V. Saving behaviour and the effectiveness of fiscal policy

OECD Economic Outlook, Dec, 2004

An error-correction procedure is used

There are several options for estimating the relationship between private and public saving. The specification preferred here is a reduced-form error-correction one, in which private saving is regressed on public saving and short- and long-term dynamics are modeled explicitly. (5) The saving equation can be estimated as follows:

(1) [S.sup.priv.sub.it] = [[alpha].sub.0] [[alpha].sub.1][S.sup.pub.sub.it] [[alpha].sub.2][X.sub.it] [e.sub.it]

(2) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Where [S.sup.priv.sub.it] and [S.sup.pub.it] denote, respectively, the private and the public saving ratios in country i at time t, [X.sub.it] is a vector of control variables, e and v are disturbance terms, and [DELTA] is the first-difference operator.

Equations (1) and (2) can be estimated jointly by solving Equation (1) for [e.sub.i,t-1] and substituting for it into Equation (2), which allows for the inclusion in the estimating equation of the right-hand-side variables in first-differences and in lagged levels, such that:

(3) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

The private saving offset is estimated at about one-half in the short term ...

The main empirical findings presented in Table V.1 control for other main determinants of private saving (see Appendix for fuller discussion). (6) The private saving offset is estimated at about one-half in the short term, while the corresponding long-term offset is estimated at about 70 per cent. (7) The magnitudes of these estimated offsets suggest that, in response to a fiscal tightening of approximately 5 per cent of GDP --comparable to that of the OECD area as a whole in the previous upturn, between 1993-2000--private saving would be expected to fall by about 3 1/2 per cent of GDP over the period. The effect on national saving of a fiscal easing of this magnitude, all else unchanged, is therefore of a rise of about 1 1/2 per cent of GDE

... abstracting from wealth effects

As noted, these estimates of the direct effects of budget deficits on saving abstract from the wealth effects of budget deficits on saving which may themselves be significant. Normally, declining deficits would be expected to have positive wealth effects. In the late 1990s, falling budget deficits were associated with a decline in private saving rates, related to increasing household net worth (Figure V.5). However, examples of higher budget deficits coinciding with growing private sector net worth can also be found, as in the United States during the 1980s. (8)

[FIGURE V.5 OMITTED]

The United States seems to be an exception

To test whether the degree of offset varies from country to country the cyclically-adjusted budget balance (measuring public saving) was interacted with a dummy variable taking value "1" for selected countries and "0" for all other countries in the panel. Based on this methodology, the private saving response appears to be positive in the United States over the longer term (Table V.2). (9) This finding should be interpreted with caution, (10) but could reflect either a greater confidence that deficits will not ultimately be reflected in higher taxes or an association between higher deficits and positive wealth effects not identified in the controls. Applying the same procedure to other major OECD countries indicates rather consistent behaviour, although differences in the level of public debt might be expected to affect the offset.