Historic Rental Prices in St. Petersburg ( ): Revolutions, World War I and Rent Control

Transcription

1 Historic Rental Prices in St. Petersburg ( ): Revolutions, World War I and Rent Control Konstantin A. Kholodilin 1 and Leonid E. Limonov 2 and Sofie R. Waltl 3 1 DIW Berlin - German Institute for Economic Research 2 Leontief Centre St. Petersburg 3 LISER - Luxembourg Institute of Socio-Economic Research LISER May 3, 2018

2 Overview 1 Research on Historic Housing Markets 2 Historical Context and Rent Control 3 Data 4 Hedonic indices 5 Spatial modelling 6 Empirical results 7 A Consumer Price Index for St. Petersburg 8 Summary and Way Forward

3 Overview 1 Research on Historic Housing Markets 2 Historical Context and Rent Control 3 Data 4 Hedonic indices 5 Spatial modelling 6 Empirical results 7 A Consumer Price Index for St. Petersburg 8 Summary and Way Forward

4 Why bother about history as an economist?

5 State of the literature Eichholtz (1997): A long run house price index: The Herengracht index, Eichholtz, Straetmans and Theebe (2012): The Amsterdam rent index: The housing market and the economy, Kholodilin (2016): War, housing rents, and free market: Berlin s rental housing during World War I Lyons (draft): Measuring House Prices in the Long Run: Insights from Dublin, Eichholtz, Korevaar and Lindenthal (draft): 500 Years of Urban Rents, Housing Quality and Affordability in Europe Knoll, Schularick and Steger (2017): No price like home: Global house prices

6 Overview 1 Research on Historic Housing Markets 2 Historical Context and Rent Control 3 Data 4 Hedonic indices 5 Spatial modelling 6 Empirical results 7 A Consumer Price Index for St. Petersburg 8 Summary and Way Forward

7 Historic context Dates (Gregorian calendar) Russian-Japanese War 1904/02/ /09/05 First World War 1914/07/ /11/11 Rent Control in St. Petersburg 1915/12/09 General Rent Control in Russia 1916/09/09 February Revolution 1917/03/08 October Revolution 1917/11/07

8 Rent Control Policies General Situation excess demand for housing due to war-related massive population movements loss of territories in the first half of 1915 led to huge migration to St. Petersburg (refugees and evacuation of civilians/government bodies/educational institutions)

9 Rent Control in St. Petersburg On December 9 (November 26), 1915 regional authorities adopted a compulsory ordinance prohibiting rent increases. Rents for apartments, rooms, corners and beds were fixed at the January 14 (1), 1915 level. Rents were allowed to be increased to offset rising costs for fuel or house personnel or in case of large refurbishments. Rental contracts were automatically prolonged when tenants regularly paid the rent.

10 Rent control in the Russian Empire In fall 1916, regulation of the housing market became a national responsibility. By September 9 (August 27), 1916 rents were frozen at the January 14 (1), 1915 level. More restrictions for contracts concluded before July Extended rights for tenants regarding prolongation of rental contracts. Restrictions regarding evictions. Expensive apartments were excluded from rent controls. Misconduct could be punished with a prison sentence! Revisions of the policy in August 1917.

11 Overview 1 Research on Historic Housing Markets 2 Historical Context and Rent Control 3 Data 4 Hedonic indices 5 Spatial modelling 6 Empirical results 7 A Consumer Price Index for St. Petersburg 8 Summary and Way Forward

12 Data collection

13 Data source: Peterburgskaya gazeta / Petrogradskaya gazeta The newspaper appeared 5 times per week until 1882, and then 7 times per week. The newspaper was published until November 1917, which is when our data ends. Digitalisation in Cyrillic (newspapers are available in paper format or on microfilms in the National Library in St. Petersburg) Transcription of data sets to the Latin alphabet / Translation to English Geo-coding of the addresses Observations missing essential information or located far away from the historic city limits were excluded

14 Description of the data set Rental price Date of advertisement Address / Longitudes and Latitudes Type of unit: apartment or room Number of rooms Dummies: bathroom, balcony, heating, electricity, furnished Total: 5,549 ads Roughly between 30 and 450 ads per year

15 Location of rental units. Street and bridge net as of 1902.

16 Overview 1 Research on Historic Housing Markets 2 Historical Context and Rent Control 3 Data 4 Hedonic indices 5 Spatial modelling 6 Empirical results 7 A Consumer Price Index for St. Petersburg 8 Summary and Way Forward

17 House price indices/rental price indices Problem: Heterogeneous good. No two dwellings are the same. Prices/rents no directly comparable. Thus: quality adjustment.

18 House price indices/rental price indices Problem: Heterogeneous good. No two dwellings are the same. Prices/rents no directly comparable. Thus: quality adjustment. 3 major approaches (Hill, 2013) Stratification Repeat-sales Hedonics - Time-dummy methods - Imputation methods

19 Time-dummy indices Hedonic equation: log p = δd + βx + ε whereas D is matrix of time-dummy variables, and X a matrix of locational and structural dwelling characteristics, i.e., characteristics are included as control variables to filter out pure time effects.

20 Time-dummy indices Hedonic equation: log p = δd + βx + ε whereas D is matrix of time-dummy variables, and X a matrix of locational and structural dwelling characteristics, i.e., characteristics are included as control variables to filter out pure time effects. Index: P t = exp(ˆδ t ) exp(ˆδ 0 )

21 The point for a continuous-time hedonic model Few observation (between 50 and 350 per year) trade-off between long period length and precision of the index dummy variables impose prior structure ( let the data speak not the statistician speak ) no averaging effect:

22 The point for a continuous-time hedonic model Figure: Waltl (2016a).

23 The point for a continuous-time hedonic model Few observation (between 50 and 350 per year) trade-off between long period length and precision of the index dummy variables impose prior structure ( let the data speak not the statistician speak ) no averaging effect: cleaner than ex-post smoothing

24 Hedonic equation: log p = α + f (time) + βx + ε The result is an Additive Model that can be estimated via penalization techniques. The function f ( ) is obtained from adaptive smoothing (speed of change can vary over the domain). The smoothing parameter (controlling the trade-off between model fit and smoothness) is estimated by minimizing a Generalized Cross Validation criterion.

25 Hedonic equation: log p = α + f (time) + βx + ε The result is an Additive Model that can be estimated via penalization techniques. The function f ( ) is obtained from adaptive smoothing (speed of change can vary over the domain). The smoothing parameter (controlling the trade-off between model fit and smoothness) is estimated by minimizing a Generalized Cross Validation criterion. Index: P t = exp(ˆf (t)) exp(ˆf (t 0 )) Can be evaluated at daily/weekly/monthly/yearly/etc. frequency. All index numbers are consistent!

26 Overview 1 Research on Historic Housing Markets 2 Historical Context and Rent Control 3 Data 4 Hedonic indices 5 Spatial modelling 6 Empirical results 7 A Consumer Price Index for St. Petersburg 8 Summary and Way Forward

27 Spatial modelling: Location, location, location No location dummies but smooth function in space: g(longitude, latitude) (Hill and Scholz, 2017; Waltl, 2016a,b) Alternatives: location dummies, spatial-lag models (same argument as before: much prior structure imposed) Spatial function is updated at regular points in time to model changes in the relative price of location. Special smoothing technique to capture St. Petersburg s special topography: soap-film smoothing (Wood et al., 2008)

28 c David Darling

29 Overview 1 Research on Historic Housing Markets 2 Historical Context and Rent Control 3 Data 4 Hedonic indices 5 Spatial modelling 6 Empirical results 7 A Consumer Price Index for St. Petersburg 8 Summary and Way Forward

30 Empirical results log r = α + β rooms rooms + β chamber (type = Chamber) + β bath bathroom + β fur furnished + β heat heating + β elec electricity + β balc balcony + f (time) + g bi (long, lat) + ε

31 Alternative indices

32 Major events

33 The price of location

34 Notes: (a) to (c): monthly imputed rents for a furnished apartment with two rooms, a bathroom, no balcony, heating and electricity. (a) 1880, (b) Dec (Nov) 1915 (rent control policy), (c) Oct (d) to (f): changes in average monthly rents per chast. (d) change over the entire period, (e) rent control policy, (e) rent control policy Oct 1917.

35 Volume index Notes: Left: Number of rental advertisements per quarter incl. seasonally-adjusted series (Loess decomposition). Vertical lines (from left to right): Russian-Japanese war, start of World War I, introduction of the rent control policy in St. Petersburg, February Revolution and October Revolution. Right: annual numbers of inhabitants.

36 Overview 1 Research on Historic Housing Markets 2 Historical Context and Rent Control 3 Data 4 Hedonic indices 5 Spatial modelling 6 Empirical results 7 A Consumer Price Index for St. Petersburg 8 Summary and Way Forward

37 A Consumer Price Index for St. Petersburg To construct a CPI, one needs: - a representative basket of goods and services - a price index for each item in the basket - expenditure shares/weights for each item in the basket

38 A Consumer Price Index for St. Petersburg To construct a CPI, one needs: - a representative basket of goods and services - a price index for each item in the basket - expenditure shares/weights for each item in the basket Mironov (2010) collected data on historic prices for St. Petersburg Price lists published by the St. Petersburg City Council (Russian National Library): prices of 200 product (from 1853 onwards)

39 Prokopovich (1909) estimates an average household budget for 1907/08 based on questionnaires sent to workers on behalf of the Russian Technical Society. 570 questionnaires were returned (including 307 married and 263 single workers). Mironov (2010) constructs a CPI including the categories: Food, Housing, and Clothing We extend Mironov s CPI (10-years frequency, until 1913) and another CPI series by Pervushin (1925) (annual, 1885 October Revolution) by including our the rental price index. We keep expenditure weights constant over time (Laspeyres Index).

40

41

42

43 Overview 1 Research on Historic Housing Markets 2 Historical Context and Rent Control 3 Data 4 Hedonic indices 5 Spatial modelling 6 Empirical results 7 A Consumer Price Index for St. Petersburg 8 Summary and Way Forward

44 Summary and Way Forward I New data set on historic rents in St. Petersburg Increase information about the pre-soviet Russian economy Rent Price index and CPI Improve index construction technique suitable for the particular topography in St. Petersburg and for historic research (low number of transactions) Analyse effect of rent control policies Effects on prices and quantities

45 Summary and Way Forward II Way forward: - Analyse rent control policy in the light of an urban economic stock-flow model - Bring together volume and price cycle - Better analyse the implications of locational rent dynamics - Make use of additional data (census data, small pieces of information recently found in the Russian National Library)

46 References I Hill, R. J. (2013). Hedonic price indexes for residential housing: A survey, evaluation and taxonomy. Journal of Economic Surveys, 27(3): Hill, R. J. and Scholz, M. (2017). Can geospatial data improve house price indexes? a hedonic imputation approach with splines. Review of Income and Wealth, doi: /roiw Mironov, B. N. (2010). Wages and prices in Imperial Russia, The Russian Review, 69: Prokopovich, S. N. (1909). Biudzhety peterburgskikh rabochikh. St. Petersburg. Waltl, S. R. (2016a). A hedonic house price index in continuous time. International Journal of Housing Markets and Analysis, 9(4):

47 References II Waltl, S. R. (2016b). Variation across price segments and locations: A comprehensive quantile regression analysis of the Sydney housing market. Real Estate Economics, doi: / Wood, S. N., Bravington, M. V., and Hedley, S. L. (2008). Soap film smoothing. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 70(5):

48 Penalized regression Including smooth components yields additive models. Estimation technique: penalization E[Y X = x] = f (x) Solution: ˆf = arg min f (y i f (x i )) 2 i } {{ } model fit λ J(f ). }{{} model smoothness where J(f ) measures wiggliness and λ the smoothing parameter controls for the trade-off between smoothness and model fit.

49 λ : Over-smoothing (leads to a straight line estimate for f ) λ = 0: Under-smoothing (interpolation of observations)

50 Exchange rate / Purchasing power 1 rouble in EUR in 2015 (consumption based; 1 rouble in 1900 could buy 0.76 gram gold. In 2015, one would need to pay EUR for the same quantity. ( Average annual wage in St. Petersburg in 1901 (enterprises with > 50 workers): roubles Mironov (2010) An average monthly salary in St. Petersburg would in 1913 (Mironov, 2010) pay for kg rye flour kg beef - 92 kg fish - 37 kg butter l milk - 70 l vodka - 31 pairs of shoes