Journal of Monetary Economics I (1975) 369-373.
8 North-Holland
Publishing Company
THE RELATIVE MPORTANCE OF AUTONOMOUS EXPENDITURES AND MONEY IN TURKEY H. Sonmez ATESOGLU* Jfiddic East Technical University,
Ankara, Turkey
1. Introduction In the early 196Os, Friedman and Meiselman (1963) (hereafter referred to as FM) made statistical comparisons of simple forms of Keynesian and quantity theory models. Their results purported to show that money, rather than autonomous expenditures, is a better explanatory variable for induced expenditures. ’ This note provides some empirical evidence from Turkey on this issue over the period 1950-1968. As may be the case with most of the developing countries, agricultural income arises, along with money or autonomous expenditures, as an additional crlicial variable in determining economic changes within the Turkish context. This is due to the dependence of agricultural income primarily on weather conditions, rather than on aggregate demand, as well as its large share in the GNP.2 The following models incorporate this characteristic of the Turkish economy. Throughout this note all variables are stated in nominal values. 2. Alternative models To avoid spurious correlation due to the strong upward trend in the data, the Keynesian and quantity theory alternatives are stated in first difference form:
*The author is an Assistant Professor of Economics. He IS grateful to the members of his dissertation committee, Professors Martin Bronfcnbrenner, Jacob Cshcn, Jrrom? Wells. Asatoshi hla.eshiro and Michael Spiro, for their guidance and assistance in the preparation of the dissertation (presented at the University of Pittsburgh) on which this paper is based. The author is also thankful for the comments of an unknown referee of the Journal. ‘FM’s result< have been challenged by various ccono?lists, including Ando and Modigliani (1965), DePrano and Mayer (1965), Hester (196-l), and. more recently, by Geijel(1970), showing that this issue is still far from settled. ‘Agricultural income on the average accounted for more than one third of the GNP in Turkey over the period 1950-1965. The weather conditions are recognized as the primary cause of changes in agricultural income by the Turkish Prime hlinistry (1971. pp. 1-75, 333-333). This view is also supported by OECD economic surveys of the Turkish economy. See, for example, OECD (1969, p. 7).
370
H.S. Aresogh, Aulonon1or4s expenditures arid money in Turkey
where AE, is rhe change in induced expenditures in period t, AA, is the change in autonomous expenditures in period t, AZ, is the change in agricultural income in period t, and BM: is the change in money supply in period t. /I, , yl , /3z and yz are regression
4 =
I+G+(X-I,,,)+F,
4 = Z,+G+(X--I,)+F, E, = Y-At
= (C+B)+D,
E, = Y-A,
= (C+B),
where A, and A, represent autonomo,tis expenditures, El :ind E, induced expenditures, I net investment, G government consumption, Xexports, I,, imports, F net factor income from abroad, Y total expenditures, C private consumption, B change in inventory, D depreciation, and IBgross investment. ’ Theoretically, I is more appropriate than Is as a component of autoromous expenditures, because D is business expenses which do not induce chai-\c; in income. However, some economists object to the use of I on statistical grolzlds, and find I9 more appropriate. They claim that the subtraction of D from I, makes 1 erroneous because of the errors of measurement in LX4For these reasons, a net as well as a gross autonomous expenditures concept has been selected. Even though A,, A,, E, and E, are irligeneral conformity with the theoretical definitions used in modern textbooks, operationally they may be far from ideal. Some parts of C and B are likely to be independent of changes in current income. Likewise, parts of I and I9 and G may be induced by changes in current 31n Turkish national income accounts, factor income from abroad is not included in exports and imports, but is given as a jcparate item. Personal consumption expenditures and change in inventories are not giver as separate items, but their sum is given which is obtained as a residual. See United Nations (1964, pp. 228-231). 4See, for example, DePrano and Mayer (1965, p. 737) and Hester (1964, p. 20).
H.S. Atesoglu, Autoilomoas expenditures and money in Turkey
371
income. But because of the unavailability of data on D and B, all of (C+B) has been considered induced; and, because of the difficulty of separating induced parts of I and le and G, all of these are considered autonomous. Another difficulty is finding an economic rationale for the inclusion of D in E, . Because of the use of IRnd A 1, the D had to be included in E, , since the sum of A 1 and E, must be equal to Y. However, the quantitative importance of the inclusion of D in E, is not likely to be large, since the value of D is very small relative to the sum of the other components of El . The (X-I,,,) is taken as autonomous, since exports are dependent on foreign demand, and imports are determined by government import quotas, in accordance with its foreign exchange holdings. The F, especially in the years following 1965, reflects the earnings of workers in foreign countries. Therefore this item is also considered autonomous. The monetary concepts selected are: M, = currency in circulation plus sight deposits: and M2 = AI, plus time deposits. Sight deposits are the closest substitute for currency in Turkey. Therefore, they are included in As,. ’ The agricultural income concept used is the farming income aggregate found in the Turkish national income accounts, 4. The empirical tests and conclusions The parameters of eqs. (l)-(3) are estimated by the ordinary least-squares method.6 The statistical results for eqs. (1) and (2) are given in table 1. It shows that the coefficients of all explanatory variables, including agricultural income, have positive signs, and that they are all statistically signiscant. These results demonstrate that agricultural income is a crucial variable, along with autonomous expenditures or mane}‘. Comparisons of the standard error of estimates of the Kej nesian equstions with those of the corresponding quantity theory equations show that the latter have a slightly better ft. Similarly, the corresponding tratios indicate that money performs a little bit better than autonomous espcnditures. In contrast to FM (1963), these results are very close and lead to the conclusion that autonomous expenditures and money perform eoually IveIl as an explanatory variable for induced expenditures. These results also imply that the roles c-f autonomous expenditures and money are not mutually exclusive, and that their interdependence is a&uslly due to the strong posit’ve correlation bet\scen th:se variab1c.s as presented 11~table 2. The regression results for eq. (3) are gikc:! in table 2. It is seen that the coefficient% of all the explanatory variables are positive, and, except for the “Friedman (1968, p. 444) suggests that currency alone is likely to be a more appropriate concept for developing countries. The results obtained by using currency alone were vastly inferior to those obtained by using M, and Mz, and were clearly against the quantity theory. Therefore these results are not reported here. (‘The d;:ta used are taken from the State lnstitutic of Swtistics (1968, 1969) and Turkish Central Bank (1970). G
H..$. Ateso&, Autonomous expenditwes and ntoncy in Twke)
372
Table 1 Regression
results for eqs. (1) and (2) (annual
data: 1950-1968,
in million Turkish liras).” _- .__._ ---
.-
-~ -
Explanatory variables Dependent variable
AA,,
A.421
A!bf;, AZ, RZ S.E. _.__ _ ~_______ _.. _-.- ____..._. ..~_ _
AM;,
1.78Mb (7.41)
A&,
1.6934b (8.21)
AE,,
1.3155b (8.36)
AEI, l.4013b (7.56)
dE2, AErr
1.5453b (8.04)
AEz,
1.2007b (8.19) __ .-._.._---__.
1.2033b 0.9567 (6.59) 1. 2906b 0.963 1 (8.11) 1.3412b 0.9642 (8.79) 1.1606b 0.9605 (6.99) 1.2629b 0.9641 (8.52) 1.3090b 0.9652 (9.21) -~--~-.-.
1162.56 1072.54 1057.26 1049.09 999.54 984.95
aFigures in parentheses are r-statistics, S.E. is the standard error of estimate. blndicates significance at the 5 percent level. Table 2 Correlations between first differences of autonomous expenditures and money (,annual data: 1950-1968).
l’lA,J.~l; -- 0.s9-t “AA ,.l.MI = 0.896
Regression results for eq. (3) (aiillual
Dependent variable
__ Jrlr,
Explanatory variables ----.__-Ll,I\1;, -1.42, L.l.W;, -
AE,,
0.6458
3E,,
(1 .OJ) 0.6117 (1.0’)
AE,, -lEzr
Table 3 data: ic)SO-1968, in millio,1 Turki+
1 1339L (1:9x, 0.9060” (2.09, 0.5955 (1.25) 3.5666 (l.?l)
0.93P (l.SO)
aIndica~es significance at the 5 percent level.
0.7563a (1 .‘I?)
-AZ, 1.23123 (7.31) 1.X78” (7.51) 1.1915a (7.61) 1.2”7”
(7.7s)
liras).
R1
S.IJ.
0.9656
IO!,0..W
0 0665
1056.22
0.9675
982.S
0.9693
970.8 1
H.S. Atesoglu, Arrtonornous expenditures arrd money in Turkey
373
coefficients of autonomous expenditures, they are all statistically significant. The comparisons of the t-ratios in each equation reveal that agricultural income is clearly more important than either money or autonomous expenditures, again demonstrating that agricultural income is a crucial variable. In contrast to FIM, the t-ratios of autonomous expenditures and money in each equation are very close and lead to the conclusion that these variables are roughly of equal importance. The findings discussed above corroborate those of critics of the FM approach for the U.S., and highlight the central role played by :,gricultural income in countries like Turkey. They also suggest that an approach which does not neglect autonomous expenditures, money, or agricultural income, is likely to be more useful for interpreting short-term economic changes in Turkey. References Ando, A. and F. Modigliani, 1965, The relative stability of monetary velocity and the investment multiplier, American Economic Review 55, Sept., 693-728. DePrano, M. and T. Mayer, 1965, Tests of the relative importance of autonomous exr,