LABOUR MARKET EFFECTS OF THE CRISIS
4. THE INFLUENCE OF THE ECONOMIC CRISIS ON SPATIAL INEQUALITIES OF UNEMPLOYMENTINEQUALITIES OF UNEMPLOYMENT
Hajnalka Lőcsei Introduction
The global financial crisis also created turbulence in the Hungarian economy during the autumn of 2008. A significant and critical sign from society’s point of view was the sudden increase in the number and share of jobseekers. The deterioration in the employment situation affected the regions unequally: the effect varied across regions with dissimilar socio-economic character or spa- tial location, and also appeared differently in time. The aim of the study is to explore these spatial and possibly temporal differences.
Today it is not necessary to explain the importance and aim of analyses which have regional aspects. Development and lagging behind appears specifically not randomly but is often concentrated spatially. In addition the regional as- pect is not just a projection of socio-economic features, but also plays a role in the development and subsistence of inequalities, therefore it has substantive explanatory power.45
The spatial duality of the Hungarian economy that emerged and stabilized following the regime change primarily due, as it was, to differences in entre- preneurial productivity, efficiency and foreign direct investments, also has a regional character (Barta 2000, Kukely 2008). Simply put, the country can be divided into two parts, one having a favourable position, competitiveness and ability to comply with the altered circumstances, and another having a much less favourable situation, hardly attracting FDI and continuously lagging be- hind. This is because FDI, which creates job opportunities, was not aimed at locations randomly, but preferred regions having an advantageous geographi- cal location and agglomeration economies and which offered a skilled labour force that had experience in modern production techniques (Fazekas 2000). It follows that FDI flowing to backward regions with unfavourable labour con- ditions is still not to be expected.
Both the volume and structure of unemployment mirrors the spatial divi- sion of the economy and society. In contrast to the central and north-western part of the country with its favourable position, in East- and South-Hungary not only is the rate of unemployment higher but also the composition of un- employed people is more unfavourable (in terms of education or age structure), additionally the share of permanent and passive unemployment is higher. The examination of the recession-caused changes of the – at least decade long es- tablished – spatial structure of unemployment can provide useful information not only for the tools and institutions established to handle unemployment,
45 Regional differences in development and the labour market are analysed by several Hungarian researchers (e.g.
Ábrahám–Kertesi 1996, Nemes Nagy–Németh 2005), and in the background to these inequali- ties, in addition to structural fac- tors, geographical location was also found. In chapter 1 Köllő used regional variables for the explanation of the decline in working hours, unemployment and wages in 2008–2009, but these variables did not prove to be significant.
but can also give general lessons. The main guideline of the study is therefore the question of what kind of effect the global economic crisis and the extreme recession had on the country’s spatial structure of development. Additionally, can the initial apparent equalisation exist over a long period, or can a further sharpening of the already considerable economic and social conflicts generat- ing differences eventually be expected?
To demonstrate the processes of the present and the rapidly changing recent past, the register of jobseekers happened to be the most appropriate vehicle.46 The applied database was provided by the PES (Public Employment Service of Hungary) on a monthly basis and with just a couple of months delay. This served as a basis for aggregating data on a microregional level, where spatial processes are most suitable to be analysed. The study therefore takes into ac- count the number of jobseekers and an indicator close to the unemployment rate, namely the changes in the share of jobseekers compared to the 15–64 years old population. In the following the rate of unemployment is identified by these indicators. In order to filter seasonal effects, changes are described in relation to the previous year (population numbers are presented in the ratio of the previous year, while the changes of the unemployment rate are interpreted in percentage point differences).
The new wave of unemployment – general trends and periods There were approximately 424 thousand jobseekers in September 2008 in the registry of the PES, and the share of jobseekers in the 15–64 years old popula- tion was 6 percent. These data – not counting a shorter transitional period with seasonal reasons – were continuously in decay until February 2010, when cca.
659 thousand people had no job; this meant a 9.4 percent share. In the longer period commencing from the change in regime this can be named as the sec- ond large wave of unemployment, since the situation worsened the last time in the same rapid way at the beginning of the ‘90s. That time the number of registered unemployed people increased practically from zero (there were 23 thousand registered in January 1990) to more than 700 thousand.
It should be pointed out that the evaluation of “raw” data is difficult due to the in-year change of unemployment. The main reason is that the employ- ment capacity of certain sectors, mainly agriculture, construction and tourism are seasonally fluctuating, and therefore it is a general phenomenon to see the number of unemployed persons increasing in February and March, decreasing in spring and summer, and then increasing again in the autumn and winter period. (Seasonality is in minor part explained by the appearance of entrants on the labour market in July and in January-February.) So the role played by seasonality in the increasing number of jobseekers in 2008–2009 and in the change for the better in spring 2010 cannot be exactly determined. From the pure changes in the number of jobseekers the development of trends can be
46 Public Employment Service (PES) registers unemployed persons according to prevail- ing Hungarian law. This regis- try was developed for operating and evaluating the toolkit es- tablished for dealing with un- employment, so the definition of unemployed workers (lately termed jobseekers) depends on the current employment policy and the rigour of the support system, so it results in higher values than the Labour Force Survey published by the Hungar- ian Statistical Office. (See Figure 16 in the introductory study of this volume written by Bálint–
Cseres-Gergely–Scharle.) The frequency of rule changing causes problems in the temporal and spatial comparison of data, because increasing or decreas- ing the number of jobseekers depends not only on the demand for a workforce. Before the crisis the number of jobseekers in the PES’ registry was increasing, while unemployment was stag- nating according to the Labour Force Survey. This difference was obviously due to the change of regulation – e.g. the raising of the retirement age, increasing the severity of disabled superan- nuation, social security reform.
(See also ÁFSZ 2008, pp. 15–16.)
concluded only through prediction and care must be taken to eliminate the seasonal effects. The difference between raw and seasonally adjusted data, as well as the increase of unemployment can be seen on Figure 4.1.
Figure 4.1: The number of registered jobseekers in 2008–2010
Note: Seasonally adjusted data, September 2008 = 100%.
Seasonal adjustment is prepared with X-11/12 ARIMA procedure by MultiRáció Kft.
on behalf of the PES.
Source of data: PES, http://kisterseg.afsz.hu/index.php?ts=00&am=1&kiigbtn=Rajt a&chxs=1&mode=kiig.
On the basis of the speed and nature of changes the past (from autumn 2008 until summer 2010) is divided into four periods.
1) From October 2008 to January 2009 seasonally adjusted data already in- creased to an appreciable extent, the increase of raw data therefore exceeded the volume normally available in previous years.
2) From January 2009 the growth rate dramatically increased, more and more people became registered by PES. This period lasted until May 2009. In May, though, a slight decrease was seen in the number of jobseekers – by 5000 people or rather 0.1 percentage point in under a month. After filtering out the seasonal effect it is obvious, however, that the trend of the rapid increase had still not been broken at this point. In May 2009 the number of jobseekers was already 33 percent (nearly 140 thousand people) higher than one year earlier.
3) From June 2009 until February 2010 the number of registered people was increasing, but the period of drastic changes had come to an end. Raw data hardly change during the summer, the seasonality caused a much smaller wave in the time series: the number of jobseekers decreased by a few thousand people in only two months (both in June and July). The stagnation in summer and the slight increase in autumn were again replaced by a large volume of in-
300 400 500 600
700 Number of jobseekers
8.
7.
6.
5.
4.
3.
2.
1.
12.
11.
10.
9.
8.
7.
6.
5.
4.
3.
2.
1.
12.
11.
10.
9.
Thousand Per cent
2008 2009 2010
90 120 Sesonally 150
adjusted data
flows. During this period, however, the increase in the number of those on the register had rather been influenced by seasonal effects, although the role of the economic downturn was still not negligible.
4) From February 2010 the negative trend seems to turn: first, the season- ally adjusted, then also the raw numbers started to decline. Between March and June the number of registered jobseekers declined to about 100 thousand people. In contrast to the previous year the positive effects of seasonality were not extinguished in 2010 by a wave of dismissals in the other sectors. However the whole improvement cannot be explained by seasonality: in “no recession”
years the outflow of only 50 thousand people is usual.
Favourable processes of the last period may warrant further explanation;
the improvement, which was also significant apart from the seasonal effects, should be treated carefully. In fact, according to the labour survey of the HCSO there is a decrease in the number and rate of unemployed people, but this im- provement is far from being of a great degree, as can be seen from PES data: 25 thousand people or 0.7 percentage points between the 1st and 2nd quarter of 2010. The decrease in the number of jobseekers registered by the PES was also potentially affected by the substantial intensity of the Back to work public em- ployment programme.47 (See the introductory study of this volume written by Bálint–Cseres-Gergely–Scharle on this programme and its effects.)
The programme, launched in 2009, aimed at improving the employability of a well determined – also incidentally spatially relatively concentrated – social group, the permanently unemployed, who receive regular social allowances.
It can be assumed that as a result of the programme individuals have dropped out of the circle of registered jobseekers, which explains the greater part of the differences between the HCSO-ILO and PES registries: passive unemployed persons, who were counted as inactive in labour surveys. To sum up, the signifi- cant improvement of the seasonally adjusted number of registered jobseekers is not necessarily a sign of recovery from the crisis, but rather reflects the results of an attempt to solve a long-existing social and employment problem. Proof of this train of thought and the evaluation of the results of the programme is not a specific consideration in a study analysing the labour market effects of the global crisis, but even so cannot be ignored in the assessment of regional processes, since it is not possible to filter out the effect.
Regional diffusion of unemployment
After the infiltration of the global crisis into the Hungarian economy48 un- employment – measured by the number and share of jobseekers – began to develop differently not only in time but also spatially. In the following section the regional differences of the effects of the crisis on the labour market are pre- sented, acting upon the previously outlined periods, by comparing spatial pat- tern and disparities experienced before and at the deepest moment of the crisis.
47 Public employment fluctuates seasonally due to the character of the works; during winters it is always less.
48 An increase in the number of jobseekers followed the decline in production and output with some delay. Industrial produc- tion began to decrease in May 2008, but a increase in unem- ployment was only perceptible in October 2008, when jobseek- ers’ numbers, over and above seasonal effects, also started to increase. Signs of improvement in the labour market also had a time-lag.
Initial state – regional pattern of unemployment before the crisis
The regional pattern of unemployment has been relatively stable since the sec- ond half of the 1990’s. The observable regional structure (Figure 4.2) clearly re- flects the spatial structure of development. In the triangle formed by the capital and its agglomeration, the M1 motorway, the western border of the country, Lake Balaton and the M7 motorway the rate of jobseekers is the lowest, un- der 5 percent. Unemployment constitutes the largest problem east of the so- called BB-line (an imaginary line between Balassagyarmat and Békéscsaba), as well as in South-Transdanubia and in micro-regions with no major cities.
Particularly critical was (and remained) the state of the micro-regions along the north-eastern border and in a small cohesive zone in Somogy and Baranya along the border of the country, as well as in settlements of the inner periphery of the Közép-Tisza area. In these areas it was not simply the high rate of job- seekers which caused problems, but also its unfavourable structure (high rate of low-skilled and long-term unemployed people). In areas burdened by social and often ethnic tensions, and with low entrepreneurial activity the demand on labour due to the absence of large enterprises is very low, additionally these areas are far distant from cities that could offer job opportunities.
By further generalisation of the above detailed picture it can be concluded that at the turn of the millennium some convergence clubs had arisen among regions and counties: the difference between the group of the 8 counties in West-Transdanubia, Central-Transdanubia, Central-Hungary and the coun- ties in the rest of Hungary became increasingly greater. The economic duality that can be derived from the location and agglomeration economies, and the significant spatial segregation of favourable and unfavourable regions is rep- resented by the strong positive spatial autocorrelation of the rate of jobseekers (Moran’s I was +0.73 in September 2008), and also by the spatial clusters out- lined upon results of Local Moran’s I 49 (Figure 4.3.).
The regional character of changes
In the following we examine how general trends influence the previously pre- sented stable spatial structure of the labour market. Since the shaping of the regional differences of unemployment is closely linked to the drop in economic performance as well as to cutbacks of companies,50 it is no wonder that accord- ing to both PES statistics and press information the number of jobseekers in the first period (namely between September 2008 and January 2009) increased the most in regions which were previously major centres of industrial production – that is the northern and north-western part of Transdanubia (Figure 4.4, January). On figures, which exemplify the changes, the index of the number of jobseekers depends also on the initial basis [part a) on Figure 4.4]. Where the number of jobseekers was relatively low prior to the crisis, the value of the index has increased rapidly, since a smaller value can be doubled much easier.
49 Moran’s I measures spatial au- tocorrelation, thus the proximity in space. A zero value indicates a random spatial pattern, negative values (up to –1) refer spatial dis- persion and positive values (up to +1) mean that similar (high or low) values concentrating in space. Local Moran’s I clustering spatial units with statistically significance spatial autocorrela- tion. (See Dusek 2004, Anselin 1995.)
50 The public sector usually pro- vides a safety net during a crisis.
Köllő highlights in Chapter 1 that at the beginning of the reces- sion initially wages were shrink- ing in the public sector, while the market sphere preferred layoffs.
See more details concerning ma- jor layoffs in Chapter 3 written by Busch and Lázár.
Figure 4.2: The share of registered jobseekers in the 15–64 years old population, September 2008 (percentage)
Source of data: PES, http://www.afsz.hu/engine.aspx?page=full_afsz_stat_telepules_ada- tok_2008.
Figure 4.3: Regional clusters defined on the basis of the rate of registered jobseekers,* September 2008
* Rook-contiguity, 5% significance level.
The rate of jobseekers in the high-high category is generally high both in the micro-re- gion and in its neighbourhood; in low-high category it is low in the micro-region but high in its surroundings, in low-low category it is low both in the micro-region and in its neighbourhood, and in high-low category it is high in the micro-region, while low on average in the neighbourhood.
Rook-contiguity: the neighbourhood relation exists if regional units share a common border with non-zero length, since association of spatial units in only one point is less relevant in connection with socio-economic relations.
Source: Own construction by the application of Geoda programme
Figure 4.4: The change of the number (a) and share (b) of registered jobseekers in relation to the same period of the previous year
a) the same period of the previous year = 100% b) in relation to the same period of the previous year in percentage points
Source of data: PES, http://www.afsz.hu/engine.aspx?page=full_afsz_stat_telepules_ada- tok_2009.
This approach well illustrates the spectacular and shocking changes. The an- nual change of the rate of jobseekers [part b) on Figure 4.4] shows a more fac- tual regional image. The changes have begun to spatially concentrate primar- ily from November, most spectacularly so in the northwest. This may easily distract attention from other large loser micro-regions in the underdeveloped parts of the country.
In centres of manufacturing industry in Komárom-Esztergom, Fejér, Győr- Moson-Sopron and Vas counties (e.g. in Komárom, Esztergom, Tatabánya, Mór, Székesfehérvár, Győr, Szombathely) as a result of the large companies’
holding back production and substantial layoffs, there were 1,5–2 times more
jobseekers registered in employment centres of the area prior to January 2009 than one year before. The record holder in January was the micro-region of Mór, where the volume of increase was 170 percent. On the contrary, if we were to look at a reverse map of development, the situation has deteriorated slightly in the poorer regions of the country and moreover the rate of jobseekers even decreased in 28 micro-regions. The strong impact of the crisis can be observed among those Northwestern-Transdanubian regions, where a major part of the workers were commuting to the nearby large cities’ manufacturing companies, i.e. inside the triangle formed by the capital, the M1 motorway, western bor- der, Lake Balaton and the M7 motorway.
The second period from February until May 2009 is the phase of the diffusion of the crisis. The number of job seekers increased from month to month, in May only very few regions remained in Hungary where the effects of the crisis were not perceptible through the significantly increasing number of registered job- seekers (Figure 4.4, March and May). The impact of the crisis was not limited to the environment of the export-oriented manufacturing facilities already affected, but became general and affected more and more sectors and regions.
At the same time it was still perceptible the most in inner peripheries of the north-western part of the country.
The maps began to darken spectacularly from March, namely the one-year rate was significantly increasing in a growing number of micro-regions by more than 3 percentage points. In this phase of diffusion an important role was played by contiguity, since impacts of the layoffs of certain large companies extended throughout the whole economy, and in all agglomerations of micro-regions with close connections the rate of jobseekers increased. (This was proven by the Moran’s I index calculated for the change of the rate of jobseekers. The in- dex reached the annual maximum in June 2009 with a value of +0.37, and then began to decrease, and strengthened its role again only in the fourth period.)
The economic crisis did not cause too great a shock on the labour market in most parts of the regions which lagged behind (beside the eastern border of the coun- try, in South-Transdanubia, in micro-regions of Bács-Kiskun country close to the Danube or to the border, or in the Central-Tisza area). In addition to the external and internal peripheries also less influenced were Paks and Tiszaújváros51 with industries of strategic importance, while in regions of Dunaújváros, Százhalom- batta, Kazincbarcika and Diósgyőr the major factories made significant layoffs, or in some cases it was just delayed by the promise of state support.52
The changes were slight in the capital and the agglomeration, however the
“unaffected” zone was getting narrower. It seems that the economy and the la- bour market of Budapest were less shocked by the recession. Although the ab- solute number of jobseekers increased to a 1.5 times higher value prior to May 2008 compared to the previous year this resulted in only a minor increase in relative indices (the increase is from 2 to 3 percent, namely only 1 percentage
51 Another important company in Tiszaújváros is the Jabil Cir- cuit, whose intention to make cutbacks became public in February 2009. It affected 900 workers.
52 E.g. in Kazincbarcika (Bor- sodchem – chemistry company) and in Dunaújváros (ISD Duna- ferr – steel production).