Economic Consequences of US Federal Land Disposal

Transcription

1 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 1 / 54 ( Economic Consequences of US Federal Land Disposal Christian Dippel UCLA Trevor O Grady Harvard trevorogrady@fas.harvard.edu Alex Whalley UC Merced NBER Summer Institute (DAE) July 10, 2014

2 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 2 / 54 ( Outline of Presentation 1. Land patent data set 2. Model of frontier settlement 3. Agricultural Adjustment Land consolidation frictions within railroad grant boundaries

3 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 3 / 54 ( US Land Disposal: Background By 1850, the federal government acquired vast territories spanning the current contiguous states By 1940, nearly half of this land had been directly transferred to private owners. Contentious congressional debates regarding land disposal Revenue motives Equity concerns Survey design Later research echoes themes of earlier debates Gates (1936, 1968); Anderson and Hill (1975) Fogel and Rutner (1972); Allen (1991)

4 US Federal Land Claims Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 4 / 54 (

5 Disposal of Federal Land Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 5 / 54 (

6 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 6 / 54 ( Novel Spatial Data Set: General Land Office (GLO) Land Patents Land patents were issued whenever federal land was sold or granted. Patents represent the original deed transferring land ownership from the federal government to an individual, corporation, or state. Each patent contains information regarding: legal description of the plot plot size patentee name date issued act of authorization

7 Land Patents Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 7 / 54 (

8 Land Patents Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 8 / 54 (

9 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 9 / 54 ( Data Set: Land Patents 5.5 million patents Recorded by General Land Office Digitized by Bureau of Land Management Dominant allocation methods observed cash sales military warrants homestead grants railroad grants

10 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 10 / 54 ( Land Disposal over Time Acres Patented (10,000) Issue Year Cash Sales Enlarged Homestead Homestead Stock Raising Homestead

11 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 11 / 54 ( Spatiotemporal Land Patent Data Set Locating patented land Nearly all of the federal land disposal was administered through the Public Land Survey System (PLSS) We use the systematic legal descriptions to connect patents to various ground-truthed shapefiles of PLSS townships 6x6 mile grid cells

12 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 12 / 54 ( Patented Land: 1830 Cash Sales Homesteads

13 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 13 / 54 ( Patented Land: 1840 Cash Sales Homesteads

14 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 14 / 54 ( Patented Land: 1850 Cash Sales Homesteads

15 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 15 / 54 ( Patented Land: 1860 Cash Sales Homesteads

16 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 16 / 54 ( Patented Land: 1870 Cash Sales Homesteads

17 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 17 / 54 ( Patented Land: 1880 Cash Sales Homesteads

18 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 18 / 54 ( Patented Land: 1890 Cash Sales Homesteads

19 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 19 / 54 ( Patented Land: 1900 Cash Sales Homesteads

20 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 20 / 54 ( Patented Land: 1910 Cash Sales Homesteads

21 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 21 / 54 ( Patented Land: 1920 Cash Sales Homesteads

22 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 22 / 54 ( Patented Land: 1930 Cash Sales Homesteads

23 ippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 23 / 54 ( Patented Land: 1940 Cash Sales Homesteads

24 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 24 / 54 ( Messy Data and a Structural Approach Difficulties we face with estimating reduced form causal effects in variation in land granting rules on outcomes: 1. Many sources of variation: Homestead Act, Kincaid Act, Enlarged Homestead Act, Graduation Act, Railways 2. Acts endogenous to geography, esp. in large spatial units 3. Sharp date of acts but no sharp spatial breaks in acts used 4. Land settlement data at grant level but county-outcomes Alternative approach: take settler decision as the outcome, and model these structurally. Advantages: Can use the many sources of exogenous variation in acts and settlement incentives Unit of observation is township Answer counterfactual questions, e.g. how would frontier have evolved if there had been no homestead act This is joint work with Connan Snider at UCLA

25 Model Motivation I Discrete/Continuous choice: Each potential farmer decides jointly on (a) optimal farm size given settlement location (township) and (b) preferred settlement location, with non-settlement the outside option Model individual decisions but estimate in data at the market-level, i.e. the township Some guidance provided by stylized facts in the data: 1. Almost all homesteaders also bought some cash sale land 2. Farms with a homestead in the mix were smaller Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 25 / 54 (

26 Model Motivation II Why should a subsidy reduce optimal farm size? Answer likely to be selection: Cash-strapped farmers, who choose a constrained-optimal smaller farm size, but would not settle at all without homestead. Consistent with the history that Homestead Act attracted immigrants, urban poor and broke farmers from behind frontier Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 26 / 54 (

27 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 27 / 54 ( Model Outline The country is populated with some number of potential settlers N. There are J possible townships to locate a farm in. Data on when a township is surveyed and offered Each potential settler chooses a township j 0, 1,..., J (0 = no farm, the outside option) and, given j, a farm size L ij Each township is described by characteristics, split in location/land quality (Zjt ) determines productivity: Soil Quality, climate, Native hostility market access (X jt ): Distance from frontier, Railroads/Waterways proximity Some of (Z jt ) and (X jt ) are endogenous to the model, but not to the individual.

28 Model - Step 1, Continuous Choice Farm output depends on farm size and exogenous characteristics (not on individual farmers) Y ijt = θ(z jt )L ρ ij c(z jt)l ij Farmer s (quasi-linear) utility depends on how much farm output he eats himself and how much he sells: Y ijt = y e,ijt + y s,ijt u(y e,ijt, y s,ijt ) = y γ e,ijt + p(x jt)y s,ijt p are net price of output, which depends on market access (evolves endogenously in the model but exogenous to farmers) Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 28 / 54 (

29 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 29 / 54 ( Model - Step 1, Continuous Choice Optimal eat/sell split yields indirect utility of a farm as a function of farm size and state variables ( ( ) γ ( ) 1 ) u = p(x jt ) γ 1 γ 1 1 γ 1 1 γ + p(x jt )Y ijt (L ij, X jt, Z jt ) γ γ Once a farmer makes settles on a farm of a given size utility accrues each period forever Value of a farm of size L ij in township j is then: V j (L ij, X jt, Z jt ) = E[ β τ t u (X jτ, Z jτ, L ij ) X jt, Z jt ] τ=t

30 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 30 / 54 ( Model - Step 1, Continuous Choice Optimal size farm solves: max Lij f (L ij 160) κ(l ij, i L ij, L j ) + V j (L ij, X jt, Z jt ) s.t. f (L ij 160) W i L j is the average size of farm possibly capturing difficulty of consolidating widely held land κ is a congestion cost capturing the fact that it may be harder to put together a contiguous farm on good land the later you arrive f is the unit land price not subsidized by Homestead Act Before 1863, the 160 subsidy drops out With data on Graduation Act, f = 1.25$ becomes fjt Budget constraint introduces selection through W i

31 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 31 / 54 ( Model - Step 1, Continuous Choice Optimal farm size: either most you can afford or solves FOC: f {L > 160} κ + ρl 1 ρ ij E[ p(x tj )θ(z tj ) c(z tj )] = 0 L ij t=τ linear dependence: L ij not in the infinite sum Over time (within a township), farm sizes will vary because of wealth constraints and date of arrival (through congestion costs and through X tj, Z tj Modeling future prices, p tj = β 0 + β 1 X tj, θ tj = β 2 + β 3 Z tj We can measure these states and model their evolution (and thus model beliefs about their evolution) t=τ

32 Model - Step 2, Discrete Choice Given L jt, calculate discrete settlement choice j Utility of settler i choosing farm in township j is Outside option is U ij = V (X jt, Z jt, W i ) + ɛ ij U i0 = g(w i ) βe[log ( V (X jt+1, Z jt+1, W i )], i.e. depends on initial wealth plus continuation value (the logit social surplus function assuming logit errors) Individual settlement probabilities: P ijt = exp(u ijt ) exp(u i0t ) + k exp(u ikt) Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 32 / 54 (

33 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 33 / 54 ( Model - Step 3, Outcomes and Counterfactual For estimation, individual choices {L ijt, j t } aggregate up to township-level { L jt, share-settled jt } The vector { L jt, share-settled jt } describes the American frontier at time t Goodness of Fit: Evolution of estimated vs. actual frontier Ask counterfactual questions: Did immigration waves push the frontier faster? How would the frontier have evolved with no homestead act or different rules How important was Native hostility? How important was European demand for agricultural imports

34 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 34 / 54 ( Farm Size and Crop Yields in the US In 1900 Average Farm Size = 147 Acres Corn Production per Corn Acre = 28.1 Wheat Production per Wheat Acre = 12.2 In 1997 Farm Size = 487 Acres Corn Production per Corn Acre = Wheat Production per Wheat Acre = 39.5

35 Farm Size and Crop Yields in the US Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 35 / 54 (

36 Farm Size and Crop Yields in the US Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 36 / 54 (

37 Farm Size and GDP Per Capita Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 37 / 54 (

38 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 38 / 54 ( (Some) Recent Work Land allocation: farm size Bleakley and Ferrie (2014) Land allocation: mechanisms Libecap and Lueck (2011) Farm Size and Tractor Diffusion Olmstead and Rhode (2001) Farm Size and External Effects Hansen and Libecap (2003)

39 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 39 / 54 ( What We Do US railroad land grants Special rules Homestead maximum: 80 Acres Minimum price doubled Checkerboarding Potential for strategic holdout in land consolidation We ask: Did checkerboard affect farm size and land value growth? How do farms adjust?

40 Checkerboard Example Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 40 / 54 (

41 Railroad Grant Boundaries Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 41 / 54 (

42 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 42 / 54 ( Data and Sample 1. Agricultural Census: Farm size, Farm Value per Acre, Wheat Yield, Corn Yield, Acres planted, etc. 2. Railroad Land Grant Map Share county land checkerboarded Counties: Public land states only First patents issued before 1890 Share county land checkerboarded> 0 (1,021 counties)

43 Cross-Sectional Comparison, 1900 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 43 / 54 (

44 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 44 / 54 ( Econometric Specification Y it = α i + α st + β t ShareCheckerboard i + γ t X i + ɛ it (1) Y it : Agricultural outcome in county i in year t α i : county fixed effect α st : state-by-year fixed effect ShareCheckerboard i : fraction of county area within RR grant boundaries interacted with year fixed effect. Omitted category: t = 1900 X i : soil characteristics, latitude, longitude, (distance to current railroads) Excluded in baseline estimates

45 logged Farm Size Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 45 / 54 (

46 logged Farm Value Farm per Acre Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 46 / 54 (

47 Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 47 / 54 ( Agricultural Adjustments Mechanization Checkerboarding may reduce intensity of lumpy capital Equipment and tractors per acre Crop mix Checkerboarding may reduce productivity and share of crops with larger efficient scale Wheat vs Corn yield and share

48 logged Farm Equipment Value per Farm Acre Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 48 / 54 (

49 logged Tractors per Farm Acre Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 49 / 54 (

50 Wheat Yield: logged Wheat Output per Wheat Acre Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 50 / 54 (

51 Corn Yield: logged Corn Output per Corn Acre Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 51 / 54 (

52 Fraction of Cropland in Wheat Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 52 / 54 (

53 Fraction of Cropland in Corn Dippel, O Grady & Whalley ( Federal trevorogrady@fas.harvard.edu Land Disposal NBER Summer July 10, 2014 Institute 53 / 54 (