. It is postulated that the quarterly periodicity of the data does not allow government consumption expenditure enough time to react, either by automatic mechanisms or by deliberate policy action, within the period. This appears plausible in view of informational and statistical lags (e.g., GDP data are only released after the end of the quarter) and internal lags in the budgetary and expenditure administration process. The second and third rows in (3) indicate restrictions on the effect within the quarter of shocks elsewhere in the system on tax collections and social transfers, respectively. The postulated structure implies that shocks to tax revenue do not affect transfers and that shocks to the REER do not affect taxes or transfers.15 Shocks to transfers, however, may affect tax revenue—for example, because some transfers are part of taxable income. The impact of shocks to GDP on taxes and transfers is given by the elasticities of these variables with respect to output: ∊TX1GDP and ∊TR1GDP. These values are set to 1.1 and -0.5 respectively, on the basis of existing estimates of the automatic stabilizers.16 Other values of the elasticities, within a reasonable range of existing estimates were also tried without significant changes in the results. Finally, the two last lines of the matrix A in (3) indicate that the impacts of all other variables on the REER are left unrestricted, whereas the contemporaneous impact of the REER on output is set to zero. The unrestricted coefficients in (3) were estimated by maximum likelihood.17
C. Results and Extensions of the Basic Model
15. The empirical results of the methodology described in the previous section indicate that the fiscal multipliers of budgetary aggregates on output and the REER are small and often statistically nonsignificant.18Figure 2 shows the impulse response functions with confidence bands corresponding to two standard deviations.19 Since the variables are logarithmically transformed, the values of the impulse response functions can be interpreted as elasticities at the corresponding lag. The only fiscal policy impact that is statistically significant is the expansionary effect on output of social transfers. The response of output to an increase in transfers is significantly positive during the first year with a peak elasticity of about 0.17. Given that social transfers have represented about ¼ of GDP in recent years, the estimated values of the impulse response function mean that an increase in transfers of 1 percent of (quarterly) GDP in one quarter is estimated to result in a GDP increase of between ½ and ⅔ of a percentage point of GDP over the following year. A possible explanation of this result is that, since social transfers are fairly well targeted in the United Kingdom, they accrue mainly to liquidity-constrained households with a high propensity to consume out of income, which offsets any potential Ricardian or confidence effects. Conversely, the progressivity of the tax system may result in tax changes accruing mainly to non-liquidity-constraint households, which are more likely to behave according to the Ricardian equivalence hypothesis. Thus, although a tax increase has a slightly contractionary effect during two quarters, the latter is subsequently offset. Government consumption has virtually no effect on output with any lag.
Figure 2.UK: Basic Model—Impulse Responses
16. The estimated impact of the three budgetary variables on the REER shows the opposite direction to that typically implied by Mundell-Fleming theory—although the values are only marginally significant and even that, only at some lags.20 Overall, increases in government spending appear associated to a depreciation of the REER, while the converse obtains for tax increases. This would be consistent with the hypothesis that confidence effects offset or even dominate conventional Keynesian effects.
17. In order to throw some light on the transmission channels of fiscal policy and the possible reasons for the results, private consumption and business investment were added, one at a time, to the basic model, exploring the effects on private demand of fiscal policies. The results are consistent with the hypothesis that contractionary fiscal policies may have an expansionary effect on private demand. The corresponding impulse-response functions are shown in Figures 3 and 4. The results from introducing private consumption indicate that the latter tends to contract (expand) following expansions (contractions) on government consumption.21 In contrast, private consumption expands (significantly at 5 percent probability) during the first eight quarters after an increase in social transfers. This behavior of consumption is consistent with and could explain the anomalous shape of the fiscal multiplier of government consumption as well as the conventional shape of the multiplier of social transfers, since the estimated impulse-response functions appear to indicate that output responds significantly to shocks in private consumption. Finally, an expansion in private consumption appreciates the REER as it could be expected. Private investment contracts following an expansion in government consumption and increases after a tax increase (slightly and for a short period, but significantly from a statistical standpoint). This apparently unconventional behavior of investment vis-à-vis taxation is consistent with expectational and confidence effects.22
Figure 3.UK: Adding Private Consumption—Impulse Responses Figure 4.UK: Adding Business Investment—Impulse Responses
18. Other extensions to the basic model were also considered. In particular, there is the possibility that low estimated values of the dynamic multipliers could result from fiscal policies conducted with a view to offset shocks in external demand. Thus, for example, if expansions in government consumption were systematically associated with negative external shocks and designed to offset the effect of these shocks on output, the multiplier of government consumption estimated with the omission of external demand would likely be close to zero. This would, in fact, reflect the effectiveness of fiscal policy and not the reverse. In order to address this possibility, the basic model was augmented by including an index of effective world demand for U.K. exports as exogenous variable (with several lag structures). Although world demand was statistically significant in the equation of government consumption, the estimates of the dynamic multipliers and results commented above did not change in any significant way. Finally, in order to explore existence of crowding out effects of fiscal policy through changes in the interest rate, the basic model was also augmented by introducing short-term and long-term interest rates.23 The results of this exercise did not provide any evidence of this channel of transmission of contractionary effects of expansionary fiscal policies (or vice versa).24
D. Conclusions and Additional Remarks
19. This paper has explored the empirical linkages between fiscal policy variables and output, the REER, and private demand in the United Kingdom during the last 23 years. We find that the impact of the fiscal policy variables is weak, with only social transfers exhibiting unequivocally standard Keynesian effects. Government consumption and taxes show, on balance, non-Keynesian effects on output, demand, and the REER. This could be due to the existence of nonlinearities in the fiscal policy transmission mechanism, stemming from confidence and expectational effects which could occasionally reverse the conventional direction of the impact of fiscal policy. If this were the case, the small value of the fiscal multipliers could be the result of offsetting opposite values during different fiscal policy episodes. Exploring these hypotheses, however, requires further evidence.
20. Existing evidence indicates that the non-Keynesian effects of government consumption and taxes are transmitted through their impact on private demand. Positive shocks to government consumption tend to depress private consumption (at least initially) and business investment. Social transfers have a stimulating effect on output and private consumption—perhaps owing to their being targeted to liquidity-constrained households—and only a very small initial negative effect on investment.
21. It must be pointed out that a small estimated size of the fiscal multipliers or the existence of nonlinearities is not necessarily inconsistent with standard Keynesian theory. The latter postulates that significant fiscal multipliers should be associated with the existence of economic slack, but not with situations when output is close to potential. For example, the full employment multipliers of government spending and tax cuts could be small as the crowding out effect of higher interest rates and spillovers through the trade balance may dominate. In practice, however, most observers would agree that actual output remained below potential in the United Kingdom for most of the period under scrutiny.25 Moreover, we did not find statistical evidence of increases in interest rates after expansionary fiscal shocks, or vice versa—rather, the opposite appears to have occurred. Thus, it is unlikely that the main statistical results stem exclusively from standard crowding out effects, although these may have been occasionally a contributing factor.
Alesina, Alberto, and SilviaArdagna,1998, “Tales of Fiscal Adjustment,” Economic Policy, Vol. 27 (October), pp. 487–545.
Alesina, Alberto and RobertoPerotti,1995, “Fiscal Expansions and Adjustments in OECD Countries,” Economic Policy, Vol. 21 (October), pp. 207–248.
Alesina, Alberto and RobertoPerotti,1997, “Fiscal Adjustments in OECD Countries: Composition and Macroeconomic Effects,” IMF Staff Papers, Vol. 44, No. 2 (June), pp. 210–248.
Baldacci, Emanele, MarcoCangiano, SelmaMahfouz, and AxelSchimmelpfenning,2001, “The Effectiveness of Fiscal Policy in Stimulating Economic Activity: An Empirical Investigation,” Mimeo, Paper presented to the Second Annual Research Conference organized by the IMF, Washington D.C., November24–30, 2001.
Bertola, Giuseppe, and AllanDrazen,1993, “Trigger Points and Budget Cuts: Explaining the Effects of Fiscal Austerity,” The American Economic Review, Vol. 83, No. 1. (March), pp. 11–26.
Blanchard, Olivier, and RobertoPerotti,1999, “An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output,” NBER Working Paper No. 7269.
Blanchard, Olivier,2000, “Commentary,” Economic Policy Review, Federal Reserve Bank of New York, Vol. 6, No. 1 (April), pp 69–73.
European Commission, 2001, “Public Finances in EMU,” Directorate-General for Economic and Financial Affairs, European Economy, No. 3.
Giavazzi, Francesco, Japelli, Tullio, and MarcoPagano,2000, “Searching for Non-linear Effects of Fiscal Policy: Evidence from Industrial and Developing Countries,” European Economic Review, Vol. 44, pp. 1259–89.
Giavazzi, Francesco and MarcoPagano,1990, “Can Severe Fiscal Contractions be Expansionary? Tales of Two Small European Countries,” NBER Macroeconomics Annual 1990.
Giavazzi, Francesco and MarcoPagano,1996, “Non-Keynesian Effects of Fiscal Policy Changes: International Evidence and the Swedish Experience,” Swedish Economic Policy Review, Vol. 3 (Spring), pp. 67–103.
Hamilton, James D.,1994, Time Series Analysis, Princeton University Press (Princeton NJ).
Hemming, Richard, MichaelKell, and SelmaMahfouz,2000, “The Effectiveness of Fiscal Policy in Stimulating Economic Activity—A Review of the Literature,” SM/00/66, IMF, March2000.
Hemming, Richard, SelmaMahfouz, and AxelSchimmelpfenning,2002, “Fiscal Policy and Economic Activity During Recessions,” Mimeo.
HM Treasury, 1999a, “Fiscal Policy: Public Finances and the Cycle,” Budget Publications, March.
van den Noord, Paul,2000, “The Size and Role of Automatic Fiscal Stabilizers in the 1990s and beyond,” OECD Economics Department Working Paper No. 230.
Perotti, Roberto,1999, “Fiscal Policy in Good Times and Bad,” Quarterly Journal of Economics, Vol. 114, No. 4 (November), pp. 1399–1436.
Perotti, Roberto,2000, “What do we Know About the Effects of Fiscal Policy?” Paper presented for the XII Conference of the Italian Society of Public Economics (SIEP), in Pavia, October6–7.
Sutherland, Alan,1996, “Fiscal Crises and Aggregate Demand: Can High Public Debt Reverse the Effects of Fiscal Policy?” Journal of Public Economics, Volume 65, Issue 2, Pages 89–244 (August).
Tanzi, Vito, and HowellZee,1997, “Fiscal Policy and Long-Run Growth,” IMF Staff Papers, Vol. 44, No. 2 (June), pp. 179–209.
Prepared by Julio Escolano. The author acknowledges the insightful suggestions and contributions from the mission team and from HM Treasury and Bank of England participants in a seminar held during the mission’s visit to London.
As has become standard-perhaps unfortunately-in the related literature, the term “Keynesian” is used loosely in this paper to indicate a significant positive effect of government expenditure, or tax cuts, on output. Nevertheless, as discussed later, a small estimated size of fiscal multipliers is not necessarily inconsistent with standard Keynesian theory.
A recent and comprehensive review of the literature and existing evidence covering both theoretical and empirical aspects can be found in Hemming, Kell, and Mahfouz (2000). For a discussion of the long-run impact of fiscal policy on growth see Tanzi and Zee (1997).
Hemming, Kell, and Mahfouz (2000) contains a summary of results across the literature from estimates of fiscal multipliers based on macroeconomic model simulations, including estimates based on the OECD’s INTERLINK model for G-7 countries. European Commission (2001) reports fiscal multipliers based on QUEST simulations for countries in the EU.
Analyses of expansionary fiscal contractions and contractionary fiscal expansions based on case studies include, among others, Giavazzi and Pagano (1990), Giavazzi and Pagano (1996), Alesina and Perotti (1997), and Alesina and Ardagna (1998).
The search for non-Keynesian effects of fiscal policy focused initially on OECD countries. For example, Giavazzi and Pagano (1990) studies the cases of Denmark and Ireland, and Giavazzi and Pagano (1996) extends the search to 19 OECD countries while focusing on Sweden. Giavazzi, Jappelli, and Pagano (2000) extended the analysis to developing countries using the World Bank dataset. Baldacci, Cangiano, Mahfouz, and Schimmelpfenning (2001) is also based on a comprehensive country set, including transition, emerging, and developing economies. Hemming, Mahfouz, and Schimmelpfenning (2002) also finds some instances of non-Keynesian effects of fiscal policy during recessions.
The output expansion during 1976-1989 was interrupted by two years of negative growth in 1980-1981.
The width of confidence intervals throughout this paper is set to two standard deviations, which under normality assumptions implies roughly 5 percent confidence.
Seasonally adjusted data were used since some variables (e.g., GDP) are not available in other form. The source of all data series used in this paper is the National Statistics office of the United Kingdom with the exception of the REER, which is from the IMF database; the index of world demand for U.K. exports, which is from the OECD database; and interest rate series, which were provided by the Bank of England. The model was also estimated with the variables set in real per capita terms without any significant change in the results.
Also, since seasonally adjusted data were not available for all taxes before 1987, the X-l 1 method was used (after deflating the series) for data before that date when necessary. Fiscal variables correspond to the general government, although taxes are not consolidated between central government and local authorities.
A more synthetic dummy variable indicating only the party description of each government in office (i.e., Labour or Conservative Party governments) was statistically less significant than dummies for each of the individual governments.
This implies in turn that the causality, if any, in contemporary comovements between the REER and taxes or transfers runs from the fiscal policy variables to the real exchange rate. Hence, any bias introduced by this assumption is likely to magnify the estimate of the fiscal multipliers.
See van den Noord (2000) and HM Treasury (1999a and 1999b). Also, since the translation of annual elasticities into quarterly elasticities is not necessarily trivial, regressions of quarterly data were run to verify the values chosen.
See Hamilton (1994).
If fiscal policy has nonlinear effects, with Keynesian effects operating in “normal” times and non-Keynesian effects dominating when confidence and expectational effects are strong, the estimates of the fiscal multipliers should probably be interpreted as an “average” over the sample period, with one effect offsetting the other.
Confidence bands were obtained by Monte Carlo simulation. The panels in Figure 2 show the impulse response functions of the variables in the horizontal axis to shocks to the variables in the vertical axis. Thus, for example, the column labeled as GDP shows the response of GDP at different lags to shocks in the variables labeled in the vertical axis, and the three first vertical panels in that column can be interpreted as the traditional fiscal multipliers of government consumption, taxes, and transfers respectively. Incidentally, the model computes as a by-product the automatic stabilizers, that is, the impact of an increase in output on the budgetary variables considered. These are the impulse response functions of the budgetary aggregates to a shock to GDP (i.e., the three first panels of the row corresponding to the GDP in Figure 2). They show the conventional signs and magnitudes consistent with existing estimates (see van den Noord (2000) and HM Treasury (1999a and 1999b)).
The REER is normalized to 100 in 1990 and has oscillated around 111 in recent times. Therefore, a one percent increase and an increase of one percentage point are roughly equivalent.
This effect is significant at 10 percent probability but not at 5 percent. Nevertheless, given that consumption represents a large proportion of GDP (about 68 percent), even a small contraction of private consumption may have large output effects.
The estimated effects of taxation on consumption and investment have opposite directions. This is consistent with Keynesian theory, which in turn is agnostic about the sign, as pointed out in Blanchard and Perotti (1999). They find that a positive shock to taxation has strong negative effects on both consumption and investment in the United States.
The interest rate variables corresponded to three-month LIBOR and three-year government maturities.
In fact, the estimated response of short-term rates to a positive shock to government expenditure (consumption and, to a lesser extent, transfers) was a statistically significant decline in rates. This could reflect the possibility that fiscal and interest rate policies were typically undertaken jointly and with the same direction during the sample period. In this regard, notice that the Bank of England did not obtain operational independence until 1997. Until then, the Treasury had ultimate responsibility for setting the policy interest rate, as well as for budgetary policies.
For example, with the exception of 1979-1980 and 1989-1990 unemployment is estimated to have been above 6 percent.