TO WHAT EXTENT ARE PUBLIC SAVINGS OFFSET BY PRIVATE SAVINGS IN THE OECD?

Journal of Economics and Finance, Fall 2006 by Holmes, Mark J

Abstract

The substitutability of private and public savings has implications for the effectiveness of fiscal policy. Using annual data for the period 1970-2004, this study re-examines long-run relationships between OECD private and public savings rates. However, unlike previous work, panel data unit root and cointegration tests are employed. The results confirm substitutability where strong Ricardian Equivalence is rejected for the entire OECD panel. There is support for weak Ricardian Equivalence with less than perfect substitutability. Indeed, it is argued that existing studies most likely overstate the extent of long-run substitutability particularly with regard to EU countries. (JEL E6)

Introduction

While there are many theories that explain private saving behaviour, it is the relationship between private and public savings that assumes a position of particular interest. The relationship is highly relevant to the effectiveness of macroeconomic stabilisation policy because the impact of fiscal policy on aggregate demand depends on the response of private saving to changes in fiscal stance. However, the impact of lower public deficits on national saving remains both theoretically and empirically controversial. The notion of perfect substitutability between private and public savings relates to the seminal work of Barro (1974) on the Ricardian Equivalence Proposition (REP). Here it is argued that rational, forward-looking consumers react only to permanent changes in government spending. However, in this context, the validity of strong REP rests on a number of assumptions whose strength may lead one to question the likelihood of perfect substitutability between private and public savings.

Following Lopez, et al. (2000), one may divide the empirical studies that address REP into two main categories. The first group of studies analyse private consumption functions derived from intertemporal optimisation and then test for the substitutability between private and public consumption [see, for example, Karras (1994), Evans and Karras (1996) and Khalid (1996)]. Ricciuti (2003) points to the inconclusiveness of the literature regarding these tests of REP arguing that it is usually rejected within a life-cycle framework, but not rejected when the empirical analysis is based on optimisation models. The second group of studies are based on reduced form equations for saving that directly measure the impact of fiscal variables thereby providing offset coefficients for private and public saving [see, for example, Edwards (1996), Masson et al. (1998)). This study considers the relationship between private and public saving ratios for a large sample of OECD countries and therefore has more in common with the second approach.

In contrast to much of the existing literature, this study explicitly focuses on estimating the long-run offset coefficient between public and private savings. However, a key contribution to the literature is offered through the application of an econometric methodology based on panel data unit root and cointegration testing recently introduced by Im et al. (2003), Bai and Ng (2004), Pesaran (2004) and Pedroni [(1999), (2001), (2004)]. While the more familiar univariate ADF unit root and cointegration tests [Engle and Granger (1987) and Johansen (1988)] have been shown to suffer from power deficiency, the panel approaches employed in this study use more observations and exploit the cross-country variations of the data in estimation thereby yielding higher test power. In addition to this, the long-run cointegrating panels are estimated using fully modified ordinary least squares (FMOLS) which provides estimates of the long-run offset coefficients between private and public savings and also facilitates formal tests of restrictions for REP for panel data.

The paper is organised as follows. The following section discusses the recent relevant literature on estimating the offset coefficient and sets out the unit root and cointegration methodology that is applied to a panel of OECD countries. This is the first study that employs savings data for a large sample of OECD countries and tests the null hypotheses of nonstationarity of national savings or non-cointegration between the private and public savings within a panel framework. This methodology enables us to reconsider the earlier estimates of the long-run offset coefficient provided by previous studies. Indeed, a formal distinction is made between strong and weak REP. The third section discusses the data and results. The data set comprises various panels of up to nineteen OECD countries over the study period 1970-2004. Using the panel data tests, there is evidence that the degree of substitutahility between private and public savings has been overestimated in previous studies. In addition to finding against strong REP, we also find evidence that the degree of substitutability has fallen over the study period with EU countries exhibiting the least degree of sensitivity between public and private savings. The implication of this is that fiscal policy may actually be more effective than hitherto thought and the degree of effectiveness may have actually increased over time. The final section concludes.