Картинки на тему «EC 2333: Transportation: Topic 4» |
Без темы | ||
<< Ease of Access and Assistive Technology on Windows 7 | ECE G201: Introductory Material >> |
Автор: RAM. Чтобы познакомиться с картинкой полного размера, нажмите на её эскиз. Чтобы можно было использовать все картинки для урока английского языка, скачайте бесплатно презентацию «EC 2333: Transportation: Topic 4.ppt» со всеми картинками в zip-архиве размером 7118 КБ.
Сл | Текст | Сл | Текст |
1 | EC 2333: Transportation: Topic #4. | 31 | DELETE. Currently have mappings for 1911, |
Professor Robert A. Margo Harvard | 1887, 1882, 1870, 1860, and 1851 | ||
University Spring 2014. | (copyright dates); 1840, and 1830 | ||
2 | Outline. Social Savings of the | (Hallberg DBF off 1851 mapping). | |
Railroad: Fogel Impact of RR on local | 32 | ||
economies: Atack, et. al. Re-assessment of | 33 | Measuring Transportation Access. | |
Fogel: Donaldson and Hornbeck (student | Original version uses dummy variables. | ||
presentation) If time: Guy Michaels on the | Railroad = 1 if county has a railroad. | ||
Interstate Highway System. | Ditto for navigable river or canal. Does | ||
3 | Fogel (1964). One of the most famous | NOT cross county boundaries. Latest | |
(perhaps most famous) of all cliometrics | version has % of county within Z miles of | ||
studies. Originally a JEH paper, then a | transportation. Crosses county boundaries. | ||
dissertation, then a book. Book is about | Not yet in general use, still | ||
more (much) than the “social savings” of | experimenting with Z parameter. | ||
the RR but, for better or worse, it is the | 34 | ||
social savings chapter that received the | 35 | ||
most attention. HUGE follow-up literature, | 36 | ||
some critical, some applying the same (or | 37 | Did RR “cause” Midwestern urbanization | |
modified) method to other countries. | + population growth? Midwest is of | ||
4 | Definition of the Social Savings. | interest because rapid RR diffusion occurs | |
Social savings (percent): savings in | in the 1850s. Percent urban and pop | ||
transportation costs in a given year by | density increase. Is this “because” of the | ||
shipping the same amount of goods by the | RR? According to Fishlow, RRs were built | ||
next best alternative/national income As | where conditions already favored | ||
defined, SS is an upper bound to the true | development, including urbanization and | ||
resource cost (because the true demand for | density. HOWEVER, stops short of | ||
transportation is NOT perfectly | estimating RR effect (not computationally | ||
inelastic). Application is to the RR. Next | feasible at the time). Atack, et. al. use | ||
best alternative is some combination of | DID and IV: economically significant, | ||
water and wagon transportation. Concept | positive effect of RR on urbanization but | ||
does not originate with Fogel but rather | not pop density. | ||
with engineers (and economists) in C19 | 38 | ||
Europe. | 39 | ||
5 | Fogel’s Algebra. Two sectors, | 40 | Specifics of our Midwestern database. |
Transportation (T) and all other things | Building on our 1860 mapping from Colton | ||
(A). Households consume A and some portion | Midwest 1848-60 from Paxson, “Railroads of | ||
of T as a final good. T can be produced in | the Old Northwest” Transactions of the | ||
two different ways, by R or by W. Overall, | Wisconsin Academy… (1914). | ||
R is more efficient than W. Holding T | 41 | Specifics of our Midwestern database. | |
fixed, if R is no longer available, then | Annual new mileage open for rail service | ||
some resources (L and K) will have to be | Supplemented with map data for Iowa and | ||
shifted from the production of A to the | Missouri Extensions to 1861 and railroad | ||
production of T. The effect of this | gauges from Taylor & Neu (1956) | ||
re-allocation on national income is, to a | Additionally, rivers used by steamboats, | ||
first approximation, the additional cost | canals in operation and Great Lakes. | ||
associated with W (i.e. it is the wage x | 42 | Rivers and Canals… | |
labor reallocated + rental price of | 43 | Data. TR database restricted to seven | |
capital x capital reallocated). Because T | Midwestern states. Link to Haines-ICPSR. | ||
is held fixed, this is an upper bound to | Basic sample consists of 278 counties | ||
the true SS. Also note that resources (L | which (a) existed in 1840-60 (b) had fixed | ||
and K) are fully employed. Note that SS | county boundaries (c) had all data needed | ||
can also be calculated as ? x dZ, where dZ | for regressions (d) got rail in the 1850s | ||
= efficiency gain of the RR in percentage | or did not have rail before the Civil War. | ||
terms (percent effect on costs) and ? = | We need fixed boundaries because | ||
transportation’s share of national income. | Haines-ICPSR data only very recently | ||
6 | corrected for boundary changes. Got rail | ||
7 | Results. Calculation performed in two | in 1850s = treatment group. No rail = | |
stages: inter-regional and intra-regional | control group. Counties in sample with RR | ||
Inter-regional: between major shipping | access in 1850 probably got railroads very | ||
points (eg. New York-Chicago) Four | late in 1840s. Not the best control sample | ||
commodities are considered: pork, beef, | for 1850s. Two outcomes: percent urban and | ||
wheat, and corn, comprise 75 percent of | log of population density | ||
commodities shipped by rail in 1890 Big | (population/square miles). Percent urban | ||
surprise: initial calculation of | is ? 2,500. Later look at 1 (percent urban | ||
inter-regional is negative, turns positive | > 0), which is, does the county have at | ||
after adjustment is made for lower | least one urban area? | ||
insurance costs (cargo losses smaller | 44 | Railroads. Sample counties: 278 | |
under rail) and speedier service | counties that had stable borders 1840-1860 | ||
(inventory costs lower) Intra-regional: | (for controls and linkage) ACCESS to | ||
farmers would have used more wagon | transportation: county boundary defined by | ||
transportation. This is the main effect. | transportation route or county traversed | ||
Total for four commodities is around 117 | by one Alternative metrics in works. | ||
million, Fogel inflates by ratio of value | 45 | Research Template So Far: Effects of | |
added = 4.7 percent of GNP in 1890. 7. | RR. Measure the “treatment” (causal) | ||
8 | Follow-up Literature. Book had a HUGE | effect of TR on economic outcomes at the | |
impact. Narrow: applications of SS concept | county level. Different from previous work | ||
to transportation improvements in other | which was aggregate (e.g. “social | ||
settings. Probably the most important is | savings”). Recent studies in urban | ||
Coatsworth on Mexico (SS is around 30 | (Baum-Snow) and development (Donaldson on | ||
percent). SS concept also applied to other | India; Bannerjee, et. al. on China). We | ||
innovations (steam engine, tractor) but | adopt a “natural experiments” (reduced | ||
not many. Broad: accelerated the shift to | form) econometric approach to measure the | ||
quantitative methods in economic history. | treatment effect of RR. Popular in applied | ||
SUBSTANTIAL critical literature at the | microeconomics, gaining ground in economic | ||
time and shortly after: McClelland, David, | history circles. Ideal natural experiment: | ||
Lebergott, Williamson, among others. Fogel | treatment vs. control group, treatment is | ||
(1979) is the response. | “as good as” random. Link TR database to | ||
9 | Lebergott: SS too small because RR | county level outcomes, typically published | |
system had large economies of scale and | census or IPUMS using FIPS codes. | ||
was never even close to capacity. Use | Completed papers to date mainly using | ||
minimum of LR AC curve instead Fogel’s | Midwest sample. Start with a “balanced” | ||
response: transport costs that were | panel of counties, 1850-1860 and, if | ||
technologically feasible were never | possible, 1840. Balanced: constant land | ||
economically feasible because the system | area, 1850-60. Approximately 280 counties. | ||
always carried a mixture of goods to many | Strategy #1: simple | ||
places. Major Criticisms at the Time (1): | difference-in-differences. Control group: | ||
Technological versus Economic Definition. | no RR access before Civil War. Treatment | ||
10 | group: no rail in 1850, got rail by 1860. | ||
11 | Major Criticisms at the Time (2): | Strategy #2: DD w/covariates (including | |
Perfectly Inelastic Demand and Marginal | pre-existing trend if possible). | ||
Social Rate of Return. Fogel assumes price | Covariates statistically related to coming | ||
elasticity of demand for RR is zero. | of the RR and (when data relevant) | ||
Clearly false, but especially for | individual/household characteristics. | ||
passenger transportation. How large is the | Primary robustness check: instrumental | ||
bias? Assume log-linear demand. Answer: | variable. Congress authorized 60+ | ||
VERY. Measure marginal social rate of | transportation surveys, 1824-1838. Draw | ||
return (Lebergott; Nerlove). Captures | straight line between starting and ending | ||
impact of the last dollar invested, not | location of survey. IV = 1 if county lies | ||
the total impact (BUT if you could measure | on this straight line. Good “first stage”. | ||
this for every dollar invested, would be | Some evidence (in SSH paper) that IV | ||
useful). | satisfies the exclusion restriction. Other | ||
12 | straight-line instruments. | ||
13 | 46 | TR and Urbanization. Coming of RR | |
14 | Major Criticisms at the Time (3): | increases trade. Midwest has comparative | |
Water Rates. Water rates too low | advantage in agriculture. Trade has to | ||
(McClelland and others). BUT Fogel used | take place somewhere. Central place = | ||
water rates appropriate to average | urban area. Population growth occurs | ||
distance traveled. Canals were subsidized | within economically feasible vicinity of | ||
and subsidy not reflected in SS | transportation hubs. Rising farm incomes | ||
calculation. NOT true, Fogel allows for | increase demand for Midwestern | ||
this and not that important | manufactures, which take place in urban | ||
quantitatively. Canals were monopolies and | areas due to agglomeration economies. When | ||
would have charged monopoly price in | rail comes to one central place, others | ||
absence of RR. TRUE in British case but | nearby grow in anticipation of future RR | ||
not true in US case. Also, effect is | expansion. Return to this (more general) | ||
offset by assumption of perfectly | point later. | ||
inelastic demand which is economically | 47 | Tables 1-3. Table 1 shows distribution | |
inconsistent under monopoly pricing. BUT | of basic sample by state Table 2 shows | ||
even if this were maintained, effect would | sample means in 1850 and 1860. Table 3, | ||
be small because canals did not provide | Panels A and B, show base | ||
much of the water-borne transportation: | differences-in-differences. Weight is land | ||
maximum effect is around $39 million, | area in 1850 for density, average | ||
which adds about 0.3 percentage points of | population for percent urban. Results for | ||
GNP to the SS. Canals had rising LRMC. | percent urban are very similar if land | ||
Regressions in the paper suggest to the | area is weight. Treatment effects are | ||
contrary, that MC was decreasing in | positive, small for population density. | ||
tonnage shipped and if more tonnage had | Table 3, last column is consistent w/ | ||
been shipped by water, canals would have | Fishlow . Treatment counties became more | ||
been designed to carry more traffic | urban and densely populated in the decade | ||
(similar to widening a modern highway). | (1840s) before arrival of railroad. | ||
15 | “Demand ahead of building” rather than the | ||
16 | reverse. | ||
17 | Major Criticisms of the Time (4): | 48 | |
Substitution Possibilities. Lebergott: RR | 49 | ||
and water were perfect substitutes. False, | 50 | ||
even on routes where the two ostensibly | 51 | ||
competed. Example: Grain shipped from | 52 | Dif-in-Dif Regression Estimates. | |
Chicago to NY could go by train or water. | Pre-treatment differences in outcomes more | ||
Cross-elasticity of substitution of grain | general phenomenon: RR did not arrive | ||
demand for water transportation with | randomly (Fishlow). Table 4 shows | ||
respect to rail rate is about 1. | observable correlates of gaining rail | ||
18 | Major Criticisms at the Time (5): RR | access: high agricultural yields, growing | |
and Long-Run Growth. Probably the most | urbanization, presence of navigable river | ||
important criticism of SS is that it is a | (-) , state dummies also matter. BTW: | ||
SR measure not a LR measure. In LR, | Table 4 specification gives good | ||
resource costs will compound. Possible | propensity score results for matching | ||
effects on factor supply (see below). BUT | estimator. Matching very close to DD. | ||
offsetting this is possibility of faster | Columns 3, 5 of Table 5 shows DID with | ||
tech change in the next best alternative. | Table 3 controls x (year = 1860 | ||
David argues that SS misses scale | interactions). No change in percent urban | ||
economies. Fogel argues back that these | treatment effect, population density | ||
cannot be at the level of the firm | effect is larger and nearly significant at | ||
(because these would already have been | 5 percent level. | ||
captured) and a reasonable assessment | 53 | ||
suggests they must be small. | 54 | ||
19 | 55 | IV estimation. Conditional DID may not | |
20 | be enough if treatment correlated with | ||
21 | Major Criticisms at the Time (6): | error term. Need an instrumental variable. | |
Williamson GE model. Williamson 1974 | “Congressional Survey” IV. Between 1824 | ||
presents CGE model of C19 American | and 1838 Congress authorizes approximately | ||
economy. Chapter on RR social savings. | 60 transportation surveys. Location | ||
Claims the effect is around 20 percent of | (starting and end points) of most are | ||
GNP. Charles Kahn reassesses. Finds errors | reported in American State Papers. Many in | ||
in computer program, reduces to 12 | Midwest. IV = 1 if county is on straight | ||
percent. Allows for resource costs of | line between starting and end point of | ||
transportation, reduces to 6.8 percent. | survey. Idea is that shortest distance | ||
Williamson does not measure RR, but rather | between two points is a straight line and | ||
effect of improvements in RR and water | low cost construction is preferred. If we | ||
transportation captured by inter-regional | regress pre-trend (1840-50) of | ||
convergence (MW and NE) in grain prices. | urbanization or log density on survey IV | ||
VERY large impact on regional production, | w/controls (1840 urbanization, density, | ||
industrial composition of output. | water transport + state dummies), | ||
22 | Lessons: The C19 Transportation | coefficients on IV are very close to zero | |
Revolution. Mistake to think that one | and insignificant. Suggests exclusion | ||
technology just replaced another: water | restriction may be ok. First stage pretty | ||
transport experienced great gains in TFP | good. Second stage coefficients are | ||
in C19 and in C20, of course, the “wagon” | positive but imprecise enough such that | ||
became the truck The key advances prior to | are not significantly different from DID | ||
the C20 were in medium and long-haul | coefficients. Results shown in Table 6. | ||
transportation Water had advantages in the | 56 | ||
very long haul but rail did better in | 57 | ||
medium haul; water + rail had an enormous | 58 | Table 7. Use Table 4 (columns 3,5) DID | |
advantage over wagon transportation. In | coefficients to predict ? in urbanization | ||
countries without adequate water | and density. RR can “explain” a little | ||
transportation, SS of railroad could be | more than half of growth in urbanization | ||
very high Example: Mexico (30 percent of | but only a small amount of population | ||
GDP in early 20th century). 22. | growth. Implications for Fogel SS. If (1) | ||
23 | Railroads Re-visited. Work on RR and | RR “caused” urbanization (2) effect of RR | |
other aspects of transportation went out | on urbanization > canals (or other | ||
of favor by the 1980s. Return to favor in | water transportation improvements) (3) | ||
mid-2000s with studies of the interstate | aggregate TFP increases because of | ||
highway system (Baum-Snow; Michaels). | urbanization, then Fogel SS is downward | ||
Important predecessor in economic history | bias. (2) and (3) are speculative at this | ||
is Craig and Palmquist (1996) which uses | point. | ||
county level data and historical maps. | 59 | ||
Atack-Margo project: uses GIS applied to | 60 | Robustness Check (1). We keep county | |
digitized historical maps to generate | boundaries fixed from 1850-60 but require | ||
county-level panel database on the TR. | ONLY that county existed in 1840. What if | ||
Linked to economic outcomes typically from | we fix county boundaries over 1840-60? | ||
census data. Basic idea is to use “natural | Sample size much smaller (188) counties. | ||
experiment” econometrics: RR is a | DD very similar for urbanization but | ||
“treatment”. Similar projects underway in | impact on population density is larger and | ||
other countries (Germany, England, China, | now significant. Explanatory power of | ||
Sweden). | gaining rail access for population density | ||
24 | Motivation. Long tradition in | ?’s 1850-60 doubles (about 10 percent) but | |
economics of studying the local impact of | still small. | ||
transportation improvements. Social | 61 | ||
savings not especially well-suited to | 62 | Robustness Checks (2). Alternative | |
this. Local effects may be useful in | measure of urbanization: 1(percent urban | ||
assessing certain outcomes not captured by | > 0). This variable = 1 if percent | ||
Fogel. Examples include technological | urban is positive, 0 otherwise. | ||
change (Sokoloff 1988 on patents and Erie | Potentially useful because there are a lot | ||
Canal), urbanization, agricultural | of zeros. See below. Base DD: positive | ||
improvements, mortality. Direct | treatment effect, fairly large and | ||
predecessor is Haines and Margo (2008). | precisely measured. Ditto with controls, | ||
Used Craig-Palmquist-Weiss (CPW) | IV results similar. Rail explains 62 | ||
transportation database (1850-60) matched | percent of increase in 1(percent | ||
to published census and IPUMS. New TR data | urban>0) between 1850 and 1860 | ||
set is a significant improvement over CPW. | (predicted = 0.115, actual = 0.173). | ||
Atack, Margo, and co-author studies focus | 63 | ||
primarily on Midwest in 1850s. Discussion | 64 | ||
today is of the primary paper (Social | 65 | Robustness Checks (3). Useful to | |
Science History). | compare DD with lagged dependent variable | ||
25 | New Data, Part One: Background. Large | (LDV). Economic reason is reverse | |
body of nineteenth century transportation | “Ashenfelter” dip. Counties that grow | ||
maps stored at various archives/libraries. | rapidly in 1840s are more likely to get a | ||
Previously used by scholars to measure | RR BUT because experience slower growth in | ||
change over time and construct data sets | 1850s because of more rapid growth in | ||
via visual matching (eg. Craig, Palmquist, | 1840s. Fishlow missed this BUT doesn’t | ||
and Weiss 1996). Very costly to access | matter much quantitatively (DD is close to | ||
maps on-site and many opportunities for | LDV, see next table). Note that DD for | ||
subtle and even gross error. Libraries and | percent urban > LDV. Indicates that | ||
archives have embraced the digital age. | percent urban is an explosive time series. | ||
On-line maps can be manipulated via GIS | 1(percent urban > 0) is ok, however. | ||
software to create numerical databases | Because percent urban has so many zeros, | ||
more cheaply and (potentially) more | good idea to try quantile regression. Need | ||
accurately than visual methods. GIS can | to go far up the distribution (90th and | ||
also be used to create data from scanned | 95th percentiles). Pooled TSCS, with | ||
paper maps. | dummies for state, water transportation. | ||
26 | New Data, Part Two: Original Version. | Treatment effect of rail access is | |
Used in all papers to date except Atack, | 0.12-0.16 (depending on weights), | ||
Margo and Perlman (in progress). | significant at 5 percent level. Still | ||
Constructed by moving “forward” in time: | significant if standard errors are | ||
1850 uses an 1850 map, 1860 uses and 1860 | bootstrapped. | ||
map, etc. Problems with this approach but | 66 | ||
apparent only after we were done. Time | 67 | Other Results: Impact of RR. In | |
frame is 1850-1880, covers entire county | manufacturing (1850-70): positive effect | ||
(BUT work to date focuses primarily on | of RR on % “factory” (firms with 16+ | ||
Midwest, 1850-60). Census year frequency. | employees). Consistent with Adam Smith | ||
Panel data at the county level. Archival | (division of labor limited by the extent | ||
version is “unbalanced” but mostly we work | of the market). Atack, Haines, Margo | ||
with a “balanced” sample (see below). RR | (2011). Midwest: positive effect on % of | ||
access is a 0-1 dummy (is there a RR in or | acres improved and average farm values. | ||
bordering the county at date t?). Other | Atack and Margo (JTLU, 2011) Midwest: | ||
types of transportation (e.g. canal, | negative effect on ownership of land, | ||
navigable waterway) coded as dummy | possibly because of increase in minimum | ||
variables as of 1850. PROBLEM: ignores | efficient scale of farms (or credit | ||
change over 1850s. Easy to link to census | constraints). Atack and Margo (2012). | ||
data (published) or IPUMS using county | Entire country (1850-70): positive effect | ||
FIPS code. BUT a major issue: changes in | on school enrollment. Atack, Margo, and | ||
county boundaries. So far have either (a) | Perlman (in progress). | ||
restrict sample to counties with constant | 68 | Michaels. Recent PhD from MIT. Work | |
boundaries (b) ignored the problem. | focuses on impact of transportation | ||
27 | New TR Data Set (2). Locate and | improvements and factor endowments. | |
download digital TR maps and/or create | Motivating framework is international | ||
digital maps from paper copies. Overlay | trade. Paper on reading list focuses on | ||
digital TR maps on digital county boundary | impact of the US federal highway system. | ||
maps. Data set starts in 1850 and | System was proposed in 1940s. Idea was to | ||
currently ends in 1880 at census year | facilitate national defense, connect major | ||
frequencies. At present GIS overlay is | metro areas, and the US to Mexico and | ||
sufficiently accurate to create 0-1 access | Canada. BUT to do this, it was necessary | ||
variables (eg. is there a rail line in a | to go through rural areas. What was the | ||
county as of a specific census date?) | effect in such areas? | ||
Water transportation access: Great Lakes | 69 | Data. Focuses on approximately half | |
frontage = 1, Ocean frontage = 1, | the mileage of the interstate highway | ||
Navigable river = 1, canal = 1. Measured | system, basically long highways (n = 18) | ||
as of 1850. Given county boundaries as of | built between 1959-1975. Eg: I-40, I-95. | ||
year t, no change over time so impact | Unit of observation is county. Has to be | ||
cannot be assessed (except maybe IV). | 50 percent rural in 1950 and essentially | ||
Railroad access = 1 if railroad line | no county boundary changes. Two | ||
passes through county. Also plan to | instruments: (a) 1944 plan (b) orientation | ||
compute total mileage, number of nodes, | of the county towards major urban areas | ||
etc. Current version was constructed | Key results: positive effects on retail | ||
working forward from 1850. New version of | trade, earnings from transportation, | ||
database in progress, working backwards | likelihood of working outside county of | ||
from 1911 map, removing lines. Fixes some | residence (statistically insignificant, | ||
measurement error in access dummy and | however). | ||
should be superior for refined measures | 70 | ||
like mileage (1911 map overlays on county | 71 | ||
boundary maps very well). | 72 | ||
28 | Digitized Maps. | 73 | |
29 | Geo-referenced Digitized Map. | 74 | |
30 | Geo-referenced Digitized Map. | 75 | |
31 | Specifics of our RR database. From | 76 | |
1911 New Century mapping, we progressively | |||
EC 2333: Transportation: Topic 4.ppt |
«Teddy bear» - Panda Bear. Bear. Bears. Signicificant studies of pandas in the wild does not exist. Some 38 million years ago bears began to go their own evolutionary way. The size of the large males Kodiak Bear exceeds 3 m and their weight can reach a ton. Teddy Bear. The polar bear can without resting 80 km by ice-cold water to swim.
«Технологии на уроках английского языка» - Содержание. При обучении аудированию: Значительно расширяют возможности предъявления учебной информации. Способствуют формированию у учащихся рефлексии. При обучении лексике: Позволяют существенно повысить мотивацию студентов к обучению. При обучении фонетике: Перенос центра тяжести с обучения на учение.
«Olympic games 2012» - Olympic Torch Relay Olympic Games. Summer Olympic Games 2012. Oscar Pistorius is a man who never gives up. Oscar on the Olympic Games in 2012. Opening Ceremony. Olympic games in london. Medal standings. Sports facilities. Closing ceremony. Talismans of Games – Wenlock and Mandeville. London became the first city to have received three Olympics.
«Russian games» - They like Football very much. Basketball is popular in Russia. Basketball appeared in the Russia in 1901. England is the home of football. Many children are playing football. In Russia football appeared at the end 19th century. Sports and games popular in Russia. Football is very popular in Russia. It is played ball basketball.
«Theme park» - Information about theme parks. Asterix Park Disneyland Paris Legoland. Have you ever been to a theme park? Урок английского языка. What is it? Asterix Park. Disneyland Paris. Тематические парки. If you want to find out more about theme parks, visit. Live shows and attractions. Legoland. 7 класс Развлечения.
«Christmas in the USA» - Рождество в Америке. Ошибка Санта Клауса. New Year. Traditions. Конкурс стихов. Jingle, Bells. Boxing Day. Беседа в классе. Merry Christmas. Чаепитие. Рождественский чулок. Mistletoe. Письмо Санта Клауса. Around the Christmas Tree. Activities. Symbols of Christmas. We Wish You a Merry Christmas. Рождество.