THE FATE that an author should dread the most is to see his/her writings ignored. While I have experienced this fate with some of my writings, this is definitely not what has happened to my articles on the underground economy in the United States. For these I have been “flattered” by more attention than I would perhaps have liked. The three comments and criticisms discussed here are quite different: they deal in part with the methodology of my work in this area and in part with the empirical results. There are several ways in which I could deal with them but perhaps the simplest is to take the authors’ comments alphabetically. I shall allocate far more space to Feige’s “comment” than to the other two, largely because his is not just a comment on my paper but is also an attempt to “sell” his work to the readers of Staff Papers. Thus, I must inevitably discuss his method and results while attempting to answer his specific criticism of my work.

Feige

Before going to the substance of Feige’s comment I would like to react to its style. I feel that the message that he wants to convey to the readers would have carried far more power of persuasion if it had not been laced throughout with such expressions as “seriously flawed,” “errors,” “erroneous results,” “inappropriate use of…,” “incorrect choice of…,” “misspecification of…,” “gross underestimates of....” These strong judgments add nothing to his basic points and are likely to predispose some readers against them.

Feige’s comment consists of three main elements: first, the specific criticism of my method; second, a presentation of his own method; and third, an attempt to convince the reader that his results are the more believable. The first and second of these elements are somewhat intermingled. The criticism of my work is presented as an aside while he describes his own method. He identifies “the specific errors in Tanzi’s method” by presenting his alternative framework for estimating “unreported income.” As I read his comment, my cardinal sins are reported to be the following:

(1) to assume “that currency is the exclusive medium of exchange” for unreported transactions (italics added);

(2) to assume “that the amount of unreported income produced by a dollar of currency transacted in the unreported sector is the same as the amount of reported income produced by a dollar of currency transacted in the reported sector” (italics added);

(3) “Tanzi’s choice of a multiplicative...functional form to estimate the currency ratio therefore violates the additive specification implied by [Feige’s] equation (12)”;

(4) “Tanzi’s. . .choice of variables and his method for estimating K₀.”

Let me take these criticisms in turn. Feige does not seem to realize that “Tanzi’s method” never assumed that “currency is the exclusive medium of exchange for unreported transactions” (italics added) or that currency usage is closely related to unreported income as Feige defines the latter.¹ Feige’s is a general definition of tax evasion, while my “estimates attempt to measure the incomes that were generated through the excessive use of currency and that presumably were not reported to the tax authorities” (Tanzi, 1983, p. 303).

Far from being the same, there is a wide difference between these two definitions. Feige’s covers all tax evasion; mine covers only those forms of tax evasion associated with the use of currency and thus more properly considered as part of the phenomenon popularly called the underground economy. Tax evasion has been discussed by tax experts since taxes were instituted. The concept of the underground economy is only a few years old. Thus, there must be a difference between tax evasion and underground economy. A detailed discussion of these differences can be found in Tanzi (1980, pp. 437–41). In that article (pp. 438–40) I explained clearly that there are many forms of tax evasion that “neither necessarily lead to greater currency use nor… bias the national accounts data.”

Suppose, for example, that Mrs. Smith receives a check from her bank for interest earned on her savings deposit but fails to report that income to the tax authorities when she files her return.² Suppose that Mr. Jones receives a check from his tenant for the rent on an apartment that he owns and also fails to declare it. Suppose that Mrs. Taylor claims a large deduction for cash contributions to a church when in fact she contributed only a small part of the amount claimed. What do these actions have in common? First, they all contribute to total tax evasion. Second, the nondeclaration of these incomes or the excessive deduction claimed have absolutely nothing to do with either currency use or with the use of checkable deposits. In other words, these forms of tax evasion cannot possibly be estimated by the use of monetary statistics; yet, they are part of Feige’s “unreported income” that he claims to be measuring with his method.³ Thus, Feige cannot estimate all tax evasion in the United States by the use of monetary statistics. He should claim that his estimates, as extraordinary as they are, are lower-bound estimates of tax evasion in the United States. He should, thus, redefine what he is measuring.

Feige seems to be bothered by two assumptions made in my work: (a) that the income velocity of money is the same in the underground economy as in the official economy; and (b) that the underground economy is fueled mainly by the use of currency rather than by checks. That is, using his symbols,

In view of the importance that he attaches to these assumptions and of his critical attitude vis-à-vis them, let us examine them in his work. As I wrote in my 1980 article, my assumption about (a) was “the result of agnosticism. The author is unable to take a position between those who would argue that the velocity of money in the underground economy must be lower than in the legal economy and those who would argue the contrary. The first alternative was backed by Nancy H. Teeters [then Governor of the Federal Reserve Board] in a recent statement before a congressional committee. See Federal Reserve Bulletin (September 1979), pp. 742–43. The second alternative (i.e., ß<1) is backed by Professor Edgar Feige…” (Tanzi, 1980, p. 449). If Feige has any strong reason to make a different assumption, he should state it and declare what precise value of ß he would use and why he chose that value. What bothers me is that while in his earlier works (Feige 1979 and 1980) he assumed, arbitrarily, that the underground income velocity was 10 percent higher than its official counterpart, that is, that ß = (1/1.1), he seems to be less confident now about that parameter. Thus, in his comment, he uses a sensitivity analysis in which he assumes figures for ß going from 0.9 to 1.1. This analysis is an interesting academic exercise, but it does not carry our knowledge of the underground economy much further.

Time also seems to have affected Feige’s position as to whether and how much the underground economy uses checks compared with currency. Here again the distinction between the underground economy and the economy connected with “unreported income” is important. As I have mentioned already, checks are, of course, used in the latter economy. In fact, in some forms of tax evasion (say, when taxpayers do not report to the tax authorities interest received from banks), only checks may be involved. But, in the underground economy, as it is usually understood, the use of checks is likely to be much more limited, although I agree with him that some use of checks is also made, so that my assumption that D_u = 0 is likely to be too extreme. But how much use? Well, in his 1980 article, he assumed, arbitrarily, that K_u = 2; that is, in the underground (unreported?) economy, there is a one-dollar use of checks for every two-dollar use of currency. In the present article, he assumes that there is a one-dollar use of checks for every three-dollar use of currency; that is, K_u has gone from 2 to 3. He justifies this change by citing a University of Michigan study of informal markets. However, using the results of the University of Michigan study as a basis for determining the value of K indicates that Feige has missed completely the point of what the underground economy is. In fact, as I shall show later, the size of the “underground economy” as defined by the University of Michigan’s survey team is only a small fraction (less than 10 percent) of Feige’s estimate. There is hardly any relation between the “informal markets” of the Michigan study and the unreported income of Feige. Therefore, the information reported by the former study is of no relevance for Feige’s study.

As explained in Feige’s note, his method for estimating the underground economy (unreported income?) requires the availability of estimates for three parameters ß, K_u, and K₀. I have already discussed how he derives estimates for ß and K_u. This leaves K₀, the ratio of official currency to official checkable deposits (C₀/D₀). Feige writes that “Tanzi’s sole contribution to the literature is his effort to relax [the] assumption… that the ratio of currency to checkable deposits remains constant except for changes induced by the growth of unreported income.” Quite apart from whether that is my “sole contribution” to the literature, the implication is that it is a minor issue. Just how important is the relaxation of this assumption? The best way to answer this question is by providing the results obtained in a study prepared by Richard Porter and Amanda Bayer (1984) of the U.S. Federal Reserve Board’s Division of Research and Statistics using Feige’s currency ratio model (described in Feige’s present comment) and Feige’s transactions-ratio method used by him in earlier work. In both of these, the “sole contribution” by Tanzi has not been incorporated. The results of these studies are reported in Table 1. It should be kept in mind that the three columns are supposed to be measuring the same thing.

I am very happy to see that Feige has abandoned the fixed-ratio approach and moved to an approach which makes K₀ a function of various behavioral determinants. While I welcome him to the spirit of the Tanzi approach, I have some difficulty with his econometric specification of K₀.

In the first place, Feige should not have included the interest rate among the determinants of K₀ in his equation (10). The rate of interest should be a determinant when the dependent variable is taken as C/M2, as it was in my 1980 paper, but it should not be a determinant when the dependent variable is CID, as it was in my 1979 paper. The reason is obvious; when the dependent variable is CID, neither the numerator, C, nor the denominator, D, receives interest payments, so that the CID ratio should be largely insensitive to the rate of interest.⁴ Thus, there is no justification to have the interest rate among the independent variables.

^{Table 1.}

Feige’s Estimates of the Underground Economy

Source: Richard D. Porter and Amanda S. Bayer, “A Monetary Perspective on Underground Economic Activity in the United States,” Federal Reserve Bulletin (Washington), March 1984, p. 178.

^aIt is assumed that underground GNP equals 5 percent of observed GNP in 1964.

^{Table 1.}

Feige’s Estimates of the Underground Economy

		Using the Transactions Approach
Year	Using the Modified Currency Ratio Model1	1939 Base	1964 Basea
(In billions of U.S. dollars)
1950	21.5	27.6	43.1
1955	15.6	1.7	21.6
1960	17.1	-3.4	21.5
1965	38.6	9,6	44.3
1970	88.5	101,0	155.2
1975	246.0	467,3	567.1
1980	666.9	1.095.6	1,280.1
1981	767.6	1,765.6	1,999.2
1982	810.5	…	…
(In percentage, as a ratio to recorded GNP)
1959	7.5	9.6	15.1
1955	3.9	0.4	5.4
1960	3.4	-0.7	4.2
1965	5.6	1.4	6.4
1970	8.9	10.2	15.6
1975	15.9	30.2	36.6
1980	25.3	41.6	48.6
1981	26.0	59.8	67.7
1982	26.4

Source: Richard D. Porter and Amanda S. Bayer, “A Monetary Perspective on Underground Economic Activity in the United States,” Federal Reserve Bulletin (Washington), March 1984, p. 178.

^aIt is assumed that underground GNP equals 5 percent of observed GNP in 1964.

View Table

^{Table 1.}

Feige’s Estimates of the Underground Economy

		Using the Transactions Approach
Year	Using the Modified Currency Ratio Model1	1939 Base	1964 Basea
(In billions of U.S. dollars)
1950	21.5	27.6	43.1
1955	15.6	1.7	21.6
1960	17.1	-3.4	21.5
1965	38.6	9,6	44.3
1970	88.5	101,0	155.2
1975	246.0	467,3	567.1
1980	666.9	1.095.6	1,280.1
1981	767.6	1,765.6	1,999.2
1982	810.5	…	…
(In percentage, as a ratio to recorded GNP)
1959	7.5	9.6	15.1
1955	3.9	0.4	5.4
1960	3.4	-0.7	4.2
1965	5.6	1.4	6.4
1970	8.9	10.2	15.6
1975	15.9	30.2	36.6
1980	25.3	41.6	48.6
1981	26.0	59.8	67.7
1982	26.4

Source: Richard D. Porter and Amanda S. Bayer, “A Monetary Perspective on Underground Economic Activity in the United States,” Federal Reserve Bulletin (Washington), March 1984, p. 178.

^aIt is assumed that underground GNP equals 5 percent of observed GNP in 1964.

I also have some difficulty with Feige’s argument that the observed currency ratio is the sum of two functions, as he claims in equation (12). Assume that the economy is divided into two neatly separated parts: one the reported, or above ground, economy and one the unreported, or underground, economy. Those in the reported economy earn their income only in that economy. Those in the underground economy earn their income only in that one. It is realistic to assume that both of these groups spend their income in both economies. Under these assumptions, I cannot understand why variables such as the appropriate measure of income (y), or the interest rate, or the share of wages and salaries should not affect the currency ratio in the underground economy. Why, for example, shouldn’t those in the underground economy economize, in their capacity as spenders, on their use of currency when economic development reduces the use of currency for everyone? But perhaps I am missing Feige’s point.

Feige also objects to my use of M2 as the denominator for the currency ratio and finds my procedure to use M1 income velocity to determine underground income “internally inconsistent.” I find this criticism puzzling and somewhat frustrating. In my first article on the underground economy (1979), I had in fact used the C/D ratio, as Feige does, rather than the C/M2 ratio. The frustrating thing is that several readers had objected to that,⁵ pointing out that the C/D ratio might rise not because the use of currency rises but because the use of checkable deposits, D, falls.⁶ This might happen, for example, if institutional changes make it easier for individuals to get money out of savings deposits when needed. In such cases individuals would increase their holdings in the banks in the form of savings deposits and decrease those in the form of checkable deposits. And, of course, this might also happen when, ceteris paribus, the rate of interest on savings deposits rises. However, transactions are still conducted either with currency or with checks. Therefore, I do not find anything inconsistent (“internally” or externally) in using M1 income velocity to determine the income behind a given “underground” use of currency. It would have been wrong to use M2 income velocity and, as I have argued,⁷ it is preferable to use C/M1 as the dependent variable.

Finally, Feige objects to my use of variables. I agree with him that the series on average effective marginal tax rate prepared by Barro and Sahasakul is probably better than the three alternative series that I used in my articles. But (a) the Barro and Sahasakul series was not available when my article was written, and (b) as the comment by Zilberfarb shows “alternative measures of the tax variable do not alter Tanzi’s basic results… “ (p. 5).

Feige presents various estimates of the unreported income “that are several times larger than those reported by Tanzi” (p. 12). In fact, if 1980 is taken as the reference point, a year for which my estimate of the underground economy was about $160 billion, his estimates range from a minimum of $448 billion (or about 17 percent of GNP) to somewhere in the trillions (see Table 1). In 1980 there were 60.3 million families in the United States with a median income of $26,500. Even the lowest estimate of “unreported income” by Feige would imply that the average family was underreporting its income by about $7,400.⁸ And, of course, some of Feige’s estimates are much higher than $448 billion. The reader should ask himself if he is really spending 28 percent of his income on the underground economy or if he is underreporting that much income. With figures as huge as that we should all be immersed in the underground up to our necks.

How do Feige’s and my estimates compare with those of other researchers? Let me start with a study cited by Feige in support of one of his points. This is the study by a team from the University of Michigan commissioned by the U.S. Internal Revenue Service to survey informal markets. Let me quote the conclusion:

This study estimated the amount of informal supplier receipts by measuring the value of household purchases from informal suppliers. Based on a national probability sample of approximately 2,100 households the upper limit of informal supplier receipts … is estimated to be about $42 billion in 1981. (See McCrohan and Smith, 1983. p. 31.)

In view of this small estimate, the authors speculate that the perception that people have large amounts of income “on the side” may be due, in part, to the pervasiveness of the informal economy in small amounts throughout the economy.

A study by Frank de Leeuw, Chief Statistician of the U.S. Department of Commerce, using a new, indirect technique for measuring the underground economy, gives a lower-bound estimate for 1977 of $88 billion. For that year Tanzi’s estimate was $99.6 billion while Feige’s estimate was several hundred billion dollars. In de Leeuw’s words: “Feige’s estimate is four or five times as large. Tanzi’s estimate, however, is about the same size” (p. 71).

A study prepared for the Joint Economic Committee of the U.S. Congress using data based directly on the current population survey estimated, for 1981, an underground economy of 7.5 percent of total output. Tanzi’s estimate for 1980, the latest year for which it was calculated, is 6.1 percent of GNP (see O’Neill, 1983).

A study on the difference between the adjusted gross income as reported to the Internal Revenue Service and the adjusted gross income as derived by the Bureau of Economic Analysis of the U.S. Department of Commerce, a difference that should be closely related to Feige’s unreported income, estimated this difference at 6.9 percent of the total in 1976 and 8.6 percent in 1980 (see Park, 1985).

Finally, since Feige has cited the IRS estimates of the so-called tax gap, let me provide some figures prepared by that service. As Table 2 shows, the tax gap calculated on the basis of income unreported by filers was estimated by the IRS at about $52 billion in 1981.⁹ A look at this table will indicate that much of the gap has little to do with the underground economy or with the use of currency or checks in that economy. For example, part of it was associated with dividends, interest, and capital gains ($14.4 billion), and part of it was associated with unreported pensions and annuities, rents, royalties, state income tax refunds, alimony income, and so on. These data should not have been used, because this type of tax evasion cannot be detected by monetary statistics.

Interestingly enough, applying, as Feige does in his comment, the recent Barro and Sahasakul estimates of the average marginal tax rate to the measure of the underground economy estimated by Tanzi gives a tax gap very close to the one reported by the Internal Revenue Service (see Table 2). In other words, taxing Tanzi’s estimate of the underground economy of $160 billion with an average marginal tax rate of about 31 percent (see Barro and Sahasakul, p. 435) would give a tax gap very close to that reported in Table 2. This is hardly surprising, as the total unreported income (for filers and nonfilers) estimated by the IRS for 1981 was $163.6 billion. This figure is remarkably close to the estimate of underground economy by Tanzi for 1980, which was $160 billion.

Thomas

The comment by Thomas relates to the results rather than to the methodology of my article. His main concerns seem to be two: first, the Durbin-Watson statistics in the equations that I used in my 1983 article (computed after the first-order Cochrane-Orcutt transformation for serial correlation) were both in the 0.95 inconclusive range, thus suggesting that the transformation may not have removed the serial correlation; second, the relationship is good for the 1929–45 period but breaks down if the successive period is taken in isolation.

Regarding the first of these points, using a second-order Cochrane-Orcutt transformation raises the level of the D- W statistic to the fully acceptable level. This change provides estimations of the underground economy that are larger than those estimated earlier. The 1980 estimate, for example, is now about 8 percent of GNP rather than the previously estimated 6 percent. This, of course, raises my estimate for tax evasion to about $20 billion, from the previously estimated figure of $15 billion. Thus, although there are some changes in the parameters, the broad conclusions remain unaffected.

^{Table 2.}

Tax Gap Related to Unreported Income of Filers, By Source of Income, 1973–81

(In millions of U.S. dollars)

Source: U.S. Internal Revenue Service, “Income Tax Compliance Research: Estimates for 1973-1981” (Washington, July 1983), p. 68.

^{Table 2.}

Tax Gap Related to Unreported Income of Filers, By Source of Income, 1973–81

(In millions of U.S. dollars)

Type of Income	1973	1976	1979	1981
Wages and salaries	2,084	2,909	4,616	6,283
Dividends	1,004	1,402	2,224	3,027
Interest	1,422	1,984	3,149	4,286
Capital gains	2,367	3,304	5,243	7,136
Nonfarm proprietor income (except informal supplier income)	4,761	6,645	10,546	14,353
Farm proprietor income	831	1,159	1,840	2,504
Partnership and small business corporation income	1,723	2,405	3,817	5,195
Informal supplier income	1,528	2,133	3,386	4,608
Pensions and annuities	393	548	870	1,184
Rents	317	442	702	995
Royalties	329	460	730	993
Estate and trust income	95	133	211	287
State income tax refunds	142	198	314	427
Alimony income	13	18	28	38
Other income	306	427	677	922
Total income	17,314	24,166	38,353	52,198

Source: U.S. Internal Revenue Service, “Income Tax Compliance Research: Estimates for 1973-1981” (Washington, July 1983), p. 68.

View Table

^{Table 2.}

Tax Gap Related to Unreported Income of Filers, By Source of Income, 1973–81

(In millions of U.S. dollars)

Type of Income	1973	1976	1979	1981
Wages and salaries	2,084	2,909	4,616	6,283
Dividends	1,004	1,402	2,224	3,027
Interest	1,422	1,984	3,149	4,286
Capital gains	2,367	3,304	5,243	7,136
Nonfarm proprietor income (except informal supplier income)	4,761	6,645	10,546	14,353
Farm proprietor income	831	1,159	1,840	2,504
Partnership and small business corporation income	1,723	2,405	3,817	5,195
Informal supplier income	1,528	2,133	3,386	4,608
Pensions and annuities	393	548	870	1,184
Rents	317	442	702	995
Royalties	329	460	730	993
Estate and trust income	95	133	211	287
State income tax refunds	142	198	314	427
Alimony income	13	18	28	38
Other income	306	427	677	922
Total income	17,314	24,166	38,353	52,198

Source: U.S. Internal Revenue Service, “Income Tax Compliance Research: Estimates for 1973-1981” (Washington, July 1983), p. 68.

The second point raised by Thomas is certainly an important one, although not a new one. In fact, it had been made by Porter and Bayer (1984). As they put it (p. 182): “The data reveal that the positive relationship between the ratio of currency to M2 and taxes is strong only for the period from 1930 to 1945” (italics in original). I must confess that I have some difficulty with this point. On purely economic grounds I do not see much justification for breaking the period. The fact is that the level of taxation rose enormously from 1929 to 1945 but then fell and did not change much after the 1945 period. It started rising again in the late 1960s. If a more recent period, say, one starting about 1970, is isolated, the tax variable becomes again highly significant.

Zilberfarb

I have no particular difficulty with the comment by Zilberfarb. His suggestion to estimate an upper-bound limit to my estimates seems valid and his re-estimation of my results, using the recently constructed average marginal tax rates by Barro and Sahasakul and the marginal federal personal and social security tax rates by Seater provides a test of whether my results were excessively influenced by the variable chosen for the tax rate. His conclusions in this context are reassuring. One small qualification to his comment is perhaps necessary. He mentions that both of the tax measures that I used are based on “average” taxes. While this is true for the variable T in my 1983 article, which represented the ratio of income tax after credit to adjusted gross income, it is not true for the variable TW, the weighted average tax rate. This is usually an average marginal tax rate just like that constructed by Barro and Sahasakul. The difference is that TW was constructed with reference only to interest incomes, while Barro and Sahasakul’s was constructed with reference to total incomes of individuals.

Summaries

A Guide to Target Zones—JACOB A. FRENKEL and MORRIS GOLDSTEIN (pages 633–73)

This paper identifies key issues surrounding the advisability and practicality of adopting “target zones” for the exchange rates of major currencies.

Four fundamental questions concerning the definition of and the rationale for target zones are addressed: first, what is generally meant by a “target zone” approach to exchange rate management and how can “hard” and “soft” versions of this approach be defined; second, what are the perceived deficiencies in the existing exchange rate system of managed floating which motivate the call for the adoption of target zones; third, how might target zones remedy these deficiencies; and fourth, what factors are behind much of the skepticism over and opposition to target zones?

In addition, the paper deals with a series of operational questions of a more technical nature that weigh heavily on the practicality of implementing a target zone approach. The issues discussed include the following: how would the target zones be calculated; what currencies would be included in the system of target zones; how wide should the target zones be and how frequently should they be revised; and what policy instruments would be employed to keep actual exchange rates within the target zones, and with what consequences for other policy objectives?

The purpose of the paper is not to make the case either for or against the adoption of target zones. Rather, the intention is to raise and discuss factors that should be considered in any serious discussion of the topic.

Profits Theory and Profits Taxation—EDMUND S. PHELPS (pages 674–96)

In what has been called the “classical” fiscal system, corporate profits are subject to heavy taxation on the ground that they can be siphoned off to the government without giving rise to costly disincentives. Corporations in turn minimize their profits tax liability for the benefit of their shareowners by debt financing their entire investment. Hence the remaining profit, if any, equals the surplus of capital income over the interest cost caused by monopolistic elements. But with regard to open economies in which firms are forced to equity-finance, this view has been challenged.

This paper considers the welfare effects of a tax on profits in a small open economy in which the firms operate in world “customer markets” so that their profits are not entirely competed away. The conclusion of the analysis is that, up to a point, a tax on corporate profits increases welfare. The principal benefit is that siphoning off some of the firms’ pure profit makes possible a net addition to total tax revenue out of which there can be a lightening of marginal tax rates on work, a consequent narrowing of the wedge between the after-tax real wage and the marginal productivity of labor, and a resulting increase in the amount of work done and output produced. The principal cost is that the consequent outflow of capital reduces the before-tax wage, which ultimately limits the optimum size of the profits tax rate to something less than 100 percent. Side benefits occur if a real appreciation of the currency results or if foreign shareowners bear some of the redistributive burden of the profits tax, and a side cost if a real exchange rate depreciation occurs instead. With the introduction of partial equity financing it is still possible, though not certain, that an increase in aftertax wages will result.

Output and Unanticipated Money with Imported Intermediate Goods and Foreign Exchange Rationing—AJAI CHOPRA and PETER MONTIEL (pages 697–721)

Short-run fluctuations in the growth of real output are an important concern of policymakers in developing countries. And yet, in spite of the considerable amount of attention that this subject has received in the context of industrial countries during the past fifty years, surprisingly little analytical—much less empirical—work has been done for developing countries.

For the industrial countries, the advent of “new classical” macroeconomics has shaken the consensus about the causes of short-run output fluctuations. Successful estimation of reduced-form output equations generated by “new classical” models contributed to the acceptance of these models. It is natural, therefore, to extend this framework to the explanation of short-run output fluctuations in developing countries, and this has been done by several authors. However, structural features that are likely to be particularly important in the developing country context have typically been either ignored or introduced in an arbitrary fashion.

This paper develops a simple “new classical” structural model that includes several features likely to be of importance in developing countries. The economy is modeled as an open economy along Mundell-Fleming lines, imported intermediate goods are introduced, and foreign exchange is assumed to be rationed. A reduced-form output equation is derived from this model, which is a generalization of its closed economy analogue. After the properties of the model are analyzed, the reduced-form output equation is estimated for the Philippines. The empirical results conform quite closely to the predictions of the model.

MINIMOD: Specification and Simulation Results— RICHARD D. HAAS and PAUL R. MASSON (pages 722–67)

This paper describes the construction and use of a small macroeconomic model, MINIMOD, of the United States and its major industrial trading partners. The goal is to have a readily understandable and transparent model of manageable size that is suitable for policy analysis. Consequently, an eclectic theoretical model has been specified for each of the two economies, with equations for aggregate demand and supply of goods and capital accumulation, and with consistent treatment of government and private sector flows of funds. The model was specified such that it has desirable long-run properties, including the neutrality of money and the property that government debt cannot grow without limit relative to output. Values for the parameters of the model were obtained, with a few exceptions, from the properties of a larger, multicountry model; the paper describes the methodology of reducing a larger model to its core interactions using partial simulation techniques.

The model is simulated to gauge the effects of changes in monetary and fiscal policies under two alternative assumptions concerning expectations of future rates of inflation, of long-term bond rates, and of the exchange rate: (i) expectations adapt to past movements in the variables, or (ii) expectations are consistent with the model’s own predictions. The simulations imply that an increase in the money supply is likely to depreciate the exchange rate and to stimulate output in the home country, as prices are slow to adjust; in the long run, however, real magnitudes will be unaffected. Government spending increases also have a temporary stimulatory effect on output in the home country, but, for unchanged money supplies, tend to appreciate the exchange rate.

These conclusions are common to many macroeconomic models. However, MINIMOD also makes it possible to see how sensitive the results are to assumptions concerning expectations. It is shown that the paths of major macro-economic variables may be quite different in the two cases mentioned above. In particular, in response to a money supply change, the exchange rate is likely to overshoot its equilibrium value under consistent expectations, though not under adaptive expectations, and output effects are likely to be smaller under consistent expectations. Government expenditure changes seem to have more similar effects in the model under the two expectations assumptions, though the changes induced in the exchange rate and in long-term bond rates are larger with model-consistent expectations. It is also shown that in this case, credible, preannounced policy changes may have substantial effects before they are actually implemented: a future fiscal contraction may in fact have a stimulatory effect on output when it is announced, because of a decline in long-term interest rates and a depreciation of the currency.

Resumes

Le point sur les zones-objectifs—JACOB A. FRENKEL et MORRIS GOLDSTEIN (pages 633–73)

Les auteurs de la présente étude identifient les problèmes clés liés à la mesure dans laquelle il convient d’adopter des zones-objectifs pour les taux de change des principales monnaies et à leur applicabilité.

Quatre questions fondamentales sont soulevées, qui se rapportent à la définition des zones-objectifs et à leur raison d’être : premièrement, qu’entend-on généralement par système de taux de change à “zones-objectifs”, et comment définir les variantes “rigides” et “souples” de ce système? Deuxièmement, quels sont les défauts du système actuel de flottement dirigé des taux de change invoqués par les partisans des zones-objectifs? Troisièmement, comment ces zones-objectifs pourraient-elles éventuellement pallier les défauts du système actuel? Quatrièmement, quels sont les facteurs qui sous-tendent en grande partie le scepticisme et l’opposition auxquels se heurtent les zones-objectifs?

Sont également passés en revue une série de points pratiques et plus techniques qui conditionnent dans une large mesure l’applicabilité d’un système de zones-objectifs. Les points traités sont les suivants : comment calculer les zones objectifs? Quelles monnaies conviendrait-il d’inclure dans le système de zones-objectifs? Quelle devrait être la largeur de ces zones et avec quelle fréquence devraient-elles être révisées? A quels instruments de la politique économique aurait-on recours pour maintenir les taux de change effectifs à l’intérieur des zones-objectifs, et quelles en seraient les conséquences probables pour les autres objectifs poursuivis?

En conclusion, cette étude n’a pas pour but de plaider en faveur de l’adoption ou du rejet des zones-objectifs, mais de faire apparaître et d’examiner les éléments à prendre en considération lors de toute discussion approfondie du sujet.

Théorie des bénéfices et imposition des bénéfices—EDMUND S. PHELPS (pages 674–96)

Dans le système de finances publiques dit “classique”, les bénéfices des sociétés sont assujettis à une lourde imposition pour la raison qu’ils peuvent être transférés à l’Etat sans que ce transfert soit un anti-stimulant coûteux. Les sociétés, quant à elles, minimisent, au profit de leurs actionnaires, leur cotisation fiscale au titre de l’impôt sur les bénéfices en finançant par l’endettement la totalité de leurs investissements. Les bénéfices qui restent, le cas échéant, correspondent donc à l’excédent du revenu des investissements par rapport au coût de l’intérêt inhérent à des éléments monopolistiques. Dans le cas, toutefois, des économies ouvertes où les entreprises sont obligées de financer leurs investissements en faisant appel aux prises de participation au capital, la validité de cette théorie a été contestée.

La présente étude est consacrée aux effets qu’un impôt sur les bénéfices peut avoir sur le bien-être dans une petite économie ouverte où les entreprises opèrent sur des marchés dominés par les acheteurs, certes, mais où la concurrence ne les empêche toutefois pas de réaliser certains bénéfices. L’analyse aboutit à la conclusion que, jusqu’à un certain point, un impôt sur les bénéfices des sociétés a pour effet d’accroître le bien-être. Le principal avantage est que le transfert à l’Etat d’une fraction des bénéfices purs des sociétés entraîne un accroissement net des recettes fiscales totales; cet accroissement des recettes permet d’alléger les taux d’imposition marginaux sur le travail, ce qui a pour effet de rétrécir l’écart entre les salaires réels après impôts et la productivité marginale de la main-d’oeuvre, et, par conséquent, d’augmenter le volume du travail et celui de la production. Le principal coût vient de ce que les sorties de capitaux qui s’ensuivent réduisent les salaires avant impôts, ce qui finit par ramener le taux optimum de l’impôt sur les bénéfices à un peu moins de 100%. L’impôt sur les bénéfices présente certains avantages secondaires si la monnaie s’apprécie en termes réels ou si les actionnaires étrangers assument une partie de la charge qu’entraîne la redistribution de l’impôt sur les bénéfices; mais il présente, en revanche, certains coûts secondaires si, au lieu de s’apprécier, la monnaie se déprécie en termes réels. Même si une partie du financement prend la forme de participations au capital, il reste possible que les salaires après impôts augmentent, bien que cela ne soit pas certain.

Production et variations imprévues de la masse monétaire dans le contexte de produits intermédiaires importés et d’un rationnement des devises—AJAI CHOPRA et PETER MONTI EL (pages 697–721)

Les fluctuations à court terme de la croissance de la production réelle sont une préoccupation importante des dirigeants des pays en développement. Pourtant, malgré toute l’attention que cette question a reçue dans le contexte des pays industrialisés au cours des cinquante dernières années, les travaux analytiques, et plus encore les travaux empiriques, sur ce sujet, ont été étonnamment limités en ce qui concerne les pays en développement.

Dans les pays industrialisés, l’avènement de la nouvelle macroéconomie “classique” a ébranlé le consensus qui régnait quant aux causes des fluctuations à court terme de la production. L’exactitude des estimations d’équations de forme réduite de la production effectuées dans les modèles “classiques” a contribué à faire accepter ces derniers. Il est par conséquent naturel d’élargir le cadre de ces modèles à l’explication des fluctuations à court terme de la production dans les pays en développement, ce qui a été fait par plusieurs auteurs. Toutefois, des aspects structurels qui sont en principe particulièrement importants dans le contexte des pays en développement ont été généralement ignorés ou introduits de manière arbitraire.

La présente étude élabore un modèle structurel simple “classique” comprenant plusieurs aspects qui sont en principe importants dans les pays en développement. Le modèle est fondé sur une économie ouverte du type Mundell-Fleming, dans laquelle des produits intermédiaires sont importés et où l’on suppose que les devises sont rationnées. Une équation de forme réduite de la production est tirée de ce modèle, qui est une généralisation de l’économie fermée qui est son analogue. Une fois les propriétés du modèle analysées, l’équation de forme réduite de la production est estimée pour les Philippines. Les résultats empiriques correspondent de très près aux prévisions du modèle.

MINIMOD : spécification et résultats des simulations — RICHARD D. HAAS et PAUL R. MASSON (pages 722–67)

La présente étude décrit la construction et l’application d’un petit modèle macroéconomique, MINIMOD, des Etats-Unis et du groupe des principaux partenaires commerciaux industriels. L’objectif est d’obtenir un modèle facilement compréhensible, transparent et de taille suffisamment réduite, qui se prête à l’analyse de la politique économique. Par conséquent, un modèle théorique éclectique a été spécifié pour chacune des deux économies, avec des équations de la demande et de l’offre globales de biens et de l’accumulation de capital, et un traitement cohérent des flux de capitaux publics et privés. Le modèle a été spécifié de manière à avoir des propriétés souhaitables à long terme, y compris la neutralité de la monnaie et la limitation de l’accroissement de la dette publique par rapport à la production. Les valeurs des paramètres du modèle ont été tirées, à quelques exceptions près, des propriétés d’un plus grand modèle pour plusieurs pays; l’étude décrit la méthode qui permet de réduire un grand modèle à ses interactions essentielles au moyen de techniques de simulation partielle.

Le modèle est simulé dans le but d’évaluer les effets de modifications de la politique monétaire et de la politique de finances publiques sur la base de deux hypothèses possibles en ce qui concerne les anticipations relatives aux taux d’inflation futurs, aux taux des obligations à long terme et aux taux de change : i) les anticipations s’adaptent aux fluctuations antérieures des variables, ou il) les anticipations correspondent aux prévisions du modèle. Les simulations impliquent qu’une augmentation de la masse monétaire entraînera vraisemblablement une dépréciation du taux de change et stimulera la production intérieure, étant donné que les prix s’ajustent lentement; toutefois, à long terme, les grandeurs réelles ne changeront pas. Les augmentations des dépenses publiques ont aussi un effet stimulant temporaire sur la production dans le pays, mais, si les agrégats monétaires ne varient pas, elles auront tendance à entraîner une appréciation du taux de change.

Ces conclusions sont celles d’un grand nombre de modèles macroéconomiques. Toutefois, MINIMOD permet aussi de voir dans quelle mesure les hypothèses relatives aux anticipations influent sur les résultats. Il est démontré que l’évolution de variables macroéconomiques importantes peut être très differente dans les deux cas susmentionnés. En particulier, il est probable qu’en réaction à une variation de la masse monétaire, il y aura surajustement du taux de change par rapport à sa valeur d’équilibre dans l’hypothèse d’anticipations conformes aux prévisions du modèle, mais non pas dans l’hypothèse d’anticipations adaptatives, et i] est probable que les effets sur la production seront plus faibles dans l’hypothèse d’anticipations conformes aux prévisions. Les variations des dépenses publiques ont apparemment des effets plus semblables dans les deux hypothèses relatives aux anticipations, bien que les variations du taux de change et des taux des obligations à long terme qui en résultent soient plus fortes dans le cas des anticipations conformes aux prévisions du modèle. Il est aussi démontré que, dans ce cas, des changements de politique crédibles et annoncés à l’avance pourront avoir des effets considérables avant même que ces changements soient effectivement apportés : la perspective d’une compression des dépenses publiques peut, en fait, avoir un effet stimulant sur la production au moment où cette compression est annoncée, en raison d’une baisse des taux d’intérêt à long terme et d’une dépréciation de la monnaie.

Resumenes

Una guía para las zonas meta—JACOB A. FRENKEL y MORRIS GOLDSTEIN (páginas 633–73)

En este trabajo se identifican cuestiones clave en torno a la conveniencia y posibilidades de poner en práctica un régimen de “zonas meta” para los tipos de cambio de las principales monedas.

En relación con la definición y fundamento lógico de las zonas meta, se abordan cuatro cuestiones fundamentales: primera, qué significa generalmente adoptar para la gestión de los tipos de cambio un régimen de “zonas meta” y cómo podrían definirse sus versiones “estricta” y “flexible”; segunda, cuáles son las deficiencias observadas en el actual sistema de flotación dirigida de los tipos de cambio que hacen proponer la adopción de zonas meta; tercera, cómo podrían las zonas meta remediar dichas deficiencias, y, cuarta, qué factores se esconden tras gran parte del escepticismo y de la oposición que las zonas meta suscitan.

Además, el trabajo se ocupa de una serie de cuestiones operativas, de naturaleza más técnica y sumamente importantes para determinar si sería práctico implantar un enfoque semejante. Entre otros aspectos tratados cabe mencionar los siguientes: cómo se calcularían as zonas meta; qué monedas deberían incluirse en el sistema; cuál debería ser la amplitud de las zonas y cada cuánto habría que revisarlas, y qué instrumentos de política económica se utilizarían para mantener los tipos de cambio observados dentro de los límites de las zonas y con qué consecuencias sobre los otros objetivos de la política económica.

Con este trabajo no se pretende justificar ni la adopción ni el rechazo del sistema de las zonas meta, sino sólo identificar y analizar los factores que han de tenerse en cuenta en cualquier discusión seria del tema que nos ocupa.

Teoría y tributación de las utilidades—EDMUND s. PHELPS (páginas 674–96)

En el sistema fiscal “clásico”, las utilidades de las empresas pagan impuestos elevados porque se cree que el trasvase de utilidades hacia el Fisco no causa un desincentivo costoso. Las empresas, a su vez, intentan minimizar la obligación tributaria sobre sus utilidades en provecho de los accionistas, financiando la inversión mediante emisión de deuda. En consecuencia, las utilidades que resten, si acaso, equivalen ai excedente de renta de capital frente al costo en intereses causados por elementos monopolísticos. Ahora bien, se han expresado opiniones contrarias cuando se trata de economías abiertas y empresas que se ven obligadas a financiar su inversión con cargo al capital propio.

En este trabajo se examinan los efectos de bienestar social del impuesto sobre las utilidades en una economía abierta pequeña en la que las empresas operan en “mercados de clientes” mundiales, de manera que las utilidades no desaparecen totalmente por efecto de la competencia. La conclusión del análisis es que, hasta un determinado nivel, el impuesto a las utilidades de las empresas aumenta el bienestar social. La ventaja principal es que la recaudación de una parte de las utilidades puras de algunas empresas permite un incremento neto de la recaudación tributaria total, lo que alivia las tasas tributarias marginales sobre el trabajo, reduce la cuña entre el salario real -una vez deducidos los impuestos-y la productividad marginal del trabajo, lo que trae como consecuencia un aumento del trabajo realizado y del producto. La principal desventaja es que la consiguiente salida de capital reduce el salario antes de deducir impuestos, lo que eventualmente limita el tamaño óptimo de la tasa tributaria sobre las utilidades a algo menos del 100%. Se pueden añadir otras ventajas si se produce una apreciación real de la moneda, o si existen accionistas en el exterior sobre los que recae parte de la carga redistributiva del impuesto a las utilidades y, por otro lado, entraña un costo si tiene lugar una depreciación real del tipo de cambio. Con introducción de la financiación parcial con capital propio es posible, aunque no seguro, que se produzca un aumento de los salarios después de deducir los impuestos.

Producción y dinero no anticipado con bienes intermedios importados y racionamiento de las divisas—AJAI CHOPRA y PETER MONTIEL (páginas 697–721)

Las fluctuaciones a corto plazo del crecimiento del producto real constituyen una cuestión muy importante para los gobiernos de los países en desarrollo. Sin embargo, pese a la considerable atención que se ha prestado a este asunto en el contexto de los países industriales durante los últimos cincuenta años, es sorprendente el poco trabajo analítico-mucho menos empírico-que se ha efectuado en relación con los países en desarrollo.

En el caso de los países industriales, el surgimiento de la macroeconomía “neoclásica” ha alterado el consenso sobre las causas de las fluctuaciones a corto plazo del producto. El éxito de la estimación de las ecuaciones de forma reducida del producto generadas por los modelos “neoclásicos” ha contribuido a la aceptación de los mismos. Por consiguiente, es natural que se aplique este marco a la explicación de las fluctuaciones a corto plazo del producto en los países en desarrollo, y así lo han hecho varios autores. Sin embargo, normalmente se ha hecho caso omiso de los aspectos estructurales que probablemente jugarán un papel importante en el contexto de los países en desarrollo o se les ha incorporado de manera arbitraria.

En el presente estudio se elabora un modelo estructural “neoclásico” sencillo que incluye diversas características que probablemente resultarán importantes en los países en desarrollo. Se elabora un modelo de una economía abierta, en el marco de la teoría de Mundell-Flemming, se incluyen los bienes intermedios importados y se supone que las divisas están racionadas. A partir de este modelo se obtiene una ecuación del producto de forma reducida que es una generalización de la ecuación correspondiente en una economía cerrada. Tras analizar las propiedades del modelo, se estima la ecuación de forma reducida del producto para el caso de Filipinas. Los resultados empíricos se ajustan en gran medida a las predicciones del modelo.

MINIMOD: Especificación y resultados de simulación—RICHARD D. HAAS y PAUL R. MASSON (páginas 722–67)

En este estudio se describe la elaboración y la utilización de un modelo macroeconómico pequeño, MINIMOD, de Estados Unidos y de los países industriales con que éste mantiene intercambios comerciales de gran volumen. El objetivo es disponer de un modelo comprensible y transparente de dimensión manejable que se pueda utilizar con fines de análisis de política. Por consiguiente, se ha especificado un modelo teórico ecléctico para cada una de las dos economías, que contiene ecuaciones de demanda y de oferta agregadas de productos y de acumulación de capital. Se han tratado de manera coherente las corrientes de fondos del sector público y del sector privado. El modelo presenta características adecuadas para el análisis a largo plazo, incluidas la neutralidad del dinero y la propiedad de que la deuda pública no puede aumentar sin límite en relación con el producto. Los valores correspondientes a los parámetros del modelo se obtuvieron, salvo en algunos casos, a partir de las propiedades de un modelo más amplio de varios países; en el estudio se describe la metodología que consiste en reducir un modelo más amplio a sus interacciones básicas empleando técnicas de simulación parcial.

Se ha simulado este modelo con el fin de determinar los efectos de las variaciones de las políticas monetaria y fiscal basándose en dos supuestos alternativos concernientes a las expectativas de la tasa de inflación futura, de los tipos de interés de los bonos a largo plazo y del tipo de cambio: i) las expectativas se adaptan a las fluctuaciones anteriores de las variables o ii) las expectativas son compatibles con las predicciones del modelo. Las simulaciones implican que un aumento de la oferta monetaria probablemente provoca una depreciación del tipo de cambio y estimulará la producción en el país de referencia dado que los precios se ajustan con lentitud; sin embargo, a largo plazo no resultarán afectadas las magnitudes reales. Los aumentos del gasto público también constituyen un estímulo temporal para la producción en el país de referencia aunque, sí no varía la oferta monetaria, tenderá a apreciarse el tipo de cambio.

Estas conclusiones son comunes a muchos modelos macroeconómicos. No obstante, con el modelo MINIMOD también se puede medir la sensibilidad de los resultados ante los supuestos relativos a las expectativas. Se demuestra que la trayectoria de las principales variables macroeconómicas puede ser muy diferente en los dos casos antes mencionados. En especial, como consecuencia de una variación de la oferta monetaria, lo más probable es que el tipo de cambio supere su valor de equilibrio si las expectativas son coherentes; esto no sucede si las expectativas son adaptables; los efectos en la producción probablemente serán menores en el caso de las expectativas coherentes. Los efectos de la variación del gasto público en el modelo son más parecidos en los dos supuestos de expectativas, aunque las variaciones inducidas en el tipo de cambio y en el tipo de interés de los bonos a largo plazo son mayores cuando las expectativas son coherentes. También se señala que, en este caso, las modificaciones de política dignas de crédito y anunciadas con anticipación pueden tener efectos considerables antes de ser aplicadas; la perspectiva de una contracción fiscal puede, de hecho, tener un efecto estimulante al anunciarse, debido a que provoca un descenso del tipo de interés a largo plazo y una depreciación de la moneda.

In statistical matter (except in the résumés and resúmenes) throughout this issue,

Components of tables may not add to totals shown because of rounding.