European Job Vacancy Surveys: The same or still different?

Transcription

1 European Job Vacancy Surveys: The same or still different? Anja Kettner & Michael Stops Institute for Employment Research (IAB) Draft paper, April 30, 2008 Summary: Data on open job vacancies are highly relevant for labour market research. Since recently they have been published by the European Commission/EUROSTAT for some European countries. Usually data are collected by business surveys. A closer look at them shows, that an international comparison of vacancy data or even their interpretation with regard to national labour market policies does not stand on a reliable base. The statistical delimitations and used methods are very different; in some cases the data quality must be doubted. One reason lies in the tremendous costs of these surveys. Another reason lies in the obvious undervaluation of reliable and comparable vacancy data, not only for research but especially for labour market policy. This paper shows the progress done with national job vacancy surveys in Europe and explains the weaknesses of the available data material on the base of examples.

2 Introduction In July 2006 a German economic research institute headlined an article Vacancies A German problem (IW 2006). A comparison of 13 national job vacancy rates based on data published by EUROSTAT showed with 3.2 the highest vacancy rate for Germany. It was more than six times as high as in France with 0.5 and Portugal with 0.6, and 1.5 times as high as in the UK. This result was interpreted by the authors as an indication for the inefficiency of the German public employment agencies and for a lack of endeavours of long term and old unemployed to take a new job. Nowhere in Europe more positions would be vacant, despite high unemployment. Authors argued that this is caused by imperfect matching of job seekers and firms offering jobs. A look behind the utilized data shows that at present they are not comparable with perspective to international analysis. Most European countries made an effort in these latter days and started collecting data on job vacancies; some are published by EUROSTAT since recently. But the definitions, covered sectors and firms, methods for sample design as well as the practical conducting of the national surveys differ strongly. Additionally an international comparison of vacancy data requires taking into account the particularities of national labour market and labour market policy. The following section will explain the importance of reliable vacancy data in economic research. Section 3 will describe why the collection of vacancy data started only recently, what steps the European countries currently do to provide data and how they are published. Section 4 will discuss the differences between national job vacancy surveys and their implications. The paper finishes with a discussion on auxiliary knowledge on national labour markets that are needed to interpret the data in the right way. 2

3 1. Vacancy data in economics 1.1. The definition of vacancies and the job vacancy rate To understand the functionality of the labour market detailed information on the supply and the demand side are necessary. The supply consists of employees and unemployed. The demand consists of employees and vacancies. Vacancies, termed as unmet labour demand are counted as the number of posts, for which employers plan to hire a person and for which they are looking for candidates (NBER 1966, Muysken 1994). The post can either be occupied by a person that will leave before the new employee starts working, or it can be an unoccupied post, or a post that will be newly created. A vacancy exists from the beginning of the search process by the employer until the decision to hire a specific candidate. As counterpart to the unemployment rate the job vacancy rate relates the number of vacancies v to the total labour demand e + v: v r = v e + v 1.2. Matching analyses and the Beveridge Curve Vacancies and unemployed are the main variables in aggregated matching functions (see Petrongoglo and Pissarides 2001, Shimer 2007 and Yashiv 2007 for a broad overview on theoretical implications and available empirical studies). According to this theory the number of new jobs M (matches) formed in a given time interval depends on the number of unemployed u looking for jobs and the number of job vacancies v: 3

4 ( v) M = m u, The probability that a vacancy is filled and that an unemployed finds a job, respectively the duration of the vacancy and the duration of unemployment, is influenced by the behaviour of job seekers and employers, the specifics of the vacancies and the macroeconomic environment. Many studies analyze the role of individual characteristics of unemployed and employed jobseekers. Their search behaviour and efficiency is first of all determined by individual search costs and the expected return from (the new) employment (Pissarides 2000). In case of unemployed job seekers their unemployment duration affects the chances for reemployment because of the loss of skills. So the percentage of long term unemployed and the percentage of unemployed among all job seekers influence the matching probability at aggregated level (Coles/Smith 1998, Blanchard/Diamond 1989, Burgess 1993, Boeri 1999). Economic research also focuses on the impact of employment protection, the benefit-income-ratio and the mobility of jobseekers for successful matching (Blanchard/Portugal 2001, Jackman, Layard and Pissarides 1989, Fahr and Sunde 2006). The role of the technological progress, sectoral shifts and the overall business cycle is analyzed too (Lilien 1982, Burgess 1993, Shimer 2005, Fisher 2006). A large discrepancy becomes evident between the huge amount of research on the role of the searching behaviour of job seekers and the very few papers focusing the employers behaviour. As well as job seekers show different searching strategies and effectiveness, firms looking for new employees are not homogenous. Depending on the specifics of the firm and the specifics of the post to be filled employers use different search ways like advertisements in newspapers, internet, public employment agencies or networks, different selection 4

5 strategies, and plan different vacancy durations (see for instance van Ours and Ridder 1991 and 1992, Weber 2000, Burdett and Cunningham 1998, Fahr and Sunde 2001). The quality of aggregated matching can be shown by the by the Beveridge curve (Layard/Nickell/Jackman 2005, Blanchard 2006). It depicts the unemployment rate and the job vacancy rate in the course of time. During the last decades in most European countries an outward shift of the curve could be observed: unemployment rates strongly increased whereas vacancy rates showed narrower upward and downward movements. The long run outward shift shows a deterioration of the matching process between vacancies and unemployed and an increasing persistence of unemployment. Beveridge Curve models analyze if this is caused by a mismatch between labour demand and labour supply or if other macroeconomic reasons lay behind (numerous studies cited in Petrongolo/Pissarides 2001, pp. 409). A closer look at the empirical work on Matching Functions and on Beveridge Curve analysis shows an important problem The problems of the usage of data on registered vacancies only Most empirical analyses are based on registered vacancies, instead of their total number. Registered vacancies are announced to the Public Employment Services. Their number depends on the registering behaviour of firms, which differs between sectors, regions and over time (Yashiv 2000, Jackman/Layard/Pissarides 1989, Kettner/Spitznagel 2007). In many countries firms are not obliged to register their vacancies to the public services. But even if they are, it might be doubted if firms do register in every case. Registered vacancies may therefore reflect only a small and not representative percentage of all vacancies. Graph 1 and 2 show the Beveridge curves for Western Germany and Eastern 5

6 Germany on base of total vacancy data (line pulled through) and on base of the registered numbers (line dotted). Only vacancies immediately to be filled 1 are considered. 11, , , job vacancy rate 9,5 9,0 8, ,0 7, ,0 0,0 1,0 2,0 3,0 4,0 5,0 unemployment rate (a) Western Germany 19, job vacancy rate 18,5 17,5 16, , ,5 0,0 1,0 2,0 3,0 4,0 5,0 unemployment rate (b) Eastern Germany Graph 1: Beveridge Curves for Western Germany (a) and Eastern Germany (b) , 4 th quarter Source: IAB Job Vacancy Surveys : vacancies immediately to be filled, Administrative data of the Public Employment Agency Considerable differences are apparent depending on the utilized job vacancy rate. Since the registering quotas differ over time, variations of the total vacancy rate can be larger or smaller than indicated by the vacancy rate based on registered numbers. The general question arises, 1 In the German Job vacancy survey firms have to divide vacancies between vacancies to be filled immediately or at the next possible date or vacancies to be filled in the distant future. For remarks on the relevance of this division see

7 how far results of econometric studies using registered vacancy data would differ, if these studies would be repeated with the total number of vacancies and the total job vacancy rate. Firms choose what kind of vacancies they report to the Public Employment Agencies and which they do not. For example the German job vacancy survey shows that over the recent three years the registering quota of vacancies immediately to be filled and not requiring any vocational qualification was about 70 percent, whereas the registering quota for posts that required a university degree was 35 percent. Among the unemployed about 38 percent 2 do not have a vocational qualification; the percentage is even higher for the long term unemployed. It can be assumed that firms anticipate their low chances to find qualified candidates among the unemployed and that they assess other recruiting ways as more efficient like search via internet or advertisements in newspapers. But: if employers mainly report vacancies with low qualifications and if a significant share of unemployed job seekers has only low vocational qualifications, an analysis on base of registered vacancy data will come up with high matching rates and might find, that qualification mismatch has a small or even no effect on the Beveridge curve. Some authors attempt to correct the number of registered vacancies by use of data on new engagements, like Jackman/Layard/Pissarides (1989) for Great Britain and Franz/Smolny (1994) for Germany. However, not every vacancy will finally be filled. As long as a decision for a candidate is not disposed, the firm can cancel the vacancy and decide not to hire a new employee. The reason can be a change in the economic situation of the firm or a demand shock in the economy. Appearing frictions also influence the relation between vacancies and hires, the higher frictions the longer a vacancy will exist. The relation between hires and vacancies is unknown, as long as data on the total number of vacancies are not referred to. 2 West Germany 46 percent, East Germany 34 percent in Actual data are not available. Institute for Employment Research Nuremberg. 7

8 Summarizing: A key reason for restrictions in empirical work lies in the lack of reliable data on total job vacancies. As Blanchard 2006, p. 30 writes: It has proven hard to learn much from the shifts in the Beveridge curve across countries; one reason may be that data on vacancies are often of poor quality... The following section will describe why data on the total number of vacancies have rarely been available in the past and which efforts the European countries actually take to collect and provide them. 2. Job Vacancy Surveys in Europe 2.1. The history Still in the 60ies European labour markets were characterized by nearly full employment and firms complains about labour shortages. The question how to bring unemployed back to the labour market did not gain the relevance that it became later. Countries built up administrative databases on registered vacancies, information on the total number was not available (see examples in NBER 1966). Beginning with the 70ies the global economy as well as the national labour markets changed rapidly. The downsizing of industrial mass production was followed by high unemployment rates mainly for persons with low vocational qualifications, and an increasing duration of unemployment. The international wage competition accelerated in the 1990ies and increased the pressure for structural economic changes in Europe. In nearly all countries an ongoing increase of unemployment with a simultaneous weak economic performance appeared (Blanchard 2006). The following questions occurred: 8

9 - How large is the total labour demand and what is its structure (by qualification, occupation, region and sector)? - Is there a mismatch between labour demand and supply that hinders firms from economic success and growing employment? What are the reasons for mismatch? - In which professions training and education need to be changed or to be intensified to meet the needs of employers? - Which qualifications should be provided to unemployed to enable them to meet the needs of employers? - Is there a need for inter-european mobility of the labour force to reduce bottlenecks and unemployment? Is there a need for immigration from outside Europe despite high unemployment rates in Europe? To answer these questions economists need to analyze the function of the labour market considering both market sides: supply and demand. But still in the middle of the 1990ies only few countries collected data on the unmet labour demand, like the Netherlands and Germany. The representativeness of registered vacancies for the total unmet labour demand became more and more disputable: The internet had brought up a new and cheap possibility for firms to search for potential employees, whereas traditional search ways lost importance The recent past In March 2002 the European countries agreed to provide national job vacancy data on a quarterly basis, by implementing new business surveys, using existing surveys or by using available administrative sources. 9

10 A Task Force and several working groups developed a definition of vacancies and discussed about the aggregation level of the data. This process was not a straightforward one, because the visions of the European Commission and the European Central Bank on disaggregation by sectors, occupations and regions were confronted by the large national cost burden of corresponding business surveys. Cost Benefit analyses showed, how much EURO an additional question in a survey would cost 3, how much the overall costs would increase if questions on occupation by ISCO would be included and if data should be disaggregated by regions. Because of the high cost burden countries finally agreed on providing only quarterly data on the total number of vacancies and the vacancy rates by 21 economic sectors (NACE Rev.2). The covering of regional and occupational breakdowns remains voluntary. Currently 4 the regulation endorsed by the majority of countries stands in the negotiation process at the European Parliament 5. According to this regulation quarterly data have to be provided from all countries from 2010 onwards. Currently all European countries except Denmark, Ireland, Malta and Austria collect data. Mostly business surveys are used, only Belgium, The Czech Republic, Luxembourg and Slovenia use administrative sources on registered vacancies which are taken as representative for the total number of vacancies. 6 countries don t publish their results, because they have high uncertainty about the data quality, 17 countries publish data at the EUROSTAT web page with different time lags to the reference quarter 6. The European Central Bank calculates the total number of job vacancies and an average of the vacancy rates on base of the available data and on estimations for those countries that don t provide data. The higher the percentage 3 As example: In case vacancies are divided by fixed termed and permanent not only one question on the number of vacancies is necessary but two questions. 4 April The regulation consists of two parts: a framework regulation and an implementing regulation

11 of missing countries the higher the inaccuracy in the European aggregates too. Table 1 shows these countries, that did not provide any data for vacancies and/or occupied posts between the first quarter 2001 and the fourth quarter 2007 (except those four who don t collect data). All other countries either provide data regularly or provided data for some quarters during this period of time. Measure Country job vacancies, total, quarterly and yearly Belgium, France, Italy, Hungary, Poland job vacancies, only 10 or more employees Belgium, France, Luxembourg, Hungary, Poland, UK Occupied posts Belgium, Denmark, France, Italy, Hungary, Netherlands, Poland Occupied posts, only 10 or more employees Belgium, France, Luxembourg, Hungary, Netherlands, Poland, United Kingdom Job rate (vacancy rate) France, Italy, Job rate (vacancy rate), only 10 or more employees Belgium, Luxembourg, Netherlands, United Kingdom Table 1: Countries not providing any data for quarterly surveys between 2001 and 2007 Source: EUROSTAT The published job vacancy rates of seven selected countries, either based on collected data for vacancies and occupied posts or based on estimations are shown in graph 2. The absolute differences between the countries are large, such as the variations over time. 11

12 4,5 4,0 3,5 3,0 vacancy rate (%) 2,5 2,0 1,5 1,0 0,5 0,0 I./2002 II./2002 III./2002 IV./2002 I./2003 II./2003 III./2003 IV./2003 I./2004 II./2004 III./2004 IV./2004 I./2005 II./2005 III./2005 IV./2005 I./2006 II./2006 III./2006 IV./2006 I./2007 II./2007 III./2007 IV./2007 Bulgaria Germany Netherlands Portugal Romania Finland United Kingdom Graph 2: Quarterly job vacancy rates published by EUROSTAT March At the time being these published national data are rarely comparable: vacancy definitions, sector and firm size covering, methods of sampling, weighting and stratification procedures differ widely. Main differences behind the published data as well as the implications of these differences will be discussed in the following section. 3. A comparison of job vacancy surveys in Europe 3.1. General remarks In the following we give an overview on differences between national job vacancy surveys in Europe. Official publications about the utilized methods are rarely available. Therefore we gleaned information out of internal papers of the Working Group on Labour Market Statistics 12

13 that regularly meets in Luxemburg at EUROSTAT and from unpublished documents discussed during the Task Force on Job Vacancy Statistics in which Germany participated. There is a large variety of quarterly survey methods: written questionnaires, questionnaires sent out by , telephone interviews, and telephone data entry. Most countries (12) have implemented an extra job vacancy survey, 11 countries added questions on job vacancies to an existing survey and 4 countries use administrative sources The definition of vacancies In the European regulation on quarterly job vacancy statistics the countries agreed on the following definition: A job vacancy is defined as a post newly created, unoccupied or about to become vacant for which the employer (i) is taking active steps to find a suitable candidate from outside the enterprise concerned and is prepared to take more steps and (ii) intends to fill either immediately or in the near future. The restrictions in (i) and (ii) are highly relevant for the use of vacancy data: Vacancies can be filled by unemployed or by employed job seekers. For the employed jobseekers these positions that shall be filled in the distant future, for instance in four or six months, can be of interest, for the unemployed these are less relevant. Labour market policy forces them to find a job as soon as possible, so they are interested to apply to real open positions, not to the long term vacancies. 13

14 Internal job vacancies are not of any relevance for the unemployed neither. The opposite is the case for the employed who want to improve their position and payment. In the long run an internal vacancy will induce an external vacancy, as long as the total employment of a firm is not decreasing or changing its professional structure. In the short run the firms unmet labour demand is not changing by such a job to job move. For economic matching analyses the usage of all vacancies would yield biased results, because a share of this unmet labour demand is not worth considering by the unemployed. Until now the term immediately or in the near future is not fleshed out in the European regulation on job vacancy statistics. Many countries include all vacancies, independently from the time horizon. Germany divides all vacancies into immediately to be filled and later to be filled. Greece and Spain only collect data on vacancies to be filled during the next three months. Some countries like Finland exclude short-term fixed-term vacancies. Internal vacancies are included by the Netherlands for instance. Without correction for the internal vacancies the Dutch job vacancy rate is not comparable to the rates of other countries which exclude the internal vacancies. The same arguments for comparisons with regard to the vacancy duration An open question is the covering of vacancies for subsidized positions. In principle they are part of the labour demand. Therefore they are covered in the German survey, but published extra. Without counting the subsidized vacancies mainly resulting from the labour market reform Hartz IV, the German vacancy rate would have increased between 2004 and 2005 only to 2.8, instead of 3.2. Information how the other countries deal with subsidized jobs is not available at the moment. 14

15 3.3. The coverage of firms and sectors The size structure of the economy should be represented in the allocation of the sample (see Särndal, Swensson and Wretman 1992, Cochran 1977). Small firms with less than 10 employees have a quite low probability of a vacancy in the reference period, therefore they need to be represented stronger, what increases the survey costs considerably. For instance in Germany small firms with less than 10 employees account for circa 32 percent of the total survey costs in the stage of data collection (shipment of questionnaires, carrying out telephone interviews). But if Germany would not include small firms in the survey, about 25 percent of all vacancies would not be covered. The high cost burden is one reason why some countries do not collect data for all firm size classes. A second reason arises from the fact that in some countries the job vacancy survey was added to an existing business survey that does not cover the very small firms at all. For instance Italy only includes establishments with 10 or more employees. The published vacancy rate therefore is not representative for the whole Italian economy and is not comparable with job vacancy rates in other countries that refer to all size classes. Following the regulation on job vacancy statistics all sectors except activities of households and extra-territorial organisations and bodies should be included 7. At the time being only sectors C to K (see abbreviations in Annex) are covered by the countries publishing data. For instance Portugal does not cover sectors A, B, L and the public bodies in sectors M, N and O.. The collection of compete data in education and health and in the public administration sector is difficult or impossible in numerous countries. Concerning the economic importance of these sectors an interpretation of data needs to be done with the utmost caution. 7 For countries with a share of employees in agriculture and fishing that counts less than 3 percent of the total employment, the coverage of these sectors is voluntary. 15

16 3.4. Reference dates The data collection method should ensure to yield results representative for the reference quarter. A continuous quarterly survey is done only by Germany, all other countries use specific reference dates, either one day in the quarter or three days (one in each month) for an average (see Table 2). There are disadvantages if one single reference day is used: First firms have to remember the number of vacancies for the reference day. The longer the time lag between the answering and the reference date the higher inaccuracies will occur in the answers. Second: one specific reference day in the quarter might not be representative for the quarter. For instance altogether 11 countries use the last day of the quarter as reference date. Depending on the choice of the reference day a different number of vacancies will be counted. See for example the number of vacancies in one of the internet job portals of the Federal Employment Agency in Germany (Graph 3). They count for about 20 percent of all vacancies 8. If the reference day would be the last day of the first month about vacancies would be counted, at the 15 th of February (the middle of the quarter) about and at the last day of the quarter The continuous survey would instead provide an average as well as a survey with several reference dates percent of these vacancies are not registered to the Public Employment Agencies. 16

17 One reference day: the last day of the middle month the 15th of the middle month the last day of the middle month average of the last day of each month the first day of the last month the first calendar day of the last month the last day of the reference quarter last working day of the last month the end of the reference quarter Country Cyprus Estonia Romania Luxembourg Greece Finland Netherlands, Poland, Portugal, Slovakia, Spain, Hungary, France, Bulgaria Italy, Latvia Lithuania Three reference days: Friday between the 2nd and 8th of each month the middle Wednesday of each the last day of each month the average of the 15th of each month United Kingdom Sweden Slovenia Czech Republic Continuous survey: continuous survey between the second week of the first month and the third week of the last month Germany Table 2: Reference dates in European Job Vacancy Surveys Source: unpublished document of the Task Force on Job Vacancy Statistics, 5-6 February 2007 in Luxembourg Graph 3: daily number of job vacancies in the internet job portal Job Roboter, quarter 1, 2008 Source: Public Employment Agency 17

18 3.5. Extrapolation methods and error calculation Only very few countries provide information on their methods for non-response-adjustment, imputation of missing values, and extrapolation. Even the available material is quite scanty and usually not officially published. Analyses on the effect of different approaches do not exist at all for job vacancy statistics. This hinders the evaluation of the accuracy of the results and possible estimation errors 9. In the regulation on job vacancy statistics countries agreed to provide information on the coefficient of variation (CV) for the total number of job vacancies and the job vacancy rates by economic sectors. Provisional calculations of the coefficient done by some countries were discussed in the Task Force on Job Vacancy statistics and showed two main problems. The calculation did not consider the specific sample design of representative job vacancy surveys, what might seriously over- or underestimate the CV (see Kiesl 2007). And the calculated CVs showed large differences between countries, but also between sectors in a country. Usually the European Commission targets a CV of 5 to 8 percent for official European statistics. This value could be achieved only in single cases. Most economic sectors showed CVs between 20 and 60 percent Implications of the differences At the time being the comparison of data coming from different national job vacancy surveys published by EUROSTAT should not be undertaken at all or only with greatest wariness. The available information at EUROSTAT s webpage does not allow a sufficient analysis of methodological differences and the evaluation of the implications for comparative studies. 9 Therefore EUROSTAT and the Institute for Employment Research are organizing a workshop for the statistical experts of job vacancy surveys for end

19 More and structured methodological information needs to be provided by the countries and published simultaneously with the vacancy data. Only data of good quality are useful for economic research and policy analysis. 4. What additional information is necessary to interpret vacancy data right? If the number of vacancies and the job vacancy rate shall be used as short term indicator for future employment like targeted by the European Commission, auxiliary information is needed to interpret the data in the right way: Replacement ratio Vacancies can be announced to replace employment or to increase it. The higher the replacement ratio in an economy, the lower the increase of the overall employment respectively the decrease of unemployment can be. The ratio is influenced by the age structure of the workforce, the percentage of female employees who break working for childcare and, especially important the percentage of temporary jobs. Vacancy duration, vacancies difficult to be filled The higher the percentage of vacancies that could not be filled in period t-1 because the vacancy duration extends as result of increasing frictions, the higher the number of vacancies in period t, given constant inflows. Without knowing the background, for instance about labour market reforms that change frictions, one could interpret this finding as the beginning of a boom, what in fact is not the case. The reverse also holds. Therefore the average vacancy durations as well as the number of vacancies difficult to be filled should be collected too. 19

20 Labour market policy, macroeconomic environment, Labour market policy can have a large influence on the unmet labour demand. The labour market reform 2005 in Germany is an example. It increased the number of subsidized jobs between 2004 and 2005 remarkably, thereby the total number of vacancies. This increase was not a precursor for an economic upturn as well as not an expression for an ineffective work of the public employment agencies. This can be shown with the German job vacancy survey, in which vacancies are divided between regular and subsidized vacancies. Knowledge about the overall business cycle is essential as well to avoid misinterpretation of vacancy data. 5. Summary This paper explained the relevance of data on the total number of job vacancies for economic research as well as for labour market policy. Most countries in Europe have been started collecting data, because registered vacancies are usually not representative for the total numbers. However, at the time being the available results of national job vacancy surveys are barely comparable. Not only the vacancy definition but also the covered sectors and firms and the utilized methods differ widely. Data should therefore not be used for comparisons of the labour market and labour market policy between countries, or should be used only with great caution concerning different methods and by use of auxiliary information on the national labour markets. 20

21 Appendix: abbreviations of the NACE classification Code Notation A, B Agriculture, farming of animals, hunting, forestry and fishing D Manufacturing C Extracting of crude petroleum and natural gas; Mining; Quarrying ; E F G H I Electricity, gas, steam and hot water supply Construction Sale, maintenance and repair of personal and household goods, cars and motorcycles; Wholesale trade and commission Hotels and Restaurants Transport and auxiliary activities, communication J K O M N L Financial intermediation, insurance, pension funding Real estate, renting and business activities, research and development Other community, social and personal service activities Education Health and social work Public administration, defence References: Blanchard, O. and Diamond, P. (1989): The Beveridge Curve. Brooking Papers on Economic Activity, 1, Blanchard, O. and Portugal, P. (2001): What hides behind an Unemployment Rate: Comparing Portuguese and U.S. Unemployment. American Economic Review, 91(1), Blanchard, O.J. (2006): European Unemployment: The Evolution of Facts and Ideas. Economic Policy, 21(45), Boeri, T. (1999): Enforcement on Employment Security Regulations, On-the-job-search and Unemployment Duration. European Economic Review, 43(1), Burdett, K. and Cunningham, E. (1998): Toward a Theory of Vacancies. Journal of Labor Economics, 16(3),

22 Burgess, S. (1993): A Model of Competition between Unemployed and Employed Job Searchers: An Application to the Unemployment Outflow Rate in Britain. Economic Journal, 103(420), Cochran, W. (1977): Sampling Techniques. 3rd ed. New York: Wiley (Wiley series in probability and mathematical statistics). Coles, M. and Smith E. (1998): Marketplaces and Matching. International Economic Review, 39(1), Fahr, R. and Sunde U. (2001): Strategic Hiring Behaviour in Empirical Matching Functions. Discussion Paper, Institute for the Study of Labour, No Fahr, R. and Sunde U. (2006): Spatial Mobility and Competition for Jobs: Some Theory and Evidence for Western Germany. Regional Science and Urban Economics, 36, Fisher, J. (2006): The Dynamic Effects of Neutral and Investment-Specific Technology Shocks. Journal of Political Economy, 114(3), Franz, W. and Smolny, W. (1994): The Measurement and Interpretation of Vacancy Data and the Dynamics of the Beveridge Curve: the German Case. In: NBER (1966), IW (2006): Offene Stellen: Ein deutsches Problem. IWD 27, Cologne. Jackman, R., Layard, R. and Pissarides (1989): On Vacancies. Oxford Bulletin of Economics and Statistics, 51(4), Jackman, R., Pissarides C. and Savouri S. (1990): Labour Market Policies and Unemployment in the OECD. Economic Policy, 11, Kettner, A. and Spitznagel, E. (2007): Gesamtwirtschaftliches Stellenangebot kräftige Konjunktur stärkt die Arbeitsnachfrage. IAB-Kurzbericht. 11, Nürnberg. Kiesl, H. (2007): Recapitulating some Facts about the Coefficient of Variation in the Context of the European Job Vacancy Survey. Presented at the Task Force on Job Vacancy Statistics, Luxembourg, 2-3 July. Layard, R., Nickell, S. and Jackman, R. (2005): Unemployment. Macroeconomic Performance and the Labour Market. Oxford: University Press. Lilien D. (1982): Sectoral Shifts and Cyclical Unemployment. Journal of Political Economy, 90 (4), Muysken, J. (Ed.) (1994): Measurement and Analysis of Job Vacancies. Avebury: Ashgate Publishing Limited. National Bureau of Economic Research (NBER) (1966): The Measurement and Interpretation of Job Vacancies. Columbia University Press, New York, London. Petrongolo, B. and Pissarides, C.A. (2001): Looking into the Black Box: A Survey of the Matching Function. Journal of Economic Literature, 39(2), Pissarides, C.A. (2000): Equilibrium Unemployment Theory. 2 nd edition, Cambridge. 22

23 Särndal, C.-E., Swensson, B. and Wretman, J. (1992): Model assisted survey sampling. New York: Springer (Springer series in statistics). Shimer, R. (2005): The Cyclical Behaviour of Equilibrium Unemployment and Vacancies. American Economic Review, 95(1), Shimer, R. (2007): Mismatch. American Economic Review, 97(4), Van Ours, J.C. and Ridder, G. (1991): Cyclical Variation in Vacancy Durations and Vacancy Flows. European Economic Review, 35(5), Van Ours, J.C. and Ridder, G. (1992): Vacancies and the Recruitment of New Employees. Journal of Labor Economics, 10(2), Weber, A. (2000): Vacancy Durations a Model for Employer s Search. Applied Economics, 32(8), Yashiv, E. (2000): The Determinants of Equilibrium Unemployment. American Economic Review. Vol. 90(5), Papers available for the Working Group on Labour Market Statistics at the European Commission; not published for the public: Godinho, M. (2004): Measuring Job Vacancies The Portuguese Experience, in: Workshop - EUROSTAT PHARE Multi - Country Programme. Vilnius. Klavers, T. (2004): The Quarterly Survey of Vacancies: Methods and Results. Centraal Bureau voor de Statistiek CBS. Machin, A. (2005): Summary of the UK Vacancy Survey, in Workshop - EUROSTAT PHARE Multi-Country Programme, Nicosia. Milosheva, E. (2005): Report on the Pilot Job Vacancy Survey, Budapest. Paun, C. (2005): Report on the Pilot Survey on Job Vacancies - Romania, in Workshop - EUROSTAT PHARE Multi-Country Programme, Nicosia. Statistics Finland (2005): Occupational Titles of the Vacancies and Calculation of Vacancy Rates in Job Vacancy Survey. Taskinen, K. (2004): Job Vacancy Survey - The Finnish Experience, in Workshop - EUROSTAT PHARE Multi - Country Programme. Vilnius. 23