Sub-sample Estimation Research Articles

Mean square consistency and improved rate of convergence of generalized subsampling estimator for non‐stationary time series

In this article, we show the mean square consistency for a generalized subsampling estimator based on the aggregation of the mean, median, and trimmed mean of some subsampling estimators for general non‐stationary time series. Consistency requires standard assumptions, including the existence of moments and ‐mixing conditions. We apply our results to the Fourier coefficients of the autocovariance function of periodically correlated time series. Furthermore, as in the i.i.d. case, we show that the generalized subsampling estimator satisfies Bernstein inequality and concentrates at an improved rate (under the condition of no or small bias) compared with the original estimator. Finally, we illustrate our results with some simulation data examples.

An optimal subsampling design for large-scale Cox model with censored data

Subsampling designs are useful for reducing computational load and storage cost for large-scale data analysis. For massive survival data with right censoring, we propose a class of optimal subsampling designs under the widely-used Cox model. The proposed designs utilize information from both the outcome and the covariates. Different forms of the design can be derived adaptively to meet various targets, such as optimizing the overall estimation accuracy or minimizing the variation of specific linear combination of the estimators. Given the subsampled data, the inverse probability weighting approach is employed to estimate the model parameters. The resultant estimators are shown to be consistent and asymptotically normally distributed. Simulation results indicate that the proposed subsampling design yields more efficient estimators than the uniform subsampling by using subsampled data of comparable sample sizes. Additionally, the subsampling estimation significantly reduces the computational load and storage cost relative to the full data estimation. An analysis of a real data example is provided for illustration.

The A-optimal subsampling approach to the analysis of count data of massive size

The uniform and the statistical leverage-scores-based (nonuniform) distributions are often used in the development of randomised algorithms and the analysis of data of massive size. Both distributions, however, are not effective in extraction of important information in data. In this article, we construct the A-optimal subsampling estimators of parameters in generalised linear models (GLM) to approximate the full-data estimators, and derive the A-optimal distributions based on the criterion of minimising the sum of the component variances of the subsampling estimators. As calculating the distributions has the same time complexity as the full-data estimator, we generalise the Scoring Algorithm introduced in Zhang, Tan, and Peng ((2023), ‘Sample Size Determination forMultidimensional Parameters and A-Optimal Subsampling in a Big Data Linear Regression Model’, To appear in the Journal of Statistical Computation and Simulation. Preprint. Available at https://math.indianapolis.iu.edu/hanxpeng/SSD_23_4.pdf) in a Big Data linear model to GLM using the iterative weighted least squares. The paper presents a comprehensive numerical evaluation of our approach using simulated and real data through the comparison of its performance with the uniform and the leverage-scores- subsamplings. The results exhibited that our approach substantially outperformed the uniform and the leverage-scores subsamplings and the Algorithm significantly reduced the computing time required for implementing the full-data estimator.

International co-movements of inflation, 1851–1913

Abstract We study inflation in a group of 15 countries before and during the classical Gold Standard using annual data spanning 1851–1913. The degree of co-movements between domestic and international inflation depends on geographical remoteness and openness to trade. Furthermore, international inflation acts as an ‘attractor’ for domestic inflation. Sub-sample estimates reveal little evidence of instability implying that international inflation was an important influence on domestic inflation throughout this time period.

Nominal and real wages in the UK, 1750–2015: mean reversion, persistence and structural breaks

This paper analyses the stochastic properties of UK nominal and real wages over the period 1750–2015 using fractional integration techniques. Both the original series and logged ones are analysed. The results generally suggest that nominal wages exhibit a higher degree of persistence, which reflects relatively long lags between inflation and wage adjustments. Endogenous break tests are also carried out and various structural breaks are identified in both series. On the whole, the corresponding subsample estimates imply an increase over time in the degree of persistence of both series.

Is manufacturing trade a stepping stone or an obstacle to service trade?

PurposeThe authors empirically examined the theoretically recognized industrial linkages between manufacturing and services from the trade perspective. In particular, they confirmed the trade effect of manufacturing on services, given that global value chain fragmentation pervades and splits manufacturing and services segments separately in developed and developing countries.Design/methodology/approachBased on observations of 47 countries with manufacturing and service trade data from 1990 to 2020 and with gravity model specification, the authors primarily used the Poisson pseudo-maximum likelihood (PPML) estimation with multiple levels of fixed effects. Considering that many zero values are included in the dependent variable and potential endogeneity, other methods such as Tobit regression, Heckman estimation and two-stage least squares estimation (2SLS) are used. Subsample estimation also supplemented the empirical research.FindingsThe results showed that manufacturing trade is a stepping-stone rather than an obstacle to service trade. This finding exhibited significant robustness under different model specifications, instrumental variable estimation and subsample checks. Moreover, in contrast to the north–north country ties, manufacturing trade between northern and southern countries has played a prominent stepping-stone role; meanwhile, manufacturing trade among core–peripheral countries has a considerably more significant impact than the outcomes of core–core and peripheral–peripheral countries.Originality/valueThe authors provided direct clarification and revealed that trade in manufacturing remains the demand basis for service trade. As trade in manufacturing and services are typical phenomena of transnational production linkages, the authors suggested exploring the underlying role of global value chain (GVC) fragmentation and the offset and even barrier effect of biased institutional arrangements on GVC fragmentation.

Green Finance Policy and ESG Performance: Evidence from Chinese Manufacturing Firms

While the literature has examined the key role of green finance policy on firms’ green innovation and environmental performance, little attention has been paid to firms’ environmental, social, and governance (ESG) performance, which is increasingly important to stakeholders. Exploiting heterogeneity in firms’ exposure to the green finance pilot zones policy in China in 2017 as a quasi-natural experiment, this paper employs the difference-in-differences model to explore the effect of green finance policy on firms’ ESG performance. Based on the data of listed manufacturing firms in China during 2013–2020, our results indicate that the green finance policy could promote firms’ ESG performance. Moreover, the overall positive effect is driven mainly by the environmental pillar. Utilizing subsample estimation and the triple differences method, we further find that the higher ESG performance is driven by firms with less financial constraints, firms in economically more developed pilot zones, and state-owned enterprises (SOEs). Mechanism analysis indicates that the pilot policy promotes firms’ ESG performance even if it worsens firms’ financial constraints. Our study contributes to the research on both the impacts of green finance policy and the relationship between financial constraints and ESG performance, as well as to the literature on ESG structure.

U.S. House Prices by Census Division: Persistence, Trends and Structural Breaks

This paper uses fractional integration methods to examine persistence, trends and structural breaks in United States house prices, more specifically the monthly Federal Housing Finance Agency House Price Index for census divisions, and the United States as a whole over the period from January 1991 to August 2022. The full sample estimates imply that the order of integration of the series is above one in all cases, and is particularly high for the aggregate series, implying high levels of persistence. However, when the possibility of structural breaks is taken into account, segmented trends are detected. The subsample estimates of the fractional differencing parameter tend to be lower, with mean reversion occurring in a number of cases. This means that shocks in the series are expected to be transitory in these subsamples, disappearing in the long run by themselves. In addition, the time trend coefficient is at its highest in the last subsample, which in most cases starts around May 2020 coincident with the beginning of the coronavirus pandemic. The results provide clear evidence of differences between census divisions, which implies that appropriate housing policies should be designed at the local (rather than at the federal) level.

Corporate governance and the outward foreign direct investment: Firm-level evidence from China

This paper investigates the role of corporate governance in China’s outward foreign direct investment (ODI) using a sample of Chinese public firms over the period 2009–2015. Different from previous studies, which mainly use country-level aggregate data to examine China’s ODI behavior, this study utilizes firm-level data to explore the impact of corporate governance on firms’ ODI decision making. Building on the integration of agency theory and foreign direct investment theories, the study finds that ownership concentration, share ownership, and executive compensation significantly influence China’s ODI flows in both full-sample and sub-sample estimations. The corporate governance has an essential effect on firms’ ODI flows in that the firms with better corporate governance are more interested in undertaking ODI projects. The results are robust across different sample periods and in both state-owned and non-state-owned firm sub-groups. The government may design some policies such as friendly tax regulations to stimulate private firms to invest overseas.

Financial development, foreign trade, regional economic development level and carbon emissions

Drawing on balanced panel data of 30 Chinese provinces in 2000–2020, this paper uses the Panel Smooth Transformation Regression (PSTR) model to explore the impact of financial development and foreign trade on carbon emissions under different regional economic development levels. The empirical results show that: 1) Financial development and foreign trade have a non-linear impact on carbon emissions under different economic development levels; 2) As the level of economic development exceeds the threshold, the positive effect of financial development on carbon emissions will weaken, while the effect of foreign trade on carbon emissions will change from negative to positive; 3) The sub-sample estimates further found that the impact on carbon emissions in southern and northern regions are different. The threshold in the south is lower than that in the north, but all the conversion speed is faster.

Sampling With Replacement vs Poisson Sampling: A Comparative Study in Optimal Subsampling

Faced with massive data, subsampling is a commonly used technique to improve computational efficiency, and using nonuniform subsampling probabilities is an effective approach to improve estimation efficiency. For computational efficiency, subsampling is often implemented with replacement or through Poisson subsampling. However, no rigorous investigation has been performed to study the difference between the two subsampling procedures such as their estimation efficiency and computational convenience. This paper performs a comparative study on these two different sampling procedures. In the context of maximizing a general target function, we first derive asymptotic distributions for estimators obtained from the two sampling procedures. The results show that the Poisson subsampling may have a higher estimation efficiency. Based on the asymptotic distributions for both subsampling with replacement and Poisson subsampling, we derive optimal subsampling probabilities that minimize the variance functions of the subsampling estimators. These subsampling probabilities further reveal the similarities and differences between subsampling with replacement and Poisson subsampling. The theoretical characterizations and comparisons on the two subsampling procedures provide guidance to select a more appropriate subsampling approach in practice. Furthermore, practically implementable algorithms are proposed based on the optimal structural results, which are evaluated through both theoretical and empirical analyses.

Consumers' food control risk attitude for blockchain traceable information seeking: Evidence from fresh fruit buyers in China

The blockchain-based traceability in agri-food marketing has brought a disruptive paradigm shift by removing the inherent information asymmetry problem. Likewise, revealing sufficient product quality and attributes information could break agricultural markets' “Lemon Market” dilemma. This study takes the fresh fruit with blockchain traceability QR label as a case and systematically investigates the influence of consumers' food control risk attitude on information-seeking intentions. We utilized online survey data of 1,058 fresh fruit buyers and simultaneously applied ordinary least square (OLS), ordered logit model (Ologit), and propensity score matching (PSM) approaches to overcome the potential self-selection biases and confounding factors. The results show that risk attitude significantly negatively impacts consumers' information seeking fresh fruits. The stronger consumers' risk preference, the lower the probability of information seeking. Furthermore, we used PSM to overcome potential sample selectivity bias; therefore, PSM reinforces the significance of OLS and Ologit results. The sub-sample estimation results show that young individuals with high school and below education have stronger blockchain information-seeking intentions. The study provides new insights into the role of food control risk attitude and agri-food information traceability and offers several measures for policy and practice to realize a border trajectory in agri-food information disclosure.

Corn Grain or Corn Silage: Effects of the Grain-to-Fodder Crop Conversion Program on Farmers’ Income in China

The Grain-to-Fodder Crop Conversion Program (GCCP) in China mainly promotes the green and sustainable development of grain crops, economic crops, and silage crops by subsidizing livestock farms to encourage farmers to plant silage crops, such as corn silage and alfalfa silage. In this context, this study assesses the impact of planting silage crops on farm household income. Based on a survey of 495 households in Henan and Hebei Provinces, China, we first constructed a theoretical model of the program’s effect on farmers’ income, and then used an ordinary least squares (OLS) method to estimate the magnitude of the GCCP on farmers’ income. To identify endogeneity and further test the stability of the results, we adopted instrumental variable estimation, subsample estimation, and matching methods. The GCCP significantly increased smallholder farm income. Compared with growing corn grain, corn silage increased income by approximately CNY 101/mu. Meanwhile, corn silage reduced the capital input cost of farmers by 10.71% per mu and labor input by 26.6% per mu. Heterogeneity analysis revealed that farmers who plant corn silage on a large scale, closer to dairy farms, have higher incomes. Few scholars have empirically analyzed the impact of GCCP on farm household income from a micro-farm household perspective. This study enriches the empirical literature on the effects of the GCCP on farmers in China, which can help policymakers understand policy implementation.

A two-stage optimal subsampling estimation for missing data problems with large-scale data

Subsampling is useful to downsize data volumes and speed up calculations for large-scale data and is well studied with completely observed data. In the presence of missing data, computation is more challenging and subsampling becomes more crucial and complex. However, there is still a lack of study on subsampling for missing data problems. This paper fills the gap by studying the subsampling method for a widely used missing data estimator, the augmented inverse probability weighting (AIPW) estimator. The response mean estimation problem with missing responses is discussed for illustration. A two-stage subsampling method is proposed via Poisson sampling framework. A small subsample of expected size n1 is used in the first stage to estimate the parameters in the propensity score and the outcome regression models, while a larger subsample of expected size n2 is used in the computationally simple second stage to calculate the final estimator. An attractive property of the resulting estimator is that its convergence rate is n2−1/2 rather than n1−1/2 when both the propensity score and the outcome regression functions are correctly specified. The rate n2−1/2 is still attainable for some important cases if only one of the two functions is correctly specified. This indicates that using a small subsample in the computationally complex first stage can reduce the computational burden with little impact on the statistical accuracy. Asymptotic normality of the resulting estimator is established and the optimal subsampling probability is derived by minimizing the asymptotic variance of the resulting estimator. Simulations and a real data analysis were conducted to demonstrate the empirical performance of the resulting estimator.

Adaptive subsample estimation for multivariate normal distributions

The minimum covariance determinant (MCD) estimate is an important subsample method in robust statistics. It is commonly used to estimate parameters of multivariate normal distributions when outliers exist because of its good robustness properties. However, MCD has some disadvantages including low efficiency when there is no outliers and poor performance when there are clustered outliers. This paper first introduces an adaptive subsample method, and shows that the adaptive MCD estimator can possess both full asymptotic efficiency and maximum breakdown value. We then propose a minimum distance subsample estimate to handle the situations where there are clustered outliers. Simulation results indicate that the adaptive version of the minimum distance subsample estimate has satisfactory performance for various types of outliers.

An empirical analysis of the global environmental impact of press freedom: The role of internet access and industry size in 153 countries

The present study attempts to empirically establish the global environmental impacts of press freedom and internet access while controlling for the role of industry. This is based on the premise that the press' role as information disseminator can be strategic to the sustainable development paradigm. The study employs a generalized method of moment technique and a bias corrected fixed effects estimator within a panel dataset of 153 countries over the 2002 to 2016 study period. The present study finds that when the moderation effect of industry is not controlled for, a 1% increase press freedom is associated with a 0.022% decrease in CO2 emissions. When the non-linear effect of internet access is not controlled for, a 1% increase in internet access is associated with a 0.018% increase in CO2 emissions. With the incorporation of non-linearities, internet access has an inverted U-shape relationship with CO2 emissions with a required access penetration of 76.2% before its carbon mitigation threshold can be attained. Also, press freedom constrains emissions in economies with a small industry size but induces emissions in economies with a large industry size. Sub-sample estimations show inter-alia —that while internet access is environmentally degrading in lower-middle income countries, it mitigates pollution in high income countries. Policy implications include the improvement of internet access penetration on a global scale to attain the threshold required to counterbalance the effect of global warming as well as engaging media stakeholders on the need to identify with cleaner production practices and transmission of same practices to the greater society in order to enhance sustainable development.

Credit bureaus and financial constraints do corruption matter?

ABSTRACT This study aims to assess whether or not the presence of credit bureaus is associated with more or fewer financing constraints while considering the interfering effect of corruption in a sample of 18 countries in Eastern Europe and the Middle East and North Africa (MENA) region during the period 2011–2014. We consider various financial constraint measures and corruption indices, and assess the stability of the relationship for different levels of economic development and corruption. The estimation outcomes suggest that countries with higher levels of corruption might produce less transparent and falsified information that would make access to sources of financing more difficult for firms. Our findings suggest that curbing corruption creates more efficient credit bureaus that, in turn, decrease financial constraints for firms. The subsample estimations confirm these findings and show that the higher and longer-term corruption in MENA countries than in Eastern European countries make credit bureaus’ less effective, imposing more financial constraints. Our findings remain robust with different corruption indices and with the addition of new control variables such as firms’ sales and size, government and exporting firms, and per-capita GDP, inflation, trade, population and human capital.

Multi-Project Assessments of Sample Quality in Cross-National Surveys: The Role of Weights in Applying External and Internal Measures of Sample Bias

AbstractThis paper examines existing methods of evaluating sample quality, showing that their practical utility and applicability to large-scale cross-project comparisons depends on whether they require auxiliary individual-level data. Among those methods that do not demand any such additional data, we differentiate between two approaches that rely on (i) external criteria, that is, comparisons of sample estimates to benchmarks derived from external population statistics, and (ii) internal criteria, that is, comparisons of subsample estimates to a theoretically derived aprioristic value. Our analyses demonstrate the advantages and limitations of both approaches based on an evaluation of 1,125 national surveys carried out in Europe between 2002 and 2016 within four survey projects: the Eurobarometer, European Quality of Life Survey, European Social Survey, and International Social Survey Programme. We show that the prevailing absence of design weights in cross-national survey datasets severely limits the applicability of external criteria evaluations. In contrast, using internal criteria without any weights proves acceptable because incorporating design weights in calculations of internal sample quality has only minor consequences for estimates of sample bias. Furthermore, applying internal criteria, we find that around 75 percent of samples in the four analyzed projects are not significantly biased. We also identify surveys with extremely high sample bias and investigate its potential sources. The paper concludes with recommendations regarding future research, which are directed at secondary data users, as well as producers of cross-national surveys.

Financial development, openness, innovation, carbon emissions, and economic growth in China

Drawing on balanced panel data of 30 Chinese provinces in 1987–2017, this paper examines the impact of carbon emissions on economic growth through the panel smooth transition regression model. Estimation is conducted based on the whole sample as well as the northern region and southern region subsamples. Empirical results reveal that: i) noticeable non-linear relationships do exist among carbon emissions, financial development, openness, innovation, and economic growth; and ii) carbon emissions attenuate the promoting effects of financial development and innovation on economic growth, which is also confirmed by using energy consumption as the transition variable; and iii) subsample estimations further discover that the impact of carbon emissions on economic growth is significantly different between the two regions, with the northern region having a lower carbon emissions threshold but quicker transition speed.

Penalized KS method to fit data sets with power law distribution over a bounded subinterval

We develop a variation of a Kolmogorov-Smirnov (KS) method for estimating a power law region, including its lower and upper bounds, of the probability density in a set of data which can be modelled as a continuous random sample. Our main innovation is to stabilize the estimation of the bounds of the power law region by introducing an adaptive penalization term involving the logarithmic length of the interval when minimizing the Kolmogorov-Smirnov distance between the random sample and the power law fit over various candidate intervals. We show through simulation studies that an adaptively penalized Kolmogorov-Smirnov (apKS) method improves the estimation of the power law interval on random samples from various theoretical probability distributions. Variability in the estimation of the bounds can be further reduced when the apKS method is applied to subsamples of the original random sample, and the subsample estimates are averaged to yield a final estimate.