WAGNER'S “LAW” AND THE DEVELOPING COUNTRIES nck DIAMOND INTRODUCTION ‘Limovatt THeRE is widespread agrooment in developing counties that the government’s role is erucial for development, there appears Tittle con- Sensus about the optimal level of public intervention in the esenomy. More than a century ago, Adolph Wegner, one of the leading German economists of the time, propounded an interesting development thesis. Loosely framed, he proposed that as a nation develops its public sector (and consequently public spending) will grow in relative importance! Wagner’s “law” of overincreasing sate expansion was derived {rom the historical experience of continental Europe, principally Germany, at the eatly stages of industilizaion. From spective, Wagner saw three factors which would cause state activity to. grow proportionately faster than other sectors of the economy. First, he projected an expansion of the governments traditional role in providing administration, Taw and order as the economy became more specialized and social and economic life more atomized as a consequence ofthe increased division of labor. Secondly, the foresaw an increase in the provision of “cultural and wellare” expenditures, ‘ost particularly education. His reasons for this expectation were not allogether clear, although it may do him litle injustice to say he thought they behaved 48 superior goods with an income elasticity of demand greater than unity. ‘Thirdly, he saw that the increasing scale of technologically efficient production would cause the government to undertake certain economic services of which the private sector would be no longer capable. Ia this he had in mind the heavy investments associated with railroad. construction, ‘Surprisingly, when other advanced counties were examined and when a longer time period was taken, Wagner’s thesis seemed to be borne out. Subsequent empirical research using time series data has amassed considerable evidence to support the contention that the relative size of the public sector hes increased ‘over time, in almost all of the currently advanced countries* But what of the LDCs? Obviously, time series data for such economies is limited in availablity and quality. What is available, however, is a body of cross-section data for countries at diflerent stages of development. What light does such evidence throw on Wagner's “Iaw"? Can we explain the relative share of the public sector in an economy by its level or type of development? What are the major 5 The most easly accessable source & found ia 26) For 2 more dase account see (36) [351A est ovew of bis thei fe sonsned In (4) 20. # See for example, (27 13) (5) C19) (8) (2 28} 30} (2) 6 738 THE DEVELOPING EcoSOMIES causel influences, and how can we account for geographical differences? What fare the implications of this evidence for development policy? This article at- tempts a preliminary exploration of such questions. I. THE EVIDENCE FOR WAGNER'S “LAW" Teally one would prefer to examine the same country at diferent levels of development and examine changes in the relative size of the public sector over time, Unfortunately, this approach encounters 2 number of practical problems. First and foremost, data are simply aot avalable for many developing countries to cover an adequate period of their economic dovelopmeat® Using available statistics would thus tend to bias the survey to the most developed of the LDCs ‘who tend to have better dats, In the absence of adequate time series we are forced to rely on cross-section data, and look st the relationship between some indicator of economic development and an indicator of the relative size of the public sector at @ particular point in time, In this more static context difer- fences in income levels may be regarded 2s a proxy for differences in the degree ‘of development since this indicator is likely to be closely associated with the ‘complex differences in economic, socal, and (sometimes) political structure that characterizes development. Unfortunately, in constructing such a test the re- searcher faces several other dificult statistical as well as conceptual problems. ‘in devising a statistical indicator of the relative sae of the public sector, as reflected in the public accounts, the usual procedure is to take some ratio of public spending 10 national income, G/Y. However, there are some doubts bout which statistical measure of public spending or income to employ. The major doubt about the numerator is whether transfer payments should be in cluded. Some would argue that if one concentrates on the public sector's role fs a consumer of resources then transfers should be excluded. However, there js no doubt that the distributive function of the government is an important source of public sector intervention in the economy. Further, such transfers fare usually financed by taxes, and as such are subject to the same kind of fiseal decision process as that involving the consumption of resources. More dificult data problems are posed by the different levels of government and the ‘vrying constitutional structure of counties, Although the definition of public spending should be as comprehensive as possible, including local governments (or in the federal case, rgional governments) as well as public agencies such fs social security funds, there aze obvious dita limitations involved in such 4 strategy. Given this lack of data and the wide differences in government structure, the definition adopted in this study excludes that fraction of local government expenditure which is Gnanced by revenue raised by local govern- sient itself ‘As for the denominator of the G/Y ratio, several options are available. For example, should we concentrate on GNP of GDP, and exclude net factor 4 Ror an exception, se (2), although his empl cocoons are Hime by the shortness ofthe tne period coveredwacwen’s “Law? 39 ‘comes from abroad? And should the chosen national income aggregate be ‘measured at market prices or factor cost? For the LDCs dae to the importance ‘of foreign ownership of factors of production, and in some sectors the employ- ment of nonnationals, perhaps income to nationals is tho more relevant agere- gate. However, since governments have the power to tax incomes, and given the “openness” of many LDCs, it has been the convention to concentrate on ‘ross domestic product rather than GNP. In valuing GDP it seems more logically ‘consistent fo measure income at market prices rather than at factor cost, since [goverment purchases are made at market prices. The substraction of indirect taxes (minus subsidies) from GNP would involve some doubiful assumptions as ‘to the shiftabilty of these taxes. Also, since available government expenditure data is typically measured gross of deprecation of the public stock, it seems consistent to choose a measure of national production which is also gross of capital depreciation. "The most widely used indicator of development is that of per capita income. Leaving aside for the moment a discussion of its adequacy as an indicator of evelopment, there are a number of statistical problems javolved in using this in international comparisons. There is no need to reiterate in great detail the ‘many empirical problems encountered. For example, national income estimates of diferent countries measured in domestic units of currency have to be con- verted into a single vurrency by use of exchange rates. This proves inadequate with exchange rate instability, exchange controls and multiple exchange rates, Moreover, foreign exchange rates fend (0 reflect the relative prices of those goods and services entering foreign trade, and are not typical of relative prices within countries, On the whole, the level of income of low income countries tend to be understated relative to high income countries* Apart from these special problems of international comparison, there are, of course, the problems encountered when using national income statistics which ate also encountered in time series studies. For example, the problem of choice of weights or prices in which output is to be measured will vary between countries. Differences in taste, need, technology, and quality, also present conceptual dificultes. Despite these numerous problems, for practical purposes a choice of development indi- cator has to be made. Although intemational comparisons of income levels are ‘undeniably suspect, there is no guarantee another indiator wil be less reliable. Tn any case, employing per capita income as a proxy index of development also appears valid, at least asa fest approximation, since the many socioeconomic variables associated with development are likely t0 be highly correlated with ‘his variable. "The above statistical dificulies may explain why, while the evidence from time series is almost wholly airmative, cross-sectional studies have led to con- ficting conclusions. Some appear to support Wagner's “law” “Thus when we we current expenditure and fotal revenue as measures of the public share in ross national product, thee is @ definite positive correlation between per capita income and the government share” [40, p.49}. Or, exprested differently: “Gov- 4A concision lo reached by (13)40 [THE DEVELOPING EcoNDMIES frument expenditures tend to rise at a faster rate than national product a8 per capita national product increases” [35, p. 23]. However, other studies appear to negate Wagner's “lav.” For example, Musgrave when examining the ratio of current public expenditures to GNP notes that the postive relationship with per capita income disappears if countries ate divided into high and low income groups and “break down for the low (per capita income below U.S:$300) and high (per capita income above US.$600) groups taken by themselves” (25, 1.120). A similar conclusion was reached by two other important studies [16] [24], In part, as has been suggested by V.P. Gandhi, this later result could be caused by combining two heterogeneous samples ‘of developed and less developed countries [13]. For our purposes, we propote to concentrate on 1 comprehensive sample of forty-one developing countries. However, since there js no one simple “average” developing country or group of LDCs, and since their individual problems are so profuse and their structural characteristics so Aiversifed, it would also seem better to look more closely at not too heterogeneous sroups of countries rather than finding useful common denominators among, all the LDCs. For this reason, we experimented with smaller ubsamples based on ‘geographical proximity. ‘The data used are derived from local sources and published by the World Bank, where every care has been taken 10 correct some of the more obvious Aiicalties of inter-country data and provide a consistent series? This set of per capita income igure, as in the case for all such data, is far from perfect, but may well be the best available. Public expenditures aze defined to include ‘expenditure of central goverament, states and provinces, municipalities and cites. It also includes spending by other agencies than the government if they collect taxes or are financed by government subsidies (the most important of these being the social security agencies). ‘The data difers from those used by previous Investigators since not only do we take » more comprehensive sample of do- veloping countries but we use averages for a mumber of years which reduces the influence of anomalies caused by unrepresentatve years. Averages for the 1961-69 period were used since the period of the 1960s was generally a period of “relative peace" without severe depression ot boom. Later years are more likely to be disturbed by the recent ‘world recession and oil crisis, and also mote prone to updating errors. The scatter diggram showing the relationship between the G/Y ratio and per capita income is shown in Figure 1, It is evident that there is no simple relationship between per capita income and the share of public spending in GNP. The forty-one country model was divided into three. geographical subgroups for comparison: Arica, Asia, and South and Central America. Upon examination of individual scatter diagrams, ‘there was every indication thatthe ft was not much improved by disaggregation sand a linear funetion was used for simple regression. analysis to verily. this, ‘As a further check of the result, 2 double logurithmic function of the form Jog G/Y=log a-+b log (Y/N), Was employed but this did not alter the con- A ted account of these aise and ther sours is contained is the Appendix 0\waaxn’s “Law” a Fie 1, The Relaonsbip between Per Capt Income and the Ratio of Tota Puc Spending 10 GDP (Average for 1961-69) Coury Codes AL Altes (©. Cental andSouth Amaia: 1" note Si. Avgetion 2 Eat 52, elie B xeora 32 Baal © Ler 20. ia 36 Colom 5. Malay 21, Indonesia 35. Cale & Mateo 2. Japan 36, Bator 3. Martie 2h Malas 37. Semaicn 8 Morocco 24 Pakisen 38 Medco 8. Nigeria 25, Phliprines 33, Pama 10, Soma 26, Singapore 40. Paraguay M1 Soden 27, $ Kores 4 Veneta 12, Tanzaaia 28, Se Lanta 1 Tunisie 23, Thailand 1 Upne So. Tatas 16 Zambia Closions. ‘The results using this functional form are shown in Table I. Henceforth, for both simple and multiple regressions the double Togarthm form was used, this having the advantage thatthe coellients measure the partial elasticities of ‘the dependent variable with respect to the independent variables, ‘Equation 1 in Table T shows that there is litle relationship between the ratio of total public spending to GDP and per capita income, and this conclusion is Tittle altered whether one takes the total sample or divides the sample into2 ‘THE DEVELOPING ECONOMIES ‘TABLE 1 “Tw ReLsvion nerWHEN PUnLIC EXYENBETORE RHO AND Pe CADE INCOME Fee Equaion 7 Equaloa 7 sump — Sagye AEGIBEa iganeinetla Yoo. InGCM=Inas bars WS) Conon 6 RE Conant 1A at -6 22 Om ool fem 0.0889 0.08, 8) 068) 695) To, 2 Alten 16 142.684 “0.02% 0.005 0.3953. 0.4028 0.22 Ea) 0 28 83618) 0513) 3. Asa 1 302.234 9.06m 0.008 1 e544 1057 0.02 GBI 368 Bh 0.418) 4, Soutané Central 141227857 O.Jom 0.08 L's D-H 0.04 ine acum) ais) OSI "Note Figures im parentbeaes are watson regional groups. In all cases, the R? is negligible and not only is the slope fnsignfiant but changes in sign between samples, On the whole, these results support V.P. Gandhi's contention that while using a combination of developed and developing countries itis possible to get a positive relationship between the expenditure ratio and income level, this is merely due to the dilference in average levels st the two ends of the scale, When one conceatrates on the developed or the developing countries individually, no frm relationship can be discerned (a conclusion also reached by [20] [25D. Further experimentation with diferent samples of countries (for example, ivided by income level into high, intermediate, and low) offered no improve- ‘ment in the results. However, using different agarogates of public spending ‘when calculating the expenditure ratio did slter these negative conclusions. Tn particular, when total civilian expenditures were caken (Le, total spending minus pending for defence purposes), there was some improvement in the overall ft and the slope coefficient was almost significant at the 10 per cent level (ee ‘equation 2 in Table 1. When the total sample was broken down by regional fsroupe a consistent postive relationship was established and surprisingly a rea- Sonable ft was obtained for the African countries (R°=0.22 and slope significant atthe 5 per cent level). However, the other twe regions showed marked dlfer- fences in slope and exhibited a poor overall fit, Given thatthe African countries fon average are at the lowest level of development, this result could suggest that ‘when the influence of war is excluded from consideration, then, Wagner's “law” holds for countries at the early stages of development when the economy is in the intial stages of industializtion. It should not be forgotten that Wagner was generalizing from Germany's transition from a rural-agricutural economy to an wan-industral one. Hl, DETERMINANTS OF PUBLIC SECTOR EXPANSION From the crost-sectional evidence presented above, the existence of Wagner's “law” seems dubious. However, this could be due to inadequacies in our testwacwen’s “LAW” “a procedure. One obvious problem arises from the cross-section approach. It requires a number of restrictive assumptions to test Wagner's thesis which is ‘a dynamic “law” describing changes over time within a couttry with cross-section fevidence which compares diferences in levels between countries at a point of time. Unfortunately, it is dificult to adopt a time-series approach due to the non-availability of dats. Secondly, these inconclusive results may refleet the fact that per capita income is not a good indicator of development. After all, the process of development involves structural changes within a country which is rellected in @ multidimensional fashion—not only in its economic but in is demographic, social, and politcal characteristics. Our results may merely con- firm the necessity of inquiring more carefully into other ealtural and economic dimensions that are not adequately reflected in simple figures of per capita fncome, but which doubtless affect the share of government in the economy. "Tas, we must face the fact that Wagner's “law” is not simply one of eco- nomics, After all, Wagner framed his “law” in general terms, encompassing institutional ehanges, industralizaton, democratization, ete. Of course, being in the natore of a sweeping generalization, the very breadth with which itis framed makes his thesis very dificult to test statistically. Thus, even if good statistical “fe” was obtained between public sector size and the level of per apita income, this isnot t0 say that per capite income could be the sole expla nation of budgetary policy. Other influences of a demographic, social, and political nature are sure to be important, but may be diffizult to separate from {he ceonomie variable. ‘Thus, e major disadvantage of trying to derive an empiti- tal relationship between some measure of G/Y and per capita income is the ‘bvious limitation in trying to interpret the relationship in eausal terms. Apart from the disadvantages of using pet capita income as an indicator of develop- ment, the fact remains that all other possible causal influences are likely to be highly correlated with this variable, ‘Thus any significant relationship between GIY and income per capita may merely relect the joint infuence of other causal ‘variables, Furthermore, even if the causal nexus between public sector expansion tnd per capita income is taken as direc, we still face a critical problem. If we take the view thatthe relative size of the public sector is predominantly demand Getermined, the influence of pet capita income is that derived from demand theory. However, obviously we have an underidenified relationship Income could just as well determined public sector expansion from the supply side since itis likely to be a prime determinant of tax revenues. For all these reasons itis as well to look at other possible determinants of public sector size. Two broed approaches are possible: one which views the ‘se in public spending as a response to demand influences, the other which views the availability of finance as crucial, Wagner seemed to have no doubts that the development of the public sector was primarily demand-determined, merely reflecting the underlying changes in the structure and stage of economic develop- ‘ment. For him, public expenditures were the principal determinant of the level ff revenues, for "in the long run the desire for development of a progressive This possity is explored friar in (41“4 ‘THE DEVELOPING ECONOMIES people will always overcome these financial difficulties” (26, p 16], ‘Thus Wagner hhad no doubt about causation: demand for public services is the propelling force fn determining the level of finance to be raised, rather than the availablity of revenues determining spending. However, it should never be forgotten that Wagner was generalizing from a particular histrical situation. For the devel- coping counties of today, it i difieult to accept his optimism about the availe bili of finance ‘The altemative proposition of taxed budgetary expansion has received much attention recently, and there has been a growing body of research on the deter- ‘minants of tx revenues in LDCs, For example, Lotz and Morss have maintained that for developing countries it is necessary to focus on the tax side because expenditures are held far below their optimal level by administrative bottlenocks associated with the mobilization of domestic resources [22]. In analyzing the eterminants of tax revenues use has been made of per capita income, again as 8 proxy for the level of economic development. Of course, it i not unreasonable to suppose taxable capacity would be greater the higher the per capita ineome because a smaller proportion of total income would be required for subsistence needs and more would be available for other purpose, including taxation [29] However, apart from the general level of prosperity, other variables have been employed to explain the growth in tax revenues, For example, in the course of evelopment, there is the observed tendency for the proportion of tax revenues raised by direct taxes to grow. These tend to be the taxes with @ broad enough ‘base and high enough income elasticity to adequately nance expenditure growing faster than GNP. Thus the structure of taxation, most partielaely the ratio of slirect to indirect taxes, can be considered a determinant of the total revenues raised, “Another characteristic ofthe tax structure in LDCs has also received emphasis Restrictions imposed by other development objectives and the narrowness of the base of some of the more income elastic direct taxes, has meant that for ‘most LDCs indirect taxes comprise the highest proportion of tax revenues, In ‘urn, taxes on foreign trade tend to form a high proportion of indirect taxes, ‘not merely because of administrative ease of collection but levying import duties oes not usually present any great political problem because of their hidden nature (24). Consequently, it has been argued that a major determinant of tax revenues in LDCs has been the degree of “opensess” of the economy as meas- tured, say, by the proportion of trade to national income. Other institutional characteristics are algo lable to affect the collection of tax revenue. For example, the degree to which the subsistence sector dominates the economy, or the degree ‘of monetization of the economy, wil affect the possibilities of levying taxes and the administrative ease in collecting them. Of couse, the relative tize of the foreign trade sector in a developing economy is aso likely to be highly cor- related with the degree of monetization in the eootomy (lor example, the impor- tance of cash crops rather than subsistence agricalture) and of the importance ff production units more amenable to taxation (such a6 large, often foreign, extractive operations).\WaaNER’s “LAW” 4s Apart from purely economic explanations, unigue istorieal circumstances have always been used 0 supplement purely economic explanations. Other institutional characteristics which have been mentioned as influencing the reve- rues raised is the administrative eficiency of the tax system whieh in tum is felt to be influenced by the county's colonial heritage. Tn particular, using data for the 19608 Richard Thorn postulated that an important influence on the size ‘of the public budget was due to the maintenance of Britsh colonial norms of expenditure and taxation (35). Hinrichs when looking at che tax structures of ‘2 wider range of countries talks of the “caltural syle” of their tax systems accent cither direct or indirect taxation. He diferentiates between two pre- dominating tax systems! the "Northwest European system” imbedded in the English-speaking world and stressing. dite taxation; and the Mediterranean system, imbedded in the Latin world, inclined towards iedirect taxation [171 Tn complete contrast to explanations of public sector expansion which rely ‘on the availabilty of revenues, there are those which stess demand factors ‘An obvious demand influence which has been singled out for study is that of ‘demographic characteristics. The importance of population size has long been appresated, hence income and public expenditures have been deflated into per capita terms for comparative purposes. However, for many of the LDCs, apart from its size the rapidity of increate, the age sirucnire, and the geographical cconcenteation of population have all been mentioned as sossible explanations fof the relative growth in the public sector. For example, Goffman and Mahar ‘conser the age structure of the population to have been an important factor in public expenditure growth in six Caribbean countries during the postwar period [14]. High growth rates have the effect of shifting the population com- positon in favor of youth, thus putting increased demands on the public sector fn such areas at eddcation. The consequences of urbanization have also been stressed in various studies [10] [35] (40) ‘Taking a demand interpretation of expenditure growth has led several writers to emphasize changes in various community needs as development progreses. For example, expenditure on education has displayed a particularly rapid growth. Some writers explain this by increasing technological requirements demanded of the Tabor force (31, Appendix E7], others by change in social values and individual preferences (25, p. 85]. Interpretation depends on whether spending fon education is regarded” #3 consumption or investment, and because of this there may be marked differences between the developed and developing countries. Another rapidly growing component of public expenditure has been in the area fof health and social services. Again this has been the subject of a mumber of interpretations. Some sce this development as a consequence of the change in cconomic, social, and politcal organization requiring greater state protection of the individual; others as a consequence of a change in ideolony with a substitution ‘of collective for individual responsibilty (1] [24], For example, Williamson ‘would argue that along with urbanization has gone the submergence of the informal security ofthe village and extended family and the emergence of formal state security [40], ‘Thosn has suggested that the growing political strength of46 ‘THE DBVELOPING ECONOMIES turban workers as capable of exacting higher per capita public socal expenditures than the more dispersed less politcal powerful rural population [35]. In cone trast, Andie and Veverka see the crucial change in economic organization as consisting of the secular decline in the size of the consumption unit, so that “as economic growth tends to reduce its size and dissolve many collective o- ganizations interposed between the consumption unit and the State, this leads to 2 general demand on the public authorities to protect the economic status of the individual members of the community” (1, p. 219) Parallel to this argument, again stemming fom Wagner’s seminal work, several ‘writers have proposed that as society develops, the cause and consequence of a greater division of labor, the concomitant inereate in the complexity of social relationships generates increasing socal friction, Musgrave suggests that duc to this increasing interdependence, externalities have increased and with them the need for greater social control {25, p.79]. ‘The requirement for greater ropu- lation, law and administration, and the provision and maintenance of stich Services and institutions would be manifested by increased expenditure” Farther, the profound impact of industrialization and tecknologial chenge on the struc- ture of the economy and its social orgenization implies atleast an indirect impact fon the growth of public spending. For example, it has been suggested that modern technology has inereased the efcient scale of production, not only in private industry but pethaps even in services like those provided by the public sector [15]. Also indusrialzation affects the structure of production in an feonomy such that as the economy develops the grestcr division and regional integration creates demand on the service secor to provide this increased interconnectedness. ‘From this survey itis apparent that there are many possible explanations for the growing share of the public sector in national income. For expository con- venience we have separated those explanations which stress the possibilities of raising revenues and the ease of administration from those which concentrate on the consequences of industrialization, urbanization, specialization, and income changes in creating demands for increased public spending. It should be recog nized, however, that these factors, even in combination, are unlikely to yield 1 total explanation for the observed differences between ‘counties in the share ‘of government in their national income. It i impossible to ignore the differences in the ideological stance of the country’s leadership. Thus, while we have con- cxntrated on those “structural” factors which work towards a larger publie sector as a consequence of development, the end result most likely to be determined by ideological commitments. Martin and Lewis, for example, have argued that it is not the level of development which is the prime determinant but rather nation’s prevailing conception of the role ofthe state [24]. Likewise, Musgrave Points out: “Low income counties today do not operate under the same tech- nical, political, and value conditions as prevailed in the past when the now * While recogiing this argument, Willamsoa doubts wheter administrative and sini sconome expendiures asin percentage of national podoct do ise ay laste social enaitres (40, p20.waawen’s “Law” a developed countries were st similar Jow levels of income, Attitudes toward growth, changed communication, the demonstration effect of aluence and wel- fare measures taken abroad, the conflict of political ideologies, all make for differences inthe historical setting” (25, p.72]. Given this admittedly restricting (qualification, an attempt was made to empirically investigate the relative impor- tance of the above “structural” factors in explaining inter-country dliferences in expenditure-income ratios. For this purpose, data was collected on twelve variables for each of the LDCs of the sample, which are listed in the Appendix: Ill, DETERMINANTS OF STATE EXPANSION: SOME EMPIRICAL, ‘EVIDENCE In economic analysis, and especially in international comparison, the constraint placed on the method and results by the nature of the data is a serious problem. "The job of statistical collection and comparison is so fraught with difficulties that no one should take these figures to be precise reflections of reality but more as rough approximations. For example, it should slvays be remembered that we are dealing with average figures forthe 1960s and methods and mean- ings both change over ime and difer between countries. However, such ad= mittedy rough statistics have their uses in generalizing about the eal world when significant common influences are indicated in the regressicn analysis for many countries. For this reason we have included some elementary statistical tests as fan aid in interpreting our regression equations and for deciding on the order of ‘magnitude of the reliability of our tentative generalizations. Table IK summarizes these results. Basically, three diferent experiments were atiempted using data for all LDCs, and then separately for our three regional groupings. First, only “demand” vatiables were used to try and explain the differences in expenditure ratios between countries (equations 1-4); then only “supply” variables were employed (equstions 5-8); lastly, combinations of both types of variables were used (equations 9-12). Following precedential traditions, we chose to include per capita income as @ demand variable, but this was only introduced as a last ‘TABLE IL Deremmnaiers oF Tae Pustic Exenvorrune Rasio: EMeRIC\ RESULTS |A. Demand Infuences eau Tadependent Varies sample. ————__—_!aepenéet Varnes __ tien " GN POR Dror «UONAN‘Ype 1. Al 1,380 —Oo1st «(OSA 0.0595" 0.1725" 0.17 {ait wat GtSoa Sat" 885) 2 Afies 2766) Ose 025 Osi Gow ome GS my Oso 3H A Asie Loakse issn gauss 28g nets 0.53, 3018) 860) GNSS (amy ema 4 South and 3.2086 oem ola Ogle ee ost SEtaE Knees Gta “estan see (te) S640)48 THE DEVELOPING ECONOMIES B. Supply Intuences Teependet Vartan AEM supe “Serene Yaris ____ ig = oN FY TR co aM =A 3507, 2. 20 CHR Basse Baier O.8 Gath “eas ees) dr Gr Si 6 Aes oan assure .0igs o.gmse —O.as 03% OS, BHR OMS, NS, 7 Asin 44416 0-060" 0,55" -0.0005 0.285% 0.0479" 0.60 SA OSES ANS Whe, ate 2 4 Sou end 7m 0392 O1MKs 246 OTe 0s Chnval America 0.982) (O30) (19K) (tte) 728) Combined Intaenes Taaeyeadeat quan Sample —_tederendeat Varieties cw o aun ° A 205 oa aa ae 0.89) 20s 1 Aten 0.179 =o ay oa Be) o.063) (0:35; mAs 2am ore oz S30 36) 595 12 South and Cental 1 59 susie 0.2514 Smctics aio, 230) (38 i sumne Independent Variables © my DIR aM «Dp a 201% oa — om oe oS aan O88) 1 Aton ome aisle oi gigsee a OS Oh aio Me Aa 02 gage er Os} ahs 12, Southard Cental 9.007 = Oe. ‘meres (05) 3.850) % Sgicant at 10 percent level, Sigicant at 5 per cet level 1 Wrong sin resort when it afforded a greater degree of explanatory power than other de- ‘mand variables. "The first important feature of these empicical results is the degree to which the total sample of LDCs requires disaggregation in order to derive meaningful results. In all cases, whether one examines demand or supply factors, or @ com bination of both, the degree of overall it is improved by breaking down the total sample by geographical groupings. Furher, in many cases where the same variables remain significant between regions, the magnitude of the coeMicients are substantially diferent, again stressing the divesied experience of LDCs a‘WAGNER'S “LAW” 9 4 group, We can thus conclude that in research of this Kind, an important problem still remains to be faced in deciding on the optimal scheme of dis- aggregation fo employ in empirical research. Although it is imeresting, as here, to compare the results by geographical groups given the possiblity of intra repional demonstration effects, it mast be admitted that many other schemes of disaggregation are possible. sto the specific influences on expenditure ratios, notwithstanding the emphasis they have received in the literature, demand factors did not offer a comprehen sive explanation of inter-country expenditure ratios. Perhaps the most notable failure was the role of demographic influences: for all LDCs as a group, and for individual regions, the growth rate of population and the size of the depend cent population was never significant at the 5 per cent level. This finding could suggest that there is some revenue constraint in those counties where the need for public spending per capita rises slower than in those countries where demo- graphic pressures are less intense, In contrast, the variable measuring the degree of urbanization was always significant atleast at the 10 per cent level, its infu fence being strongest for Asia. Our indicator of the speed of industalization (the growth in manufacturing) also appeared to exert a significant ‘nfluence, and Seems to confirm the increased pressure for public intervention in the process of industrialzaton. As can be seen from equation 1, per capita income appears a significant variable for the total sample of LDCs, but examination by region revealed that thi was primarily due to the strong association in Asian countries (equation 3). Supply influences seemed to offer a better explanation of expenditure ratios in the African and Asian regions. Remembering that these regions contain coun- tees with very low income levels, cis would seem to suggest that for the poorer developing repions the ability to aise finance is erucial in determining the level ‘of public spending. As can be seen in equation 8, for the richer countries of South and Central America supply factors by themselves aflorded litle expla- nation of the diferences in expenditure ratio, it being easier to isolate the influence cof demand factors, Given the lengthy discussion of the degree of “openness” of an economy as 4 prime determinant of revenues, the relative unimportance of this variable in explaining expenditure ratios seems remarkable, Only in the Asian region is it possible to detect some jafluence for this variable, In contrast, the growth in ‘means of payment variable, defined to include money in cireulation as well as quasi-money, has a marked influence on the expenditure ratios in Aftica and Asis, However, although the generic nature ofthis variable makes interpretation dificult, it may be folt to represent the importance, for example, of eash crops, and other production activity organized on a wage basis and hence amenable (o taxation. As was previously suggested, this variable i likely to be highly coreelated with the dogree of “openness” of the economy and a simple corre- lation test did indeed reveal a positive association. One may speculate, therefore, that the relative failure of the depree of “openness” variable may in part be caused by including the monetization variable in the same equation which eap-50 {THE DEVELOPING ECONOMIES ‘tures ther joint influence. However, there is no reaton to rejet the proposition ‘that the degree of monetization rather than the “openness” of an economy is the more important influence on expenditure ratios, ‘As for the other supply variables, the ratio of direct taxes to total revenues ‘which is taken as an indicator of the dogree of elatcty of the tax system seemed 1 important influence in African countries, agnin supporting the idea that in ‘these countries spending is supply constrained. The degree of centralization of ‘budgetary decisions offers litle explanation of diferences in expenditure ratios, Which would lead one to conclude that this fastor is not a good index of the ‘government’ sbilty to aise revenues and increase spending in felation to income. ‘The influence of colonial heritage gives confictng results between regions. Te is most significant for the Asian region, confirming the conclusion reached by ‘Thorn that countries which were previously Briish colonies tend to have higher spending ratios. Again, however, this result stresses the danger of generalization ‘rom a heterogeneous ‘sample: while this variable also appears significant for the total sample, this is almost certainly due to the individual influence of the ‘Asian countries. ‘As a final attempt to try and improve the overall level of “ft,” supply and demind variables were combined and in each case per eapite income was itro= ‘duced to see if its influence could bo detected, On the whole, the overall regression results were not much improved by the income variable, as is evident from the size of the R*. For Africa, however, per capita income did appeae significant, and confirms the result presented in Table I, equation 2. Surprisingly, this variable also added to the explanation of the variation in expenditure ratio in South and Central American countries. For each region the relative contti= bution of supply and demand factors was reemphasized: for Africa, the impor- tance of the elasticity of the tax system was aotable; for Asia, the degree of “openness” and colonial heritage remained sigiiieant; for South and Cental ‘America, demand factors dominated. ‘What then of the general tenor of Wagner's “law” as a thesis of demand-led budgetary expansion? Our results suggest his besc presumption that the relative size of the public sector is determined by “Sstru:tual” factors without any con straints from the revenue sie is not universally vali for the currently developing ‘countries, Although one should not be lured by the false precision of the regres~ sion results, their order of magnitude does not contradict the idea that supply inluences predominate in the poorer (pareularly Aftican) countries and demand influences are more important in the richer countries (eg, of South and Central ‘America). Certainly it should be stressed that epart from the “structural” iaflu- ences examined here, the spending policy in each country will be infiuenced by political preferences and cizcumstances which are likely to account for a great eal of the “unexplained” variation in expenditure ratios between counties. IV, STATE EXPANSION AND ECONOMIC GROWTH For the advanced countries, public spending and the relative size of the govern[WAGNER'S “LAW” st ‘ment sector have expanded at an increasing rate for many years. This growth ‘was based on an underlying philosophy which contended that greater direct overnment activity was the best way, if not the only way, to achieve certain ‘economic and social goals. Tn recent years, many countries aave questioned the {ality ofthis philosophy. Not only bas there been growing skepticism about the achievements of increased public spending, but also some have wondered ‘whether undesirable “side effects" have seriously undermined the desirability of such policies, In the LDCs, the task of reorganizing the economic structure of their economies and promoting faster growth has led to programs of higher and higher public spending and greater state intervention, As yet the full implications of these developments has aot recived careful appraisal. For this reason it is interesting to reverse Wagner's chain of causation. Instead of enquiring into the impact of development on the expenditure ratio, Tet us examine the question ‘whether the relative size of public sector has had a significant impact on the rate of development, oF the type of development experience. Tn the literature it is posible to detect two opposing views about the impact ‘of public sector expansion on economic growth. There is the opinion, which finds eurreney in the majority of LDCs, that the growth rate will be raised. On the other hand, there is the argument often expressed in the advanced counties, ‘that the growth rate will be slowed down. Let us examine the fist postion, Teas almost become axiomatic in the discussion of developing countries that ‘government intervention is one of the most important factors promoting economic trowth, It is frequently argued that the state has played av indispensable role in providing various forms of economic and social overhead capital which has contributed to growth, even when this capital has long gone underutilized. The {reat importance of public policies for the introduction of advanced technology ‘and the training of unskilled labor forces ie dificult to dispute. In many of these countries, a major part of public spending ig viewed ss simply another input of specialized services in 2 generally technologically determined production funetion® Parenthtically, it is worth remembering a rather dated controversy among early designers of social accounting frameworks concerning the question ‘whether public expenditure represented an intermediate or a final output. Tn this controversy, Kuznets took the former position and argued that “national income is a measure ofthe net output of economic activity within the given social framework, aot of what it would be in a hypothetical absence of the latter. The maintenance and modifications of this framework, even though it employs scarce resources that ean be secured on business markets, cannot in itself con stitute part ofthe final product of economic activity...” [19]. Kuznets implied, theretore, that @ lange part of public spending can be regarded as an input t0 the economic system. But can all public spending be seen in this light, as positively aflecting economic productivity? Although for the advanced counties Where there is greater emphesis on public spending of a “consumption” nature social security, socal expenditures of all kinds astociated with the maintenance of the welfare state—for the LDCs, distinctions between economic and social ora stlet appli ofthis lew te (27)2 ‘THE DEVELOPING ECONOMIES polices, or investment and consumption spending, become blurred. A govern: ‘ment health program is an instance of “social golicy,” but its impact on actual ‘or potential economic growth of society may be far-reaching. Similarly transfer payments in 2 society where poverty and maleutrition are prevalent may have 8 considerable impact on productivity. But ever in advanced countries as Shoup has noted: “most government activity is a producers’ good rather than a con sumets’ good since it reduces the cost of doing business” (34, p. 494], In complete contrast, to this view, there is a large body of opinion, especially in the more advanced countries, that argues that increased public spending has bbeen deleterious fo economic growth. In general, wo dillereat types of argument hhave been forwarded, There is the view that the public sector lags behind other sectors in productivity. For example, many commentators see a possible expla- nation for the rising government share in nation income as eaused by diflerential productivity: that, the labor-intensive pubic sector being relatively less productive land less amenable to technological ianovation implies a rising proportion of resources must flow into the public sector in order to maintain constant publi services per unit of growing output [2]. The indirect consequence of @ growing public sector is thus to deprive high productivity sectors of resources. Hence stating the argument diffeendy, tis implies the greater the share of resources absorbed by the public sector, the lower will be the aggregate level of produc tivity in the ecosomy and growth will suffer. A second type of argument con- ‘centrates on the need to finance public spending increasing faster than income, ‘which may have important disincentive effects. This argument has been asso- ciated with the “crtiealtimits” hypothesis of Colin Clark who maintains that increasing the tax burden, especially at high levels of taxation, discourages productive effort on the part both of labor and capitalists [8} (9]. Further, taxation tends to Iead to inefficiency and rising costs as industrialists discover that out of any increase in costs a larger propertion will in fact be paid by the public sector in reduced tax burden. However, such arguments rely heavily on ‘the importance of direct taxes in the tax system, and taxes falling on the produc- tive members of society—assumptions which may not be applicable in many LDCs. Given these conflicting hypotheses concerning the impact of public sector ‘expansion on economic growth, it seems important to examine the empirical ‘evidence for each viewpoint, A$ a first experiment, the average growth rates of the diferent LDCs for the 1960s were employed as & dependent variable and regressed on the ratio of total public spending to GDP. For the total sample 4 strong negative relationship was discovered, a can be seen from equation Al and Table IIL. Given the natare af our inquiry, one should disregard the size of the R? as of secondary importance and concantrate on the significance of the slope of our regression line. As indicated by the £ statistic in parenthesis, this is significant atthe S per cont level. However, this significant relationship breaks ‘down for our geographical groupings (equations A2-A4), which may throw doubt on the usefulness of our scheme for disageegating the total sample. At ‘the same time, the sign is always negative, never postive, which suggeste that33 ‘TABLE I ‘Ta Inrtueves oF Posie Expsnorroms ow Fcononte Grown AL lniiyain ab iG) Equation Sanpie © 2 1 All a 2.2570 8s) 2 Abies 6 2 De am 3 Ade a 22s aa, 4 Sout Cena America M3. 18 BL InGxyymin a8 Igor gation Sample N Al a Alsen 16 3 Asie “ 4. South and Central Ameria UL 98, vo.sizy rey a8, 6.86 ‘oa, 05318 © in=in ad wor “Egoation ‘Sale W at a 2 Abies 6 Ale 4 4. South and Cental Ameria 1 Ped 0-200) 223 sor) 2.812. aah 20m sa) D. Inia + XG) Equation ‘Sane w 2 all a haa 235i) fren er ort Aaa got 4. Southand Central America 1 aB0h 08 e 88 aioe -a.2mi6 3810) 02589 Gast ~o.am ioe, > Cec) Zan ag 3 ow. 5st {.om0 7 oo to 28 (es) 1459 49) 2.0035 atta, aa aS, “Leo was) or i) = 39 a) 0.8 0.0 ou D8 0.05 0.05 oatsé "THE DEVELOPING ECONOMIES there is indeed @ tendency for the growth rate in an economy to be Inversely ‘elated to the size of the public sector. Of couse, there is always the possibility in correlation analysis that the direction of causation can be reversed: a low ‘growth rate in an economy may indicate the nesd for greater public intervention to raise it and consequently hieh expenditure ratio. However, this being the fase and remembering that our data depict averages for a number of years, this negative association would lead one to conclude that such public intervention hhad not been successful. “To investigate this possiblity, it is necessary to reexamine the reasons for postulating @ negative relationship between economic growth and the extent of the public sector in the economy. Tt will be remembered that two main argu ments have been advanced for this effec: a “productivity Iag” on the part of the public sector and the disincentive effets on the rest of the economy as ‘consequence of financing an increased expenditure ratio. Obviously, these hypotheses are dificult to test empirically. The productivity tag hypothesis presents particularly intractable problems given the nature of most public serv- {ees which are not sold on markets, which cannet easy be priced and the derived Denes are difiealt to assign individually, Due to this dificulty in defining units of ouiput in the public sector, there aro few reliable measures of public sector productivity changes and consequently the beliet that the public sector's tech nology is relatively unprogressive has not been empirically verified even in the ftraint on such testing, As for disincentive effets, itis interesting to investigate Wether the relative size of the public sector has exerted any influence on broad ategores of economic activity which one might expect to be erucial to economic growth. "To this end, three further repression equations were estimated. Given the fmportance of the foreign exchange eap # a constraint on the growth of many LDCs, in equation BI of Table HI the ratio of exports to national income was regressed on the expenditure ratio. For all developing counties as a. group 4 positive relationship was established which was significant at the 10 per cent Ievel. However, on disaggregation, it was discovered that this result was almost tentiely due to the effect of the more advanced South and Central Americen land Asian countries while for the poorer Aisican countries the relationship fppeared negative. One might offer the tentative suggestion that this retlests the priority given to export promotion in the former regions. A similar regres- ‘Sion equation was then estimated fo show the relationship between the expenditure fatio and the ratio of domestic investment to GNP. The results displayed by fcquation C Table TIT are inconelusive, For all LDCs as a group there is a non ‘Significant postive relationship, Disaggreyation revealed that for all geographical {groups the relalioiship remained positive, but with exception of the African ounries this was not significant at the 10 per cent level. OF course, itis not Surprising that one should obtsin this postive relationship since for most LDCs public investment forms a large part of total domestic investment. Our result * Howove, se (61WaoweR’s “LAW” 55 ‘may well hide the fact that public investment has been at the expense of, or substitute for, private investment, To investigate the effects on the private sector, private sector savings (comprising the saving of private corporations, unincorpo- rated enterprises, households and private nonprolit institutions) was taken as the ‘dependent variable and a strong negative relationship with respect to expenditure ratio was discovered (see equation D), Disaggregation revealed the relationship tvas negative for all geographical groups but especially strong for the Asian and South and Central American couatries, Inevitably, interpretation of these results js speculative, However, one might hazard the guess that the latter regions contain the more advanced of the developing countries which have more highly ‘developed tax systems with greater emphasis on direct tsxation and face a higher potential for disincentive effects, ‘On the whole, our results are not conclusive nor would one expect them to be given the obvious limitations of the data employed, However, remembering the divergent hypotheses encountered in the literature, some survey of the em= pirical evidence seems desirable. For our sample of forty-one LDCs, these findings would lead one to place emphasis on a negative rather than postive relationship between the relative sizeof the public sector in the economy and the rate of economic growth, We would also beled to speculate that this negative relationship was brought about by the discentive effects on domestic savings and 4 possible “productivity lag” inthe public sector, rather tha in any discourage- ment fo agzrepate investment or exports. Obviously, this has been a preliminary fanelysis and much more work necds to be undertaken. For some repions the fits were quite poor (eg. Arica) and this may indicate a need for further dis- aggregation (eg, into countries north and south of the Satara) or the adoption of a diflereat scheme of disaggregation (¢., grouping by income level). Another fobvious direction for future research isto inguire into the relationship of diferent functional categories of public expenditure with economic growth. However, in this paper our concern has been to illuminate some of the implications of Wagner's “law” for economic development and development policy, hence we have concentrated on aggregate public spending. CONCLUSIONS In this analysis we have attempted to delineate empirical regularities and single cout potentially interesting relationships concerning the relative size of the public sector which have tended to be discussed in an descriptive fashion, Tt is hoped the dificltes encountered in trying to reconcile previously maintained specific hypotheses with the discovered empirical relationships wil be of use to sub- sequent researchers. Unfortunately, such analysis should be considered no more than a preliminary “ground-elering” operation. Inevitably, in making this at- tempt, we continually come up against the great disadvantage of grand specu- Tations like Wagner's “law”: the impossibility of subjectng them to precise cmpitical testing, Essentially, Wagner's is not really a “law” or a theory, but rather a philosophizing about development based on the underlying idea that56 “THE DEVELOPING BeONOMIES ‘this process is fundamentally similar in diferent countries at dliferent historical periods. Form our empirical results, there is every indication that such a pre- sumption is unjustifed—not only are there marked discrepancies between regions ‘but also within regions which account for the poorness of “ft in the individual repression equations. ‘Thus, while itis possible to isolate influences which are ‘common for all countries, especially within regions, at the same time one cannot ‘ignore the obvious fact that the spending policy in each country is influenced by political preferences and ideological comimitwents which cannot be expinined by our empirical analysis. Because of this Wagner’s thesis is too “deterministic” to afford an adequate explanation of the relative size of the public sector in currenlly developing countries. ‘While there is widespread agreement that the government's role is important, if not critical, in development, there has been litle analysis of the impact of increased public spending on the overall performance of developing economies. In this paper we have considered the empirical evidence concerning the impact of a rising public expenditure share on economic growth for forty-one developing countries. On the whole, the relationship was discovered to be negative. Une fortunately, data limitations prevented us from a rigorous investigation of the reasons for ths inverse relationship. Our atempss to relate the expenditure ratio to certain magnitudes. generally accepted as crucial to economic growth were inevitably crude and hence causal interpretation was of necessity speculative. ‘Accepting these quaifcation, the picture presented was that on average devel- ‘opment policies while increasing the relative scale of public intervention in the economy had maintained if not increased the lovel of exports and aggregate investment. However, the need to increase taxation to finance this spending ‘ay have had some disincentive effects on privae saving, specially in the richer countries. At the same time, the probable relative productivity backwardness of the public sector may also have contributed te a slowing down of economic growth. Certainly the inverse relationship between public spending and economic growth in the developing economies is food for thought, questioning as it does the conventional wisdom of the necessity of increased public intervention to stimulato growth. Unfortunately, at best these results merely indicate empirical sociation and to have confidence in our cause interpretation necessitates do- tailed understanding about the mechanism by which public spending, operates ‘on the economy. Unfortunately, the theory of public expenditure groweh is sill in its infancy and, itis not too pessimistic to conclude given the nature of the subject matter, itis destined to remain a fascinating although elusive problem, [REFERENCES 1. Anno, $, and Vovines, J. "The Growth of Goveramest Expenditure in Germany floc the Uniention” Flnarzarch, n= Vol. 23, No. 3 Oanvay 1969). 2 Bowol, W.J. "Macroscosomis of Unbalanced’ Grove,” American’ Economie. Re- view, Vol 57, No. 3 Gane 196. 3, Ro, RING The Growth of Goverment Spending in Canada Toronto: Canadian Tax Foundatin, 1970,u w. Ey 1. 2, m, 2, \WaaNeR’s “LAW” 1 ‘Wagers ‘Law’ of Bapanding State Activity Puble Finance, Vol. 26, No.1 lanuary 1970 ‘towoat, G. "The Growth of Pubic Bxzeniure fa Ieland"Scondnavian Economic Hisor) Review, Vol. 8, No.2 use 1969). Boout, BH and Guevear, D. “"Wagne?’s ‘Law and the Growth of State and Lose Governnect” mols of Relonal Scince, Vol. 1, No. 2 Jue 1963. Bhasrowo, DP; Mats, RA: and Ousss, W.E.” “The Rising Cost of Local Public ‘Sic Soe Bndence and Retestions” Navional Tax Jounal, Va. 2, No, 2 Gone 196). Gane, © "Public Finance and Changes inthe Valve of More,” Eeonomle Journal, Vol. 85, No. 4 (December 1949, SESE exmanship, Hober Paper No. 26 (London: Tutte of Heonomic At fais, 196, Drvreca, “Socal Mobilition im Foie! Development” American Poa! Seence Revlon, WO. 8, No. 3 September 1961). Bu, K, Goverment Pca detviy and Economie Growth tx Jepan, 1869-1960, (To- jet Kinkniya Bookstore Co, 1963) Enownze, C. “Sttare of Pubic Expendires Ia Selected Developing Counts: A ‘Time Series Sty" Manchester Sehol, Vol. #, No. 4 Deseaber 1973). Gavoun, VP. “Wagners ‘Law’ of Public Expenditure Do. Reet CrosSetion Sie Condom 12" Public Finance, Vol. 26, No.1 Ganuary 197 Gorman, 1, and ass, Do. "The Growth of Pubic Expenditures in Seloted Developing Nsioa: Sk Carkeen Couetrs, 1940-1965" Pull Finance, Vol. 26, Nov} Gasuary 1970. ‘Gurra, SP. “Pubic Pspendare and Economie Growth: A Time Seles Ansys” Pale Pnance, Vol 22, No.2 (May 1967). ‘Pubic. Expenditure” ead Economie Development—A Cross-Section ‘Ronis Finararchiy, a. Vol. 28, No. 1 (October 1968, Rinoniens, Hoth General Theory of Tax Stuer Chane daring Economic Dex Yelopment (Cambridse, Mas: Harvard Law Schoo), 1968) Hoor, "The Expuaion of the Pubic Sesoe—A Study of the Development of ‘ubtl Cian Expeoitres im Sweden Going the Year 1912-1956” Pube Finance, Wal. i, No.4 (tober 1952). cus, S: “On the Valation of Socal acom,” Economica ns. Vo. 15, Nos 1 and 2 (February and May 194. Lat, 8. 2A Note.on Goverament Expendtues ia Deteloplag Counties,” Economic Jura, Vol. 79, No. 3 Gane 1969). wis Sim “Government Revere from Foreign Tradet Aa Tateratonal Com: parlon Mencheser Schoo, Val 38, No. 2 Guse 1970). Fore 1 Rand Mons, Fi “Mesiting “Tox Bilort in Developing Coustie,” IMF Stop Popts Vol. 1 No. 2 Ouse 1967. Maun Di, and Rezavot, FA. “Tho Growth end Pater of Pubic Rxpendiae In Breal 1520-1963," mimsogaphed (Rio se Tansco: Iatiuto de Plana eonomico e Social, 1979. Marr, A.M and Laws, W-A, “Patuens of Public Reweoe and Expenditure” Manchenor Sool, Vo. 24, No. 3 (September 1956) Musou RA. Flee! Sysioms (Now Haven: Yale Uniersiy Pres 1969, Murenave, RcAy and Puscoee, AVE, oie Clair inthe Theory of Public Plnance ondone Machin, 1958. Nowe, T, sPatteras of Government Copal Formation in the Economie Development Ue Tapa, 18781967 in Ete ta Honours of Ural Hicks el, W-L- Div (Lanion: Mackin, 1973) (ODenoenvn, My end Tar, ALA. The Growth of Public Resente and Expendiare bx elon? (Dublia: Insite of Puble Adminstration, 1968.58 ‘THE DEVELOPING ECONOMIES 29, Osancs, HLT. “Share of Goverment in Gross Natonal Product for Vasious Count” imerican Economie Review, Va. 47, No. 3 ane 195), Ba, Pesoocx, A.T, and Wistoan, J. The Growih of Public Bspendiare i the Unite inom, cov. ed (Lando: Alen and Unwin, 1967). 31. Pavon FL. Pubile-Expendlars ta Communit end Capit Nations (London: Alen 2d Unt, 156 82, Kiscaronn, BMV. Public Expenditures in Ausra (Duebam: Duke University Press, 1939) 53. Rom ALR, "The Government Revenve Share Ja Poorer Attisin Covatise—A Com neni” Beonomie Journal, Vl. 98, No.3 use 1565. 34 Stour, C.8, Publ Finance (Cicgo” Aldine Pushing Co, 196). 35, Thoms, RS, "Tho Broluton of Patio Fomsce: doing Esoomie Development—A Cos Sestion Analyse" Fnaearhiy, Vol 27, 36, Tinos, H. "Das Gust der Wacken Stasaugaben” Finansarchiv, ns, Va. 21, Nov Ganuaey 1961), ST, Tusane, A.D, abd Hao, JA. “Longoa Growth of Nondefense Goverment Expendiores {2 te Unied Sates” Publle Phance Quarerly, Vol. 2, Now 2 (Apa i, 38 Waowen, A. Crunlegng der polischen Okononi, Sd cd. (Lapis: CX. Wine, 185, 39, —"__. rinanawsenchat ed ed Lage, 1890) 40, Wanuoson, 1.6. "Pablle Expenditures and Reve: An Yteceation Cnmparn,” Morchestr Sobol, Vol 29, Newt Gamary 1961. 41, Wists, J, sod Duniont, J."*Commeat oa th Lospua Growth of Nondefense Governmeot “Expenditures fo tie United State” Puble Finance Quarter. Vol. No.4 (Ocober 1975). 42, Vovnts, J. "The Growh of Government Fependture inthe Unlted Kingdom since 1910" Scorish Journal of Poiteal Beonomy, Vel. 10, No, 1 Gantary 1963, ‘APPENDIX THE DATA AND SOURCES Vee Destinton Some Y pie. Per capt gros domestic product ODP tad wherever omible A,B, C.D ‘on estinates coast mart pices Wher soc evtiates were ‘ot avila ales of Contant factor cost were wed Por. Population BB G—Totalspending of seerl government Wherelformationon azar A ovemment was at vale onal goverameat data Ud. (Gc Tota pbc spending mins defence spend. Defence spending A, G ‘acude intra! sect. aPoP. Growth rae of population, Aa werage of he rites inctuding F ind ao terminal years D.POP. Dependent population, deed 13 thse Yelow cigeen yeas of | F Sgr and above aby years _of ape, Derived from Tabor force tistics wid woempoyed persons and und tay workersWAGNER'S “LAW” Degree of urbanization a indcated by the percentage of population Ting i urban arest The defiition of turban populaton is not Uniform forall counsl, However, the data_do_peovde ah i ‘Seaton ofthe conottrailon of sleriy no-reral population AMAW. Growth rate of manufacturing industry. These have generally er prin a xe ‘Som computed onthe basi of the country ines of anata fring prodocton published bythe U.N. Stale OMe or te nicl series of ie varoot countries Where sich indi=s Srere not avallabe, ase hasbeen made of ince compled by he US. Aseny of Interasonal Development or of the vale added st constant pies. Exort plus imports asa percentage of GDP. Based on current US dollar values defined fo exlode factor and wens payments to sed from abroad Direct axe a8 8 percentage of total government revenue, Direc. farce comprise all tanes and ertanes levied a2 8 chage oo the income of hotacolds and plate nonprofit institons corporate Income and excess pots face; tee on ensued pots oF fm capital stock whch ae levied at regular ital Stare of central government im total curteat revemst For the purposes of comping these ratios, ceial goverment current {auton to and foo non-centat government ageniesand to and from the rst ofthe world Growth la the means of payment. Defined to include the money ‘ony toney i ection otride te basking system. and de Imand depos held by the tongovermmest sector) and guste Ihoney fim, svings depons, ee Bold by the nongovernment ‘Sui Grow sale are average based on domes ciency wales ‘omy variable indesting colonial Reriage, =I for former Belish Colony? 0 fr ars ‘Growh rate of GDP. These are average compounéed rates of (gow between inl and terminal years for mot countrles ‘oven te period 1961-68, Ratio of exports to GNP. GNP measured at mavket prces and ‘Exports converted at cotent US. liar vabes Domestic lavestment, bated on domes currency vues af ete en torket price Average period Agures ae catated 13 ‘Smple means or individual. yrs with change in inventories Srauded, Private sector savings, comprising the saving of privte corpo thon unicorprald enerrve, and ousebols and private noa- oft nations 59 ABH Ac ‘Surses! As UN. National Accomm Yearbooks. W: WN. Monthly Bulletin of Sai ‘©: IBRD Country Reports D: OECD pablcaions; B: IMP Financial Stati FUN, Demogrphic Yearbooks; Q: UN, Statice! Yearboots, 1:IMP Balance of Payments Yearbooks