pith. sign in

arxiv: 2509.16115 · v3 · submitted 2025-09-19 · 💰 econ.EM · stat.AP

A Korean Macroeconomic Database for Data-Rich Policy Analysis and U.S.--Korea Dependence

Changryong Baek , Seunghyun Moon , Seunghyeon Lee This is my paper

Pith reviewed 2026-05-18 15:31 UTC · model grok-4.3

classification 💰 econ.EM stat.AP
keywords Korean macroeconomicsfactor-augmented VARmonetary policy transmissionU.S.-Korea dependencetensor autoregressionprincipal componentscross-country spilloversfinancial conditions
0
0 comments X

The pith

New Korean database shows U.S. monetary tightening transmits strongly to Korea via financial factors.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces KRED, a monthly macroeconomic database with 125 series for Korea that matches the structure of U.S. FRED-MD data and reaches back to 1960. Principal components analysis on a balanced 2009-2025 panel extracts four factors that track financial conditions, real activity, housing and real-estate credit, and labor-market and price pressures. These factors are inserted into VAR models, where U.S. monetary tightening produces larger and more coherent effects on Korean variables than in low-dimensional alternatives. A grouped tensor autoregression then isolates the strongest cross-country linkages inside financially oriented blocks and finds that spillovers run more forcefully from the U.S. financial block to Korea than in the opposite direction.

Core claim

KRED supplies 125 Korean series for data-rich work; four principal components from the recent balanced panel capture financial conditions, real activity, housing credit, and labor-price pressures; factor-augmented VARs document strong transmission of U.S. monetary tightening to Korea together with clearer inflation responses; grouped U.S.-Korea tensor autoregression concentrates cross-country dependence in financial blocks and shows asymmetric transmission from the U.S. financial block to Korea.

What carries the argument

Four principal components extracted from the Korean panel that enter factor-augmented VARs and grouped tensor autoregressions to measure monetary transmission and cross-border spillovers.

Load-bearing premise

The four principal components extracted from the 2009-2025 panel correspond to the labeled economic concepts and stay stable enough to support the transmission conclusions.

What would settle it

If the diffusion indices built from the four factors fail to track the timing or direction of major Korean episodes such as the 2020 pandemic shock or the 2022 inflation surge, the economic interpretation and the reported transmission results would be undermined.

Figures

Figures reproduced from arXiv: 2509.16115 by Changryong Baek, Seunghyeon Lee, Seunghyun Moon.

Figure 1
Figure 1. Figure 1: Scree plot [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Time plot of four factors. With all four factors together, [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Explanatory power of the factors in R2 . T5YFFM BAAFFM AAAFFM INDPRO GS5 GS1 PERMITMW PERMITW PERMITS T10YFFM PAYEMS CLF16OV SRVPRD USGOV IPDCONGD CE16OV CES3000000008 CES0600000007 ACOGNO IPFUELS IPNCONGD CES0600000008 MANEMP CP3Mx HWI INVEST M1SL USTPU CPIMEDSL DTCTHFNM HWIURATIO CUSR0000SAS USWTRADE EXJPUSx CPITRNSL UEMP15OV UEMP15T26 UEMP27OV EXSZUSx PPICMM Sorted R²(4) Values R 2 i (4) 0.0 0.2 0.4 0.6… view at source ↗
Figure 4
Figure 4. Figure 4: Explanatory power of the factors in R2 (4). 8 [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Diffusion indices. South Korea’s macroeconomic fluctuations over the 2009–2024 period, effectively capturing both global shocks and domestic cycles. Major downturns such as the 2020 COVID-19 recession are clearly visible across multiple indexes, with F1 (monetary conditions) and F3 (real activity) showing the most pronounced responses. The 2015–2016 slowdown, often underemphasized in headline data, emerges… view at source ↗
read the original abstract

We introduce KRED (Korea Research Economic Database), a FRED-MD-compatible monthly macroeconomic database for Korea designed for data-rich policy analysis and cross-country comparison. KRED contains 125 monthly series from ECOS, KOSIS, and administrative labor-market sources, with coverage back to 1960. Using a balanced panel of 104 series over 2009:06--2025:12, principal-components analysis extracts four factors that explain about 30% of total variation. These factors correspond to financial conditions, real activity, housing and real-estate credit, and labor-market and price pressures, and their diffusion indices summarize major Korean macroeconomic episodes. We then use KRED in two empirical applications. First, factor-augmented VARs show that U.S. monetary tightening transmits strongly to Korea and that factor augmentation yields a more coherent inflation response than a low-dimensional VAR. Second, a grouped U.S.--Korea tensor autoregression shows that cross-country dependence is concentrated in financially oriented blocks, with stronger transmission from the U.S. financial block to Korea than in the reverse direction, while spillovers in real activity and housing are much weaker. KRED thus provides a transparent public database for Korean macroeconomic research and a useful building block for comparative work on macro-financial dependence in Asia.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 3 minor

Summary. The paper introduces KRED, a FRED-MD-compatible monthly macroeconomic database for Korea containing 125 series back to 1960. From a balanced panel of 104 series over 2009:06--2025:12, PCA extracts four factors explaining ~30% of variation, interpreted via diffusion indices as financial conditions, real activity, housing and real-estate credit, and labor-market and price pressures. These factors are then used in FAVARs to document strong transmission of U.S. monetary tightening to Korea with more coherent inflation responses than low-dimensional VARs, and in a grouped U.S.--Korea tensor autoregression to show cross-country dependence concentrated in financial blocks with asymmetric spillovers favoring U.S.-to-Korea transmission.

Significance. If the factor interpretations prove robust, the paper's primary contribution is the public release of a transparent, high-frequency Korean macro database that enables data-rich policy analysis and facilitates U.S.--Korea comparative work. The empirical applications illustrate how factor augmentation and tensor methods can sharpen evidence on macro-financial spillovers, which is relevant for Asian policy research. The modest 30% explained variation and reliance on post-extraction labeling are acknowledged limitations but do not negate the database's standalone value.

major comments (3)
  1. [Principal components extraction and factor labeling] The interpretation that the four principal components correspond to financial conditions, real activity, housing/real-estate credit, and labor/price pressures rests on narrative summaries of diffusion indices without reported factor loadings, variable-specific R² contributions, or subsample stability diagnostics over 2009:06--2025:12. This is load-bearing for both applications: the FAVAR transmission results and the tensor-AR block spillovers cannot be reliably attributed to the labeled concepts if the financial-conditions factor mixes real and nominal series or shifts post-2020.
  2. [Factor-augmented VAR application] The FAVAR section claims that factor augmentation produces a more coherent inflation response to U.S. monetary tightening than a low-dimensional VAR, yet provides no explicit criteria for retaining exactly four factors, no robustness checks across alternative factor counts, and no description of how error bands are constructed for the impulse responses. These omissions directly affect the strength of the transmission and coherence conclusions.
  3. [Grouped U.S.--Korea tensor autoregression] In the grouped tensor autoregression, the finding that cross-country dependence is concentrated in financially oriented blocks with stronger U.S.-to-Korea transmission is presented without subsample stability tests or break diagnostics for the underlying factors. Given that only ~30% of variation is explained, it is unclear whether the remaining idiosyncratic components alter the reported block-specific spillover patterns.
minor comments (3)
  1. [Data and database construction] The database construction section would benefit from an explicit list of transformations applied to achieve stationarity, paralleling the FRED-MD documentation.
  2. [Introduction and literature] Standard references to the FAVAR literature (e.g., Bernanke, Boivin, and Eliasz 2005) and to tensor autoregressive models should be added for context.
  3. [Figures and tables] Impulse-response and diffusion-index figures should include explicit confidence bands and legends to improve readability.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the thoughtful and constructive comments on our manuscript. We address each of the major comments point by point below and describe the revisions we plan to make to improve the paper's transparency and robustness.

read point-by-point responses

  1. Referee: The interpretation that the four principal components correspond to financial conditions, real activity, housing/real-estate credit, and labor/price pressures rests on narrative summaries of diffusion indices without reported factor loadings, variable-specific R² contributions, or subsample stability diagnostics over 2009:06--2025:12. This is load-bearing for both applications: the FAVAR transmission results and the tensor-AR block spillovers cannot be reliably attributed to the labeled concepts if the financial-conditions factor mixes real and nominal series or shifts post-2020.

    Authors: We agree that explicit documentation of the factor loadings and their contributions would strengthen the interpretation. In the revised manuscript, we will include a table reporting the factor loadings for the top variables in each factor, along with the R-squared contributions of individual series to each factor. Additionally, we will conduct and report subsample stability checks, splitting the sample at 2020 to assess whether the factor interpretations hold post-pandemic. These additions will allow readers to evaluate the robustness of the labeling. revision: yes

  2. Referee: The FAVAR section claims that factor augmentation produces a more coherent inflation response to U.S. monetary tightening than a low-dimensional VAR, yet provides no explicit criteria for retaining exactly four factors, no robustness checks across alternative factor counts, and no description of how error bands are constructed for the impulse responses. These omissions directly affect the strength of the transmission and coherence conclusions.

    Authors: We acknowledge the need for greater transparency in the factor selection and inference procedures. The choice of four factors is motivated by the cumulative explained variance reaching approximately 30% and by standard information criteria such as the Bai-Ng criteria, which we will now report explicitly. We will add robustness checks using three, five, and six factors to confirm that the main transmission results are not sensitive to this choice. For the impulse response error bands, we will clarify that they are constructed using the bootstrap procedure described in the appendix, and we will provide more details on the implementation. revision: yes

  3. Referee: In the grouped tensor autoregression, the finding that cross-country dependence is concentrated in financially oriented blocks with stronger U.S.-to-Korea transmission is presented without subsample stability tests or break diagnostics for the underlying factors. Given that only ~30% of variation is explained, it is unclear whether the remaining idiosyncratic components alter the reported block-specific spillover patterns.

    Authors: We will incorporate subsample stability tests for the tensor autoregression by re-estimating the model on pre- and post-2020 subsamples and comparing the block-specific spillover estimates. Regarding the idiosyncratic components, we note that the tensor autoregression is applied to the factors which capture the common variation across series; the idiosyncratic parts are by construction orthogonal to the factors and thus do not directly enter the cross-country dependence analysis. However, to address the concern, we will discuss the potential implications and add a robustness exercise using a larger number of factors that explain more variation. revision: partial

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained empirical application

full rationale

The paper constructs a new database KRED, extracts four factors via standard principal components analysis on the 2009:06--2025:12 balanced panel of 104 series, supplies narrative labels and diffusion-index summaries for those factors, and applies off-the-shelf FAVAR and grouped tensor autoregression models to the resulting factors and data. No equation or step reduces the reported transmission or spillover results to quantities defined by the paper's own fitted parameters or by a self-citation chain. Factor interpretation rests on post-estimation narrative rather than self-definition, and the central claims remain falsifiable against external benchmarks or alternative factor specifications. This is the normal, non-circular outcome for a data-construction-plus-application paper.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claims rest on standard econometric assumptions for principal components and VAR models plus the interpretive mapping of factors to economic concepts; no new entities are postulated and only one obvious free parameter (number of factors) is chosen.

free parameters (1)
  • Number of retained factors = 4
    Four factors selected because they explain about 30% of total variation in the 104-series panel.
axioms (2)
  • domain assumption Principal components extracted from the balanced panel correspond to interpretable macroeconomic concepts (financial conditions, real activity, housing/real-estate credit, labor/price pressures).
    Invoked when labeling the four factors and using them in subsequent VAR and tensor models.
  • domain assumption The 2009:06--2025:12 balanced panel is representative for studying US-Korea transmission and dependence.
    Underlies both empirical applications.

pith-pipeline@v0.9.0 · 5770 in / 1663 out tokens · 42782 ms · 2026-05-18T15:31:27.114785+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

6 extracted references · 6 canonical work pages

  1. [1]

    and Ng, S

    Bai, J. and Ng, S. (2002), ‘Determining the number of factors in approximate factor models’,Econometrica 70(1), 191–221

    work page 2002

  2. [2]

    McCracken, M. W. and Ng, S. (2016), ‘Fred-md: A monthly database for macroeconomic research’,Journal of Business & Economic Statistics34(4), 574–589. Also available as Federal Reserve Bank of St. Louis Working Paper 2015-012B

    work page 2016

  3. [3]

    (2024), ‘Everybody loves fred: How america fell for a data tool’,The New York Times

    Smialek, J. (2024), ‘Everybody loves fred: How america fell for a data tool’,The New York Times. Available online

    work page 2024

  4. [4]

    Stock, J. H. and Watson, M. W. (1996), ‘Evidence on structural instability in macroeconomic time series relations’,Journal of Business & Economic Statistics14(1), 11–30

    work page 1996

  5. [5]

    Stock, J. H. and Watson, M. W. (2002), ‘Macroeconomic forecasting using diffusion indexes’,Journal of Business & Economic Statistics20(2), 147–162

    work page 2002

  6. [6]

    Stock, J. H. and Watson, M. W. (2005), ‘Implications of dynamic factor models for var analysis’,NBER Working Paper Series(11467). URL:https://www.nber.org/papers/w11467 17

    work page 2005