The saving-investment relationship is at the center of one the oldest debates in economics. Even though all economists agree on the existence of the saving-investment identity, disputes start once the direction of causality is sought. In fact, almost all the debates between (effective) demand-side vs. supply-side economics can be summarized in the saving-investment debate.
There are various theoretical as well as empirical studies on this topic, yet one aspect of this issue remains unnoticed; how saving is estimated. A close study of data shows that (gross) national saving data does not exist ex ante and it is derived from investment data (Ozgur, 2017). In a closed economy, national saving equals investment (S = I), and in an open economy national saving equals investment plus the current account balance (S = I + CAB) (Krugman et al., 2012: 303). Alternatively, gross national saving (S) equals net saving (NS) plus consumption of fixed capital (CFC) (SNA, 1993). A close study of how saving is estimated can show that, in both methods, saving is derived from investment data, and we can use the United Nation’s System of National Accounts (SNA, 1993) for this purpose. According to SNA, (gross national) saving data is estimated ex post as a residual, and (gross national) saving does not exist ex ante. The difference between ex post and ex ante saving is crucial for understanding the nature of macroeconomic events, and developing macroeconomic policies -as promoting national saving is often seen as a remedy for current account deficits, government and foreign debt, and sluggish economic growth. However, if saving is an outcome, not the cause, of economic events, the order of economic policies should be rearranged.
If we start with saving-investment identity, IMF’s World Economic Outlook (WEO) defines gross national saving as follows: “Gross national saving is gross disposable income less final consumption expenditure after taking account of an adjustment for pension funds (SNA, 1993). For many countries, the estimates of national saving are built up from national accounts data on gross domestic investment and from balance of payments-based data on net foreign investment” (IMF, 2014). Though the WEO database has limited information about the derivation of savings, SNA shows that gross national saving data can be derived through basic income accounting identities (SNA, 1993: 55, 256, 274). Following SNA methodology, gross domestic product equals the sum of consumption, investment, government spending, and net exports:
GDP = C + I + G + NX (1)
GDP + F = C + I + G + NX + F, (2)
where F is net factor incomes from abroad.
GNI = C + I + G + NX + F, (3)
where GNI = GDP + F is gross national income.
GNI + TR = C + I + G + NX + F + TR, (4)
where TR is all current transfers in cash or in kind receivable by resident institutional units from non-residents, net of those payable by residents to non-residents.
GNDI = C + I + G +CAB, (5)
where GNDI = GNI + TR is gross national disposable income, and CAB = NX + F +TR is current account balance
GDI – C – G = I + CAB, (6)
And, as a result,
S = I + CAB, (7)
where S = GDI – C – G is gross national saving.
Equation (7) is still an identity and it shows gross national saving can be derived through investment and current account balance. Anyone familiar with the WEO database can find this identity for various economies. It can still be argued that, even though saving can be derived ex post, it must have existed trough economic units’ decisions to spend and save. And for economic units’ decisions we should look at the relationship between net saving and gross saving. This relationship can be better understood from the U.S. Bureau of Economic Analysis (BEA) “NIPA Table 5.1. Saving and Investment by Sector.” As summarized in Table 1 below, gross national saving equals gross domestic investment, capital account transactions, and net lending (with a statistical discrepancy). The second column of Table 1 is similar to right-hand of equation (7). Net capital account transactions and net lending and borrowing equals the current account balance (Ozgur, 2017: 8). In this summary table, the second column represents the ex post data, and the net saving in the first column represents the net savings of domestic sectors such as domestic business, households, and government. For domestic business, net saving equals undistributed corporate profits, inventory valuation adjustment, and capital consumption adjustment; for households, it is disposable income minus consumption; for the government, taxes minus government spending.
Table 1. NIPA: Gross saving and Investment
Figure 1 shows the two columns of Table 1 for the U.S. economy in nominal terms and as a ratio of GDP between 1952 and 2012. Gross saving and gross investment in this graph represents both sides of equation (7).
Figure 1. Saving-investment identity, U.S. , 1952-2012
Source: NIPA, Table 5.1
The relationship between these variables becomes more interesting once the details of Table 1 are opened up. Gross saving has two components: net saving of domestic sectors plus consumption of fixed capital (CFC). Net saving represents actual or ex ante savings of domestic sectors whereas consumption of fixed capital is a residual and an estimated value for depreciation; it is a major balancing item, and “is one of the most important elements in the System” (SNA, 1993: 187). Figure 1 shows that the saving-investment identity holds—with statistical discrepancy—since 1952, yet net saving has continuously declined and it has never been sufficient to cover investment. The largest component of gross saving is CFC, an imputed value, to balance the equality of gross saving and investment.
Figure 2. Components of gross saving and investment
Source: NIPA, Table 5.1
Thus, equation (7) holds at the aggregate level only after the CFC is imputed. Moreover, CFC does not represent a saving decision and it is an imputed value to represent depreciation. In the U.S. economy, BEA uses a geometric pattern to estimate the CFC, or depreciation of all U.S. fixed assets for the overall service life of assets (Fraumeni, 1997). BEA uses a table for the rate of depreciation and service life of all types of fixed assets (Fraumeni, 1997: 18-19). Depreciation is high in the early years of an asset, it declines as the asset gets older, and follows a geometric pattern. For one dollar of investment, depreciation, d_(i,G), of an assets is as follows (equation 8):i = 1, 2, 3, …, where i is the age of the asset.
In equation (8), δ represents the rate of depreciation, and d represents the depreciation of a physical asset in a given year. Over the course of its lifetime, an asset will lose a fraction of its value each year, and its value will become zero at the end of its lifetime. Here, CFC or the largest component of gross saving is imputed based on past investment data. In order to show that CFC is built on past years’ investment data we can use a simple exercise based on BEA’s methodology. In this exercise, let’s assume the average lifetime of all the physical capital in the U.S. is twenty years, and then we can randomly pick depreciation rates between 1 and 10 percent for all these capital assets. In NIPA Table 5.1, U.S. investment data starts in 1952, and as a result, our estimated CFC starts in 1972. That includes the CFC of physical assets invested in 1952, 1953, and all other years up to 1971. In our exercise, the investments of 1952 were 20 years old, the investments of 1953 were 19 years old, and the investments of 1971 were 1 year old by year 1972. Starting from 1972, we can estimate depreciation for all the age groups by using the BEA’s formula and add them together in order to find overall depreciation or CFC for that year. If we repeat this exercise for every year between 1972 and 2012, we can find our estimated CFC. We can estimate different CFC series based on different depreciation rates. Finally, we can plot the estimated series of three different depreciation rates together with the BEA’s CFC for comparison.
As can be seen in the graphs of Figure 3 below, using past years’ investment data can give very similar results with actual CFC even under very unrealistic assumptions, i.e. all assets have the same depreciation rate and service life. Out of these three, a 8% depreciation rate for a 20-year lifetime gave an estimated value of CFC very similar to that of the BEA. The BEA is using different depreciation rates and life times for various assets, but this simple exercise shows that CFC can be estimated by using past values of investment. As a result, the biggest component of gross saving does not depend on what economic units actually save, but on investment data of previous years. And in this sense, neither CFC nor gross national saving can be sources of funds.
Figure 3. Consumption of fixed capital
Source: Author’s estimations from NIPA, Table 5.1
After reviewing the details of saving data, a review of SNA methodology can be helpful in understanding the conceptual basis of saving. Similar to any economic data, saving data is based on the methodology of System of National Accounts of the United Nations, which was first developed in 1953. Even though this methodology evolved and changed in 1968 and 1993, standard macroeconomic analyses did not follow these developments. According to Godley and Lavoie (2007: 23) The 1953 version of SNA had “left the monetary and financial phenomena in dark” as the focus was “saving must equal to investment.” Even though this notion is valid at the aggregate level, the real issue is who finances whom, and through which instruments. In SNA methodology, financial markets and institutions are not passive but active participants of an economic system. In 1968, a new SNA “provided a theoretical scheme that stressed the integration of the national income accounts with financial transactions, capital stocks and balance sheets” (Godley and Lavoie, 2007: 24), and this new system was also updated in 1993 (SNA, 1993). However, National Income and Product Accounts (NIPA) in the U.S. and similar macro data sources all around the world did not incorporate such developments into their systems. And many economists were similarly reluctant to use this new methodology in their models (Godley and Lavoie, 2007: 25). As a result, the financial transactions remained outside of the system, and these transactions were represented under saving as if it were a black box. This approach also enabled the classical dichotomy between real and monetary variables to survive (Godley and Lavoie, 2007: 24). The questions of who finances whom, and how an investment is financed are often ignored.
In SNA framework, saving can emerge for an economic unit as a negative or positive amount as a residual. And once it emerges, the next step is the direction of change in terms of a change in liabilities or assets. As a result, saving, by itself, is not a constraint for any economic unit.
As a result, building macroeconomic policies on the concept of saving can be misleading. Promoting national saving is often seen as a solution for excessive government debt, current account deficits, and boosting economic growth. However, lack of gross national saving is usually lack of investment or current account deficits or, usually, both of these. Spending less cannot make a nation’s goods more competitive in international markets; these issues should be addressed individually. The concept of national saving hides many macroeconomic problems. A deficit unit—a firm or a national economy—can always spend above its disposable income. Yet, it does not mean that such an economic unit can continuously increase its liabilities. Even though running a deficit may not be a problem, running chronic deficits can lead to accumulation of liabilities, and eventually creating financial instability. In a macroeconomic framework, financial positions of domestic private sectors, government, and the rest of the world are interdependent (Parenteau, 2004; Zezza, 2009). The interaction between these sectors, and the changes in assets and liabilities of these sectors can yield more information than gross national saving.
 CFC “does not represent the aggregate value of a set of transactions. It is an imputed value whose economic significance is different from entries in the accounts based mainly on market transactions. (…) Its value may deviate considerably from depreciation as recorded in business accounts or as allowed for taxation purposes, especially when there is inflation. Consumption of fixed capital should reflect underlying resource costs and relative demands at the time the production takes place. It should therefore be calculated using the actual or estimated prices and rentals of fixed assets prevailing at that time and not at the times the goods were originally acquired” (SNA 1993: 182).
 For details of CFC, see Ozgur, 2017.
Godley, W. and M. Lavoie (2007) Monetary Economics: An Integrated Approach to Credit, Money, Income Production and Wealth, New York: Palgrave MacMillan.
Krugman, P. R., M. Obstfeld, and M. J. Melitz (2012) International Economics, 9th Ed., Boston, MA.: The Pearson Series in Economics.
IMF (2014) World Economic Outlook April, Washington, D.C., IMF.
Ozgur, G. (2017) How Saving Data is Estimated?, University of Massachussetts-Amherst Political Economy Research Institute, Working Paper 436.
Parenteau, R. (2004) “Exploring the Economics of Euphoria: Using Post Keynesian Tools to Understand the US Bubble and Its Aftermath.” In L. R. Wray and M. Forstater (eds.), Contemporary Post Keynesian Analysis. Northampton, MA: Edward Elgar, pp. 44-66.
United Nations (1993) System of National Accounts. Washington, D.C.: United Nations.
Zezza, G. (2009) “Fiscal Policy and the Economics of Financial Balances.” Levy Economics Institute of Bard College Working Paper 569. (http://www.levyinstitute.org/pubs/wp_569.pdf)
Triple Crisis welcomes your comments. Please share your thoughts below.