IMMIGRANTS IN THE ISRAELI HI- TECH INDUSTRY: COMPARISON TO NATIVES AND THE EFFECT OF TRAINING

B2v8:0f XML:ver::0: RLEC V024 : 2400 /0/0 :4 Prod:Type:com pp:2ðcol:fig::nilþ ED:SeemaA:P PAGN: SCAN: 2 IMMIGRANTS IN THE ISRAELI HI- TECH INDUSTRY: COMPARISON TO NATIVES AND THE EFFECT OF TRAINING Sarit Cohen-Goldner ABSTRACT During the 0s, the Israeli economy experienced two major events. First, starting in the fall of 8, a large wave of relatively highly skilled immigrants arrived from the former Soviet Union (CIS) increasing the population and the labor force by considerable magnitude. Second, the hitech sector has grown substantially and reached a peak in growth and level in 2000. This paper provides a descriptive analysis of the integration of immigrants from the CIS in the Israeli labor market and, specifically, in the hitech sector. Based on a unique panel data that follows immigrants for up to 2 years in Israel we find a significant positive correlation between immigrants participation in Israeli government-provided training programs and the propensity to work as professionals in the hi-tech industry and to work in white-collar occupations in other sectors. However, this correlation diminishes with time since participation such that recent participants face a higher probability to work in hi-tech and white-collar jobs than those who participated in training earlier. Research in Labor Economics: The Economics of Immigration and Social Diversity Research in Labor Economics, Volume 24, Copyright r 200 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 04-2/doi:0.06/S04-2(0)2400-8

2 SARIT COHEN-GOLDNER 2. INTRODUCTION During the decade that started in 0, the Israeli economy experienced two major events. First, starting in the fall of 8, a large wave of relatively highly skilled immigrants arrived from the former Soviet Union (CIS), increasing the population and the labor force by considerable magnitude. Second, the hi-tech (HT) industry had grown substantially and reached a peak in growth and level in 2000. In this study, we try to explore the interaction between the growth of the HT industry and the integration of CIS immigrants in the Israeli economy and, specifically, in the HT industry. In particular, we compare the integration of immigrants in the HT sector to their integration in other occupations as well as to the presence of natives in this industry. Furthermore, since the immigrants were offered governmentsponsored vocational training, we study the determinants of training attendance and the role of training in the process of labor market employment and occupational dynamics of the immigrants. The analysis is based on two sources of data that provide the best description of the Israeli labor market dynamics. First, we use annual crosssectional Labor Force Survey (LFS) of 0 2000. The LFS is conducted by the Central Bureau of Statistics (CBS) and provides a very good aggregate description of the labor force distribution by occupation, industry, years since arrival, country of birth, gender and age. In addition, we are the first to use a unique panel data that tracks a sample of immigrants for a period of up to 2 years since their arrival. All immigrants in this sample studied engineering in the CIS (hereafter engineers survey ). The two databases, therefore, differ in their structure (cross section vs. panel) and in the investigated population (general population of CIS immigrants in the LFS vs. immigrant engineers in the engineers survey). In both surveys the analysis is restricted to CIS immigrants who arrived in Israel between 8 and 4. The cross-sectional data is used mainly for the comparison between immigrants and native Israelis, while the engineers panel data is used to describe the patterns of training attendance among immigrants and the transitions of immigrants between various labor market states. In particular, the engineers survey enables us to study the role of training in the occupational choice of immigrant engineers. Cohen-Goldner and Eckstein (2002, 2004) used a panel data on CIS immigrants during their first five years in Israel and found that the slow transition to white-collar (WC) occupations stems from the low availability of WC job offers. However, participation in

Immigrants in the Israeli Hi-Tech Industry 2 Israeli government-subsidized training had a substantial positive impact on the transition of male and female immigrants to WC occupations. Based on the cross-sectional data, we find that the integration of immigrants in the HT industry is similar to that of native Israelis. Furthermore, the share of immigrants who work in high-skill occupations in the HT industry is roughly the same as that of natives. One of the explanations to this phenomenon is that the Israeli HT industry grew parallel to the arrival of immigrants in the 0s, such that it absorbed new workers both among natives and immigrants. It is not clear, however, that this observation implies that immigrants integration in the HT sector was due to a good match between their imported human capital and the skills demanded by this specific industry. One of the novel aspects of the engineers survey is the detailed information on the participation of immigrants in government-sponsored vocational training programs. These programs were offered by The Ministry of Labor and The Ministry of Absorption and many programs were especially designed for the specific population of highly skilled immigrants. The courses were offered in software engineering, programming, electronic engineering, computers, etc. The average length of these programs was approximately 6 months, which is substantially longer than the average months length of classroom training in the USA. Approximately, 40% of the males in the engineers survey have participated in government-subsidized vocational training since their arrival. These participants are, on average, younger at arrival and a higher share of them worked in WC occupations in CIS. Females participation rate in training was 8%, and similar to males, female participants were, on average, younger at arrival and most of them worked in the CIS in WC occupations or as professionals in the HT industry. To explore the participation of immigrants in training programs and, in particular, the timing of participation, we estimate a hazard regression for time in Israel until participation in training (duration to training). We find that there is no significant difference in the duration to training of males and females. For males, the duration to training increases with age at arrival, while the knowledge of Hebrew and work in WC occupations before migration lead to a shorter duration. For females, the duration to training also decreases with the knowledge of Hebrew and with the number of children aged less than 8. The most striking effects on the duration to training are the changes in previous labor market states. For example, immigrants who moved from employment to unemployment during the 6 months prior to the training program face a substantial shorter duration to training. These

4 SARIT COHEN-GOLDNER 2 findings suggest that the participation of immigrants in training is not motivated solely by considerations of investment in local human capital, but rather immigrants take advantage of the opportunity to attend training after spells of unemployment. We further estimate a Cox hazard regression for time until work as a professional in the HT industry and find that females face a significantly longer duration. Immigrants who are proficient in English face a shorter duration to work as professionals in the HT industry, reflecting the international orientation of the Israeli HT industry. Participation in training leads to a longer duration to HT, as the length of the Israeli training programs is usually around 6 months. Imported education (years of schooling) does not have a significant effect on the duration to work as a professional in the HT industry, possibly due to the low variance of this variable in the engineers survey. A Multinomial-Logit analysis for the occupational choice of immigrant engineers shows that participation in training significantly (at 0% level) increases the propensity to work as a professional in HT and to work in WC occupations in other sectors, but this effect declines with time since participation, such that recent participants face a higher probability than participants who participated in training earlier. This result suggests that the accumulated knowledge in the training programs is subject to depreciation and, therefore, standard before after estimates for the impact of training are sensitive to the time intervals chosen. The rest of the paper is organized as follows. In the next section, we provide the cross-sectional descriptive analysis and in the third section we analyze the panel data. Section 4 concludes. 2. IMMIGRANTS AND NATIVES IN THE LABOR MARKET: CROSS-SECTIONAL DATA In this section, we describe the labor market characteristics of CIS immigrants who arrived in Israel in 8 4 and of native Israelis, with focus on the HT industry. 2 The analysis is based on the Israeli national crosssection labor force survey, which is conducted annually by the Israeli CBS among approximately,000 households. Eckstein and Weiss (2002, 2004) provide an extensive analysis of the integration process of the immigrants and a comparison to native Israelis. In this section, we build on the above papers and extend the analysis to the integration of immigrants and native

Immigrants in the Israeli Hi-Tech Industry 2 Israelis in the HT industry while in the next section we explore the panel data and incorporate training into the analysis. In order to describe the labor market dynamics of immigrants and natives we define three occupational categories: WC occupations which include professionals and managers such as engineers, physicians, teachers, nurses, etc.; Blue-collar occupations (BC) include sales agents, electricians, etc. and HT occupations in the HT industry (see Appendix (B) for the list of economic branches which comprise the HT industry and the list of three-digit occupations, which are considered as HT occupations within the HT industry). 4 Fig. presents the aggregate proportion of immigrants employment in WC, HT and BC occupations and of immigrants unemployment. The figure demonstrates the rapid decline in unemployment and the fast increase in employment in BC occupations. Furthermore, after five years in Israel there is a gradual shift from employment in BC jobs to WC occupations. These transitions are explained by Cohen-Goldner and Eckstein (2002, 2004), who showed that as immigrants accumulate Israeli human capital via language acquisition, vocational training and on the job learning, they are able to shift to better jobs. Fig. also indicates that there is no substantial difference in the integration of male and female immigrants in WC occupations. After 0 years in Israel 2% (4%) of male (female) immigrants work in WC occupations. The integration of immigrants in WC occupations is a gradual process and the share of immigrants in these occupations was substantially lower than that of native Israelis during the 0s (Eckstein & Weiss, 2004). Cohen- Goldner and Eckstein (2002, 2004) found that the slow and low entrance of immigrants to WC is due to low availability of job offers in these occupations. Participation in training, however, accelerated these offer probabilities, considerably. In the next section we investigate the role of training in the integration of immigrant engineers in the HT industry. The rapid growth of the HT industry occurred almost parallel to the arrival of immigrants from the CIS. During 2000 two main factors led to the growth of employment in the HT industry: new entrants to the labor market (mainly native Israelis who graduated from college) and immigrants. As most of these immigrants were college graduates and a sizable share of them worked in the CIS as engineers, it is challenging to study the integration of this population in the growing HT industry. Fig. 2 presents the share of employed in the HT industry among the general population (natives and immigrants) and among natives and immigrants, separately. The figure demonstrates that there are only minor dif-

6 SARIT COHEN-GOLDNER a. Males 0.8 0. Hi-Tech Occupations White Collar Occupations Blue Collar Occupations Unemployed 0.6 percentage 0. 0.4 0. 0.2 0. 0 0 2 4 6 8 0 years since arrival b. Females 0.8 0. Hi-Tech Occupations White Collar Occupations Blue Collar Occupations Unemployed 2 0.6 percentage 0. 0.4 0. 0.2 0. 0 0 2 4 6 8 0 years since arrival Fig.. Labor Force Employment and Unemployment of Immigrants. Source: LFS 2000. (a) Males (b) Females. ferences between immigrants and natives, such that employment in the HT industry is an identical share from the reference groups. Furthermore, Fig. shows that the share of male immigrants in HT industry is roughly the same as that of native males, while the share of female immigrants is higher than that of native females. 6

Immigrants in the Israeli Hi-Tech Industry 2 percentage 0.4 0.2 0. 0.08 0.06 0.04 all population natives 0.02 all immigrants immigrants after year 8 0 2 4 6 8 2000 year 2. % 2.86 % 2.08 %.8 % Fig. 2. Employment in the Hi-Tech Industry (Present). Source: LFS 2000. percentage Fig.. 0. 0.2 0. 0. 0.0 male natives male immigrants female natives female immigrants 0 2 4 6 8 2000 year. % 4. % 0.4 %.2 % Employment in the Hi-Tech Industry by Gender and Origin (Persent). Source: LFS 2000. Since not all workers in the HT industry work in high-skilled occupations, we divide workers within the HT industry to workers in HT occupations (e.g., physicists, system analysts, electronic engineers, etc.) and in non-ht occupations (see Appendix (B) for further details on HT-related occupa-

8 SARIT COHEN-GOLDNER 2 percentage 0.6 0. 0. 0.4 0.4 0. 0. 0. HT occupations - natives HT occupations - immigrants 0.2 2 4 6 8 2000 year 48. % 4.6 % Fig. 4. Employment in the Hi-Tech Industry by Occupation. Source: LFS 2000. tions). The majority of workers in the HT industry work in non-ht occupations. However, as Fig. 4 shows, the total share of workers in HT occupations grew from % to 4% through the 0s. Furthermore, Fig. 4 shows that the share and trend of immigrants who work in HT occupations within the industry is roughly the same as the share among natives. The overall composition of employment in the HT industry by occupation and origin is presented in Fig.. We see that immigrants comprise about 20% of the industry labor force in 2000 (0% work in HT occupations and 0% work in non-ht occupations). As previously pointed out in Fig. 4, the labor composition in the industry has changed, such that the share of workers in HT occupations increases. We can summarize that unlike the slow integration of immigrants compared to natives in WC occupations, documented by Eckstein and Weiss (2004), we find that the integration of immigrants in the HT industry as well as their occupational distribution within the industry is very similar to that of native Israelis. One of the explanations to this phenomenon is that the Israeli HT industry grew parallel to the arrival of immigrants in the 0s, such that it absorbed new workers both among natives and immigrants. Yet, it is not clear whether this observation implies that immigrants integration in the HT industry was due to a good match between their human capital and the skills demanded by the HT industry.

Immigrants in the Israeli Hi-Tech Industry 0.6 0. percentage 0.4 0. 0.2 non HT occupations - immigrants HT occupations - immigrants non HT occupations - natives HT occupations - natives 4.46 % 6.4 % 2 0. Fig.. 0 2 4 6 8 2000 year 0. %.6 % Composition of the Hi-Tech Industry by Occupations and Origin. Source: LFS 2000. Given the fact that immigrants were about 0% of the new entrants to the Israeli labor market during 0 2, and considering the rapid growth of the HT industry, one could expect a larger share of immigrants in the HT industry. On the other hand, the immigrants relatively higher rate of integration (compared to natives) into the HT industry and, in particular, to HT occupations indicates that there was a positive match between the immigrants and the sector demand for labor. The question we face now is whether government-sponsored training had any influence on these observations.. OCCUPATIONAL DISTRIBUTION AND TRAINING ATTENDANCE AMONG IMMIGRANT ENGINEERS: PANEL DATA In order to investigate the role of training in the integration of immigrants in HT, we use a unique panel data on immigrant engineers. The engineers survey is based on two interviews. The Brookdale Institute conducted the first interview in, and the second interview of the same sample was conducted in 200 2002. 8 The surveys included men and women who had immigrated during 8 4 from the CIS, and reported that they have an

0 SARIT COHEN-GOLDNER 2 engineering diploma when entering Israel. In all, 42 individuals (824 males and 608 females) were interviewed in, and individuals of the original sample (4 males and 20 females) were interviewed in 200 2002. The current study focuses on 446 men and 04 women aged between 24 and 60 years at arrival in Israel who actively looked for a job at some stage since arrival. The surveys include information on: age, years of schooling, country of origin, occupation in the country of origin, knowledge of Hebrew before immigration, marital status, size of family, etc. The two interviews enable us to construct a work-history profile of immigrants from time of arrival in Israel until the date of the second interview (200 2002), that is, a maximum period of 2 years since arrival. For each job that the immigrant worked in Israel, information is available on wages, starting and ending dates, the weekly working hours, occupation and industry. The surveys also provide detailed information on immigrants participation in Israeli governmentsponsored vocational classroom training and on their participation in Hebrew classes ( Ulpan in Hebrew). Each immigrant who arrived from the CIS during the 0s received an absorption package that included a set of monetary and non-monetary benefits. One of these benefits was the eligibility to participate in a government-sponsored vocational training program. These programs were offered by The Ministry of Labor and The Ministry of Absorption and many of them were especially designed for the specific population of highly skilled immigrants. The courses were offered in software engineering, programming, electronic engineering, computers, etc. Most of the male and female immigrants who participated in training, attended courses which lasted for four months or more. The average weekly hours of these programs was about 22 hours. Despite the long duration of the Israeli training programs, less than % of the participants dropped out from training. In the previous section (cross-sectional data) we made a distinction between workers in HT and in non-ht occupations within the HT industry. In this section, however, we distinguish between professionals and non-professionals in the HT industry, where professionals refer to workers who work in WC occupations in the HT industry. The change in definition is due to the lack of three-digit occupations classification in the engineers surveys. 0 Table provides summary statistics of the engineers survey for males and females. The average age at arrival of males (Panel A, col. ) is almost 42 and the average years of imported schooling is 6.. Since all the immigrants in this sample studied engineering, the variance of education is very low.

Immigrants in the Israeli Hi-Tech Industry Table. Summary Statistics Engineers Survey. Variables Entire Sample Participated in Training Course Non-Participants in Training Course 2 (A) Males Number of 446 26 observations Age on arrival in 4. 40. 42. Israel (years) (8.) a (8.2) (.) Education (years) 6. 6.4 6. (.) (.6) (.) Worked in WC in.0 4.. CIS (%) Worked as a 6.6 6.2 6. professional in HT in CIS (%) Hebrew knowledge 0.8. 0. before immigration (%) Number of children 0.6 0.2 0.60 (0.2) (0.2) (0.2) Married (%).0 8.4 2. Time in Israel at 22.2 22.4 22.0 latest survey (6.) (6.8) (6.) (months) Number of jobs in 2.. 2. Israel since arrival (.) (.) (.) Unemployed 0.2 0.6 0.0 throughout entire sample period (%) Time from arrival to. 8.2. first job (months) (.) (.0) (6.) Time from arrival to start of training course (months) (B) Females.2 (4.) Number of 04 6 88 observations Age on arrival in 4..8 42.2 Israel (years) (8.6) (8.0) (8.8) Education (years) 6.0. 6. (.4) (.4) (.4) Worked in WC in CIS (%) 8.8 4.4.2

2 SARIT COHEN-GOLDNER Table. (Continued ) Variables Entire Sample Participated in Training Course Non-Participants in Training Course 2 Worked as a 24.0 28.4 2. professional in HT in CIS (%) Hebrew knowledge 0.66 0.86 0. before immigration (%) Number of children 0.46 0. 0.8 (0.) (0.8) (0.) Married (%) 4. 6. 2. Time in Israel at latest survey (months) Number of jobs in Israel since arrival Unemployed throughout entire sample period (%) Time from arrival to first job (months) Time from arrival to start of training course (months) 20. 20. 2.0 (6.) (.2) (6.2) 2.4.0 2.0 (.4) (.) (.) 2.0 0.0.2.2.. (.2) (4.2) (.) 4. (.) Source: Engineers survey. a Standard deviation in parentheses. QA :2 About % of the males worked in WC jobs prior to immigration and 6% were employed as professionals in the HT industry in the CIS. The mean duration to first job in Israel is. months. Only 0.2% of the male engineers were unemployed through the entire sample period, and all of them have participated in training at some point since arrival. Around 40% ( ¼ / 446) of all males have participated in government-subsidized vocational training since their arrival. These participants are, on average, younger at arrival and a higher share of them worked in WC in CIS (Panel A, col. 2 ). The average age on arrival of females (Panel B) is 4. and the average years of schooling is 6.. Almost % of the females were employed in WC occupations in the CIS and 24% were employed as professionals in HT prior to migration. The average duration of females to the first job in Israel is months, which is twice the duration of males. Females participation rate in training is 8%. Similar to males, female participants are, on average,

Immigrants in the Israeli Hi-Tech Industry 2 younger at arrival and most of them worked in the CIS in WC occupations or as professionals in HT. Fig. 6 describes the dynamics of immigrants employment by occupations, unemployment and participation in training for males (Fig. 6a6) and females (Fig. 6b6). The general trends derived from the engineers panel data are not much different from those obtained from the cross-sectional LFS (Fig. ), though the levels are different. Unemployment among immigrant engineers declines sharply and employment in BC occupations increases rapidly during the first two years since arrival. The transition to WC occupations is gradual and steady since arrival and up to 0 years later. The shift of male engineers to WC jobs is more rapid and occurs earlier than the shift of female engineers, while according to the cross-sectional data there is no substantial difference in the integration of male and female immigrants in WC occupations. The share of male (female) engineers who work as professionals in HT grows slowly and reaches % (%) after 0 years. The occupational integration of engineers in the labor market is faster and different than that of the general population of CIS immigrants of the same cohort. After 0 years in Israel 0% (%) of engineer males (females) work in WC occupations, compared to 2% (4%) among the general population of male (female) immigrants. On the other hand, the unemployment rate of engineers is higher than the unemployment rate of the immigrants population, mainly among females. 2 This finding may indicate that immigrant engineers are more selective than the general population of immigrants and invest more in job-search while unemployed. Cohen-Goldner and Eckstein (2002, 2004) find a similar result for highly educated immigrants. Fig. 6 also shows that there is no trend in the participation of male engineers in training programs. However, the patterns of training attendance of females is consistent with the theory of investment in human capital, as their participation rate in training is higher close to their arrival and it declines with time spent in Israel... Participation in Training and Its Timing Table 2 presents the number of participants in the Israeli training programs by occupation in the CIS and by occupation in the training programs. Based on the occupation that the immigrant studied in the program, we distinguish between HT-related training and non-ht-related training (see Appendix (C) for the list of training programs which are HT-related). The table shows that both among males and females, training attendance does not vary consid-

4 SARIT COHEN-GOLDNER A. Males 00% 0% 80% 0% Percentage 60% 0% 40% 0% 20% Professionals in HT White-Collar Blue-Collar training unemployed 0% 0% 4 4 6 6 8 0 0 2 months since arrival B. Females 00% 2 0% Percentage 80% 0% 60% 0% 40% Professionals in HT White-Collar Blue-Collar training unemployed 0% 20% 0% 0% Fig. 6. 4 4 6 8 8 0 2 months since arrival Labor Force Composition Immigrant Engineers. Source: Brookdale Engineers surveys (A) Males (B) Females.

Immigrants in the Israeli Hi-Tech Industry Table 2. Participation in Training by Type and Occupation in the CIS a. Type of Occupation in CIS Type of Course Did not Participate in Training Course Hi-tech related Not-Hi-tech related Total 2 (A) Males Hi-tech professional 4 4 4 (20.) (8.2) (60.8) (00) White collar (non-ht) 22 24 4 00 (22.0) (24.0) (4.0) (00) Blue collar 4 62 64 26 (.6) (.22) (6.42) (00) Did not work in CIS 0 4 (0.0) (20.0) (80.0) (00) (B) Females Hi-tech professional 6 40 (.) (2.2) (4.) (00) White collar (non-ht) 4 0 4 (8.8) (24.44) (66.6) (00) Blue collar 2 46 84 (.4) (.0) (6.) (00) Did not work in CIS 0 2 (0.0) (0.0) (0.0) (00) Source: Engineers survey. a Values within parentheses are percent of the total in the row. erably with respect to the occupation held in the CIS. However, among immigrants who worked in BC occupations before migration, there is a higher tendency to attend non-ht training programs. In Table, we present the number of transitions between the occupation in the last job prior to training and the occupation in the first job following training. It is interesting to note that for both males and females, everyone who was unemployed before training, found a job after attending the training course. However, most of these jobs were in BC occupations. Among the training participants, the share of males employed in BC occupations declined from 6.4% before training to 4.% in the first job after training, while the share of employed in WC occupations grew from 28.% to 44%. Among female participants, the share of employed in BC occupations did

6 SARIT COHEN-GOLDNER Table. Occupational Transitions after Training a. Occupation in First-Job after Training Professional in HT White collar (not in HT) Blue collar Unemployed after Training Total 2 Occupation in last job prior to training (A) Males Professional in HT 0 (0.00) (60.00) (20.00) (20.00) White collar (non 2 4 HT) (.2) (88.24) (.88) (.6) Blue collar 28 68 4 0 (0.) (.2) (6.) (.6) Unemployed before 22 training (4.) (.64) (.26) (4.) Total 4 8 (B) Females Professional in HT 0 0 2 (0.00) (0.00) (0.00) (0.00) White collar (non 0 2 8 HT) (0.00) (66.66) (.8) (.6) Blue collar 4 4 6 (.) (22.) (6.6) (4.48) Unemployed before 0 8 0 training (.4) (4.48) (62.0) (0.00) Total 8 6 4 6 Source: Engineers survey. a Actual numbers. Values within parentheses are percent of the total in the row. not change due to participation, but the share of employed in WC jobs has doubled from.% before participation to 2.8% afterwards. This increase is mainly due to the reduction of unemployed females after training attendance. In order to describe the dynamic decision to participate in training, conditional of observed state variables, we run a Cox hazard rate regression for the duration to training. The hazard function has the form: HðtÞ ¼H 0 ðtþ expðx 0 bþ where the dependent variable H(t) is duration in Israel until training (in

Immigrants in the Israeli Hi-Tech Industry 2 Table 4. Cox Hazard Regression for Time until Participation in Training. Males Females All Dummy for females 0.842 (0.0) Hebrew. (0.).82 (0.26).408 (0.0) Number of children.6 (0.22) Years of schooling.0 (0.04) 0.6 (0.06).02 (0.0) Age on arrival 0.84 (0.00).08 (0.08) 0.88 (0.008) Worked in WC in CIS.42 (0.24) 0.82 (0.8).8 (0.6) Experience in WC.002 (0.008).00 (0.00).00 (0.00) Experience as a professional in HT 0. (0.02).00 (0.0).004 (0.00) Experience in BC.008 (0.00).008 (0.00).0 (0.00) Employed unemployed 6.26 (.60) 4.08 (.6).0 (.) Unemployed employed 0.68 (0.) 0.606 (0.) 0.28 (0.248) Unemployed unemployed.8 (0.2).2 (0.44) 2.68 (0.84) Log likelihood 0.44 2. 4.8 Source: Engineers survey. Significant at % level. Significant at 0% level. months), H 0 (t) the baseline hazard and x a vector of the state variables. This regression corrects for right censoring, as not all the immigrants have participated in training during the sample period. The results of the regressions for male and female immigrants separately and jointly are presented in Table 4. The panel structure of our data allows us to study the interactions between the dynamic decisions if and when to attend training and the dynamic employment decisions. To capture the possibility that the timing of training attendance is closely related to several labor market transitions between states, we include in x indicators related to transitions between different labor market states prior to training. The variable Employed Unemployed equals if the immigrant has moved from employment to unemployment during the 6 months prior to training, and 0 otherwise. Similarly, the variable Unemployed Employed equals if the immigrant moved from unemployment to employment during the 6 months prior to training, and 0 otherwise. The variable Unemployed Unemployed indicates that the immigrant was unemployed all through the 6 months prior to training. Hence, the reference group consists of immigrants who were employed all through the 6 months. 4

8 SARIT COHEN-GOLDNER 2 According to the joint regression there is no significant difference between the duration to training of males and females. For males, the duration to training increases with age at arrival while the knowledge of Hebrew and work in WC occupation before migration lead to a shorter duration to training. For females, the duration to training decreases with the knowledge of Hebrew and with the number of children aged less than 8 years of age. Previous occupation-specific accumulated experience does not significantly affect duration to training. The most striking effect on the duration to training is the impact of previous labor market transitions. Immigrants (males and females) who moved from employment to unemployment during the six months prior to the program have a substantial shorter duration to training. The same holds for immigrants who were constantly unemployed during the six months prior to the program. On the other hand, immigrants who moved from unemployment to employment during this period have a longer duration to training, though this effect is not significant. The finding that changes in the labor market state provide the strongest predictors for the timing of training implies that the participation of immigrants in training is not motivated solely by considerations of investment in local human capital, but rather immigrants take advantage of the opportunity to attend training and avoid temporarily unemployment..2. The Entry of Immigrants to Hi-Tech Table presents the estimates from a Cox hazard regression for the duration until the first entry to the HT industry as a professional. The dependent variable is time in Israel until work as professional in HT (in months) and the regression corrects for right censoring. The hazard model is estimated for male and female immigrants separately and jointly. From the joint regression it turns that females face a significantly longer duration to HT. English proficiency leads to a shorter duration to HT, while the knowledge of Hebrew has no significant effect. These results reflect the fact that many Israeli HT companies have an international orientation and some of them work closely with HT companies in the USA. Years of schooling do not affect the duration to HT significantly. Surprisingly, immigrants who were younger at arrival and immigrants who worked in WC occupations in the CIS have a longer duration to HT. This might result from the fact that these immigrants (i.e., younger and those who worked in WC in the CIS) tend to first participate in training and only later are integrated in high-skilled WC and HT occupations. Work in

Immigrants in the Israeli Hi-Tech Industry Table. Cox Hazard Regression for Time until Work as a Professional in Hi-Tech. Males Females All 2 Dummy for females 0.6 (0.68) Hebrew. (0.40) 2.68 (.).42 (0.40) English. (0.8) 0.80 (0.).8 (0.2) Years of schooling.04 (0.04).08 (0.26).06 (0.08) Age on arrival 0.4 (0.022) 0.6 (0.0) 0. (0.0) Worked in WC in CIS 0. (0.88) 0.26 (0.86) 0.260 (0.8) Worked as a professional in HT.42 (0.608) 2.4 (.04).04 (0.4) in CIS Participated in training 2.06 (.24) 2.406 (8.2) 4.4 (.66) Log likelihood 80.4 80.846 00.868 Source: Engineers survey. Significant at % level. Significant at 0% level. HT occupation in the CIS and participation in training shortens the duration to HT... Training and the Occupational Choice of Immigrant Engineers To study the role of imported and local human capital, and specifically of training on the labor market absorption of immigrant engineers, we run multinomial-logit regressions for the three occupational employment states (BC, WC and HT professionals) and unemployment as a function of various human capital variables. As the decision to participate in training is interrelated with the occupational choice, the multinomial-logit regressions provide correlation between the explanatory variable (i.e., training) and the occupational choices of the immigrants. Hence, one should not interpret the results as causal effects. In order to study the causal effect of training on occupational choices one needs to specify a model for the decision to participate in training and its affect on the occupation chosen. Previous work by Cohen-Goldner and Eckstein (2002, 2004) suggests that the simple correlations obtained from simple multinomial-logit regression retained after controlling for the selectivity to training and occupations. Table 6 presents the estimates obtained from two specifications of the explanatory variables. The regressions are pooled over time, such that each

20 SARIT COHEN-GOLDNER 2 Table 6. Occupational Choice and Unemployment of Immigrant Engineers. Hi-Tech Professionals (A) First specification (Log likelihood: 40.4) White-Collar Occupations Unemployed Dummy for females 0.248 (0.6) 0.6 (0.4) 0.6 (0.22) Hebrew 0.82 (0.2) 0.8 (0.22) 0. (0.0) English 0.6 (0.) 0.062 (0.04) 0.02 (0.066) Age on arrival 0. (0.0) 0.06 (0.0) 0.00 (0.00) Years of schooling 0.04 (0.0) 0.0 (0.04) 0.0 (0.04) Participated in training. (0.8). (0.24) 0.8 (0.) Years since training 0.4 (0.82) 0.406 (0.02) 0.0 (0.048) Experience as a professional 0.2 (0.04) 0.0 (0.08) 0.06 (0.0) in HT Experience as a professional 0.002 (0.000) 0.00 (0.000) 0.00 (0.0002) in HT-squared Experience in WC 0.6 (0.0) 0.26 (0.022) 0.08 (0.02) Experience in WC-squared 0.00 (0.0002) 0.002 ( 0.0002) 0.00 (0.000) Constant.4 (2.04) 2.4 (0.866) 0.46 (0.) (B) Second specification (Log likelihood: 6.0) Dummy for females 0.2 (0.6) 0.44 (0.6) 0.60 (0.) Hebrew 0.44 (0.) 0.4 (0.) 0. (0.088) English 0.24 (0.2) 0.46 (0.08) 0.08 (0.068) Age on arrival 0.08 (0.0) 0.0 (0.0) 0.0002 (0.00) Years of schooling 0.08 (0.0) 0.00 (0.04) 0.024 (0.042) Participated in training 0. (0.44) 0.446 (0.) 0.24 (0.20) Years since training 0.0 (0.02) 0.06 (0.0) 0.04 (0.0) Time in Israel 0.0 (0.00) 0.02 (0.00) 0.008 (0.002) Constant.48 (.82). (0.4) 0. (0.8) The reference group consists of immigrants who work in BC occupations. Source: Engineers survey. immigrant appears in each regression the number of months she/he appears in the sample. The comparison group is employment in BC occupations. Both specifications include indicator for females, years of schooling and age at arrival. The levels of Hebrew and English proficiency both range from (no knowledge) to 4 (perfect knowledge). To capture the possibility that the impact of training on the four potential outcomes changes over time, the regressions include the indicator for training participation as well as the interaction of training participation with years since training. In the first regression, we also include variables for the accumulated experience in HT

Immigrants in the Israeli Hi-Tech Industry 2 2 and WC occupations, but as these variables are likely to be endogenous, we replace them in the second specification with the variable time in Israel. According to the first specification, the propensity to work as a professional in HT decreases significantly with age at arrival and increases with previous accumulated experience in WC jobs or as a professional in HT. There is no significant difference in the propensity of females to work as a professional in HT compared to that of males. Hebrew proficiency does not appear to have a significant effect on the probability to work as a professional in HT, but proficiency in English does increase this probability significantly. Training significantly (at 0% level) increases the propensity to work as a professional in HT, but this effect declines with time since participation, such that recent participants face a higher probability than participants who participated years ago. This result suggests that the knowledge that is accumulated in the training programs is subject to depreciation and, therefore, before after estimates for the impact of training are sensitive to the choice of time since participation. The probability to work in WC jobs also increases with accumulated experience in WC jobs or as a professional in HT. Hebrew knowledge also has a positive significant effect on work in WC jobs. Like in HT, training has a positive impact on work in WC and this impact decreases with time since participation in the training program. Last, the probability of being unemployed also increases with accumulated experience in WC jobs or as a professional in HT, which suggest that immigrants with higher levels of human capital invest more in search from unemployment. Proficiency in Hebrew lowers the probability to be unemployed (compared to work in BC jobs). Females seem to have a significantly higher propensity than males to be unemployed. As was documented in previous studies on Russian immigration to Israel (Eckstein & Weiss, 2004; Cohen-Goldner & Eckstein, 2002, 2004; Weiss, Sauer, & Gotlibovski, 200), we find that conditional on local accumulated human capital, imported schooling has no significant impact on local labor market activities of Russian immigrants. To illustrate the impact of training on the probability to work as a professional in HT, we present in Fig. this probability conditional on the participation in training in the first year in Israel, and as a function of experience as a professional in HT for the average male/female immigrant engineer. 6 The result is that the impact of training on the probability to work as a professional in HT is substantial for both male and female immigrants. The probability to work in HT is doubled for those who participated in training during the last 2 months.

22 SARIT COHEN-GOLDNER probability to work as a professional in HT 0. 0.6 0. 0.4 0. 0.2 0. average male who participated in training (immediately after arrival) average female who participated in training (immediately after arrival) average male who did not participate in training (immediately after arrival) average female who did not participate in training (immediately after arrival) 0 0 2 4 6 8 0 2 4 6 8 20 Experience in Hi-Tech (in months) Fig.. Probability to Works as a Professional in the Hi-Tech Industry by Gender. 2 The results from the second specification suggest that only age at arrival and time in Israel significantly affect the probability to work as a professional in HT, such that this probability decreases with age at arrival, as in the first specification, and it increases with time spent in Israel. The probability to work in WC is significantly lower for females than for males. Knowledge of Hebrew and English, and time in Israel increase the probability to work in WC. Training also significantly increases the probability to work in WC jobs and its effect is independent of time since participation in the program. Similar to in the first specification, females face a higher probability to be unemployed, while Hebrew proficiency leads to a lower probability of unemployment. As expected, the probability to be unemployed, compared to work in BC jobs, significantly declines with time in Israel. Training also has a negative impact on the probability to be unemployed, though the effect is not significant..4. Transitions between Labor Market States In Table, we present the annual transitions between the three employment states (BC, WC and HT professionals), training and unemployment. Since the frequency of these transitions may change over time, we present the transitions that occurred during the first five years in Israel and during the

Immigrants in the Israeli Hi-Tech Industry Table. Number of Annual Transitions between Labor Market States a. TO HT Professional White Collar Blue Collar Training Unemployed 2 FROM (A) Males first five years HT professional 28 0 (%) (4) () () (0) (8) White collar 40 26 (%) () (8) (6) (2) () Blue collar 62 280 2 (%) (2) () (6) () (4) Training 6 4 0 2 (%) () () (42) (0) (2) Unemployed 44 48 (%) () (2) (48) () (4) (A) Males second five years HT professional 26 2 (%) () (20) (8) (2) () White collar 8 84 4 (%) (2) (8) (6) (2) () Blue collar 6 40 26 4 (%) () () (6) (2) () Training 0 (%) (2) (6) (4) (0) () Unemployed 8 6 8 (%) (4) (24) () () (4) (B) Females first five years HT professional 6 0 0 (%) () (4) () (0) (0) White collar 0 20 0 6 (%) (0) (8) () (0) (4) Blue collar 2 0 42 66 (%) (0) (20) () () (20) Training 6 2 0 4 (%) () (4) (40) (0) () Unemployed 80 6 2 4 (%) () (24) (0) () (2)

24 SARIT COHEN-GOLDNER Table. (Continued ) TO HT Professional White Collar Blue Collar Training Unemployed 2 (B) Females second five years HT professional 0 0 (%) (68) () (0) (0) () White collar 4 26 6 6 (%) () (8) (4) (2) (6) Blue collar 2 204 8 2 224 (%) (2) () (6) () () Training 0 4 (%) (8) () (46) (0) () Unemployed 0 24 (%) (2) (4) () (0) () Source: Engineers survey. a Actual number of individuals transitions between month t and month t+2 and row percentage. QA :2 subsequent five years ( 0), separately. The entries in the table are the number of individuals transitions between month t and month t+2. The table reveals substantial differences between the two sub-periods, mainly with respect to the transitions from unemployment. During the first five years in Israel, 48% (0%) of unemployed males (females) move to BC jobs, while 2% (24%) move to WC jobs and 4% (2%) are unemployed 2 months later. The transitions from unemployment to HT professionals are minor for both males and females. In the second sub-period (th to the 0th year), however, most of the unemployed immigrants remain unemployed. This finding suggests that during the first five years unemployment among immigrants is more transitory. That is, immigrants are unemployed between jobs. However, in the next five years, unemployment is more persistent, such that 4% (%) of the male (female) immigrants who were unemployed in a given month are unemployed a year later. During the two periods, we find a high persistence in employment in WC jobs and a lower persistence in BC and in HT jobs. Among the three employment states, professionals in HT is the less stable occupational category with persistence rate of 4% (%) among males (females) in the first period and, respectively, % (68%) in the second period. The transitions from HT

Immigrants in the Israeli Hi-Tech Industry 2 professionals to WC in the two periods are non-negligible and reflect the fluctuational nature of the HT industry during the 0s. The transitions from training to the three employment states are substantial, while the transitions to unemployment are low. During the first period % (4%) of the males (females) who attended training moved to WC jobs, while 42% (40%) moved to BC jobs and % (%) moved to HT professionals. Only 2% (%) of the males (females) moved from training to unemployment during the first period. In the second period, the transitions from training to WC declines to 6% (%) among males (females) and the transitions to BC increases to 4% (46%). About 2% (8%) of the males (females) moved from training to HT professionals during this period and % (%) moved to unemployment. Overall, it seems that the transitions to low-skill BC jobs occur mainly from unemployment, while the transitions to high-skill WC and HT jobs occur also through training and employment in BC jobs. 4. CONCLUSION This paper provides a descriptive analysis of the integration of CIS immigrants in the Israeli labor market in comparison to natives and with emphasis on their training attendance and employment in the HT industry. This description is informative for the researcher who is interested in evaluating the potential benefit for the growth of the Israeli HT industry from the imported and locally accumulated human capital of the recent immigrants. However, to further understand the link between the skills of the immigrants and the growth of the industry, one would need to formulate and estimate a model that controls for the dynamic selection of choices made by workers and firms. This challenging research is left for the future. NOTES. Cohen-Goldner and Eckstein (2002, 2004) found that upon arrival, unemployed immigrant with no work experience in Israel has a very low probability to receive a job offer in WC jobs. The estimated probability to receive such an offer was approximately 2% per quarter for the average male immigrant and 6% per quarter for the average female immigrant. Training increased these probabilities by 0 00%.

26 SARIT COHEN-GOLDNER 2 2. We define immigrant as a person who was born in the CIS and arrived in Israel after 8 at the age of 4 or above. Native is defined as a person who was born in Israel or immigrated to Israel before age 4 prior to 8.. See Appendix (A) for further description of the LFS. 4. The definitions of HT occupations in the HT industry are based on Feldman and Abouganem (2002).. The average years of schooling of the immigrants in Fig. is 4 for both males and females. 6. Immigrants in the HT industry are, on average, older than natives and have more years of schooling.. No data on training are available from the cross-sectional LFS. Therefore, the analysis in this section is based only on the engineers surveys. 8. The 200 2002 survey was conducted by the PORI survey company under the supervision of Sarit Cohen-Goldner and Zvi Eckstein.. The first survey does not provide wage data. Not all individuals reported their wages in the second survey. 0. The integration of immigrant engineers in the HT industry is similar to the integration of the general population of immigrants. However, within the HT industry, there is a substantial difference between the occupational distribution of engineers and the general population of immigrants. In particular, the vast majority of engineers work as professionals in the industry. For example, in 2000, more than 80% of immigrant engineers (both among males and female) in the HT industry worked as professionals. In contrast, among the general population of CIS immigrants 60% of the males and 40% of the females in the HT industry worked as professionals (LFS, 2000).. Logit estimates for participation in training confirm these effects, although only age at arrival has a significant negative effect on participation. 2. Unemployment in the LFS (Fig. ) includes also individuals who attend training programs.. Table 4 reports hazard ratios such that a coefficient greater than indicates that the variable shortens the duration to training and a coefficient lower than means that the variable leads to a longer duration to training. 4. Heckman and Smith () studied the role of these variables as predictors in the static decision to participate in training, while we study their effect on the decision when to participate.. Standard errors are clustered for each individual. 6. The characteristics of the average male/female immigrant are taken from Table. ACKNOWLEDGMENTS Support for the paper from the Science Technology and the Economy Program (STE), at the Samuel Neaman Institute for Advanced Studies in Science and Technology is gratefully acknowledged. I wish to thank Zvi Eckstein for valuable comments. I also thank two anonymous referees and

Immigrants in the Israeli Hi-Tech Industry the participants at the STE program at The Samuel Neaman Institute at the Technion. Marina Agranove, Yaniv Idid-Levi and Tali Larom provided excellent research assistance. 2 REFERENCES Cohen-Goldner, S., & Eckstein, Z. (2002). Labour mobility of immigrants: Training, experience, language and opportunities. CEPR Discussion paper series no. 42. Cohen-Goldner, S., & Eckstein, Z. (2004). Estimating the return to training and occupational experience: The case of females. IZA DP no. 2. Eckstein, Z., & Weiss, Y. (2002). The integration of immigrants from the former Soviet Union in the Israeli labor market. In: B.-B. Avi (Ed.), The Israeli economy, 8 8: From government intervention to market economics, essays in memory of Prof. Michael Bruno. Cambridge, MA: MIT Press. Eckstein, Z., & Weiss, Y. (2004). On the wage growth of immigrants: Israel 0 2000. Journal QA : of European Economic Association, 66 6. Feldman, M., & Abouganem, M. (2002). Development of the high-tech industry in Israel : Labour force and wages. Central Bureau of Statistics (Israel) Working Paper no., April. Heckman, J., & Smith, J. (). The pre-program earning dip and the determinants of participation in social program: Implications for simple program evaluation strategies. NBER Working Paper no. 68. Weiss, Y., Sauer, R. M., & Gotlibovski, M. (200). Immigration, search, and loss of skill. Journal of Labor Economics, 2(),. APPENDIX: DATA DEFINITIONS (A). The Labor Force Survey The LFS is an annual household survey, which is conducted by the Israeli Central Bureau of Statistics (CBS). The data is collected from roughly,000 households that are interviewed four times over a period of 8 months. Each household is interviewed for two consecutive quarters, followed by a break for two quarters, and is interviewed again for two consecutive quarters. The LFS provides information on labor market participation, occupation, education, country of origin, year of immigration and other demographic variables as well as details on workplace.