The demand for assets through a low-interest rate environment

ABSTRACT

Asset demand is important in many areas of economics. To investigate this, we estimate a globally flexible dynamic Fourier asset demand specification and calculate Morishima elasticities of substitution for six asset categories for the U.S. This specification produces negative own price (user cost) elasticities. The Morishima elasticities show that each of the six asset groups is substitutes. These results suggest that monetary policy efficacy increases in recessions and support the Barnett Critique that central banks should not ignore substitution effects and portfolio adjustments among financial assets and Friedman and Schwartz’s view that saving deposit assets should be included in money.

KEYWORDS:

• Asset demand
• elasticities of substitution
• long-run adjustment
• own-user cost elasticities

JEL CLASSIFICATION:

• E51
• E41
• E52

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ This alternative approach to estimating elasticities of substitution is to use a locally flexible functional form, which provides a local approximation to the data generating function in a delta neighbourhood of an unknown and possibly small size. Early monetary studies using locally flexible forms include the translog for U.S. cite above and an Almost Ideal Demand System by Barr and Cuthbertson (1991). White noted, however, that such forms may provide biased and inconsistent elasticity estimates. Gallant introduced the Fourier form in response to White’s critique. Fisher, Fleissig, and Serletis (2001) compare numerous locally flexible forms, effectively global flexible forms and globally flexible forms.

² The Barnett (1980) critique is composed of two parts. One part concerns the composition of monetary aggregates and our results speak to that part of the Barnett (1980) critique. A second aspect of the Barnett (1980) critique concerns the index number used in the construction of monetary aggregates and the results in this paper do not speak on that latter issue.

³ A semi-nonparametric function refers to the use of a truncated series expansion that is dense in a Sobolev norm; see El Badawi, Gallant, and Souza (1983). The functional form, of course, is parametric; the nonparametric feature is due to using a flexible number of terms in the expansion. The number of terms must be chosen carefully if the estimates are to be used for statistical inference. Generally, adaptive rules (that respond to sample size) are preferred to fixed rules (e.g., Eastwood and Gallant 1991, and reference therein).

⁴ Some of the U.S. data is publicly available on the CFS website and some of the data we obtained in personal contact with the CFS.

⁵ The data used in this research begins in September 1991 and starts after the late 1970s and early 1980s which are the structural breaks analysed by Dai and Serletis (2019).

⁶ These papers do not cover the exact same time frame as the current article so their separability results can only be suggestive for our sub-aggregation decisions.

⁷ That is the assets in A₁ along with the assets in A₂ and A₃ make up the assets in M2.

⁸ The full results are extensive and all results are available upon request.

⁹ Thus all 3,816 point estimates, 318 monthly observations times 6 asset categories for both the long and short-run, of the own-price elasticities are negative.

¹⁰ As with the own-price elasticities, the Morishima elasticities of substitution are similar in the long-run and the short-run. The coefficient of correlation of the short and long-run Morishima elasticities of substitution are greater than.9 for 17 out 30 cases and greater than.75 for 26 out of 30 cases. The lowest coefficient of correlation is above.48. So, again we only display the long-run elasticities of substitution, but the full results are available upon request.

¹¹ This is an example of where the non-symmetric nature of the Morishima elasticities of substitution pays off, as the flight to safety is naturally more pronounced in the elasticity ME₅₁, i.e. from the riskier marketable assets in A₅ to the safer capital certain assets in A₁, but not from ME₁₅ where the movement would be towards the riskier assets.