Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Discovery Grants Program - Individual

Carleton University

$28,000.00

Multivariate confidence procedures with application to dynamic responses and tail forecasts

FY 2024-25

Summary
In economics, many questions of interest involve multiple functions of parameters that define some model. When such questions are analyzed statistically, their multiplicity has to be taken into account, alongside a comparison of the model implications to what actually happened in the data, in and out-of-sample. Underlying models are often grounded on economic theory and policy foundations, which complicates links to actual data. This research program concerns quantifying estimation uncertainty in such contexts, where the model-to-data connection (i.e. "identification") is non-standard, and when the distribution of relevant estimators is complicated yet can be simulated. Specifically, contributions will focus on: (1) economic theory-founded (i.e. "structural") macro-econometric models that explain several observables using a smaller number of disturbances (i.e. are "stochastically singular"); (2) time series forecasting models that generate predictions for several measures of extreme outcomes. Empirical applications will cover: (3) structural models of the response to monetary policy and inflation shocks; (4) extreme value criteria that are the target of international regulatory accords; (5) forecasting models for environmentally responsible exchange traded funds, and for inflation. As we have seen over the past couple of years, modeling inflation is instrumental for the success of monetary policy. In parallel, the measurement of extreme outcomes is central to financial management and to the regulatory process of the financial system. Structural models are at the core of economic analysis since compatible data can readily be simulated, which enables policy experiments. The proposed work will take advantage of this property through (6) simulation-based solutions. The parameters of a singular model can be estimated using several subgroups of variables. This research program thus targets (7) identification-robust confidence procedures, to enable aggregation of information across potential data subsets. Extreme value analysis is typically based on tail measures, defined as specific metrics that quantify exposure to extreme events beyond given probability thresholds. Emphasis in this proposal is on the multiplicity of measures, that can convey different possibly conflicting assessments. Financial regulations often require large institutions to "backtest" such measures, which refers to formal comparisons of forecasts against realised values. The proposed work is motivated by such requirements. Specifically, this research program aims for a unified approach to confidence estimation of forecasts and their (8) out-of-sample validation, in a single step. This will permit joint interpretation of all measures at all probability thresholds. Macro-economic samples are typically short. While financial data sets are larger, backtests are often conducted on short assessment periods. Conformably, the proposed work will favour (9) finite sample validity.
Program
Discovery Grants Program - Individual
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
RGPIN-2024-05322
Name nom
Khalaf, Lynda
Department département
Economics
Organization id
23
Institution établissement
Carleton University
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2024
Award amount
28000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1508
Committee name en
Mathematics and Statistics
Committee name fr
Mathématiques et statistique
Area of application code
1212
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Social sciences
Area of application fr
Sciences sociales
Research subject code
3004
Research subject group en
Statistics and probability
Research subject group fr
Statistique et probabilité
Research subject en
Applied statistics
Research subject fr
Statistique appliquée
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Multivariate confidence procedures with application to dynamic responses and tail forecasts
Keywords
multivariate confidence sets; simultaneous inference; simulation-based inference; out-of-sample stability; tail forecasting; structural singularity
Application summary
In economics, many questions of interest involve multiple functions of parameters that define some model. When such questions are analyzed statistically, their multiplicity has to be taken into account, alongside a comparison of the model implications to what actually happened in the data, in and out-of-sample. Underlying models are often grounded on economic theory and policy foundations, which complicates links to actual data. This research program concerns quantifying estimation uncertainty in such contexts, where the model-to-data connection (i.e. "identification") is non-standard, and when the distribution of relevant estimators is complicated yet can be simulated. Specifically, contributions will focus on: (1) economic theory-founded (i.e. "structural") macro-econometric models that explain several observables using a smaller number of disturbances (i.e. are "stochastically singular"); (2) time series forecasting models that generate predictions for several measures of extreme outcomes. Empirical applications will cover: (3) structural models of the response to monetary policy and inflation shocks; (4) extreme value criteria that are the target of international regulatory accords; (5) forecasting models for environmentally responsible exchange traded funds, and for inflation. As we have seen over the past couple of years, modeling inflation is instrumental for the success of monetary policy. In parallel, the measurement of extreme outcomes is central to financial management and to the regulatory process of the financial system. Structural models are at the core of economic analysis since compatible data can readily be simulated, which enables policy experiments. The proposed work will take advantage of this property through (6) simulation-based solutions. The parameters of a singular model can be estimated using several subgroups of variables. This research program thus targets (7) identification-robust confidence procedures, to enable aggregation of information across potential data subsets. Extreme value analysis is typically based on tail measures, defined as specific metrics that quantify exposure to extreme events beyond given probability thresholds. Emphasis in this proposal is on the multiplicity of measures, that can convey different possibly conflicting assessments. Financial regulations often require large institutions to "backtest" such measures, which refers to formal comparisons of forecasts against realised values. The proposed work is motivated by such requirements. Specifically, this research program aims for a unified approach to confidence estimation of forecasts and their (8) out-of-sample validation, in a single step. This will permit joint interpretation of all measures at all probability thresholds. Macro-economic samples are typically short. While financial data sets are larger, backtests are often conducted on short assessment periods. Conformably, the proposed work will favour (9) finite sample validity.
Carleton University - $28,000 Grant (2024-25) | Canada Spends