The prices of residential land in German counties

Highlights

• We estimate German land prices.at the county-level from 2014 to 2018.

• We build a database for German residential land values at the county-level.

• We show that the average land values increased by 28%.

• We show that the average structure and home values increased by 13% and 20%.

• West German states experienced a higher increase in land shares than the East.

Abstract

We estimate changes in the value and price of residential land for 379 German counties (“Landkreise”) from 2014 to 2018 using a total of 42,647 observations. In doing so, we build a database for the cost of housing structures and residential land values at the county-level. We use the two-step residual method that decomposes the value of a home into its structure cost and land value. Despite the short time series, we show that the price of residential land has increased significantly more than the structure costs and house prices in most of the German counties. Specifically, the average land values increased by 27%, while the average structure and home values increased only by 14% and 20%, respectively. Our findings thus imply that cycles in the German land values are more likely to affect the evolution of house prices in the future. Furthermore, we show that our estimated land prices vary significantly to the current land price valuation in two of the German federal states (Hessen and Thüringen).

Introduction

Given the recent macroeconomic experience of most developed countries, few students of the economy would argue that the housing sector is a critical component for financial stability, aggregate economic behavior, and household net wealth. One of the key lessons from the Great Financial Crisis of 2008 is that sustained imbalances in housing valuation can jeopardize the financial sector's soundness due to banks' central role as mortgage lenders and the use of real estate as collateral.¹ Moreover, on a micro-level, European Central Bank (2018) reports that post-2008 EU area house price run-ups suggest worsening of housing affordability for new buyers. Consequently, understanding the housing price variation is paramount in assessing the housing and financial sectors' risks. Recent studies on land prices such as Ahlfeldt and McMillen (2020), Albouy et al. (2018) and Davis et al. (2021), with different objectives in mind, have shown for the U.S. that most of the variations in house prices are due to the underlying land values and shares. Knoll et al. (2017) also document that rising land values and shares mainly drive house price increases for many developed economies since World War II.² We also argue, in this paper, that the movements in the land - and structure values are crucial for understanding the development of German housing markets and policy measures that accompany these movements.

Bonnet et al. (2021) show that the housing component of national private wealth explains the spectacular rise of wealth relative to national income in several countries: the historical increase in housing wealth is in significant part due to the increase in land share. The OECD data on Germany also confirms the observation by Bonnet et al. (2021). The recent development in the housing wealth (including both structures and land) of households and housing prices for Germany from 1999 to2017³ show three salient features. First, housing accounts for 62% of national wealth in Germany at the end of 2017. Second, housing prices have been sharply increasing in the last ten years. Lastly, the fraction of land in total wealth has been increasing over the last six years, but the fraction of dwellings’ value in total wealth has been steadily decreasing. These stylized facts suggest that housing value changes may significantly affect the German macroeconomy and could be due to the movements in the land - and structure values.

Consequently, this paper estimates changes in the value and price of residential land for the 379 German counties (“Landkreise”)⁴ from 2014 to 2018 by combining data from several sources that are publicly available for academic researchers. The absence of reliable data on actual residential land transaction prices in the past makes it necessary to estimate land prices using other data sources.⁵ In doing so, we build a database for the cost of housing structures and residential land values at the county-level. The framework of our approach is that of Davis and Heathcote (2007), who decompose the value of a home into the value of the structure and land value on the aggregate level for the U.S., and Davis and Palumbo (2008), who estimate the land prices for 46 large U.S. metropolitan areas. To account for capital depreciation, previous works on German land prices such as Kolbe et al. (2012, 2015, 2019) use land value assessment methods employing transaction data from local surveyor commissions. Kolbe et al. (2012, 2015, 2019), who focus only on the city of Berlin, generate highly correlated land value estimates to expert-based assessments and show that, using semi - and non-parametric methods, land price estimation can be harmonized across different German federal states. Their methods, however, require high computational power and some financial costs suggesting that the methods are mostly applicable on a very local level. We, however, estimate German land prices on county levels using housing data that is publicly available for research purposes.⁶

There are three main reasons to estimate German residential land prices at the county-level using Davis and Palumbo (2008) framework to build up the dataset. First, there are no publicly available county-level transaction residential land prices in all the federal German states. The Federal Statistics Office provides only the publicly available vacant land prices on a national level from 2010. Second, seven out of sixteen federal states provide land prices at the county-level but at costs.⁷ Lastly, unlike many other countries that use statistical standard mass land appraisal systems for their use in property taxation and real estate development projects,⁸ Germany still uses independent appraisal methods for each of the Federal states. Despite being based on detailed guidelines, these existing land price valuations rely heavily on surveyors' knowledge and expertise.⁹ And given Germany's federal state structure, there is no assurance that each state's valuation meets the national standard quality. Thus, our motivation to estimate the German land prices on the county level is to provide a systematic method and to build-up a harmonized land price valuation database that can address and help some of the German housing markets' measurement issues. For example, besides surveyors' land valuation, the existing large-scale land values measurements in Germany only refer to undeveloped land.¹⁰ Hence, these measurements are not comparable for all single-/double family housing units throughout Germany. Subsequently, our land price estimates could further provide an alternate view for real estate market participants in evaluating their projects' economic feasibility and help policymakers address some of the housing taxing laws. In particular, Bonnet et al. (2021) using a Ramsey taxation model show that the first best optimal taxation is achieved with a property tax on land without resorting to tax on capital. Consequently, having systematic land price estimates could help to improve the current German property tax reform that requires a frequent, systematic, and cost-effective land price valuation methodology.¹¹

One of the main results from both Davis and Heathcote (2007) and Davis and Palumbo (2008) is that the prices of both components of the housing bundle evolve quite differently. Consequently, they shed light on the importance of distinguishing between the values of construction costs and residential land when analyzing home prices. Without the decomposition of housing prices, Davis and Palumbo (2008) argue that regression results from housing market-related variables on the fundamentals in determining housing prices can lead to misguiding conclusions. Furthermore, both Davis and Heathcote (2007) and Davis and Palumbo (2008) show that land prices appear to be three times more volatile at business cycle frequencies than the value of the structural component. Moreover, the average share of a home's value attributed to residential land increased significantly between 1984 and 2004. While the inflation-adjusted land price increased by more than 400%, the price of housing structures appreciated by only 33%. Consequently, to estimate and identify land prices in a systematic method is crucial in understanding housing prices.

As with Davis and Palumbo (2008), we also show that despite the short time series, the price of residential land has become relatively more expensive in the majority of German counties. The counties around urban centers such as Munich, Stuttgart, Berlin, Hamburg, Dresden, and the Ruhr area cities experienced the highest land price increases. However, in general, an upward shift in home values, land values, and residential land share occurred in almost every state. The most significant differences in the changes in land prices are between the new and old federal states of Germany. Although the differences in the average land's share of home value increased between East (“new”) and West (“old”), the absolute differences in home values, the value of residential land, and replacement costs did not decrease. Specifically, the cumulative change in land values from 2014 to 2018 varies between 4% and 56%. On average, the home values increased by 23% and 13% for West and East German counties, respectively. In contrast, the land values increased by 31% and 17% for West and East Germany and structure values increased by only 15% and 11% for West and East. Our findings imply that cycles in the German land values are likely to influence house prices more in the future than they did in the past. Lastly, we also show that our land prices vary significantly to the expert-based land price valuation in two German state governments (Hessen and Thüringen). Our results are also robust to various sensitivity analyses. In particular, our robustness checks on the land share estimates using different cost functions, the structure intensities of new housing, and housing stock growth measures all show almost no variations in the land shares. However, the effect of depreciation rates for structure on land shares is significant as we account for capital depreciation for various structure ages in our sample: an increase in the depreciation rate from 2% to 4% results in a rise of 45.4%–62.3% in the land shares.

The next section briefly overviews the literature. We then outline the methodology in Section 3 as well as the detailed description of Davis and Heathcote (2007) and Davis and Palumbo (2008) version for Germany. Our empirical results and discussions are in Section 4. Section 5 concludes, followed by the Appendix.