One of the principal issues of Colombian macroeconomic policy in the 1990s has been the deterioration of private savings (Figure 1). The decline in the private saving rate was usually related to consumers real expenditure, which grew by an average of 3.9% in the period 1991-1993, compared to 1.9% per annum during 1950-1990. The trade reform, the acceleration of the financial liberalization process, a reassessment of permanent income due to the oil bonanza and a sterilization policy that increased real interest rates, and caused expectations of real exchange rate appreciation, were seen as the main determinants of the high consumption growth experienced in this decade (see, among others, Urrutia an López (1994-1995), Echeverry (1996)).
López (1996) and López et. Al (1996) pointed out new avenues of research. Looking closer to the Colombian National Accounts and correcting data inadequacies, they suggested that the behavior of private consumption was not the main factor behind the sharp drop in private savings in the 1990s. According to these studies, the consumption of durable goods after the trade reform could not be blamed for the decline in the private saving rate. In fact, the latter was falling since 1998 and, until 1993, the trade reform did not cause a stock adjustment of durable goods. Moreover, they argued that the private consumption boom of the 1990s did not happen , leaving without foundation most of the traditional explanations of the decline in private savings. In contrast to the prolonged and significant increases of private savings. In contrast to the prolonged and significant increases of private consumption observed in the contrast to the prolonged and significant increases of private consumption observed in the early 1960s and 1980s, this aggregate only increased 2% of GNP in 1992. The higher growth of consumption between 1991 and 1993 was matched by a similar increase on the rate of growth of labor per-capita income, wich rose 4.3% in 1991-1993 compared to 2.4% per annum during 1950-1990. As a consequence, the potencial explanatory power of the a reassessment of permanent income and financial liberalization as determinants of private consumption in the 1990s was reduced.
The behavior of the private saving rate in the 1990s was mainly seen in these studies as the result of a secular decline in private disposable income, closely linked to tax increases. Howevwr, given that disposable income of corporations is equivalent to corporate savings, to have a better understanding of Colombian private savings. Figure 1 shows that the stability of the National Account measure of private saving observed between 1970 and 1990 was hiding a small but secular decline in household savings (almost equivalent to 4% of GNP) which was compesated by an increase in corporate savings returned to the levels attained during 1970 and 1985.
The study by Sanchez et. Al (1996) has been the first attempt to understand the rise and fall of corporate savings observed in the period 1983-1994. Using micro evidence, their results indicate that collapse of this aggregate in the early 1990s was not reflected in investment because there was almost a perfect substitution of domestic savings by collapse of the private saving rate (i.e. trade and financial liberalization) played a significant role through their effects on corporate behavior, as they relaxed liquidity corporate savins should remain an open question for future research.
The aim of this paper is to have a first look at the remaining part of Colombia's private saving puzzle: the secular deterioration of household savings. Following the literature that has developed since the paper by Davidson, Hendry, srba an Yeo (1978) - DHSY-, particular attention will be paid to error correction mechanisms (ECMs). ECMs have been successful in accounting the short run behavior of aggregate private consumption in the U.K. This in interesting since agents in those models have been usually assumed to be myopic. ECMs, however, do not necessarily imply a myopic or backward looking behavior of agents . Muellnauer and Bover (1986) showed that they could be interpreted as an Euler equation with liquidity constarined consumers. Still they are not incompatible with the pure Life Cycle Permanent Income Rational expectations Hypothesis (LCPIREH) (Campbell (1987), Campbell and Shiller (1988)). The paper is structured as follows. Section 2 considers the theoretical background under the main models of consumer's behavior. Section 3 presentes several ECMs representations for Colombian consumption and test for weak, strong and super exogeneity. This in important to ensure efficient and consistent inferences, and accurate forecasts and policy simulations (Engle, Hendry and Richard (1983)). Section 4 summarizes the main findings and presents some final remarks.