FRAME - Documents and notes

Terms and items Notes
TBTF proxy "TBTF proxy" refers to effects of too-big-to-fail (TBTF) reforms, bail-in events, the effect of a bank being systemically important and/or possible differences between systemically important banks (SIBs) and non-SIBs of which the study assesses the impact. Note, for example, "Reform effect" can refer to a TBTF reform effect on banks (irrespective of the SIB-status of a bank) or exclusively on SIBs; using the breakdown option menu allows to identify the respective treatment group. In contrast, the "SIB x Reform effect" refers to the difference of a TBTF reform effect on the target variable of SIBs relative to non-SIBs.
Target Refers to the variable on which the study assesses the impact (e.g. bank funding cost, bank credit rating or bank contingent claims).
Target in levels/changes In the case of a regression, the target (dependent variable) is measured in "changes" if the estimate measures the amount by which the target has changed due to an event (the difference before and after an event); for example the change in funding costs due to a reform or a bail-in event. The target is measured in "levels", if the estimate represents the target level during a certain period; for example the level of funding cost advantage of systemically important banks during the pre-reform or the post-reform period.
Method (Target detail) This is the methodology and identification strategy used in the study; e.g. estimation of funding cost advantages using prices of financial market instruments (bond yield spreads, CDS, deposit rates, etc.), estimation of "issuer credit rating" uplift, estimation of contingent claims, etc. 
Regime (period) This is determined by the (sub-)sample period over which the impact has been estimated, e.g. "pre-crisis", "post-crisis", "crisis times", "pre-reform", "post-reform", "reform impact" and "bail-in impact". Example (1): "reform impact" corresponds to the impact estimated on a sample period that coincides with a reform event. In a linear regression, this would be the case if, for example, the "SIB effect" would be interacted with a reform dummy. In this linear regression, the "reform impact" estimate would be the coefficient of the "SIB dummy x reform dummy" interaction. The "pre-reform" estimate would be the coefficient of the "SIB dummy" variable, and the "post-reform" estimate would be the sum of the "SIB dummy" and the interaction "SIB dummy x reform dummy" coefficients. Example (2): The period before the global financial crisis could be defined as "pre-crisis", the global financial crisis period can be defined as "crisis times" and the period after the crisis can be defined as "post-crisis" period. Example (3): "bail-in impact" corresponds to the impact estimated on a sample period that coincides with a bail-in event. In a linear regression, this would be the case if, for example, the "SIB effect" would be interacted with a bail-in dummy. In this linear regression, the "bail-in impact" estimate would be the coefficient of the "SIB dummy x bail-in dummy" interaction. 
Sample Year Sample year is reported if the impact estimate corresponds to a specific year, otherwise is classified as N/A. For example:  If the study shows one estimate for 2017 and one for 2018, then the two different estimates are reported separately; adding in "Sample year" 2017 for the first estimate and 2018 for the second estimate. If the study only reports one estimate for the sample period 2017-2018, then the estimate is reported only once and the "Sample year" section has a "_N/A_".
Country Country corresponds to the ISO 2 digit code of the country, where the data come from. If the sample is composed of several countries, the acronym of the group of countries (e.g. Global, OECD, European Countries, AEs, EMEs, etc.) is reported.
Statistical significance This breakdown option reports the p-value buckets of the impact estimates as "p<0.01", "0.01<p<0.05", "0.05<p<0.1", or "p>0.1". When p-values are not available in the study a "_N/A_" is reported.
Data type (time series dimension) Time dimension of the data (monthly, yearly, etc.).
Treatment Group "Treatment group" corresponds to those banks or institutions that are in the focus of the respective study. For example, systemically important banks (SIBs) are the treatment group if the study estimates the reform effect on SIBs. In contrast, banks are the treatment group if the study estimates the reform effect on banks in general without distinguishing between SIBs and non-SIBs.
Control Group "Control group" corresponds to those banks or institutions that the study uses as comparison to the treatment group. If SIBs are the treatment group then non-SIBs are usually in the control group. Note however, that a control group is not reported for all studies. Examples without a control group are studies that analyse only a sample of SIBs in particular or banks in general (without differentiation between SIBs and non-SIBs). In such a case, the "Control group" is reported as "without control group".
Year of publication Year of the most recent (or published) version of the study/article/paper.
Source Source is determined by the editor of the publication, not by the affiliation of the authors: (1) A public institution; "Central banks" (e.g. working paper, quarterly review, annual report, etc.). (2) A refereed academic journal; "Academia". (3) A private institution (think tank, bank, etc.); "Private financial sector". For example, the study is considered academic if it has been published in an academic journal, even if the authors are from central banks.
Disclosure statement Indicates whether the authors of the study had relevant or material financial interests that relate to their analysis (e.g. research sponsored or commissioned by lobbying groups, trade associations, NGOs, or governmental bodies).

Additional information can be found in the reporting template.