Discussion Paper

No. 2017-96 | November 14, 2017
What do aggregate saving rates (not) show?


The aggregate saving indicator does not directly reflect changes in individuals’ microeconomic behavior. From the official statistics’ point of view, households choose
between spending, which generates additional income and consumption in the economy, and setting money aside, which does not. Formally, households may not (if the authors disregard housing investment) choose to save, because the aggregate saving statistical indicator is a residual concept defined as the ensuing difference between aggregate disposable income and consumption. It measures the change in net worth, which, in a closed economy, may only be generated by the production of capital goods and an increase in inventories. Using an agentbased model, the authors show that shocks unrelated to structural changes in households’ behavior may generate positively correlated fluctuations in the aggregate saving rate, productivity growth and lending. Meanwhile, a genuine increase in the average individual propensity to save is not necessarily associated with a higher aggregate saving rate.

JEL Classification:

C63, G21, O16, O40


  • Downloads: 167


Cite As

Alexey A. Ponomarenko and Alexey N. Ponomarenko (2017). What do aggregate saving rates (not) show? Economics Discussion Papers, No 2017-96, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2017-96

Comments and Questions

Anonymous - Referee Report 1
December 05, 2017 - 14:54

See attached file

Alexey Ponomarenko - Reply to Referee Report 1
December 10, 2017 - 09:58

see attached file

Anonymous - Referee report 2
December 28, 2017 - 09:35

The work focus on an important aggregation problem: the link between the individual saving rate and the aggregate saving statistical indicator. In particular explore the aggregation relation using an agent-based model showing that shocks unrelated to structural changes in individual behavior generate positively correlated fluctuations in the aggregate saving rate, ...[more]

... productivity and lending. Overall, I think that the work is interesting and the analysis is competent. Obviously, the saving rate aggregation problem is not new in the literature. Among others there are the works in Boskin (1992) “Issue in the Measurement and Interpretation of Saving and Wealth” in an NBER paper and Blundell-Stocker (2005) “Heterogeneity and Aggregation” in the Journal of Economic Literature analyzing the problem. The strength of the paper is the investigation of the problem using an agent-based model with a correct analysis in my opinion.

Alexey Ponomarenko - Reply to Referee Report 2
December 28, 2017 - 11:52

Thank you for your comments!

We intend to add these references to the paper.

Yougui Wang - Meaningful topic but scrutinized description of the modelling is needed
January 15, 2018 - 20:48

The aggregate saving is defined as the difference between aggregate income and consumption. This concept of saving is associated with the accumulation of capital, and thus as elaborated in Solow model, it is always equal to investment. In contrast, the saving behaviour of household can be characterized by the propensity ...[more]

... to save, resulting from an abstention from consumption. This concept is the basis of Keynesian multiplier process. Based on the former one, it has been proposed that saving is a good conduct for the resulting more capital will enhance the economic growth and bring about more welfare. But with the latter one in mind, more saving would correspond to the lower equilibrium income and may cause a reduction in welfare. The confusion of the two concepts of saving has raised a long debated “paradox of thrift” or “paradox of saving”. Therefore, the clarification of the two savings is very meaningful and crucial for understanding how a macroeconomic economy works. I believe that this attempt is also helpful for comprehending the causes of the secular stagnation, which is currently a great challenge to macroeconomists.

In order to distinguish the two measures of saving, the authors first illustrate the distinction of them by using the balance sheet and flows of fund approaches. I think this is an explicit illustration on the difference between aggregate saving and the behavioural saving. Then they employed agent-based modelling approach, which is the most appropriate tool for the object of this work. The model incorporates not only goods market, labor market, but also credit one. The integration of the credit creation seems make the artificial economy so complicated that the behavioural saving would have impacts on both capital accumulation and credit creation. The measure of aggregate saving should reflect the change of capital, which might be different from the change in credit volume. Actually, the impacts of the behavioural saving on the two sorts of variables have been examined in this work. Nevertheless, the authors should specify which kind of saving they refer to exactly in the last two paragraphs of Section 2. The description of the model is hard to follow and some innovative ideas are not well highlighted.

As assumed by the authors, the investment is financed by both remaining profit and borrowing from the bank. The proportion of financing sources varies over time and is dependent on the setting of economic conditions. I guess this is the reason why the relationship between the behavioural saving and the aggregate measure is uncertainty.

Alexey Ponomarenko - Reply
January 16, 2018 - 17:34

Thank you for your comments!

Indeed, the main purpose of the paper is to discuss the distinction between aggregate saving (accumulation of capital) and saving behaviour of household (abstention from consumption). Although I guess that modelling the credit market is not strictly necessary to illustrate this difference and we ...[more]

... have included it into the model in order to incorporate another concept that is sometimes erroneously associated with households’ saving – the process of money creation.

We intend to describe the conformity between the elements of our model and the standard specifications (Ashraf et al. 2017 and Popoyan et al. 2017) in more detail to facilitate understanding of the modelling set-up. We will also make the appropriate clarifications in Section 2.