Wolfgang Brunauer , Ronald Weberndorfer , Karin Wagner (p), Fabrice Kreuzbichler (p)

Jean Bonnet

Reliable residential property price indices at micro-geographic scales are increasingly important inputs into housing policy and economic research. The Austrian Central Bank (OeNB) has been publishing the Austrian residential property price index (RPPI) for many years now. In principle, the so-called double imputation methodology, which this index is based on, and which was developed by DataScience Service (DSS) with the support of OeNB following Eurostat's specifications, allows the index to be disaggregated to almost any depth because the data on which the calculation is based is available at the individual transaction level. The reason for the currently only very rough evaluation lies in the very different density distribution of the data and the change in these densities over time: While in some regions, especially urban areas, there is a relatively high transaction density, there are districts in which only a few transactions are available for a certain object category even more as quarterly indices have to be published; in some cases, markets for certain transaction types have "dried up" for various reasons. A highly regionalized index differentiated according to object features must therefore manage the “bias-variance tradeoff”. In other words: How much can an index be disaggregated (reduction of bias), and how can the results be regularized in a data-driven manner (reduction of variance)? The aim of this paper is to use information from neighboring units on the one hand and, on the other hand, historical information from the same and/or from neighboring units, in order to derive time series that are as robust as possible. Thus, we derive a spatially granular panel of indices for prices and rents, which can serve as input for early warning systems. Furthermore, we investigate the possible impact on the construction of regional early warning systems for the Austrian housing markets, which can in turn be used for systemic risk analyses.

Joao-pedro Ferreira (p), William Shobe, Elizabeth Marshall, Terry Rephann, Andre Avelino, Alberto Franco-Solis

Fabrice Kreuzbichler

In the case of large-scale solar facilities (LSS), some practices associated with economic impact assessment have become methodologically outdated. This is particularly true when researchers or consultants attempt to estimate local impacts at the county or city level using local input-output multipliers in three key ways. First, as the regional scale decreases and the underlying input-output tables cover smaller geographical regions, the models produced are more sensitive to the use of national data and rely on strong assumptions in estimating the local consumption of inputs, the local provision of labor and commuting flows, the leakage of capital to other regions, or the local fiscal policy and its complementary effects on local government expenditures. Secondly, most studies have simply neglected the impact of trade-offs with other land-intensive activities (like agriculture or forestry) that support the economic base of many rural economies. Finally, since solar development is still a relatively new technology in a phase of rapid expansion, national production technology data might not accurately reflect the impacts of the operation phase and instead overestimate the need for labor and other inputs primarily used in the construction phase. Unfortunately, many of the aspects above are absent from studies that use standardized applications of IMPLAN, REMI, or other top-down regional input-output models.

Given this, we start by summarizing the recent literature that addresses the local economic impact of solar development and highlight how they have been overestimating the impacts both at the state and local levels, with the impacts of solar sometimes representing almost 10% of the total employment of some US states or local counties. Next, we highlight the potential methodological aspects that lead to these results and suggest different ways to improve the estimations and the data sources that can be used for this purpose. For example, in the case of IMPLAN, one particular aspect is that the coefficient of employment per unit of output is significantly outdated and ignores the differences between the construction and operation phases. Finally, we will list several areas where best practices should be implemented to increase the accuracy of local impact assessments.

Toshimori Otazawa (p), Koya Morotome

Joao-Pedro Ferreira

This study aims to examine the heterogeneity in the causal impact of highway development on regional economic growth. To this end, we employ causal forests, that is a machine learning algorithm for causal inference, to data on Japanese municipalities from 1971 to 2011 and estimate the conditional average treatment effect (CATE) of highway interchange openings on the growth of value-added per employee in manufacturing sector. We then find evidence that an opening of highway interchange improves regional labor productivity by an average of 9.9% from 1971 to 1991, while there is no significant average effect from 1991 to 2011. This result is consistent with the fact that Japan experienced rapid growth in the former period and long stagnation in the latter period. We also identify regional characteristics that affect the heterogeneity in the causal effects for each period, and reveal that those differ between periods of economic growth and stagnation. This regional and temporal heterogeneity is likely to improve the efficiency of transportation infrastructure policies by targeting regions where significant benefits can be expected. Finally, by comparing the result of causal forests with those of regression models and propensity score matching DID, we demonstrated the usefulness of the non-parametric method.

Jean Bonnet (p), Bruno Drouot

Toshimori Otazawa

One of the main problems of sustainable development is to reduce the consumption of natural lands. The reform of “Zero Net Artificialisation” (ZNA) in France, provides a framework for a downward trend in the observed consumption of Natural, Agricultural and Forestry Land, but which needs to be amplified. The aim of ZNA is to compensate for any new artificialisation by measures to restore, renature or enhance other areas.
As far as business parks are concerned, we want to define several sub-indexes to promote the densification of areas and more sustainable businesses, with buildings that respect environmental criteria, better governance and equality between employees.
- The Land Sobriety dimension takes into account the need to reduce land use. For example, we are encouraging companies to make an effort to build taller buildings. Yet, two other dimensions must be taken into account,
- The Energy Sobriety and the Circular dimension, that refer to all the incentives and constraints that exist regarding new buildings that must be environmentally performant.
- The Employ and Societal dimension that refer to all dimensions that new installations must comply, regarding density of employs but also diversity, minimum wage, access to education etc.
The construction of aggregate composite indexes for each of the three selected dimensions of sustainable development in the ZNA can be described in several stages.
- 1 Choice of variables
- 2 Normalisation method (Min-Max, Desirability)
- 3 Aggregation method (arithmetic)
- 4. Choice of weighting
The calibration is important. The index should be see as a mean towards sustainability. There is also the question of the social acceptability of entrepreneurs.
1. Multi-factor, the index provides flexibility (in particular, some companies will find it easier to meet one criterion than another, and some form of compensation can be accepted).
2. Adaptable, while maintaining the general principle, certain variables may be weighted differently or even may be different, depending on the branch of activity.
3. Self-improvement: by taking into account the total size of the plot and not just the floor area ratio, existing businesses can be encouraged to release surplus land - land reserve - (which can then be used to densify the zone).
Of course, all these new constraints are costly, but we will try to offer the best way for a company to gain in sustainability according to the specifics of its activities.