Minjeong Kwak (p), Kwansoo Kim

Ioanna Tziolas

South Korea has achieved rapid economic growth through urbanization and industrialization, leading to a continuous decline in the proportion of rural population and agricultural income relative to urban area. To counteract this trend, the Korean government has implemented various agricultural support policies, including the Farmland Gift Tax Exemption Policy for Successor Farmers. This policy allows parents to transfer farmland to their children tax-free, reducing successor farmers’ land acquisition costs and supporting the sustainability of farming.
This study empirically examines whether this policy effectively fosters successor farmers. Specifically, it evaluates (1) the policy’s impact on maintaining the farming population and increasing farm income and (2) its effectiveness over multiple time periods.
Previous studies applied Generalized Propensity Score (GPS) to overcome the binary classification limitation of Propensity Score Matching (PSM) and used Difference-in-Differences (DID) to compare changes in the farming population and income before and after policy implementation. However, these approaches failed to isolate the specific effect of this policy, due to the simultaneous implementation of multiple farm income support programs. To address this issue, this study adopts Coarsened Exact Matching (CEM) to control for multiple policy benefits as a covariate, followed by GPS to analyze the policy effects as a continuous treatment variable. Additionally, Difference-in-Differences with multiple periods method will be used to examine changes in farm household age and income across two time periods, distinguishing between exclusive recipients, multiple-policy beneficiaries, or non-recipients.
Covering 2016 to 2024, the study divides this period into three four-year intervals to track policy effects. Specifically, it assesses whether the scale of farmland gift tax exemptions influences regional farm household age and income. Data from the National Tax Service (NTS) and Statistics Korea (KOSTAT) will be used to analyze tax benefit scales and farm household characteristics.
This research provides empirical evidence on the policy’s impact on successor farmer retention and farm income growth. Additionally, it evaluates whether the policy should be sustained and suggests future improvements. The findings will contribute to reevaluating successor farmer policies in Korea and globally, serving as a foundation for more effective agricultural policies to combat rural population decline.

Paul Elhorst, Petros Milionis, Ioanna Tziolas (p)

Alexandre Florindo Alves

We develop computationally efficient methods to estimate large three-dimensional spatial panel data models with multiple fixed effects using quasi-maximum likelihood (QML). We consider the spatial autoregressive (SAR) model with four common fixed effects specifications, including the most extensive with origin-time, destination-time, and pair fixed effects. In addition, we consider three extensions of the SAR model, either with multiple spatial lags in the regressand, spatial lags in the regressors, or a spatial lag in the error term.

To avoid biased parameter estimates due to incidental parameter problems, the estimation approach is based on the orthogonal transformation to demean the regressand and the regressors for the fixed effects. Furthermore, in order to use computationally efficient algorithms already developed for such models at one point in time, a lemma is developed which expresses the log-likelihood function of the model based on the orthogonal transformation into its counterpart based on the standard within transformation.

The algorithms already available cover shortcuts that reduce the number of parameters to be estimated and simplify matrix operations, such as the computation of determinants, inverses, and traces in the case of large spatial weight matrices, as well as algorithms to speed up the computation of direct and indirect effects of the regressors.

We illustrate the computational feasibility of our proposed methods by estimating several variants of the gravity model of trade with spatial lags, using a data set on regional trade flows among EU NUTS-II regions from 2000 to 2010. In all cases, we find empirical evidence in favor of local spatial dependence between trade flows affecting the estimated direct and indirect effects of common trade policy variables.

Alexandre Florindo Alves (p), Aline Andreotti Pancera

Flavio Vieira

The objective of the research is to study the distribution and concentration of employment in Knowledge Intensive Activities (KIA) and the Manufacturing Activities in Paraná State, focusing on the five most employment-intensive activities in each category. The discussion is anchored in the broader debate on the industrial decentralization process in Brazil and the extent to which Knowledge-Intensive Activities (KIA) can serve as an alternative for economic diversification. Methodologically, the research relies on data from the Annual Social Information Report (RAIS/MTE - 2023), with employment figures analyzed for all 399 municipalities. The selection of the top five KIA was based on Eurostat’s (2025) definition, considering the highest employment volumes, and these were compared with the five most representative industrial activities, using the same criteria. The analysis incorporated data normalization per 100 inhabitants and evaluated concentration levels using the Hirschman-Herfindahl (HH) Index and the CR10 Index. To avoid biases related to universal public-sector employment, "Public administration, defense, and social security" and "Education" Activities were excluded from the KIA dataset. The exclusion of education is particularly relevant, as state universities register employment data in their headquarters’ municipalities, which could distort the real geographical distribution of higher-education employment. Findings reveal that while the Manufacturing Activities accounts for 11.26% of total formal employment, KIA contribute 4.24%. However, KIA exhibit greater spatial dispersion, with only two municipalities without employment in these activities, as opposed to 19 municipalities without industrial employment. A key structural asymmetry is the high concentration of KIAs in Curitiba, which alone accounts for 38.68% of employment in the top five KIA activities, positioning the state capital as an outlier in the dataset. By excluding Curitiba, the HH and CR10 measures became significantly adjusted. In addition, employment in the manufacturing sector remains highly concentrated, with only 1 in 5 municipalities having a significant presence. The standard deviation of industrial employment per 100 inhabitants (5.6939) far exceeds that of KIA (0.4398), highlighting the polarized distribution of manufacturing jobs, where certain locations concentrate employment while others are excluded from the manufacturing activities in question. The main contribution of this study lies in the analysis of dispersion and concentration patterns of Knowledge-Intensive Activities (KIA) and the Manufacturing Activities in the state of Paraná. This is a preliminary study, and near-future research will focus on developing more robust spatial indicators that will allow a more accurate assessment of KIA dispersion and its potential role in statewide economic diversification.

Flavio Vieira (p), Laura Silveira Paiva

Minjeong Kwak

The goal of this study is to investigate the role of economic complexity for exchange rate volatility and misalignment for the BRICS. We estimate three autoregressive distributed lag (ARDL) models of exchange rate volatility, exchange rate misalignment and misalignment based on Hodrick-Prescott filter for each country using data from 1995 to 2021. The results indicates that economic complexity has a long run role for the exchange rate misalignment models for Brazil, China, Russia and South Africa. Other variables such as the interest rate is significant for Brazil, China and India. Inflation is significant for all five countries. For the volatility models, economic complexity, inflation and interest rate are significant for China, and inflation for Russia. Regarding the error correction mechanism, for the volatility models the average is -0.564, for the misalignment models is -0.665 and for the misalignment HP models is -0.788, suggesting that in average the correction towards the equilibrium is slightly faster for the exchange rate volatility models compared to the exchange rate misalignment models.