<<  Ease of Access and Assistive Technology on Windows 7 ECE G201: Introductory Material  >>
EC 2333: Transportation: Topic #4
EC 2333: Transportation: Topic #4
Outline
Outline
Fogel (1964)
Fogel (1964)
Definition of the Social Savings
Definition of the Social Savings
Fogels Algebra
Fogels Algebra
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Results
Results
Follow-up Literature
Follow-up Literature
Lebergott: SS too small because RR system had large economies of scale
Lebergott: SS too small because RR system had large economies of scale
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Major Criticisms at the Time (2): Perfectly Inelastic Demand and
Major Criticisms at the Time (2): Perfectly Inelastic Demand and
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Major Criticisms at the Time (3): Water Rates
Major Criticisms at the Time (3): Water Rates
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Major Criticisms of the Time (4): Substitution Possibilities
Major Criticisms of the Time (4): Substitution Possibilities
Major Criticisms at the Time (5): RR and Long-Run Growth
Major Criticisms at the Time (5): RR and Long-Run Growth
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Major Criticisms at the Time (6): Williamson GE model
Major Criticisms at the Time (6): Williamson GE model
Lessons: The C19 Transportation Revolution
Lessons: The C19 Transportation Revolution
Railroads Re-visited
Railroads Re-visited
Motivation
Motivation
New Data, Part One: Background
New Data, Part One: Background
New Data, Part Two: Original Version
New Data, Part Two: Original Version
New TR Data Set (2)
New TR Data Set (2)
Digitized Maps
Digitized Maps
Geo-referenced Digitized Map
Geo-referenced Digitized Map
Geo-referenced Digitized Map
Geo-referenced Digitized Map
Specifics of our RR database
Specifics of our RR database
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Measuring Transportation Access
Measuring Transportation Access
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Did RR cause Midwestern urbanization + population growth
Did RR cause Midwestern urbanization + population growth
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Specifics of our Midwestern database
Specifics of our Midwestern database
Specifics of our Midwestern database
Specifics of our Midwestern database
Rivers and Canals
Rivers and Canals
Data
Data
Railroads
Railroads
Research Template So Far: Effects of RR
Research Template So Far: Effects of RR
TR and Urbanization
TR and Urbanization
Tables 1-3
Tables 1-3
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Dif-in-Dif Regression Estimates
Dif-in-Dif Regression Estimates
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
IV estimation
IV estimation
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Table 7
Table 7
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Robustness Check (1)
Robustness Check (1)
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Robustness Checks (2)
Robustness Checks (2)
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Robustness Checks (3)
Robustness Checks (3)
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
Other Results: Impact of RR
Other Results: Impact of RR
Michaels
Michaels
Data
Data
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4

: EC 2333: Transportation: Topic 4. : RAM. : EC 2333: Transportation: Topic 4.ppt. zip-: 7118 .

EC 2333: Transportation: Topic 4

EC 2333: Transportation: Topic 4.ppt
1 EC 2333: Transportation: Topic #4

EC 2333: Transportation: Topic #4

Professor Robert A. Margo Harvard University Spring 2014

2 Outline

Outline

Social Savings of the Railroad: Fogel Impact of RR on local economies: Atack, et. al. Re-assessment of Fogel: Donaldson and Hornbeck (student presentation) If time: Guy Michaels on the Interstate Highway System

3 Fogel (1964)

Fogel (1964)

One of the most famous (perhaps most famous) of all cliometrics studies. Originally a JEH paper, then a dissertation, then a book. Book is about more (much) than the social savings of the RR but, for better or worse, it is the social savings chapter that received the most attention. HUGE follow-up literature, some critical, some applying the same (or modified) method to other countries.

4 Definition of the Social Savings

Definition of the Social Savings

Social savings (percent): savings in transportation costs in a given year by shipping the same amount of goods by the next best alternative/national income As defined, SS is an upper bound to the true resource cost (because the true demand for transportation is NOT perfectly inelastic). Application is to the RR. Next best alternative is some combination of water and wagon transportation. Concept does not originate with Fogel but rather with engineers (and economists) in C19 Europe.

5 Fogels Algebra

Fogels Algebra

Two sectors, Transportation (T) and all other things (A). Households consume A and some portion of T as a final good. T can be produced in two different ways, by R or by W. Overall, R is more efficient than W. Holding T fixed, if R is no longer available, then some resources (L and K) will have to be shifted from the production of A to the production of T. The effect of this re-allocation on national income is, to a first approximation, the additional cost associated with W (i.e. it is the wage x labor reallocated + rental price of capital x capital reallocated). Because T is held fixed, this is an upper bound to the true SS. Also note that resources (L and K) are fully employed. Note that SS can also be calculated as ? x dZ, where dZ = efficiency gain of the RR in percentage terms (percent effect on costs) and ? = transportations share of national income.

6 EC 2333: Transportation: Topic 4
7 Results

Results

Calculation performed in two stages: inter-regional and intra-regional Inter-regional: between major shipping points (eg. New York-Chicago) Four commodities are considered: pork, beef, wheat, and corn, comprise 75 percent of commodities shipped by rail in 1890 Big surprise: initial calculation of inter-regional is negative, turns positive after adjustment is made for lower insurance costs (cargo losses smaller under rail) and speedier service (inventory costs lower) Intra-regional: farmers would have used more wagon transportation. This is the main effect. Total for four commodities is around 117 million, Fogel inflates by ratio of value added = 4.7 percent of GNP in 1890

7

8 Follow-up Literature

Follow-up Literature

Book had a HUGE impact. Narrow: applications of SS concept to transportation improvements in other settings. Probably the most important is Coatsworth on Mexico (SS is around 30 percent). SS concept also applied to other innovations (steam engine, tractor) but not many. Broad: accelerated the shift to quantitative methods in economic history. SUBSTANTIAL critical literature at the time and shortly after: McClelland, David, Lebergott, Williamson, among others. Fogel (1979) is the response.

9 Lebergott: SS too small because RR system had large economies of scale

Lebergott: SS too small because RR system had large economies of scale

and was never even close to capacity. Use minimum of LR AC curve instead Fogels response: transport costs that were technologically feasible were never economically feasible because the system always carried a mixture of goods to many places.

Major Criticisms at the Time (1): Technological versus Economic Definition

10 EC 2333: Transportation: Topic 4
11 Major Criticisms at the Time (2): Perfectly Inelastic Demand and

Major Criticisms at the Time (2): Perfectly Inelastic Demand and

Marginal Social Rate of Return

Fogel assumes price elasticity of demand for RR is zero. Clearly false, but especially for passenger transportation. How large is the bias? Assume log-linear demand. Answer: VERY. Measure marginal social rate of return (Lebergott; Nerlove). Captures impact of the last dollar invested, not the total impact (BUT if you could measure this for every dollar invested, would be useful).

12 EC 2333: Transportation: Topic 4
13 EC 2333: Transportation: Topic 4
14 Major Criticisms at the Time (3): Water Rates

Major Criticisms at the Time (3): Water Rates

Water rates too low (McClelland and others). BUT Fogel used water rates appropriate to average distance traveled. Canals were subsidized and subsidy not reflected in SS calculation. NOT true, Fogel allows for this and not that important quantitatively. Canals were monopolies and would have charged monopoly price in absence of RR. TRUE in British case but not true in US case. Also, effect is offset by assumption of perfectly inelastic demand which is economically inconsistent under monopoly pricing. BUT even if this were maintained, effect would be small because canals did not provide much of the water-borne transportation: maximum effect is around $39 million, which adds about 0.3 percentage points of GNP to the SS. Canals had rising LRMC. Regressions in the paper suggest to the contrary, that MC was decreasing in tonnage shipped and if more tonnage had been shipped by water, canals would have been designed to carry more traffic (similar to widening a modern highway).

15 EC 2333: Transportation: Topic 4
16 EC 2333: Transportation: Topic 4
17 Major Criticisms of the Time (4): Substitution Possibilities

Major Criticisms of the Time (4): Substitution Possibilities

Lebergott: RR and water were perfect substitutes. False, even on routes where the two ostensibly competed. Example: Grain shipped from Chicago to NY could go by train or water. Cross-elasticity of substitution of grain demand for water transportation with respect to rail rate is about 1.

18 Major Criticisms at the Time (5): RR and Long-Run Growth

Major Criticisms at the Time (5): RR and Long-Run Growth

Probably the most important criticism of SS is that it is a SR measure not a LR measure. In LR, resource costs will compound. Possible effects on factor supply (see below). BUT offsetting this is possibility of faster tech change in the next best alternative. David argues that SS misses scale economies. Fogel argues back that these cannot be at the level of the firm (because these would already have been captured) and a reasonable assessment suggests they must be small.

19 EC 2333: Transportation: Topic 4
20 EC 2333: Transportation: Topic 4
21 Major Criticisms at the Time (6): Williamson GE model

Major Criticisms at the Time (6): Williamson GE model

Williamson 1974 presents CGE model of C19 American economy. Chapter on RR social savings. Claims the effect is around 20 percent of GNP. Charles Kahn reassesses. Finds errors in computer program, reduces to 12 percent. Allows for resource costs of transportation, reduces to 6.8 percent. Williamson does not measure RR, but rather effect of improvements in RR and water transportation captured by inter-regional convergence (MW and NE) in grain prices. VERY large impact on regional production, industrial composition of output.

22 Lessons: The C19 Transportation Revolution

Lessons: The C19 Transportation Revolution

Mistake to think that one technology just replaced another: water transport experienced great gains in TFP in C19 and in C20, of course, the wagon became the truck The key advances prior to the C20 were in medium and long-haul transportation Water had advantages in the very long haul but rail did better in medium haul; water + rail had an enormous advantage over wagon transportation. In countries without adequate water transportation, SS of railroad could be very high Example: Mexico (30 percent of GDP in early 20th century)

22

23 Railroads Re-visited

Railroads Re-visited

Work on RR and other aspects of transportation went out of favor by the 1980s. Return to favor in mid-2000s with studies of the interstate highway system (Baum-Snow; Michaels). Important predecessor in economic history is Craig and Palmquist (1996) which uses county level data and historical maps. Atack-Margo project: uses GIS applied to digitized historical maps to generate county-level panel database on the TR. Linked to economic outcomes typically from census data. Basic idea is to use natural experiment econometrics: RR is a treatment. Similar projects underway in other countries (Germany, England, China, Sweden).

24 Motivation

Motivation

Long tradition in economics of studying the local impact of transportation improvements. Social savings not especially well-suited to this. Local effects may be useful in assessing certain outcomes not captured by Fogel. Examples include technological change (Sokoloff 1988 on patents and Erie Canal), urbanization, agricultural improvements, mortality. Direct predecessor is Haines and Margo (2008). Used Craig-Palmquist-Weiss (CPW) transportation database (1850-60) matched to published census and IPUMS. New TR data set is a significant improvement over CPW. Atack, Margo, and co-author studies focus primarily on Midwest in 1850s. Discussion today is of the primary paper (Social Science History).

25 New Data, Part One: Background

New Data, Part One: Background

Large body of nineteenth century transportation maps stored at various archives/libraries. Previously used by scholars to measure change over time and construct data sets via visual matching (eg. Craig, Palmquist, and Weiss 1996). Very costly to access maps on-site and many opportunities for subtle and even gross error. Libraries and archives have embraced the digital age. On-line maps can be manipulated via GIS software to create numerical databases more cheaply and (potentially) more accurately than visual methods. GIS can also be used to create data from scanned paper maps.

26 New Data, Part Two: Original Version

New Data, Part Two: Original Version

Used in all papers to date except Atack, Margo and Perlman (in progress). Constructed by moving forward in time: 1850 uses an 1850 map, 1860 uses and 1860 map, etc. Problems with this approach but apparent only after we were done. Time frame is 1850-1880, covers entire county (BUT work to date focuses primarily on Midwest, 1850-60). Census year frequency. Panel data at the county level. Archival version is unbalanced but mostly we work with a balanced sample (see below). RR access is a 0-1 dummy (is there a RR in or bordering the county at date t?). Other types of transportation (e.g. canal, navigable waterway) coded as dummy variables as of 1850. PROBLEM: ignores change over 1850s. Easy to link to census data (published) or IPUMS using county FIPS code. BUT a major issue: changes in county boundaries. So far have either (a) restrict sample to counties with constant boundaries (b) ignored the problem.

27 New TR Data Set (2)

New TR Data Set (2)

Locate and download digital TR maps and/or create digital maps from paper copies. Overlay digital TR maps on digital county boundary maps. Data set starts in 1850 and currently ends in 1880 at census year frequencies. At present GIS overlay is sufficiently accurate to create 0-1 access variables (eg. is there a rail line in a county as of a specific census date?) Water transportation access: Great Lakes frontage = 1, Ocean frontage = 1, Navigable river = 1, canal = 1. Measured as of 1850. Given county boundaries as of year t, no change over time so impact cannot be assessed (except maybe IV). Railroad access = 1 if railroad line passes through county. Also plan to compute total mileage, number of nodes, etc. Current version was constructed working forward from 1850. New version of database in progress, working backwards from 1911 map, removing lines. Fixes some measurement error in access dummy and should be superior for refined measures like mileage (1911 map overlays on county boundary maps very well).

28 Digitized Maps

Digitized Maps

29 Geo-referenced Digitized Map

Geo-referenced Digitized Map

30 Geo-referenced Digitized Map

Geo-referenced Digitized Map

31 Specifics of our RR database

Specifics of our RR database

From 1911 New Century mapping, we progressively DELETE. Currently have mappings for 1911, 1887, 1882, 1870, 1860, and 1851 (copyright dates); 1840, and 1830 (Hallberg DBF off 1851 mapping)

32 EC 2333: Transportation: Topic 4
33 Measuring Transportation Access

Measuring Transportation Access

Original version uses dummy variables. Railroad = 1 if county has a railroad. Ditto for navigable river or canal. Does NOT cross county boundaries. Latest version has % of county within Z miles of transportation. Crosses county boundaries. Not yet in general use, still experimenting with Z parameter.

34 EC 2333: Transportation: Topic 4
35 EC 2333: Transportation: Topic 4
36 EC 2333: Transportation: Topic 4
37 Did RR cause Midwestern urbanization + population growth

Did RR cause Midwestern urbanization + population growth

Midwest is of interest because rapid RR diffusion occurs in the 1850s. Percent urban and pop density increase. Is this because of the RR? According to Fishlow, RRs were built where conditions already favored development, including urbanization and density. HOWEVER, stops short of estimating RR effect (not computationally feasible at the time). Atack, et. al. use DID and IV: economically significant, positive effect of RR on urbanization but not pop density.

38 EC 2333: Transportation: Topic 4
39 EC 2333: Transportation: Topic 4
40 Specifics of our Midwestern database

Specifics of our Midwestern database

Building on our 1860 mapping from Colton Midwest 1848-60 from Paxson, Railroads of the Old Northwest Transactions of the Wisconsin Academy (1914)

41 Specifics of our Midwestern database

Specifics of our Midwestern database

Annual new mileage open for rail service Supplemented with map data for Iowa and Missouri Extensions to 1861 and railroad gauges from Taylor & Neu (1956) Additionally, rivers used by steamboats, canals in operation and Great Lakes

42 Rivers and Canals

Rivers and Canals

43 Data

Data

TR database restricted to seven Midwestern states. Link to Haines-ICPSR. Basic sample consists of 278 counties which (a) existed in 1840-60 (b) had fixed county boundaries (c) had all data needed for regressions (d) got rail in the 1850s or did not have rail before the Civil War. We need fixed boundaries because Haines-ICPSR data only very recently corrected for boundary changes. Got rail in 1850s = treatment group. No rail = control group. Counties in sample with RR access in 1850 probably got railroads very late in 1840s. Not the best control sample for 1850s. Two outcomes: percent urban and log of population density (population/square miles). Percent urban is ? 2,500. Later look at 1 (percent urban > 0), which is, does the county have at least one urban area?

44 Railroads

Railroads

Sample counties: 278 counties that had stable borders 1840-1860 (for controls and linkage) ACCESS to transportation: county boundary defined by transportation route or county traversed by one Alternative metrics in works

45 Research Template So Far: Effects of RR

Research Template So Far: Effects of RR

Measure the treatment (causal) effect of TR on economic outcomes at the county level. Different from previous work which was aggregate (e.g. social savings). Recent studies in urban (Baum-Snow) and development (Donaldson on India; Bannerjee, et. al. on China). We adopt a natural experiments (reduced form) econometric approach to measure the treatment effect of RR. Popular in applied microeconomics, gaining ground in economic history circles. Ideal natural experiment: treatment vs. control group, treatment is as good as random. Link TR database to county level outcomes, typically published census or IPUMS using FIPS codes. Completed papers to date mainly using Midwest sample. Start with a balanced panel of counties, 1850-1860 and, if possible, 1840. Balanced: constant land area, 1850-60. Approximately 280 counties. Strategy #1: simple difference-in-differences. Control group: no RR access before Civil War. Treatment group: no rail in 1850, got rail by 1860. Strategy #2: DD w/covariates (including pre-existing trend if possible). Covariates statistically related to coming of the RR and (when data relevant) individual/household characteristics. Primary robustness check: instrumental variable. Congress authorized 60+ transportation surveys, 1824-1838. Draw straight line between starting and ending location of survey. IV = 1 if county lies on this straight line. Good first stage. Some evidence (in SSH paper) that IV satisfies the exclusion restriction. Other straight-line instruments.

46 TR and Urbanization

TR and Urbanization

Coming of RR increases trade. Midwest has comparative advantage in agriculture. Trade has to take place somewhere. Central place = urban area. Population growth occurs within economically feasible vicinity of transportation hubs. Rising farm incomes increase demand for Midwestern manufactures, which take place in urban areas due to agglomeration economies. When rail comes to one central place, others nearby grow in anticipation of future RR expansion. Return to this (more general) point later.

47 Tables 1-3

Tables 1-3

Table 1 shows distribution of basic sample by state Table 2 shows sample means in 1850 and 1860. Table 3, Panels A and B, show base differences-in-differences. Weight is land area in 1850 for density, average population for percent urban. Results for percent urban are very similar if land area is weight. Treatment effects are positive, small for population density. Table 3, last column is consistent w/ Fishlow . Treatment counties became more urban and densely populated in the decade (1840s) before arrival of railroad. Demand ahead of building rather than the reverse.

48 EC 2333: Transportation: Topic 4
49 EC 2333: Transportation: Topic 4
50 EC 2333: Transportation: Topic 4
51 EC 2333: Transportation: Topic 4
52 Dif-in-Dif Regression Estimates

Dif-in-Dif Regression Estimates

Pre-treatment differences in outcomes more general phenomenon: RR did not arrive randomly (Fishlow). Table 4 shows observable correlates of gaining rail access: high agricultural yields, growing urbanization, presence of navigable river (-) , state dummies also matter. BTW: Table 4 specification gives good propensity score results for matching estimator. Matching very close to DD. Columns 3, 5 of Table 5 shows DID with Table 3 controls x (year = 1860 interactions). No change in percent urban treatment effect, population density effect is larger and nearly significant at 5 percent level.

53 EC 2333: Transportation: Topic 4
54 EC 2333: Transportation: Topic 4
55 IV estimation

IV estimation

Conditional DID may not be enough if treatment correlated with error term. Need an instrumental variable. Congressional Survey IV. Between 1824 and 1838 Congress authorizes approximately 60 transportation surveys. Location (starting and end points) of most are reported in American State Papers. Many in Midwest. IV = 1 if county is on straight line between starting and end point of survey. Idea is that shortest distance between two points is a straight line and low cost construction is preferred. If we regress pre-trend (1840-50) of urbanization or log density on survey IV w/controls (1840 urbanization, density, water transport + state dummies), coefficients on IV are very close to zero and insignificant. Suggests exclusion restriction may be ok. First stage pretty good. Second stage coefficients are positive but imprecise enough such that are not significantly different from DID coefficients. Results shown in Table 6.

56 EC 2333: Transportation: Topic 4
57 EC 2333: Transportation: Topic 4
58 Table 7

Table 7

Use Table 4 (columns 3,5) DID coefficients to predict ? in urbanization and density. RR can explain a little more than half of growth in urbanization but only a small amount of population growth. Implications for Fogel SS. If (1) RR caused urbanization (2) effect of RR on urbanization > canals (or other water transportation improvements) (3) aggregate TFP increases because of urbanization, then Fogel SS is downward bias. (2) and (3) are speculative at this point.

59 EC 2333: Transportation: Topic 4
60 Robustness Check (1)

Robustness Check (1)

We keep county boundaries fixed from 1850-60 but require ONLY that county existed in 1840. What if we fix county boundaries over 1840-60? Sample size much smaller (188) counties. DD very similar for urbanization but impact on population density is larger and now significant. Explanatory power of gaining rail access for population density ?s 1850-60 doubles (about 10 percent) but still small.

61 EC 2333: Transportation: Topic 4
62 Robustness Checks (2)

Robustness Checks (2)

Alternative measure of urbanization: 1(percent urban > 0). This variable = 1 if percent urban is positive, 0 otherwise. Potentially useful because there are a lot of zeros. See below. Base DD: positive treatment effect, fairly large and precisely measured. Ditto with controls, IV results similar. Rail explains 62 percent of increase in 1(percent urban>0) between 1850 and 1860 (predicted = 0.115, actual = 0.173).

63 EC 2333: Transportation: Topic 4
64 EC 2333: Transportation: Topic 4
65 Robustness Checks (3)

Robustness Checks (3)

Useful to compare DD with lagged dependent variable (LDV). Economic reason is reverse Ashenfelter dip. Counties that grow rapidly in 1840s are more likely to get a RR BUT because experience slower growth in 1850s because of more rapid growth in 1840s. Fishlow missed this BUT doesnt matter much quantitatively (DD is close to LDV, see next table). Note that DD for percent urban > LDV. Indicates that percent urban is an explosive time series. 1(percent urban > 0) is ok, however. Because percent urban has so many zeros, good idea to try quantile regression. Need to go far up the distribution (90th and 95th percentiles). Pooled TSCS, with dummies for state, water transportation. Treatment effect of rail access is 0.12-0.16 (depending on weights), significant at 5 percent level. Still significant if standard errors are bootstrapped.

66 EC 2333: Transportation: Topic 4
67 Other Results: Impact of RR

Other Results: Impact of RR

In manufacturing (1850-70): positive effect of RR on % factory (firms with 16+ employees). Consistent with Adam Smith (division of labor limited by the extent of the market). Atack, Haines, Margo (2011). Midwest: positive effect on % of acres improved and average farm values. Atack and Margo (JTLU, 2011) Midwest: negative effect on ownership of land, possibly because of increase in minimum efficient scale of farms (or credit constraints). Atack and Margo (2012). Entire country (1850-70): positive effect on school enrollment. Atack, Margo, and Perlman (in progress).

68 Michaels

Michaels

Recent PhD from MIT. Work focuses on impact of transportation improvements and factor endowments. Motivating framework is international trade. Paper on reading list focuses on impact of the US federal highway system. System was proposed in 1940s. Idea was to facilitate national defense, connect major metro areas, and the US to Mexico and Canada. BUT to do this, it was necessary to go through rural areas. What was the effect in such areas?

69 Data

Data

Focuses on approximately half the mileage of the interstate highway system, basically long highways (n = 18) built between 1959-1975. Eg: I-40, I-95. Unit of observation is county. Has to be 50 percent rural in 1950 and essentially no county boundary changes. Two instruments: (a) 1944 plan (b) orientation of the county towards major urban areas Key results: positive effects on retail trade, earnings from transportation, likelihood of working outside county of residence (statistically insignificant, however).

70 EC 2333: Transportation: Topic 4
71 EC 2333: Transportation: Topic 4
72 EC 2333: Transportation: Topic 4
73 EC 2333: Transportation: Topic 4
74 EC 2333: Transportation: Topic 4
75 EC 2333: Transportation: Topic 4
76 EC 2333: Transportation: Topic 4
EC 2333: Transportation: Topic 4
http://900igr.net/prezentacija/anglijskij-jazyk/ec-2333-transportation-topic-4-148876.html
c

661

29
900igr.net > > > EC 2333: Transportation: Topic 4