OECD WORKSHOP ON THE MEASUREMENT OF THE HIGH-GROWTH ENTERPRISES THE IDENTIFICATION OF HG ENTERPRISES The Italian case Paris, November 19, 2007 Identification of HG Let Nxx be the active population in year xx, reference year (xx=2005) Step1: a merge by identification code of the population Nxx and Nxx-1 to xx-3 → AA=N2002 ∩ N2003 ∩ N2004 ∩ N2005 Step2: we exclude the real births in xx-3 i.e. Rb(2002) from AA, then Poten_population=AA-Rb(2002) Comparing distributions Potential population AA versus Active population N By size class N2005 0 1 N2002 2-4 5-9 10-14 AA2005 15-19 20+ AA2002 0% 20% 40% 60% Dissimilarity index : Nxx versus AA z NAA 1 K | f Ni f AAi | 0.05 2 i 1 80% 100% Calculated by nace&size Potential population AA Empl2005 empl2002 0 1 2-4 5-9 10-14 15-19 20+ Total 0 1672471 111596 43514 6503 1479 526 851 1836940 1 70794 149104 68222 5597 789 204 224 294934 2-4 25676 44242 201802 43417 3432 689 633 319891 5-9 6621 3551 25942 88183 18540 2557 1448 146842 10-14 2311 716 2230 11952 28572 8252 2959 56992 15-19 986 239 552 1517 6002 9991 5903 25190 1997 370 653 1076 1825 3896 48402 58219 1780856 309818 342915 158245 60639 26115 60420 2739008 20+ Total Size threshold : 10 or more employees in the beginning Step3: If employees(2002) ≥ 10 → reference for HG =140,401 for year 2005 Growth threshold : 20% per annum Step4: HG are obtained applying to the population of reference (step 3) the following rules to employees and to turnover: HG_empl - yes IF [empl(2005)/empl(2002)] ≥ 1.728 HG_tur - yes IF [turnover(2005)/turnover(2002)] ≥ 1.728 Growth measured by employees or turnover hg_turnovr hg_empl no No yes total 0 8262 (59%) 8262 (59%) yes 2749 (20%) 3022 (22%) 5771 (41%) Total 2749 (20%) 11284 (80%) 14033 Ratio= HG/pop_ref_hg= 14033/140401 *100= 10.0% HG by type of growth (employees or/and turnover) by sector of activity (%) Tot K74 K72 K70.71.73 J hg_tur I hg_emp hg_both H G F D C+E 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Growth by events Step5: HG can grow because events of takeover. Takeover from units that cease Takeover from units that transfer activity Looking for such events in the period 2002-2005, we have identified 979 (7%) of HG with events. By excluding such events we obtain the “pure” HG, Nace empl_02 empl_03 empl_04 empl_05 date of event HG_dip 4512 335 376 480 541 200407 . 3430 26 31 36 189 200501 Y 7230 1556 1533 1436 2873 200501 Y HG_tur Tur_02 Tur_05 Y 45766 113639 . 37631 31173 Y 148118 312504 HG*= 13054 Ratio*= HG*/pop_ref_hg= 13054/140401 *100= 9.3% Empl_02_ Empl_04_i id1 d1 411 243 206 172 1767 1566 HG: with or without events?(1) HG with events and “pure HG”, by sector of activity (%) 40.0 35.0 30.0 HG HG* pure 25.0 20.0 15.0 10.0 5.0 74 K 72 K J 70 .7 1. 73 K I H G F D C +E 0.0 HG: with or without events?(2) hgeve hg_empl . . . Y . Y Y . Y Y Y Y Total sub_HG_pure hg_tur Y . Y Y . Y _FREQ_ 7961 2485 2608 301 264 414 14033 13054 empl(2002) empl(2005) average 48.2 55.3 25.9 57.8 38.3 98.8 253.8 305.9 84.3 198.1 111.0 454.4 HG : Turnover and employment Growth rates by size and sectors Sector size_class Industry 10-14 15-19 20+ 10-14 15-19 20+ 10-14 15-19 20+ 10-14 15-19 20+ Construction Trade Services Total Enterprises 2034 958 2195 1419 544 814 837 353 766 1387 672 2054 14033 empl_02 23,802 15,939 160,922 16,367 9,002 41,804 9,783 5,866 69,562 16,219 11,179 311,678 692,124 empl_05 40,928 25,369 238,819 25,345 14,129 61,654 19,017 11,554 129,485 31,061 23,060 554,004 1,174,425 Turnover_02 Turnover_05 4698193 3207346 44395225 2074931 1355835 8611734 5398630 3002411 33647696 2400627 4205122 28127045 141124795 10360214 7339735 103518126 4687596 2854768 18058326 10565335 5874710 73235754 7667780 7473291 192895109 444530744 GROWTH empl tur 72.0 59.2 48.4 54.9 57.0 47.5 94.4 97.0 86.1 91.5 106.3 77.7 69.7 120.5 128.8 133.2 125.9 110.6 109.7 95.7 95.7 117.7 219.4 77.7 585.8 215.0 Gazelle It is that subset of HG that are real births in xx-5 and xx-4. For xx=2005 gazelles are obtained by merging the real births 2000 and the real births 2001 with HG by identification code Step1: potential gazelle are identified as the potential population of HG born in 2000 and 2001 (the subset of real births in 2000 and 2001 surviving in 2005, with employees(2002) ≥10) Gazelle NACE C+E D F G H I J K70.71.73 K72 K74 Total poten_gazelle % 54 2462 1390 888 485 694 64 70 258 977 7342 0.74 33.53 18.93 12.09 6.62 9.45 0.87 0.95 3.51 13.31 100.00 gazelle 17 483 348 173 70 214 19 18 53 283 1678 % 1.01 28.78 20.74 10.31 4.17 12.75 1.13 1.07 3.16 16.87 100.00 Gazelle % Gazelle/ Without poten_gaz events 17 1.05 31.5 464 28.75 19.6 343 21.25 25.0 163 10.1 19.5 68 4.21 14.4 207 12.83 30.8 18 1.12 29.7 17 1.05 25.7 47 2.91 20.5 270 16.73 29.0 1614 100 22.9 Working in progress Changing Growth threshold 45.00 40.00 35.00 30.00 15% 25.00 20% 25% 20.00 30% 15.00 40% 10.00 5.00 74 K 72 K J 70 .7 1. 73 K I H G F D C +E 0.00 Working in progress Alternative bases for measurements Varying threshold of growth BIRCH index m=(xt1-xt0) xt1/xt0 where xt1=employees(2005) xt0=employees(2002) HG by Birch are: the AA having m>=m[90%] OR m>=m[95%] of the distribution in the whole economy Working in progress ENTERPRISES 1 2-4 5-9 10-14 15-19 20+ Total GROWTH empl. GROWTH turn. 5% 10% 5% 10% 5% 10% 2962 3292 1647 1003 660 5162 14726 6814 7299 3652 1987 1378 8342 29472 1104.6 650.8 396.3 260.1 204.0 47.4 60.0 726.4 417.4 256.6 178.1 137.1 43.4 57.5 204.4 180.4 144.8 130.2 54.1 39.6 45.1 160.2 142.1 115.8 103.2 53.9 37.3 43.0 Problem: which base to be used as reference population? AA size [1+] ? AA=902068 Working in progress Comparison between HG and Birch top 10% 14% 18% HG & Birch HG only Birch only empl<10 20% 48% Birch only empl>=10