Appendix 3A
Table 1
Changing Ethnic
Composition of U.S. 1960-2000
Total
Population
|
White
|
%
|
Black
|
%
|
American
Indian,
Eskimo,
and Aleut
|
%
|
Asian
and
Pacific
Islander
|
%
|
|
1960
|
179,323,175
|
158,831,732
|
88.57
|
18,871,831
|
10.52
|
551,669
|
0.31
|
980,337
|
0.55
|
1970
|
203,211,926
|
177,748,975
|
87.47
|
22,580,289
|
11.11
|
827,255
|
0.41
|
1,538,721
|
0.76
|
1980
|
226,545,805
|
188,371,622
|
83.15
|
26,495,025
|
11.70
|
1,420,400
|
0.63
|
3,500,439
|
1.55
|
1990
|
248,709,873
|
199,686,070
|
80.29
|
29,986,060
|
12.06
|
1,959,234
|
0.79
|
7,273,662
|
2.92
|
2000
|
281,421,906
|
211,353,725
|
75.10
|
34,361,740
|
12.21
|
2,447,989
|
0.87
|
11,569,280
|
4.11
|
Mexican
|
%
|
Other Hispanic
|
%
|
Other
race
|
%
|
Hispanic
origin
(of any
race)
|
%
|
White, not
of Hispanic
origin
|
%
|
|
1960
|
87,606
|
|||||||||
1970
|
4,532,435
|
2.23
|
5,056,781
|
2.49
|
516,686
|
0.25
|
9,589,216
|
4.72
|
169,023,068
|
83.18
|
1980
|
8,740,439
|
3.86
|
6,296,351
|
2.78
|
6,758,319
|
2.98
|
15,036,790
|
6.64
|
180,256,366
|
79.57
|
1990
|
13,495,938
|
5.43
|
8,858,121
|
3.56
|
9,804,847
|
3.94
|
22,354,059
|
8.99
|
188,128,296
|
75.64
|
2000
|
20,640,711
|
7.33
|
14,597,770
|
5.19
|
15,436,924
|
5.49
|
35,238,481
|
12.52
|
194,514,140
|
69.12
|
Lee, Martin and Fogel, Immigrant Stock’s Share of U.S.
Population Growth 1970-2004
Table 2
2010 Population
Total population
308,745,538
|
Hispanic or
Latino
50,477,594
16.35%
|
Non-Hispanic White
196,817,552
63.75%
|
Black
37,685,848
12.21%
|
Asian
14,465,124
4.69%
|
Other
9,299,420
3.01%
|
|
Source: U.S. Census Bureau, 2010 Census
Table 3
Post 1970 Foreign Stock
U.S. Total
|
Foreign Stock Total
|
Foreign Stock % of Total
|
Growth of FS as % of Growth of Pop.
|
Foreign Stock: White, not Hispanic
|
Foreign Stock: Mexican
|
Foreign Stock: Other Hispanic
|
Foreign Stock: Black
|
Foreign Stock: Asian
|
Foreign Stock: Other
|
|
1970
|
203,211,926
|
|||||||||
1980
|
226,545,805
|
8,614,549
|
3.80%
|
36.92%
|
1,610,349
|
2,105,405
|
1,196,689
|
791,717
|
2,267,044
|
643,345
|
1990
|
248,709,873
|
19,926,534
|
8.01%
|
51.04%
|
1,838,476
|
5,700,216
|
3,866,493
|
1,869,415
|
5,786,066
|
865,869
|
2000
|
281,421,906
|
38,287,425
|
13.60%
|
56.13%
|
3,687,870
|
13,525,719
|
6,848,792
|
3,093,964
|
10,028,185
|
1,102,895
|
2004
|
293,645,630
|
45,857,158
|
15.62%
|
61.93%
|
4,399,345
|
16,208,619
|
8,224,327
|
3,710,067
|
11,994,610
|
1,320,190
|
Adapted from Lee, Martin and Fogel, Immigrant Stock’s
Share of U.S. Population Growth 1970-2004
Appendix 3B
Gini Coefficient
The above graph shows the Lorenz curve which plots the cumulative
proportion of the total income of a population. The Gini coefficient is the
ratio of the area that lies between the line of perfect equality and the Lorenz
curve. The line of perfect equality is the 45 degree line; if all members of
the population had the same income the cumulative share would lie along this
line. The Gini coefficient is the area between the line of equality and the
Lorenz curve as a proportion of the total area under the line of equality. It
is therefore given by G=E/(E+F) where E is the area between equality and Lorenz
and F is the area under the Lorenz curve. A low Gini coefficient indicates a more equal distribution; a value of zero
represents complete income equality. Higher Gini coefficients indicate more income
inequality with a value of one corresponding to complete inequality. Notice
that the area of the triangle underneath the line of perfect inequality is ½.
Thus the Gini statistic is:
where the
Lorenz curve is represented by the function Y = f(X).
Appendix 3C
Immigration and Gini
Coefficients
I.
Cross Section State Data 2010 Hispanic Population
Data Sources:
Hispanic Population: US Census Bureau
Gini: 2010 American Community Survey 1-Year Estimates
Regression:
Gini on Percent Hispanic Population by State 2010
Constant Hispanic %
Coefficient
|
0.448
|
0.056
|
S.E.
|
0.004
|
0.029
|
t
statistic
|
106.760
|
1.926
|
F
statistic
|
3.711
|
|
r2
|
0.070
|
II.
U.S. data 1970 – 2010 Hispanic Foreign Stock
Data Sources:
Gini: US Census Bureau Current Population Survey Annual
Social and Economic Supplements
Hispanic Foreign Stock interpolated annually from the
following table:
Post 1970 Foreign Stock
US
|
Mexico
|
Other
Hispanic
|
Total
Hispanic
|
Total
Population
|
Hispanic
FS %
|
1970
|
0
|
0
|
0
|
203,211,926
|
0
|
1980
|
2,105,405
|
1,196,689
|
3,302,094
|
226,545,805
|
0.014576
|
1990
|
5,700,216
|
3,866,493
|
9,566,709
|
248,709,873
|
0.038465
|
2000
|
13,525,719
|
6,848,792
|
20,374,511
|
281,421,906
|
0.072398
|
2004
|
16,208,619
|
8,224,327
|
24,432,946
|
293,656,355
|
0.083203
|
2006
|
16,796,497
|
8,469,111
|
25,265,607
|
298,280,698
|
0.084704
|
2010
|
20,282,604
|
9,413,835
|
29,696,439
|
308,745,538
|
0.096184
|
1970
– 2004 from Lee, Martin and Fogel, Immigrant Stock’s Share of U.S. Population
Growth
2006 from Martin
and Fogel, Projecting the U.S. Population to 2050, unpublished calculation
2010
calculation based on 2010 Census and Martin and Fogel estimates
Regression:
Gini on Percent Hispanic Post Foreign Stock 1970 – 2010
Constant Hispanic %
Coefficient
|
0.395
|
0.873
|
S.E.
|
0.002
|
0.030
|
t
statistic
|
251.255
|
29.450
|
F
statistic
|
867.302
|
|
r2
|
0.957
|
Therefore, Gini = .395 + .873*(Hispanic FS %) and if the
Hispanic percentage doubles from the 10% to 20% the Gini would rise to .57,
levels characteristic of Latin America.
III.
U.S. data 1970 – 2009 Foreign Stock, Trade Balance
Data Sources:
Gini: US Census Bureau Current Population Survey Annual
Social and Economic Supplements
Foreign Stock from
Appendix 3A Table 3 plus Martin and Fogel estimates
Trade Balance from U.S. Census Bureau, Foreign Trade
Division
Regression:
Gini on Percent Foreign Stock 1970 - 2009
Constant FS %
Coefficient
|
0.390
|
0.506
|
S.E.
|
0.002
|
0.015
|
t
statistic
|
254.733
|
32.664
|
F
statistic
|
1066.931
|
|
r2
|
0.966
|
Regression:
Gini on Trade Balance 1970 - 2009
Constant TB
Coefficient
|
0.413
|
-8.7E-05
|
S.E.
|
0.003
|
9.49E-06
|
t
statistic
|
126.504
|
-9.151
|
F
statistic
|
83.745
|
|
r2
|
0.688
|
In this case the large negative trade balance causes the estimated
coefficient to be very small and negative. The fit is reasonably strong but
much less so than in the foreign stock regression.
Regression:
Gini on TB and FS 1970 - 2009
Constant TB FS%
Coefficient
|
0.388
|
1.47E-05
|
0.569
|
S.E.
|
0.002
|
6.19E-06
|
0.030
|
t
statistic
|
232.829
|
2.370
|
18.703
|
F
statistic
|
601.113
|
||
r2
|
0.970
|
||
Multicollinearity causes the trade balance variable to
have the wrong sign making the results uninterpretable. To assess the contribution of each variable
the partial correlation coefficients are computed:
Partial correlation between Gini and foreign stock
holding trade balance constant = 0.951
Partial correlation between Gini and trade balance
holding foreign stock constant = 0.363
Thus holding the trade balance constant the foreign stock
is still highly significant. Holding the foreign stock constant the trade
balance is somewhat significant. The conclusion is that immigration has the
largest effect on income inequality.
Appendix 3D
Immigration
Surplus
The following graph represents the economic analysis of
the benefits of immigration:
W1:
Wage at time 1
Q1:
Quantity of labor demanded at time 1
W2:
Wage at time 2 following high immigration
Q2: Quantity of labor demanded at time 2
The area of the triangle under the demand curve
and above the line W1-A represents the labor market surplus at time
1. The area of the triangle under the demand curve and above the line W2-C
represents the labor market surplus at time 2.
The surplus due to immigration is represented by triangle ABC. The size
of the immigration surplus depends on the steepness of the demand curve. A
horizontal line representing an infinite demand for labor means that there
would be no immigrant surplus for the native population. The evidence indicates
that, in fact, there is at least some immigrant surplus but it goes to the
employers of immigrants. Even so the immigrant surplus is far outweighed by all
of the other costs of immigration.
Appendix 3E
Regression Model: Impact of Immigration
on State and Local Expenditures
There are two important ways in which immigration may
affect expenditures. First there is the interaction between immigration and the
welfare state. Immigrants and their offspring have a demand for social
services, including; welfare, medical assistance and housing as do native born
immigrants. However, the higher level of poverty among recent immigrants has a
greater impact than is the case for the population as a whole. Similarly,
immigrants have the same need for education as do other population groups. However,
a higher level of poverty leads to a higher reliance on public education.
Cultural and language differences require additional and, often expensive,
compensatory programs. The second impact of immigration is due to the effects
on spending for infrastructure and the quality of life. Congestion and
environmental effects due to the increase in population creates the demand for
additional public spending. Maintaining public safety and the quality of life
requires large increases in expenditures. Public utilities and transportation
facilities are severely impacted by an increasing population. In addition, the
most recent immigrant cohorts appear to have a higher need for police, fire and
public transportation than the general population magnifying the effect on
public spending. Furthermore, the higher level of poverty in recent immigrant
groups also means that these groups have less tax paying ability and,
therefore, impose a higher tax burden on the general population.
Estimation Procedure:
It is expected that certain expenditures are more likely
to be sensitive to immigration. Anticipated sensitive categories are education,
medical, social services, public safety, environment and infrastructure.
Expenditures for administration, pensions, social insurance, debt service etc.
are excluded. It seems likely that the impact of immigration on the latter
would be indirect and not out of proportion to the percent of immigrants within
the population. An attempt is made throughout the study to, if anything,
understate the estimated impacts.
A linear regression model was specified. Expenditures for
each category are regressed on populations for various cohorts of the foreign
born and offspring of recent immigrants; as well as age groups of the white and
black populations. The coefficients obtained are a direct measure of the
proportion of expenditures allocated to increases in each group. A preliminary
examination of the data raises two issues to be addressed. In the first place,
a time series analysis is inadequate for data provided by the Census
Bureau. A feasible approach is to look
at a cross section made up of recent data presented for the 50 states. Furthermore,
a simple regression combining all the states into one dataset would obscure the
results since there are different political cultures and institutional
arrangements among the states. Therefore, a set of regional dummy variables was
specified to group states into those with similar attitudes or needs regarding
public assistance, environment, transportation etc.
The estimated coefficients from the model were used to
calculate the impact of the immigrant groups on each expenditure category. The
impacts for each category were summed giving totals for each state. Using the
percentage of the foreign born cohorts, a crude estimate of the net impacts
after tax receipts, were obtained. Applying per capita income data then refined
these crude estimates.
Model
X = C + D + w1 PNHW<18
+ w2 PNHW18-45 + w3
PNHW45< + b1PB<18
+ b2 PB18-45 + b3 PB45<
+ f1 PFB90-00
+ f2 PFB80-90 + f3 PFB<80 + f4 PNBO
where,
X: Expenditures in $thousands for one of these categories: education,
education capital, welfare, hospitals, health, public safety, housing,
environment; years 1992, 1999
C: Constant term measures the level of the Northeast effect
D: Dummy variables: South, North, Midwest, West, Pacific measured as a
deviation from the constant, C representing the northeast states.
P’s: Population numbers for
Ethnic groups – Non-Hispanic
Whites, Blacks, Foreign-born, Other
NHW and Black broken into
following age groups: <18, 18-45, 45<
FB broken into year of
entry: <80, 80-90, 90-2000
w’s: coefficients for white cohorts
b’s: coefficients for black cohorts
f’s: coefficients for foreign cohorts
Estimating the coefficients for each expenditure category
began with running an initial regression on the full series of independent
variables. Variables were discarded for two reasons. In the first place, it was
obvious that a “displacement” effect was present. This effect caused the
appearance of anomalous negative coefficients for certain variables. This could
be caused by the inter-state migration of certain groups. For example, an older
population, e.g. whites, may displace a younger one making it appear as if
there is a negative impact on education. Alternatively if a young immigrant
cohort moves in and older whites move out this may cause a negative correlation
and make it appear as if immigration depresses spending on health. A more
important effect was probably due to the extreme multicollinearity within group
age or time cohorts. One variable in each of the three age cohorts within a
population category would “soak up” all the explanatory power due to that
group. The strategy followed was to eliminate the negative coefficients due to
multicollinearity and displacement effects. In a number of instances,
non-significant coefficients were kept or inserted for the white or black
population, which diminished the relative impact of the immigrant contingent.
This was one more way of keeping the estimates of the immigrant impact
conservative. On the other hand, in
every category significant coefficients for immigrant variables would always be
found; these were usually highly significant. High R2 values were
expected and observed; large populations will engender high spending. The
crucial measures to look at are the coefficients of the various population
groups relative to each other.
Education
XEDU = -1,615,922 DSOUTH + 7.393
PNHW<18 + 5.087 PB<18 + 6.993 PFB90-00 +.
866 PNBO
(-2.786)
(13.528) (3.261) (7.407) (2.524)
R2 = .9901
Education Capital
XEDC = -167,240 + 213,534 DPACIFIC +
170,230DWEST + 79,848 DMIDWEST + 84,923
DNORTH +
(-1.979)
(1.424)
(1.608) (.728) (.611)
.867PNHW<18
+ 1.158PB<18 +.253PFB90-00 + .265 PNBO
(8.595) (4.021) (1.449) (4.177)
R2 = .9750
Welfare
XWEL = 852,663 - 1,877,880 DWEST - 2,208,927 DNORTH - 1,612,748 DSOUTH
(1.167) (-2.204) (-1.795) (-1.736)
+ 3.569 PNHW<18
+ 1.470 PB<18 + 4.747 PFB90-00
(3.954)
(0.586) (4.526)
R2 =
.8975
Hospitals
XHOS = 554,193DPACIFIC + 416,653DWEST
+ 568,527DMIDWEST + 568,803DSOUTH
(1.302)
(1.490)
(1.805)
(1.977)
+ .151 PNHW45< + 4.024 PB45< + 1.821 PFB90-00
(0.840) (5.203) (6.066)
R2 = .8957
Health
XHEA = 497,463 DPACIFIC + 471,124 DNORTH + .418
PNHW45< + .061 PB45< + 1.416 PFB80-90
(2.582)
(2.775)
(6.002) (.152) (9.291)
R2 = .9602
Public Safety
XPS = -265,116 DMIDWEST - 505,236 DSOUTH + 1.243 PNHW18-45 + 2.029 PB<18 + 5.050 PFB80-90
(-1.253) (-2.404) (10.032) (3.160) (6.092)
R2 = .9924
Housing
XHOU = 130,055 – 208,969 DWEST - 150,060 DMIDWEST - 176,939 DNORTH - 240,036 DSOUTH
(2.004) (-2.713) (-1.702) (-1.631) (-3.015)
+ .193 PNHW18-45
+.291PB45< + 1.037 PFB80-90
(4.034) (1.144) (11.445)
R2 = .9594
Environment, Utilities and Infrastructure
XENV =
1,968,325DPACIFIC + 2.256PNHW18-45 +
1.988 PB45< + 11.646 PFB90-00
(1.394)
(3.771)
(.642) (9.664)
R2 = .9601
It is possible that for spending on environment,
infrastructure and housing a density factor may cause the immigration effect to
appear larger than it in fact is. The immigrant contingent tends to settle in
more densely populated urban and suburban areas, magnifying the effect of
congestion on these variables. These expenditures may also be extremely
sensitive to increases in population. The rate of population growth, in
addition to the level of the population, is a factor in these expenditures.
Immigration is the major source of population increase in the U.S. Health
expenditures may also be sensitive to the rate of growth accounting for the
large immigrant coefficients relative to the other groups. In every equation at
least one immigrant coefficient was significant with t statistics ranging from
1.449 to 11.445. For whites one insignificant coefficient was inserted; for
blacks three insignificant coefficients were left in the equations. The pre
1980 foreign-born cohort did not make it into any equation; this may be an
indicator of the changing characteristics of the immigrant flow. The
coefficients, being derivatives of a linear function, are a measure of the
marginal impact on expenditures of each variable.
Appendix 3F
Regression Model: Impact of Immigration
on Mortgage Foreclosures
The following data set was used:
Foreclosure Rustbelt % Hispanic
Rate Foreign Stock
Alabama 0.0383% 0 1.254%
Alaska 0.0582% 0 1.581%
Arizona 0.5264% 0 13.002%
Arkansas 0.0893% 0 2.647%
California 0.7595% 0 23.520%
Colorado 0.2153% 0 7.152%
Connecticut 0.1246% 0 4.120%
Delaware 0.0919% 0 2.805%
DC 0.2274% 0 6.441%
Florida 0.5000% 0 13.967%
Georgia 0.2175% 0 4.825%
Hawaii 0.0655% 0 0.958%
Idaho 0.1203% 0 4.490%
Illinois 0.2038% 1 9.826%
Indiana 0.1888% 1 2.081%
Iowa 0.0501% 0 1.909%
Kansas 0.0591% 0 4.551%
Kentucky 0.0651% 0 1.034%
Louisiana 0.0336% 0 1.542%
Maine 0.0333% 0 0.236%
Maryland 0.1379% 0 3.762%
Massachusetts 0.0887% 0 3.935%
Michigan 0.3000% 1 1.548%
Minnesota 0.0927% 0 1.999%
Mississippi 0.0187% 0 0.800%
Missouri 0.1410% 0 1.118%
Montana 0.0308% 0 0.330%
Nebraska 0.0490% 0 3.880%
Nevada 1.0386% 0 13.840%
New Hampshire 0.1092% 0 0.885%
New Jersey 0.1841% 0 10.097%
New Mexico 0.0566% 0 10.374%
New York 0.0686% 0 9.766%
North Carolina 0.1095% 0 4.414%
North Dakota 0.0140% 0 0.364%
Ohio 0.2236% 1 0.635%
Oklahoma 0.0837% 0 3.225%
Oregon 0.1136% 0 6.085%
Pennsylvania 0.0895% 1 1.036%
Rhode Island 0.1217% 0 7.347%
South Carolina 0.0673% 0 1.913%
South Dakota 0.0119% 0 0.554%
Tennessee 0.1692% 0 1.739%
Texas 0.1132% 0 15.987%
Utah 0.1706% 0 5.413%
Vermont 0.0058% 0 0.303%
Virginia 0.1609% 0 3.848%
Washington 0.1136% 0 4.785%
West Virginia 0.0049% 0 0.209%
Wisconsin 0.0866% 1 2.081%
Wyoming 0.0235% 0 1.514%
Source:
Foreclosure Rate: RealtyTrac® (realtytrac.com) August 2008 U.S.
Foreclosure Market Report and Total Housing Units, July 1, 2008 from US Census
Bureau, Population Estimates Program T2 Housing Unit Estimates
% Hispanic Foreign Stock: US Census Bureau, GCT-T1 Population Estimates Data Set: 2008
Population Estimates
and Martin and Fogel, Projecting the U.S. Population to 2050, FAIR, 2006
Regression: FR
on RD and HFS 1970 - 2009
Constant RD HFS
Coefficient
|
0.000121
|
0.000894
|
0.028117
|
S.E.
|
0.000269
|
0.000576
|
0.003866
|
t statistic
|
0.448431
|
1.551382
|
7.272305
|
F statistic
|
26.64139
|
||
r2
|
0.526079
|
||
The results are
summarized in the following equation:
FR = .0001 + .0009RD + .0281HFS
(1.55)
(7.27)
R2 = 0.53
FR: foreclosure rate for each state = Total Foreclosures
August 2008 from RealtyTrac® (realtytrac.com) divided by Total Housing Units,
July 1, 2008 from US Census Bureau, Population Estimates Program T2 Housing
Unit Estimates
RD: rustbelt dummy variable representing the difference
between the states with depressed industries (PA, OH, IN, IL, MI, WI) and the
remaining states
HFS: Hispanic Foreign Stock from the FAIR demographic
projections for 2006 divided by the July 1, 2008 CB GCT-T1. Population
Estimates for each state
The HFS coefficient is highly
significant while the RD coefficient is somewhat significant (10% level). A 1 percent increase in Hispanic foreign
stock implies a 2.8% rate of increase in the foreclosure rate (not the
foreclosure rate percentage as such). The rustbelt states have a .09
percent foreclosure differential. The R2 of .53 is a fairly high fit
for cross section data. Regressing the
foreclosure rate on the % Hispanic foreign stock without the rustbelt variable
yields a slightly lower R2 of .502 indicating that most of the
explanatory power is due to the immigrant population.
No comments:
Post a Comment