Formula allocation: The regional allocation of budgetary funds for measures of active labour market policy in Germany

Transcription

1 Uwe Blen Insttute for Employment Research (IAB) (Insttut fuer Arbetsmarkt und Berufsforschung der Bundesanstalt fuer Arbet) uwe.blen@ab.de Formula allocaton: The regonal allocaton of budgetary funds for measures of actve labour market polcy n Germany Paper prepared for presentaton at the European Regonal Scence Assocaton (ERSA) 2003 Conference n Jyväskylä (Fnland) Summary: A common problem of regonal polcy s the dstrbuton of funds to regonal unts. To follow the specfc purpose of ths budgetng process n a ratonal way ths s commonly done n a formula allocaton. In the paper ths s shown wth the example of funds for actve labour market polcy. In Germany, measures of actve labour market polcy e.g. tranng measures and job creaton schemes are pad from a common budget. For the allocaton of these budgetary funds (amountng to 13 bllon every year) to the regons of the Federal Republc of Germany, a formula was developed whch was to be based essentally on a labour market ndcator. The crtera for the development of a formula allocaton were: most accurate ft to the legal gudelnes of the Socal Code, transparency of the procedure, openness for necessary polcy decsons, scentfc correctness n mplementaton, effcency of the whole process. Here the procedure used n constructng the dstrbuton process s explaned and the dstrbuton result s set out. Both was fnally passed by the Executve Board of the Federal Employment Servces n Germany. 1 Karen Scott-Leutertz s thanked for very valuable support. Thanks also go to F. Hrschenauer (IAB), H. Rudolph (IAB), M. Wedenbeck (ZUMA) and M. Zaengle (Unversty of Regensburg) for advce regardng methodology. The basc structure of the allocaton scheme was developed n a workng group of D. Blaschke, C. Brnkmann, W. Karr and the author of the paper (all IAB). The development was done n co-operaton wth the admnstraton of the Federal Employment Servces. D. Hebmueller and M. Musat s thanked for the effectveness of ths co-operaton. All responsblty for ths paper and the ncluded analyss remans wth the author. 1

2 1. Introducton Ths paper treats a problem that s of a famlar nature n many contexts (see the artcles n the Journal of Offcal Statstcs no 18/3, e.g. Downes, Pogue 2002): The dstrbuton of a fxed budgetary funds to a number of regonal unts. To follow the specfc purpose of ths budgetng process n a ratonal way ths s commonly done n a formula allocaton. The specfc structure of the formula depends on the purpose, but there s the addtonal requrement to transfer ths nto a formal structure that s correctly specfed. The crtera for the development of the method are: most accurate ft to the am of the polcy and to general legal gudelnes, transparency of the procedure, openness for necessary polcy decsons, scentfc correctness n mplementaton, effcency of the whole process. It s necessary to open the scentfc dscusson over the degree to whch the formula allocatons mplemented n the polcy process corresponds to these requrements. The example we concentrate on s the dstrbuton of money for the purposes of labour market polcy n Germany, whch amounts to a sum of about 13 bllon every year. A specal method was developed n 1997/8 and has been used every year snce Durng ths tme the method has changed slghtly, but because the experences wth ts applcaton has been promsng, the method s descrbed here. In 2004 the allocaton scheme wll be assessed agan by the Board of the Federal Employment Servces n Germany (Blen 1998). The results of a scentfc debate about the ratonalty of the allocaton process can be used n a revson of the method. Snce there has been only one budget ttle for the man nstruments of labour market polcy, the so-called ntegraton ttle. It s left to the employment offces to decde how to dstrbute the funds among the ndvdual types of measure. Ths s an mportant element of the regonal responsblty n the control of labour market polcy, whch the legslator now prefers. In the Socal Code IV the crtera accordng to whch the funds are to be dstrbuted are already stpulated relatvely precsely. In 71b(2) t says: "When allocatng the funds, n partcular the regonal development of employment, the demand for labour, the type and extent of unemployment as well as the partcular expendture development n the precedng fnancal year are to be taken nto account. (talcs added by the author). 2. Basc decsons regardng the allocaton process The regonal unts the allocaton process refers to are the 182 admnstratve unts of the sngle employment offces (labour market regons, Arbetsamtsbezrke ). But the dstrbuton s done n three steps not drectly n one. In the frst step the budget s splt between western and eastern Germany. Snce n the East the labour market stuaton s stll very bad and especally very dfferent form the one n the West a fxed proporton of currently 48,5 % of the budget s allocated to the regons of eastern Germany far more than the proporton of the populaton. Then, n the next step the budget s dstrbuted among 12 (8 n the East, 4 n the West) large regons (the Landesarbetsamtsbezrke ), separately n the East and n the West. Ths s done by the formula descrbed n the followng. Fnally, by applyng the same procedure as n the second step, the budget s allocated to sngle 2

3 labour market regons, wthn each large regon. For brevty, we concentrate here on the second step. Four ndcator components were used n the formaton of the total ndcator whch s the bass for the dstrbuton formula. Each ndcator component operatonalses a dmenson of the crtera whch are decsve n the allocaton of funds accordng to the Socal Code IV. The four ndcator components are - Rate of change of employment - Rate of underemployment - Rate of unemployed wth specal labour market problems - Outflows from unemployment nto regular employment The defntons of the sngle ndcator components are: 1. Rate of change of employment calculated for two years. Ths ndcator usng data from the quarterly record of the employment statstcs whch ncludes all employment subject to socal securty contrbutons was easy to defne and to realse, snce the legal provson could be converted drectly nto operatonal terms. In order to obtan the same drecton as the other ndcator components, the sgn of the rate of change of employment was changed. Studes have shown that ths ndcator component demonstrates relatvely unstable behavour, as n a relatvely short tme consderable shfts can occur between the regons n the rate of change of employment subject to socal securty contrbutons. The ndvdual Employment Servce Regons are affected to dfferng degrees by cyclcal effects dependng on the strength and tme of ther occurrence. Therefore a two-year perod s now used for the calculaton of ths ndcator to smooth up abrupt changes. 2. Forecast rate of underemployment, made up of the forecast unemployment rate and the partcpants n relevant labour market polcy measures. By usng the measures, part of the underemployment n the partcular regon s absorbed and open unemployment s prevented. For ths reason structural adjustment measures, job creaton measures and full-tme tranng measures were also ncluded. The regonal values for the measures and unemployment are shown n Table Rate of unemployed wth specal labour market problems. Here people are counted, who are long-term unemployed. In addton - among the unemployed - dsabled people, elderly (above the year of 55), not formally qualfed people and those, who came recently back to the labour market are ncluded. All persons are counted only once, even f they have more than one of the mentoned attrbutes. The number of these people belongng to the these groups s weghted by the denomnator of the underemployment rate. 4. Outflows from unemployment nto regular employment n order to take nto consderaton the demand for labour or the capacty of the labour market to absorb workers. Ths ndcator does not measure the problem stuaton of the labour market and thus portrays a dfferent dmenson of the 3

4 chosen total ndcator. The ncluson of ths dmenson can be explaned by the ncorporaton of funds for tranng measures nto the ntegraton ttle. The budget for actve labour market polcy should be spent where there are good prospects for success. One problem of ths ndcator component s that seasonal fluctuatons are reflected especally strongly n the outflows. In order to avod regons wth hgh proportons of seasonal employment, n other words wth large proportons of the toursm and buldng ndustres, beng reflected dsproportonately and undesrably n the ndcator values, the month of June was used for calculatng the ndcator values, as ths month s not partcularly affected by seasonal fluctuatons. Solely for reasons of clarty, the values were multpled by the factor 12. As the regonal proportons do not change as a result of ths transformaton, the further calculatons are not affected. When determnng the values for the outflow ndcator, transfers nto job creaton schemes or structural adjustment measures were not taken nto account snce ths does not sgnalse the desred type of capacty of the labour market to absorb workers. The data requred for the four ndcator components were made avalable by the statstcs department of the Federal Employment Servce n the state desred as regards currentness and demarcaton. After completng the calculatons, the bass data (here ncluded n Tables 1 and 2) were addtonally checked once agan by the sectons responsble. In ths way t was possble to guarantee that solely the correct offcal data from the Federal Employment Servce were used. 3. Basc concept used n the constructon of the formula The procedure adopted for the constructon of a formula to be used for the dstrbuton of the funds of the dscretonary payments for actve employment promoton s to be descrbed brefly n the followng, before the formally exact calculaton s presented n the next secton. The dstrbuton formula s based on a total labour market ndcator whch combnes the four ndcator components dscussed earler. Table 1 contans the basc data for all four of the ndcator components whch are to be used n constructng the total ndcator, and provdes an overvew of the defnton of the ndvdual ndcator components. The underemployment rate, whch s one of the relevant ndcator components, s n turn made up of dfferent components, whch are shown n more detal n Table 1. 4

5 Table 1 Basc ndcator components n percent Calculaton for budget year of 2002 Rate of. Rate of under- Rate of specal Rate of out- Reference change of em- ployment - groups of flow from un- Quantty(regon ployment unemployed employment Sze: number (2 years)) nto employm. of people) Regon (Landesarbetsamt) Schleswg-Holsten- 1, ,347 6,906 6, Hamburg Nedersachsen- Bremen 1, ,618 7,947 6, Nordrhen-Westfalen 1,4745 9,836 7,683 4, Hessen 1,9354 7,404 5,848 4, Rhenland-Pfalz- 1,5210 8,098 6,220 5, Saarland Baden-Wuerttemberg 1,8271 5,375 4,337 4, Bayern 2,1511 6,166 4,391 5, Westberln 0, ,411 12,757 6, Western Germany 1,7553 8,684 6,405 5, Mecklenburg- -1, ,806 11,970 11, Vorpommern Berln-Brandenburg (o. -1, ,769 11,853 9, Westberln) Sachsen-Anhalt- -1, ,162 12,000 10, Thuerngen Sachsen -1, ,897 12,061 10, Eastern Germany -1, ,808 11,977 10, Defnton Indcators: Rate of change of employment {(6/00-6/99)/6/99 + (6/99-6/98)/6/98}/2 Underemployment rate (Unemployed(forecast 2002) + full-tme tranng measures (FbW)+ structural adjustment measures(sam) + job creaton measures (ABM) (all 8/00-7/01)) / Reference quantty Rate of spezal groups of unemployed: Long term unemployed., unemployed over 50, wthout formal qualfcatons, dsabled und people who come back nto the labour market. (8/00-7/01) / Reference quantty Rate of outflow of unemployed nto employment (wthout ABM, SAM 6/01) / Reference quantty * 12 Reference quantty: Dependent Workforce+ full-tme tranng measures (FbW) + unemployed Regons (Landesarbetsamtsbezrke are dvded between eastern and western Germany The four ndcator components rate of change of employment subject to socal securty contrbutons (wth a reversed sgn), underemployment rate, rate of specal groups of unemployed and rate of outflow from unemployment nto employment can not be combned smply e.g. by calculatng the average. They have a dfferent varaton and a dfferent range of values. If ths effect s not controlled for, mplct weghtngs of the ndcators wll result. For ths reason t s necessary to standardse the ndcator components beforehand,.e. they must be transformed n such a way that they show a mean value of 0 and a standard devaton of 1. 5

6 By means of the standardsaton t s taken nto account that the ndcator components ncluded show entrely dfferent characterstcs as regards ther defnton and ther character. It s not possble to make a drect comparson of a growth rate, such as that for employment, and proportonal values. Standardsaton gves them a common bass. 2 The standardsed ndcators could be combned by formng an average. In ths way, however, the fxed varaton of the ndcator components of 1 would have to be reflected n the values of the total ndcator. Ths would be an effect of the ndcator constructon and not one of the emprcal realty. As one wshes to take nto consderaton the actual dfferences n the problem stuatons of the ndvdual regons, the average of the standard devatons weghted by the partcular mean values s calculated (ths s the varaton coeffcent) and ths value s carred over to the total ndcator. As all calculaton operatons are carred out separately for eastern and western Germany, the dfferent varaton of the labour market problems n the two parts of the country can also be reflected n the values of the total ndcator. Ths s of some relevance snce a look at the orgnal values of the ndcators n Table 2 shows that accordng to Employment Servce Regons the labour market stuaton n eastern Germany s much more homogeneous than n the west. The total ndcator can then be used for the dstrbuton of funds. The allocaton scheme takes nto account on the one hand the sze of the regon n queston, and on the other hand the pressure of the problem stuaton. The choce of the weght of the two factors s a decson of regonal labour polcy; t does not result automatcally from the procedure. A need for decsons also exsts n another respect: the ndvdual ndcator components can be weghted dfferently. 4. Exact descrpton of the allocaton In the followng the formally exact descrpton s gven. A total ndcator G s formed by lnkng ndvdual ndcators E. The smplest procedures that can be appled here are multplcaton and addton. A multplcaton lnk has the pecularty of gvng extreme values a hgher weght. As there s no cause for ths n ths case, addton s used as a lnkage here. The total ndcator G s determned as follows, when t s addtonally taken nto account that specfc weghts w are allocated to the ndvdual ndcators, and r s an ndex for the regonal unt n queston. G = w E r r (1) 2 Smlar rules are appled n many areas of lfe, for nstance n sport. In the case of mult-dscplne sports each performance n one ndvdual dscplne s gven a score whch s mportant for the overall result. Only n ths way s t possble to compare results n metres for long-jump and tmes for runnng dscplnes, e.g. n decathlon. Ths also apples for results n the same unt of measurement, as the varaton of the results s mportant. If one were smply to add up the results of a 100-metre run and a 1500-metre run, the 100-metre specalst would be hopelessly at a dsadvantage. 6

7 Pror to combnaton, the ndcator components have to be standardsed as they show dfferent ranges of values and also the mean range of varaton of the values resultng for them,.e. ther standard devaton, fluctuates. If no standardsaton were carred out, ndcators wth a hgher standard devaton σ would mplctly receve a hgher weght. Wth the followng formula for the standardsaton, values are produced whch show the mean value of zero and the standard devaton of one: e r = E E σ r (2) Here E denotes the natonal value for the partcular ndcator component and σ ts standard devaton whch s calculated accordng to the followng formula n whch R stands for the number of Employment Servce Regons: R σ = B r E r E r ( ) 2 (3) Here B r = b r /Σb r whch s denoted as reference proporton where b gves the absolute number of persons called the reference quantty, whch s formed as the sum of persons n dependent employment and partcpants n measures (cf. appendx Table 1). The reference quantty s therefore a measure for the sze of the regons and s used as denomnator n the calculaton of the respectve rates whch serve as ndcators. The standard devaton and mean values are calculated separately for eastern and western Germany, snce the labour market stuaton n the two parts of the country contnues to dffer greatly. The above formulae dffer from those usually found n text books on statstcs only n that the natonal value s used nstead of the arthmetcal mean and that weghtngs are set accordng to the sze of the regon when calculatng the standard devaton. One of the purposes of the precautons s for changes n the boundares of a regon to have as lttle effect as possble on the total ndcator. The standardsaton produces ndcator values wth the standard devaton of one and thus oblterates the nformaton contaned n the orgnal data about the varaton of the problem stuatons. In order to avod ths effect t s approprate to apply the average of the varatons back to the ndcators. However, the standard devaton can not be used for ths as the orgnal data have dfferent mean values and ranges of values. Instead, t s advsable to use the varaton coeffcent V, whch s the standard devaton weghted by the natonal value: V = σ / E (4) However, there are two problems whch have to be consdered here: the varaton coeffcent s only defned for values ether greater than or smaller than zero. Ths s not always gven, however, n the case of the rate of change of employment E b. For ths reason, only the varaton coeffcent of the other three ndcators s used. Therefore, the ndcator change of employment does not contrbute to the ncluded assessment of the deepness of regonal dspartes on labour markets. 7

8 Secondly, the applcaton of the value of the varaton coeffcent to the standardsed values can only be an approxmaton of the soluton, snce the standardsed values show a mean value of zero. In spte of ths restrcton, the ncorporaton of an automatsm for transferrng the varaton range of the emprcal problem stuatons to the calculated ndcator values s an mportant advantage of the selected procedure. If dfferent weghts for the ndcators are taken nto consderaton, the followng formula results for the total ndcator GI: w er GI r= V (5) 3 It must also be borne n mnd that wth the standardsed ndcators at frst only a herarchy for the ndvdual regons s produced. Here, however, t s a matter of obtanng a formula for the dstrbuton of funds. For ths the total ndcator must addtonally be multpled by a measure) for the sze of the regon concerned. If to ths end, as would be approprate, the relatve number of workers (or the reference value) B r s calculated, the dstrbuton of funds can be gven accordng to the followng formula: M r = B r + B r GI r S (6) It can be seen that the proporton of the avalable funds that s gven to a regon s equal to two components whch are lnked by addton. The frst smply gves the sze of the regon concerned, by means of whch the funds are dstrbuted solely accordng to the sze of the regon. The second component on the other hand, whch s n turn made up of three factors multpled together, gves the pressure of the problems of the labour market stuaton 3. A global control factor S determnes the relatve weght of the two components. Ths factor can be set freely, f t s very small, the dstrbuton of funds s determned almost solely by the relatve sze of the regons; f t s large, the problem stuaton on the partcular labour markets, whch s shown by the total ndcator, comes through more strongly. It must be kept n mnd that as a result of the standardsaton the second component s negatve for those regons whch are n a comparatvely good poston. These regons are therefore allocated fewer funds than they would be enttled to accordng to ther sze. Thus there remans only one fnal step n the process of constructng the ndcator and ts applcaton for the allocaton of funds: the sum M = Σ r M r s only approxmately equal to one, so that for the correcton t s necessary to dvde by ths very total M. Then t results M the basc allocaton quantty. 3 Snce an ndcator component for the capacty of the labour market to absorb workers s also ncorporated nto the overall ndcator, the statement as to the pressure of problems s modfed accordngly. 8

9 5. Dscusson of the procedure chosen The adopted procedure s ntended to mplement the ntenton of the legslator of the Socal Act and to take nto account formal aspects of the ndcator constructon n accordance wth scentfc standards, n order to produce a result that s adequate. In prncple the funds could also be dstrbuted followng other procedures; the method selected here uses sutablty as a crteron, not exclusvty. The proposed procedure has, among other thngs, the advantage of smplcty; for example there s no need for multvarate methods, whch would make t more dffcult to explan the method to the decson-makers. Ths explanaton of the method s necessary as there are stll nterventon possbltes and these are vrtually revealed by the method. One of these nterventon possbltes concerns the control quantty S, whch can not be fxed n advance e. g. by a statstcal crteron. The ncorporated ndcator components refer to dfferent elements of economc realty; n order to nterpret them t s necessary to have knowledge about labour market processes, whch are not least substantated theoretcally. Ths must be reflected n the choce of the factor S. What s ultmately behnd the choce of the factor s a queston as to the fundamental effects of labour market polcy. Does t seem more favourable accordng to poltcal purposes to employ measures more at the focal ponts of the labour market or s t better to dstrbute the funds more equally? For the decson problem there was addtonal help n so far as that the varaton of the total ndcator was modfed n such a way that s orentated towards the emprcal varaton of the ndcator components. If one assumes that ths varaton reproduces the dfferences n the problem stuatons n a certan way, then an orentaton quantty s provded here. Accordng to the mathematcal vewpont, varaton could also be taken nto account by means of a modfed control factor S', whch ncludes the weghtng by means of varaton coeffcents (n other words S' = S(Σ V )/3). Ths can be seen when equatons (4) und (5) are used: M r = B + B r r 4 e r V S' S = Br ( e r ) (7) When takng the varaton nto account, however, t s necessary to be aware that ths s at varance wth the standardsaton, as the latter s based on the standardsaton of the varaton of the ndcator components. For takng the varaton nto account, however, exactly ths varaton s used agan whch necessarly dffers for the sngle ndcator components. The problem s ultmately nevtable as every procedure-related soluton has to decde n the descrbed sense between standardsaton and mantanng the varaton. For ths reason the development of the ndcator components should n future be carefully observed and f necessary the choce of the control parameter S should be adapted. Fnally t s necessary to pont out that the permssble range of values for the control parameter has an upper lmt. Formally the problem arses because the total ndcator assumes negatve values for regons wth a comparatvely favourable labour market stuaton. As can be seen from equaton 9

10 (6), f B r < -B r GI r S, the allocaton of funds becomes negatve from a certan pont. In ths case, whch s reached when S > -1/GI ro, ndvdual regons r o would not only receve no money, but would even have to submt funds. Ths s, of course, not a sensble result that hnts at strong outlers and ndcates that the factor S s not chosen sensbly. In many test calculatons no such nadmssble effects for the dstrbuton of funds occurred n the area of the values that were consdered for the control factor. Table 2 Total ndcator constructed wth four standardsed ndcator components (budget year 2002) Indcator components wth equal weghts, control factor S = 1, Reference basc Allocaton of Devaton Devaton quantty (%) allocaton funds (Pre- from regon from allocat. B quantty (%) M vous year 50%) M X sze prevous year Regons Wth respect to M Schleswg-Holsten- 6,62 8,37 8,18 26,47 3,19 Hamburg Nedersachsen- 12,67 16,40 16,22 29,47-0,47 Bremen Nordrhen-Westfalen 26,64 30, ,39 2,02 Hessen 9,06 7,27 7,74-19,74-9,47 Rhenland-Pfalz- 7,46 7,81 7,50 4,71 8,82 Saarland Baden- 15,99 9,88 10,48-38,18-5,85 Wuerttemberg Bayern 18,46 12,62 13,36-31,63-3,02 Westberln 3,11 6,90 6,55 122,13 5,26 Western Germany 100,00 100,00 100,00 0,00 0,00 Mecklenburg- 11,86 12,11 12,59 2,19-0,60 Vorpommern Berln-Brandenburg 25,39 24,23 23,90-4,58-6,10 (o Westberln) Sachsen-Anhalt- 33,77 34,64 34,92 2,56 0,23 Thuerngen Sachsen 28,98 29,02 28,59 0,13 5,70 Eastern Germany 100,00 100,00 100,00 0,00 0,00 Devaton proportonal allocaton (West)* 27,46 Devaton proportonal allocaton (East)* 2,32 The last two lnes show separately for East and West how much the allocaton of funds devates from an allocaton accordng to the regon sze B (summed up over all regons) 10

11 6. The result Table 2 shows the results of the allocaton procedure for the budget year It was done accordng to a decson of the Board of the Federal Employment Servces to use the procedure developed for the allocaton of funds. Ths decson ncluded further to use equal weghts for all ndcator components and to set the control factor S to 1,5. Later on t was decded to used the drectly computed values of M only wth a global weght of 50 %. The other 50 % s the allocaton of the prevous year. Ths was done to smooth up the allocaton of funds. The last two columns of table 2 compare the dstrbuton gven by M wth the szes of the regon and wth the M of the prevous year. The devatons A are calculated as A r = (M r - B r )/ B r. The devatons A are relatvely small; ths becomes clear n a summarsng quantty D x, whch s contaned n Table 2. The last lnes of the table shows how much the generated dstrbuton of funds dffers from a proportonal dstrbuton. For ths the absolute values of the dfferences between the frst column of the table and the proportonal values for the quantty B are smply added together wth the column relevant for the dstrbuton of funds; ths s done separately for east and west. Ths s done n accordance wth the followng formula: D x = Σ r B r - M r x The smallness of the values for D x can be explaned by the fact that also the second term n equaton (6) ncludes the sze of the regon concerned B r. Even f one wshed to orentate oneself radcally towards the problem stuaton and not towards proportonalty when dstrbutng the funds, an allocaton of budgetary funds has to take nto account the sze of the regon. The values for D x dffer for eastern and western Germany even when the same control factor S s used. Ths can be attrbuted to two characterstcs of the selected ndcator constructon. Frstly the dfferent values for the average varaton coeffcents V Ost und V West have an effect. Secondly the correlatons between the ndcator components affect the result. If they are hgh, then there are serous regonal dspartes, the values for the total ndcator dffer more consderably and the funds are redstrbuted to a greater extent. It s thus clear that the total ndcator reflects such correlatons. The results obtaned by the procedure descrbed was afterwards used by the regonal unts of the Federal Employment Servces (Landesarbetsaemter) to allocate funds to the smaller unts of local employment offces (Arbetsamtsbezrke). The same procedure was appled respectvely, whereas t was possble to use the optons of weghtng the ndcator components dfferently and to chose a dfferent control factor S. 7. Outlook Snce the begnnng of 1998 the labour market polcy funds have been dstrbuted accordng to the method descrbed. Meanwhle the law that regulates the workng of the Federal Employment Servces has been changed several tmes. Up to now only slght correctons of the formula for the allocaton have been necessary (Blen 2002). But now there s a fundamental reform of labour market polcy gong on, ncludng a complete reorganzaton of the Federal Employment Servces. The budgetng process wll be changed fundamentally. Therefore changes of the allocaton process are 11

12 unavodable. It s necessary to start the debate agan, thnk about the ntegraton of labour market and structural polcy measures and about many other aspects. In 2004 a new proposal of the allocaton scheme wll be assessed by the Board of the Federal Employment Servces n Germany. The results of a scentfc debate about the ratonalty of the allocaton process can be used n a revson of the method. References: Blen, Uwe (1998): "De regonale Mttelvertelung fuer de aktve Arbetsmarktpoltk ", n: Mttelungen aus der Arbetsmarkt- und Berufsforschung 31/4: Blen, Uwe (2002): "En Arbetsmarktgesamtndkator zur Mttelvertelung fuer de aktve Arbetsmarktpoltk", n: Klenhenz, Gerhard (ed.): "IAB Kompendum Arbetsmarkt- und Berufsforschung", (Betraege zur Arbetsmarkt- und Berufsforschung 250), Nuernberg: Downes, Thomas A.; Pogue, Thomas F. (2002): "How Best to Hand Out Money: Issues n the Desgn and Structure of Intergovernmental Ad Formulas", n: Journal of Offcal Statstcs 18/3: