Trams, Income, and Urban Structure: Munich before World War I
John C. Brown
Clark University

A key development in the structure of the cities of industrializing Europe and the United States after the mid-19th century was the transformation of the "walking city", which was typically only a few miles in diameter, into far-flung urban agglomerations such as New York or Berlin. Many historians argue that the suburbanization that accelerated after World War II reflected primarily a continuation of this development. The most outstanding feature of the transformed cities in the United States and Europe was the emergence of spatial separation by income group. In the United States and Great Britain, this segregation resulted as high and middle income suburban areas came to surround a core of lower-income neighborhoods. Elsewhere, the center was more likely to be held by upper income residents with lower income residents living at the periphery in neighborhoods that later became centers of working class agitation(Hohenberg and Lees[1985: 293-307] and Wischermann[1987]). This paper examines two alternative explanations for the spatial structure of Munich in the early 1900s. The findings suggest that explanations for the changes in structure must focus on changes in urban labor markets as well transport innovation.

Urban economists and many historians argue that the pattern of urban structure can be best explained by the standard urban model that posits a central business district containing all of the employment in the city(Mieszkowski and Mills [1993: 136-141]). If marginal costs of commuting reflect primarily the opportunity cost of time and speed of available transportation and housing is a normal good, higher-income residents will be willing to give up the savings in leisure received by living close in exchange for the increased consumption of (space-intensive) housing available most cheaply at the periphery.

Leroy and Sonstelie[1983] formalize the arguments found in studies of "street car suburbs" and the 19th-century journey to work (Hershberg, et. al.[1981]) within the context of this model. They introduce transport innovation into the model and focus on the relationship between modal choice and residential location. The horse-drawn streetcar, for example, implied a higher fixed cost per day than walking in exchange for speeds twice as high. Beyond a particular distance, the wealthy (with the higher valuation of time) would choose the faster over the slower mode since the fixed cost would be covered. Since this break-even distance would lie closer to the center for the wealthy than for the poor, higher income workers would outbid lower income workers in more distant locations, and lower income workers would outbid higher income workers at the center. Gin and Sonstelie[1992] offers the most sophisticated test of this hypothesis data from 1880 in Philadelphia, when the price of using the street car system is said to have restricted its use to middle- and upper-income residents. They find that higher income groups would generally have outbid low-income groups for access to the center if both had access only to walking, but lower income groups would outbid upper income groups if the wealthy could afford the streetcar and the poor could not. Labor Markets and Urban Structure

This paper argues that heavy reliance on the standard urban model to explain the transformation of urban structure leads to an under-recognition of two important developments in labor markets in late 19th-century cities: the relocation of and dispersion of employment and the life-cycle pattern of employment relationships. After 1880, periodic occupational and business censuses revealed that two processes were shaping urban structure in European cities(Wischermann [1987] and Hohenberg and Lees[1985: 206-209]). Reductions in transportation costs as well as localized agglomeration economies increased the attractiveness of city-center locations in larger cities relative to the periphery and smaller towns. The process of "city-building" resulted in the simultaneous depopulation of city centers and the expansion of firms with a decided advantage in the center, including finance, publishing, wholesaling, and retail sales. At the same time, improvements in inter-urban transportation as well as technological change increased the average size of manufacturing firms and their demand for land. Unable to compete with the firms of the "City", these firms moved towards the periphery. Data from Munich bear out this characterization of these developments. The center city's share of employment fell from thirty to twenty percent between 1882 and 1907, even as the total number employed in the city quadrupled. The chief sources of center-city employment by 1907 were in clothing and apparel, wholesale and retail sales, banking, insurance, publishing, and hotels and restaurants. The city's major industrial employers in machine-making, metalworking, and food processing had for the most part left the city center.

Observers point to another important feature of pre-war labor markets that exerted an important influence on the housing market: the length of job tenure. The short job tenures of younger workers recently arrived in cities tended to compel them to seek out employment with a wide variety of firms in the city. A center-city location--typically a room or bed in another household--minimized the average commuting distance for these workers as well as the search time for employment. With increasing age, job tenures tended to lengthen and the relative attractiveness of a central location to the worker declined. Testing for the Determinants of Urban Residential Structure

The model used by this study to explain what influenced urban structure in the pre- automobile era retains the notion common to the standard urban model that any one location in the city will be taken by the highest bidder. It departs from the standard urban model by incorporating a richer specification of the household head's employment status and age. The analysis draws upon a unique data set from the 1904-1907 Munich housing census: a sample of 1600 apartments with detailed information on rent, housing characteristics, and location; part of the data set could be matched with detailed data from other sources on household composition, origins, and in many cases, income. Archival sources provided additional information on tram service. Munich is well- suited for such a study, since its employment pattern was relatively concentrated and the cost of travel on the tram Š the alternative to walking Š was relatively high. Munich had an unusually high share of employment in trade, hotels, and restaurants; all these are industries for which the city center was increasingly attractive by 1910. In addition, while tram use increased steadily up to World War I, it still remained a relatively expensive means of transportation. Roundtrip fares in Munich equaled 5-7.5% of the daily wage paid unskilled workers and 3-5% of the wage for skilled workers in 1905, shares that are not too much less the 6-9% reported for Philadelphia as far back as 1880.

The test of alternative hypotheses uses the two-step hedonic approach to estimate the influences on marginal bids of households for locations in the city. This approach offers a key advantage: it views housing as a bundle of characteristics(z). In this application, the focus is on two locational characteristics of the apartment--its distance from the center of the city and its distance from the nearest tram stop--as well as its size. In equilibrium, if P is the market rent paid, z* is the utility-maximizing choice of characteristics, and the household's bid is B, then P = B(z*) and the slope of P, pi, is also the slope of the household's bid function with respect to zi. The higher the implicit market price for the characteristic, pi, the higher the household's marginal willingness to pay for zi. The first step of hedonic analysis identifies the values of pi for each household by regressing rent on housing characteristics. The second step of the analysis identifies the influence of household characteristics on pi. This study tests four hypotheses suggested by the transport cost and labor market approaches to explaining household locational decisions. The transport cost approach suggests that so long as tram usage was affordable for upper income residents, higher income households would be willing to pay less at the margin for access to the center of the city than lower-income households. In addition, higher income households would place a higher marginal valuation on access to the tram than would lower income households. The labor market view of demand for location suggests that the shorter the typical job tenure, the higher the household head's marginal valuation of access to the center. In addition, the older the household head, the more likely he is in secure employment and the lower the marginal willingness to pay for access.

Preliminary estimation of the relationship f(Rent)=g(z) used Box-Cox transformations of the rent and twenty characteristics, which had been corrected for heteroskedasticity. The estimation results suggest that access to the tram system and the city center played highly significant roles in influencing rents. Rents declined at a rate of about 20 percent per mile from the city center, well above rates of discount reported for third-world cities today. Apartments located one-half mile from a tram stop rented at a two to three percent discount relative to those within a short distance.

Estimation of the marginal willingness to pay functions for access to the tram line, access to the center, and space of the apartment offered an opportunity to test for the importance of income (non-housing expenditures) and characteristics of the tenant. The transport cost view of urban structure argues that income would be an important determinant of the marginal willingness to pay functions for access to the center and access to tram service. The higher the income, the lower the marginal willingness to pay for access to the center and the higher the marginal willingness to pay for access to tram services. A focus on centralized access to employment would suggest that those employed in lower-skilled jobs with shorter tenures would tend to have a higher marginal willingness to pay for access to both the center and tram service. Data from the Munich local insurance fund suggest that the average completed job tenure of day laborers was shortest, about one-half year, while the tenure of semi-skilled shop help was 1.3 years and the tenure of messengers was 1.6 years. All other workers averaged 2.5 years. Dummy variables control for each of these groups. The specification is rounded out with the age of the household head and a dummy for home workers, who tended to rely heavily on centrally-located merchants for raw materials and marketing. Estimation of the marginal willingness to pay functions for access and space employed a systems approach to increase efficiency. As Epple[1987] notes, estimation of this kind of system must recognize that the quantities purchased are endogenous, which requires the use of instrumental variables. The coefficients reported in Table 1 resulted from three stage least squares estimation. The dependent variable in each case is the natural log of the marginal willingness to pay.

The most notable result in Table 1 is the weak negative influence of non-housing expenditures on bids for access and the strong influence of other household characteristics. The dummies for household heads who were employed as day laborers or as home workers significantly raised on bids for access, particularly for the tram. The age of the household head exerted a significant negative influence on demands for access. The marginal bid of a forty year- old worker would be ten percent lower than the bid of a twenty year-old worker. Of particular interest as well, space and access tended to be strong complements, a result in keeping with the relative attractiveness of the Munich Center. Higher non-housing expenditures increased the marginal willingness to pay for space as did servants. The complementarity of space and access provided another route by which income could influence marginal bids for access; Table 2 offers estimates of conditional marginal bids that incorporate this influence. Household demands for space rose steadily with income, while the marginal willingness to pay for access followed a U- shape. Youth, uncertain employment, and wealth all raised the marginal valuation of access to the tram system. Homeworkers, the young and the very rich showed stronger marginal bids for access to the center.

Overall, the results suggest that a richer characterization of what influenced the attractiveness of center-city locations is needed before we can fully understand the forces that resulted in the transformation of urban structure in the late 19h century. They also point to the need to look more closely at the potential importance of employment decentralization and the stabilization of employment relationships in the emergence of the modern urban structure.

Table 1
Estimated Marginal Willingness to Pay for Access and Space
Independent Variable	Mean	Tram 	Center	Space 
Access to Tram 
(Km. to Stop)	.383	-2.47 
(11.5)	-1.21 
(6.12)	-.258 
Access to Center
(Km. to Marienplatz)	2.31	-.178 
(5.12)	-.376 
(11.8)	-.059 
(Cubic meters)	164	.004
(5.11)	 .004 
(5.56)	-.009 
Day Laborer	.07	.264 
(1.98)	 .152 
(1.24)	-.027 
Shop helper	.16	-.056 
(0.66)	-.086 
(1.09)	-.069 
Messenger	.15	 .039 
(0.43)	-.027 
(0.32)	-.072 
Home worker	.04	.665
(3.51)	.577
(3.30)	 .017 
Non-Housing Expenditures 
(in 1,000 Marks)	1.533	-.029 
(0.66)	-.025 
(0.60)	.024
Age	42.6	-.004 
(1.63)	-.005 
(2.36)	-.004 
Age2				 .0003  
Number in household over
14	2.45	 .007  (0.40)	-.001 
(0.06)	-.007 
Room Renters or Sublettors	.30			 .017 
Servants	.13			.088
Constant		-4.05 
(22.9)	 2.00
(12.33)	-1.22 
F 		86.0	74.1	28.4
Adjusted R2		.81	.79	.62 

Source: Results of three stage least squares estimation. The dependent variable is the natural log of the respective marginal price or demand. Asymptotic t-statistics are in parentheses. N=218.

Table 2
Conditional Marginal Willingness to Pay for Access
Occupation and income
group	Marginal Willingess to Pay for		Demand for
	Tram	Center	
Day-Laborer (Y=800)	8.22	2.86	 76
Home-Worker (Y=800)	13.66 	5.06	100
Y=2000	7.94	3.10	135
Y=4000	9.36	3.68	184

Source:Results of estimation in Table 1. The marginal bids are calculated conditional upon the household living one-third mile from a tram stop and two miles from the center. The demand for space is calculated from separate regressions of the inverted marginal willingness to pay.

Selected References

Epple, Dennis[1987]. "Hedonic Prices and Implicit Markets: Estimating Demand and Supply Functions for Differentiated Products." Journal of Political Economy 95: 51-80.

Gin, Alan and Jon Sonstelie[1992]. "The Streetcar and Residential Location in Nineteenth Century Philadelphia." Journal of Urban Economics 32: 92-107.

Hershberg, Theodore, et. al.[1981]."The 'Journey-to-Work': An Empirical Investigation of Work, Residence, and Transportation, Philadelphia, 1850 and 1880." in Theodore Hersberg, ed. Philadelphia: Work, Space, and Group Experience in the Nineteenth Century. New York: Oxford University Press.

Hohenberg, Paul and Lynn Hollen Lees[1985]. The Making of Urban Europe. Cambridge, MA: Harvard University Press.

LeRoy, Stephen F. and Jon Sonstelie[1983]. "Paradise Lost and Regained: Transportation Innovation, Income, and Residential Location." Journal of Urban Economics 13: 67-89.

Mieszkowski, Peter and Edwin S. Mills[1993]. "The Causes of Metropolitan Suburbanization." Journal of Economic Perspectives 7: 135-148.

Wischermann, Clemens[1983]. Wohnen im Hamburg vor dem Ersten Weltkrieg. Munster: Coppenrath.

Wischermann, Clemens[1987]. "Wohnung und Wohnquartier: Zur innerstŠdtischen Differenzierung der Wohnbedingungen in deutschen Gro§stŠdten des spŠten 19. Jahrhunderts." in Heinz Heineberg, ed. InnerstŠdtische Differenzierung und Prozesse im 19. und 20. Jahrhundert. Cologne: Bšhlau Verlag.