THE LAST FEW YEARS have seen a large and interesting development of financial statistics. Starting out from certain well-established bases—the statistics of banks, of government finance, and of the balance of payments—financial statistics have stretched their tentacles upward and sideways, and intertwined them, until they now cover, in principle, an area as broad as that of the national income and social accounting statistics—in principle, for the new area is only thinly held. In fact, not much more than the bare lattice work of the concepts is visible at this stage, with empty squares for the statistics to occupy in the future.1
Many countries leading in the field of national income statistics are now in the process of matching the “upper halves” of their accounts, containing data on income, investment, and saving by sectors, with similarly articulated “bottom halves,” showing borrowings and lendings. In Germany, France, the Netherlands, and Sweden (to mention only a few countries), these financing statistics have been tied on to the existing national income accounts; in the United States, a wholly new set of accounts, integrating income and financing accounts, has been developed by the Board of Governors of the Federal Reserve System in the form of the “Flow of Funds.”
These developments have breathed a welcome new life into social accounting statistics. This new field with its new vistas has rekindled interest in social accounting work, which, having slid down a steep slope of diminishing returns, seemed to be keeping barely alive on minuscule definitional crumbs.
The development of financial statistics has occurred broadly at the same time as the revival of interest in financial, and in particular monetary, policy. It seems plausible to assume that there must be a connection between these two developments, in particular since in a number of countries (the United States as well as European countries) the central banks have been the agencies in the forefront of this statistical work. But with the notable exception of the stimulating Reports of the Netherlands Bank, there does not in general, on closer inspection, appear to be an intimate connection between the new data and the revived policies.
This paper concentrates on that very connection—the need for financial statistics as an aid to monetary and financial policy. A good deal of what I have to say applies to countries in general. But insofar as this subject involves important distinctions between countries with more, and those with less, developed financial systems, I shall deal primarily with the numerically much larger second category of countries. In at least 50 of the 68 countries that are members of the International Monetary Fund, financial policy has to be operated in the framework of an elementary financial system.
The essential unifying element between statistics and policy is, of course, theory—a coherent set of assumptions regarding the behavior of the economy. These assumptions (which obviously must have as firm a basis in observation as can be achieved) will indicate how the economy is expected to respond to changes of particular variables on which policy action concentrates. The absence of agreed financial theory has so far hampered the development of financial statistics and has also raised questions as to the theoretical basis of financial policy.
By way of contrast, there is no question as to the theoretical basis of the existing national accounting statistics. These statistics (in their modern form) originated in the 1930’s and 1940’s in conjunction with the elaboration of the Keynesian income theory. Their link to the full employment policies of the 1940’s was even more evident than this: there was a direct physical connection between the national accounts statistics and the annual reviews of policies in such documents as the U.K. Economic Survey and the early Reports of the Council of Economic Advisers in the United States.
It was income theory that gave the tremendous stimulus to the development of national income statistics throughout the world and provided, at the same time, important principles of classification for these statistics. The distinction in final expenditures between consumption and investment, which was not needed to arrive at a total for GNP or national income, found its origin in the concentration of Keynesian economics on investment as the primary autonomous variable. This theoretical consideration, reinforced by the interest in postwar years in questions of economic growth and development, is no doubt responsible for the fact that most national income presentations are still organized around the saving-investment principle.
While these statistics have grown and spread, it is well to realize that the underlying choice of investment as the main autonomous variable in income theory has not stood up particularly well. In the first place, and particularly in the less developed countries, business saving may be determined, to a large extent, by the same factors that determine business investment, so that autonomous fluctuations in investment may not produce corresponding effects on income. Secondly, consumer expenditure itself has been found to be by no means as docile a follower of income as all our propensity exercises would imply. Thirdly, and above all, the two most important autonomous variables in most economies fall essentially outside the field of real capital formation, the variable with which investment must inevitably be associated statistically. In open economies, where foreign trade plays a large role, the principal factor affecting income is contributed by fluctuations in exports. Exports do not appear explicitly at all in many consolidated social accounting statements; all there is is “net foreign investment,” the difference between exports and imports. The other important autonomous determinant of income is the surplus or deficit in the accounts of the government, a variable which has been associated with “investment” by Keynes and dubbed “honorary investment” by Robertson; in the national accounts, however, a government deficit would normally appear not as positive investment, but as negative saving.
Financing statistics provide a welcome opportunity to approach this problem from another angle. The basic problem is the same: to find, in terms of statistically operable concepts, the best multiplicand for a multiplier-type approach to the determination of income. For this purpose we want to isolate autonomous spending, that is, spending that does not constitute a mere passing on in the next round of income received in the previous round of the income stream. Fluctuations of expenditure not associated with fluctuations in income by the same sector are precisely what financing statistics focus on. The main source from which any firm, household, or other sector can outspend its current income is borrowing (we shall deal later with a secondary source in the form of the use of existing cash balances). Time series on borrowing and lending by sectors thus provide the most interesting clues for finding the sectors in which the causes of expansion and contraction have been at work. Fluctuations in the government deficit or surplus also show up in financing statistics; and if balance of payments data are used in conjunction with financial data, exports can be separated out as one of the determinants in the gross increase in the supply of money.
The accounts of the money and banking and financial system can provide a large part of the required financing statistics in a highly reliable form. In all economies a large part of borrowing and lending is indirect. The institutions in the middle of the indirect process are necessarily few, compared with the number of primary borrowers and lenders; and their records are far more easily accessible, and probably also better, than those of the primary borrowers and lenders. Financing accounts constructed by assembling, side by side with money and banking statistics, the consolidated accounts of the other parts of the financial system, with assets (and so far as possible, liabilities) classified by the economic sector against which they constitute claims (or toward which they constitute liabilities), go a long way toward the completion of a matrix of borrowings and lendings. While data on the assets and liabilities of financial institutions are relatively easily available and provide a large part of the data required for financing statistics, they have not, except for the accounts of the money and banking system, been assembled in most countries.
Insofar as the purpose of financing statistics is the measurement of the net borrowing or lending of each of the nonfinancial sectors, it is of course desirable to have a complete picture of the lending activities of all financial institutions. In most countries, the first place among these financial institutions in terms of the magnitude of net lending would be occupied by the banking system (the central bank and the deposit-money banks), while the second place would be held by the life insurance companies and, in some of the underdeveloped countries, the development banks. In most underdeveloped countries, such a large proportion of total net borrowing goes through the banking system that the monetary statistics alone would give a nearly complete picture of the net borrowing of all other sectors; in these countries, the role of life insurance and savings institutions is small and transfers of savings through a bond market or through the stock exchange are also small.
Interesting though financing statistics may be in completing and improving our information on the transactions of the nonfinancial sectors, they do not by themselves focus attention on the role of finance in the economy. It is in the nature of financial institutions that they hold financial assets against financial liabilities; during any period of time, their lending to all other sectors must therefore equal the sum of the increases in their liabilities to all other sectors; and in the picture of intersector net lending or net borrowing, the financial institutions wash out completely.
To approach their meaning, these institutions must be looked at by themselves, not as “financial intermediaries” but as the creators of particular types of liabilities which the public is willing to hold in certain proportions and at certain prices. The difference between the structures of their liabilities and their assets distinguishes one category of financial institution from another; and these variously structured institutions form together the financial link between the desires of economic subjects to hold a variety of assets and the needs of the same and other economic subjects to attract outside resources.
Among the financial institutions, the banking system stands out, not merely because of its relative magnitude but because of its ability to create its own liabilities. Both banks and life insurance companies grant credit; that is not where the difference lies. But by granting credit, banks create money; by granting credit, life insurance companies do not create life insurance. On account of this difference in nature between banks and other institutions, the banking system has in itself power to expand credit; other financial institutions do not have that power. The banking system can expand credit because a large proportion of any credits granted by it will not drain resources from the system but will remain as deposits with it. The other financial institutions, on the contrary, cannot count on any significant return flow to them of moneys put at the disposal of their borrowers. It is thus almost exactly true to say that the lending of financial institutions other than banks is determined by, and equal to, the sums received from their depositors, shareholders (in the case of savings and loan institutions), and policy holders (in the case of life insurance companies). With the stream of their lending activities thus closely tied to the stream of their receipts from borrowing, these institutions, whatever their structural importance in the capital market, do not play an original role in the process of income formation.
While the banking system as a whole has considerable power of expansion, this power is not unlimited. The deposit-money banks are limited (1) by the fact that credit expansion, by raising deposits, lowers the ratio of reserves to deposits and (2) by the absolute reduction in their reserves owing to the “currency drain” brought about by the desire of the economy to increase its holdings of currency in some ratio to the increase in its holdings of deposits. If the central bank is willing to do so, it can offset the resulting pressures on the commercial banks by the creation of reserves or the reduction of reserve requirements. The central bank itself is also subject to certain limitations because of the effect of the expansion on imports and hence on international reserves. But whatever limitations there may be to credit expansion by the banking system—and there are no limitations of the same nature to credit contraction—the banking system must be considered the central locus of expansion and contraction of money income in the economy. Because of the limitations affecting the deposit-money banks, the role of the central bank in this whole process is crucial; the central bank is also the agency through which fluctuations in income from abroad, through exports, affect the domestic money supply and, unless offset by central bank action, the domestic income stream.
For another reason, the monetary system is of primary importance in the process of income formation. In most countries, it is the lever by means of which financial and monetary policy is exercised, and hence the point of attack of a large part of any policy to control fluctuations in income. Insofar as this control is exercised by means of budgetary policy, it will also find its reflection in the holdings by the banking system of credit to the government.
The crucial place of the monetary system in these various respects makes it necessary to focus the consideration of financial statistics on the statistics of the money and banking system in the first place. It is fortunate that in almost all countries statistics on the operations of the banking system are available, statistics in far better shape and far more current than the social accounts statistics can be, at best.2 It would, indeed, be unfortunate if the interest in expanding financial statistics toward a fuller picture of financing activities were to lead to loss of this essential and available part in a statistical morass of larger, less certain, less meaningful totals which attempted, with less chance of success, to give us that picture of global financing which from a policy point of view is of only secondary importance.
A simple model of the economy can be built on the basis of injections of income that can be observed from available statistics.3 The model starts from the gross creation of money, through credit creation (banking statistics) and exports (balance of payments statistics) as the predominantly autonomous expansionary or contractionary factors. These factors then determine income and imports through a conventional multiplier process, and at the same time explain the extent to which the original creation of money will ultimately distribute itself between an increase in the quantity of money and a fall in foreign exchange reserves.
While simple, this model appears useful as a description of the main workings of the economies of many countries. It is particularly suitable to indicate the effect of credit creation on the balance of payments.
In situations where this model is rigidly applicable, monetary statistics would be the only financial statistics required. Any other financial statistics would convey interesting descriptive detail, similar to statistics of the production of individual commodities; but they would not contribute to the understanding of short-run changes in the economy. The model is nowhere fully applicable; and for that reason a wider array of financial statistics is required.
The proposition that the expansion and contraction of income can be discussed and controlled by additions to, and subtractions from, the quantity of money presupposes that there are no injections into or withdrawals from the income stream by means of changes in cash balances; the proposition implies, in other words, that the ratio of money holdings to income is a constant.
There is nothing unusual in the assumption of a fixed ratio of money to income as a first approximation. In one form or another, this assumption—of a constant velocity of circulation of money—can be found in even the earliest stages of classical monetary theory. This ratio does not differ from others which economists have assumed constant in order to construct a workable model of the economy, such as the propensity to consume, the ratio of stocks to sales, or the capital-output ratio.
The assumption that the amount of money which the economy is willing to hold is a function of income alone implies that money in that economy performs only the function of means of payment and not that of a store of value. This implication is not too far from reality in a large number of underdeveloped countries, in which the amount of wealth held in financial assets of any sort is limited, and most of that so held is money, for the simple reason that the holding of money is necessary for transactions.
Even in economies in which financial assets other than money are of some or even of great importance, the ratio of money to income may still be close to a constant. But a full understanding of why this ratio is a near constant, and under what circumstances it may change, would require a broad theory of assets, in which money is seen as a function not only of income but also of wealth. It would seem obvious that in such a theory equilibrium between the holdings of money and other assets must depend, inter alia, on the prices of the assets—in the case of fixed income financial assets, on interest rates.
Such, surprisingly rather limited, work as has been done on the factors that determine the velocity of circulation points indeed to the rate of interest as the primary factor. But this is clearly only a first approximation. For its full elaboration a new dimension has to be added to monetary analysis, which will require a further set of statistics.
A theory of this nature would be an important ingredient of the theoretical background of monetary policy in a country with a highly articulated capital market. It would also require the kind of statistics on holdings of assets of various types by various holders which are not at present generally available and which the “financial accounts” to which reference was made earlier, the “bottom halves” of the income accounts, would not automatically provide. Since the value of assets outstanding changes not merely by the issuance and redemption of new securities but also, and in a shorter run far more importantly, by the change in prices of outstanding securities, the systematic elaboration of assets statistics in conjunction with financing statistics would also have to deal with the important problem of unrealized capital gains and losses. This would be an important task in itself, since these gains and losses have, up to now, been far too readily worked out of the way as “valuation adjustments,” instead of being taken into account on their own merits.
There is thus a wide field for continued work by many of us on financial statistics. With an equal degree of sophistication, this work can proceed at various levels of approximation. First, for many countries and at many times, we may operate with monetary statistics on the assumption of a constant velocity of circulation and expect to be helpful in explaining the past, forecasting the near future, and advising on policy. Secondly, in attempting to understand the role of the different sectors of the economy in expansion and contraction, we should attempt to branch out from money and banking statistics and attempt to build up as complete a picture of borrowings and lendings as we can achieve. Thirdly, we should explore the validity of the assumption of a constant velocity of money without the development of a full theory of assets. And fourthly, we should develop statistics of assets in terms of market values as the basis for a more general theory of the holding of financial assets and of the structure of interest rates.
Mr. Polak, Director of the Department of Research and Statistics, is a graduate of the University of Amsterdam. He was formerly a member of the League of Nations Secretariat, Economist at the Netherlands Embassy in Washington, and Economic Adviser at UNRRA. He is the author of An International Economic System and of several other books and numerous articles in economic journals.
This paper was presented on December 28, 1958, at the Annual Meeting of the American Statistical Association.
For the development of these statistics from various origins, see Earl Hicks, Graeme S. Dorrance, and Gerard R. Aubanel, “Monetary Analyses,” Staff Papers, Vol. V (1956–57), pp. 342–433.
Data for 65 countries are published monthly in International Financial Statistics.
J. J. Polak, “Monetary Analysis of Income Formation and Payments Problems,” Staff Papers, Vol. VI (1957–58), pp. 1–50.