PISA (Programme for International Student Assessment) is an international survey organised by the OECD. Its aim is to create indicators to measure the skills and abilities of 15-year-old pupils in reading, mathematics and science, as well as learning strategies, motivation and computer skills. The results are based on national representative samples. The indicators are calculated according to the characteristics of the system, schools and pupils. In order to be able to make time comparisons, these competence measurements are repeated every three years, with a new focus every three years (2000 reading, 2003 mathematics with the addition problem solving, 2006 natural sciences). Around 60 countries participated in the 2006 survey. The test takes about 120 minutes. It is followed by a 45-minute questionnaire. This questionnaire contains mandatory international questions that shed light on the socio-economic environment of pupils, as well as optional international and national questions. Switzerland chose two international options for the student questionnaire: the module on interdisciplinary competences and the ICT module on familiarity with the new technologies. In addition, national modules were added. Another questionnaire for schools aims to collect information about the characteristics of the school and its teaching. Around 24,700 pupils from 510 schools throughout Switzerland took part in the 2006 survey. The national project management 2000-2003-2006 was carried out by the Federal Statistical Office in Neuchâtel which was responsible for the project coordination in close cooperation with the international PISA consortium and 4 regional centres in Switzerland.
PISA (Programme for International Student Assessment) is an international survey organised by the OECD. Its aim is to create indicators to measure the skills and abilities of 15-year-old pupils in reading, mathematics and science, as well as learning strategies, motivation and computer skills. The results are based on national representative samples. The indicators are calculated according to the characteristics of the system, schools and pupils. In order to be able to make time comparisons, these competence measurements are repeated every three years, with a new focus every three years (2000 reading, 2003 mathematics with the addition problem solving, 2006 natural sciences). Around 60 countries participated in the 2006 survey. The test takes about 120 minutes. It is followed by a 45-minute questionnaire. This questionnaire contains mandatory international questions that shed light on the socio-economic environment of pupils, as well as optional international and national questions. Switzerland chose two international options for the student questionnaire: the module on interdisciplinary competences and the ICT module on familiarity with the new technologies. In addition, national modules were added. Another questionnaire for schools aims to collect information about the characteristics of the school and its teaching. Around 24,700 pupils from 510 schools throughout Switzerland took part in the 2006 survey. The national project management 2000-2003-2006 was carried out by the Federal Statistical Office in Neuchâtel which was responsible for the project coordination in close cooperation with the international PISA consortium and 4 regional centres in Switzerland.
PISA (Programme for International Student Assessment) is an international survey organised by the OECD. Its aim is to create indicators to measure the skills and abilities of 15-year-old pupils in reading, mathematics and science, as well as learning strategies, motivation and computer skills. The results are based on national representative samples. The indicators are analysed according to the characteristics of the school system, schools and pupils. In order to be able to make chronological comparisons, these competence measurements are repeated every three years, with a new focus every three years (2000 reading, 2003 mathematics with the addition problem solving, 2006 natural sciences).
Around 40 countries participated in the 2003 survey. The test takes about 120 minutes. It is followed by a 45-minute questionnaire. This questionnaire contains mandatory international questions that shed light on the socio-economic environment of pupils, as well as optional international and national questions. Switzerland chose two international options for the student questionnaire: the module on interdisciplinary competences and the ICT module on familiarity with the new technologies. In addition, national modules were added. Another questionnaire for schools aims to collect information about the characteristics of the school and and its didactics.
Around 24,700 pupils from 451 schools took part in the 2003 survey. The national project management 2000-2003-2006 was carried out by the Federal Statistical Office in Neuchâtel which was responsible for coordination in close cooperation with the international PISA consortium and 4 regional centres in Switzerland.
PISA (Programme for International Student Assessment) is an international survey organised by the OECD. Its aim is to create indicators to measure the skills and abilities of 15-year-old pupils in reading, mathematics and science, as well as learning strategies, motivation and computer skills. The results are based on national representative samples. The indicators are analysed according to the characteristics of the school system, schools and pupils. In order to be able to make time comparisons, these competence measurements are repeated every three years, with a new focus every three years (2000 reading, 2003 mathematics with a supplement on problem solving, 2006 natural sciences).
Around 40 countries participated in the 2000 survey. The test takes about 120 minutes. This is followed by a 45-minute questionnaire. This questionnaire contains mandatory international questions that shed light on the socio-economic environment of pupils, as well as optional international and national questions. Switzerland chose two international options for the student questionnaire: the module on interdisciplinary competences and the ICT module on familiarity with the new technologies. In addition, national modules were added. Another questionnaire for schools aims to collect information about the characteristics of the school and the lessons.
Around 13,500 students from 360 schools throughout Switzerland took part in the first survey in 2000.
The national project management 2000-2003-2006 was carried out by the Federal Statistical Office in Neuchâtel and was responsible for coordination in close cooperation with the international PISA consortium and 4 regional centres in Switzerland.
Das "Occupational Panel on Tasks and Education (OPTE)" beschreibt für die Jahre von 1973 bis 2011 Tätigkeitsprofile, Bildungsinvestitionszeiten und das Ausbildungsverhalten differenziert nach 179 harmonisierten Berufsgruppen. Es wurde für das Dissertationsprojekt "Die Anwendbarkeit des Erlernten in den wandelnden Bildungs- und Arbeitslandschaften der 1970er bis 2000er Jahre" erstellt. Die Dissertationsschrift ist unter https://kops.uni-konstanz.de/handle/123456789/49897 frei zugänglich und beschreibt (im Anhang) ausführlich die Erstellung der mit diesem Panel veröffentlichten Variablen.
Die Datenbasis für das Panel auf Berufsebene bilden die Scientific Use Files (SUF) des deutschen Mikrozensus. Diese erfassen den ausgeübten Beruf bis zum Jahr 1993 nach der Klassifikation der Berufe des Jahres 1975 (KldB75). In den nachfolgenden Erhebungsjahren erfolgt die Erfassung nach der Klassifikation der Berufe des Jahres 1992 (KldB92). Beide Berufsklassifikationen wurden nach dem Prinzip des kleinsten gemeinsamen Nenners so aggregiert, dass über den gesamten Zeitraum von 1973 bis 2011 eine homogene Messung von Berufsordnungen erfolgt. Zudem wurden die Daten auch auf die Klassifikation der Berufe des Jahres 1988 (KldB88) umgeschlüsselt, um ein Zuspielen der Berufsinformationen zu anderen Datensätzen zu ermöglichen, welche den Beruf nach der KldB75, KldB88 oder KldB92 erfassen. Das Excel-Dokument "Transition_key_occupational_groups_KldB75_88_92.xlsx" gibt die Zuordnung der Berufscodes der KldB75, KldB88 oder KldB92 zur harmonisierten KldB88h wieder.
Auf der Ebene der 179 harmonisierten Berufsordnungen werden Veränderungen im Tätigkeitsprofil, in den Bildungsinvestitionen und im Ausbildungsverhalten über die Zeit beschrieben. Diese werden aus folgenden Informationen der Mikrozensus-SUF's erhalten:
Tätigkeitsprofile: In den Jahren 1973, 1976, 1978, 1980, 1982, 1985, 1987, 1989, 1991, 1993, 1995, 1996, 2000, 2004, 2007, 2011 wird jeweils die Frage nach der "überwiegend ausgeübten Tätigkeit" in der Haupterwerbstätigkeit gestellt. Die möglichen Antwortvorgaben unterscheiden sich in den einzelnen Erhebungsjahren. Grob gesagt kann zwischen drei Perioden (1973 bis 1980, 1982 bis 1995 und 1996 bis 2011) der Tätigkeitsmessung unterschieden werden. Die erfassten Haupttätigkeitsschwerpunkte können jedoch harmonisiert werden, so dass für jede harmonisierte Berufsgruppe über die Zeit nachvollziehbar ist, wie hoch der Anteil an Personen in einem Beruf ist, die in einem Jahr folgende elf Haupttätigkeitsschwerpunkte ausgeübt haben:
Das Vorgehen zur Harmonisierung wird in der Dissertation ab Seite 299 (Anhang A) und in doi.org/10.1007/s11135-021-01158-y beschrieben. Die Tätigkeitsprofile in den "Zwischenjahren", in welchen keine SUFs des Mikrozensus zur Verfügung stehen, wurden interpoliert. Anschließend wurden die Tätigkeitsanteile mit einem Moving-Average (t-3, t, t+3) geglättet. Die mit diesem Panel veröffentlichten Tätigkeitsanteile unterscheiden sich von der in der Dissertation verwendeten Tätigkeitsanteilen, indem auch Nichtdeutsche und Erwerbstätige mit weniger als zehn Wochenstunden Arbeit berücksichtigt werden. Zudem werden die Tätigkeitsanteile nach den Arbeitsstunden der Erwerbstätigen gewichtet und anonymisiert.
Anonymisierung der Tätigkeitsprofile: Die Fallzahl "N" gibt die hochgerechneten, interpolierten und mit Moving-Average (t-3, t, t+3) geglättete Anzahl an Erwerbstätigen in der Berufsordnung wieder. Wird eine Aggregation der Berufsordnungen angestrebt, kann "N" genutzt werden, um z.B. gewichtete Durchschnitte zu berechnen. Multipliziert man "N" mit den jeweiligen Tätigkeitsanteilen "taskshare_..." erhält man eine "fiktive" Zahl an Erwerbstätigen, die diese Haupttätigkeit im Beruf ausüben. Die Zahl ist fiktiv, weil es sich aufgrund der Harmonisierung um geschätzte Tätigkeitsanteile handelt, die zudem mit der jeweiligen Stundenanzahl der Erwerbstätigen gewichtet sind. Einzelfälle können deshalb sowieso nicht zweifelsfrei identifiziert werden. Um eine mögliche Deanonymisierung faktisch weiter zu erschweren, wurden des Weiteren sichergestellt, dass hinter jeder genannten Tätigkeit mindestens drei "fiktive Personen" stehen. Haupttätigkeiten in einem Beruf wurden deshalb mit ein oder zwei weiteren Haupttätigkeiten zusammengefasst, bis in Summe über drei "fiktive Personen" diese Haupttätigkeiten ausübten. Die ursprüngliche "fiktive Personenanzahl" in diesen Haupttätigkeiten wurden anschließend mit der durchschnittlichen Anzahl der "fiktiven Personen" aus diesen Haupttätigkeiten ersetzt. War eine Zusammenfassung im Querschnitt nicht sinnvoll, weil sich der nächstgrößte Tätigkeitsanteil stärker vom kleinsten Tätigkeitsanteil unterschied (weil dieser mehr als 10 "fiktive Personen" enthielt) wurde eine Aggregation über die Erhebungsjahre gewählt. In diesem Fall wurden die Erhebungsjahre solange zusammengefasst, bis in jeder Tätigkeit des Berufs mindestens drei "fiktive Personen" enthalten waren. Die Tätigkeitsanteile des Berufs wurden anschließend mit den durchschnittlichen Tätigkeitsanteilen der zusammengefassten Erhebungsjahre ersetzt. Zuletzt wurden alle Tätigkeitsanteile gerundet. Aufgrund dieser Rundung ergibt die Summe aller Tätigkeitsanteilen einer Berufsgruppe nicht immer den Wert 1. Ist dies für die weiteren Analysen notwendig, sollten die Tätigkeitsanteile so skaliert werden, dass sie in Summe 1 ergeben.
Die Variable "N_soc" gibt die Anzahl der hochgerechneten, interpolierten und mit einem Moving-Average (t-3, t, t+3) geglätteten abhängig Beschäftigten "Angestellte, Arbeiter, Heimarbeiter" (ohne Auszubildende) aus dem Mikrozensus wieder. Die Variable "taskshare_socsec_..." gibt die dazugehörigen Tätigkeitsanteile der abhängig Beschäftigten wieder. Die Anonymisierung erfolgte in derselben Weise wie bei den Tätigkeitsanteilen "taskshare_..." mit allen Erwerbstätigen. Um Einzelfallidentifikationen durch die Subtraktion von "N_socsec" von "N" zu vermeiden, wurden die Tätigkeitsanteile "taskshare_socsec_..." mit den Tätigkeitsanteilen "taskshare_..." aller Erwerbstätigen ersetzt, sofern N-N_socsecEnglish version ================================================================================
The "Occupational Panel on Tasks and Education (OPTE)" describes task profiles, education investment periods and training behavior differentiated by 179 harmonized occupational groups for the years from 1973 to 2011. It was prepared for the dissertation project "The Applicability of the Learned in the Changing Educational and Labor Landscapes of the 1970s to 2000s." The dissertation paper (in German) is freely available at https://kops.uni-konstanz.de/handle/123456789/49897 and describes in detail (in the appendix) the creation of the variables published with this panel. The creation of the task variables is also decribed in English in doi.org/10.1007/s11135-021-01158-y
The data basis for the occupation-level panel are the Scientific Use Files (SUF) of the German Microcensus. These record the occupation up to 1993 according to the 1975 classification of occupations (KldB75). In subsequent survey years, the occupation is recorded according to the 1992 classification of occupations (KldB92). Both occupational classifications were aggregated according to the principle of the lowest common denominator in such a way that there is a homogeneous measurement of occupational classifications over the entire period from 1973 to 2011. In addition, the data were also recoded to the 1988 Classification of Occupations (KldB88) to allow matching of occupational information to other datasets that record the occupation according to KldB75, KldB88, or KldB92. The Excel document "Transition_key_occupational_groups_KldB75_88_92.xlsx" shows the mapping of the occupation codes of KldB75, KldB88 or KldB92 to the harmonized KldB88h.
At the level of the 179 harmonized occupational codes, changes in task profile, educational investments and educational behavior over time are described. These are obtained from the following information from the Microcensus SUF's:
Task profiles: In each of the years 1973, 1976, 1978, 1980, 1982, 1985, 1987, 1989, 1991, 1993, 1995, 1996, 2000, 2004, 2007, 2011, the question about the "predominantly performed activity" in the main job is asked. The possible answer specifications differ in the individual survey years. Roughly speaking, a distinction can be made between three periods (1973 to 1980, 1982 to 1995, and 1996 to 2011) of task measurement. However, the main task recorded can be harmonized so that for each harmonized occupational group it is possible to track over time the proportion of people in an occupation who performed the following eleven main activity foci in a given year:
• taskshare 11: "nursing/treating medically or cosmetically."
The procedure for harmonization is described in doi.org/10.1007/s11135-021-01158-y .
Anonymization of task profiles: The case number "N" reflects the extrapolated, interpolated and moving-average (t-3, t, t+3) smoothed number of employed persons in the occupational group. If aggregation of occupational groups is desired, "N" can be used to calculate weighted averages, for example. Multiplying "N" by the respective activity shares "taskshare_..." yields a "fictitious" number of employed persons performing this main activity in the occupation. The number is fictitious because, due to harmonization, it is an estimated activity share, which is also weighted with the respective number of hours of the employed persons. Individual cases can therefore not be identified beyond doubt anyway. Furthermore, in order to make deanonymization even more difficult, it was ensured that at least three "fictitious" persons are behind each activity mentioned. Main activities in an occupation were therefore combined with one or two other main activities until a total of more than three "fictitious persons" performed these main activities. The original "notional number of persons" in these main activities were then replaced with the average number of "notional persons" from these main activities. If a cross-sectional aggregation did not make sense because the next largest activity share was more different from the smallest activity share (because the latter contained more than 10 "fictitious persons"), an aggregation over the survey years was chosen. In this case, survey years were aggregated until each activity in the occupation contained at least three "notional persons". The occupation's activity shares were then replaced with the average activity shares of the aggregated survey years. Finally, all task shares were rounded. Due to this rounding, the sum of all task shares of an occupational group does not always add up to 1. If this is necessary for further analyses, the activity shares should be scaled so that they add up to 1.
The variable "N_soc" reflects the number of extrapolated, interpolated and moving-average (t-3, t, t+3) smoothed dependent employees "white-collar workers, blue-collar workers, homeworkers" (without apprentices) from the microcensus. The variable "taskshare_socsec_..." reflects the corresponding activity shares of the dependent employees. Anonymization was carried out in the same way as for the activity shares "taskshare_..." with all employed persons. To avoid individual case identifications by subtracting "N_socsec" from "N", the activity shares "taskshare_socsec_..." were replaced with the activity shares "taskshare_..." of all employed persons, if N-N_socsec<5000. The corresponding cases are labeled with the variable "anonymous_socsec".
Educational investment: For the variable "educ_invest", the education time in months formally required to obtain the general education and last/highest vocational qualification was calculated from the Microcensus SUF's of 1973, 1976, 1978, 1980, 1982, 1985, 1987, 1989, 1991, 1993 and 1995 to 2011 for all employed persons. For example, a secondary general school certificate was measured as 108 months (9 years) and a "completion of apprenticeship training or equivalent vocational school qualification" as 36 months (3 years). A detailed list and justification of the education periods assigned to each by degree can be found in the dissertation beginning on page 308 (Appendix B). The "average formal education time" of an occupation was calculated using the average education time of all employed persons in the harmonized occupational group. The "intermediate years" in which no SUF was available were interpolated. Subsequently, the values were smoothed with a moving average (t-3, t, t+3).
Training behavior (supply-demand relation): The Federal Institute for Vocational Education and Training (BIBB) converted the major field of the highest vocational qualification in combination with the training institution into a learned occupation according to KldB92. The heuristic procedure for this is described in Maier and Helmrich (2012). To calculate the supply-demand relation ("sdr"), the microcensuses (on-site) from 2005 to 2012 are pooled and a relative distribution of vocational degrees according to the harmonized occupational classification KldB88h is calculated for all degree years from 1973 to 2012. This distribution is contrasted with the relative distribution of employment shares according to KldB88h for the respective years. The procedure is described in the dissertation on page 86 and 145-147 and plausibilized starting on page 328 (Appendix D). The variable "ln_sdr" corresponds to ln(sdr).
The responsiveness of democratic institutions is a topic of fundamental importance to researchers, citizens, and decision-makers. The PolicyVotes project aimed to assemble a dataset that facilitates investigation of the responsiveness of political parties and governments to public preferences. The data collection efforts were motivated by the interest to allow researchers to examine, among many others, the following questions: Are governments responsive to citizen demands? Do we see policy changing in response to changing public preferences over time? Is a government's responsiveness to public demands more pronounced in some policy areas than in others and at some points in time than others? What is the mediating role of political institutions such as electoral systems, government types (coalition versus single-party) and executive-legislative structures? How does the degree of responsiveness of national governments compare to responsiveness of European institutions? What are the interdependencies of legislative decision-making between the national and the European level? Do national policies influence the development of European level public policies and vice versa?
The data collection we have assembled facilitates addressing these questions and others. It allows researchers to use large-N statistical methodologies to empirically test theoretical models of dynamic representation in a multilevel system of governance. It allows longitudinal comparative empirical analysis of the triangular relationship between preferences of the electorate, policy positions of parties and governments, and legislative outputs of national governments and the EU. With this data collection we are introducing efficiencies that enable researchers to examine how and under what circumstances responsiveness can be achieved in different institutional settings.
For individual-level data, we have harmonized Eurobarometers from 1970 to the 2011, the ISSP Role of Government surveys, and the EES voter Study. For measurements of party positions, we have harmonized and cross-linked the Chapel Hill Expert Survey, the Party Policy in Modern Democracies Dataset, the Comparative Manifesto Project data, and the EES Euromanifesto Study. For the measurements of policy output we have collected and cross-linked data for legislative output and budget outlays of 15 EU governments and the European Union.
Please refer to the How-to-Guide and the user guides in the individual trendfile folders (see Downloads/Datasets) for detailed information and citation instructions. Following trendfiles and user guides are available:
- Arnold, Christine, Franklin, Mark, Wlezien, Christopher, Russo, Luana & Palacios, Irene (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes Eurobarometer Trendfile. Data File Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, Wlezien, Christopher, Russo, Luana & Palacios, Irene (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes Eurobarometer Trendfile User Guide. Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, Wlezien, Christopher, Sapir, Eliyahu & Williams, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes EES Voter Study Trendfile. Data File Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, Wlezien, Christopher, Sapir, Eliyahu & Williams, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes EES Voter Study Trendfile User Guide. Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, Wlezien, Christopher, Sapir, Eliyahu & Williams, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes ISSP Role of Government Trendfile. Data File Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, Wlezien, Christopher, Sapir, Eliyahu & Williams, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes ISSP Role of Government Trendfile User Guide. Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, Wlezien, Christopher, Sapir, Eliyahu & Williams, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes Party Positions Trendfile. Data File Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, Wlezien, Christopher, Sapir, Eliyahu & Williams, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes Party Positions Trendfile User Guide. Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, & Wlezien, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes National Budgets Trendfile. Data File Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, & Wlezien, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes National Budgets Trendfile User Guide. Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, & Wlezien, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes European Union Budget Trendfile. Data File Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, & Wlezien, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes European Union Budget Trendfile User Guide. Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, & Wlezien, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes European Union Legislation Trendfile. Data File Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, & Wlezien, Christopher (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes European Union Legislation Trendfile User Guide. Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, Wlezien, Christopher, & Rahmani, Hossein (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes National Legislation Trendfile. Data File Version 1.0.0, https://doi.org/10.7802/2618 - Arnold, Christine, Franklin, Mark, Wlezien, Christopher, & Rahmani, Hossein (2023): PolicyVotes Database on Political Responsiveness. PolicyVotes National Legislation Trendfile User Guide. Version 1.0.0, https://doi.org/10.7802/2618
Monetary policy is generally regarded as a central element in the attempts of policy makers to attenuate business-cycle fluctuations. According to the New Keynesian paradigm, central banks are able to stimulate or depress aggregate demand in the short run by adjusting their nominal interest rate targets. The effects of interest rate changes on aggregate consumption, the largest component of aggregate demand, are well understood in the context of this paradigm, on which the canonical "workhorse'' model used in monetary policy analysis is grounded. A key feature of the model is that aggregate consumption is fully described by the amount of goods consumed by a representative household. A decline in the policy rate for instance implies that the real interest rate declines, the representative household saves less and hence increase its demand for consumption. At the same time, general equilibrium effects let labour income grow causing consumption to increase further. However, the mechanism outlined above ignores a considerable amount of empirically-observed heterogeneity among households. For example, households with a higher earnings elasticity to interest rate changes benefit more from a rate cut than those with a lower elasticity; households with large debt positions are at a relative advantage over households with large bond holdings; and households with low exposure to inflation are relatively better off than those holding a sizeable amount of nominal assets. As a result, the contribution to the aggregate consumption response differs substantially across households, implying that monetary expansions and tightenings produce relative "winners'' and relative "losers''.
The aim of the project laid out in this proposal is to give a disaggregated account of the heterogeneous effects of monetary-policy induced interest rate changes on household consumption and a detailed analysis of the channels underlying them. Additionally, it seeks to draw conclusions about the determinants of the strength of the transmission mechanism of monetary policy. To do so, it relies on a large panel comprising detailed data from the universe of all households residing in Norway between 1993 and 2015 supplemented with additional micro-data provided by the European Commission. I will be assisted by two project partners, Pascal Paul who is a member of the Research Department of the Federal Reserve Bank of San Francisco and Martin Holm who is affiliated with the Research Unit of Statistics Norway and the University of Oslo. In addition, I would like to collaborate with and help train a doctoral student based at the University of Lausanne on this project.
Existing empirical studies of the consumption response to monetary policy at the micro level rely on survey data. Therefore, they are subject to a number of severe data limitations. The surveys employed typically have either no or only a short panel dimension, suffer from attrition, include only limited information on income and wealth, are top-coded, and contain a significant amount of measurement error. The administrative data set provided to us by Statistics Norway suffers from none of these issues, implying that we are in a unique position to evaluate the household-level effects of policy rate changes. In a first step, we use forecasts published by the Norwegian central bank to derive monetary policy shocks that are robust to the simultaneity problem inherent in the identification of the effects of monetary policy following Romer and Romer (2004). We then confront the micro-data with the estimated shocks to study the consumption response along different segments of the income and wealth distribution and to test the importance of heterogeneity in labour earnings, financial income, liquid assets, inflation exposure and interest rate exposure among others. The findings will be of high relevance as they will not only allow us to evaluate channels hypothesised in the analytical literature, improve our understanding of the monetary policy transmission mechanism and its distributional consequences but also serve as a benchmark for structural models built both by theorists and practitioners.