Li Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Li Chen has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Deshui Yu
College of Finance and Statistics, Hunan University
6 shared publications - Difang Huang
Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China
4 shared publications - Luyang Li
Monash University - Department of Economics
3 shared publications - Haiqiang Chen
Wang Yanan Institute for Studies in Economics, Xiamen University, Xiamen, China
2 shared publications Bo Zhang
1 shared publicationDeshui Yu
1 shared publication
Latest Publications
- Open Access
- 23 September 2026
- Published by Elsevier in SSRN Electronic Journal
Does Training on Future Data Pay? Look-Ahead Bias in Forecasting with Pretrained Models
We examine whether post-origin training information inflates the measured accuracy and economic value of financial forecasts. We evaluate five sets of financial time-series foundation models, each comprising independently trained annual vintages under U.S., global, and factor-augmented training environments, across 14 equity markets and four forecast horizons. Rolling comparisons vary the annual vintage for a fixed forecast; fixed-vintage comparisons hold the vintage fixed as target windows move across its training cutoff. Each alternative forecast is paired with an origin-aligned point-in-time (PIT) benchmark using identical numerical histories and inference protocols. In the U.S.-trained reference environment, post-origin vintages materially revise informative PIT forecasts but generally reduce accuracy in both designs. Pooled rolling comparisons yield higher mean squared forecast errors in 18 of 20 U.S. model-set-horizon combinations. The origin-crossing update also performs worse on average than an equally long pre-origin update. Under a common constrained allocation rule using one-month forecasts, median exposed-minus-PIT differences in annualized certainty-equivalent returns are -1.77 percentage points in the United States and -2.14 points internationally. Global and factor-augmented training produce more mixed predictive effects. An exact squared-error decomposition shows that revisions improve accuracy when their error-correcting benefit exceeds their mean squared magnitude; under U.S. training, alignment with PIT errors generally falls short of this requirement. Temporal exposure therefore establishes an information-set violation, not sufficient evidence of inflated predictive accuracy or investor value.
- 23 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessDoes Training on Future Data Pay? Look-Ahead Bias in Forecasting with Pretrained Models
We examine whether post-origin training information inflates the measured accuracy and economic value of financial forecasts. We evaluate five sets of financial time-series foundation models, each comprising independently trained annual vintages under U.S., global, and factor-augmented training environments, across 14 equity markets and four forecast horizons. Rolling comparisons vary the annual vintage for a fixed forecast; fixed-vintage comparisons hold the vintage fixed as target windows move across its training cutoff. Each alternative forecast is paired with an origin-aligned point-in-time (PIT) benchmark using identical numerical histories and inference protocols. In the U.S.-trained reference environment, post-origin vintages materially revise informative PIT forecasts but generally reduce accuracy in both designs. Pooled rolling comparisons yield higher mean squared forecast errors in 18 of 20 U.S. model-set-horizon combinations. The origin-crossing update also performs worse on average than an equally long pre-origin update. Under a common constrained allocation rule using one-month forecasts, median exposed-minus-PIT differences in annualized certainty-equivalent returns are -1.77 percentage points in the United States and -2.14 points internationally. Global and factor-augmented training produce more mixed predictive effects. An exact squared-error decomposition shows that revisions improve accuracy when their error-correcting benefit exceeds their mean squared magnitude; under U.S. training, alignment with PIT errors generally falls short of this requirement. Temporal exposure therefore establishes an information-set violation, not sufficient evidence of inflated predictive accuracy or investor value.
- Open Access
- 27 April 2026
- Published by Elsevier in SSRN Electronic Journal
General-Purpose Technology and Speculative Bubble Detection
We show that the leading bubble test suffers severe size distortion when fundamentals incorporate general-purpose technology adoption. Embedding a hump-shaped technology shock in the Campbell-Shiller present-value model, we prove that the fundamental price becomes locally explosive during adoption, contaminating the test's limit distribution with a non-centrality parameter proportional to the shock's peak. We propose a fundamental-versus-speculative decomposition that projects prices onto observable technology proxies and applies the test to the residual. Empirically, the decomposition eliminates evidence of speculation in the 2020-2025 AI rally while confirming a speculative peak confined to December 1999-March 2000 in the dot-com episode.
- 27 April 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessGeneral-Purpose Technology and Speculative Bubble Detection
We show that the leading bubble test suffers severe size distortion when fundamentals incorporate general-purpose technology adoption. Embedding a hump-shaped technology shock in the Campbell-Shiller present-value model, we prove that the fundamental price becomes locally explosive during adoption, contaminating the test's limit distribution with a non-centrality parameter proportional to the shock's peak. We propose a fundamental-versus-speculative decomposition that projects prices onto observable technology proxies and applies the test to the residual. Empirically, the decomposition eliminates evidence of speculation in the 2020-2025 AI rally while confirming a speculative peak confined to December 1999-March 2000 in the dot-com episode.
- Open Access
- 3 February 2026
- Published by Elsevier in SSRN Electronic Journal
Testing Quantile Predictability with a Persistent Regressor
This paper proposes a consistent nonparametric test for quantile predictability in the presence of a persistent predictor. We construct a kernel-weighted U-statistic that aggregates information across multiple quantile levels, capturing joint predictability over a specified region of the conditional distribution. We demonstrate that, under the null hypothesis, the test statistic converges to a mixed normal distribution and remains robust to the embedded endogeneity commonly associated with persistent financial predictors. We further establish the consistency of the test against a broad class of local alternatives. To facilitate inference given the nonstandard limiting distribution, we develop a multiplier bootstrap procedure to obtain the distribution of the test statistic under the null hypothesis. Extensive Monte Carlo simulations demonstrate that the bootstrap-based test achieves accurate size control and strong power across varying degrees of regressor persistence under both linear and nonlinear alternatives. Finally, an empirical application to U.S. stock returns reveals significant predictability concentrated in the right tail of the distribution - a signal otherwise masked by standard predictive mean regressions.
- 3 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessTesting Quantile Predictability with a Persistent Regressor
This paper proposes a consistent nonparametric test for quantile predictability in the presence of a persistent predictor. We construct a kernel-weighted U-statistic that aggregates information across multiple quantile levels, capturing joint predictability over a specified region of the conditional distribution. We demonstrate that, under the null hypothesis, the test statistic converges to a mixed normal distribution and remains robust to the embedded endogeneity commonly associated with persistent financial predictors. We further establish the consistency of the test against a broad class of local alternatives. To facilitate inference given the nonstandard limiting distribution, we develop a multiplier bootstrap procedure to obtain the distribution of the test statistic under the null hypothesis. Extensive Monte Carlo simulations demonstrate that the bootstrap-based test achieves accurate size control and strong power across varying degrees of regressor persistence under both linear and nonlinear alternatives. Finally, an empirical application to U.S. stock returns reveals significant predictability concentrated in the right tail of the distribution - a signal otherwise masked by standard predictive mean regressions.
- 19 June 2025
- Published by Elsevier in Economics Letters
- Vol. 254
Testing for common trends and patterns in functional time series data
- 19 June 2025
- Published by Elsevier in Economics Letters
- Vol. 254
Testing for common trends and patterns in functional time series data
- 15 March 2024
- Published by Elsevier in Journal of Empirical Finance
- Vol. 77
Local predictability of stock returns and cash flows
- 15 March 2024
- Published by Elsevier in Journal of Empirical Finance
- Vol. 77
Local predictability of stock returns and cash flows
- 28 February 2023
- Published by Elsevier in Journal of Empirical Finance
- Vol. 72, 36-53
Stock return predictability and cyclical movements in valuation ratios
- 28 February 2023
- Published by Elsevier in Journal of Empirical Finance
- Vol. 72, 36-53
Stock return predictability and cyclical movements in valuation ratios
- 22 February 2023
- Published by Elsevier in Economics Letters
- Vol. 225
Nonparametric modeling for the time-varying persistence of inflation
- 22 February 2023
- Published by Elsevier in Economics Letters
- Vol. 225
Nonparametric modeling for the time-varying persistence of inflation
- 10 February 2023
- Published by Elsevier in Economics Letters
- Vol. 224
Time-varying predictability of the long horizon equity premium based on semiparametric regressions
- 10 February 2023
- Published by Elsevier in Economics Letters
- Vol. 224
Time-varying predictability of the long horizon equity premium based on semiparametric regressions
- Open Access
- 1 January 2023
- Published by Elsevier in SSRN Electronic Journal
Local Predictability of Stock Returns and Cash Flows
Motivated by the present-value framework, this article proposes a novel and flexible semiparametric time-varying model to examine the so-called `pockets of predictability,' i.e., stock returns or cash flows are significantly predictable in a given local period. We apply a semiparametric profile method to estimate both time-varying and constant parameters. The predictive ability of the dividend-price ratio for dividend growth is slightly stronger for stock returns, but the conclusion reverses at longer horizons of three and five years. In contrast, at both short and long horizons, the predictive ability of the earnings-price ratio for earnings growth is significant in all local periods, stronger for stock returns and dividend payout ratios. Surprisingly, dividend smoothing matters for cash flow predictability at the low frequency only. Moreover, the two valuation ratios predict either expected future discount rates or cash flows, or both, for most sample periods, indicating that the present-value relation holds precisely in local periods.
- 1 January 2023
- Published by Elsevier in SSRN Electronic Journal
Open AccessLocal Predictability of Stock Returns and Cash Flows
Motivated by the present-value framework, this article proposes a novel and flexible semiparametric time-varying model to examine the so-called `pockets of predictability,' i.e., stock returns or cash flows are significantly predictable in a given local period. We apply a semiparametric profile method to estimate both time-varying and constant parameters. The predictive ability of the dividend-price ratio for dividend growth is slightly stronger for stock returns, but the conclusion reverses at longer horizons of three and five years. In contrast, at both short and long horizons, the predictive ability of the earnings-price ratio for earnings growth is significant in all local periods, stronger for stock returns and dividend payout ratios. Surprisingly, dividend smoothing matters for cash flow predictability at the low frequency only. Moreover, the two valuation ratios predict either expected future discount rates or cash flows, or both, for most sample periods, indicating that the present-value relation holds precisely in local periods.
- Open Access
- 1 January 2023
- Published by Elsevier in SSRN Electronic Journal
Time-Varying Predictability of the Long Horizon Equity Premium Based on Semiparametric Regressions
This paper proposes a novel semiparametric time-varying model for long-horizon predictive regressions in which the coefficients are allowed to change over time with unspecified functional forms. A linear projection method is employed to deal with the embedded endogeneity issue. We pursue an indirect approach to estimate the long-horizon coefficients through the implication of the short-horizon coefficients, which is less subject to overlapping observations. Monte Carlo simulation illustrates that the nonparametric estimation method works well in finite and large samples. Empirically, the predictability of return or dividend growth changes significantly over time. The dividend-price ratio predicts stock return in most of our samples, but the dividend growth is much less predictable than the stock return. Moreover, the dividend-price ratio predicts either stock return or dividend growth, or both in a local period. Our findings are robust to several forecast horizons.
- 1 January 2023
- Published by Elsevier in SSRN Electronic Journal
Open AccessTime-Varying Predictability of the Long Horizon Equity Premium Based on Semiparametric Regressions
This paper proposes a novel semiparametric time-varying model for long-horizon predictive regressions in which the coefficients are allowed to change over time with unspecified functional forms. A linear projection method is employed to deal with the embedded endogeneity issue. We pursue an indirect approach to estimate the long-horizon coefficients through the implication of the short-horizon coefficients, which is less subject to overlapping observations. Monte Carlo simulation illustrates that the nonparametric estimation method works well in finite and large samples. Empirically, the predictability of return or dividend growth changes significantly over time. The dividend-price ratio predicts stock return in most of our samples, but the dividend growth is much less predictable than the stock return. Moreover, the dividend-price ratio predicts either stock return or dividend growth, or both in a local period. Our findings are robust to several forecast horizons.