Proceedings from the ECFIN Workshop "The budgetary implications of structural reforms" - Brussels, 2 December 2005

Table 1. The costing of Budget policy decisions in the UK

Budgetary measures as a per cent of GDP¹

1. A plus sign indicates an Exchequer yield.

Source : HM Treasury, Budget 2001, 2002, 2003, 2004.

10. What this spending has achieved is open to debate, as the new metrics to measure government
outputs, following the Atkinson Review, are not yet fully in place. Assessments of feedbacks of tax and
spending changes on economic activity are even rarer. HMT’s effects of budget measures, for instance, do
not include effects on overall levels of income and spending. This is not surprising, because there is often
no consensus on effects. Depending on the underlying model and empirics, a rise in payroll taxes can have
a negative short, medium or long-term impact on structural unemployment, or none at all. Effects of R&D
tax credits or grants on R&D spending and ultimately growth are notoriously difficult to quantify. Or the
effect of savings incentives on substitution between different savings vehicles versus aggregate savings is
usually impossible to pin down. And sometimes, long lags imply that any discernible effects of reforms
would only show up after many years. Still, in the United States, attempts are underway to include feedback
effects in assessing tax and spending proposals (Box 1).

Box 1. Dynamic scoring

In the United States, legislative proposals are scrutinised by the Congressional Budget Office (CBO) and the
Joint Committee on Taxation (JCT). They provide a baseline and forecasts of the changes in expenditure and
revenues that would result from proposed legislation over the following ten years.¹ The forecasts provide a cost
estimate or “score”, for each piece of legislation that is reported by a Congressional committee (Page, 2005). In the
past, the scoring was static, not taking into account macroeconomic feedback effects. Dynamic scoring, which is still in
its infancy, takes into account induced changes on output, inflation, interest rates or other macroeconomic feedbacks.
The scoring is important as it influences how favourably initiatives are viewed in Congress and the Senate. A tax cut,
for instance, could raise output significantly, with a large cut in tax rates having little implication for net government
revenues. If this feedback is not included, the tax cut will be viewed less favourably by Congress, which is usually
constrained to keep the total revenue cost of a tax package within pre-set targets (Altshuler et al., 2005)

The outcome of dynamic scoring depends on the models used and on assumptions about macroeconomic policy
reactions. In assessing the 2004 budget, forward-looking, life-cycle growth models and more traditional
macroeconomic forecasting models were used. For the latter, various monetary policy reactions to the fiscal stimulus
were simulated, while the growth models differ in various respects. Given different models and assumptions, the
outcomes differed widely. Concerning the JCT’s analysis, incorporating dynamic effects reduced the net revenue cost
of one proposal by 6 to 28% over the first five years and 3 to 23% over the second five years. Auerbach (2005)
concludes from these first attempts, that "... it seems clear that dynamic scoring analysis has value, but also that
adjustments to estimates are smaller than some might have expected. The process to date offers some support to
those on both sides of the debate. On the one hand, the ability of CBO and JCT to produce dynamic analyses of
complex, realistic proposals lends credence to the argument that dynamic analysis and, indeed, dynamic scoring may
be feasible. On the other hand, the many models used and the many assumptions needed leave many with doubts
about the quality of these estimates and how they would fit into the budget scoring process as currently structured.”

1. On the expenditure side, the CBO provides the baseline and the scoring, on the revenue side the CBO provides
the baseline and the JCT the scoring.

46

<<   <   49   50   51   52   53   54   55   >   >>