Knowledge that Transforms

Lending institutions usually operate in a set pattern with respect to FHA and VA lending. That is, life insurance companies and commercial banks are stronger supporters of federal housing finance programs than savings and loan associations. Therefore, the impact of FHA and VA programs on specific markets is predictable. That is, in those markets where savings and loan associations are the predominant lenders the proportion of insured lending will be lower than in areas where life insurance companies and/or commercial banks predominate. The object of this note is to test the applicability of these conclusions to the city of Ann Arbor, Michigan, a non-metropolitan mortgage market. For the purpose of this note, “market structure” is defined in terms of the percentage distribution of home mortgage recordings among the various types of lenders.22 Gillies and Curtis defined market structure in terms of the proportion of outstanding mortgages held by each type of lender (ibid., p. 364). Since in the Ann Arbor mortgage market the volume of assignments has been small relative to total recordings, the two definitions of market structure are roughly interchangeable for operational purposes. Thus two markets would be “structurally” identical if each type of lender records the same proportions of mortgages in both markets. Under this definition, it is clear from Table 1 that in 1956 the structure of the Ann Arbor mortgage market differed in a number of ways from that of the national market as a whole. With the exception of mutual savings banks, all types of institutional lenders operating in the national mortgage market were repre-sented in the 1956 Ann Arbor market. About 38 per cent of the value of Ann Arbor's home mortgages was furnished by three savings and loan associations—the most important single source of funds. On a national basis, savings associations accounted for 35 per cent of home mortgage recordings. Next in importance to these mutual organizations were four commercial banks, which as a group recorded close to 24 per cent of Ann Arbor's home mortgages. The corresponding figure for commercial banks in the national market was 20 per cent. Life insurance companies, which recorded about 7 per cent of home mortgages on a national basis, were twice as important in the Ann Arbor market. As a group, they underwrote approximately $2.6 million of home mortgages, or 14 per cent of the Ann Arbor 1956 total. Because of the relative abundance of long-term funds from local and national institutions, individuals have been insignificant as a source of mortgage money in Ann Arbor. In 1956, only 3 per cent of the value of home loans on properties within the city were supplied by individuals. The individuals' corresponding share in the national market was 13 per cent. “Other” lenders, who accounted for the remaining 21 per cent of 1956 home mortgage recordings in Ann Arbor, included one trust company, one educational institution, and several mortgage and realty companies. So far, we have described the structural disparities between the Ann Arbor and the national mortgage markets. The question is, with what degree of accuracy can these disparities be utilized to forecast differences in the use of government mortgage insurance programs in the two markets? We know that life insurance companies and commercial banks are strong national supporters of FHA and VA programs; we have seen the greater combined importance of these groups in the Ann Arbor market. Can one predict, on the basis of these points, that the proportion of insured lending would be higher in Ann Arbor, and that of conventional lending accordingly lower, than the national average? Table 2 shows that in 1956 the ratio of conventional lending to total lending in Ann Arbor was actually 73 per cent, 4 percentage points higher than the corresponding national average of 69 per cent. What we have here is apparently a local exception to the general tendency described by the Gillies and Curtis' hypothesis. How can we account for this exception? A closer examination of Table 2 reveals two important circumstances. First, commercial bank activities with respect to conventional lending in Ann Arbor were quite a bit above the national average. While 66 per cent of the total volume of home mortgages recorded by all commercial banks in the United States were of the conventional type, the corresponding figure for Ann Arbor was 77 per cent. Second, “other lenders”-not an unimportant market force-were significantly heavier conventional lenders in Ann Arbor as compared with the national pattern. Thus, while only 25 per cent of the total volume recorded by “other lenders” in the United States was of the conventional type, the comparable figure for Ann Arbor was 76 per cent.33 According to detailed statistics, the predominance of a local educational institution (which restricts itself to conventional lending only) over trust, mortgage, and realty companies was largely responsible for the unusually heavy conventional lending on the part of Ann Arbor's “other lenders.” These two factors combined were apparently more than enough to offset the unusually low conventional lending by life insurance companies in Ann Arbor and to raise the city's ratio of conventional lending to total lending above the national average. The fact that conventional lending in Ann Arbor was proportionately higher than the national average should not lead one to conclude that Ann Arborites were not getting their share of FHA financing. As can be seen from Table 2, 18 per cent of Ann Arbor's 1956 home mortgage recordings were FHA-insured, whereas the corresponding figure for the nation as a whole was only 10 per cent. However, VA financing was considerably more scarce in Ann Arbor. Only 9 per cent of the city's home mortgage recordings in 1956 were VA-guaranteed, whereas the corresponding national figure was as high as 22 per cent. What has caused this peculiar difference in the relative impacts of FHA and VA programs? Perhaps the most important single explanation of the relatively small volume of VA financing in Ann Arbor can be traced to the VA's own appraisal policy. Under the law on VA financing, a home loan is insurable only if the purchase price paid by the veteran for the property does not exceed the “reasonable value” of the property as determined by VA appraisers. However, VA appraisers operating in the Ann Arbor area have been overconservative in their determination of “reasonable value”-overconservative, that is, in relation to Ann Arbor's high construction costs and in relation to what the home buyers are willing and capable of paying. On the other hand, the FHA and conventional lenders have been more realistic about local conditions and are not concerned with setting legal upper limits on purchase prices. As a result of all this, sellers frequently find it more profitable to deal with home buyers who do not need VA financing. This disinclination toward VA financing on the part of sellers is responsible to a considerable degree for the different impacts of FHA and VA programs on the local market.44 This paragraph is based on personal interviews with builders, real estate brokers, and mortgage lending officers operating in the Ann Arbor area. Finally, it should be added that the willingness of Ann Arbor home buyers to pay prices higher than are considered “reasonable” by VA appraisers and their ability to meet the higher down payments under FHA and conventional financing partly reflect their above-average income and wealth positions.55 In 1956, Ann Arbor “households” (defined here as an occupied dwelling unit) had an average disposable income of $10,856, almost twice as high as the corresponding national average of $5,720 (see Standard Rate and Data Service, Inc., Spot Television Rates and Data, XXXIX [May, 1957], 27, 260). In a recent article, Gillies and Curtis advanced the hypothesis that the extent of government-insured mortgage lending in various areas could be predicted on the basis of the structure of local mortgage markets. Their argument was based on the primary assumption that lenders operating in local markets exhibit a clearly delineated set of actions or attitudes with respect to FHA or VA lending. This paper has shown that in 1956, at least, the market structure did not provide us with an accurate basis for predicting the impact of FHA and VA programs on the Ann Arbor market. This is due to the fact that major types of lenders there deviated significantly from their respective national patterns regarding the extent of conventional versus insured lending. The foregoing analysis also suggests that factors other than market structure could be significant in determining the impact of federal housing finance programs on specific markets. Other factors which may have been operative in the 1956 Ann Arbor market include the lending policies of local institutions; the income and wealth positions of home buyers; the level of local construction costs; and the appraisal policy of the VA. It seems clear, therefore, that what the predictability hypothesis of Gillies and Curtis describes is a general tendency—a tendency which is subject to local exceptions such as those which occurred in the Ann Arbor market.

As soon as the increased real value of assets raises the consumption function sufficiently to start employment and output up again, prices and wages will cease falling, and the real value of assets will cease rising.… Thus the deflation of prices (which à la Pigou is supposed to continue to cause an ever increasing real value of liquid assets until full employment is reached) stops at the lower turning point. The force that is supposed to drive the economy into full employment already peters out once the lower turning point is reached.2 Hansen derived his criticism from the well-known article on the Pigou effect by Don Patinkin, “Price Flexibility and Full Employment.”3 Assuming a fixed stock of money, no adverse liquidity effects, and a flexible price level, Patinkin's argument runs, a falling price level will augment the real value of the stock of money and lead to a shifting of the consumption function to the left (less will be saved out of given alternative incomes). In effect, Hansen questions whether this shift will be large enough to re-establish a full-employment level of spending and, therefore, income. The extent to which an individual wishes to save out of current income for reasons other than the desire for future income is inversely related to the real value of his cash balances. If this is sufficiently large, all his secondary desires for saving will be fully satisfied. At this point the only reason he will continue to save out of current income is the primary one of anticipated future interest payments. In other words, if the real value of cash balances is sufficiently large, the savings function becomes zero at a positive rate of interest, regardless of the income level.4 Probably a diagram of the relevant functions at this point would clarify the arguments. In order to keep the ideas clear, let us assume straight-line functions. We start from a full-employment income, Y a , a result of consumption spending and investment spending from the functions C 1 and I 1 . The savings and investment functions are simply “rolled” 180° from their normal position and made integral with the consumption function. Thus the vertical axis in Figure 1 is either +C or —S above the xȁaxis and is +S or +I below the xȁaxis. All values are in “real” terms. The consumption function is drawn for a given real stock of money, M 0 / P 1 and the savings function is drawn for a given real stock of money and a given interest rate, r a . Thus both savings and consumption are functions of three variables: income, the rate of interest, and the real stock of money. The relationship of savings to the rate of interest is presented in Figure 2, in which savings are graphed as a function of the rate of interest, given income, and the real stock of money. The real stock of money itself is simply a given nominal stock of money with respect to some given price level. If the price level rises (falls), the real stock of money would then fall (rise). Let us start by assuming that the economy is operating at a full-employment income, Y a with an exogenous investment function, I 1 , intersecting the savings function, S 1 , at this income. Now let investment fall to I 2 —the normal Keynesian shift that is supposed to give us an equilibrium income of less-than-full employment. We should note at this point that we do not “need” the real balances effect to get an upward turning point. That is, even the pure Keynesian model will stabilize at income level Y b . Therefore, Hansen's argument uses up the real balance effect before it is even needed. Now let us assume that the price level falls to P 2 . With a fixed nominal stock of money,5 M 0 , the real stock of money will increase to M 0 / P 2 , shifting the consumption function up or to the left, simply indicating that more will be consumed at various levels of income. The rationale of this position is that the increased real stock of money is quasi-capital gains income and will thus result in greater consumption. Since a lower turning point had already been reached, may we not presume that we are now headed back toward full employment? At least we shall be at some intermediate position between Y b and Y a . Finally, we have the effect of cash-balance changes on savings. As the real stock of money increases with the falling price level, the public's “secondary” desires for savings will be satisfied by the greater stock of money it is holding. Therefore, there will be less savings at various possible interest rates, implying that the savings function in Figure 2 will shift from S 1 to S 2 .6 But a shift in the savings function must also be shown in Figure 1. Such is shown by a shift in the savings function in Figure 1 from S 2 to S 3 . Of course, the consumption must also shift up to C 3 . And income is now in equilibrium at Y a ′ , a position which appears on the diagram to be “over” full employment but which is not necessarily any specific level of income until we know the magnitudes of the variables involved. That is, we must know the actual flexibility of prices and the reactions of consumption and savings to them. If we allow ourselves to introduce an accelerator, that is, if we presume that investment is some function of consumption, we have cause for a shift in the investment function back toward its original position. But we are not limited to relying on this accelerator for an upward push after the turning point. Both the consumption and the savings functions will provide a stimulus; and it is this that Hansen failed to take account of. In brief, there are many combinations of investment (= savings) and consumption spending which could give us a full-employment level of national income. And even though we cannot say that full employment is necessarily reached, we have forces in reserve which do not “peter out once the lower turning point is reached.”7 With respect to monetary policy, the real-value-of-balances theory in effect says that we can get any price level and any level of business activity we want by manipulating the stock of money. (A priori, it would seem that if we could get a real stock of money and corresponding level of activity by altering the nominal stock of money, we should be able to get the same effect by letting the price level fluctuate around a given stock of money.) Practically, we should note, first, that we do not have a fixed stock of money and, second, that we do not have a fixed stock of “liquidity,” primarily because of the operations of the fractional reserve banking system.8 Further, our economy is not static but shows year-to-year increases in both total productivity per capita and the total number of people sharing this greater productivity. Therefore, in order that economic stability may be maintained at full employment with a reasonably stable price level, some part of the stock of liquid assets must be increased at a fairly constant rate.9 Since an increase in the real value of liquid balances may take place either by an increase in M or a decrease in P, or a combination of changes in both, we have no good reason for following the austere alternative which calls for P to do all the “work.” In fact, the analysis of the Pigou effect is simply an exercise for getting the relevant variables delineated and does not imply recourse to any particular policy. It shows only what may be expected under the most simplified conditions. At present, Federal Reserve System purchases of government securities in the open market is the principal means that our economy has for augmenting the stock of money. The alchemy, “painlessly” exercised by the Federal Reserve banks, of purchasing assets of a lower order of liquidity—i.e., government securities—and emitting liquid liabilities of a higher order—i.e., Federal Reserve Notes or member—bank reserves-is certainly capable of giving the economy in the long run, as well as in the short run, the liquidity impulses it needs to satisfy the cash-balance demands of the public without recourse to falling prices, even if other minor factors in the balance sheet behave contrariwise.10 One common charge against the effectiveness of monetary policy on the banking system is that greater reserves in the hands of the banks will result only in “bank hoarding.” We should note, however, that, although the behavior of banks is less predictable than that of people—since banks have to hold only fractional reserves against liquid liabilities—they are not likely to build up an infinite stock of liquidity either. They, too, will be faced with an inducement to disinvest cash balances for more attractive, income-earning assets. Hansen, in his discussion of policy, surprisingly seems to regard the banking system as dynamically “neutral.” He says: “Bank created money involves both assets and liabilities which balance each other off.”11 And on the next page he implies the same kind of error when he says: “A mere monetization of publicly owned securities will, however, not increase total liquid asset holdings.“12 Although Hansen is perfectly correct in holding that a Pigou effect would not be involved as the banking system expands and contracts, dramatic positive shifts in the coefficients of liquidity of the total asset structure of the economy would certainly take place. The same would hold for monetization of the public debt. These qualitative shifts in liquidity may be even more important, under some conditions, than the quantitative changes subsumed under the Pigou analysis.