2. EVOLUTION OF THE HUNGARIAN RESEARCH AND DEVELOP- MENT POTENTIAL FROM THE REGIME CHANGE TO OUR DAYS

Transcription

1 ANALYSIS OF THE HUNGARIAN RESEARCH AND DEVELOPMENT POTENTIAL AND STATISTICAL METHODS OF ITS PROGNOSIS 1. INTRODUCTION László Molnár Ph.D. student Unversty of Mskolc, Insttute of Economc Theory In our days research and development (R&D) ncreases n mportance n all sectors of the economy. The government, enterprses and publc opnon take a growng nterest n ths topc. Recognzng the decsve role n boostng economc compettveness and productvty of R&D, n March 2002 the European Councl decded on ncreasng R&D expendture to 3% of GDP n order to brdge the gap between the EU and ts maor compettors lke Unted States and Japan. By passng a law on research and development and nnovaton, and ntroducng a compulsory nnovaton contrbuton for companes, Hungary has taken mportant steps towards ths obectve. Yet, havng reached halftme, the proposed goal seems extremely remote. In ths study I wll nvestgate the composton and evoluton from the regme change to our days of the Hungaran research and development potental, ncludng the number of research unts, research ntensty and R&D expendture as a percentage of GDP. By means of tme-seres analyss models I wll reveal the Hungaran R&D trends of the past 15 years and also predct the feasblty of achevng the EU obectve. At the same tme I wll use advanced statstcal methods to analyze the correlaton between the components of the R&D potental and the ndex numbers of the natonal economy. Towards the end of my study I wll nvestgate the evoluton and composton of down-to-earth results of the Hungaran research and development actvty and suggest ndex numbers for comparson and to measure the results. 2. EVOLUTION OF THE HUNGARIAN RESEARCH AND DEVELOP- MENT POTENTIAL FROM THE REGIME CHANGE TO OUR DAYS Before the analyss of the Hungaran research and development trends, I fnd t necessary to defne exactly what the keyword of my study, research and development means. Research and development s the term for all creatve actvtes whch am to ncrease the stock of knowledge, ncludng knowledge of nature, humanty, culture and socety and the use of ths stock of knowledge to devse new applcatons (Hungaran Central Statstcal Offce (KSH), 2004, p. 9.). I wll deal wth the man components of research and development potental: the number of research unts, staff number n R&D unts and R&D expendture. These factors llustrate accurately the research and development effcency of a natonal economy. Furthermore, statstcal yearbooks provde obectve and numercal data about them.

2 Research and development does not end n tself. The new stock of knowledge manfests tself n scentfc publcatons and patents. In the last quarter of ths chapter I wll treat n detal the alternatves of the measurement of these results Number of research and development unts One of the mportant ndcators characterzng the research & development potental and performance of natonal economes s the total number of those nsttutes, enterprses and other places whch do basc and appled research and appled development n an effectve way. The dagram n Fgure 1 descrbes the trend of total number of R&D unts from 1990 to 2005 n Hungary Number of R&D unts Number of R&D unts Logstc trend fucton Fgure 1 Total number of research and development unts and the logstc trend functon Source: compled by author from KSH In the frst half decade ( ) after the change of the poltcal regme n Hungary the number of R&D unts was n secular stagnaton. There were some R&D unts durng ths perod. The second half of the 1990s saw a powerful growth of R&D actvtes, whch resulted n some 60% ncrease n the number of R&D unts durng the next 5 years ( ). But ths quck growth came to a sudden halt n 2000, and durng the next 3 years ( ) the growth of actvtes dd not reach 10%. Between 2004 and 2005 the number of R&D unts dd not only fal to ncrease, but started to decrease slowly by 1%. In order to make a precse predcton of the number of R&D unts I used the analytc trend calculaton method from math-statstcal tme seres analyss 1. Among the tested trend functons a logstc one has proved far the best, due to ts strong explanatory power (R 2 =99.1%). 1 The analytc trend calculaton was carred out wth Solver extenson pack for Mcrosoft Offce 2003.

3 b1 (1.) y = b0 + b3 1+ b e Based on the analytcal trend functon the number of R&D unts converges to , namely, accordng to my estmaton ths number was n 2006, t s n 2007 and t wll be n The applcaton of lnear (R 2 =94.2%) and cubc models (R 2 =98.3%) for calculatons results n a good approxmaton, but n my opnon the former model underestmates, the latter one overestmates the ncrease n the number of R&D unts. I drew an addtonal concluson n connecton wth the unts strongly nvolved n the extenson of knowledge, after my nvestgaton of the composton and evoluton of R&D unts of hgher educaton, enterprses and R&D nsttutes. These results can be seen n Table 1. Table 1 Total number of research and development unts 2 Year R&D nsttutes, other research unts R&D unts of hgher educaton R&D unts of enterprses Total number of R&D unts % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % Source: compled by author from KSH The largest segment of R&D unts (n %) can be found at hgher educaton 3. The rate of ncrease of these unts has been balanced snce 1990, but ther proporton s slowly decreasng. R&D unts of enterprses have been ganng more and more ground (29.8% n 2005). The evoluton n ths feld appears even more obvous, f we take nto account that the proporton of R&D unts of enterprses was 13.9% n 1990, and durng the last one and the half decades ther number has been ncreased by 4.3 tmes. The 2 The value of b 0 coeffcent s , the value of b 1 coeffcent s , therefore the lmt value of the logstc functon at nfnty: lm f ( x) = x 3 Accordng to the KSH defnton R&D unts of hgher educaton are unverstes, colleges and ther departments, clncs and laboratores.

4 number and proporton of R&D nsttutes and other research unts have been stagnatng n the last 15 years Staff number n research and development unts Another mportant ndex for characterzng the R&D potental s the calculated staff number n R&D unts 4. These employees are scentsts and engneers, techncans, R&D assstants drectly promotng research and development wth ther work, as well as other manual and non-manual workers, provdng for workng condtons (KSH, p. 367.) Total calculated staff number n R&D unts Total calculated staff number n R&D unts Quntc trend functon Fgure 2 Total calculated staff number n R&D unts and the quntc trend functon Source: compled by author from KSH The number of R&D persons was dramatcally cut n the early 1990s. Durng ths perod ( ) ther number approxmately halved from to Snce then there has not been consderable change n the data. The ndex of research ntensty, whch s the rato of R&D employees to the total number of employees n a natonal economy, has had a smlar character durng the last one and a half decades. After a gradual decrease (from 0.75% to 0.54%) n the early 1990s the researchntensty ndex has been stagnant. Furthermore, the correlaton between the number of R&D persons and natonal employment fgures has a postve and strong character whch s also ndcated by the value of the correlaton coeffcent 5 (r=0.98). 4 The actual staff number converted to full-tme employees,.e. staff number weghted by the rato of tme spent wth actual research and development to the total number of workng hours (KSH, p. 367.). 5 cov( x, y) To characterze the correlaton I used Pearson s lnear correlaton coeffcent: r =, where var( x) var( y) d x d y x y 2 2 d x d y cov(, ) =, var( x) =, and var( y) = (Hunyad Vta, p. 588., 589.) n n n

5 Attemptng to balance the numbers of R&D persons I used dfferent trend functons. The lnear model although t has a very poor matchng (R 2 =18.2%) was able to demonstrate the negatve tendency n the evoluton of the number of R&D persons. Usng a cubc model (R 2 =96.5%) I managed to prove ths tendency statstcally as well. For further optmzaton the followng quntc model (R 2 =97.8%) was used because the cubc model seems to underestmate the evoluton of the number of R&D persons. (2.) y = b b1 + b2 + b3 + b4 + b5 Based on the observable stagnaton of the staff number n R&D unts n the last few years and on the prognoss of the quntc trend functon I came to the concluson that no sgnfcant change can be expected n the data n the near future. My estmaton s that the probable total calculated staff number n R&D unts wll be around thousand, wth the present slghtly droppng tendency contnung n the next two years. Addtonally, I estmate a 0.6% research-ntensty ndex assumng that there wll be no dramatc change n the natonal employment fgures. Table 2 represents the evoluton of the staff number fgures n R&D unts of hgher educaton, enterprses and R&D nsttutes between 1990 and I came to a few further conclusons by nvestgatng the composton of the total calculated staff number. Table 2 Total calculated staff number n R&D unts R&D nsttutes, R&D unts of hgher Total calculated staff Research Year R&D unts of enterprses other research unts educaton number n R&D unts ntensty % % % % 0.75% % % % % 0.65% % % % % 0.59% % % % % 0.59% % % % % 0.59% % % % % 0.53% % % % % 0.54% % % % % 0.57% % % % % 0.55% % % % % 0.56% % % % % 0.61% % % % % 0.59% % % % % 0.61% % % % % 0.59% % % % % 0.59% % % % % 0.60% Source: compled by author from KSH The dstrbuton of R&D persons s approxmately unform among the R&D unts of hgher educaton, enterprses and R&D nsttutes. The R&D unts of hgher educaton, the R&D unts of enterprses and the R&D nsttutes represent 35.3%, 31.8% and 32.9% respectvely of the total staff number. From these fgures and from the

6 prevously mentoned nner ratos of the R&D unts, one can conclude that the average staff number n R&D unts of hgher educaton s far behnd the average staff number of R&D unts of enterprses, whch n turn s exceeded by the average staff number of R&D nsttutes and other research unts exclusvely nvolved n research and development. The estmated value of the concentraton ndex 6 of R&D unts (L=34.3%) represents a weak-medum level of concentraton n terms of staff number R&D expendture One of the most mportant fgures n the statstcs of research and development as a creatve actvty s besde the number of R&D unts and the calculated staff number the sum of the current and captal expendtures from domestc and foregn sources. The European Unon, ncludng Hungary, has decded on ncreasng R&D expendture to 3% of GDP by Fgure 3 below shows the evoluton of the total R&D expendture between 1990 and Total R&D expendture, mllon HUF Total R&D expendture Cubc trend functon Fgure 3 Total R&D expendture and the cubc trend functon Source: compled by author from KSH In the early and mddle 1990s Hungary s total R&D expendture ranged around bllon HUF per year aggregatng % of the country s GDP calculated at the current rate. After the turnng-pont n 1996 n close relaton wth the rate of growth of the GDP R&D expendtures started to ncrease unsteadly. The strength and 6 To characterze the concentraton I used the concentraton coeffcent: 1 G = N ( N 1) k k = 1 = 1 f f Y Y (Hunyad Vta, p. 101., 111.) G L =, where 2 Y 7 Accordng to If. Smon the planned growth wll probably not come true, because the EU average has been under 2% for years and n Hungary R&D expendtures are even lower. The latter s also confrmed by my followng prognoss.

7 drecton of the connecton can readly be characterzed by Pearson s lnear correlaton coeffcent (r=0.98). However, the rate of growth of the sum of current and captal expendtures could not exceed the rate of growth of the GDP, thus the rate of the R&D expendtures n GDP remaned between 0.74 and 1.00%. I made the statstcal fttng of the tme seres of the R&D expendtures ncreasng year by year wth three alternatves. Although the fts of both the lnear (R 2 =89.0%) and the exponental (R 2 =96.4%) functons are acceptable, I used the followng cubc model for forecastng. Ths trend-functon s actually the golden mean between the lnear and exponental functons, and that s confrmed by ts ndex of ft (R 2 =98.1%). (3.) y = b b1 + b2 + b3 Accordng to the prognoss made by the cubc model, the total R&D expendture was around 260 bllon HUF n 2006, and t wll be 300 bllon HUF n 2007 and 350 bllon HUF n If the GDP s rate of growth contnues, the rate of R&D expendtures as a percentage of GDP s excepted to stay under 1.00%. The largest part of the R&D expendtures s comng from the current expendtures, but nvestments also mean several bllons of expense. Table 3 below shows the evoluton of volumes of current R&D expendtures and nvestments, as well as ther percentage n the total expendtures. Table 3 Total R&D expendture Year R&D current expendture, mllon HUF Captal expendture, mllon HUF Total R&D expendture, mllon HUF 8 R&D expendture as a percentage of GDP % % % 1.61% % % % 1.08% % % % 1.07% % % % 0.99% % % % 0.92% % % % 0.75% % % % 0.67% % % % 0.74% % % % 0.71% % % % 0.69% % % % 0.80% % % % 0.92% % % % 1.00% % % % 0.93% % % % 0.88% % % % 0.94% Source: compled by author from KSH 8 Includng amounts from state budgetary sources utlzed outsde R&D unts and ncludng the honorarum, salary complements of scentfc degrees, and also the amounts of the state scentfc scholarshps.

8 After a stagnaton n the early 1990s (21-39 bllon HUF/year), R&D current expendtures started to ncrease steadly n the mddle of the decade. In 9 years ( ) the yearly volume of R&D costs ncreased 4.3 tmes. However durng the whole tme the porton of current expendtures was unchanged, gvng 75-85% of the R&D expendtures. In the last 16 years captal expendtures have followed nearly the same trend. Stagnaton n the early 1990s, then a 6-fold ncrease from Further mportant questons of the analyss of R&D expendtures are as follows. How are they dvded among the dfferent R&D unts, what are ther sources and what are they used for? In 2005, nearly 45% of the total R&D expendture was used by enterprses R&D unts wth the number of these unts beng only one thrd of the all R&D unts and complement. The research nsttutes expended 29%, whle the hgher educaton segment wth the most R&D unts and largest complement expended hardly more than a quarter of the expendtures. The R&D unts of the tertary educaton segment have the smallest budget for research and development, an average of 33 mllon HUF/year, whle the research nsttutes and the R&D unts of enterprses have a budget of 290 mllon HUF and 120 mllon HUF per year respectvely. Regardng the fnancal sources, near half of the expendtures s comng from state budget. Further 40% s comng from the enterprses own budget or from other enterprses they have contract wth, or from supports. The sums from nternatonal organzatons through order, support, ad or tender gve 10% of the expendtures. (There s also some natonal support, but ts percentage s less than 0.5%.) In 2005 nearly 60% of total R&D expendture was expended on research. Wthn ths, the resources were used nearly n the same proporton for basc and appled research. For expermental development 40% of the expendtures were used Results of research and development and the methods of ts measurement The Hungaran statstcal practce uses the quanttatve data of scentfc publcatons and patents to measure the results of the research and development unts. Scentfc publcatons are books, chapters, artcles n Hungaran or foregn scentfc ournals and accepted theses n Hungaran or foregn language wrtten by employees of research nsttutes. Patents are all ndustrally applcable nventons based on new nvent actvty. (KSH, 2004, p ). The number of scentfc publcatons has not changed for fve years ( ), stayng around thousand each year percent of these publcatons were books n Hungaran or foregn language, the other percent beng artcles n scentfc ournals. Patents show a dfferent trend, namely the number of patent applcatons has dramatcally decreased n the last years. Whle between 2000 and 2003 the Hungaran Patent Offce (MSZH) fled 5 thousand patent applcatons a year, ths number has reduced to 2.7 thousand n 2004 and only 1.3 thousand n 2005 (MSZH, 2006, p. 31.).

9 Technologcal Actvty Index There are mult-component ndex numbers to measure the results of research and development unts whereof I wll ntroduce the Technologcal Actvty Index (TAI). Ths s the frst component of the Innovaton Capablty Index (UNICI) developed by the Unted Natons Conference on Trade and Development (UNCTAD). Ths ndex number ncludes the followng components wth equal weghts: - R&D personnel per mllon populaton - Unted States patents granted per mllon populaton - Scentfc publcatons per mllon populaton (UNCTAD, 2005, p. 113.) The prmal advantage of the Technologcal Actvty Index s that we can draw nternatonal comparsons wth ts help. However, nterpretng ths ndex number s extremely dffcult, because t averages populaton deflated values of factors measured n dfferent unts R&D actvty ndex To measure the down-to-earth results of research and development unts I have developed a much easer renderable method. Ths statstcal method contans two ndex numbers: a raw and a pure ndex number Raw ndex number The raw ndex number of R&D actvty shows the weghted sum of scentfc publcatons and patent applcatons per populaton n a gven year. The concrete value can be calculated wth the followng formula. (4.) PPE = n W Pu Pu + E m W Pa Pa E Employment Pa Number of the -type patents PPE Raw ndex number of R&D actvty (Publcatons plus Patents per Employment) Pu Number of the -type publcaton W Weght of -type patent Pa W Pu Weght of the -type publcaton In my opnon weghtng s necessary because wrtng an artcle, a chapter, a book, brngng out a conference publcaton or flng a patent applcaton ndcates dfferent research and development actvty.

10 Pure ndex number The pure ndex of R&D actvty dvdes the weghted sum of scentfc publcatons and patents by the total staff number n R&D unts. Its formula vares from that of the raw ndex only n the denomnator. (5.) PPP = n W Pu Pu + P m W Pa Pa P R&D Personnel PPP Pure ndex number of R&D actvty (Publcatons plus Patents per R&D Personnel) It s easy to see that the quotent of my R&D actvty ndex numbers wll gve the research ntensty, whch s the rato of the number of R&D personnel among the employed. In my analyss of the Hungaran research and development actvty I calculated the concrete values of the raw and pure ndex numbers between 2000 and I summarzed the results n the followng Table 4. Table 4 Raw and pure ndex number of R&D actvty Year Papers Books Patents PPE PPP Research ntensty % % % % % % Source: compled by author from KSH The evoluton of the Hungaran R&D actvty showed a curve-lke tendency at the start of the new mllennum. The ndces of the actvty were ncreasng gradually from 2000 to 2002 (PPE: from 1.09 to 1.19; PPP: from 179 to 195). After that the R&D actvty decreased below the ntal level durng the same length of tme (PPE: from 1.19 to 1.00; PPP: from 195 to 170). Between 2004 and 2005 there was no change n the Hungaran R&D actvty.

11 3. CONCLUSION AND RECOMMENDATION Fnally I brefly summarze and conclude the results obtaned from my studes n connecton wth R&D actvtes n Hungary. 1. In the frst 5 years after the poltcal changes the number of R&D unts stagnated. After that, a perod of progressve ncrease was followed by a regressve perod n the past few years. Ths tendency can be perfectly descrbed by logstc trend functons. Accordng to my prognoss the ncreasng trend wll contnue but wth a decreasng rate. 2. The number of R&D persons dropped dramatcally n the md 1990s and t has practcally remaned unchanged snce then. The same s vald for the research ntensty ndex, whch shows the rato of the number of R&D persons to the total number of employees. In my analyss I proved that there s a close and statstcally sgnfcant correlaton between the number of R&D persons and the total number of employees. 3. Smlarly to R&D unts, R&D expendture n Hungary started to grow progressvely from the md 1990s. However, the EU recommendaton that R&D expendture has to reach 3% of the GDP by 2010 seems mpossble to fulfll. The correlaton between the expendtures and the GDP s also strong and statstcally sgnfcant. 4. In Hungaran statstcal practce natural unts are used for measurng the effcency of R&D actvtes, but nternatonal lterature also apples complex parameters such as the Technologcal Actvty Index ntroduced by UNCTAD. 5. On the other hand I recommend the applcaton of the raw R&D actvty ndex (PPE) for measurng scentfc publcatons, patents and other R&D results on the level of the natonal economy. Ths ndex enables comparatve nvestgaton n tme and space. 6. I recommend that the pure R&D actvty ndex (PPP) should be appled for the characterzaton of R&D persons scentfc actvty. Both above mentoned ndces are easy to calculate and nterpret, and they perfectly descrbe the subects under nvestgaton.

12 REFERENCES Books HUNYADI MUNDRUCZÓ VITA (1997) Statsztka, Budapest, Aula HUNYADI VITA (2002) Statsztka közgazdászoknak, Budapest, KSH JÁNOSA (2006) Adatelemzés számítógéppel, Budapest, Perfekt KSH (2004) A K+F statsztka módszertana, Budapest KSH ( ) Kutatás és felesztés, Budapest KSH (2006) Magyar Statsztka Évkönyv 2005, Budapest, p MSZH (2006) Éves elentés 2005 Budapest NYITRAI FERECNÉ (1996) Gazdaságstatsztka, Budapest, KSH RAPPAI (2001) Üzlet statsztka Excellel, Budapest, KSH UNCTAD (2005) World nvestment report 2005: Transnatonal corporatons and nternatonalzaton of R&D, Unted Natons, New York, Geneva Revews IFJ. SIMON (2006) Gazdaság növekedés és kutatás-felesztés: a svéd és fnn példa tanulsága, Statsztka Szemle, december, p Internet KSH, Stadat-rendszer, portal &_schema=portal