1. General This publication presents the results of the part of the 2010–2012 Innovation Survey which covered Research and Development in the business sector in 2012. The survey was conducted by the Central Bureau of Statistics, and was financed in part by the Office of the Chief Scientist at the Ministry of Economy. The survey data were classified by division according to the Standard Industrial Classification of All Economic Activities 2011,1 based on the principal activity of each company. Some of the data, relating to the years 2002–2009, were classified according to the previous Classification (1993). An examination of the transition to the new Classification revealed no substantial effect on the series of research and development data. Therefore the same series of data can be continued with no break. The survey covers the entire business sector (excluding agriculture and diamonds industries), which includes the Manufacturing, mining, and quarrying industries and the service industries. It was carried out on the base of a sample of companies in the survey population that were active in the business sector and had 10 or more jobs.2 Until 2008, the surveys of research and development covered the Manufacturing, mining and quarrying industries, the Computer programming, consultancy, and related activities division,3 and the Scientific research and development division, which included research institutes, R&D centres, technology incubators, Fabless companies,4 and start-up companies. The estimates of expenditure on R&D in the other service industries were calculated as part of the national expenditure on civilian research and development.5 In view of this, data from before 2008 for these industries are not presented in this publication. The aims of the R&D surveys in the business sector included: 1. Estimating expenditure, inputs, and outputs of businesses in the R&D field, in order to provide information on this activity, as well as to calculate the national expenditure on R&D. 1 CBS (2012). Standard Industrial Classification of All Economic Activities 2011. Technical Publication No. 80. Jerusalem: Author. 2 In the Computer programming, consultancy, and related activities division (62) and the Scientific research and development division (72), companies with 5 or more job positions were included. 3 In the 1993 Standard Industrial Classification, this division was called the Computer and related services division (72). 4 Fabless companies are companies that focus on planning and development, and which manufacture through sub-contractors in Israel or abroad. The companies in Israel are only responsible for R&D, marketing, and the management of the manufacturing process. These companies are included in the Scientific research and development division (72). 5 Data and information regarding these expenditures are provided in: CBS (2014). National Expenditure on Civilian Research and Development 1989-2012. Publication 1550. Jerusalem: Author. ( 17 ) 2. Producing a series of data on R&D that will allow continuous evaluation of the growth potential generated by technological innovations. 3. Providing the government, research institutions, and other planning bodies with a resource to help them guide Israeli research and development. In the past, R&D surveys were published in the Current Statistics series and special publications of the CBS. Updates 2010: In this survey, a new, updated questionnaire was used that included new measures. The process of constructing the questionnaire lasted a year and a half, and involved consulting with various parties at government offices, in academia, and in the business sector. The questionnaire underwent an initial examination at 11 companies from various industries in the business sector. The data for 2002–2009 were updated in this publication, for two major reasons: 1) The addition of companies that carry out a large amount of R&D, but which traditionally did not report R&D activity in previous R&D surveys. Based on clarification of this issue with company representatives (during the 2010 business R&D survey), the data were also updated for past years. 2) Updating of the industrial classifications of the companies as required for the entire period, as was done in the 2005 update. 2005: The classifications of the companies participating in the R&D survey were updated as a result of applying the recommendations of the Helfman Commission,6 which convened in 2005 to characterize and define the national expenditure on civilian R&D. The update was also part of an ongoing effort to adapt the data series to international definitions, in order to enable international comparisons. The classifications were corrected retroactively as of 2003, and as a result, there is a discrepancy between data series for the period up to 2002 and data series for the period from 2003 onwards. Data from the previous series that relate to 2003 and 2004 are presented in Appendix A. Part 2 of the Introduction presents data on business R&D by industries, and Part 3 presents R&D data by geographic regions. Part 4 focuses on R&D expenditures in R&D centres, and Parts 5 and 6 provide terms, definitions and explanations, and address the statistical methodology. 6 The recommendations of the Helfman Commission appear in the Report of the Committee for Characterization and Definition of National Expenditure on Civilian R&D, which was submitted to TELEM (Forum for National Infrastructures for Research & Development) in April 2005. ( 18 ) 2. Research and Development in the Business Sector by Industry 2.1 Total Expenditure on Business R&D Total expenditure on business R&D includes all expenditure on R&D in companies (internal – directly within companies; and external – by acquiring R&D services). The expenditure items included in this indicator are labour cost, purchase of raw materials, payments to external factors, other expenditures and overhead, and fixed capital formation for R&D activity. In 2012, companies in the business sector spent approximately NIS 39.9 billion on R&D activity, an increase of 14% compared to 2011 (NIS 34.9 billion). This increase occurred in every industry group, led by the knowledge-intensive services divisions (Computer programming, consultancy, and related activities division and the Scientific research and development division), which grew by 18.3%. In the Computer programming, consultancy, and related activities division, the total expenditure on R&D increased by 25.4% – from NIS 10.2 billion in 2011 to NIS 12.8 billion in 2012. Expenditure in the Scientific research and development division, which grew by 12.5%, still comprised the largest share of the expenditure in the knowledge-intensive services divisions (NIS 14.2 billion). In the Manufacturing, mining and quarrying industries, there was a 6.5% increase – from NIS 11.7 billion in 2011 to NIS 12.5 billion in 2012. Diagram 1 presents the total real expenditure on R&D by industries during 2003–2012. During that period, expenditure in the Scientific research and development division increased by 47% – a 4% average annual increase. Note that the Scientific research and development division includes start-up companies, which increased substantially in number. It also includes R&D centres of multi-national companies, whose activities increased appreciably during that period. This increase reflects the considerable change in the nature of R&D in general and Israeli R&D in particular, as well as the increase in the scope of R&D services. Expenditure in the Computer programming, consultancy, and related activities division increased by 65% during that period – an average annual increase of 5%; and expenditure in the Manufacturing, mining, and quarrying industries increased by 31% during that period – an average annual increase of 3%. ( 19 ) 1. TOTAL EXPENDITURE ON R&D, BY INDUSTRY 2003-2012 (2010 prices) 16 14 NIS Billion 12 10 8 6 4 2 0 2003 2004 2005 2006 2007 2008 2009 2010* 2011 2012 Computer programming, consultancy, and related activities division Scientific research and development division Manufacturing, mining, and quarrying industries *As of 2010 - according to the 2011 classification. In 2012, the expenditure on R&D by companies in the Scientific research and development division constituted 36% of the total R&D expenditure in the entire business sector; the expenditure on R&D by companies in the Computer programming, consultancy, and related services division constituted 32%; and R&D in the High-technology industries constituted 24% of the total R&D expenditure in the entire business sector (Diagram 2). 2. TOTAL EXPENDITURE ON R&D BY ECONOMIC Manufacturing medium-high technology* 4% 2012 Manufacturing - high technology* 24% Other service industries 1% Manufacturing medium-low technology* 2% Manufacturing - low technology* 1% Scientific research and development division 36% NIS 39.9 billion *Manufacturing, mining and quarrying divisions, by technological intensity. ( 20 ) Computer programming, consultancy, and related activities division 32% 2.2 Labour cost Human capital is the main production factor in research and development. Therefore, the main item in expenditure on internal R&D is labour cost. In 2012, labour cost in the business sector amounted to NIS 23.9 billion – a 12% increase compared with 2011. In 2012, the labour cost for a full-time equivalent (FTE – see the Terms, Definitions and Explanations chapter) job in R&D was NIS 331,000 (Table A) – a decrease of 2% compared to 2011 (NIS 339,000). Table A.- Labour Cost per FTE Job, by Industry (2011 Classification) NIS Thousands, 2010 prices 2009 2010 2011 2012 Total business sector 355 357 339 331 Manufacturing, mining and quarrying 297 302 290 295 Computer programming, consultancy, and related services 323 340 306 273 Scientific research and development 450 431 435 448 The highest labour cost per FTE job (NIS 448,000) was recorded in the Scientific research and development division. A substantial decrease occurred in the Computer programming, consultancy, and related services division (11%), following a decrease of 10% in 2011. It should be noted that labour cost included the labour cost for all jobs in R&D during 2012, whereas the number of FTE jobs was for the month of December 2012 only. 2.3 Jobs in R&D in the Business Sector Jobs in R&D are those of researchers, developers, technicians, programmers, and other workers who engage in R&D in companies or establishments, or who directly supply services or support to R&D activity. For example, the job of an accountant or bookkeeper who manages the accounts of an R&D unit is considered a job in R&D. However, the job of the director of the company or establishment, who does not deal directly with matters concerning the R&D unit, is not considered a job in R&D. Nonetheless, the labour cost of the director's job is included in the overhead item. In 2012, there were 68,700 jobs in R&D, an increase of 9.1% compared to 2011. The number of FTE jobs in R&D increased by 8.8%, from 59,500 in 2011 to 64,700 in 2012. In the Manufacturing, mining, and quarrying industries, there were 18,800 jobs in R&D. In the Computer programming, consultancy, and related activities division there were 26,000 jobs in R&D (an increase of 21% compared to 2011), and in the Scientific research and development division, there were 19,100 jobs in R&D (Diagram 3). ( 21 ) Thousands 80 70 60 50 40 30 20 10 0 2002 3. JOBS IN R&D, BY INDUSTRY 2002-2012 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Total Manufacturing, mining, and quarrying industries Computer programming, consultancy, and related activities division Scientific research and development division In 2012, 82% of the jobs in R&D were for academics holding any degree (including engineers), and 5% were for holders of third degrees or higher. In the Scientific research and development division, 86% of the jobs were for persons holding an academic degree (Table B). Practical engineers, technicians, and persons with other technological education held 10.2% of the jobs in R&D in the entire business sector. Workers with no relevant education (Others) held only 7.5% of the jobs in the entire business sector. Table B.- Jobs in the Entire Business Sector, by Industry 2102 Thousands Industry Total business sector Manufacturing, mining, and quarrying industries Computer programming, consultancy, and related activities division Scientific research and development division Total jobs in R&D Total academics Other workers 56.6 Practical engineers and technicians 7.0 68.7 20.0 15.3 2.9 1.8 27.5 23.0 2.4 2.1 20.3 17.6 1.6 1.2 5.1 Most of the jobs that were added in the entire business sector were for academics (including engineers). Part of the increase in labour cost per FTE job might be explained by this increase, because wages of academic workers are higher than those of non-academics. In 2012, FTE jobs in R&D constituted 94.2% of all R&D jobs. This is similar to the percentage published in 2011 (94.1%). ( 22 ) 2.4 Income from the sale of R&D The income from the sale of R&D in 2012 amounted to NIS 22.3 billion, an increase of 17.0% compared to 2011 (NIS 19.1 billion). In the Computer programming, consultancy, and related activities division income, income from the sale of R&D increased by 18.6% (NIS 1,608 million more than in 2011). In the Manufacturing, mining, and quarrying industries, income from the sale of R&D increased by 11% (NIS 301 million more than in 2011), and in the Scientific research and development division, this income increased by 13.6%. 4. INCOME FROM SALE OF R&D 2011-2012 24,000 Scientific research and development division 20,000 NIS Million 16,000 12,000 Computer programming, consultancy, and related activities division 8,000 4,000 0 R2011 2012 Income from sale of R&D R2011 2012 Income from sale of R&D to affiliated companies abroad* Manufacturing, mining, and quarrying industries * This figure is included in the total income from sale of R&D. 2.5 Income from the sale of new or improved products (goods or services) Income from the sale of new or improved products is an indicator that reflects the influence of R&D and innovation on the companies' output. Although innovation does not necessarily result from R&D, in companies engaging in R&D it forms a substantive part of the innovative process. Out of the total income of companies that engaged in R&D in 2012, NIS 22.6 billion resulted from the sale of new or improved products (11% of income), a decrease of 33.8% compared to 2011 (NIS 49.2 billion). In comparison, only 6% of the income of companies not engaged in R&D resulted from the sale of new or improved products – more than 3 times lower.7 7 See CBS. (2012). Innovation in the Business Sector 2006–2008. Publication 1463. Jerusalem: Author. ( 23 ) Table C.- Income from the Sale of New or Significantly Improved Products, by Industry 2012 Industry Total income in companies engaged in R&D Total business sector Manufacturing, mining, and quarrying industries Computer programming, consultancy, and related activities division Scientific R&D division Other service industries 2.6 NIS Million 299,597 Distribution of Income from the income in sale of new or companies improved products engaged in R&D Percentages 100 11 173,560 58 10 42,176 20,107 63,753 14 7 21 15 44 1 Concentration of R&D R&D activity in Israel is very concentrated (Table D). The 10 establishments (CR10) with the highest expenditure are responsible for 30%–50% of the total expenditure on R&D. While in the Manufacturing, mining, and quarrying industries this percentage is increased in 2012, in the Computer programming, consultancy, and related activities division and in the Scientific R&D division the percentage is decreased compared to 2011. These data are identical for the other indices (CR20 and CR50). Table D.- Concentration of Total Expenditure on R&D in the Business Sector Percentages Industry Year CR10 CR20 CR50 Manufacturing, mining, and quarrying industries 2011 2012 41 43 55 56 73 75 Computer programming, consultancy, and related activities division 2011 2012 34 32 42 40 57 54 Scientific research and development division 2011 2012 49 46 59 55 73 68 ( 24 ) 3. Business R&D by Geographic Location This section presents data on R&D activity according to a geographic division, and presents relative indicators for the purposes of comparison and analysis of gaps among the various districts in Israel. The Jerusalem district also included the R&D activity in the Judea and Samaria Area (Israeli localities). The data reveal great variations in R&D activity among the different districts, with most of the R&D activity concentrated in the centre of the country. Table E.– Selected R&D Indicators, by District 2012 Percentages, unless otherwise specified Indicator Total R&D expenditure (NIS million) Distribution of expenditure on internal R&D Total District Haifa Central Jerusalem Northern Tel Aviv Southern 23,887 1,981 3,047 6,720 16,794 10,179 1,166 100 5 7 18 40 27 2 13 24 7 12 14 24 4 17 10 20 7 23 12 39 8.2 2.9 4.4 11.7 12.4 14.2 1.6 5,042 1,511 2,325 7,208 8,791 7,765 1,027 369 347 269 418 390 360 259 Intensity of internal R&D (1) Ratio of external to internal R&D FTE jobs in R&D per 1,000 residents (whole numbers) R&D expenditure per 1,000 residents (NIS thousand) Labour cost per FTE job (NIS thousand) (1) R&D intensity is the ratio between expenditure on R&D and income from sales. ( 25 ) Percentages Diagram 5 presents the distribution of expenditure on internal R&D, financing from government sources and from international foundations, and the percentage of this financing out of the expenditure on internal R&D. The diagram indicates that the Northern and Southern Districts enjoyed a larger portion of the financing from government sources and from international foundations compared to their share in the expenditure on internal R&D. For example, companies in the Southern District were responsible for 2.5% of the expenditure on internal R&D in the business sector, and received 11.9% of the total financing from government sources and international foundations. 45 40 35 30 25 20 15 10 5 0 5. FINANCING FROM GOVERNMENT SOURCES AND INTERNATIONAL FOUNDATIONS, AND DISTRIBUTION OF EXPENDITURE ON INTERNAL R&D, BY DISTRICT 2012 Jerusalem Northern Haifa Central Tel Aviv Southern Percentage of financing from government sources and international foundations Distribution of expenditure on internal R&D Distribution of financing from government sources and international foundations ( 26 ) 4. R&D Centres 4.1 Foreign ownership and multi-national companies’ R&D centres An R&D centre is a company in which most of the R&D services of the R&D department are provided to a group of companies abroad with which it is affiliated. Israel's human resources draw many multi-national companies to the country. These companies establish development centres, which they use only as R&D units, in order to benefit from the innovative and creative thinking of Israeli researchers and developers. This mode of activity of the R&D centres forms a significant component of the amount of R&D activity in Israel, and the phenomenon has increased in recent years. The vast majority of the R&D centres are concentrated in the Computer programming, consultancy, and related activities division (62) and in the Scientific research and development division (72). The main activity of these centres is provision of R&D services to their parent companies or to other companies in their company group. Sometimes the activity includes the marketing and sale of the parent corporation's products as well. There are also R&D centres in the Manufacturing, mining, and quarrying industries, in which the R&D activity is carried out separately from the manufacturing activity, and the intellectual property is not transferred directly to the production lines but is instead transferred first to the parent company. Generally, this type of centre is the result of mergers and acquisitions of Israeli companies by foreign multi-national companies, and of the removal of the production line to outside of Israel. Occasionally these centres are established by parent companies without there having been any previous merger or acquisition. 4.2 Total expenditure on R&D in the R&D centres In 2012, the R&D centres spent a total of NIS 16.9 billion on R&D activity, an increase of 13.2% compared to 2011 (NIS 14.9 billion). In the R&D centres, total expenditure on R&D increased steadily during 2005–2012. During these years, total expenditure on R&D increased by 97%, from NIS 7.6 billion in 2005 to NIS 15.0 billion in 2012 (at 2005 prices). R&D centres constitute a major share of R&D activity in the business sector. During the period 2005–2012, the share of R&D centres out of the total business sector expenditure on R&D increased by 10 percentage points. In 2005, this percentage was 28% and by 2012 it increased to 38%. ( 27 ) 25 6. TOTAL EXPENDITURE ON R&D IN R&D CENTERES AND IN THE OTHER COMPANIES IN THE BUSINESS SECTOR 2005-2012 (at 2010 prices) NIS Billion 20 15 10 5 0 2005 2006 2007 2008 Other companies in the business sector 4.3 2009 2010 2011 2012 R&D centres Jobs in R&D in the R&D centres Jobs in R&D are those of researchers, developers, technicians, programmers, and other workers who engage in R&D in companies or establishments, or who directly supply services or support to R&D activity. For example, the job of an accountant or bookkeeper who manages the accounts of the R&D unit is considered a job in R&D. However, the job of the director of the company or establishment, who does not deal directly with matters concerning the R&D unit, is not considered a job in R&D. Nonetheless, the labour cost of the director's job is included in the overhead item. In 2012, there were 26,800 jobs in R&D in the R&D centres, an increase of 13% compared to 2011 (23,600 jobs). This increase occurred in all industries in which the R&D centres were classified. During the period 2005–2012, the R&D centres' percentage of employment out of R&D jobs in the business sector increased by 11 percentage points, from 22% in 2005 to 33% in 2012. ( 28 ) Table F.– Jobs in R&D, by Education and Full-Time Equivalent (FTE) Jobs, 2012 R&D centres Other companies in the entire business sector Thousands Jobs in R&D Jobs for holders of academic degrees Thereof: Third degree and above First and second degree (including engineers) Practical engineers and technicians Others FTE jobs Share of R&D centres' contribution to entire business sector Percentage 26.8 41.9 39 22.8 33.8 40 1.1 2.7 28 21.7 31.1 41 2.4 4.6 34 1.6 3.5 32 25.6 39.2 39 ( 29 ) 5. Terms, Definitions and Explanations 5.1 The survey population The survey was carried out on the basis of a sampling of companies active in the business sector with 10 or more jobs that were included in the survey population. In the Computer programming, consultancy, and related activities division and in the Scientific research and development division, companies with 5 or more jobs were sampled. The survey sample base was comprised of all industries in which the business sector operates (excluding agriculture and diamonds), among them the Manufacturing, mining, and quarrying industries and the service industries, which include the Computer programming, consultancy, and related activities division (62) and the Scientific research and development division (72) (which includes research institutes, R&D centres, technological incubators, Fabless companies, and start-up companies). Also included were Financial activities, Professional, scientific and technical activities, Trade, etc. 5.2 Economic classification The companies and establishments in the 2012 Survey of Research and Development were classified according to the Standard Industrial Classification of All Economic Activities 2011.8 5.3 The investigation unit The investigation unit was the establishment. An establishment is an economic unit that engages in one economic activity or one main activity and is situated in one location. It maintains separate accounts, and sells most of its services and/or goods to external factors. 5.4 The reporting period The establishments and companies were requested to report on their research and development activities during 2012 (January through December). Data on the number of jobs are for the month of December 2012. 5.5 Definitions Foreign-owned company: Any company for which at least 50% of the ownership is held by foreign residents. Research and development: Systematic and creative activity which is intended to increase the amount of knowledge of mankind, culture, and society. It also includes the use of knowledge in order to create new applications (the OECD "Frascati Manual", 2002). The definitions allocated to the sampling units can be found in the questionnaire, which appears as an appendix in this publication. 8 Central Bureau of Statistics (2012). Standard Industrial Classification of All Economic Activities 2011. Technical Publication 80. Jerusalem: Author. ( 30 ) Total expenditure on R&D: Includes labour cost, other expenditure including overhead, expenditure on raw and auxiliary materials for R&D, capital formation in construction and equipment for R&D, and payments to external factors for research and development activity. Current expenditure on R&D: Includes labour cost, other expenditures including overhead, expenditure on raw and auxiliary materials for R&D, and payments to external factors for research and development activity. This expenditure does not include investments in buildings and machinery for R&D. Expenditure on raw and auxiliary materials for R&D: Total expenditures on materials used for research and development. This item includes: raw materials, auxiliary materials, tools and instruments with a durability of less than one year, fuel, water, electricity, etc. Other expenditures, including overhead: The establishment’s expenditures related to R&D that are not included in the labour cost item, such as overhead, maintenance, travel abroad related to R&D, etc. Capital formation in construction and equipment for research and development: Capital formation in buildings, machinery, and equipment, whose durability exceeds one year, for purposes of research and development. Labour cost: In addition to wages, also includes expenses supplementary to wages and additional labour cost components, such as the employer’s share of National Insurance payments, study funds, provident funds, severance pay expenses borne by the employer, worker transportation, maintaining a cafeteria, worker training expenditures, etc. Payments to external factors for R&D work (excluding payments for patent acquisition): Research and development commissioned from external factors. Jobs: Employee jobs, owners, and unpaid family members; kibbutz members and workers hired through employment agencies. Jobs in R&D: Jobs in R&D are jobs directly dealing with R&D activity or that directly supply services or support to R&D activity, such as jobs of engineers, R&D managers, and administrative workers. Jobs that supply indirect services to the R&D department, such as information security, accountants, and directors are not included in the count of R&D jobs even when part of their wages is included in the R&D department's expenditure. Number of full-time jobs (full-time equivalent jobs): Includes a calculation of part-time jobs as full-time jobs (e.g., 6 half-time engineers = 3 full-time jobs). Jobs for academic professionals: Jobs for persons with an academic degree (first, second and third degrees, including engineers), or those with an equivalent level of education (i.e., jobs of persons who have not yet completed their academic studies, but are employed in jobs requiring an academic degree). Jobs for practical engineers and technicians: Jobs for persons with practical engineer/technician diplomas or for persons with equivalent qualification (e.g., those who have practical experience, laboratory technicians, and draftsmen). ( 31 ) R&D centres of multi-national companies: Companies whose R&D units provide R&D services mainly to a company group abroad with whom they are affiliated. Financing from government sources (government financing for R&D) and international foundations: Financing for conducting R&D activities that is provided by the Chief Scientist of the Ministry of Economy, government ministries, international foundations, and other governmental sources. Price indices: Until 2002, prices were amended based on the Consumer Price Index. Since 2003, prices have been deducted based on combined indices for income, capital formation, raw materials, and wages. Indices of concentration (CR10, CR20 and CR50): The concentration of R&D is measured by two indices, which represent the 10, 20 and 50 companies with the largest expenditure on R&D. ( 32 ) 6. Methodology 6.1 Sampling method In accordance with the definition of the survey population, the sampling units were VAT entities. If an entity had two (or more) National Insurance Institute files, the salary and employee job data of the multiple files were combined. The size variable used for planning the sample was the average number of employee jobs for the year 2012. A stratified simple random sample, without replacement, was taken, with strata defined by groupings of sample industries x size groups. (Sample industry refers to the industry to which the units belong.) The boundaries of the size groups were generally 10, 50, 100, and 250 employee jobs. All businesses with 250 or more employee jobs were sampled, in take-all strata. In a small number of industries, all businesses with 100 or more employee jobs were also sampled, in take-all strata. In addition, in most of the industries, all businesses with 50 or more employee jobs which also had recognition from the Office of the Chief Scientist were sampled in take-all strata (even if the recognition only involved a past request for financing from the Office of the Chief Scientist). Many of the questions in the R&D Survey were intended to determine what percentage of businesses had certain characteristics – for example, the percentage of businesses that engage in R&D. (These questions are in contrast to quantitative questions such as those concerning total expenditure). Therefore, the sampling was performed using (almost) equal probabilities in each stratum within each industry (with the exception of the take-all strata); the businesses in the take-some strata were sampled with probability in proportion to the number of business in the stratum (separately in each industry grouping). 6.2 Investigation method The data for the Research and Development Survey were collected by telephone and by mail. Telephone calls were used to locate reporting parties and to provide explanations regarding the survey, its requirements, and the manner in which the questionnaires should be answered. Questionnaires were sent and returned by mail. Data collection process The data-collection process included a number of stages, as described below: Stage 1: Location of a company and mailing of the questionnaire – At this stage the survey interviewers located the companies in the sample and the contact persons at the companies. In most cases, the contact persons were accountants, R&D directors, and human resources personnel. Also during this stage, questionnaires were mailed out to the contact persons. Stage 2: Clarifications and explanations – After the questionnaires were returned to the Central Bureau of Statistics, and after they were reviewed by the appropriate parties, there were some instances of problems and misunderstandings that needed to be clarified with the reporting parties at the companies. Reporting parties were contacted again in order to obtain corrections and supplementary information. ( 33 ) 6.3 Response In all, 2,515 companies were sampled in the 2010–2012 Innovation survey, and 99% of the companies responded to the survey in full or in part. Approximately 789 companies responded that they engaged in research and development activity in 2012, and data for 36 companies were imputed based on administrative data. 6.4 Data handling After receiving the reports from the companies, the data underwent a quality control procedure that included logic checks, comparison to administrative data from different sources (such as the Tax Authority, the Chief Scientist of the Ministry of Economy, and various multi-national research funds), as well as comparison to the companies’ data from previous years (where relevant). During this stage, 20% of the companies in the sample received a second request, for clarifications and supplemental data. In certain cases, it was impossible to receive full data from the companies (for instance, if a company closed), or some of the data (if the company’s accounting system was unable to provide a specific figure, or data according to a specific breakdown). Missing data were handled by means of: 1. Imputation of missing data: In cases in which the company provided data but some of the data were missing, the data were completed based on imputation of averages in the company’s sampling stratum (according to industry, company size, ownership, etc.). The missing datum was calculated using one of two weights – workers and/or revenue. 2. Imputation of full data, in cases of non-response: For companies whose response to the survey could not be obtained, for various reasons, the “nearest neighbour” model was used, in which the non-responding company receives the data of a company similar to it on R&D-related parameters. Note that data were imputed only for companies for which there was a clear indication of R&D activity in the relevant year (newspaper articles, data from the Chief Scientist at the Ministry of Economy, reports from previous years, etc.). In order to determine the “close neighbor” for each company, logistic regression was used to calculate the probability of R&D activity for all companies in the sample. The regression, which was used to examine a binary parameter for R&D activity, was based on administrative data (such as number of workers, industry, rate of export, and sector). The company closest in probability for R&D activity to the non-responding company was chosen as its “nearest neighbour”. ( 34 ) 6.5 Reliability and limitations of the data The results of this survey, like those of any other statistical survey, were subject to a number of limitations. The limitations in the present survey were: (a) Lack of accurate recording by the establishments of expenditures related to R&D (especially labour cost, when a considerable portion of jobs in R&D also included other activities, for which there was no distinction between the labour inputs in the various activities); (b) The definition of “Research and Development” was not fully understood. As a result, some establishments failed to include R&D activities, whereas activities not necessarily related to R&D were included. (c) Lack of complete, up-to-date information on changes in the nature of the activities that the companies engaged in (particularly small businesses in the Manufacturing, mining, and quarrying industries), which occasionally led to a less than optimal classification, because the classification was based mainly on the previous survey. (d) For a variety of reasons, some companies did not respond to one or more sections of the questionnaire, which necessitated imputation of the missing data (based on information from previous years, as well as on other information). (e) It should be noted that the smaller the size of the industry (in terms of the number of companies engaged in R&D), the greater the influence of the problems mentioned in the two previous paragraphs. Ultimately, this leads to greater variance when the distribution among industries is broader. (f) The fact that a sample was used rather than a full enumeration might lead to less accurate expenditures on R&D. It should be noted that most of the establishments and companies that were approached responded affirmatively, and filled out the questionnaires. ( 35 )