Patrick Arni, Rafael Lalive, and Jan C. van Ours, "How Effective are Unemployment Benefit Sanctions? Looking Beyond Unemployment Exit", Journal of Applied Econometrics, Vol. 28, No. 7, 2013, pp. 1153-1178. The file appendix-alo.pdf contains a technical appendix. The file alo-program.zip contains the Stata file simulation_exp_earn_dur.do, which is an ASCII file in DOS format. The file alo-dta.zip contains sim_values.dta, which is a Stata dataset that is described below. The data used in this article are confidential register data, owned by the Swiss government (SECO and ZAS). A slightly smaller data set is available to anyone wanting to work with the data. The public release data does not contain personal ID number, municipality IDs, or PES IDs. Contact Rafael Lalive in case you are interested in the public release file (see contact details below). IMPORTANT: The file is for your personal use ONLY and must NOT be shared with other researchers. This is to ensure that we can communicate who is working with the data to the owners of the administrative data. The data are merged from two sources: the unemployment insurance register databases (original names: AVAM and ASAL) and the social security administration database (original name: AHV). From the unemployment insurance register, we use, first, information about unemployment spell duration as well as data about sociodemographics, former occupation and regional location (inflow and outflow records of AVAM). Second, we extract information about benefit eligibility and payments (ASAL). Third, we use a specific sub-database of the unemployment insurance register that records the warnings, legal reasons and impositions of benefit sanctions. All the time-related information are in daily precision. From the social security administration database, we extract all the information on employment earnings; this allows as well the definition of the employment or non-employment status (employment earnings bigger than zero or not). These data are in monthly precision. For more details, please see the data sections in the paper and the online appendix. A complete list of the original variables is provided below. The unemployment insurance register data are administered by: Jonathan Gast Eidgenössisches Volkswirtschaftsdepartement EVD Staatssekretariat für Wirtschaft (SECO) Ressort Arbeitsmarktstatistik Effingerstr. 31-35 CH-3003 Bern The social security administration data are administered by: David Sanchez Centrale de Compensation CdC/ZAS Statistique et registres centraux Avenue Edmond-Vaucher 18 CP 3000 CH-1211 Genève 2 ***** List of the used variables (original names)" From AVAM and ASAL: PERSNR (not released) KA DTANM DTABM KANT (not released) ABGRU ARAMT (not released) SPREG PLZ (not released) ZIVIL GESLE STAAT AUSTA ALTER ALTR10 QUALIN MUSPR FUNKT STATU SITUA ANZAN ERVER DTEIN FREMD1 FREMD2 PRBES VERMS AUBN3 ANSPR HGREF CDIV VVERD UPERS ANZPE PROZENT Sanctions information from AVAM/ASAL: persnr (not released) datver datrecht beginn enzugab datguelt sart grund sankstat einsttag From AHV: persnr (not released) noaff ext debut fin revenu gcot bonass cosp year (monthly files - created) month (monthly files - created) ***** The file sim_values.dta contains, for each spell, information on actual warning, simulated sanctions, predicted index shaping the transition intensities and probabilities to be of a latent class type. Specifically, the variables are the following (with reference to the pages in the online appendix that describe their role in the simulation) adur duration from entry to warning (data) warn =1 if warned (data) yp24 Y>0 group: =1 if sum of earnings over 24mt > 0 (& duration not censored; data) y024 Y=0 group: =1 if sum of earnings over 24mt = 0 (& duration not censored;data) Ds2 =1 if enforcement observed (enf=1) or if simulation of enforcement duration yields an enforcement (if dur0 process estimation; this is the estimated index in theta_{y24>0}, see Appendix p.21 xbne prediction x'beta based on parameters from exit Y=0 process estimation; this is the estimated index in theta_{y24=0}, see Appendix p.21 xbe24 prediction x'beta based on parameters from earnings process estimation; this is the estimated index relevant for the earnings hazard, theta_{y24}; Appendix p.8 p_ty_a Prob(v_a|Ty = ty, T0 > ty): unobs.het.cond.on exit to Y>0 (see Appendix p.22),non-treated p_ty_b Prob(v_b|Ty = ty, T0 > ty): unobs.het.cond.on exit to Y>0 (seeAppendix p.22),non-treated pt_ty_a Prob(v_a|Ty = ty, T0 > ty): unobs.het.cond.on exit to Y>0 (see Appendix p.22),treated pt_ty_b Prob(v_b|Ty = ty, T0 > ty): unobs.het.cond.on exit to Y>0 (see Appendix p.22),treated ***** Please address the requests to work with the data and any other questions to: Rafael Lalive University of Lausanne Department of Economics Internef 553 CH-1015 Lausanne Rafael.Lalive |at| unil.ch